diff options
author | Sebastian Thiel <byronimo@gmail.com> | 2014-02-09 21:23:51 +0100 |
---|---|---|
committer | Sebastian Thiel <byronimo@gmail.com> | 2014-02-09 21:23:51 +0100 |
commit | 15dd52cd578691930cea194e003fa80dd02f40eb (patch) | |
tree | ef4c9c5f705dd1ca743b7ceefe5b91b11ad15010 /git/test | |
parent | 660bdca125aa9dcca7a7730535bec433edb8ba02 (diff) | |
download | gitpython-15dd52cd578691930cea194e003fa80dd02f40eb.tar.gz |
tabs to 4 spaces - overall state of this branch is desolate, but fixable. Needs plenty of work
Diffstat (limited to 'git/test')
51 files changed, 6199 insertions, 6199 deletions
diff --git a/git/test/__init__.py b/git/test/__init__.py index a29d96a7..f71cbdf0 100644 --- a/git/test/__init__.py +++ b/git/test/__init__.py @@ -7,8 +7,8 @@ import git.util def _init_pool(): - """Assure the pool is actually threaded""" - size = 2 - print "Setting ThreadPool to %i" % size - git.util.pool.set_size(size) + """Assure the pool is actually threaded""" + size = 2 + print "Setting ThreadPool to %i" % size + git.util.pool.set_size(size) diff --git a/git/test/db/base.py b/git/test/db/base.py index 0f5eebe4..202a4353 100644 --- a/git/test/db/base.py +++ b/git/test/db/base.py @@ -23,619 +23,619 @@ from git.db.compat import RepoCompatibilityInterface class RepoGlobalsItemDeletorMetaCls(GlobalsItemDeletorMetaCls): - ModuleToDelete = 'RepoBase' - + ModuleToDelete = 'RepoBase' + class RepoBase(TestDBBase): - """Basic test for everything a fully implemented repository should support""" - __metaclass__ = RepoGlobalsItemDeletorMetaCls - - def test_new_should_raise_on_invalid_repo_location(self): - self.failUnlessRaises(InvalidGitRepositoryError, self.RepoCls, tempfile.gettempdir()) + """Basic test for everything a fully implemented repository should support""" + __metaclass__ = RepoGlobalsItemDeletorMetaCls + + def test_new_should_raise_on_invalid_repo_location(self): + self.failUnlessRaises(InvalidGitRepositoryError, self.RepoCls, tempfile.gettempdir()) - def test_new_should_raise_on_non_existant_path(self): - self.failUnlessRaises(NoSuchPathError, self.RepoCls, "repos/foobar") + def test_new_should_raise_on_non_existant_path(self): + self.failUnlessRaises(NoSuchPathError, self.RepoCls, "repos/foobar") - def test_repo_creation_from_different_paths(self): - r_from_gitdir = self.RepoCls(self.rorepo.git_dir) - assert r_from_gitdir.git_dir == self.rorepo.git_dir - assert r_from_gitdir.git_dir.endswith('.git') - assert not self.rorepo.git.working_dir.endswith('.git') - assert r_from_gitdir.git.working_dir == self.rorepo.git.working_dir + def test_repo_creation_from_different_paths(self): + r_from_gitdir = self.RepoCls(self.rorepo.git_dir) + assert r_from_gitdir.git_dir == self.rorepo.git_dir + assert r_from_gitdir.git_dir.endswith('.git') + assert not self.rorepo.git.working_dir.endswith('.git') + assert r_from_gitdir.git.working_dir == self.rorepo.git.working_dir - def test_description(self): - txt = "Test repository" - self.rorepo.description = txt - assert_equal(self.rorepo.description, txt) + def test_description(self): + txt = "Test repository" + self.rorepo.description = txt + assert_equal(self.rorepo.description, txt) - def test_heads_should_return_array_of_head_objects(self): - for head in self.rorepo.heads: - assert_equal(Head, head.__class__) + def test_heads_should_return_array_of_head_objects(self): + for head in self.rorepo.heads: + assert_equal(Head, head.__class__) - def test_heads_should_populate_head_data(self): - for head in self.rorepo.heads: - assert head.name - assert isinstance(head.commit,Commit) - # END for each head - - assert isinstance(self.rorepo.heads.master, Head) - assert isinstance(self.rorepo.heads['master'], Head) - - def test_tree_from_revision(self): - tree = self.rorepo.tree('0.1.6') - assert len(tree.hexsha) == 40 - assert tree.type == "tree" - assert self.rorepo.tree(tree) == tree - - # try from invalid revision that does not exist - self.failUnlessRaises(BadObject, self.rorepo.tree, 'hello world') - - def test_commit_from_revision(self): - commit = self.rorepo.commit('0.1.4') - assert commit.type == 'commit' - assert self.rorepo.commit(commit) == commit + def test_heads_should_populate_head_data(self): + for head in self.rorepo.heads: + assert head.name + assert isinstance(head.commit,Commit) + # END for each head + + assert isinstance(self.rorepo.heads.master, Head) + assert isinstance(self.rorepo.heads['master'], Head) + + def test_tree_from_revision(self): + tree = self.rorepo.tree('0.1.6') + assert len(tree.hexsha) == 40 + assert tree.type == "tree" + assert self.rorepo.tree(tree) == tree + + # try from invalid revision that does not exist + self.failUnlessRaises(BadObject, self.rorepo.tree, 'hello world') + + def test_commit_from_revision(self): + commit = self.rorepo.commit('0.1.4') + assert commit.type == 'commit' + assert self.rorepo.commit(commit) == commit - def test_commits(self): - mc = 10 - commits = list(self.rorepo.iter_commits('0.1.6', max_count=mc)) - assert len(commits) == mc - - c = commits[0] - assert_equal('9a4b1d4d11eee3c5362a4152216376e634bd14cf', c.hexsha) - assert_equal(["c76852d0bff115720af3f27acdb084c59361e5f6"], [p.hexsha for p in c.parents]) - assert_equal("ce41fc29549042f1aa09cc03174896cf23f112e3", c.tree.hexsha) - assert_equal("Michael Trier", c.author.name) - assert_equal("mtrier@gmail.com", c.author.email) - assert_equal(1232829715, c.authored_date) - assert_equal(5*3600, c.author_tz_offset) - assert_equal("Michael Trier", c.committer.name) - assert_equal("mtrier@gmail.com", c.committer.email) - assert_equal(1232829715, c.committed_date) - assert_equal(5*3600, c.committer_tz_offset) - assert_equal("Bumped version 0.1.6\n", c.message) + def test_commits(self): + mc = 10 + commits = list(self.rorepo.iter_commits('0.1.6', max_count=mc)) + assert len(commits) == mc + + c = commits[0] + assert_equal('9a4b1d4d11eee3c5362a4152216376e634bd14cf', c.hexsha) + assert_equal(["c76852d0bff115720af3f27acdb084c59361e5f6"], [p.hexsha for p in c.parents]) + assert_equal("ce41fc29549042f1aa09cc03174896cf23f112e3", c.tree.hexsha) + assert_equal("Michael Trier", c.author.name) + assert_equal("mtrier@gmail.com", c.author.email) + assert_equal(1232829715, c.authored_date) + assert_equal(5*3600, c.author_tz_offset) + assert_equal("Michael Trier", c.committer.name) + assert_equal("mtrier@gmail.com", c.committer.email) + assert_equal(1232829715, c.committed_date) + assert_equal(5*3600, c.committer_tz_offset) + assert_equal("Bumped version 0.1.6\n", c.message) - c = commits[1] - assert isinstance(c.parents, tuple) + c = commits[1] + assert isinstance(c.parents, tuple) - def test_trees(self): - mc = 30 - num_trees = 0 - for tree in self.rorepo.iter_trees('0.1.5', max_count=mc): - num_trees += 1 - assert isinstance(tree, Tree) - # END for each tree - assert num_trees == mc + def test_trees(self): + mc = 30 + num_trees = 0 + for tree in self.rorepo.iter_trees('0.1.5', max_count=mc): + num_trees += 1 + assert isinstance(tree, Tree) + # END for each tree + assert num_trees == mc - def _assert_empty_repo(self, repo): - # test all kinds of things with an empty, freshly initialized repo. - # It should throw good errors - - # entries should be empty - assert len(repo.index.entries) == 0 - - # head is accessible - assert repo.head - assert repo.head.ref - assert not repo.head.is_valid() - - # we can change the head to some other ref - head_ref = Head.from_path(repo, Head.to_full_path('some_head')) - assert not head_ref.is_valid() - repo.head.ref = head_ref - - # is_dirty can handle all kwargs - for args in ((1, 0, 0), (0, 1, 0), (0, 0, 1)): - assert not repo.is_dirty(*args) - # END for each arg - - # we can add a file to the index ( if we are not bare ) - if not repo.bare: - pass - # END test repos with working tree - + def _assert_empty_repo(self, repo): + # test all kinds of things with an empty, freshly initialized repo. + # It should throw good errors + + # entries should be empty + assert len(repo.index.entries) == 0 + + # head is accessible + assert repo.head + assert repo.head.ref + assert not repo.head.is_valid() + + # we can change the head to some other ref + head_ref = Head.from_path(repo, Head.to_full_path('some_head')) + assert not head_ref.is_valid() + repo.head.ref = head_ref + + # is_dirty can handle all kwargs + for args in ((1, 0, 0), (0, 1, 0), (0, 0, 1)): + assert not repo.is_dirty(*args) + # END for each arg + + # we can add a file to the index ( if we are not bare ) + if not repo.bare: + pass + # END test repos with working tree + - def test_init(self): - prev_cwd = os.getcwd() - os.chdir(tempfile.gettempdir()) - git_dir_rela = "repos/foo/bar.git" - del_dir_abs = os.path.abspath("repos") - git_dir_abs = os.path.abspath(git_dir_rela) - try: - # with specific path - for path in (git_dir_rela, git_dir_abs): - r = self.RepoCls.init(path=path, bare=True) - assert isinstance(r, self.RepoCls) - assert r.bare == True - assert os.path.isdir(r.git_dir) - - self._assert_empty_repo(r) - - # test clone - clone_path = path + "_clone" - rc = r.clone(clone_path) - self._assert_empty_repo(rc) - - - try: - shutil.rmtree(clone_path) - except OSError: - # when relative paths are used, the clone may actually be inside - # of the parent directory - pass - # END exception handling - - # try again, this time with the absolute version - rc = self.RepoCls.clone_from(r.git_dir, clone_path) - self._assert_empty_repo(rc) - - shutil.rmtree(git_dir_abs) - try: - shutil.rmtree(clone_path) - except OSError: - # when relative paths are used, the clone may actually be inside - # of the parent directory - pass - # END exception handling - - # END for each path - - os.makedirs(git_dir_rela) - os.chdir(git_dir_rela) - r = self.RepoCls.init(bare=False) - r.bare == False - - self._assert_empty_repo(r) - finally: - try: - shutil.rmtree(del_dir_abs) - except OSError: - pass - os.chdir(prev_cwd) - # END restore previous state - - def test_bare_property(self): - if isinstance(self.rorepo, RepoCompatibilityInterface): - self.rorepo.bare - #END handle compatability - self.rorepo.is_bare + def test_init(self): + prev_cwd = os.getcwd() + os.chdir(tempfile.gettempdir()) + git_dir_rela = "repos/foo/bar.git" + del_dir_abs = os.path.abspath("repos") + git_dir_abs = os.path.abspath(git_dir_rela) + try: + # with specific path + for path in (git_dir_rela, git_dir_abs): + r = self.RepoCls.init(path=path, bare=True) + assert isinstance(r, self.RepoCls) + assert r.bare == True + assert os.path.isdir(r.git_dir) + + self._assert_empty_repo(r) + + # test clone + clone_path = path + "_clone" + rc = r.clone(clone_path) + self._assert_empty_repo(rc) + + + try: + shutil.rmtree(clone_path) + except OSError: + # when relative paths are used, the clone may actually be inside + # of the parent directory + pass + # END exception handling + + # try again, this time with the absolute version + rc = self.RepoCls.clone_from(r.git_dir, clone_path) + self._assert_empty_repo(rc) + + shutil.rmtree(git_dir_abs) + try: + shutil.rmtree(clone_path) + except OSError: + # when relative paths are used, the clone may actually be inside + # of the parent directory + pass + # END exception handling + + # END for each path + + os.makedirs(git_dir_rela) + os.chdir(git_dir_rela) + r = self.RepoCls.init(bare=False) + r.bare == False + + self._assert_empty_repo(r) + finally: + try: + shutil.rmtree(del_dir_abs) + except OSError: + pass + os.chdir(prev_cwd) + # END restore previous state + + def test_bare_property(self): + if isinstance(self.rorepo, RepoCompatibilityInterface): + self.rorepo.bare + #END handle compatability + self.rorepo.is_bare - def test_daemon_export(self): - orig_val = self.rorepo.daemon_export - self.rorepo.daemon_export = not orig_val - assert self.rorepo.daemon_export == ( not orig_val ) - self.rorepo.daemon_export = orig_val - assert self.rorepo.daemon_export == orig_val + def test_daemon_export(self): + orig_val = self.rorepo.daemon_export + self.rorepo.daemon_export = not orig_val + assert self.rorepo.daemon_export == ( not orig_val ) + self.rorepo.daemon_export = orig_val + assert self.rorepo.daemon_export == orig_val - def test_alternates(self): - cur_alternates = self.rorepo.alternates - # empty alternates - self.rorepo.alternates = [] - assert self.rorepo.alternates == [] - alts = [ "other/location", "this/location" ] - self.rorepo.alternates = alts - assert alts == self.rorepo.alternates - self.rorepo.alternates = cur_alternates + def test_alternates(self): + cur_alternates = self.rorepo.alternates + # empty alternates + self.rorepo.alternates = [] + assert self.rorepo.alternates == [] + alts = [ "other/location", "this/location" ] + self.rorepo.alternates = alts + assert alts == self.rorepo.alternates + self.rorepo.alternates = cur_alternates - def test_repr(self): - assert_equal('<git.Repo "%s">' % rorepo_dir(), repr(self.rorepo)) + def test_repr(self): + assert_equal('<git.Repo "%s">' % rorepo_dir(), repr(self.rorepo)) - def test_is_dirty_with_bare_repository(self): - orig_value = self.rorepo._bare - self.rorepo._bare = True - assert_false(self.rorepo.is_dirty()) - self.rorepo._bare = orig_value + def test_is_dirty_with_bare_repository(self): + orig_value = self.rorepo._bare + self.rorepo._bare = True + assert_false(self.rorepo.is_dirty()) + self.rorepo._bare = orig_value - def test_is_dirty(self): - self.rorepo._bare = False - for index in (0,1): - for working_tree in (0,1): - for untracked_files in (0,1): - assert self.rorepo.is_dirty(index, working_tree, untracked_files) in (True, False) - # END untracked files - # END working tree - # END index - orig_val = self.rorepo._bare - self.rorepo._bare = True - assert self.rorepo.is_dirty() == False - self.rorepo._bare = orig_val + def test_is_dirty(self): + self.rorepo._bare = False + for index in (0,1): + for working_tree in (0,1): + for untracked_files in (0,1): + assert self.rorepo.is_dirty(index, working_tree, untracked_files) in (True, False) + # END untracked files + # END working tree + # END index + orig_val = self.rorepo._bare + self.rorepo._bare = True + assert self.rorepo.is_dirty() == False + self.rorepo._bare = orig_val - def test_head(self): - assert self.rorepo.head.reference.object == self.rorepo.active_branch.object + def test_head(self): + assert self.rorepo.head.reference.object == self.rorepo.active_branch.object - def test_index(self): - index = self.rorepo.index - assert isinstance(index, IndexFile) - - def test_tag(self): - assert self.rorepo.tag('0.1.5').commit - assert self.rorepo.tag('refs/tags/0.1.5').commit - - def test_archive(self): - tmpfile = os.tmpfile() - self.rorepo.archive(tmpfile, '0.1.5') - assert tmpfile.tell() - - @patch_object(Git, '_call_process') - def test_should_display_blame_information(self, git): - git.return_value = fixture('blame') - b = self.rorepo.blame( 'master', 'lib/git.py') - assert_equal(13, len(b)) - assert_equal( 2, len(b[0]) ) - # assert_equal(25, reduce(lambda acc, x: acc + len(x[-1]), b)) - assert_equal(hash(b[0][0]), hash(b[9][0])) - c = b[0][0] - assert_true(git.called) - assert_equal(git.call_args, (('blame', 'master', '--', 'lib/git.py'), {'p': True})) - - assert_equal('634396b2f541a9f2d58b00be1a07f0c358b999b3', c.hexsha) - assert_equal('Tom Preston-Werner', c.author.name) - assert_equal('tom@mojombo.com', c.author.email) - assert_equal(1191997100, c.authored_date) - assert_equal('Tom Preston-Werner', c.committer.name) - assert_equal('tom@mojombo.com', c.committer.email) - assert_equal(1191997100, c.committed_date) - assert_equal('initial grit setup', c.message) - - # test the 'lines per commit' entries - tlist = b[0][1] - assert_true( tlist ) - assert_true( isinstance( tlist[0], basestring ) ) - assert_true( len( tlist ) < sum( len(t) for t in tlist ) ) # test for single-char bug - - def test_blame_real(self): - c = 0 - for item in self.rorepo.head.commit.tree.traverse( - predicate=lambda i, d: i.type == 'blob' and i.path.endswith('.py')): - c += 1 - b = self.rorepo.blame(self.rorepo.head, item.path) - #END for each item to traverse - assert c - - def test_untracked_files(self): - base = self.rorepo.working_tree_dir - files = ( join_path_native(base, "__test_myfile"), - join_path_native(base, "__test_other_file") ) - num_recently_untracked = 0 - try: - for fpath in files: - fd = open(fpath,"wb") - fd.close() - # END for each filename - untracked_files = self.rorepo.untracked_files - num_recently_untracked = len(untracked_files) - - # assure we have all names - they are relative to the git-dir - num_test_untracked = 0 - for utfile in untracked_files: - num_test_untracked += join_path_native(base, utfile) in files - assert len(files) == num_test_untracked - finally: - for fpath in files: - if os.path.isfile(fpath): - os.remove(fpath) - # END handle files - - assert len(self.rorepo.untracked_files) == (num_recently_untracked - len(files)) - - def test_config_reader(self): - reader = self.rorepo.config_reader() # all config files - assert reader.read_only - reader = self.rorepo.config_reader("repository") # single config file - assert reader.read_only - - def test_config_writer(self): - for config_level in self.rorepo.config_level: - try: - writer = self.rorepo.config_writer(config_level) - assert not writer.read_only - except IOError: - # its okay not to get a writer for some configuration files if we - # have no permissions - pass - # END for each config level - - def test_creation_deletion(self): - # just a very quick test to assure it generally works. There are - # specialized cases in the test_refs module - head = self.rorepo.create_head("new_head", "HEAD~1") - self.rorepo.delete_head(head) - - tag = self.rorepo.create_tag("new_tag", "HEAD~2") - self.rorepo.delete_tag(tag) - self.rorepo.config_writer() - remote = self.rorepo.create_remote("new_remote", "git@server:repo.git") - self.rorepo.delete_remote(remote) - - def test_comparison_and_hash(self): - # this is only a preliminary test, more testing done in test_index - assert self.rorepo == self.rorepo and not (self.rorepo != self.rorepo) - assert len(set((self.rorepo, self.rorepo))) == 1 - - def test_git_cmd(self): - # test CatFileContentStream, just to be very sure we have no fencepost errors - # last \n is the terminating newline that it expects - l1 = "0123456789\n" - l2 = "abcdefghijklmnopqrstxy\n" - l3 = "z\n" - d = "%s%s%s\n" % (l1, l2, l3) - - l1p = l1[:5] - - # full size - # size is without terminating newline - def mkfull(): - return Git.CatFileContentStream(len(d)-1, StringIO(d)) - - ts = 5 - def mktiny(): - return Git.CatFileContentStream(ts, StringIO(d)) - - # readlines no limit - s = mkfull() - lines = s.readlines() - assert len(lines) == 3 and lines[-1].endswith('\n') - assert s._stream.tell() == len(d) # must have scrubbed to the end - - # realines line limit - s = mkfull() - lines = s.readlines(5) - assert len(lines) == 1 - - # readlines on tiny sections - s = mktiny() - lines = s.readlines() - assert len(lines) == 1 and lines[0] == l1p - assert s._stream.tell() == ts+1 - - # readline no limit - s = mkfull() - assert s.readline() == l1 - assert s.readline() == l2 - assert s.readline() == l3 - assert s.readline() == '' - assert s._stream.tell() == len(d) - - # readline limit - s = mkfull() - assert s.readline(5) == l1p - assert s.readline() == l1[5:] - - # readline on tiny section - s = mktiny() - assert s.readline() == l1p - assert s.readline() == '' - assert s._stream.tell() == ts+1 - - # read no limit - s = mkfull() - assert s.read() == d[:-1] - assert s.read() == '' - assert s._stream.tell() == len(d) - - # read limit - s = mkfull() - assert s.read(5) == l1p - assert s.read(6) == l1[5:] - assert s._stream.tell() == 5 + 6 # its not yet done - - # read tiny - s = mktiny() - assert s.read(2) == l1[:2] - assert s._stream.tell() == 2 - assert s.read() == l1[2:ts] - assert s._stream.tell() == ts+1 - - def _assert_rev_parse_types(self, name, rev_obj): - rev_parse = self.rorepo.rev_parse - - if rev_obj.type == 'tag': - rev_obj = rev_obj.object - - # tree and blob type - obj = rev_parse(name + '^{tree}') - assert obj == rev_obj.tree - - obj = rev_parse(name + ':CHANGES') - assert obj.type == 'blob' and obj.path == 'CHANGES' - assert rev_obj.tree['CHANGES'] == obj - - - def _assert_rev_parse(self, name): - """tries multiple different rev-parse syntaxes with the given name - :return: parsed object""" - rev_parse = self.rorepo.rev_parse - orig_obj = rev_parse(name) - if orig_obj.type == 'tag': - obj = orig_obj.object - else: - obj = orig_obj - # END deref tags by default - - # try history - rev = name + "~" - obj2 = rev_parse(rev) - assert obj2 == obj.parents[0] - self._assert_rev_parse_types(rev, obj2) - - # history with number - ni = 11 - history = [obj.parents[0]] - for pn in range(ni): - history.append(history[-1].parents[0]) - # END get given amount of commits - - for pn in range(11): - rev = name + "~%i" % (pn+1) - obj2 = rev_parse(rev) - assert obj2 == history[pn] - self._assert_rev_parse_types(rev, obj2) - # END history check - - # parent ( default ) - rev = name + "^" - obj2 = rev_parse(rev) - assert obj2 == obj.parents[0] - self._assert_rev_parse_types(rev, obj2) - - # parent with number - for pn, parent in enumerate(obj.parents): - rev = name + "^%i" % (pn+1) - assert rev_parse(rev) == parent - self._assert_rev_parse_types(rev, parent) - # END for each parent - - return orig_obj - - @with_rw_repo('HEAD', bare=False) - def test_rw_rev_parse(self, rwrepo): - # verify it does not confuse branches with hexsha ids - ahead = rwrepo.create_head('aaaaaaaa') - assert(rwrepo.rev_parse(str(ahead)) == ahead.commit) - - def test_rev_parse(self): - rev_parse = self.rorepo.rev_parse - - # try special case: This one failed at some point, make sure its fixed - assert rev_parse("33ebe").hexsha == "33ebe7acec14b25c5f84f35a664803fcab2f7781" - - # start from reference - num_resolved = 0 - - for ref in Reference.iter_items(self.rorepo): - path_tokens = ref.path.split("/") - for pt in range(len(path_tokens)): - path_section = '/'.join(path_tokens[-(pt+1):]) - try: - obj = self._assert_rev_parse(path_section) - assert obj.type == ref.object.type - num_resolved += 1 - except BadObject: - print "failed on %s" % path_section - # is fine, in case we have something like 112, which belongs to remotes/rname/merge-requests/112 - pass - # END exception handling - # END for each token - # END for each reference - assert num_resolved - - # it works with tags ! - tag = self._assert_rev_parse('0.1.4') - assert tag.type == 'tag' - - # try full sha directly ( including type conversion ) - assert tag.object == rev_parse(tag.object.hexsha) - self._assert_rev_parse_types(tag.object.hexsha, tag.object) - - - # multiple tree types result in the same tree: HEAD^{tree}^{tree}:CHANGES - rev = '0.1.4^{tree}^{tree}' - assert rev_parse(rev) == tag.object.tree - assert rev_parse(rev+':CHANGES') == tag.object.tree['CHANGES'] - - - # try to get parents from first revision - it should fail as no such revision - # exists - first_rev = "33ebe7acec14b25c5f84f35a664803fcab2f7781" - commit = rev_parse(first_rev) - assert len(commit.parents) == 0 - assert commit.hexsha == first_rev - self.failUnlessRaises(BadObject, rev_parse, first_rev+"~") - self.failUnlessRaises(BadObject, rev_parse, first_rev+"^") - - # short SHA1 - commit2 = rev_parse(first_rev[:20]) - assert commit2 == commit - commit2 = rev_parse(first_rev[:5]) - assert commit2 == commit - - - # todo: dereference tag into a blob 0.1.7^{blob} - quite a special one - # needs a tag which points to a blob - - - # ref^0 returns commit being pointed to, same with ref~0, and ^{} - tag = rev_parse('0.1.4') - for token in (('~0', '^0', '^{}')): - assert tag.object == rev_parse('0.1.4%s' % token) - # END handle multiple tokens - - # try partial parsing - max_items = 40 - for i, binsha in enumerate(self.rorepo.odb.sha_iter()): - assert rev_parse(bin_to_hex(binsha)[:8-(i%2)]).binsha == binsha - if i > max_items: - # this is rather slow currently, as rev_parse returns an object - # which requires accessing packs, it has some additional overhead - break - # END for each binsha in repo - - # missing closing brace commit^{tree - self.failUnlessRaises(ValueError, rev_parse, '0.1.4^{tree') - - # missing starting brace - self.failUnlessRaises(ValueError, rev_parse, '0.1.4^tree}') - - # REVLOG - ####### - head = self.rorepo.head - - # need to specify a ref when using the @ syntax - self.failUnlessRaises(BadObject, rev_parse, "%s@{0}" % head.commit.hexsha) - - # uses HEAD.ref by default - assert rev_parse('@{0}') == head.commit - if not head.is_detached: - refspec = '%s@{0}' % head.ref.name - assert rev_parse(refspec) == head.ref.commit - # all additional specs work as well - assert rev_parse(refspec+"^{tree}") == head.commit.tree - assert rev_parse(refspec+":CHANGES").type == 'blob' - #END operate on non-detached head - - # the most recent previous position of the currently checked out branch - - try: - assert rev_parse('@{1}') != head.commit - except IndexError: - # on new checkouts, there isn't even a single past branch position - # in the log - pass - #END handle fresh checkouts - - # position doesn't exist - self.failUnlessRaises(IndexError, rev_parse, '@{10000}') - - # currently, nothing more is supported - self.failUnlessRaises(NotImplementedError, rev_parse, "@{1 week ago}") - - def test_submodules(self): - assert len(self.rorepo.submodules) == 2 # non-recursive - # in previous configurations, we had recursive repositories so this would compare to 2 - # now there is only one left, as gitdb was merged, but we have smmap instead - assert len(list(self.rorepo.iter_submodules())) == 2 - - assert isinstance(self.rorepo.submodule("async"), Submodule) - self.failUnlessRaises(ValueError, self.rorepo.submodule, "doesn't exist") - - @with_rw_repo('HEAD', bare=False) - def test_submodule_update(self, rwrepo): - # fails in bare mode - rwrepo._bare = True - # special handling: there are repo implementations which have a bare attribute. IN that case, set it directly - if not rwrepo.bare: - rwrepo.bare = True - self.failUnlessRaises(InvalidGitRepositoryError, rwrepo.submodule_update) - rwrepo._bare = False - if rwrepo.bare: - rwrepo.bare = False - #END special repo handling - - # test create submodule - sm = rwrepo.submodules[0] - sm = rwrepo.create_submodule("my_new_sub", "some_path", join_path_native(self.rorepo.working_tree_dir, sm.path)) - assert isinstance(sm, Submodule) - - # note: the rest of this functionality is tested in test_submodule - - + def test_index(self): + index = self.rorepo.index + assert isinstance(index, IndexFile) + + def test_tag(self): + assert self.rorepo.tag('0.1.5').commit + assert self.rorepo.tag('refs/tags/0.1.5').commit + + def test_archive(self): + tmpfile = os.tmpfile() + self.rorepo.archive(tmpfile, '0.1.5') + assert tmpfile.tell() + + @patch_object(Git, '_call_process') + def test_should_display_blame_information(self, git): + git.return_value = fixture('blame') + b = self.rorepo.blame( 'master', 'lib/git.py') + assert_equal(13, len(b)) + assert_equal( 2, len(b[0]) ) + # assert_equal(25, reduce(lambda acc, x: acc + len(x[-1]), b)) + assert_equal(hash(b[0][0]), hash(b[9][0])) + c = b[0][0] + assert_true(git.called) + assert_equal(git.call_args, (('blame', 'master', '--', 'lib/git.py'), {'p': True})) + + assert_equal('634396b2f541a9f2d58b00be1a07f0c358b999b3', c.hexsha) + assert_equal('Tom Preston-Werner', c.author.name) + assert_equal('tom@mojombo.com', c.author.email) + assert_equal(1191997100, c.authored_date) + assert_equal('Tom Preston-Werner', c.committer.name) + assert_equal('tom@mojombo.com', c.committer.email) + assert_equal(1191997100, c.committed_date) + assert_equal('initial grit setup', c.message) + + # test the 'lines per commit' entries + tlist = b[0][1] + assert_true( tlist ) + assert_true( isinstance( tlist[0], basestring ) ) + assert_true( len( tlist ) < sum( len(t) for t in tlist ) ) # test for single-char bug + + def test_blame_real(self): + c = 0 + for item in self.rorepo.head.commit.tree.traverse( + predicate=lambda i, d: i.type == 'blob' and i.path.endswith('.py')): + c += 1 + b = self.rorepo.blame(self.rorepo.head, item.path) + #END for each item to traverse + assert c + + def test_untracked_files(self): + base = self.rorepo.working_tree_dir + files = ( join_path_native(base, "__test_myfile"), + join_path_native(base, "__test_other_file") ) + num_recently_untracked = 0 + try: + for fpath in files: + fd = open(fpath,"wb") + fd.close() + # END for each filename + untracked_files = self.rorepo.untracked_files + num_recently_untracked = len(untracked_files) + + # assure we have all names - they are relative to the git-dir + num_test_untracked = 0 + for utfile in untracked_files: + num_test_untracked += join_path_native(base, utfile) in files + assert len(files) == num_test_untracked + finally: + for fpath in files: + if os.path.isfile(fpath): + os.remove(fpath) + # END handle files + + assert len(self.rorepo.untracked_files) == (num_recently_untracked - len(files)) + + def test_config_reader(self): + reader = self.rorepo.config_reader() # all config files + assert reader.read_only + reader = self.rorepo.config_reader("repository") # single config file + assert reader.read_only + + def test_config_writer(self): + for config_level in self.rorepo.config_level: + try: + writer = self.rorepo.config_writer(config_level) + assert not writer.read_only + except IOError: + # its okay not to get a writer for some configuration files if we + # have no permissions + pass + # END for each config level + + def test_creation_deletion(self): + # just a very quick test to assure it generally works. There are + # specialized cases in the test_refs module + head = self.rorepo.create_head("new_head", "HEAD~1") + self.rorepo.delete_head(head) + + tag = self.rorepo.create_tag("new_tag", "HEAD~2") + self.rorepo.delete_tag(tag) + self.rorepo.config_writer() + remote = self.rorepo.create_remote("new_remote", "git@server:repo.git") + self.rorepo.delete_remote(remote) + + def test_comparison_and_hash(self): + # this is only a preliminary test, more testing done in test_index + assert self.rorepo == self.rorepo and not (self.rorepo != self.rorepo) + assert len(set((self.rorepo, self.rorepo))) == 1 + + def test_git_cmd(self): + # test CatFileContentStream, just to be very sure we have no fencepost errors + # last \n is the terminating newline that it expects + l1 = "0123456789\n" + l2 = "abcdefghijklmnopqrstxy\n" + l3 = "z\n" + d = "%s%s%s\n" % (l1, l2, l3) + + l1p = l1[:5] + + # full size + # size is without terminating newline + def mkfull(): + return Git.CatFileContentStream(len(d)-1, StringIO(d)) + + ts = 5 + def mktiny(): + return Git.CatFileContentStream(ts, StringIO(d)) + + # readlines no limit + s = mkfull() + lines = s.readlines() + assert len(lines) == 3 and lines[-1].endswith('\n') + assert s._stream.tell() == len(d) # must have scrubbed to the end + + # realines line limit + s = mkfull() + lines = s.readlines(5) + assert len(lines) == 1 + + # readlines on tiny sections + s = mktiny() + lines = s.readlines() + assert len(lines) == 1 and lines[0] == l1p + assert s._stream.tell() == ts+1 + + # readline no limit + s = mkfull() + assert s.readline() == l1 + assert s.readline() == l2 + assert s.readline() == l3 + assert s.readline() == '' + assert s._stream.tell() == len(d) + + # readline limit + s = mkfull() + assert s.readline(5) == l1p + assert s.readline() == l1[5:] + + # readline on tiny section + s = mktiny() + assert s.readline() == l1p + assert s.readline() == '' + assert s._stream.tell() == ts+1 + + # read no limit + s = mkfull() + assert s.read() == d[:-1] + assert s.read() == '' + assert s._stream.tell() == len(d) + + # read limit + s = mkfull() + assert s.read(5) == l1p + assert s.read(6) == l1[5:] + assert s._stream.tell() == 5 + 6 # its not yet done + + # read tiny + s = mktiny() + assert s.read(2) == l1[:2] + assert s._stream.tell() == 2 + assert s.read() == l1[2:ts] + assert s._stream.tell() == ts+1 + + def _assert_rev_parse_types(self, name, rev_obj): + rev_parse = self.rorepo.rev_parse + + if rev_obj.type == 'tag': + rev_obj = rev_obj.object + + # tree and blob type + obj = rev_parse(name + '^{tree}') + assert obj == rev_obj.tree + + obj = rev_parse(name + ':CHANGES') + assert obj.type == 'blob' and obj.path == 'CHANGES' + assert rev_obj.tree['CHANGES'] == obj + + + def _assert_rev_parse(self, name): + """tries multiple different rev-parse syntaxes with the given name + :return: parsed object""" + rev_parse = self.rorepo.rev_parse + orig_obj = rev_parse(name) + if orig_obj.type == 'tag': + obj = orig_obj.object + else: + obj = orig_obj + # END deref tags by default + + # try history + rev = name + "~" + obj2 = rev_parse(rev) + assert obj2 == obj.parents[0] + self._assert_rev_parse_types(rev, obj2) + + # history with number + ni = 11 + history = [obj.parents[0]] + for pn in range(ni): + history.append(history[-1].parents[0]) + # END get given amount of commits + + for pn in range(11): + rev = name + "~%i" % (pn+1) + obj2 = rev_parse(rev) + assert obj2 == history[pn] + self._assert_rev_parse_types(rev, obj2) + # END history check + + # parent ( default ) + rev = name + "^" + obj2 = rev_parse(rev) + assert obj2 == obj.parents[0] + self._assert_rev_parse_types(rev, obj2) + + # parent with number + for pn, parent in enumerate(obj.parents): + rev = name + "^%i" % (pn+1) + assert rev_parse(rev) == parent + self._assert_rev_parse_types(rev, parent) + # END for each parent + + return orig_obj + + @with_rw_repo('HEAD', bare=False) + def test_rw_rev_parse(self, rwrepo): + # verify it does not confuse branches with hexsha ids + ahead = rwrepo.create_head('aaaaaaaa') + assert(rwrepo.rev_parse(str(ahead)) == ahead.commit) + + def test_rev_parse(self): + rev_parse = self.rorepo.rev_parse + + # try special case: This one failed at some point, make sure its fixed + assert rev_parse("33ebe").hexsha == "33ebe7acec14b25c5f84f35a664803fcab2f7781" + + # start from reference + num_resolved = 0 + + for ref in Reference.iter_items(self.rorepo): + path_tokens = ref.path.split("/") + for pt in range(len(path_tokens)): + path_section = '/'.join(path_tokens[-(pt+1):]) + try: + obj = self._assert_rev_parse(path_section) + assert obj.type == ref.object.type + num_resolved += 1 + except BadObject: + print "failed on %s" % path_section + # is fine, in case we have something like 112, which belongs to remotes/rname/merge-requests/112 + pass + # END exception handling + # END for each token + # END for each reference + assert num_resolved + + # it works with tags ! + tag = self._assert_rev_parse('0.1.4') + assert tag.type == 'tag' + + # try full sha directly ( including type conversion ) + assert tag.object == rev_parse(tag.object.hexsha) + self._assert_rev_parse_types(tag.object.hexsha, tag.object) + + + # multiple tree types result in the same tree: HEAD^{tree}^{tree}:CHANGES + rev = '0.1.4^{tree}^{tree}' + assert rev_parse(rev) == tag.object.tree + assert rev_parse(rev+':CHANGES') == tag.object.tree['CHANGES'] + + + # try to get parents from first revision - it should fail as no such revision + # exists + first_rev = "33ebe7acec14b25c5f84f35a664803fcab2f7781" + commit = rev_parse(first_rev) + assert len(commit.parents) == 0 + assert commit.hexsha == first_rev + self.failUnlessRaises(BadObject, rev_parse, first_rev+"~") + self.failUnlessRaises(BadObject, rev_parse, first_rev+"^") + + # short SHA1 + commit2 = rev_parse(first_rev[:20]) + assert commit2 == commit + commit2 = rev_parse(first_rev[:5]) + assert commit2 == commit + + + # todo: dereference tag into a blob 0.1.7^{blob} - quite a special one + # needs a tag which points to a blob + + + # ref^0 returns commit being pointed to, same with ref~0, and ^{} + tag = rev_parse('0.1.4') + for token in (('~0', '^0', '^{}')): + assert tag.object == rev_parse('0.1.4%s' % token) + # END handle multiple tokens + + # try partial parsing + max_items = 40 + for i, binsha in enumerate(self.rorepo.odb.sha_iter()): + assert rev_parse(bin_to_hex(binsha)[:8-(i%2)]).binsha == binsha + if i > max_items: + # this is rather slow currently, as rev_parse returns an object + # which requires accessing packs, it has some additional overhead + break + # END for each binsha in repo + + # missing closing brace commit^{tree + self.failUnlessRaises(ValueError, rev_parse, '0.1.4^{tree') + + # missing starting brace + self.failUnlessRaises(ValueError, rev_parse, '0.1.4^tree}') + + # REVLOG + ####### + head = self.rorepo.head + + # need to specify a ref when using the @ syntax + self.failUnlessRaises(BadObject, rev_parse, "%s@{0}" % head.commit.hexsha) + + # uses HEAD.ref by default + assert rev_parse('@{0}') == head.commit + if not head.is_detached: + refspec = '%s@{0}' % head.ref.name + assert rev_parse(refspec) == head.ref.commit + # all additional specs work as well + assert rev_parse(refspec+"^{tree}") == head.commit.tree + assert rev_parse(refspec+":CHANGES").type == 'blob' + #END operate on non-detached head + + # the most recent previous position of the currently checked out branch + + try: + assert rev_parse('@{1}') != head.commit + except IndexError: + # on new checkouts, there isn't even a single past branch position + # in the log + pass + #END handle fresh checkouts + + # position doesn't exist + self.failUnlessRaises(IndexError, rev_parse, '@{10000}') + + # currently, nothing more is supported + self.failUnlessRaises(NotImplementedError, rev_parse, "@{1 week ago}") + + def test_submodules(self): + assert len(self.rorepo.submodules) == 2 # non-recursive + # in previous configurations, we had recursive repositories so this would compare to 2 + # now there is only one left, as gitdb was merged, but we have smmap instead + assert len(list(self.rorepo.iter_submodules())) == 2 + + assert isinstance(self.rorepo.submodule("async"), Submodule) + self.failUnlessRaises(ValueError, self.rorepo.submodule, "doesn't exist") + + @with_rw_repo('HEAD', bare=False) + def test_submodule_update(self, rwrepo): + # fails in bare mode + rwrepo._bare = True + # special handling: there are repo implementations which have a bare attribute. IN that case, set it directly + if not rwrepo.bare: + rwrepo.bare = True + self.failUnlessRaises(InvalidGitRepositoryError, rwrepo.submodule_update) + rwrepo._bare = False + if rwrepo.bare: + rwrepo.bare = False + #END special repo handling + + # test create submodule + sm = rwrepo.submodules[0] + sm = rwrepo.create_submodule("my_new_sub", "some_path", join_path_native(self.rorepo.working_tree_dir, sm.path)) + assert isinstance(sm, Submodule) + + # note: the rest of this functionality is tested in test_submodule + + diff --git a/git/test/db/cmd/test_base.py b/git/test/db/cmd/test_base.py index cbb4a339..890c0232 100644 --- a/git/test/db/cmd/test_base.py +++ b/git/test/db/cmd/test_base.py @@ -14,78 +14,78 @@ from git.db.cmd.base import * from git.refs import TagReference, Reference, RemoteReference class TestBase(RepoBase): - RepoCls = CmdCompatibilityGitDB + RepoCls = CmdCompatibilityGitDB - def test_basics(self): - gdb = self.rorepo - - # partial to complete - works with everything - hexsha = bin_to_hex(gdb.partial_to_complete_sha_hex("0.1.6")) - assert len(hexsha) == 40 - - assert bin_to_hex(gdb.partial_to_complete_sha_hex(hexsha[:20])) == hexsha - - # fails with BadObject - for invalid_rev in ("0000", "bad/ref", "super bad"): - self.failUnlessRaises(BadObject, gdb.partial_to_complete_sha_hex, invalid_rev) - - def test_fetch_info(self): - self.failUnlessRaises(ValueError, CmdCmdFetchInfo._from_line, self.rorepo, "nonsense", '') - self.failUnlessRaises(ValueError, CmdCmdFetchInfo._from_line, self.rorepo, "? [up to date] 0.1.7RC -> origin/0.1.7RC", '') - - - def test_fetch_info(self): - # assure we can handle remote-tracking branches - fetch_info_line_fmt = "c437ee5deb8d00cf02f03720693e4c802e99f390 not-for-merge %s '0.3' of git://github.com/gitpython-developers/GitPython" - remote_info_line_fmt = "* [new branch] nomatter -> %s" - fi = CmdFetchInfo._from_line(self.rorepo, - remote_info_line_fmt % "local/master", - fetch_info_line_fmt % 'remote-tracking branch') - - # we wouldn't be here if it wouldn't have worked - - # handles non-default refspecs: One can specify a different path in refs/remotes - # or a special path just in refs/something for instance - - fi = CmdFetchInfo._from_line(self.rorepo, - remote_info_line_fmt % "subdir/tagname", - fetch_info_line_fmt % 'tag') - - assert isinstance(fi.ref, TagReference) - assert fi.ref.path.startswith('refs/tags') - - # it could be in a remote direcftory though - fi = CmdFetchInfo._from_line(self.rorepo, - remote_info_line_fmt % "remotename/tags/tagname", - fetch_info_line_fmt % 'tag') - - assert isinstance(fi.ref, TagReference) - assert fi.ref.path.startswith('refs/remotes/') - - # it can also be anywhere ! - tag_path = "refs/something/remotename/tags/tagname" - fi = CmdFetchInfo._from_line(self.rorepo, - remote_info_line_fmt % tag_path, - fetch_info_line_fmt % 'tag') - - assert isinstance(fi.ref, TagReference) - assert fi.ref.path == tag_path - - # branches default to refs/remotes - fi = CmdFetchInfo._from_line(self.rorepo, - remote_info_line_fmt % "remotename/branch", - fetch_info_line_fmt % 'branch') - - assert isinstance(fi.ref, RemoteReference) - assert fi.ref.remote_name == 'remotename' - - # but you can force it anywhere, in which case we only have a references - fi = CmdFetchInfo._from_line(self.rorepo, - remote_info_line_fmt % "refs/something/branch", - fetch_info_line_fmt % 'branch') - - assert type(fi.ref) is Reference - assert fi.ref.path == "refs/something/branch" - - - + def test_basics(self): + gdb = self.rorepo + + # partial to complete - works with everything + hexsha = bin_to_hex(gdb.partial_to_complete_sha_hex("0.1.6")) + assert len(hexsha) == 40 + + assert bin_to_hex(gdb.partial_to_complete_sha_hex(hexsha[:20])) == hexsha + + # fails with BadObject + for invalid_rev in ("0000", "bad/ref", "super bad"): + self.failUnlessRaises(BadObject, gdb.partial_to_complete_sha_hex, invalid_rev) + + def test_fetch_info(self): + self.failUnlessRaises(ValueError, CmdCmdFetchInfo._from_line, self.rorepo, "nonsense", '') + self.failUnlessRaises(ValueError, CmdCmdFetchInfo._from_line, self.rorepo, "? [up to date] 0.1.7RC -> origin/0.1.7RC", '') + + + def test_fetch_info(self): + # assure we can handle remote-tracking branches + fetch_info_line_fmt = "c437ee5deb8d00cf02f03720693e4c802e99f390 not-for-merge %s '0.3' of git://github.com/gitpython-developers/GitPython" + remote_info_line_fmt = "* [new branch] nomatter -> %s" + fi = CmdFetchInfo._from_line(self.rorepo, + remote_info_line_fmt % "local/master", + fetch_info_line_fmt % 'remote-tracking branch') + + # we wouldn't be here if it wouldn't have worked + + # handles non-default refspecs: One can specify a different path in refs/remotes + # or a special path just in refs/something for instance + + fi = CmdFetchInfo._from_line(self.rorepo, + remote_info_line_fmt % "subdir/tagname", + fetch_info_line_fmt % 'tag') + + assert isinstance(fi.ref, TagReference) + assert fi.ref.path.startswith('refs/tags') + + # it could be in a remote direcftory though + fi = CmdFetchInfo._from_line(self.rorepo, + remote_info_line_fmt % "remotename/tags/tagname", + fetch_info_line_fmt % 'tag') + + assert isinstance(fi.ref, TagReference) + assert fi.ref.path.startswith('refs/remotes/') + + # it can also be anywhere ! + tag_path = "refs/something/remotename/tags/tagname" + fi = CmdFetchInfo._from_line(self.rorepo, + remote_info_line_fmt % tag_path, + fetch_info_line_fmt % 'tag') + + assert isinstance(fi.ref, TagReference) + assert fi.ref.path == tag_path + + # branches default to refs/remotes + fi = CmdFetchInfo._from_line(self.rorepo, + remote_info_line_fmt % "remotename/branch", + fetch_info_line_fmt % 'branch') + + assert isinstance(fi.ref, RemoteReference) + assert fi.ref.remote_name == 'remotename' + + # but you can force it anywhere, in which case we only have a references + fi = CmdFetchInfo._from_line(self.rorepo, + remote_info_line_fmt % "refs/something/branch", + fetch_info_line_fmt % 'branch') + + assert type(fi.ref) is Reference + assert fi.ref.path == "refs/something/branch" + + + diff --git a/git/test/db/dulwich/lib.py b/git/test/db/dulwich/lib.py index 56734064..a58469f1 100644 --- a/git/test/db/dulwich/lib.py +++ b/git/test/db/dulwich/lib.py @@ -1,23 +1,23 @@ """dulwich specific utilities, as well as all the default ones""" from git.test.lib import ( - InheritedTestMethodsOverrideWrapperMetaClsAutoMixin, - needs_module_or_skip - ) + InheritedTestMethodsOverrideWrapperMetaClsAutoMixin, + needs_module_or_skip + ) __all__ = ['needs_dulwich_or_skip', 'DulwichRequiredMetaMixin'] #{ Decoorators def needs_dulwich_or_skip(func): - """Skip this test if we have no dulwich - print warning""" - return needs_module_or_skip('dulwich')(func) + """Skip this test if we have no dulwich - print warning""" + return needs_module_or_skip('dulwich')(func) #}END decorators #{ MetaClasses class DulwichRequiredMetaMixin(InheritedTestMethodsOverrideWrapperMetaClsAutoMixin): - decorator = [needs_dulwich_or_skip] + decorator = [needs_dulwich_or_skip] #} END metaclasses diff --git a/git/test/db/dulwich/test_base.py b/git/test/db/dulwich/test_base.py index 78416518..ed2f8975 100644 --- a/git/test/db/dulwich/test_base.py +++ b/git/test/db/dulwich/test_base.py @@ -9,24 +9,24 @@ from git.test.db.base import RepoBase try: - import dulwich + import dulwich except ImportError: - # om this case, all other dulwich tests will be skipped - # Need to properly initialize the class though, otherwise it would fail - from git.db.complex import PureCompatibilityGitDB as DulwichDB + # om this case, all other dulwich tests will be skipped + # Need to properly initialize the class though, otherwise it would fail + from git.db.complex import PureCompatibilityGitDB as DulwichDB else: - # now we know dulwich is available, to do futher imports - from git.db.dulwich.complex import DulwichCompatibilityGitDB as DulwichDB - + # now we know dulwich is available, to do futher imports + from git.db.dulwich.complex import DulwichCompatibilityGitDB as DulwichDB + #END handle imports class TestDulwichDBBase(RepoBase): - __metaclass__ = DulwichRequiredMetaMixin - RepoCls = DulwichDB - - @needs_dulwich_or_skip - @with_rw_repo('HEAD', bare=False) - def test_basics(self, rw_repo): - db = DulwichDB(rw_repo.working_tree_dir) - - + __metaclass__ = DulwichRequiredMetaMixin + RepoCls = DulwichDB + + @needs_dulwich_or_skip + @with_rw_repo('HEAD', bare=False) + def test_basics(self, rw_repo): + db = DulwichDB(rw_repo.working_tree_dir) + + diff --git a/git/test/db/lib.py b/git/test/db/lib.py index 2b3ddde5..df9fec76 100644 --- a/git/test/db/lib.py +++ b/git/test/db/lib.py @@ -4,21 +4,21 @@ # the New BSD License: http://www.opensource.org/licenses/bsd-license.php """Base classes for object db testing""" from git.test.lib import ( - with_rw_directory, - with_packs_rw, - ZippedStoreShaWriter, - fixture_path, - TestBase, - rorepo_dir, - ) + with_rw_directory, + with_packs_rw, + ZippedStoreShaWriter, + fixture_path, + TestBase, + rorepo_dir, + ) from git.stream import Sha1Writer from git.base import ( - IStream, - OStream, - OInfo - ) - + IStream, + OStream, + OInfo + ) + from git.exc import BadObject from git.typ import str_blob_type @@ -28,220 +28,220 @@ from struct import pack __all__ = ('TestDBBase', 'with_rw_directory', 'with_packs_rw', 'fixture_path') - + class TestDBBase(TestBase): - """Base Class providing default functionality to all tests such as: - - - Utility functions provided by the TestCase base of the unittest method such as:: - self.fail("todo") - self.failUnlessRaises(...) - - - Class level repository which is considered read-only as it is shared among - all test cases in your type. - Access it using:: - self.rorepo # 'ro' stands for read-only - - The rorepo is in fact your current project's git repo. If you refer to specific - shas for your objects, be sure you choose some that are part of the immutable portion - of the project history ( to assure tests don't fail for others ). - - Derived types can override the default repository type to create a different - read-only repo, allowing to test their specific type - """ - - # data - two_lines = "1234\nhello world" - all_data = (two_lines, ) - - #{ Configuration - # The repository type to instantiate. It takes at least a path to operate upon - # during instantiation. - RepoCls = None - - # if True, a read-only repo will be provided and RepoCls must be set. - # Otherwise it may remain unset - needs_ro_repo = True - #} END configuration - - @classmethod - def setUpAll(cls): - """ - Dynamically add a read-only repository to our actual type. This way - each test type has its own repository - """ - if cls.needs_ro_repo: - if cls is not TestDBBase: - assert cls.RepoCls is not None, "RepoCls class member must be set in %s" % cls - cls.rorepo = cls.RepoCls(rorepo_dir()) - #END handle rorepo - - def _assert_object_writing_simple(self, db): - # write a bunch of objects and query their streams and info - null_objs = db.size() - ni = 250 - for i in xrange(ni): - data = pack(">L", i) - istream = IStream(str_blob_type, len(data), StringIO(data)) - new_istream = db.store(istream) - assert new_istream is istream - assert db.has_object(istream.binsha) - - info = db.info(istream.binsha) - assert isinstance(info, OInfo) - assert info.type == istream.type and info.size == istream.size - - stream = db.stream(istream.binsha) - assert isinstance(stream, OStream) - assert stream.binsha == info.binsha and stream.type == info.type - assert stream.read() == data - # END for each item - - assert db.size() == null_objs + ni - shas = list(db.sha_iter()) - assert len(shas) == db.size() - assert len(shas[0]) == 20 - - - def _assert_object_writing(self, db): - """General tests to verify object writing, compatible to ObjectDBW - :note: requires write access to the database""" - # start in 'dry-run' mode, using a simple sha1 writer - ostreams = (ZippedStoreShaWriter, None) - for ostreamcls in ostreams: - for data in self.all_data: - dry_run = ostreamcls is not None - ostream = None - if ostreamcls is not None: - ostream = ostreamcls() - assert isinstance(ostream, Sha1Writer) - # END create ostream - - prev_ostream = db.set_ostream(ostream) - assert type(prev_ostream) in ostreams or prev_ostream in ostreams - - istream = IStream(str_blob_type, len(data), StringIO(data)) - - # store returns same istream instance, with new sha set - my_istream = db.store(istream) - sha = istream.binsha - assert my_istream is istream - assert db.has_object(sha) != dry_run - assert len(sha) == 20 - - # verify data - the slow way, we want to run code - if not dry_run: - info = db.info(sha) - assert str_blob_type == info.type - assert info.size == len(data) - - ostream = db.stream(sha) - assert ostream.read() == data - assert ostream.type == str_blob_type - assert ostream.size == len(data) - else: - self.failUnlessRaises(BadObject, db.info, sha) - self.failUnlessRaises(BadObject, db.stream, sha) - - # DIRECT STREAM COPY - # our data hase been written in object format to the StringIO - # we pasesd as output stream. No physical database representation - # was created. - # Test direct stream copy of object streams, the result must be - # identical to what we fed in - ostream.seek(0) - istream.stream = ostream - assert istream.binsha is not None - prev_sha = istream.binsha - - db.set_ostream(ZippedStoreShaWriter()) - db.store(istream) - assert istream.binsha == prev_sha - new_ostream = db.ostream() - - # note: only works as long our store write uses the same compression - # level, which is zip_best - assert ostream.getvalue() == new_ostream.getvalue() - # END for each data set - # END for each dry_run mode - - def _assert_object_writing_async(self, db): - """Test generic object writing using asynchronous access""" - ni = 5000 - def istream_generator(offset=0, ni=ni): - for data_src in xrange(ni): - data = str(data_src + offset) - yield IStream(str_blob_type, len(data), StringIO(data)) - # END for each item - # END generator utility - - # for now, we are very trusty here as we expect it to work if it worked - # in the single-stream case - - # write objects - reader = IteratorReader(istream_generator()) - istream_reader = db.store_async(reader) - istreams = istream_reader.read() # read all - assert istream_reader.task().error() is None - assert len(istreams) == ni - - for stream in istreams: - assert stream.error is None - assert len(stream.binsha) == 20 - assert isinstance(stream, IStream) - # END assert each stream - - # test has-object-async - we must have all previously added ones - reader = IteratorReader( istream.binsha for istream in istreams ) - hasobject_reader = db.has_object_async(reader) - count = 0 - for sha, has_object in hasobject_reader: - assert has_object - count += 1 - # END for each sha - assert count == ni - - # read the objects we have just written - reader = IteratorReader( istream.binsha for istream in istreams ) - ostream_reader = db.stream_async(reader) - - # read items individually to prevent hitting possible sys-limits - count = 0 - for ostream in ostream_reader: - assert isinstance(ostream, OStream) - count += 1 - # END for each ostream - assert ostream_reader.task().error() is None - assert count == ni - - # get info about our items - reader = IteratorReader( istream.binsha for istream in istreams ) - info_reader = db.info_async(reader) - - count = 0 - for oinfo in info_reader: - assert isinstance(oinfo, OInfo) - count += 1 - # END for each oinfo instance - assert count == ni - - - # combined read-write using a converter - # add 2500 items, and obtain their output streams - nni = 2500 - reader = IteratorReader(istream_generator(offset=ni, ni=nni)) - istream_to_sha = lambda istreams: [ istream.binsha for istream in istreams ] - - istream_reader = db.store_async(reader) - istream_reader.set_post_cb(istream_to_sha) - - ostream_reader = db.stream_async(istream_reader) - - count = 0 - # read it individually, otherwise we might run into the ulimit - for ostream in ostream_reader: - assert isinstance(ostream, OStream) - count += 1 - # END for each ostream - assert count == nni - - + """Base Class providing default functionality to all tests such as: + + - Utility functions provided by the TestCase base of the unittest method such as:: + self.fail("todo") + self.failUnlessRaises(...) + + - Class level repository which is considered read-only as it is shared among + all test cases in your type. + Access it using:: + self.rorepo # 'ro' stands for read-only + + The rorepo is in fact your current project's git repo. If you refer to specific + shas for your objects, be sure you choose some that are part of the immutable portion + of the project history ( to assure tests don't fail for others ). + + Derived types can override the default repository type to create a different + read-only repo, allowing to test their specific type + """ + + # data + two_lines = "1234\nhello world" + all_data = (two_lines, ) + + #{ Configuration + # The repository type to instantiate. It takes at least a path to operate upon + # during instantiation. + RepoCls = None + + # if True, a read-only repo will be provided and RepoCls must be set. + # Otherwise it may remain unset + needs_ro_repo = True + #} END configuration + + @classmethod + def setUpAll(cls): + """ + Dynamically add a read-only repository to our actual type. This way + each test type has its own repository + """ + if cls.needs_ro_repo: + if cls is not TestDBBase: + assert cls.RepoCls is not None, "RepoCls class member must be set in %s" % cls + cls.rorepo = cls.RepoCls(rorepo_dir()) + #END handle rorepo + + def _assert_object_writing_simple(self, db): + # write a bunch of objects and query their streams and info + null_objs = db.size() + ni = 250 + for i in xrange(ni): + data = pack(">L", i) + istream = IStream(str_blob_type, len(data), StringIO(data)) + new_istream = db.store(istream) + assert new_istream is istream + assert db.has_object(istream.binsha) + + info = db.info(istream.binsha) + assert isinstance(info, OInfo) + assert info.type == istream.type and info.size == istream.size + + stream = db.stream(istream.binsha) + assert isinstance(stream, OStream) + assert stream.binsha == info.binsha and stream.type == info.type + assert stream.read() == data + # END for each item + + assert db.size() == null_objs + ni + shas = list(db.sha_iter()) + assert len(shas) == db.size() + assert len(shas[0]) == 20 + + + def _assert_object_writing(self, db): + """General tests to verify object writing, compatible to ObjectDBW + :note: requires write access to the database""" + # start in 'dry-run' mode, using a simple sha1 writer + ostreams = (ZippedStoreShaWriter, None) + for ostreamcls in ostreams: + for data in self.all_data: + dry_run = ostreamcls is not None + ostream = None + if ostreamcls is not None: + ostream = ostreamcls() + assert isinstance(ostream, Sha1Writer) + # END create ostream + + prev_ostream = db.set_ostream(ostream) + assert type(prev_ostream) in ostreams or prev_ostream in ostreams + + istream = IStream(str_blob_type, len(data), StringIO(data)) + + # store returns same istream instance, with new sha set + my_istream = db.store(istream) + sha = istream.binsha + assert my_istream is istream + assert db.has_object(sha) != dry_run + assert len(sha) == 20 + + # verify data - the slow way, we want to run code + if not dry_run: + info = db.info(sha) + assert str_blob_type == info.type + assert info.size == len(data) + + ostream = db.stream(sha) + assert ostream.read() == data + assert ostream.type == str_blob_type + assert ostream.size == len(data) + else: + self.failUnlessRaises(BadObject, db.info, sha) + self.failUnlessRaises(BadObject, db.stream, sha) + + # DIRECT STREAM COPY + # our data hase been written in object format to the StringIO + # we pasesd as output stream. No physical database representation + # was created. + # Test direct stream copy of object streams, the result must be + # identical to what we fed in + ostream.seek(0) + istream.stream = ostream + assert istream.binsha is not None + prev_sha = istream.binsha + + db.set_ostream(ZippedStoreShaWriter()) + db.store(istream) + assert istream.binsha == prev_sha + new_ostream = db.ostream() + + # note: only works as long our store write uses the same compression + # level, which is zip_best + assert ostream.getvalue() == new_ostream.getvalue() + # END for each data set + # END for each dry_run mode + + def _assert_object_writing_async(self, db): + """Test generic object writing using asynchronous access""" + ni = 5000 + def istream_generator(offset=0, ni=ni): + for data_src in xrange(ni): + data = str(data_src + offset) + yield IStream(str_blob_type, len(data), StringIO(data)) + # END for each item + # END generator utility + + # for now, we are very trusty here as we expect it to work if it worked + # in the single-stream case + + # write objects + reader = IteratorReader(istream_generator()) + istream_reader = db.store_async(reader) + istreams = istream_reader.read() # read all + assert istream_reader.task().error() is None + assert len(istreams) == ni + + for stream in istreams: + assert stream.error is None + assert len(stream.binsha) == 20 + assert isinstance(stream, IStream) + # END assert each stream + + # test has-object-async - we must have all previously added ones + reader = IteratorReader( istream.binsha for istream in istreams ) + hasobject_reader = db.has_object_async(reader) + count = 0 + for sha, has_object in hasobject_reader: + assert has_object + count += 1 + # END for each sha + assert count == ni + + # read the objects we have just written + reader = IteratorReader( istream.binsha for istream in istreams ) + ostream_reader = db.stream_async(reader) + + # read items individually to prevent hitting possible sys-limits + count = 0 + for ostream in ostream_reader: + assert isinstance(ostream, OStream) + count += 1 + # END for each ostream + assert ostream_reader.task().error() is None + assert count == ni + + # get info about our items + reader = IteratorReader( istream.binsha for istream in istreams ) + info_reader = db.info_async(reader) + + count = 0 + for oinfo in info_reader: + assert isinstance(oinfo, OInfo) + count += 1 + # END for each oinfo instance + assert count == ni + + + # combined read-write using a converter + # add 2500 items, and obtain their output streams + nni = 2500 + reader = IteratorReader(istream_generator(offset=ni, ni=nni)) + istream_to_sha = lambda istreams: [ istream.binsha for istream in istreams ] + + istream_reader = db.store_async(reader) + istream_reader.set_post_cb(istream_to_sha) + + ostream_reader = db.stream_async(istream_reader) + + count = 0 + # read it individually, otherwise we might run into the ulimit + for ostream in ostream_reader: + assert isinstance(ostream, OStream) + count += 1 + # END for each ostream + assert count == nni + + diff --git a/git/test/db/py/test_base.py b/git/test/db/py/test_base.py index 6b06bbe9..5d076bb2 100644 --- a/git/test/db/py/test_base.py +++ b/git/test/db/py/test_base.py @@ -8,9 +8,9 @@ from git.test.db.base import RepoBase from git.db.complex import PureCompatibilityGitDB class TestPyDBBase(RepoBase): - - RepoCls = PureCompatibilityGitDB - - def test_basics(self): - pass - + + RepoCls = PureCompatibilityGitDB + + def test_basics(self): + pass + diff --git a/git/test/db/py/test_git.py b/git/test/db/py/test_git.py index ecaa5c8f..4f5b5fb5 100644 --- a/git/test/db/py/test_git.py +++ b/git/test/db/py/test_git.py @@ -12,40 +12,40 @@ from git.util import hex_to_bin, bin_to_hex import os class TestGitDB(TestDBBase): - needs_ro_repo = False - - def test_reading(self): - gdb = PureGitODB(os.path.join(rorepo_dir(), 'objects')) - - # we have packs and loose objects, alternates doesn't necessarily exist - assert 1 < len(gdb.databases()) < 4 - - # access should be possible - git_sha = hex_to_bin("5aebcd5cb3340fb31776941d7e4d518a712a8655") - assert isinstance(gdb.info(git_sha), OInfo) - assert isinstance(gdb.stream(git_sha), OStream) - assert gdb.size() > 200 - sha_list = list(gdb.sha_iter()) - assert len(sha_list) == gdb.size() - - - # This is actually a test for compound functionality, but it doesn't - # have a separate test module - # test partial shas - # this one as uneven and quite short - assert gdb.partial_to_complete_sha_hex('5aebcd') == hex_to_bin("5aebcd5cb3340fb31776941d7e4d518a712a8655") - - # mix even/uneven hexshas - for i, binsha in enumerate(sha_list[:50]): - assert gdb.partial_to_complete_sha_hex(bin_to_hex(binsha)[:8-(i%2)]) == binsha - # END for each sha - - self.failUnlessRaises(BadObject, gdb.partial_to_complete_sha_hex, "0000") - - @with_rw_directory - def test_writing(self, path): - gdb = PureGitODB(path) - - # its possible to write objects - self._assert_object_writing(gdb) - self._assert_object_writing_async(gdb) + needs_ro_repo = False + + def test_reading(self): + gdb = PureGitODB(os.path.join(rorepo_dir(), 'objects')) + + # we have packs and loose objects, alternates doesn't necessarily exist + assert 1 < len(gdb.databases()) < 4 + + # access should be possible + git_sha = hex_to_bin("5aebcd5cb3340fb31776941d7e4d518a712a8655") + assert isinstance(gdb.info(git_sha), OInfo) + assert isinstance(gdb.stream(git_sha), OStream) + assert gdb.size() > 200 + sha_list = list(gdb.sha_iter()) + assert len(sha_list) == gdb.size() + + + # This is actually a test for compound functionality, but it doesn't + # have a separate test module + # test partial shas + # this one as uneven and quite short + assert gdb.partial_to_complete_sha_hex('5aebcd') == hex_to_bin("5aebcd5cb3340fb31776941d7e4d518a712a8655") + + # mix even/uneven hexshas + for i, binsha in enumerate(sha_list[:50]): + assert gdb.partial_to_complete_sha_hex(bin_to_hex(binsha)[:8-(i%2)]) == binsha + # END for each sha + + self.failUnlessRaises(BadObject, gdb.partial_to_complete_sha_hex, "0000") + + @with_rw_directory + def test_writing(self, path): + gdb = PureGitODB(path) + + # its possible to write objects + self._assert_object_writing(gdb) + self._assert_object_writing_async(gdb) diff --git a/git/test/db/py/test_loose.py b/git/test/db/py/test_loose.py index 0c9b4831..cfb0ca3a 100644 --- a/git/test/db/py/test_loose.py +++ b/git/test/db/py/test_loose.py @@ -6,31 +6,31 @@ from git.test.db.lib import TestDBBase, with_rw_directory from git.db.py.loose import PureLooseObjectODB from git.exc import BadObject from git.util import bin_to_hex - + class TestLooseDB(TestDBBase): - - needs_ro_repo = False - - @with_rw_directory - def test_basics(self, path): - ldb = PureLooseObjectODB(path) - - # write data - self._assert_object_writing(ldb) - self._assert_object_writing_async(ldb) - - # verify sha iteration and size - shas = list(ldb.sha_iter()) - assert shas and len(shas[0]) == 20 - - assert len(shas) == ldb.size() - - # verify find short object - long_sha = bin_to_hex(shas[-1]) - for short_sha in (long_sha[:20], long_sha[:5]): - assert bin_to_hex(ldb.partial_to_complete_sha_hex(short_sha)) == long_sha - # END for each sha - - self.failUnlessRaises(BadObject, ldb.partial_to_complete_sha_hex, '0000') - # raises if no object could be foudn - + + needs_ro_repo = False + + @with_rw_directory + def test_basics(self, path): + ldb = PureLooseObjectODB(path) + + # write data + self._assert_object_writing(ldb) + self._assert_object_writing_async(ldb) + + # verify sha iteration and size + shas = list(ldb.sha_iter()) + assert shas and len(shas[0]) == 20 + + assert len(shas) == ldb.size() + + # verify find short object + long_sha = bin_to_hex(shas[-1]) + for short_sha in (long_sha[:20], long_sha[:5]): + assert bin_to_hex(ldb.partial_to_complete_sha_hex(short_sha)) == long_sha + # END for each sha + + self.failUnlessRaises(BadObject, ldb.partial_to_complete_sha_hex, '0000') + # raises if no object could be foudn + diff --git a/git/test/db/py/test_mem.py b/git/test/db/py/test_mem.py index bc98dc56..bb879554 100644 --- a/git/test/db/py/test_mem.py +++ b/git/test/db/py/test_mem.py @@ -5,26 +5,26 @@ from git.test.db.lib import TestDBBase, with_rw_directory from git.db.py.mem import PureMemoryDB from git.db.py.loose import PureLooseObjectODB - + class TestPureMemoryDB(TestDBBase): - - needs_ro_repo = False + + needs_ro_repo = False - @with_rw_directory - def test_writing(self, path): - mdb = PureMemoryDB() - - # write data - self._assert_object_writing_simple(mdb) - - # test stream copy - ldb = PureLooseObjectODB(path) - assert ldb.size() == 0 - num_streams_copied = mdb.stream_copy(mdb.sha_iter(), ldb) - assert num_streams_copied == mdb.size() - - assert ldb.size() == mdb.size() - for sha in mdb.sha_iter(): - assert ldb.has_object(sha) - assert ldb.stream(sha).read() == mdb.stream(sha).read() - # END verify objects where copied and are equal + @with_rw_directory + def test_writing(self, path): + mdb = PureMemoryDB() + + # write data + self._assert_object_writing_simple(mdb) + + # test stream copy + ldb = PureLooseObjectODB(path) + assert ldb.size() == 0 + num_streams_copied = mdb.stream_copy(mdb.sha_iter(), ldb) + assert num_streams_copied == mdb.size() + + assert ldb.size() == mdb.size() + for sha in mdb.sha_iter(): + assert ldb.has_object(sha) + assert ldb.stream(sha).read() == mdb.stream(sha).read() + # END verify objects where copied and are equal diff --git a/git/test/db/py/test_pack.py b/git/test/db/py/test_pack.py index 5043f446..54dc2e2c 100644 --- a/git/test/db/py/test_pack.py +++ b/git/test/db/py/test_pack.py @@ -13,64 +13,64 @@ import os import random class TestPackDB(TestDBBase): - - needs_ro_repo = False - - @with_packs_rw - def test_writing(self, path): - pdb = PurePackedODB(path) - - # on demand, we init our pack cache - num_packs = len(pdb.entities()) - assert num_packs - assert pdb._st_mtime != 0 - - # test pack directory changed: - # packs removed - rename a file, should affect the glob - pack_path = pdb.entities()[0].pack().path() - new_pack_path = pack_path + "renamed" - os.rename(pack_path, new_pack_path) - - pdb.update_cache(force=True) - assert len(pdb.entities()) == num_packs - 1 - - # packs added - os.rename(new_pack_path, pack_path) - pdb.update_cache(force=True) - assert len(pdb.entities()) == num_packs - - # bang on the cache - # access the Entities directly, as there is no iteration interface - # yet ( or required for now ) - sha_list = list(pdb.sha_iter()) - assert len(sha_list) == pdb.size() - - # hit all packs in random order - random.shuffle(sha_list) - - for sha in sha_list: - info = pdb.info(sha) - stream = pdb.stream(sha) - # END for each sha to query - - - # test short finding - be a bit more brutal here - max_bytes = 19 - min_bytes = 2 - num_ambiguous = 0 - for i, sha in enumerate(sha_list): - short_sha = sha[:max((i % max_bytes), min_bytes)] - try: - assert pdb.partial_to_complete_sha(short_sha, len(short_sha)*2) == sha - except AmbiguousObjectName: - num_ambiguous += 1 - pass # valid, we can have short objects - # END exception handling - # END for each sha to find - - # we should have at least one ambiguous, considering the small sizes - # but in our pack, there is no ambigious ... - # assert num_ambiguous - - # non-existing - self.failUnlessRaises(BadObject, pdb.partial_to_complete_sha, "\0\0", 4) + + needs_ro_repo = False + + @with_packs_rw + def test_writing(self, path): + pdb = PurePackedODB(path) + + # on demand, we init our pack cache + num_packs = len(pdb.entities()) + assert num_packs + assert pdb._st_mtime != 0 + + # test pack directory changed: + # packs removed - rename a file, should affect the glob + pack_path = pdb.entities()[0].pack().path() + new_pack_path = pack_path + "renamed" + os.rename(pack_path, new_pack_path) + + pdb.update_cache(force=True) + assert len(pdb.entities()) == num_packs - 1 + + # packs added + os.rename(new_pack_path, pack_path) + pdb.update_cache(force=True) + assert len(pdb.entities()) == num_packs + + # bang on the cache + # access the Entities directly, as there is no iteration interface + # yet ( or required for now ) + sha_list = list(pdb.sha_iter()) + assert len(sha_list) == pdb.size() + + # hit all packs in random order + random.shuffle(sha_list) + + for sha in sha_list: + info = pdb.info(sha) + stream = pdb.stream(sha) + # END for each sha to query + + + # test short finding - be a bit more brutal here + max_bytes = 19 + min_bytes = 2 + num_ambiguous = 0 + for i, sha in enumerate(sha_list): + short_sha = sha[:max((i % max_bytes), min_bytes)] + try: + assert pdb.partial_to_complete_sha(short_sha, len(short_sha)*2) == sha + except AmbiguousObjectName: + num_ambiguous += 1 + pass # valid, we can have short objects + # END exception handling + # END for each sha to find + + # we should have at least one ambiguous, considering the small sizes + # but in our pack, there is no ambigious ... + # assert num_ambiguous + + # non-existing + self.failUnlessRaises(BadObject, pdb.partial_to_complete_sha, "\0\0", 4) diff --git a/git/test/db/py/test_ref.py b/git/test/db/py/test_ref.py index c5374dc9..dfaf9644 100644 --- a/git/test/db/py/test_ref.py +++ b/git/test/db/py/test_ref.py @@ -6,57 +6,57 @@ from git.test.db.lib import * from git.db.py.ref import PureReferenceDB from git.util import ( - NULL_BIN_SHA, - hex_to_bin - ) + NULL_BIN_SHA, + hex_to_bin + ) import os - + class TestPureReferenceDB(TestDBBase): - - needs_ro_repo = False - - def make_alt_file(self, alt_path, alt_list): - """Create an alternates file which contains the given alternates. - The list can be empty""" - alt_file = open(alt_path, "wb") - for alt in alt_list: - alt_file.write(alt + "\n") - alt_file.close() - - @with_rw_directory - def test_writing(self, path): - NULL_BIN_SHA = '\0' * 20 - - alt_path = os.path.join(path, 'alternates') - rdb = PureReferenceDB(alt_path) - assert len(rdb.databases()) == 0 - assert rdb.size() == 0 - assert len(list(rdb.sha_iter())) == 0 - - # try empty, non-existing - assert not rdb.has_object(NULL_BIN_SHA) - - - # setup alternate file - # add two, one is invalid - own_repo_path = fixture_path('../../../.git/objects') # use own repo - self.make_alt_file(alt_path, [own_repo_path, "invalid/path"]) - rdb.update_cache() - assert len(rdb.databases()) == 1 - - # we should now find a default revision of ours - git_sha = hex_to_bin("5aebcd5cb3340fb31776941d7e4d518a712a8655") - assert rdb.has_object(git_sha) - - # remove valid - self.make_alt_file(alt_path, ["just/one/invalid/path"]) - rdb.update_cache() - assert len(rdb.databases()) == 0 - - # add valid - self.make_alt_file(alt_path, [own_repo_path]) - rdb.update_cache() - assert len(rdb.databases()) == 1 - - + + needs_ro_repo = False + + def make_alt_file(self, alt_path, alt_list): + """Create an alternates file which contains the given alternates. + The list can be empty""" + alt_file = open(alt_path, "wb") + for alt in alt_list: + alt_file.write(alt + "\n") + alt_file.close() + + @with_rw_directory + def test_writing(self, path): + NULL_BIN_SHA = '\0' * 20 + + alt_path = os.path.join(path, 'alternates') + rdb = PureReferenceDB(alt_path) + assert len(rdb.databases()) == 0 + assert rdb.size() == 0 + assert len(list(rdb.sha_iter())) == 0 + + # try empty, non-existing + assert not rdb.has_object(NULL_BIN_SHA) + + + # setup alternate file + # add two, one is invalid + own_repo_path = fixture_path('../../../.git/objects') # use own repo + self.make_alt_file(alt_path, [own_repo_path, "invalid/path"]) + rdb.update_cache() + assert len(rdb.databases()) == 1 + + # we should now find a default revision of ours + git_sha = hex_to_bin("5aebcd5cb3340fb31776941d7e4d518a712a8655") + assert rdb.has_object(git_sha) + + # remove valid + self.make_alt_file(alt_path, ["just/one/invalid/path"]) + rdb.update_cache() + assert len(rdb.databases()) == 0 + + # add valid + self.make_alt_file(alt_path, [own_repo_path]) + rdb.update_cache() + assert len(rdb.databases()) == 1 + + diff --git a/git/test/db/pygit2/lib.py b/git/test/db/pygit2/lib.py index 356df9dc..fab762e7 100644 --- a/git/test/db/pygit2/lib.py +++ b/git/test/db/pygit2/lib.py @@ -1,23 +1,23 @@ """pygit2 specific utilities, as well as all the default ones""" from git.test.lib import ( - InheritedTestMethodsOverrideWrapperMetaClsAutoMixin, - needs_module_or_skip - ) + InheritedTestMethodsOverrideWrapperMetaClsAutoMixin, + needs_module_or_skip + ) __all__ = ['needs_pygit2_or_skip', 'Pygit2RequiredMetaMixin'] #{ Decoorators def needs_pygit2_or_skip(func): - """Skip this test if we have no pygit2 - print warning""" - return needs_module_or_skip('pygit2')(func) + """Skip this test if we have no pygit2 - print warning""" + return needs_module_or_skip('pygit2')(func) #}END decorators #{ MetaClasses class Pygit2RequiredMetaMixin(InheritedTestMethodsOverrideWrapperMetaClsAutoMixin): - decorator = [needs_pygit2_or_skip] + decorator = [needs_pygit2_or_skip] #} END metaclasses diff --git a/git/test/db/pygit2/test_base.py b/git/test/db/pygit2/test_base.py index 246a1643..52ee24f5 100644 --- a/git/test/db/pygit2/test_base.py +++ b/git/test/db/pygit2/test_base.py @@ -9,24 +9,24 @@ from git.test.db.base import RepoBase try: - import pygit2 + import pygit2 except ImportError: - # om this case, all other pygit2 tests will be skipped - # Need to properly initialize the class though, otherwise it would fail - from git.db.complex import PureCompatibilityGitDB as Pygit2DB + # om this case, all other pygit2 tests will be skipped + # Need to properly initialize the class though, otherwise it would fail + from git.db.complex import PureCompatibilityGitDB as Pygit2DB else: - # now we know pygit2 is available, to do futher imports - from git.db.pygit2.complex import Pygit2CompatibilityGitDB as Pygit2DB - + # now we know pygit2 is available, to do futher imports + from git.db.pygit2.complex import Pygit2CompatibilityGitDB as Pygit2DB + #END handle imports class TestPyGit2DBBase(RepoBase): - __metaclass__ = Pygit2RequiredMetaMixin - RepoCls = Pygit2DB - - @needs_pygit2_or_skip - @with_rw_repo('HEAD', bare=False) - def test_basics(self, rw_repo): - db = Pygit2DB(rw_repo.working_tree_dir) - - + __metaclass__ = Pygit2RequiredMetaMixin + RepoCls = Pygit2DB + + @needs_pygit2_or_skip + @with_rw_repo('HEAD', bare=False) + def test_basics(self, rw_repo): + db = Pygit2DB(rw_repo.working_tree_dir) + + diff --git a/git/test/db/test_base.py b/git/test/db/test_base.py index 2a882d0a..78da9f04 100644 --- a/git/test/db/test_base.py +++ b/git/test/db/test_base.py @@ -7,14 +7,14 @@ from git.db import RefSpec class TestBase(TestDBBase): - needs_ro_repo = False + needs_ro_repo = False - @with_rw_directory - def test_basics(self, path): - self.failUnlessRaises(ValueError, RefSpec, None, None) - rs = RefSpec(None, "something") - assert rs.force == False - assert rs.delete_destination() - assert rs.source is None - assert rs.destination == "something" - + @with_rw_directory + def test_basics(self, path): + self.failUnlessRaises(ValueError, RefSpec, None, None) + rs = RefSpec(None, "something") + assert rs.force == False + assert rs.delete_destination() + assert rs.source is None + assert rs.destination == "something" + diff --git a/git/test/lib/base.py b/git/test/lib/base.py index bc160783..298e8e05 100644 --- a/git/test/lib/base.py +++ b/git/test/lib/base.py @@ -5,14 +5,14 @@ """Utilities used in ODB testing""" from git.base import OStream from git.stream import ( - Sha1Writer, - ZippedStoreShaWriter - ) + Sha1Writer, + ZippedStoreShaWriter + ) from git.util import ( - zlib, - dirname - ) + zlib, + dirname + ) import sys import random @@ -30,171 +30,171 @@ import gc #{ Decorators def with_rw_directory(func): - """Create a temporary directory which can be written to, remove it if the - test suceeds, but leave it otherwise to aid additional debugging""" - def wrapper(self): - path = maketemp(prefix=func.__name__) - os.mkdir(path) - keep = False - try: - try: - return func(self, path) - except Exception: - print >> sys.stderr, "Test %s.%s failed, output is at %r" % (type(self).__name__, func.__name__, path) - keep = True - raise - finally: - # Need to collect here to be sure all handles have been closed. It appears - # a windows-only issue. In fact things should be deleted, as well as - # memory maps closed, once objects go out of scope. For some reason - # though this is not the case here unless we collect explicitly. - if not keep: - gc.collect() - shutil.rmtree(path) - # END handle exception - # END wrapper - - wrapper.__name__ = func.__name__ - return wrapper + """Create a temporary directory which can be written to, remove it if the + test suceeds, but leave it otherwise to aid additional debugging""" + def wrapper(self): + path = maketemp(prefix=func.__name__) + os.mkdir(path) + keep = False + try: + try: + return func(self, path) + except Exception: + print >> sys.stderr, "Test %s.%s failed, output is at %r" % (type(self).__name__, func.__name__, path) + keep = True + raise + finally: + # Need to collect here to be sure all handles have been closed. It appears + # a windows-only issue. In fact things should be deleted, as well as + # memory maps closed, once objects go out of scope. For some reason + # though this is not the case here unless we collect explicitly. + if not keep: + gc.collect() + shutil.rmtree(path) + # END handle exception + # END wrapper + + wrapper.__name__ = func.__name__ + return wrapper def with_rw_repo(func): - """Create a copy of our repository and put it into a writable location. It will - be removed if the test doesn't result in an error. - As we can currently only copy the fully working tree, tests must not rely on - being on a certain branch or on anything really except for the default tags - that should exist - Wrapped function obtains a git repository """ - def wrapper(self, path): - src_dir = dirname(dirname(dirname(__file__))) - assert(os.path.isdir(path)) - os.rmdir(path) # created by wrapper, but must not exist for copy operation - shutil.copytree(src_dir, path) - target_gitdir = os.path.join(path, '.git') - assert os.path.isdir(target_gitdir) - return func(self, self.RepoCls(target_gitdir)) - #END wrapper - wrapper.__name__ = func.__name__ - return with_rw_directory(wrapper) - + """Create a copy of our repository and put it into a writable location. It will + be removed if the test doesn't result in an error. + As we can currently only copy the fully working tree, tests must not rely on + being on a certain branch or on anything really except for the default tags + that should exist + Wrapped function obtains a git repository """ + def wrapper(self, path): + src_dir = dirname(dirname(dirname(__file__))) + assert(os.path.isdir(path)) + os.rmdir(path) # created by wrapper, but must not exist for copy operation + shutil.copytree(src_dir, path) + target_gitdir = os.path.join(path, '.git') + assert os.path.isdir(target_gitdir) + return func(self, self.RepoCls(target_gitdir)) + #END wrapper + wrapper.__name__ = func.__name__ + return with_rw_directory(wrapper) + def with_packs_rw(func): - """Function that provides a path into which the packs for testing should be - copied. Will pass on the path to the actual function afterwards - - :note: needs with_rw_directory wrapped around it""" - def wrapper(self, path): - src_pack_glob = fixture_path('packs/*') - print src_pack_glob - copy_files_globbed(src_pack_glob, path, hard_link_ok=True) - return func(self, path) - # END wrapper - - wrapper.__name__ = func.__name__ - return with_rw_directory(wrapper) + """Function that provides a path into which the packs for testing should be + copied. Will pass on the path to the actual function afterwards + + :note: needs with_rw_directory wrapped around it""" + def wrapper(self, path): + src_pack_glob = fixture_path('packs/*') + print src_pack_glob + copy_files_globbed(src_pack_glob, path, hard_link_ok=True) + return func(self, path) + # END wrapper + + wrapper.__name__ = func.__name__ + return with_rw_directory(wrapper) #} END decorators #{ Routines def rorepo_dir(): - """:return: path to our own repository, being our own .git directory. - :note: doesn't work in bare repositories""" - base = os.path.join(dirname(dirname(dirname(dirname(__file__)))), '.git') - assert os.path.isdir(base) - return base + """:return: path to our own repository, being our own .git directory. + :note: doesn't work in bare repositories""" + base = os.path.join(dirname(dirname(dirname(dirname(__file__)))), '.git') + assert os.path.isdir(base) + return base def maketemp(*args, **kwargs): - """Wrapper around default tempfile.mktemp to fix an osx issue""" - tdir = tempfile.mktemp(*args, **kwargs) - if sys.platform == 'darwin': - tdir = '/private' + tdir - return tdir + """Wrapper around default tempfile.mktemp to fix an osx issue""" + tdir = tempfile.mktemp(*args, **kwargs) + if sys.platform == 'darwin': + tdir = '/private' + tdir + return tdir def fixture_path(relapath=''): - """:return: absolute path into the fixture directory - :param relapath: relative path into the fixtures directory, or '' - to obtain the fixture directory itself""" - test_dir = os.path.dirname(os.path.dirname(__file__)) - return os.path.join(test_dir, "fixtures", relapath) - + """:return: absolute path into the fixture directory + :param relapath: relative path into the fixtures directory, or '' + to obtain the fixture directory itself""" + test_dir = os.path.dirname(os.path.dirname(__file__)) + return os.path.join(test_dir, "fixtures", relapath) + def fixture(name): - return open(fixture_path(name), 'rb').read() + return open(fixture_path(name), 'rb').read() def absolute_project_path(): - return os.path.abspath(os.path.join(os.path.dirname(__file__), "..", "..")) + return os.path.abspath(os.path.join(os.path.dirname(__file__), "..", "..")) def copy_files_globbed(source_glob, target_dir, hard_link_ok=False): - """Copy all files found according to the given source glob into the target directory - :param hard_link_ok: if True, hard links will be created if possible. Otherwise - the files will be copied""" - for src_file in glob.glob(source_glob): - if hard_link_ok and hasattr(os, 'link'): - target = os.path.join(target_dir, os.path.basename(src_file)) - try: - os.link(src_file, target) - except OSError: - shutil.copy(src_file, target_dir) - # END handle cross device links ( and resulting failure ) - else: - shutil.copy(src_file, target_dir) - # END try hard link - # END for each file to copy - + """Copy all files found according to the given source glob into the target directory + :param hard_link_ok: if True, hard links will be created if possible. Otherwise + the files will be copied""" + for src_file in glob.glob(source_glob): + if hard_link_ok and hasattr(os, 'link'): + target = os.path.join(target_dir, os.path.basename(src_file)) + try: + os.link(src_file, target) + except OSError: + shutil.copy(src_file, target_dir) + # END handle cross device links ( and resulting failure ) + else: + shutil.copy(src_file, target_dir) + # END try hard link + # END for each file to copy + def make_bytes(size_in_bytes, randomize=False): - """:return: string with given size in bytes - :param randomize: try to produce a very random stream""" - actual_size = size_in_bytes / 4 - producer = xrange(actual_size) - if randomize: - producer = list(producer) - random.shuffle(producer) - # END randomize - a = array('i', producer) - return a.tostring() + """:return: string with given size in bytes + :param randomize: try to produce a very random stream""" + actual_size = size_in_bytes / 4 + producer = xrange(actual_size) + if randomize: + producer = list(producer) + random.shuffle(producer) + # END randomize + a = array('i', producer) + return a.tostring() def make_object(type, data): - """:return: bytes resembling an uncompressed object""" - odata = "blob %i\0" % len(data) - return odata + data - + """:return: bytes resembling an uncompressed object""" + odata = "blob %i\0" % len(data) + return odata + data + def make_memory_file(size_in_bytes, randomize=False): - """:return: tuple(size_of_stream, stream) - :param randomize: try to produce a very random stream""" - d = make_bytes(size_in_bytes, randomize) - return len(d), StringIO(d) + """:return: tuple(size_of_stream, stream) + :param randomize: try to produce a very random stream""" + d = make_bytes(size_in_bytes, randomize) + return len(d), StringIO(d) #} END routines #{ Stream Utilities class DummyStream(object): - def __init__(self): - self.was_read = False - self.bytes = 0 - self.closed = False - - def read(self, size): - self.was_read = True - self.bytes = size - - def close(self): - self.closed = True - - def _assert(self): - assert self.was_read + def __init__(self): + self.was_read = False + self.bytes = 0 + self.closed = False + + def read(self, size): + self.was_read = True + self.bytes = size + + def close(self): + self.closed = True + + def _assert(self): + assert self.was_read class DeriveTest(OStream): - def __init__(self, sha, type, size, stream, *args, **kwargs): - self.myarg = kwargs.pop('myarg') - self.args = args - - def _assert(self): - assert self.args - assert self.myarg + def __init__(self, sha, type, size, stream, *args, **kwargs): + self.myarg = kwargs.pop('myarg') + self.args = args + + def _assert(self): + assert self.args + assert self.myarg #} END stream utilitiess diff --git a/git/test/lib/helper.py b/git/test/lib/helper.py index 2045f9d3..3d7da451 100644 --- a/git/test/lib/helper.py +++ b/git/test/lib/helper.py @@ -16,295 +16,295 @@ import warnings from nose import SkipTest from base import ( - maketemp, - rorepo_dir - ) + maketemp, + rorepo_dir + ) __all__ = ( - 'StringProcessAdapter', 'GlobalsItemDeletorMetaCls', 'InheritedTestMethodsOverrideWrapperMetaClsAutoMixin', - 'with_rw_repo', 'with_rw_and_rw_remote_repo', 'TestBase', 'TestCase', 'needs_module_or_skip' - ) + 'StringProcessAdapter', 'GlobalsItemDeletorMetaCls', 'InheritedTestMethodsOverrideWrapperMetaClsAutoMixin', + 'with_rw_repo', 'with_rw_and_rw_remote_repo', 'TestBase', 'TestCase', 'needs_module_or_skip' + ) - + #{ Adapters - + class StringProcessAdapter(object): - """Allows to use strings as Process object as returned by SubProcess.Popen. - Its tailored to work with the test system only""" - - def __init__(self, input_string): - self.stdout = cStringIO.StringIO(input_string) - self.stderr = cStringIO.StringIO() - - def wait(self): - return 0 - - poll = wait - + """Allows to use strings as Process object as returned by SubProcess.Popen. + Its tailored to work with the test system only""" + + def __init__(self, input_string): + self.stdout = cStringIO.StringIO(input_string) + self.stderr = cStringIO.StringIO() + + def wait(self): + return 0 + + poll = wait + #} END adapters #{ Decorators def _rmtree_onerror(osremove, fullpath, exec_info): - """ - Handle the case on windows that read-only files cannot be deleted by - os.remove by setting it to mode 777, then retry deletion. - """ - if os.name != 'nt' or osremove is not os.remove: - raise - - os.chmod(fullpath, 0777) - os.remove(fullpath) + """ + Handle the case on windows that read-only files cannot be deleted by + os.remove by setting it to mode 777, then retry deletion. + """ + if os.name != 'nt' or osremove is not os.remove: + raise + + os.chmod(fullpath, 0777) + os.remove(fullpath) def with_rw_repo(working_tree_ref, bare=False): - """ - Same as with_bare_repo, but clones the rorepo as non-bare repository, checking - out the working tree at the given working_tree_ref. - - This repository type is more costly due to the working copy checkout. - - To make working with relative paths easier, the cwd will be set to the working - dir of the repository. - """ - assert isinstance(working_tree_ref, basestring), "Decorator requires ref name for working tree checkout" - def argument_passer(func): - def repo_creator(self): - prefix = 'non_' - if bare: - prefix = '' - #END handle prefix - repo_dir = maketemp("%sbare_%s" % (prefix, func.__name__)) - rw_repo = self.rorepo.clone(repo_dir, shared=True, bare=bare, n=True) - - rw_repo.head.commit = rw_repo.commit(working_tree_ref) - if not bare: - rw_repo.head.reference.checkout() - # END handle checkout - - prev_cwd = os.getcwd() - os.chdir(rw_repo.working_dir) - try: - try: - return func(self, rw_repo) - except: - print >> sys.stderr, "Keeping repo after failure: %s" % repo_dir - repo_dir = None - raise - finally: - os.chdir(prev_cwd) - rw_repo.git.clear_cache() - if repo_dir is not None: - shutil.rmtree(repo_dir, onerror=_rmtree_onerror) - # END rm test repo if possible - # END cleanup - # END rw repo creator - repo_creator.__name__ = func.__name__ - return repo_creator - # END argument passer - return argument_passer - + """ + Same as with_bare_repo, but clones the rorepo as non-bare repository, checking + out the working tree at the given working_tree_ref. + + This repository type is more costly due to the working copy checkout. + + To make working with relative paths easier, the cwd will be set to the working + dir of the repository. + """ + assert isinstance(working_tree_ref, basestring), "Decorator requires ref name for working tree checkout" + def argument_passer(func): + def repo_creator(self): + prefix = 'non_' + if bare: + prefix = '' + #END handle prefix + repo_dir = maketemp("%sbare_%s" % (prefix, func.__name__)) + rw_repo = self.rorepo.clone(repo_dir, shared=True, bare=bare, n=True) + + rw_repo.head.commit = rw_repo.commit(working_tree_ref) + if not bare: + rw_repo.head.reference.checkout() + # END handle checkout + + prev_cwd = os.getcwd() + os.chdir(rw_repo.working_dir) + try: + try: + return func(self, rw_repo) + except: + print >> sys.stderr, "Keeping repo after failure: %s" % repo_dir + repo_dir = None + raise + finally: + os.chdir(prev_cwd) + rw_repo.git.clear_cache() + if repo_dir is not None: + shutil.rmtree(repo_dir, onerror=_rmtree_onerror) + # END rm test repo if possible + # END cleanup + # END rw repo creator + repo_creator.__name__ = func.__name__ + return repo_creator + # END argument passer + return argument_passer + def with_rw_and_rw_remote_repo(working_tree_ref): - """ - Same as with_rw_repo, but also provides a writable remote repository from which the - rw_repo has been forked as well as a handle for a git-daemon that may be started to - run the remote_repo. - The remote repository was cloned as bare repository from the rorepo, wheras - the rw repo has a working tree and was cloned from the remote repository. - - remote_repo has two remotes: origin and daemon_origin. One uses a local url, - the other uses a server url. The daemon setup must be done on system level - and should be an inetd service that serves tempdir.gettempdir() and all - directories in it. - - The following scetch demonstrates this:: - rorepo ---<bare clone>---> rw_remote_repo ---<clone>---> rw_repo - - The test case needs to support the following signature:: - def case(self, rw_repo, rw_remote_repo) - - This setup allows you to test push and pull scenarios and hooks nicely. - - See working dir info in with_rw_repo - """ - assert isinstance(working_tree_ref, basestring), "Decorator requires ref name for working tree checkout" - def argument_passer(func): - def remote_repo_creator(self): - remote_repo_dir = maketemp("remote_repo_%s" % func.__name__) - repo_dir = maketemp("remote_clone_non_bare_repo") - - rw_remote_repo = self.rorepo.clone(remote_repo_dir, shared=True, bare=True) - rw_repo = rw_remote_repo.clone(repo_dir, shared=True, bare=False, n=True) # recursive alternates info ? - rw_repo.head.commit = working_tree_ref - rw_repo.head.reference.checkout() - - # prepare for git-daemon - rw_remote_repo.daemon_export = True - - # this thing is just annoying ! - crw = rw_remote_repo.config_writer() - section = "daemon" - try: - crw.add_section(section) - except Exception: - pass - crw.set(section, "receivepack", True) - # release lock - del(crw) - - # initialize the remote - first do it as local remote and pull, then - # we change the url to point to the daemon. The daemon should be started - # by the user, not by us - d_remote = Remote.create(rw_repo, "daemon_origin", remote_repo_dir) - d_remote.fetch() - remote_repo_url = "git://localhost%s" % remote_repo_dir - - d_remote.config_writer.set('url', remote_repo_url) - - # try to list remotes to diagnoes whether the server is up - try: - rw_repo.git.ls_remote(d_remote) - except GitCommandError,e: - print str(e) - if os.name == 'nt': - raise AssertionError('git-daemon needs to run this test, but windows does not have one. Otherwise, run: git-daemon "%s"' % os.path.dirname(_mktemp())) - else: - raise AssertionError('Please start a git-daemon to run this test, execute: git-daemon "%s"' % os.path.dirname(_mktemp())) - # END make assertion - #END catch ls remote error - - # adjust working dir - prev_cwd = os.getcwd() - os.chdir(rw_repo.working_dir) - try: - return func(self, rw_repo, rw_remote_repo) - finally: - os.chdir(prev_cwd) - rw_repo.git.clear_cache() - rw_remote_repo.git.clear_cache() - shutil.rmtree(repo_dir, onerror=_rmtree_onerror) - shutil.rmtree(remote_repo_dir, onerror=_rmtree_onerror) - # END cleanup - # END bare repo creator - remote_repo_creator.__name__ = func.__name__ - return remote_repo_creator - # END remote repo creator - # END argument parsser - - return argument_passer - + """ + Same as with_rw_repo, but also provides a writable remote repository from which the + rw_repo has been forked as well as a handle for a git-daemon that may be started to + run the remote_repo. + The remote repository was cloned as bare repository from the rorepo, wheras + the rw repo has a working tree and was cloned from the remote repository. + + remote_repo has two remotes: origin and daemon_origin. One uses a local url, + the other uses a server url. The daemon setup must be done on system level + and should be an inetd service that serves tempdir.gettempdir() and all + directories in it. + + The following scetch demonstrates this:: + rorepo ---<bare clone>---> rw_remote_repo ---<clone>---> rw_repo + + The test case needs to support the following signature:: + def case(self, rw_repo, rw_remote_repo) + + This setup allows you to test push and pull scenarios and hooks nicely. + + See working dir info in with_rw_repo + """ + assert isinstance(working_tree_ref, basestring), "Decorator requires ref name for working tree checkout" + def argument_passer(func): + def remote_repo_creator(self): + remote_repo_dir = maketemp("remote_repo_%s" % func.__name__) + repo_dir = maketemp("remote_clone_non_bare_repo") + + rw_remote_repo = self.rorepo.clone(remote_repo_dir, shared=True, bare=True) + rw_repo = rw_remote_repo.clone(repo_dir, shared=True, bare=False, n=True) # recursive alternates info ? + rw_repo.head.commit = working_tree_ref + rw_repo.head.reference.checkout() + + # prepare for git-daemon + rw_remote_repo.daemon_export = True + + # this thing is just annoying ! + crw = rw_remote_repo.config_writer() + section = "daemon" + try: + crw.add_section(section) + except Exception: + pass + crw.set(section, "receivepack", True) + # release lock + del(crw) + + # initialize the remote - first do it as local remote and pull, then + # we change the url to point to the daemon. The daemon should be started + # by the user, not by us + d_remote = Remote.create(rw_repo, "daemon_origin", remote_repo_dir) + d_remote.fetch() + remote_repo_url = "git://localhost%s" % remote_repo_dir + + d_remote.config_writer.set('url', remote_repo_url) + + # try to list remotes to diagnoes whether the server is up + try: + rw_repo.git.ls_remote(d_remote) + except GitCommandError,e: + print str(e) + if os.name == 'nt': + raise AssertionError('git-daemon needs to run this test, but windows does not have one. Otherwise, run: git-daemon "%s"' % os.path.dirname(_mktemp())) + else: + raise AssertionError('Please start a git-daemon to run this test, execute: git-daemon "%s"' % os.path.dirname(_mktemp())) + # END make assertion + #END catch ls remote error + + # adjust working dir + prev_cwd = os.getcwd() + os.chdir(rw_repo.working_dir) + try: + return func(self, rw_repo, rw_remote_repo) + finally: + os.chdir(prev_cwd) + rw_repo.git.clear_cache() + rw_remote_repo.git.clear_cache() + shutil.rmtree(repo_dir, onerror=_rmtree_onerror) + shutil.rmtree(remote_repo_dir, onerror=_rmtree_onerror) + # END cleanup + # END bare repo creator + remote_repo_creator.__name__ = func.__name__ + return remote_repo_creator + # END remote repo creator + # END argument parsser + + return argument_passer + def needs_module_or_skip(module): - """Decorator to be used for test cases only. - Print a warning if the given module could not be imported, and skip the test. - Otherwise run the test as usual - :param module: the name of the module to skip""" - def argpasser(func): - def wrapper(self, *args, **kwargs): - try: - __import__(module) - except ImportError: - msg = "Module %r is required to run this test - skipping" % module - warnings.warn(msg) - raise SkipTest(msg) - #END check import - return func(self, *args, **kwargs) - #END wrapper - wrapper.__name__ = func.__name__ - return wrapper - #END argpasser - return argpasser - + """Decorator to be used for test cases only. + Print a warning if the given module could not be imported, and skip the test. + Otherwise run the test as usual + :param module: the name of the module to skip""" + def argpasser(func): + def wrapper(self, *args, **kwargs): + try: + __import__(module) + except ImportError: + msg = "Module %r is required to run this test - skipping" % module + warnings.warn(msg) + raise SkipTest(msg) + #END check import + return func(self, *args, **kwargs) + #END wrapper + wrapper.__name__ = func.__name__ + return wrapper + #END argpasser + return argpasser + #} END decorators #{ Meta Classes class GlobalsItemDeletorMetaCls(type): - """Utiltiy to prevent the RepoBase to be picked up by nose as the metacls - will delete the instance from the globals""" - #{ Configuration - # Set this to a string name of the module to delete - ModuleToDelete = None - #} END configuration - - def __new__(metacls, name, bases, clsdict): - assert metacls.ModuleToDelete is not None, "Invalid metaclass configuration" - new_type = super(GlobalsItemDeletorMetaCls, metacls).__new__(metacls, name, bases, clsdict) - if name != metacls.ModuleToDelete: - mod = __import__(new_type.__module__, globals(), locals(), new_type.__module__) - try: - delattr(mod, metacls.ModuleToDelete) - except AttributeError: - pass - #END skip case that people import our base without actually using it - #END handle deletion - return new_type - - + """Utiltiy to prevent the RepoBase to be picked up by nose as the metacls + will delete the instance from the globals""" + #{ Configuration + # Set this to a string name of the module to delete + ModuleToDelete = None + #} END configuration + + def __new__(metacls, name, bases, clsdict): + assert metacls.ModuleToDelete is not None, "Invalid metaclass configuration" + new_type = super(GlobalsItemDeletorMetaCls, metacls).__new__(metacls, name, bases, clsdict) + if name != metacls.ModuleToDelete: + mod = __import__(new_type.__module__, globals(), locals(), new_type.__module__) + try: + delattr(mod, metacls.ModuleToDelete) + except AttributeError: + pass + #END skip case that people import our base without actually using it + #END handle deletion + return new_type + + class InheritedTestMethodsOverrideWrapperMetaClsAutoMixin(object): - """Automatically picks up the actual metaclass of the the type to be created, - that is the one inherited by one of the bases, and patch up its __new__ to use - the InheritedTestMethodsOverrideWrapperInstanceDecorator with our configured decorator""" - - #{ Configuration - # decorator function to use when wrapping the inherited methods. Put it into a list as first member - # to hide it from being created as class method - decorator = [] - #}END configuration - - @classmethod - def _find_metacls(metacls, bases): - """emulate pythons lookup""" - mcls_attr = '__metaclass__' - for base in bases: - if hasattr(base, mcls_attr): - return getattr(base, mcls_attr) - return metacls._find_metacls(base.__bases__) - #END for each base - raise AssertionError("base class had not metaclass attached") - - @classmethod - def _patch_methods_recursive(metacls, bases, clsdict): - """depth-first patching of methods""" - for base in bases: - metacls._patch_methods_recursive(base.__bases__, clsdict) - for name, item in base.__dict__.iteritems(): - if not name.startswith('test_'): - continue - #END skip non-tests - clsdict[name] = metacls.decorator[0](item) - #END for each item - #END for each base - - def __new__(metacls, name, bases, clsdict): - assert metacls.decorator, "'decorator' member needs to be set in subclass" - base_metacls = metacls._find_metacls(bases) - metacls._patch_methods_recursive(bases, clsdict) - return base_metacls.__new__(base_metacls, name, bases, clsdict) - + """Automatically picks up the actual metaclass of the the type to be created, + that is the one inherited by one of the bases, and patch up its __new__ to use + the InheritedTestMethodsOverrideWrapperInstanceDecorator with our configured decorator""" + + #{ Configuration + # decorator function to use when wrapping the inherited methods. Put it into a list as first member + # to hide it from being created as class method + decorator = [] + #}END configuration + + @classmethod + def _find_metacls(metacls, bases): + """emulate pythons lookup""" + mcls_attr = '__metaclass__' + for base in bases: + if hasattr(base, mcls_attr): + return getattr(base, mcls_attr) + return metacls._find_metacls(base.__bases__) + #END for each base + raise AssertionError("base class had not metaclass attached") + + @classmethod + def _patch_methods_recursive(metacls, bases, clsdict): + """depth-first patching of methods""" + for base in bases: + metacls._patch_methods_recursive(base.__bases__, clsdict) + for name, item in base.__dict__.iteritems(): + if not name.startswith('test_'): + continue + #END skip non-tests + clsdict[name] = metacls.decorator[0](item) + #END for each item + #END for each base + + def __new__(metacls, name, bases, clsdict): + assert metacls.decorator, "'decorator' member needs to be set in subclass" + base_metacls = metacls._find_metacls(bases) + metacls._patch_methods_recursive(bases, clsdict) + return base_metacls.__new__(base_metacls, name, bases, clsdict) + #} END meta classes - + class TestBase(TestCase): - """ - Base Class providing default functionality to all tests such as: - - Utility functions provided by the TestCase base of the unittest method such as:: - self.fail("todo") - self.failUnlessRaises(...) - """ - - @classmethod - def setUpAll(cls): - """This method is only called to provide the most basic functionality - Subclasses may just override it or implement it differently""" - cls.rorepo = Repo(rorepo_dir()) - - def _make_file(self, rela_path, data, repo=None): - """ - Create a file at the given path relative to our repository, filled - with the given data. Returns absolute path to created file. - """ - repo = repo or self.rorepo - abs_path = os.path.join(repo.working_tree_dir, rela_path) - fp = open(abs_path, "w") - fp.write(data) - fp.close() - return abs_path + """ + Base Class providing default functionality to all tests such as: + - Utility functions provided by the TestCase base of the unittest method such as:: + self.fail("todo") + self.failUnlessRaises(...) + """ + + @classmethod + def setUpAll(cls): + """This method is only called to provide the most basic functionality + Subclasses may just override it or implement it differently""" + cls.rorepo = Repo(rorepo_dir()) + + def _make_file(self, rela_path, data, repo=None): + """ + Create a file at the given path relative to our repository, filled + with the given data. Returns absolute path to created file. + """ + repo = repo or self.rorepo + abs_path = os.path.join(repo.working_tree_dir, rela_path) + fp = open(abs_path, "w") + fp.write(data) + fp.close() + return abs_path diff --git a/git/test/objects/lib.py b/git/test/objects/lib.py index fe1d9f9d..e3860ba5 100644 --- a/git/test/objects/lib.py +++ b/git/test/objects/lib.py @@ -1,14 +1,14 @@ """Provide customized obhject testing facilities""" from git.test.lib import ( - rorepo_dir, - TestBase, - assert_equal, - assert_not_equal, - with_rw_repo, - StringProcessAdapter, - ) + rorepo_dir, + TestBase, + assert_equal, + assert_not_equal, + with_rw_repo, + StringProcessAdapter, + ) class TestObjectBase(TestBase): - """Provides a default read-only repository in the rorepo member""" - pass + """Provides a default read-only repository in the rorepo member""" + pass diff --git a/git/test/objects/test_blob.py b/git/test/objects/test_blob.py index 58ac25b7..978ab931 100644 --- a/git/test/objects/test_blob.py +++ b/git/test/objects/test_blob.py @@ -9,15 +9,15 @@ from git.objects.blob import * from git.util import hex_to_bin class TestBlob(TestObjectBase): - - def test_mime_type_should_return_mime_type_for_known_types(self): - blob = Blob(self.rorepo, **{'binsha': Blob.NULL_BIN_SHA, 'path': 'foo.png'}) - assert_equal("image/png", blob.mime_type) + + def test_mime_type_should_return_mime_type_for_known_types(self): + blob = Blob(self.rorepo, **{'binsha': Blob.NULL_BIN_SHA, 'path': 'foo.png'}) + assert_equal("image/png", blob.mime_type) - def test_mime_type_should_return_text_plain_for_unknown_types(self): - blob = Blob(self.rorepo, **{'binsha': Blob.NULL_BIN_SHA,'path': 'something'}) - assert_equal("text/plain", blob.mime_type) + def test_mime_type_should_return_text_plain_for_unknown_types(self): + blob = Blob(self.rorepo, **{'binsha': Blob.NULL_BIN_SHA,'path': 'something'}) + assert_equal("text/plain", blob.mime_type) - def test_nodict(self): - self.failUnlessRaises(AttributeError, setattr, self.rorepo.tree()['AUTHORS'], 'someattr', 2) - + def test_nodict(self): + self.failUnlessRaises(AttributeError, setattr, self.rorepo.tree()['AUTHORS'], 'someattr', 2) + diff --git a/git/test/objects/test_commit.py b/git/test/objects/test_commit.py index 80326fe9..1b8b69c7 100644 --- a/git/test/objects/test_commit.py +++ b/git/test/objects/test_commit.py @@ -10,9 +10,9 @@ from git.objects.commit import * from git.base import IStream from git.util import ( - hex_to_bin, - Actor, - ) + hex_to_bin, + Actor, + ) from cStringIO import StringIO import time @@ -20,260 +20,260 @@ import sys def assert_commit_serialization(rwrepo, commit_id, print_performance_info=False): - """traverse all commits in the history of commit identified by commit_id and check - if the serialization works. - :param print_performance_info: if True, we will show how fast we are""" - ns = 0 # num serializations - nds = 0 # num deserializations - - st = time.time() - for cm in rwrepo.commit(commit_id).traverse(): - nds += 1 - - # assert that we deserialize commits correctly, hence we get the same - # sha on serialization - stream = StringIO() - cm._serialize(stream) - ns += 1 - streamlen = stream.tell() - stream.seek(0) - - istream = rwrepo.odb.store(IStream(Commit.type, streamlen, stream)) - assert istream.hexsha == cm.hexsha - - nc = Commit(rwrepo, Commit.NULL_BIN_SHA, cm.tree, - cm.author, cm.authored_date, cm.author_tz_offset, - cm.committer, cm.committed_date, cm.committer_tz_offset, - cm.message, cm.parents, cm.encoding) - - assert nc.parents == cm.parents - stream = StringIO() - nc._serialize(stream) - ns += 1 - streamlen = stream.tell() - stream.seek(0) - - # reuse istream - istream.size = streamlen - istream.stream = stream - istream.binsha = None - nc.binsha = rwrepo.odb.store(istream).binsha - - # if it worked, we have exactly the same contents ! - assert nc.hexsha == cm.hexsha - # END check commits - elapsed = time.time() - st - - if print_performance_info: - print >> sys.stderr, "Serialized %i and deserialized %i commits in %f s ( (%f, %f) commits / s" % (ns, nds, elapsed, ns/elapsed, nds/elapsed) - # END handle performance info - + """traverse all commits in the history of commit identified by commit_id and check + if the serialization works. + :param print_performance_info: if True, we will show how fast we are""" + ns = 0 # num serializations + nds = 0 # num deserializations + + st = time.time() + for cm in rwrepo.commit(commit_id).traverse(): + nds += 1 + + # assert that we deserialize commits correctly, hence we get the same + # sha on serialization + stream = StringIO() + cm._serialize(stream) + ns += 1 + streamlen = stream.tell() + stream.seek(0) + + istream = rwrepo.odb.store(IStream(Commit.type, streamlen, stream)) + assert istream.hexsha == cm.hexsha + + nc = Commit(rwrepo, Commit.NULL_BIN_SHA, cm.tree, + cm.author, cm.authored_date, cm.author_tz_offset, + cm.committer, cm.committed_date, cm.committer_tz_offset, + cm.message, cm.parents, cm.encoding) + + assert nc.parents == cm.parents + stream = StringIO() + nc._serialize(stream) + ns += 1 + streamlen = stream.tell() + stream.seek(0) + + # reuse istream + istream.size = streamlen + istream.stream = stream + istream.binsha = None + nc.binsha = rwrepo.odb.store(istream).binsha + + # if it worked, we have exactly the same contents ! + assert nc.hexsha == cm.hexsha + # END check commits + elapsed = time.time() - st + + if print_performance_info: + print >> sys.stderr, "Serialized %i and deserialized %i commits in %f s ( (%f, %f) commits / s" % (ns, nds, elapsed, ns/elapsed, nds/elapsed) + # END handle performance info + class TestCommit(TestObjectBase): - def test_bake(self): + def test_bake(self): - commit = self.rorepo.commit('2454ae89983a4496a445ce347d7a41c0bb0ea7ae') - # commits have no dict - self.failUnlessRaises(AttributeError, setattr, commit, 'someattr', 1) - commit.author # bake + commit = self.rorepo.commit('2454ae89983a4496a445ce347d7a41c0bb0ea7ae') + # commits have no dict + self.failUnlessRaises(AttributeError, setattr, commit, 'someattr', 1) + commit.author # bake - assert_equal("Sebastian Thiel", commit.author.name) - assert_equal("byronimo@gmail.com", commit.author.email) - assert commit.author == commit.committer - assert isinstance(commit.authored_date, int) and isinstance(commit.committed_date, int) - assert isinstance(commit.author_tz_offset, int) and isinstance(commit.committer_tz_offset, int) - assert commit.message == "Added missing information to docstrings of commit and stats module\n" + assert_equal("Sebastian Thiel", commit.author.name) + assert_equal("byronimo@gmail.com", commit.author.email) + assert commit.author == commit.committer + assert isinstance(commit.authored_date, int) and isinstance(commit.committed_date, int) + assert isinstance(commit.author_tz_offset, int) and isinstance(commit.committer_tz_offset, int) + assert commit.message == "Added missing information to docstrings of commit and stats module\n" - def test_stats(self): - commit = self.rorepo.commit('33ebe7acec14b25c5f84f35a664803fcab2f7781') - stats = commit.stats - - def check_entries(d): - assert isinstance(d, dict) - for key in ("insertions", "deletions", "lines"): - assert key in d - # END assertion helper - assert stats.files - assert stats.total - - check_entries(stats.total) - assert "files" in stats.total - - for filepath, d in stats.files.items(): - check_entries(d) - # END for each stated file - - # assure data is parsed properly - michael = Actor._from_string("Michael Trier <mtrier@gmail.com>") - assert commit.author == michael - assert commit.committer == michael - assert commit.authored_date == 1210193388 - assert commit.committed_date == 1210193388 - assert commit.author_tz_offset == 14400, commit.author_tz_offset - assert commit.committer_tz_offset == 14400, commit.committer_tz_offset - assert commit.message == "initial project\n" - - def test_unicode_actor(self): - # assure we can parse unicode actors correctly - name = "Üäöß ÄußÉ".decode("utf-8") - assert len(name) == 9 - special = Actor._from_string(u"%s <something@this.com>" % name) - assert special.name == name - assert isinstance(special.name, unicode) - - def test_traversal(self): - start = self.rorepo.commit("a4d06724202afccd2b5c54f81bcf2bf26dea7fff") - first = self.rorepo.commit("33ebe7acec14b25c5f84f35a664803fcab2f7781") - p0 = start.parents[0] - p1 = start.parents[1] - p00 = p0.parents[0] - p10 = p1.parents[0] - - # basic branch first, depth first - dfirst = start.traverse(branch_first=False) - bfirst = start.traverse(branch_first=True) - assert dfirst.next() == p0 - assert dfirst.next() == p00 - - assert bfirst.next() == p0 - assert bfirst.next() == p1 - assert bfirst.next() == p00 - assert bfirst.next() == p10 - - # at some point, both iterations should stop - assert list(bfirst)[-1] == first - stoptraverse = self.rorepo.commit("254d04aa3180eb8b8daf7b7ff25f010cd69b4e7d").traverse(as_edge=True) - l = list(stoptraverse) - assert len(l[0]) == 2 - - # ignore self - assert start.traverse(ignore_self=False).next() == start - - # depth - assert len(list(start.traverse(ignore_self=False, depth=0))) == 1 - - # prune - assert start.traverse(branch_first=1, prune=lambda i,d: i==p0).next() == p1 - - # predicate - assert start.traverse(branch_first=1, predicate=lambda i,d: i==p1).next() == p1 - - # traversal should stop when the beginning is reached - self.failUnlessRaises(StopIteration, first.traverse().next) - - # parents of the first commit should be empty ( as the only parent has a null - # sha ) - assert len(first.parents) == 0 - - def test_iteration(self): - # we can iterate commits - all_commits = Commit.list_items(self.rorepo, self.rorepo.head) - assert all_commits - assert all_commits == list(self.rorepo.iter_commits()) - - # this includes merge commits - mcomit = self.rorepo.commit('d884adc80c80300b4cc05321494713904ef1df2d') - assert mcomit in all_commits - - # we can limit the result to paths - ltd_commits = list(self.rorepo.iter_commits(paths='CHANGES')) - assert ltd_commits and len(ltd_commits) < len(all_commits) - - # show commits of multiple paths, resulting in a union of commits - less_ltd_commits = list(Commit.iter_items(self.rorepo, 'master', paths=('CHANGES', 'AUTHORS'))) - assert len(ltd_commits) < len(less_ltd_commits) - - def test_iter_items(self): - # pretty not allowed - self.failUnlessRaises(ValueError, Commit.iter_items, self.rorepo, 'master', pretty="raw") - - def test_rev_list_bisect_all(self): - """ - 'git rev-list --bisect-all' returns additional information - in the commit header. This test ensures that we properly parse it. - """ - revs = self.rorepo.git.rev_list('933d23bf95a5bd1624fbcdf328d904e1fa173474', - first_parent=True, - bisect_all=True) + def test_stats(self): + commit = self.rorepo.commit('33ebe7acec14b25c5f84f35a664803fcab2f7781') + stats = commit.stats + + def check_entries(d): + assert isinstance(d, dict) + for key in ("insertions", "deletions", "lines"): + assert key in d + # END assertion helper + assert stats.files + assert stats.total + + check_entries(stats.total) + assert "files" in stats.total + + for filepath, d in stats.files.items(): + check_entries(d) + # END for each stated file + + # assure data is parsed properly + michael = Actor._from_string("Michael Trier <mtrier@gmail.com>") + assert commit.author == michael + assert commit.committer == michael + assert commit.authored_date == 1210193388 + assert commit.committed_date == 1210193388 + assert commit.author_tz_offset == 14400, commit.author_tz_offset + assert commit.committer_tz_offset == 14400, commit.committer_tz_offset + assert commit.message == "initial project\n" + + def test_unicode_actor(self): + # assure we can parse unicode actors correctly + name = "Üäöß ÄußÉ".decode("utf-8") + assert len(name) == 9 + special = Actor._from_string(u"%s <something@this.com>" % name) + assert special.name == name + assert isinstance(special.name, unicode) + + def test_traversal(self): + start = self.rorepo.commit("a4d06724202afccd2b5c54f81bcf2bf26dea7fff") + first = self.rorepo.commit("33ebe7acec14b25c5f84f35a664803fcab2f7781") + p0 = start.parents[0] + p1 = start.parents[1] + p00 = p0.parents[0] + p10 = p1.parents[0] + + # basic branch first, depth first + dfirst = start.traverse(branch_first=False) + bfirst = start.traverse(branch_first=True) + assert dfirst.next() == p0 + assert dfirst.next() == p00 + + assert bfirst.next() == p0 + assert bfirst.next() == p1 + assert bfirst.next() == p00 + assert bfirst.next() == p10 + + # at some point, both iterations should stop + assert list(bfirst)[-1] == first + stoptraverse = self.rorepo.commit("254d04aa3180eb8b8daf7b7ff25f010cd69b4e7d").traverse(as_edge=True) + l = list(stoptraverse) + assert len(l[0]) == 2 + + # ignore self + assert start.traverse(ignore_self=False).next() == start + + # depth + assert len(list(start.traverse(ignore_self=False, depth=0))) == 1 + + # prune + assert start.traverse(branch_first=1, prune=lambda i,d: i==p0).next() == p1 + + # predicate + assert start.traverse(branch_first=1, predicate=lambda i,d: i==p1).next() == p1 + + # traversal should stop when the beginning is reached + self.failUnlessRaises(StopIteration, first.traverse().next) + + # parents of the first commit should be empty ( as the only parent has a null + # sha ) + assert len(first.parents) == 0 + + def test_iteration(self): + # we can iterate commits + all_commits = Commit.list_items(self.rorepo, self.rorepo.head) + assert all_commits + assert all_commits == list(self.rorepo.iter_commits()) + + # this includes merge commits + mcomit = self.rorepo.commit('d884adc80c80300b4cc05321494713904ef1df2d') + assert mcomit in all_commits + + # we can limit the result to paths + ltd_commits = list(self.rorepo.iter_commits(paths='CHANGES')) + assert ltd_commits and len(ltd_commits) < len(all_commits) + + # show commits of multiple paths, resulting in a union of commits + less_ltd_commits = list(Commit.iter_items(self.rorepo, 'master', paths=('CHANGES', 'AUTHORS'))) + assert len(ltd_commits) < len(less_ltd_commits) + + def test_iter_items(self): + # pretty not allowed + self.failUnlessRaises(ValueError, Commit.iter_items, self.rorepo, 'master', pretty="raw") + + def test_rev_list_bisect_all(self): + """ + 'git rev-list --bisect-all' returns additional information + in the commit header. This test ensures that we properly parse it. + """ + revs = self.rorepo.git.rev_list('933d23bf95a5bd1624fbcdf328d904e1fa173474', + first_parent=True, + bisect_all=True) - commits = Commit._iter_from_process_or_stream(self.rorepo, StringProcessAdapter(revs)) - expected_ids = ( - '7156cece3c49544abb6bf7a0c218eb36646fad6d', - '1f66cfbbce58b4b552b041707a12d437cc5f400a', - '33ebe7acec14b25c5f84f35a664803fcab2f7781', - '933d23bf95a5bd1624fbcdf328d904e1fa173474' - ) - for sha1, commit in zip(expected_ids, commits): - assert_equal(sha1, commit.hexsha) + commits = Commit._iter_from_process_or_stream(self.rorepo, StringProcessAdapter(revs)) + expected_ids = ( + '7156cece3c49544abb6bf7a0c218eb36646fad6d', + '1f66cfbbce58b4b552b041707a12d437cc5f400a', + '33ebe7acec14b25c5f84f35a664803fcab2f7781', + '933d23bf95a5bd1624fbcdf328d904e1fa173474' + ) + for sha1, commit in zip(expected_ids, commits): + assert_equal(sha1, commit.hexsha) - def test_count(self): - assert self.rorepo.tag('refs/tags/0.1.5').commit.count( ) == 143 - - def test_list(self): - assert isinstance(Commit.list_items(self.rorepo, '0.1.5', max_count=5)[hex_to_bin('5117c9c8a4d3af19a9958677e45cda9269de1541')], Commit) + def test_count(self): + assert self.rorepo.tag('refs/tags/0.1.5').commit.count( ) == 143 + + def test_list(self): + assert isinstance(Commit.list_items(self.rorepo, '0.1.5', max_count=5)[hex_to_bin('5117c9c8a4d3af19a9958677e45cda9269de1541')], Commit) - def test_str(self): - commit = Commit(self.rorepo, Commit.NULL_BIN_SHA) - assert_equal(Commit.NULL_HEX_SHA, str(commit)) + def test_str(self): + commit = Commit(self.rorepo, Commit.NULL_BIN_SHA) + assert_equal(Commit.NULL_HEX_SHA, str(commit)) - def test_repr(self): - commit = Commit(self.rorepo, Commit.NULL_BIN_SHA) - assert_equal('<git.Commit "%s">' % Commit.NULL_HEX_SHA, repr(commit)) + def test_repr(self): + commit = Commit(self.rorepo, Commit.NULL_BIN_SHA) + assert_equal('<git.Commit "%s">' % Commit.NULL_HEX_SHA, repr(commit)) - def test_equality(self): - commit1 = Commit(self.rorepo, Commit.NULL_BIN_SHA) - commit2 = Commit(self.rorepo, Commit.NULL_BIN_SHA) - commit3 = Commit(self.rorepo, "\1"*20) - assert_equal(commit1, commit2) - assert_not_equal(commit2, commit3) - - def test_iter_parents(self): - # should return all but ourselves, even if skip is defined - c = self.rorepo.commit('0.1.5') - for skip in (0, 1): - piter = c.iter_parents(skip=skip) - first_parent = piter.next() - assert first_parent != c - assert first_parent == c.parents[0] - # END for each - - def test_base(self): - name_rev = self.rorepo.head.commit.name_rev - assert isinstance(name_rev, basestring) - - @with_rw_repo('HEAD', bare=True) - def test_serialization(self, rwrepo): - # create all commits of our repo - assert_commit_serialization(rwrepo, '0.1.6') - - def test_serialization_unicode_support(self): - assert Commit.default_encoding.lower() == 'utf-8' - - # create a commit with unicode in the message, and the author's name - # Verify its serialization and deserialization - cmt = self.rorepo.commit('0.1.6') - assert isinstance(cmt.message, unicode) # it automatically decodes it as such - assert isinstance(cmt.author.name, unicode) # same here - - cmt.message = "üäêèß".decode("utf-8") - assert len(cmt.message) == 5 - - cmt.author.name = "äüß".decode("utf-8") - assert len(cmt.author.name) == 3 - - cstream = StringIO() - cmt._serialize(cstream) - cstream.seek(0) - assert len(cstream.getvalue()) - - ncmt = Commit(self.rorepo, cmt.binsha) - ncmt._deserialize(cstream) - - assert cmt.author.name == ncmt.author.name - assert cmt.message == ncmt.message - # actually, it can't be printed in a shell as repr wants to have ascii only - # it appears - cmt.author.__repr__() - + def test_equality(self): + commit1 = Commit(self.rorepo, Commit.NULL_BIN_SHA) + commit2 = Commit(self.rorepo, Commit.NULL_BIN_SHA) + commit3 = Commit(self.rorepo, "\1"*20) + assert_equal(commit1, commit2) + assert_not_equal(commit2, commit3) + + def test_iter_parents(self): + # should return all but ourselves, even if skip is defined + c = self.rorepo.commit('0.1.5') + for skip in (0, 1): + piter = c.iter_parents(skip=skip) + first_parent = piter.next() + assert first_parent != c + assert first_parent == c.parents[0] + # END for each + + def test_base(self): + name_rev = self.rorepo.head.commit.name_rev + assert isinstance(name_rev, basestring) + + @with_rw_repo('HEAD', bare=True) + def test_serialization(self, rwrepo): + # create all commits of our repo + assert_commit_serialization(rwrepo, '0.1.6') + + def test_serialization_unicode_support(self): + assert Commit.default_encoding.lower() == 'utf-8' + + # create a commit with unicode in the message, and the author's name + # Verify its serialization and deserialization + cmt = self.rorepo.commit('0.1.6') + assert isinstance(cmt.message, unicode) # it automatically decodes it as such + assert isinstance(cmt.author.name, unicode) # same here + + cmt.message = "üäêèß".decode("utf-8") + assert len(cmt.message) == 5 + + cmt.author.name = "äüß".decode("utf-8") + assert len(cmt.author.name) == 3 + + cstream = StringIO() + cmt._serialize(cstream) + cstream.seek(0) + assert len(cstream.getvalue()) + + ncmt = Commit(self.rorepo, cmt.binsha) + ncmt._deserialize(cstream) + + assert cmt.author.name == ncmt.author.name + assert cmt.message == ncmt.message + # actually, it can't be printed in a shell as repr wants to have ascii only + # it appears + cmt.author.__repr__() + diff --git a/git/test/objects/test_submodule.py b/git/test/objects/test_submodule.py index 2b7c7f40..bfafb150 100644 --- a/git/test/objects/test_submodule.py +++ b/git/test/objects/test_submodule.py @@ -17,580 +17,580 @@ import sys # to keep file handles open. On windows we get problems as they are not properly # closed due to mmap bugs on windows (as it appears) if sys.platform == 'win32': - try: - import smmap.util - smmap.util.MapRegion._test_read_into_memory = True - except ImportError: - sys.stderr.write("The submodule tests will fail as some files cannot be removed due to open file handles.\n") - sys.stderr.write("The latest version of gitdb uses a memory map manager which can be configured to work around this problem") + try: + import smmap.util + smmap.util.MapRegion._test_read_into_memory = True + except ImportError: + sys.stderr.write("The submodule tests will fail as some files cannot be removed due to open file handles.\n") + sys.stderr.write("The latest version of gitdb uses a memory map manager which can be configured to work around this problem") #END handle windows platform class TestRootProgress(RootUpdateProgress): - """Just prints messages, for now without checking the correctness of the states""" - - def update(self, op, index, max_count, message='', input=''): - print message - + """Just prints messages, for now without checking the correctness of the states""" + + def update(self, op, index, max_count, message='', input=''): + print message + prog = TestRootProgress() class TestSubmodule(TestObjectBase): - k_subm_current = "468cad66ff1f80ddaeee4123c24e4d53a032c00d" - k_subm_changed = "394ed7006ee5dc8bddfd132b64001d5dfc0ffdd3" - k_no_subm_tag = "0.1.6" - k_github_gitdb_url = 'git://github.com/gitpython-developers/gitdb.git' - env_gitdb_local_path = "GITPYTHON_TEST_GITDB_LOCAL_PATH" - - def _generate_async_local_path(self): - return to_native_path_linux(join_path_native(self.rorepo.working_tree_dir, 'git/ext/async')) + k_subm_current = "468cad66ff1f80ddaeee4123c24e4d53a032c00d" + k_subm_changed = "394ed7006ee5dc8bddfd132b64001d5dfc0ffdd3" + k_no_subm_tag = "0.1.6" + k_github_gitdb_url = 'git://github.com/gitpython-developers/gitdb.git' + env_gitdb_local_path = "GITPYTHON_TEST_GITDB_LOCAL_PATH" + + def _generate_async_local_path(self): + return to_native_path_linux(join_path_native(self.rorepo.working_tree_dir, 'git/ext/async')) - def _rewrite_gitdb_to_local_path(self, smgitdb): - """Rewrites the given submodule to point to the local path of the gitdb repository, if possible. - Otherwise it leaves it unchanged - :return: new clone path, or None if no new path was set""" - new_smclone_path = os.environ.get(self.env_gitdb_local_path) - if new_smclone_path is not None: - writer = smgitdb.config_writer() - writer.set_value('url', new_smclone_path) - del(writer) - assert smgitdb.config_reader().get_value('url') == new_smclone_path - assert smgitdb.url == new_smclone_path - else: - sys.stderr.write("Submodule tests need the gitdb repository. You can specify a local source setting the %s environment variable. Otherwise it will be downloaded from the internet" % self.env_gitdb_local_path) - #END handle submodule path - return new_smclone_path + def _rewrite_gitdb_to_local_path(self, smgitdb): + """Rewrites the given submodule to point to the local path of the gitdb repository, if possible. + Otherwise it leaves it unchanged + :return: new clone path, or None if no new path was set""" + new_smclone_path = os.environ.get(self.env_gitdb_local_path) + if new_smclone_path is not None: + writer = smgitdb.config_writer() + writer.set_value('url', new_smclone_path) + del(writer) + assert smgitdb.config_reader().get_value('url') == new_smclone_path + assert smgitdb.url == new_smclone_path + else: + sys.stderr.write("Submodule tests need the gitdb repository. You can specify a local source setting the %s environment variable. Otherwise it will be downloaded from the internet" % self.env_gitdb_local_path) + #END handle submodule path + return new_smclone_path - def _do_base_tests(self, rwrepo): - """Perform all tests in the given repository, it may be bare or nonbare""" - # manual instantiation - smm = Submodule(rwrepo, "\0"*20) - # name needs to be set in advance - self.failUnlessRaises(AttributeError, getattr, smm, 'name') - - # iterate - 1 submodule - sms = Submodule.list_items(rwrepo, self.k_subm_current) - assert len(sms) == 1 - sm = sms[0] - - # at a different time, there is None - assert len(Submodule.list_items(rwrepo, self.k_no_subm_tag)) == 0 - - assert sm.path == 'git/ext/gitdb' - assert sm.path != sm.name # in our case, we have ids there, which don't equal the path - assert sm.url == self.k_github_gitdb_url - assert sm.branch_path == 'refs/heads/master' # the default ... - assert sm.branch_name == 'master' - assert sm.parent_commit == rwrepo.head.commit - # size is always 0 - assert sm.size == 0 - # the module is not checked-out yet - self.failUnlessRaises(InvalidGitRepositoryError, sm.module) - - # which is why we can't get the branch either - it points into the module() repository - self.failUnlessRaises(InvalidGitRepositoryError, getattr, sm, 'branch') - - # branch_path works, as its just a string - assert isinstance(sm.branch_path, basestring) - - # some commits earlier we still have a submodule, but its at a different commit - smold = Submodule.iter_items(rwrepo, self.k_subm_changed).next() - assert smold.binsha != sm.binsha - assert smold != sm # the name changed - - # force it to reread its information - del(smold._url) - smold.url == sm.url - - # test config_reader/writer methods - sm.config_reader() - new_smclone_path = None # keep custom paths for later - new_csmclone_path = None # - if rwrepo.bare: - self.failUnlessRaises(InvalidGitRepositoryError, sm.config_writer) - else: - # for faster checkout, set the url to the local path - # Note: This is nice but doesn't work anymore with the latest git-python - # version. This would also mean we need internet for this to work which - # is why we allow an override using an environment variable - new_smclone_path = self._rewrite_gitdb_to_local_path(sm) - # END handle bare repo - smold.config_reader() - - # cannot get a writer on historical submodules - if not rwrepo.bare: - self.failUnlessRaises(ValueError, smold.config_writer) - # END handle bare repo - - # make the old into a new - this doesn't work as the name changed - prev_parent_commit = smold.parent_commit - self.failUnlessRaises(ValueError, smold.set_parent_commit, self.k_subm_current) - # the sha is properly updated - smold.set_parent_commit(self.k_subm_changed+"~1") - assert smold.binsha != sm.binsha - - # raises if the sm didn't exist in new parent - it keeps its - # parent_commit unchanged - self.failUnlessRaises(ValueError, smold.set_parent_commit, self.k_no_subm_tag) - - # TEST TODO: if a path in the gitmodules file, but not in the index, it raises - - # TEST UPDATE - ############## - # module retrieval is not always possible - if rwrepo.bare: - self.failUnlessRaises(InvalidGitRepositoryError, sm.module) - self.failUnlessRaises(InvalidGitRepositoryError, sm.remove) - self.failUnlessRaises(InvalidGitRepositoryError, sm.add, rwrepo, 'here', 'there') - else: - # its not checked out in our case - self.failUnlessRaises(InvalidGitRepositoryError, sm.module) - assert not sm.module_exists() - - # currently there is only one submodule - assert len(list(rwrepo.iter_submodules())) == 1 - assert sm.binsha != "\0"*20 - - # TEST ADD - ########### - # preliminary tests - # adding existing returns exactly the existing - sma = Submodule.add(rwrepo, sm.name, sm.path) - assert sma.path == sm.path - - # no url and no module at path fails - self.failUnlessRaises(ValueError, Submodule.add, rwrepo, "newsubm", "pathtorepo", url=None) - - # CONTINUE UPDATE - ################# - - # lets update it - its a recursive one too - newdir = os.path.join(sm.abspath, 'dir') - os.makedirs(newdir) - - # update fails if the path already exists non-empty - self.failUnlessRaises(OSError, sm.update) - os.rmdir(newdir) - - # dry-run does nothing - sm.update(dry_run=True, progress=prog) - assert not sm.module_exists() - - assert sm.update() is sm - sm_repopath = sm.path # cache for later - assert sm.module_exists() - assert isinstance(sm.module(), git.Repo) - assert sm.module().working_tree_dir == sm.abspath - - # INTERLEAVE ADD TEST - ##################### - # url must match the one in the existing repository ( if submodule name suggests a new one ) - # or we raise - self.failUnlessRaises(ValueError, Submodule.add, rwrepo, "newsubm", sm.path, "git://someurl/repo.git") - - - # CONTINUE UPDATE - ################# - # we should have setup a tracking branch, which is also active - assert sm.module().head.ref.tracking_branch() is not None - - # delete the whole directory and re-initialize - shutil.rmtree(sm.abspath) - assert len(sm.children()) == 0 - # dry-run does nothing - sm.update(dry_run=True, recursive=False, progress=prog) - assert len(sm.children()) == 0 - - sm.update(recursive=False) - assert len(list(rwrepo.iter_submodules())) == 2 - assert len(sm.children()) == 1 # its not checked out yet - csm = sm.children()[0] - assert not csm.module_exists() - csm_repopath = csm.path - - # adjust the path of the submodules module to point to the local destination - # In the current gitpython version, async is used directly by gitpython - new_csmclone_path = self._generate_async_local_path() - csm.config_writer().set_value('url', new_csmclone_path) - assert csm.url == new_csmclone_path - - # dry-run does nothing - assert not csm.module_exists() - sm.update(recursive=True, dry_run=True, progress=prog) - assert not csm.module_exists() - - # update recursively again - sm.update(recursive=True) - assert csm.module_exists() - - # tracking branch once again - csm.module().head.ref.tracking_branch() is not None - - # this flushed in a sub-submodule - assert len(list(rwrepo.iter_submodules())) == 2 - - - # reset both heads to the previous version, verify that to_latest_revision works - smods = (sm.module(), csm.module()) - for repo in smods: - repo.head.reset('HEAD~2', working_tree=1) - # END for each repo to reset - - # dry run does nothing - sm.update(recursive=True, dry_run=True, progress=prog) - for repo in smods: - assert repo.head.commit != repo.head.ref.tracking_branch().commit - # END for each repo to check - - sm.update(recursive=True, to_latest_revision=True) - for repo in smods: - assert repo.head.commit == repo.head.ref.tracking_branch().commit - # END for each repo to check - del(smods) - - # if the head is detached, it still works ( but warns ) - smref = sm.module().head.ref - sm.module().head.ref = 'HEAD~1' - # if there is no tracking branch, we get a warning as well - csm_tracking_branch = csm.module().head.ref.tracking_branch() - csm.module().head.ref.set_tracking_branch(None) - sm.update(recursive=True, to_latest_revision=True) - - # to_latest_revision changes the child submodule's commit, it needs an - # update now - csm.set_parent_commit(csm.repo.head.commit) - - # undo the changes - sm.module().head.ref = smref - csm.module().head.ref.set_tracking_branch(csm_tracking_branch) - - # REMOVAL OF REPOSITOTRY - ######################## - # must delete something - self.failUnlessRaises(ValueError, csm.remove, module=False, configuration=False) - # We have modified the configuration, hence the index is dirty, and the - # deletion will fail - # NOTE: As we did a few updates in the meanwhile, the indices were reset - # Hence we create some changes - csm.set_parent_commit(csm.repo.head.commit) - sm.config_writer().set_value("somekey", "somevalue") - csm.config_writer().set_value("okey", "ovalue") - self.failUnlessRaises(InvalidGitRepositoryError, sm.remove) - # if we remove the dirty index, it would work - sm.module().index.reset() - # still, we have the file modified - self.failUnlessRaises(InvalidGitRepositoryError, sm.remove, dry_run=True) - sm.module().index.reset(working_tree=True) - - # make sure sub-submodule is not modified by forcing it to update - # to the revision it is supposed to point to. - for subitem in sm.traverse(): - subitem.update() - #END checkout to right commit - - # this would work - assert sm.remove(dry_run=True) is sm - assert sm.module_exists() - sm.remove(force=True, dry_run=True) - assert sm.module_exists() - - # but ... we have untracked files in the child submodule - fn = join_path_native(csm.module().working_tree_dir, "newfile") - open(fn, 'w').write("hi") - self.failUnlessRaises(InvalidGitRepositoryError, sm.remove) - - # forcibly delete the child repository - prev_count = len(sm.children()) - assert csm.remove(force=True) is csm - assert not csm.exists() - assert not csm.module_exists() - assert len(sm.children()) == prev_count - 1 - # now we have a changed index, as configuration was altered. - # fix this - sm.module().index.reset(working_tree=True) - - # now delete only the module of the main submodule - assert sm.module_exists() - sm.remove(configuration=False) - assert sm.exists() - assert not sm.module_exists() - assert sm.config_reader().get_value('url') - - # delete the rest - sm.remove() - assert not sm.exists() - assert not sm.module_exists() - - assert len(rwrepo.submodules) == 0 - - # ADD NEW SUBMODULE - ################### - # add a simple remote repo - trailing slashes are no problem - smid = "newsub" - osmid = "othersub" - nsm = Submodule.add(rwrepo, smid, sm_repopath, new_smclone_path+"/", None, no_checkout=True) - assert nsm.name == smid - assert nsm.module_exists() - assert nsm.exists() - # its not checked out - assert not os.path.isfile(join_path_native(nsm.module().working_tree_dir, Submodule.k_modules_file)) - assert len(rwrepo.submodules) == 1 - - # add another submodule, but into the root, not as submodule - osm = Submodule.add(rwrepo, osmid, csm_repopath, new_csmclone_path, Submodule.k_head_default) - assert osm != nsm - assert osm.module_exists() - assert osm.exists() - assert os.path.isfile(join_path_native(osm.module().working_tree_dir, 'setup.py')) - - assert len(rwrepo.submodules) == 2 - - # commit the changes, just to finalize the operation - rwrepo.index.commit("my submod commit") - assert len(rwrepo.submodules) == 2 - - # needs update as the head changed, it thinks its in the history - # of the repo otherwise - nsm.set_parent_commit(rwrepo.head.commit) - osm.set_parent_commit(rwrepo.head.commit) - - # MOVE MODULE - ############# - # invalid inptu - self.failUnlessRaises(ValueError, nsm.move, 'doesntmatter', module=False, configuration=False) - - # renaming to the same path does nothing - assert nsm.move(sm.path) is nsm - - # rename a module - nmp = join_path_native("new", "module", "dir") + "/" # new module path - pmp = nsm.path - abspmp = nsm.abspath - assert nsm.move(nmp) is nsm - nmp = nmp[:-1] # cut last / - nmpl = to_native_path_linux(nmp) - assert nsm.path == nmpl - assert rwrepo.submodules[0].path == nmpl - - mpath = 'newsubmodule' - absmpath = join_path_native(rwrepo.working_tree_dir, mpath) - open(absmpath, 'w').write('') - self.failUnlessRaises(ValueError, nsm.move, mpath) - os.remove(absmpath) - - # now it works, as we just move it back - nsm.move(pmp) - assert nsm.path == pmp - assert rwrepo.submodules[0].path == pmp - - # TODO lowprio: test remaining exceptions ... for now its okay, the code looks right - - # REMOVE 'EM ALL - ################ - # if a submodule's repo has no remotes, it can't be added without an explicit url - osmod = osm.module() - - osm.remove(module=False) - for remote in osmod.remotes: - remote.remove(osmod, remote.name) - assert not osm.exists() - self.failUnlessRaises(ValueError, Submodule.add, rwrepo, osmid, csm_repopath, url=None) - # END handle bare mode - - # Error if there is no submodule file here - self.failUnlessRaises(IOError, Submodule._config_parser, rwrepo, rwrepo.commit(self.k_no_subm_tag), True) - - @with_rw_repo(k_subm_current) - def test_base_rw(self, rwrepo): - self._do_base_tests(rwrepo) - - @with_rw_repo(k_subm_current, bare=True) - def test_base_bare(self, rwrepo): - self._do_base_tests(rwrepo) - - @with_rw_repo(k_subm_current, bare=False) - def test_root_module(self, rwrepo): - # Can query everything without problems - rm = RootModule(rwrepo) - # test new constructor - assert rm.parent_commit == RootModule(self.rorepo, self.rorepo.commit(self.k_subm_current)).parent_commit - assert rm.module() is rwrepo - - # try attributes - rm.binsha - rm.mode - rm.path - assert rm.name == rm.k_root_name - assert rm.parent_commit == self.rorepo.commit(self.k_subm_current) - rm.url - rm.branch - - assert len(rm.list_items(rm.module())) == 1 - rm.config_reader() - rm.config_writer() - - # deep traversal git / async - rsmsp = [sm.path for sm in rm.traverse()] - assert len(rsmsp) == 1 # gitdb only - its not yet uptodate so it has no submodule - - # cannot set the parent commit as root module's path didn't exist - self.failUnlessRaises(ValueError, rm.set_parent_commit, 'HEAD') - - # TEST UPDATE - ############# - # setup commit which remove existing, add new and modify existing submodules - rm = RootModule(rwrepo) - assert len(rm.children()) == 1 - - # modify path without modifying the index entry - # ( which is what the move method would do properly ) - #================================================== - sm = rm.children()[0] - pp = "path/prefix" - fp = join_path_native(pp, sm.path) - prep = sm.path - assert not sm.module_exists() # was never updated after rwrepo's clone - - # assure we clone from a local source - self._rewrite_gitdb_to_local_path(sm) - - # dry-run does nothing - sm.update(recursive=False, dry_run=True, progress=prog) - assert not sm.module_exists() - - sm.update(recursive=False) - assert sm.module_exists() - sm.config_writer().set_value('path', fp) # change path to something with prefix AFTER url change - - # update fails as list_items in such a situations cannot work, as it cannot - # find the entry at the changed path - self.failUnlessRaises(InvalidGitRepositoryError, rm.update, recursive=False) - - # move it properly - doesn't work as it its path currently points to an indexentry - # which doesn't exist ( move it to some path, it doesn't matter here ) - self.failUnlessRaises(InvalidGitRepositoryError, sm.move, pp) - # reset the path(cache) to where it was, now it works - sm.path = prep - sm.move(fp, module=False) # leave it at the old location - - assert not sm.module_exists() - cpathchange = rwrepo.index.commit("changed sm path") # finally we can commit - - # update puts the module into place - rm.update(recursive=False, progress=prog) - sm.set_parent_commit(cpathchange) - assert sm.module_exists() - - # add submodule - #================ - nsmn = "newsubmodule" - nsmp = "submrepo" - async_url = self._generate_async_local_path() - nsm = Submodule.add(rwrepo, nsmn, nsmp, url=async_url) - csmadded = rwrepo.index.commit("Added submodule").hexsha # make sure we don't keep the repo reference - nsm.set_parent_commit(csmadded) - assert nsm.module_exists() - # in our case, the module should not exist, which happens if we update a parent - # repo and a new submodule comes into life - nsm.remove(configuration=False, module=True) - assert not nsm.module_exists() and nsm.exists() - - - # dry-run does nothing - rm.update(recursive=False, dry_run=True, progress=prog) - - # otherwise it will work - rm.update(recursive=False, progress=prog) - assert nsm.module_exists() - - - - # remove submodule - the previous one - #==================================== - sm.set_parent_commit(csmadded) - smp = sm.abspath - assert not sm.remove(module=False).exists() - assert os.path.isdir(smp) # module still exists - csmremoved = rwrepo.index.commit("Removed submodule") - - # an update will remove the module - # not in dry_run - rm.update(recursive=False, dry_run=True) - assert os.path.isdir(smp) - - rm.update(recursive=False) - assert not os.path.isdir(smp) - - - # change url - #============= - # to the first repository, this way we have a fast checkout, and a completely different - # repository at the different url - nsm.set_parent_commit(csmremoved) - nsmurl = os.environ.get(self.env_gitdb_local_path, self.k_github_gitdb_url) - - # Note: We would have liked to have a different url, but we cannot - # provoke this case - assert nsm.url != nsmurl - nsm.config_writer().set_value('url', nsmurl) - csmpathchange = rwrepo.index.commit("changed url") - nsm.set_parent_commit(csmpathchange) - - prev_commit = nsm.module().head.commit - # dry-run does nothing - rm.update(recursive=False, dry_run=True, progress=prog) - assert nsm.module().remotes.origin.url != nsmurl - - rm.update(recursive=False, progress=prog) - assert nsm.module().remotes.origin.url == nsmurl - # head changed, as the remote url and its commit changed - assert prev_commit != nsm.module().head.commit - - # add the submodule's changed commit to the index, which is what the - # user would do - # beforehand, update our instance's binsha with the new one - nsm.binsha = nsm.module().head.commit.binsha - rwrepo.index.add([nsm]) - - # change branch - #================= - # we only have one branch, so we switch to a virtual one, and back - # to the current one to trigger the difference - cur_branch = nsm.branch - nsmm = nsm.module() - prev_commit = nsmm.head.commit - for branch in ("some_virtual_branch", cur_branch.name): - nsm.config_writer().set_value(Submodule.k_head_option, git.Head.to_full_path(branch)) - csmbranchchange = rwrepo.index.commit("changed branch to %s" % branch) - nsm.set_parent_commit(csmbranchchange) - # END for each branch to change - - # Lets remove our tracking branch to simulate some changes - nsmmh = nsmm.head - assert nsmmh.ref.tracking_branch() is None # never set it up until now - assert not nsmmh.is_detached - - #dry run does nothing - rm.update(recursive=False, dry_run=True, progress=prog) - assert nsmmh.ref.tracking_branch() is None - - # the real thing does - rm.update(recursive=False, progress=prog) - - assert nsmmh.ref.tracking_branch() is not None - assert not nsmmh.is_detached - - # recursive update - # ================= - # finally we recursively update a module, just to run the code at least once - # remove the module so that it has more work - assert len(nsm.children()) >= 1 # could include smmap - assert nsm.exists() and nsm.module_exists() and len(nsm.children()) >= 1 - # assure we pull locally only - nsmc = nsm.children()[0] - nsmc.config_writer().set_value('url', async_url) - rm.update(recursive=True, progress=prog, dry_run=True) # just to run the code - rm.update(recursive=True, progress=prog) - - assert len(nsm.children()) >= 2 and nsmc.module_exists() - + def _do_base_tests(self, rwrepo): + """Perform all tests in the given repository, it may be bare or nonbare""" + # manual instantiation + smm = Submodule(rwrepo, "\0"*20) + # name needs to be set in advance + self.failUnlessRaises(AttributeError, getattr, smm, 'name') + + # iterate - 1 submodule + sms = Submodule.list_items(rwrepo, self.k_subm_current) + assert len(sms) == 1 + sm = sms[0] + + # at a different time, there is None + assert len(Submodule.list_items(rwrepo, self.k_no_subm_tag)) == 0 + + assert sm.path == 'git/ext/gitdb' + assert sm.path != sm.name # in our case, we have ids there, which don't equal the path + assert sm.url == self.k_github_gitdb_url + assert sm.branch_path == 'refs/heads/master' # the default ... + assert sm.branch_name == 'master' + assert sm.parent_commit == rwrepo.head.commit + # size is always 0 + assert sm.size == 0 + # the module is not checked-out yet + self.failUnlessRaises(InvalidGitRepositoryError, sm.module) + + # which is why we can't get the branch either - it points into the module() repository + self.failUnlessRaises(InvalidGitRepositoryError, getattr, sm, 'branch') + + # branch_path works, as its just a string + assert isinstance(sm.branch_path, basestring) + + # some commits earlier we still have a submodule, but its at a different commit + smold = Submodule.iter_items(rwrepo, self.k_subm_changed).next() + assert smold.binsha != sm.binsha + assert smold != sm # the name changed + + # force it to reread its information + del(smold._url) + smold.url == sm.url + + # test config_reader/writer methods + sm.config_reader() + new_smclone_path = None # keep custom paths for later + new_csmclone_path = None # + if rwrepo.bare: + self.failUnlessRaises(InvalidGitRepositoryError, sm.config_writer) + else: + # for faster checkout, set the url to the local path + # Note: This is nice but doesn't work anymore with the latest git-python + # version. This would also mean we need internet for this to work which + # is why we allow an override using an environment variable + new_smclone_path = self._rewrite_gitdb_to_local_path(sm) + # END handle bare repo + smold.config_reader() + + # cannot get a writer on historical submodules + if not rwrepo.bare: + self.failUnlessRaises(ValueError, smold.config_writer) + # END handle bare repo + + # make the old into a new - this doesn't work as the name changed + prev_parent_commit = smold.parent_commit + self.failUnlessRaises(ValueError, smold.set_parent_commit, self.k_subm_current) + # the sha is properly updated + smold.set_parent_commit(self.k_subm_changed+"~1") + assert smold.binsha != sm.binsha + + # raises if the sm didn't exist in new parent - it keeps its + # parent_commit unchanged + self.failUnlessRaises(ValueError, smold.set_parent_commit, self.k_no_subm_tag) + + # TEST TODO: if a path in the gitmodules file, but not in the index, it raises + + # TEST UPDATE + ############## + # module retrieval is not always possible + if rwrepo.bare: + self.failUnlessRaises(InvalidGitRepositoryError, sm.module) + self.failUnlessRaises(InvalidGitRepositoryError, sm.remove) + self.failUnlessRaises(InvalidGitRepositoryError, sm.add, rwrepo, 'here', 'there') + else: + # its not checked out in our case + self.failUnlessRaises(InvalidGitRepositoryError, sm.module) + assert not sm.module_exists() + + # currently there is only one submodule + assert len(list(rwrepo.iter_submodules())) == 1 + assert sm.binsha != "\0"*20 + + # TEST ADD + ########### + # preliminary tests + # adding existing returns exactly the existing + sma = Submodule.add(rwrepo, sm.name, sm.path) + assert sma.path == sm.path + + # no url and no module at path fails + self.failUnlessRaises(ValueError, Submodule.add, rwrepo, "newsubm", "pathtorepo", url=None) + + # CONTINUE UPDATE + ################# + + # lets update it - its a recursive one too + newdir = os.path.join(sm.abspath, 'dir') + os.makedirs(newdir) + + # update fails if the path already exists non-empty + self.failUnlessRaises(OSError, sm.update) + os.rmdir(newdir) + + # dry-run does nothing + sm.update(dry_run=True, progress=prog) + assert not sm.module_exists() + + assert sm.update() is sm + sm_repopath = sm.path # cache for later + assert sm.module_exists() + assert isinstance(sm.module(), git.Repo) + assert sm.module().working_tree_dir == sm.abspath + + # INTERLEAVE ADD TEST + ##################### + # url must match the one in the existing repository ( if submodule name suggests a new one ) + # or we raise + self.failUnlessRaises(ValueError, Submodule.add, rwrepo, "newsubm", sm.path, "git://someurl/repo.git") + + + # CONTINUE UPDATE + ################# + # we should have setup a tracking branch, which is also active + assert sm.module().head.ref.tracking_branch() is not None + + # delete the whole directory and re-initialize + shutil.rmtree(sm.abspath) + assert len(sm.children()) == 0 + # dry-run does nothing + sm.update(dry_run=True, recursive=False, progress=prog) + assert len(sm.children()) == 0 + + sm.update(recursive=False) + assert len(list(rwrepo.iter_submodules())) == 2 + assert len(sm.children()) == 1 # its not checked out yet + csm = sm.children()[0] + assert not csm.module_exists() + csm_repopath = csm.path + + # adjust the path of the submodules module to point to the local destination + # In the current gitpython version, async is used directly by gitpython + new_csmclone_path = self._generate_async_local_path() + csm.config_writer().set_value('url', new_csmclone_path) + assert csm.url == new_csmclone_path + + # dry-run does nothing + assert not csm.module_exists() + sm.update(recursive=True, dry_run=True, progress=prog) + assert not csm.module_exists() + + # update recursively again + sm.update(recursive=True) + assert csm.module_exists() + + # tracking branch once again + csm.module().head.ref.tracking_branch() is not None + + # this flushed in a sub-submodule + assert len(list(rwrepo.iter_submodules())) == 2 + + + # reset both heads to the previous version, verify that to_latest_revision works + smods = (sm.module(), csm.module()) + for repo in smods: + repo.head.reset('HEAD~2', working_tree=1) + # END for each repo to reset + + # dry run does nothing + sm.update(recursive=True, dry_run=True, progress=prog) + for repo in smods: + assert repo.head.commit != repo.head.ref.tracking_branch().commit + # END for each repo to check + + sm.update(recursive=True, to_latest_revision=True) + for repo in smods: + assert repo.head.commit == repo.head.ref.tracking_branch().commit + # END for each repo to check + del(smods) + + # if the head is detached, it still works ( but warns ) + smref = sm.module().head.ref + sm.module().head.ref = 'HEAD~1' + # if there is no tracking branch, we get a warning as well + csm_tracking_branch = csm.module().head.ref.tracking_branch() + csm.module().head.ref.set_tracking_branch(None) + sm.update(recursive=True, to_latest_revision=True) + + # to_latest_revision changes the child submodule's commit, it needs an + # update now + csm.set_parent_commit(csm.repo.head.commit) + + # undo the changes + sm.module().head.ref = smref + csm.module().head.ref.set_tracking_branch(csm_tracking_branch) + + # REMOVAL OF REPOSITOTRY + ######################## + # must delete something + self.failUnlessRaises(ValueError, csm.remove, module=False, configuration=False) + # We have modified the configuration, hence the index is dirty, and the + # deletion will fail + # NOTE: As we did a few updates in the meanwhile, the indices were reset + # Hence we create some changes + csm.set_parent_commit(csm.repo.head.commit) + sm.config_writer().set_value("somekey", "somevalue") + csm.config_writer().set_value("okey", "ovalue") + self.failUnlessRaises(InvalidGitRepositoryError, sm.remove) + # if we remove the dirty index, it would work + sm.module().index.reset() + # still, we have the file modified + self.failUnlessRaises(InvalidGitRepositoryError, sm.remove, dry_run=True) + sm.module().index.reset(working_tree=True) + + # make sure sub-submodule is not modified by forcing it to update + # to the revision it is supposed to point to. + for subitem in sm.traverse(): + subitem.update() + #END checkout to right commit + + # this would work + assert sm.remove(dry_run=True) is sm + assert sm.module_exists() + sm.remove(force=True, dry_run=True) + assert sm.module_exists() + + # but ... we have untracked files in the child submodule + fn = join_path_native(csm.module().working_tree_dir, "newfile") + open(fn, 'w').write("hi") + self.failUnlessRaises(InvalidGitRepositoryError, sm.remove) + + # forcibly delete the child repository + prev_count = len(sm.children()) + assert csm.remove(force=True) is csm + assert not csm.exists() + assert not csm.module_exists() + assert len(sm.children()) == prev_count - 1 + # now we have a changed index, as configuration was altered. + # fix this + sm.module().index.reset(working_tree=True) + + # now delete only the module of the main submodule + assert sm.module_exists() + sm.remove(configuration=False) + assert sm.exists() + assert not sm.module_exists() + assert sm.config_reader().get_value('url') + + # delete the rest + sm.remove() + assert not sm.exists() + assert not sm.module_exists() + + assert len(rwrepo.submodules) == 0 + + # ADD NEW SUBMODULE + ################### + # add a simple remote repo - trailing slashes are no problem + smid = "newsub" + osmid = "othersub" + nsm = Submodule.add(rwrepo, smid, sm_repopath, new_smclone_path+"/", None, no_checkout=True) + assert nsm.name == smid + assert nsm.module_exists() + assert nsm.exists() + # its not checked out + assert not os.path.isfile(join_path_native(nsm.module().working_tree_dir, Submodule.k_modules_file)) + assert len(rwrepo.submodules) == 1 + + # add another submodule, but into the root, not as submodule + osm = Submodule.add(rwrepo, osmid, csm_repopath, new_csmclone_path, Submodule.k_head_default) + assert osm != nsm + assert osm.module_exists() + assert osm.exists() + assert os.path.isfile(join_path_native(osm.module().working_tree_dir, 'setup.py')) + + assert len(rwrepo.submodules) == 2 + + # commit the changes, just to finalize the operation + rwrepo.index.commit("my submod commit") + assert len(rwrepo.submodules) == 2 + + # needs update as the head changed, it thinks its in the history + # of the repo otherwise + nsm.set_parent_commit(rwrepo.head.commit) + osm.set_parent_commit(rwrepo.head.commit) + + # MOVE MODULE + ############# + # invalid inptu + self.failUnlessRaises(ValueError, nsm.move, 'doesntmatter', module=False, configuration=False) + + # renaming to the same path does nothing + assert nsm.move(sm.path) is nsm + + # rename a module + nmp = join_path_native("new", "module", "dir") + "/" # new module path + pmp = nsm.path + abspmp = nsm.abspath + assert nsm.move(nmp) is nsm + nmp = nmp[:-1] # cut last / + nmpl = to_native_path_linux(nmp) + assert nsm.path == nmpl + assert rwrepo.submodules[0].path == nmpl + + mpath = 'newsubmodule' + absmpath = join_path_native(rwrepo.working_tree_dir, mpath) + open(absmpath, 'w').write('') + self.failUnlessRaises(ValueError, nsm.move, mpath) + os.remove(absmpath) + + # now it works, as we just move it back + nsm.move(pmp) + assert nsm.path == pmp + assert rwrepo.submodules[0].path == pmp + + # TODO lowprio: test remaining exceptions ... for now its okay, the code looks right + + # REMOVE 'EM ALL + ################ + # if a submodule's repo has no remotes, it can't be added without an explicit url + osmod = osm.module() + + osm.remove(module=False) + for remote in osmod.remotes: + remote.remove(osmod, remote.name) + assert not osm.exists() + self.failUnlessRaises(ValueError, Submodule.add, rwrepo, osmid, csm_repopath, url=None) + # END handle bare mode + + # Error if there is no submodule file here + self.failUnlessRaises(IOError, Submodule._config_parser, rwrepo, rwrepo.commit(self.k_no_subm_tag), True) + + @with_rw_repo(k_subm_current) + def test_base_rw(self, rwrepo): + self._do_base_tests(rwrepo) + + @with_rw_repo(k_subm_current, bare=True) + def test_base_bare(self, rwrepo): + self._do_base_tests(rwrepo) + + @with_rw_repo(k_subm_current, bare=False) + def test_root_module(self, rwrepo): + # Can query everything without problems + rm = RootModule(rwrepo) + # test new constructor + assert rm.parent_commit == RootModule(self.rorepo, self.rorepo.commit(self.k_subm_current)).parent_commit + assert rm.module() is rwrepo + + # try attributes + rm.binsha + rm.mode + rm.path + assert rm.name == rm.k_root_name + assert rm.parent_commit == self.rorepo.commit(self.k_subm_current) + rm.url + rm.branch + + assert len(rm.list_items(rm.module())) == 1 + rm.config_reader() + rm.config_writer() + + # deep traversal git / async + rsmsp = [sm.path for sm in rm.traverse()] + assert len(rsmsp) == 1 # gitdb only - its not yet uptodate so it has no submodule + + # cannot set the parent commit as root module's path didn't exist + self.failUnlessRaises(ValueError, rm.set_parent_commit, 'HEAD') + + # TEST UPDATE + ############# + # setup commit which remove existing, add new and modify existing submodules + rm = RootModule(rwrepo) + assert len(rm.children()) == 1 + + # modify path without modifying the index entry + # ( which is what the move method would do properly ) + #================================================== + sm = rm.children()[0] + pp = "path/prefix" + fp = join_path_native(pp, sm.path) + prep = sm.path + assert not sm.module_exists() # was never updated after rwrepo's clone + + # assure we clone from a local source + self._rewrite_gitdb_to_local_path(sm) + + # dry-run does nothing + sm.update(recursive=False, dry_run=True, progress=prog) + assert not sm.module_exists() + + sm.update(recursive=False) + assert sm.module_exists() + sm.config_writer().set_value('path', fp) # change path to something with prefix AFTER url change + + # update fails as list_items in such a situations cannot work, as it cannot + # find the entry at the changed path + self.failUnlessRaises(InvalidGitRepositoryError, rm.update, recursive=False) + + # move it properly - doesn't work as it its path currently points to an indexentry + # which doesn't exist ( move it to some path, it doesn't matter here ) + self.failUnlessRaises(InvalidGitRepositoryError, sm.move, pp) + # reset the path(cache) to where it was, now it works + sm.path = prep + sm.move(fp, module=False) # leave it at the old location + + assert not sm.module_exists() + cpathchange = rwrepo.index.commit("changed sm path") # finally we can commit + + # update puts the module into place + rm.update(recursive=False, progress=prog) + sm.set_parent_commit(cpathchange) + assert sm.module_exists() + + # add submodule + #================ + nsmn = "newsubmodule" + nsmp = "submrepo" + async_url = self._generate_async_local_path() + nsm = Submodule.add(rwrepo, nsmn, nsmp, url=async_url) + csmadded = rwrepo.index.commit("Added submodule").hexsha # make sure we don't keep the repo reference + nsm.set_parent_commit(csmadded) + assert nsm.module_exists() + # in our case, the module should not exist, which happens if we update a parent + # repo and a new submodule comes into life + nsm.remove(configuration=False, module=True) + assert not nsm.module_exists() and nsm.exists() + + + # dry-run does nothing + rm.update(recursive=False, dry_run=True, progress=prog) + + # otherwise it will work + rm.update(recursive=False, progress=prog) + assert nsm.module_exists() + + + + # remove submodule - the previous one + #==================================== + sm.set_parent_commit(csmadded) + smp = sm.abspath + assert not sm.remove(module=False).exists() + assert os.path.isdir(smp) # module still exists + csmremoved = rwrepo.index.commit("Removed submodule") + + # an update will remove the module + # not in dry_run + rm.update(recursive=False, dry_run=True) + assert os.path.isdir(smp) + + rm.update(recursive=False) + assert not os.path.isdir(smp) + + + # change url + #============= + # to the first repository, this way we have a fast checkout, and a completely different + # repository at the different url + nsm.set_parent_commit(csmremoved) + nsmurl = os.environ.get(self.env_gitdb_local_path, self.k_github_gitdb_url) + + # Note: We would have liked to have a different url, but we cannot + # provoke this case + assert nsm.url != nsmurl + nsm.config_writer().set_value('url', nsmurl) + csmpathchange = rwrepo.index.commit("changed url") + nsm.set_parent_commit(csmpathchange) + + prev_commit = nsm.module().head.commit + # dry-run does nothing + rm.update(recursive=False, dry_run=True, progress=prog) + assert nsm.module().remotes.origin.url != nsmurl + + rm.update(recursive=False, progress=prog) + assert nsm.module().remotes.origin.url == nsmurl + # head changed, as the remote url and its commit changed + assert prev_commit != nsm.module().head.commit + + # add the submodule's changed commit to the index, which is what the + # user would do + # beforehand, update our instance's binsha with the new one + nsm.binsha = nsm.module().head.commit.binsha + rwrepo.index.add([nsm]) + + # change branch + #================= + # we only have one branch, so we switch to a virtual one, and back + # to the current one to trigger the difference + cur_branch = nsm.branch + nsmm = nsm.module() + prev_commit = nsmm.head.commit + for branch in ("some_virtual_branch", cur_branch.name): + nsm.config_writer().set_value(Submodule.k_head_option, git.Head.to_full_path(branch)) + csmbranchchange = rwrepo.index.commit("changed branch to %s" % branch) + nsm.set_parent_commit(csmbranchchange) + # END for each branch to change + + # Lets remove our tracking branch to simulate some changes + nsmmh = nsmm.head + assert nsmmh.ref.tracking_branch() is None # never set it up until now + assert not nsmmh.is_detached + + #dry run does nothing + rm.update(recursive=False, dry_run=True, progress=prog) + assert nsmmh.ref.tracking_branch() is None + + # the real thing does + rm.update(recursive=False, progress=prog) + + assert nsmmh.ref.tracking_branch() is not None + assert not nsmmh.is_detached + + # recursive update + # ================= + # finally we recursively update a module, just to run the code at least once + # remove the module so that it has more work + assert len(nsm.children()) >= 1 # could include smmap + assert nsm.exists() and nsm.module_exists() and len(nsm.children()) >= 1 + # assure we pull locally only + nsmc = nsm.children()[0] + nsmc.config_writer().set_value('url', async_url) + rm.update(recursive=True, progress=prog, dry_run=True) # just to run the code + rm.update(recursive=True, progress=prog) + + assert len(nsm.children()) >= 2 and nsmc.module_exists() + diff --git a/git/test/objects/test_tree.py b/git/test/objects/test_tree.py index bc8d3f97..6317f4db 100644 --- a/git/test/objects/test_tree.py +++ b/git/test/objects/test_tree.py @@ -7,140 +7,140 @@ from lib import * from git.objects.fun import ( - traverse_tree_recursive, - traverse_trees_recursive - ) + traverse_tree_recursive, + traverse_trees_recursive + ) from git.objects.blob import Blob from git.objects.tree import Tree from cStringIO import StringIO import os class TestTree(TestObjectBase): - - def test_serializable(self): - # tree at the given commit contains a submodule as well - roottree = self.rorepo.tree('6c1faef799095f3990e9970bc2cb10aa0221cf9c') - for item in roottree.traverse(ignore_self=False): - if item.type != Tree.type: - continue - # END skip non-trees - tree = item - # trees have no dict - self.failUnlessRaises(AttributeError, setattr, tree, 'someattr', 1) - - orig_data = tree.data_stream.read() - orig_cache = tree._cache - - stream = StringIO() - tree._serialize(stream) - assert stream.getvalue() == orig_data - - stream.seek(0) - testtree = Tree(self.rorepo, Tree.NULL_BIN_SHA, 0, '') - testtree._deserialize(stream) - assert testtree._cache == orig_cache - - - # TEST CACHE MUTATOR - mod = testtree.cache - self.failUnlessRaises(ValueError, mod.add, "invalid sha", 0, "name") - self.failUnlessRaises(ValueError, mod.add, Tree.NULL_HEX_SHA, 0, "invalid mode") - self.failUnlessRaises(ValueError, mod.add, Tree.NULL_HEX_SHA, tree.mode, "invalid/name") - - # add new item - name = "fake_dir" - mod.add(testtree.NULL_HEX_SHA, tree.mode, name) - assert name in testtree - - # its available in the tree immediately - assert isinstance(testtree[name], Tree) - - # adding it again will not cause multiple of them to be presents - cur_count = len(testtree) - mod.add(testtree.NULL_HEX_SHA, tree.mode, name) - assert len(testtree) == cur_count - - # fails with a different sha - name exists - hexsha = "1"*40 - self.failUnlessRaises(ValueError, mod.add, hexsha, tree.mode, name) - - # force it - replace existing one - mod.add(hexsha, tree.mode, name, force=True) - assert testtree[name].hexsha == hexsha - assert len(testtree) == cur_count - - # unchecked addition always works, even with invalid items - invalid_name = "hi/there" - mod.add_unchecked(hexsha, 0, invalid_name) - assert len(testtree) == cur_count + 1 - - del(mod[invalid_name]) - assert len(testtree) == cur_count - # del again, its fine - del(mod[invalid_name]) - - # have added one item, we are done - mod.set_done() - mod.set_done() # multiple times are okay - - # serialize, its different now - stream = StringIO() - testtree._serialize(stream) - stream.seek(0) - assert stream.getvalue() != orig_data - - # replaces cache, but we make sure of it - del(testtree._cache) - testtree._deserialize(stream) - assert name in testtree - assert invalid_name not in testtree - # END for each item in tree - - def test_traverse(self): - root = self.rorepo.tree('0.1.6') - num_recursive = 0 - all_items = list() - for obj in root.traverse(): - if "/" in obj.path: - num_recursive += 1 - - assert isinstance(obj, (Blob, Tree)) - all_items.append(obj) - # END for each object - assert all_items == root.list_traverse() - - # limit recursion level to 0 - should be same as default iteration - assert all_items - assert 'CHANGES' in root - assert len(list(root)) == len(list(root.traverse(depth=1))) - - # only choose trees - trees_only = lambda i,d: i.type == "tree" - trees = list(root.traverse(predicate = trees_only)) - assert len(trees) == len(list( i for i in root.traverse() if trees_only(i,0) )) - - # test prune - lib_folder = lambda t,d: t.path == "lib" - pruned_trees = list(root.traverse(predicate = trees_only,prune = lib_folder)) - assert len(pruned_trees) < len(trees) - - # trees and blobs - assert len(set(trees)|set(root.trees)) == len(trees) - assert len(set(b for b in root if isinstance(b, Blob)) | set(root.blobs)) == len( root.blobs ) - subitem = trees[0][0] - assert "/" in subitem.path - assert subitem.name == os.path.basename(subitem.path) - - # assure that at some point the traversed paths have a slash in them - found_slash = False - for item in root.traverse(): - assert os.path.isabs(item.abspath) - if '/' in item.path: - found_slash = True - # END check for slash - - # slashes in paths are supported as well - assert root[item.path] == item == root/item.path - # END for each item - assert found_slash + + def test_serializable(self): + # tree at the given commit contains a submodule as well + roottree = self.rorepo.tree('6c1faef799095f3990e9970bc2cb10aa0221cf9c') + for item in roottree.traverse(ignore_self=False): + if item.type != Tree.type: + continue + # END skip non-trees + tree = item + # trees have no dict + self.failUnlessRaises(AttributeError, setattr, tree, 'someattr', 1) + + orig_data = tree.data_stream.read() + orig_cache = tree._cache + + stream = StringIO() + tree._serialize(stream) + assert stream.getvalue() == orig_data + + stream.seek(0) + testtree = Tree(self.rorepo, Tree.NULL_BIN_SHA, 0, '') + testtree._deserialize(stream) + assert testtree._cache == orig_cache + + + # TEST CACHE MUTATOR + mod = testtree.cache + self.failUnlessRaises(ValueError, mod.add, "invalid sha", 0, "name") + self.failUnlessRaises(ValueError, mod.add, Tree.NULL_HEX_SHA, 0, "invalid mode") + self.failUnlessRaises(ValueError, mod.add, Tree.NULL_HEX_SHA, tree.mode, "invalid/name") + + # add new item + name = "fake_dir" + mod.add(testtree.NULL_HEX_SHA, tree.mode, name) + assert name in testtree + + # its available in the tree immediately + assert isinstance(testtree[name], Tree) + + # adding it again will not cause multiple of them to be presents + cur_count = len(testtree) + mod.add(testtree.NULL_HEX_SHA, tree.mode, name) + assert len(testtree) == cur_count + + # fails with a different sha - name exists + hexsha = "1"*40 + self.failUnlessRaises(ValueError, mod.add, hexsha, tree.mode, name) + + # force it - replace existing one + mod.add(hexsha, tree.mode, name, force=True) + assert testtree[name].hexsha == hexsha + assert len(testtree) == cur_count + + # unchecked addition always works, even with invalid items + invalid_name = "hi/there" + mod.add_unchecked(hexsha, 0, invalid_name) + assert len(testtree) == cur_count + 1 + + del(mod[invalid_name]) + assert len(testtree) == cur_count + # del again, its fine + del(mod[invalid_name]) + + # have added one item, we are done + mod.set_done() + mod.set_done() # multiple times are okay + + # serialize, its different now + stream = StringIO() + testtree._serialize(stream) + stream.seek(0) + assert stream.getvalue() != orig_data + + # replaces cache, but we make sure of it + del(testtree._cache) + testtree._deserialize(stream) + assert name in testtree + assert invalid_name not in testtree + # END for each item in tree + + def test_traverse(self): + root = self.rorepo.tree('0.1.6') + num_recursive = 0 + all_items = list() + for obj in root.traverse(): + if "/" in obj.path: + num_recursive += 1 + + assert isinstance(obj, (Blob, Tree)) + all_items.append(obj) + # END for each object + assert all_items == root.list_traverse() + + # limit recursion level to 0 - should be same as default iteration + assert all_items + assert 'CHANGES' in root + assert len(list(root)) == len(list(root.traverse(depth=1))) + + # only choose trees + trees_only = lambda i,d: i.type == "tree" + trees = list(root.traverse(predicate = trees_only)) + assert len(trees) == len(list( i for i in root.traverse() if trees_only(i,0) )) + + # test prune + lib_folder = lambda t,d: t.path == "lib" + pruned_trees = list(root.traverse(predicate = trees_only,prune = lib_folder)) + assert len(pruned_trees) < len(trees) + + # trees and blobs + assert len(set(trees)|set(root.trees)) == len(trees) + assert len(set(b for b in root if isinstance(b, Blob)) | set(root.blobs)) == len( root.blobs ) + subitem = trees[0][0] + assert "/" in subitem.path + assert subitem.name == os.path.basename(subitem.path) + + # assure that at some point the traversed paths have a slash in them + found_slash = False + for item in root.traverse(): + assert os.path.isabs(item.abspath) + if '/' in item.path: + found_slash = True + # END check for slash + + # slashes in paths are supported as well + assert root[item.path] == item == root/item.path + # END for each item + assert found_slash diff --git a/git/test/performance/db/looseodb_impl.py b/git/test/performance/db/looseodb_impl.py index 6d3c1fa6..0c451906 100644 --- a/git/test/performance/db/looseodb_impl.py +++ b/git/test/performance/db/looseodb_impl.py @@ -4,18 +4,18 @@ from git.base import * from git.stream import * from async import ChannelThreadTask from git.util import ( - pool, - bin_to_hex - ) + pool, + bin_to_hex + ) import os import sys from time import time from git.test.lib import ( - GlobalsItemDeletorMetaCls, - make_memory_file, - with_rw_repo - ) + GlobalsItemDeletorMetaCls, + make_memory_file, + with_rw_repo + ) from git.test.performance.lib import TestBigRepoR @@ -23,110 +23,110 @@ from git.test.performance.lib import TestBigRepoR #{ Utilities def read_chunked_stream(stream): - total = 0 - while True: - chunk = stream.read(chunk_size) - total += len(chunk) - if len(chunk) < chunk_size: - break - # END read stream loop - assert total == stream.size - return stream - - + total = 0 + while True: + chunk = stream.read(chunk_size) + total += len(chunk) + if len(chunk) < chunk_size: + break + # END read stream loop + assert total == stream.size + return stream + + class TestStreamReader(ChannelThreadTask): - """Expects input streams and reads them in chunks. It will read one at a time, - requireing a queue chunk of size 1""" - def __init__(self, *args): - super(TestStreamReader, self).__init__(*args) - self.fun = read_chunked_stream - self.max_chunksize = 1 - + """Expects input streams and reads them in chunks. It will read one at a time, + requireing a queue chunk of size 1""" + def __init__(self, *args): + super(TestStreamReader, self).__init__(*args) + self.fun = read_chunked_stream + self.max_chunksize = 1 + #} END utilities class PerfBaseDeletorMetaClass(GlobalsItemDeletorMetaCls): - ModuleToDelete = 'TestLooseDBWPerformanceBase' + ModuleToDelete = 'TestLooseDBWPerformanceBase' class TestLooseDBWPerformanceBase(TestBigRepoR): - __metaclass__ = PerfBaseDeletorMetaClass - - large_data_size_bytes = 1000*1000*10 # some MiB should do it - moderate_data_size_bytes = 1000*1000*1 # just 1 MiB - - #{ Configuration - LooseODBCls = None - #} END configuration - - @classmethod - def setUpAll(cls): - super(TestLooseDBWPerformanceBase, cls).setUpAll() - if cls.LooseODBCls is None: - raise AssertionError("LooseODBCls must be set in subtype") - #END assert configuration - # currently there is no additional configuration - - @with_rw_repo("HEAD") - def test_large_data_streaming(self, rwrepo): - # TODO: This part overlaps with the same file in git.test.performance.test_stream - # It should be shared if possible - objects_path = rwrepo.db_path('') - ldb = self.LooseODBCls(objects_path) - - for randomize in range(2): - desc = (randomize and 'random ') or '' - print >> sys.stderr, "Creating %s data ..." % desc - st = time() - size, stream = make_memory_file(self.large_data_size_bytes, randomize) - elapsed = time() - st - print >> sys.stderr, "Done (in %f s)" % elapsed - - # writing - due to the compression it will seem faster than it is - st = time() - binsha = ldb.store(IStream('blob', size, stream)).binsha - elapsed_add = time() - st - assert ldb.has_object(binsha) - hexsha = bin_to_hex(binsha) - db_file = os.path.join(objects_path, hexsha[:2], hexsha[2:]) - fsize_kib = os.path.getsize(db_file) / 1000 - - - size_kib = size / 1000 - print >> sys.stderr, "%s: Added %i KiB (filesize = %i KiB) of %s data to loose odb in %f s ( %f Write KiB / s)" % (self.LooseODBCls.__name__, size_kib, fsize_kib, desc, elapsed_add, size_kib / elapsed_add) - - # reading all at once - st = time() - ostream = ldb.stream(binsha) - shadata = ostream.read() - elapsed_readall = time() - st - - stream.seek(0) - assert shadata == stream.getvalue() - print >> sys.stderr, "%s: Read %i KiB of %s data at once from loose odb in %f s ( %f Read KiB / s)" % (self.LooseODBCls.__name__, size_kib, desc, elapsed_readall, size_kib / elapsed_readall) - - - # reading in chunks of 1 MiB - cs = 512*1000 - chunks = list() - st = time() - ostream = ldb.stream(binsha) - while True: - data = ostream.read(cs) - chunks.append(data) - if len(data) < cs: - break - # END read in chunks - elapsed_readchunks = time() - st - - stream.seek(0) - assert ''.join(chunks) == stream.getvalue() - - cs_kib = cs / 1000 - print >> sys.stderr, "%s: Read %i KiB of %s data in %i KiB chunks from loose odb in %f s ( %f Read KiB / s)" % (self.LooseODBCls.__name__, size_kib, desc, cs_kib, elapsed_readchunks, size_kib / elapsed_readchunks) - - # del db file so git has something to do - os.remove(db_file) - # END for each randomization factor - + __metaclass__ = PerfBaseDeletorMetaClass + + large_data_size_bytes = 1000*1000*10 # some MiB should do it + moderate_data_size_bytes = 1000*1000*1 # just 1 MiB + + #{ Configuration + LooseODBCls = None + #} END configuration + + @classmethod + def setUpAll(cls): + super(TestLooseDBWPerformanceBase, cls).setUpAll() + if cls.LooseODBCls is None: + raise AssertionError("LooseODBCls must be set in subtype") + #END assert configuration + # currently there is no additional configuration + + @with_rw_repo("HEAD") + def test_large_data_streaming(self, rwrepo): + # TODO: This part overlaps with the same file in git.test.performance.test_stream + # It should be shared if possible + objects_path = rwrepo.db_path('') + ldb = self.LooseODBCls(objects_path) + + for randomize in range(2): + desc = (randomize and 'random ') or '' + print >> sys.stderr, "Creating %s data ..." % desc + st = time() + size, stream = make_memory_file(self.large_data_size_bytes, randomize) + elapsed = time() - st + print >> sys.stderr, "Done (in %f s)" % elapsed + + # writing - due to the compression it will seem faster than it is + st = time() + binsha = ldb.store(IStream('blob', size, stream)).binsha + elapsed_add = time() - st + assert ldb.has_object(binsha) + hexsha = bin_to_hex(binsha) + db_file = os.path.join(objects_path, hexsha[:2], hexsha[2:]) + fsize_kib = os.path.getsize(db_file) / 1000 + + + size_kib = size / 1000 + print >> sys.stderr, "%s: Added %i KiB (filesize = %i KiB) of %s data to loose odb in %f s ( %f Write KiB / s)" % (self.LooseODBCls.__name__, size_kib, fsize_kib, desc, elapsed_add, size_kib / elapsed_add) + + # reading all at once + st = time() + ostream = ldb.stream(binsha) + shadata = ostream.read() + elapsed_readall = time() - st + + stream.seek(0) + assert shadata == stream.getvalue() + print >> sys.stderr, "%s: Read %i KiB of %s data at once from loose odb in %f s ( %f Read KiB / s)" % (self.LooseODBCls.__name__, size_kib, desc, elapsed_readall, size_kib / elapsed_readall) + + + # reading in chunks of 1 MiB + cs = 512*1000 + chunks = list() + st = time() + ostream = ldb.stream(binsha) + while True: + data = ostream.read(cs) + chunks.append(data) + if len(data) < cs: + break + # END read in chunks + elapsed_readchunks = time() - st + + stream.seek(0) + assert ''.join(chunks) == stream.getvalue() + + cs_kib = cs / 1000 + print >> sys.stderr, "%s: Read %i KiB of %s data in %i KiB chunks from loose odb in %f s ( %f Read KiB / s)" % (self.LooseODBCls.__name__, size_kib, desc, cs_kib, elapsed_readchunks, size_kib / elapsed_readchunks) + + # del db file so git has something to do + os.remove(db_file) + # END for each randomization factor + diff --git a/git/test/performance/db/odb_impl.py b/git/test/performance/db/odb_impl.py index 677cf6a8..887604c0 100644 --- a/git/test/performance/db/odb_impl.py +++ b/git/test/performance/db/odb_impl.py @@ -5,68 +5,68 @@ import sys import stat from git.test.performance.lib import ( - TestBigRepoR, - GlobalsItemDeletorMetaCls - ) + TestBigRepoR, + GlobalsItemDeletorMetaCls + ) class PerfBaseDeletorMetaClass(GlobalsItemDeletorMetaCls): - ModuleToDelete = 'TestObjDBPerformanceBase' - + ModuleToDelete = 'TestObjDBPerformanceBase' + class TestObjDBPerformanceBase(TestBigRepoR): - __metaclass__ = PerfBaseDeletorMetaClass - - #{ Configuration - RepoCls = None # to be set by subclass - #} END configuration - - def test_random_access_test(self): - repo = self.rorepo - - # GET COMMITS - st = time() - root_commit = repo.commit(self.head_sha_2k) - commits = list(root_commit.traverse()) - nc = len(commits) - elapsed = time() - st - - print >> sys.stderr, "%s: Retrieved %i commits from ObjectStore in %g s ( %f commits / s )" % (type(repo.odb), nc, elapsed, nc / elapsed) - - # GET TREES - # walk all trees of all commits - st = time() - blobs_per_commit = list() - nt = 0 - for commit in commits: - tree = commit.tree - blobs = list() - for item in tree.traverse(): - nt += 1 - if item.type == 'blob': - blobs.append(item) - # direct access for speed - # END while trees are there for walking - blobs_per_commit.append(blobs) - # END for each commit - elapsed = time() - st - - print >> sys.stderr, "%s: Retrieved %i objects from %i commits in %g s ( %f objects / s )" % (type(repo.odb), nt, len(commits), elapsed, nt / elapsed) - - # GET BLOBS - st = time() - nb = 0 - too_many = 15000 - data_bytes = 0 - for blob_list in blobs_per_commit: - for blob in blob_list: - data_bytes += len(blob.data_stream.read()) - # END for each blobsha - nb += len(blob_list) - if nb > too_many: - break - # END for each bloblist - elapsed = time() - st - - print >> sys.stderr, "%s: Retrieved %i blob (%i KiB) and their data in %g s ( %f blobs / s, %f KiB / s )" % (type(repo.odb), nb, data_bytes/1000, elapsed, nb / elapsed, (data_bytes / 1000) / elapsed) - - + __metaclass__ = PerfBaseDeletorMetaClass + + #{ Configuration + RepoCls = None # to be set by subclass + #} END configuration + + def test_random_access_test(self): + repo = self.rorepo + + # GET COMMITS + st = time() + root_commit = repo.commit(self.head_sha_2k) + commits = list(root_commit.traverse()) + nc = len(commits) + elapsed = time() - st + + print >> sys.stderr, "%s: Retrieved %i commits from ObjectStore in %g s ( %f commits / s )" % (type(repo.odb), nc, elapsed, nc / elapsed) + + # GET TREES + # walk all trees of all commits + st = time() + blobs_per_commit = list() + nt = 0 + for commit in commits: + tree = commit.tree + blobs = list() + for item in tree.traverse(): + nt += 1 + if item.type == 'blob': + blobs.append(item) + # direct access for speed + # END while trees are there for walking + blobs_per_commit.append(blobs) + # END for each commit + elapsed = time() - st + + print >> sys.stderr, "%s: Retrieved %i objects from %i commits in %g s ( %f objects / s )" % (type(repo.odb), nt, len(commits), elapsed, nt / elapsed) + + # GET BLOBS + st = time() + nb = 0 + too_many = 15000 + data_bytes = 0 + for blob_list in blobs_per_commit: + for blob in blob_list: + data_bytes += len(blob.data_stream.read()) + # END for each blobsha + nb += len(blob_list) + if nb > too_many: + break + # END for each bloblist + elapsed = time() - st + + print >> sys.stderr, "%s: Retrieved %i blob (%i KiB) and their data in %g s ( %f blobs / s, %f KiB / s )" % (type(repo.odb), nb, data_bytes/1000, elapsed, nb / elapsed, (data_bytes / 1000) / elapsed) + + diff --git a/git/test/performance/db/packedodb_impl.py b/git/test/performance/db/packedodb_impl.py index b95a8d13..1ca7c8a0 100644 --- a/git/test/performance/db/packedodb_impl.py +++ b/git/test/performance/db/packedodb_impl.py @@ -4,9 +4,9 @@ # the New BSD License: http://www.opensource.org/licenses/bsd-license.php """Performance tests for object store""" from git.test.performance.lib import ( - TestBigRepoR, - GlobalsItemDeletorMetaCls - ) + TestBigRepoR, + GlobalsItemDeletorMetaCls + ) from git.exc import UnsupportedOperation @@ -17,91 +17,91 @@ import random class PerfBaseDeletorMetaClass(GlobalsItemDeletorMetaCls): - ModuleToDelete = 'TestPurePackedODBPerformanceBase' + ModuleToDelete = 'TestPurePackedODBPerformanceBase' class TestPurePackedODBPerformanceBase(TestBigRepoR): - __metaclass__ = PerfBaseDeletorMetaClass - - #{ Configuration - PackedODBCls = None - #} END configuration - - @classmethod - def setUpAll(cls): - super(TestPurePackedODBPerformanceBase, cls).setUpAll() - if cls.PackedODBCls is None: - raise AssertionError("PackedODBCls must be set in subclass") - #END assert configuration - cls.ropdb = cls.PackedODBCls(cls.rorepo.db_path("pack")) - - def test_pack_random_access(self): - pdb = self.ropdb - - # sha lookup - st = time() - sha_list = list(pdb.sha_iter()) - elapsed = time() - st - ns = len(sha_list) - print >> sys.stderr, "PDB: looked up %i shas by index in %f s ( %f shas/s )" % (ns, elapsed, ns / elapsed) - - # sha lookup: best-case and worst case access - pdb_pack_info = pdb._pack_info - # END shuffle shas - st = time() - for sha in sha_list: - pdb_pack_info(sha) - # END for each sha to look up - elapsed = time() - st - - # discard cache - del(pdb._entities) - pdb.entities() - print >> sys.stderr, "PDB: looked up %i sha in %i packs in %f s ( %f shas/s )" % (ns, len(pdb.entities()), elapsed, ns / elapsed) - # END for each random mode - - # query info and streams only - max_items = 10000 # can wait longer when testing memory - for pdb_fun in (pdb.info, pdb.stream): - st = time() - for sha in sha_list[:max_items]: - pdb_fun(sha) - elapsed = time() - st - print >> sys.stderr, "PDB: Obtained %i object %s by sha in %f s ( %f items/s )" % (max_items, pdb_fun.__name__.upper(), elapsed, max_items / elapsed) - # END for each function - - # retrieve stream and read all - max_items = 5000 - pdb_stream = pdb.stream - total_size = 0 - st = time() - for sha in sha_list[:max_items]: - stream = pdb_stream(sha) - stream.read() - total_size += stream.size - elapsed = time() - st - total_kib = total_size / 1000 - print >> sys.stderr, "PDB: Obtained %i streams by sha and read all bytes totallying %i KiB ( %f KiB / s ) in %f s ( %f streams/s )" % (max_items, total_kib, total_kib/elapsed , elapsed, max_items / elapsed) - - def test_correctness(self): - pdb = self.ropdb - # disabled for now as it used to work perfectly, checking big repositories takes a long time - print >> sys.stderr, "Endurance run: verify streaming of objects (crc and sha)" - for crc in range(2): - count = 0 - st = time() - for entity in pdb.entities(): - pack_verify = entity.is_valid_stream - sha_by_index = entity.index().sha - for index in xrange(entity.index().size()): - try: - assert pack_verify(sha_by_index(index), use_crc=crc) - count += 1 - except UnsupportedOperation: - pass - # END ignore old indices - # END for each index - # END for each entity - elapsed = time() - st - print >> sys.stderr, "PDB: verified %i objects (crc=%i) in %f s ( %f objects/s )" % (count, crc, elapsed, count / elapsed) - # END for each verify mode - + __metaclass__ = PerfBaseDeletorMetaClass + + #{ Configuration + PackedODBCls = None + #} END configuration + + @classmethod + def setUpAll(cls): + super(TestPurePackedODBPerformanceBase, cls).setUpAll() + if cls.PackedODBCls is None: + raise AssertionError("PackedODBCls must be set in subclass") + #END assert configuration + cls.ropdb = cls.PackedODBCls(cls.rorepo.db_path("pack")) + + def test_pack_random_access(self): + pdb = self.ropdb + + # sha lookup + st = time() + sha_list = list(pdb.sha_iter()) + elapsed = time() - st + ns = len(sha_list) + print >> sys.stderr, "PDB: looked up %i shas by index in %f s ( %f shas/s )" % (ns, elapsed, ns / elapsed) + + # sha lookup: best-case and worst case access + pdb_pack_info = pdb._pack_info + # END shuffle shas + st = time() + for sha in sha_list: + pdb_pack_info(sha) + # END for each sha to look up + elapsed = time() - st + + # discard cache + del(pdb._entities) + pdb.entities() + print >> sys.stderr, "PDB: looked up %i sha in %i packs in %f s ( %f shas/s )" % (ns, len(pdb.entities()), elapsed, ns / elapsed) + # END for each random mode + + # query info and streams only + max_items = 10000 # can wait longer when testing memory + for pdb_fun in (pdb.info, pdb.stream): + st = time() + for sha in sha_list[:max_items]: + pdb_fun(sha) + elapsed = time() - st + print >> sys.stderr, "PDB: Obtained %i object %s by sha in %f s ( %f items/s )" % (max_items, pdb_fun.__name__.upper(), elapsed, max_items / elapsed) + # END for each function + + # retrieve stream and read all + max_items = 5000 + pdb_stream = pdb.stream + total_size = 0 + st = time() + for sha in sha_list[:max_items]: + stream = pdb_stream(sha) + stream.read() + total_size += stream.size + elapsed = time() - st + total_kib = total_size / 1000 + print >> sys.stderr, "PDB: Obtained %i streams by sha and read all bytes totallying %i KiB ( %f KiB / s ) in %f s ( %f streams/s )" % (max_items, total_kib, total_kib/elapsed , elapsed, max_items / elapsed) + + def test_correctness(self): + pdb = self.ropdb + # disabled for now as it used to work perfectly, checking big repositories takes a long time + print >> sys.stderr, "Endurance run: verify streaming of objects (crc and sha)" + for crc in range(2): + count = 0 + st = time() + for entity in pdb.entities(): + pack_verify = entity.is_valid_stream + sha_by_index = entity.index().sha + for index in xrange(entity.index().size()): + try: + assert pack_verify(sha_by_index(index), use_crc=crc) + count += 1 + except UnsupportedOperation: + pass + # END ignore old indices + # END for each index + # END for each entity + elapsed = time() - st + print >> sys.stderr, "PDB: verified %i objects (crc=%i) in %f s ( %f objects/s )" % (count, crc, elapsed, count / elapsed) + # END for each verify mode + diff --git a/git/test/performance/db/test_looseodb_cmd.py b/git/test/performance/db/test_looseodb_cmd.py index 9738278c..9147eff6 100644 --- a/git/test/performance/db/test_looseodb_cmd.py +++ b/git/test/performance/db/test_looseodb_cmd.py @@ -4,8 +4,8 @@ from looseodb_impl import TestLooseDBWPerformanceBase import sys class TestCmdLooseDB(TestLooseDBWPerformanceBase): - LooseODBCls = CmdCompatibilityGitDB - - def test_info(self): - sys.stderr.write("This test does not check the write performance of the git command as it is implemented in pure python") - + LooseODBCls = CmdCompatibilityGitDB + + def test_info(self): + sys.stderr.write("This test does not check the write performance of the git command as it is implemented in pure python") + diff --git a/git/test/performance/db/test_looseodb_dulwich.py b/git/test/performance/db/test_looseodb_dulwich.py index e123ebf1..174be83d 100644 --- a/git/test/performance/db/test_looseodb_dulwich.py +++ b/git/test/performance/db/test_looseodb_dulwich.py @@ -1,13 +1,13 @@ try: - from git.db.dulwich.complex import DulwichGitODB + from git.db.dulwich.complex import DulwichGitODB except ImportError: - from git.db.py.complex import PureGitODB as DulwichGitODB + from git.db.py.complex import PureGitODB as DulwichGitODB #END handle import from git.test.db.dulwich.lib import DulwichRequiredMetaMixin from looseodb_impl import TestLooseDBWPerformanceBase class TestPureLooseDB(TestLooseDBWPerformanceBase): - __metaclass__ = DulwichRequiredMetaMixin - LooseODBCls = DulwichGitODB - + __metaclass__ = DulwichRequiredMetaMixin + LooseODBCls = DulwichGitODB + diff --git a/git/test/performance/db/test_looseodb_pure.py b/git/test/performance/db/test_looseodb_pure.py index 46f39d5e..bb080612 100644 --- a/git/test/performance/db/test_looseodb_pure.py +++ b/git/test/performance/db/test_looseodb_pure.py @@ -2,5 +2,5 @@ from git.db.py.loose import PureLooseObjectODB from looseodb_impl import TestLooseDBWPerformanceBase class TestPureLooseDB(TestLooseDBWPerformanceBase): - LooseODBCls = PureLooseObjectODB - + LooseODBCls = PureLooseObjectODB + diff --git a/git/test/performance/db/test_looseodb_pygit2.py b/git/test/performance/db/test_looseodb_pygit2.py index 326af9fb..a9661111 100644 --- a/git/test/performance/db/test_looseodb_pygit2.py +++ b/git/test/performance/db/test_looseodb_pygit2.py @@ -1,13 +1,13 @@ try: - from git.db.pygit2.complex import Pygit2GitODB + from git.db.pygit2.complex import Pygit2GitODB except ImportError: - from git.db.py.complex import PureGitODB as Pygit2GitODB + from git.db.py.complex import PureGitODB as Pygit2GitODB #END handle import from git.test.db.pygit2.lib import Pygit2RequiredMetaMixin from looseodb_impl import TestLooseDBWPerformanceBase class TestPureLooseDB(TestLooseDBWPerformanceBase): - __metaclass__ = Pygit2RequiredMetaMixin - LooseODBCls = Pygit2GitODB - + __metaclass__ = Pygit2RequiredMetaMixin + LooseODBCls = Pygit2GitODB + diff --git a/git/test/performance/db/test_odb_cmd.py b/git/test/performance/db/test_odb_cmd.py index acd55cc9..37af34fd 100644 --- a/git/test/performance/db/test_odb_cmd.py +++ b/git/test/performance/db/test_odb_cmd.py @@ -2,5 +2,5 @@ from git.db.complex import CmdCompatibilityGitDB from odb_impl import TestObjDBPerformanceBase class TestCmdDB(TestObjDBPerformanceBase): - RepoCls = CmdCompatibilityGitDB - + RepoCls = CmdCompatibilityGitDB + diff --git a/git/test/performance/db/test_odb_dulwich.py b/git/test/performance/db/test_odb_dulwich.py index 6802483c..33abc88c 100644 --- a/git/test/performance/db/test_odb_dulwich.py +++ b/git/test/performance/db/test_odb_dulwich.py @@ -1,13 +1,13 @@ try: - from git.db.dulwich.complex import DulwichCompatibilityGitDB + from git.db.dulwich.complex import DulwichCompatibilityGitDB except ImportError: - from git.db.complex import PureCompatibilityGitDB as DulwichCompatibilityGitDB + from git.db.complex import PureCompatibilityGitDB as DulwichCompatibilityGitDB #END handle dulwich compatibility from git.test.db.dulwich.lib import DulwichRequiredMetaMixin from odb_impl import TestObjDBPerformanceBase class TestDulwichDB(TestObjDBPerformanceBase): - __metaclass__ = DulwichRequiredMetaMixin - RepoCls = DulwichCompatibilityGitDB - + __metaclass__ = DulwichRequiredMetaMixin + RepoCls = DulwichCompatibilityGitDB + diff --git a/git/test/performance/db/test_odb_pure.py b/git/test/performance/db/test_odb_pure.py index 6ed3585d..93139c57 100644 --- a/git/test/performance/db/test_odb_pure.py +++ b/git/test/performance/db/test_odb_pure.py @@ -2,5 +2,5 @@ from git.db.complex import PureCompatibilityGitDB from odb_impl import TestObjDBPerformanceBase class TestPureDB(TestObjDBPerformanceBase): - RepoCls = PureCompatibilityGitDB - + RepoCls = PureCompatibilityGitDB + diff --git a/git/test/performance/db/test_odb_pygit2.py b/git/test/performance/db/test_odb_pygit2.py index bb7ed8a9..c5911ae3 100644 --- a/git/test/performance/db/test_odb_pygit2.py +++ b/git/test/performance/db/test_odb_pygit2.py @@ -1,13 +1,13 @@ try: - from git.db.pygit2.complex import Pygit2CompatibilityGitDB + from git.db.pygit2.complex import Pygit2CompatibilityGitDB except ImportError: - from git.db.complex import PureCompatibilityGitDB as Pygit2CompatibilityGitDB + from git.db.complex import PureCompatibilityGitDB as Pygit2CompatibilityGitDB #END handle pygit2 compatibility from git.test.db.pygit2.lib import Pygit2RequiredMetaMixin from odb_impl import TestObjDBPerformanceBase class TestPygit2DB(TestObjDBPerformanceBase): - __metaclass__ = Pygit2RequiredMetaMixin - RepoCls = Pygit2CompatibilityGitDB - + __metaclass__ = Pygit2RequiredMetaMixin + RepoCls = Pygit2CompatibilityGitDB + diff --git a/git/test/performance/db/test_packedodb_pure.py b/git/test/performance/db/test_packedodb_pure.py index 11497d9d..90e8381f 100644 --- a/git/test/performance/db/test_packedodb_pure.py +++ b/git/test/performance/db/test_packedodb_pure.py @@ -17,73 +17,73 @@ from nose import SkipTest class CountedNullStream(NullStream): - __slots__ = '_bw' - def __init__(self): - self._bw = 0 - - def bytes_written(self): - return self._bw - - def write(self, d): - self._bw += NullStream.write(self, d) - + __slots__ = '_bw' + def __init__(self): + self._bw = 0 + + def bytes_written(self): + return self._bw + + def write(self, d): + self._bw += NullStream.write(self, d) + class TestPurePackedODB(TestPurePackedODBPerformanceBase): - #{ Configuration - PackedODBCls = PurePackedODB - #} END configuration - - def test_pack_writing_note(self): - sys.stderr.write("test_pack_writing should be adjusted to support different databases to read from - see test for more info") - raise SkipTest() - - def test_pack_writing(self): - # see how fast we can write a pack from object streams. - # This will not be fast, as we take time for decompressing the streams as well - # For now we test the fast streaming and slow streaming versions manually - ostream = CountedNullStream() - # NOTE: We use the same repo twice to see whether OS caching helps - for rorepo in (self.rorepo, self.rorepo, self.ropdb): - - ni = 5000 - count = 0 - total_size = 0 - st = time() - for sha in rorepo.sha_iter(): - count += 1 - rorepo.stream(sha) - if count == ni: - break - #END gather objects for pack-writing - elapsed = time() - st - print >> sys.stderr, "PDB Streaming: Got %i streams from %s by sha in in %f s ( %f streams/s )" % (count, rorepo.__class__.__name__, elapsed, count / elapsed) - - st = time() - PackEntity.write_pack((rorepo.stream(sha) for sha in rorepo.sha_iter()), ostream.write, object_count=ni) - elapsed = time() - st - total_kb = ostream.bytes_written() / 1000 - print >> sys.stderr, "PDB Streaming: Wrote pack of size %i kb in %f s (%f kb/s)" % (total_kb, elapsed, total_kb/elapsed) - #END for each rorepo - - - def test_stream_reading(self): - raise SkipTest("This test was only used for --with-profile runs") - pdb = self.ropdb - - # streaming only, meant for --with-profile runs - ni = 5000 - count = 0 - pdb_stream = pdb.stream - total_size = 0 - st = time() - for sha in pdb.sha_iter(): - if count == ni: - break - stream = pdb_stream(sha) - stream.read() - total_size += stream.size - count += 1 - elapsed = time() - st - total_kib = total_size / 1000 - print >> sys.stderr, "PDB Streaming: Got %i streams by sha and read all bytes totallying %i KiB ( %f KiB / s ) in %f s ( %f streams/s )" % (ni, total_kib, total_kib/elapsed , elapsed, ni / elapsed) - + #{ Configuration + PackedODBCls = PurePackedODB + #} END configuration + + def test_pack_writing_note(self): + sys.stderr.write("test_pack_writing should be adjusted to support different databases to read from - see test for more info") + raise SkipTest() + + def test_pack_writing(self): + # see how fast we can write a pack from object streams. + # This will not be fast, as we take time for decompressing the streams as well + # For now we test the fast streaming and slow streaming versions manually + ostream = CountedNullStream() + # NOTE: We use the same repo twice to see whether OS caching helps + for rorepo in (self.rorepo, self.rorepo, self.ropdb): + + ni = 5000 + count = 0 + total_size = 0 + st = time() + for sha in rorepo.sha_iter(): + count += 1 + rorepo.stream(sha) + if count == ni: + break + #END gather objects for pack-writing + elapsed = time() - st + print >> sys.stderr, "PDB Streaming: Got %i streams from %s by sha in in %f s ( %f streams/s )" % (count, rorepo.__class__.__name__, elapsed, count / elapsed) + + st = time() + PackEntity.write_pack((rorepo.stream(sha) for sha in rorepo.sha_iter()), ostream.write, object_count=ni) + elapsed = time() - st + total_kb = ostream.bytes_written() / 1000 + print >> sys.stderr, "PDB Streaming: Wrote pack of size %i kb in %f s (%f kb/s)" % (total_kb, elapsed, total_kb/elapsed) + #END for each rorepo + + + def test_stream_reading(self): + raise SkipTest("This test was only used for --with-profile runs") + pdb = self.ropdb + + # streaming only, meant for --with-profile runs + ni = 5000 + count = 0 + pdb_stream = pdb.stream + total_size = 0 + st = time() + for sha in pdb.sha_iter(): + if count == ni: + break + stream = pdb_stream(sha) + stream.read() + total_size += stream.size + count += 1 + elapsed = time() - st + total_kib = total_size / 1000 + print >> sys.stderr, "PDB Streaming: Got %i streams by sha and read all bytes totallying %i KiB ( %f KiB / s ) in %f s ( %f streams/s )" % (ni, total_kib, total_kib/elapsed , elapsed, ni / elapsed) + diff --git a/git/test/performance/lib.py b/git/test/performance/lib.py index 758d402d..c82ccb16 100644 --- a/git/test/performance/lib.py +++ b/git/test/performance/lib.py @@ -1,9 +1,9 @@ """Contains library functions""" import os from git.test.lib import ( - TestBase, - GlobalsItemDeletorMetaCls - ) + TestBase, + GlobalsItemDeletorMetaCls + ) import shutil import tempfile @@ -16,12 +16,12 @@ k_env_git_repo = "GIT_PYTHON_TEST_GIT_REPO_BASE" #{ Utilities def resolve_or_fail(env_var): - """:return: resolved environment variable or raise EnvironmentError""" - try: - return os.environ[env_var] - except KeyError: - raise EnvironmentError("Please set the %r envrionment variable and retry" % env_var) - # END exception handling + """:return: resolved environment variable or raise EnvironmentError""" + try: + return os.environ[env_var] + except KeyError: + raise EnvironmentError("Please set the %r envrionment variable and retry" % env_var) + # END exception handling #} END utilities @@ -29,46 +29,46 @@ def resolve_or_fail(env_var): #{ Base Classes class TestBigRepoR(TestBase): - """TestCase providing access to readonly 'big' repositories using the following - member variables: - - * gitrorepo - - * a big read-only git repository + """TestCase providing access to readonly 'big' repositories using the following + member variables: + + * gitrorepo + + * a big read-only git repository """ - - #{ Invariants - head_sha_2k = '235d521da60e4699e5bd59ac658b5b48bd76ddca' - head_sha_50 = '32347c375250fd470973a5d76185cac718955fd5' - #} END invariants - - #{ Configuration - RepoCls = Repo - #} END configuration - - @classmethod - def setUpAll(cls): - super(TestBigRepoR, cls).setUpAll() - if cls.RepoCls is None: - raise AssertionError("Require RepoCls in class %s to be set" % cls) - #END assert configuration - cls.rorepo = cls.RepoCls(resolve_or_fail(k_env_git_repo)) + + #{ Invariants + head_sha_2k = '235d521da60e4699e5bd59ac658b5b48bd76ddca' + head_sha_50 = '32347c375250fd470973a5d76185cac718955fd5' + #} END invariants + + #{ Configuration + RepoCls = Repo + #} END configuration + + @classmethod + def setUpAll(cls): + super(TestBigRepoR, cls).setUpAll() + if cls.RepoCls is None: + raise AssertionError("Require RepoCls in class %s to be set" % cls) + #END assert configuration + cls.rorepo = cls.RepoCls(resolve_or_fail(k_env_git_repo)) class TestBigRepoRW(TestBigRepoR): - """As above, but provides a big repository that we can write to. - - Provides ``self.rwrepo``""" - - @classmethod - def setUpAll(cls): - super(TestBigRepoRW, cls).setUpAll() - dirname = tempfile.mktemp() - os.mkdir(dirname) - cls.rwrepo = cls.rorepo.clone(dirname, shared=True, bare=True) - - @classmethod - def tearDownAll(cls): - shutil.rmtree(cls.rwrepo.working_dir) - + """As above, but provides a big repository that we can write to. + + Provides ``self.rwrepo``""" + + @classmethod + def setUpAll(cls): + super(TestBigRepoRW, cls).setUpAll() + dirname = tempfile.mktemp() + os.mkdir(dirname) + cls.rwrepo = cls.rorepo.clone(dirname, shared=True, bare=True) + + @classmethod + def tearDownAll(cls): + shutil.rmtree(cls.rwrepo.working_dir) + #} END base classes diff --git a/git/test/performance/objects/test_commit.py b/git/test/performance/objects/test_commit.py index 685fba2f..e342e6b3 100644 --- a/git/test/performance/objects/test_commit.py +++ b/git/test/performance/objects/test_commit.py @@ -13,88 +13,88 @@ from time import time import sys class TestPerformance(TestBigRepoRW): - - # ref with about 100 commits in its history - ref_100 = 'v0.99' + + # ref with about 100 commits in its history + ref_100 = 'v0.99' - def _query_commit_info(self, c): - c.author - c.authored_date - c.author_tz_offset - c.committer - c.committed_date - c.committer_tz_offset - c.message - c.parents - - def test_iteration(self): - no = 0 - nc = 0 - - # find the first commit containing the given path - always do a full - # iteration ( restricted to the path in question ), but in fact it should - # return quite a lot of commits, we just take one and hence abort the operation - - st = time() - for c in self.rorepo.iter_commits(self.ref_100): - nc += 1 - self._query_commit_info(c) - for obj in c.tree.traverse(): - obj.size - no += 1 - # END for each object - # END for each commit - elapsed_time = time() - st - assert no, "Should have traversed a few objects" - print >> sys.stderr, "Traversed %i Trees and a total of %i unchached objects in %s [s] ( %f objs/s )" % (nc, no, elapsed_time, no/elapsed_time) - - def test_commit_traversal(self): - # bound to cat-file parsing performance - nc = 0 - st = time() - for c in self.rorepo.commit(self.head_sha_2k).traverse(branch_first=False): - nc += 1 - self._query_commit_info(c) - # END for each traversed commit - elapsed_time = time() - st - print >> sys.stderr, "Traversed %i Commits in %s [s] ( %f commits/s )" % (nc, elapsed_time, nc/elapsed_time) - - def test_commit_iteration(self): - # bound to stream parsing performance - nc = 0 - st = time() - for c in Commit.iter_items(self.rorepo, self.head_sha_2k): - nc += 1 - self._query_commit_info(c) - # END for each traversed commit - elapsed_time = time() - st - print >> sys.stderr, "Iterated %i Commits in %s [s] ( %f commits/s )" % (nc, elapsed_time, nc/elapsed_time) - - def test_commit_serialization(self): - assert_commit_serialization(self.rwrepo, self.head_sha_2k, True) - - rwrepo = self.rwrepo - make_object = rwrepo.store - # direct serialization - deserialization can be tested afterwards - # serialization is probably limited on IO - hc = rwrepo.commit(self.head_sha_2k) - - commits = list() - nc = 5000 - st = time() - for i in xrange(nc): - cm = Commit( rwrepo, Commit.NULL_BIN_SHA, hc.tree, - hc.author, hc.authored_date, hc.author_tz_offset, - hc.committer, hc.committed_date, hc.committer_tz_offset, - str(i), parents=hc.parents, encoding=hc.encoding) - - stream = StringIO() - cm._serialize(stream) - slen = stream.tell() - stream.seek(0) - - cm.binsha = make_object(IStream(Commit.type, slen, stream)).binsha - # END commit creation - elapsed = time() - st - - print >> sys.stderr, "Serialized %i commits to loose objects in %f s ( %f commits / s )" % (nc, elapsed, nc / elapsed) + def _query_commit_info(self, c): + c.author + c.authored_date + c.author_tz_offset + c.committer + c.committed_date + c.committer_tz_offset + c.message + c.parents + + def test_iteration(self): + no = 0 + nc = 0 + + # find the first commit containing the given path - always do a full + # iteration ( restricted to the path in question ), but in fact it should + # return quite a lot of commits, we just take one and hence abort the operation + + st = time() + for c in self.rorepo.iter_commits(self.ref_100): + nc += 1 + self._query_commit_info(c) + for obj in c.tree.traverse(): + obj.size + no += 1 + # END for each object + # END for each commit + elapsed_time = time() - st + assert no, "Should have traversed a few objects" + print >> sys.stderr, "Traversed %i Trees and a total of %i unchached objects in %s [s] ( %f objs/s )" % (nc, no, elapsed_time, no/elapsed_time) + + def test_commit_traversal(self): + # bound to cat-file parsing performance + nc = 0 + st = time() + for c in self.rorepo.commit(self.head_sha_2k).traverse(branch_first=False): + nc += 1 + self._query_commit_info(c) + # END for each traversed commit + elapsed_time = time() - st + print >> sys.stderr, "Traversed %i Commits in %s [s] ( %f commits/s )" % (nc, elapsed_time, nc/elapsed_time) + + def test_commit_iteration(self): + # bound to stream parsing performance + nc = 0 + st = time() + for c in Commit.iter_items(self.rorepo, self.head_sha_2k): + nc += 1 + self._query_commit_info(c) + # END for each traversed commit + elapsed_time = time() - st + print >> sys.stderr, "Iterated %i Commits in %s [s] ( %f commits/s )" % (nc, elapsed_time, nc/elapsed_time) + + def test_commit_serialization(self): + assert_commit_serialization(self.rwrepo, self.head_sha_2k, True) + + rwrepo = self.rwrepo + make_object = rwrepo.store + # direct serialization - deserialization can be tested afterwards + # serialization is probably limited on IO + hc = rwrepo.commit(self.head_sha_2k) + + commits = list() + nc = 5000 + st = time() + for i in xrange(nc): + cm = Commit( rwrepo, Commit.NULL_BIN_SHA, hc.tree, + hc.author, hc.authored_date, hc.author_tz_offset, + hc.committer, hc.committed_date, hc.committer_tz_offset, + str(i), parents=hc.parents, encoding=hc.encoding) + + stream = StringIO() + cm._serialize(stream) + slen = stream.tell() + stream.seek(0) + + cm.binsha = make_object(IStream(Commit.type, slen, stream)).binsha + # END commit creation + elapsed = time() - st + + print >> sys.stderr, "Serialized %i commits to loose objects in %f s ( %f commits / s )" % (nc, elapsed, nc / elapsed) diff --git a/git/test/performance/test_utils.py b/git/test/performance/test_utils.py index 19c1e84a..8637af48 100644 --- a/git/test/performance/test_utils.py +++ b/git/test/performance/test_utils.py @@ -4,171 +4,171 @@ import sys import stat from lib import ( - TestBigRepoR - ) + TestBigRepoR + ) class TestUtilPerformance(TestBigRepoR): - - def test_access(self): - # compare dict vs. slot access - class Slotty(object): - __slots__ = "attr" - def __init__(self): - self.attr = 1 - - class Dicty(object): - def __init__(self): - self.attr = 1 - - class BigSlotty(object): - __slots__ = ('attr', ) + tuple('abcdefghijk') - def __init__(self): - for attr in self.__slots__: - setattr(self, attr, 1) - - class BigDicty(object): - def __init__(self): - for attr in BigSlotty.__slots__: - setattr(self, attr, 1) - - ni = 1000000 - for cls in (Slotty, Dicty, BigSlotty, BigDicty): - cli = cls() - st = time() - for i in xrange(ni): - cli.attr - # END for each access - elapsed = time() - st - print >> sys.stderr, "Accessed %s.attr %i times in %s s ( %f acc / s)" % (cls.__name__, ni, elapsed, ni / elapsed) - # END for each class type - - # check num of sequence-acceses - for cls in (list, tuple): - x = 10 - st = time() - s = cls(range(x)) - for i in xrange(ni): - s[0] - s[1] - s[2] - # END for - elapsed = time() - st - na = ni * 3 - print >> sys.stderr, "Accessed %s[x] %i times in %s s ( %f acc / s)" % (cls.__name__, na, elapsed, na / elapsed) - # END for each sequence - - def test_instantiation(self): - ni = 100000 - max_num_items = 4 - for mni in range(max_num_items+1): - for cls in (tuple, list): - st = time() - for i in xrange(ni): - if mni == 0: - cls() - elif mni == 1: - cls((1,)) - elif mni == 2: - cls((1,2)) - elif mni == 3: - cls((1,2,3)) - elif mni == 4: - cls((1,2,3,4)) - else: - cls(x for x in xrange(mni)) - # END handle empty cls - # END for each item - elapsed = time() - st - print >> sys.stderr, "Created %i %ss of size %i in %f s ( %f inst / s)" % (ni, cls.__name__, mni, elapsed, ni / elapsed) - # END for each type - # END for each item count - - # tuple and tuple direct - st = time() - for i in xrange(ni): - t = (1,2,3,4) - # END for each item - elapsed = time() - st - print >> sys.stderr, "Created %i tuples (1,2,3,4) in %f s ( %f tuples / s)" % (ni, elapsed, ni / elapsed) - - st = time() - for i in xrange(ni): - t = tuple((1,2,3,4)) - # END for each item - elapsed = time() - st - print >> sys.stderr, "Created %i tuples tuple((1,2,3,4)) in %f s ( %f tuples / s)" % (ni, elapsed, ni / elapsed) - - def test_unpacking_vs_indexing(self): - ni = 1000000 - list_items = [1,2,3,4] - tuple_items = (1,2,3,4) - - for sequence in (list_items, tuple_items): - st = time() - for i in xrange(ni): - one, two, three, four = sequence - # END for eac iteration - elapsed = time() - st - print >> sys.stderr, "Unpacked %i %ss of size %i in %f s ( %f acc / s)" % (ni, type(sequence).__name__, len(sequence), elapsed, ni / elapsed) - - st = time() - for i in xrange(ni): - one, two, three, four = sequence[0], sequence[1], sequence[2], sequence[3] - # END for eac iteration - elapsed = time() - st - print >> sys.stderr, "Unpacked %i %ss of size %i individually in %f s ( %f acc / s)" % (ni, type(sequence).__name__, len(sequence), elapsed, ni / elapsed) - - st = time() - for i in xrange(ni): - one, two = sequence[0], sequence[1] - # END for eac iteration - elapsed = time() - st - print >> sys.stderr, "Unpacked %i %ss of size %i individually (2 of 4) in %f s ( %f acc / s)" % (ni, type(sequence).__name__, len(sequence), elapsed, ni / elapsed) - # END for each sequence - - def test_large_list_vs_iteration(self): - # what costs more: alloc/realloc of lists, or the cpu strain of iterators ? - def slow_iter(ni): - for i in xrange(ni): - yield i - # END slow iter - be closer to the real world - - # alloc doesn't play a role here it seems - for ni in (500, 1000, 10000, 20000, 40000): - st = time() - for i in list(xrange(ni)): - i - # END for each item - elapsed = time() - st - print >> sys.stderr, "Iterated %i items from list in %f s ( %f acc / s)" % (ni, elapsed, ni / elapsed) - - st = time() - for i in slow_iter(ni): - i - # END for each item - elapsed = time() - st - print >> sys.stderr, "Iterated %i items from iterator in %f s ( %f acc / s)" % (ni, elapsed, ni / elapsed) - # END for each number of iterations - - def test_type_vs_inst_class(self): - class NewType(object): - pass - - # lets see which way is faster - inst = NewType() - - ni = 1000000 - st = time() - for i in xrange(ni): - inst.__class__() - # END for each item - elapsed = time() - st - print >> sys.stderr, "Created %i items using inst.__class__ in %f s ( %f items / s)" % (ni, elapsed, ni / elapsed) - - st = time() - for i in xrange(ni): - type(inst)() - # END for each item - elapsed = time() - st - print >> sys.stderr, "Created %i items using type(inst)() in %f s ( %f items / s)" % (ni, elapsed, ni / elapsed) + + def test_access(self): + # compare dict vs. slot access + class Slotty(object): + __slots__ = "attr" + def __init__(self): + self.attr = 1 + + class Dicty(object): + def __init__(self): + self.attr = 1 + + class BigSlotty(object): + __slots__ = ('attr', ) + tuple('abcdefghijk') + def __init__(self): + for attr in self.__slots__: + setattr(self, attr, 1) + + class BigDicty(object): + def __init__(self): + for attr in BigSlotty.__slots__: + setattr(self, attr, 1) + + ni = 1000000 + for cls in (Slotty, Dicty, BigSlotty, BigDicty): + cli = cls() + st = time() + for i in xrange(ni): + cli.attr + # END for each access + elapsed = time() - st + print >> sys.stderr, "Accessed %s.attr %i times in %s s ( %f acc / s)" % (cls.__name__, ni, elapsed, ni / elapsed) + # END for each class type + + # check num of sequence-acceses + for cls in (list, tuple): + x = 10 + st = time() + s = cls(range(x)) + for i in xrange(ni): + s[0] + s[1] + s[2] + # END for + elapsed = time() - st + na = ni * 3 + print >> sys.stderr, "Accessed %s[x] %i times in %s s ( %f acc / s)" % (cls.__name__, na, elapsed, na / elapsed) + # END for each sequence + + def test_instantiation(self): + ni = 100000 + max_num_items = 4 + for mni in range(max_num_items+1): + for cls in (tuple, list): + st = time() + for i in xrange(ni): + if mni == 0: + cls() + elif mni == 1: + cls((1,)) + elif mni == 2: + cls((1,2)) + elif mni == 3: + cls((1,2,3)) + elif mni == 4: + cls((1,2,3,4)) + else: + cls(x for x in xrange(mni)) + # END handle empty cls + # END for each item + elapsed = time() - st + print >> sys.stderr, "Created %i %ss of size %i in %f s ( %f inst / s)" % (ni, cls.__name__, mni, elapsed, ni / elapsed) + # END for each type + # END for each item count + + # tuple and tuple direct + st = time() + for i in xrange(ni): + t = (1,2,3,4) + # END for each item + elapsed = time() - st + print >> sys.stderr, "Created %i tuples (1,2,3,4) in %f s ( %f tuples / s)" % (ni, elapsed, ni / elapsed) + + st = time() + for i in xrange(ni): + t = tuple((1,2,3,4)) + # END for each item + elapsed = time() - st + print >> sys.stderr, "Created %i tuples tuple((1,2,3,4)) in %f s ( %f tuples / s)" % (ni, elapsed, ni / elapsed) + + def test_unpacking_vs_indexing(self): + ni = 1000000 + list_items = [1,2,3,4] + tuple_items = (1,2,3,4) + + for sequence in (list_items, tuple_items): + st = time() + for i in xrange(ni): + one, two, three, four = sequence + # END for eac iteration + elapsed = time() - st + print >> sys.stderr, "Unpacked %i %ss of size %i in %f s ( %f acc / s)" % (ni, type(sequence).__name__, len(sequence), elapsed, ni / elapsed) + + st = time() + for i in xrange(ni): + one, two, three, four = sequence[0], sequence[1], sequence[2], sequence[3] + # END for eac iteration + elapsed = time() - st + print >> sys.stderr, "Unpacked %i %ss of size %i individually in %f s ( %f acc / s)" % (ni, type(sequence).__name__, len(sequence), elapsed, ni / elapsed) + + st = time() + for i in xrange(ni): + one, two = sequence[0], sequence[1] + # END for eac iteration + elapsed = time() - st + print >> sys.stderr, "Unpacked %i %ss of size %i individually (2 of 4) in %f s ( %f acc / s)" % (ni, type(sequence).__name__, len(sequence), elapsed, ni / elapsed) + # END for each sequence + + def test_large_list_vs_iteration(self): + # what costs more: alloc/realloc of lists, or the cpu strain of iterators ? + def slow_iter(ni): + for i in xrange(ni): + yield i + # END slow iter - be closer to the real world + + # alloc doesn't play a role here it seems + for ni in (500, 1000, 10000, 20000, 40000): + st = time() + for i in list(xrange(ni)): + i + # END for each item + elapsed = time() - st + print >> sys.stderr, "Iterated %i items from list in %f s ( %f acc / s)" % (ni, elapsed, ni / elapsed) + + st = time() + for i in slow_iter(ni): + i + # END for each item + elapsed = time() - st + print >> sys.stderr, "Iterated %i items from iterator in %f s ( %f acc / s)" % (ni, elapsed, ni / elapsed) + # END for each number of iterations + + def test_type_vs_inst_class(self): + class NewType(object): + pass + + # lets see which way is faster + inst = NewType() + + ni = 1000000 + st = time() + for i in xrange(ni): + inst.__class__() + # END for each item + elapsed = time() - st + print >> sys.stderr, "Created %i items using inst.__class__ in %f s ( %f items / s)" % (ni, elapsed, ni / elapsed) + + st = time() + for i in xrange(ni): + type(inst)() + # END for each item + elapsed = time() - st + print >> sys.stderr, "Created %i items using type(inst)() in %f s ( %f items / s)" % (ni, elapsed, ni / elapsed) diff --git a/git/test/refs/test_reflog.py b/git/test/refs/test_reflog.py index afa00d92..2ac19de9 100644 --- a/git/test/refs/test_reflog.py +++ b/git/test/refs/test_reflog.py @@ -9,92 +9,92 @@ import os class TestRefLog(TestBase): - def test_reflogentry(self): - nullhexsha = IndexObject.NULL_HEX_SHA - hexsha = 'F' * 40 - actor = Actor('name', 'email') - msg = "message" - - self.failUnlessRaises(ValueError, RefLogEntry.new, nullhexsha, hexsha, 'noactor', 0, 0, "") - e = RefLogEntry.new(nullhexsha, hexsha, actor, 0, 1, msg) - - assert e.oldhexsha == nullhexsha - assert e.newhexsha == hexsha - assert e.actor == actor - assert e.time[0] == 0 - assert e.time[1] == 1 - assert e.message == msg - - # check representation (roughly) - assert repr(e).startswith(nullhexsha) - - def test_base(self): - rlp_head = fixture_path('reflog_HEAD') - rlp_master = fixture_path('reflog_master') - tdir = tempfile.mktemp(suffix="test_reflogs") - os.mkdir(tdir) - - rlp_master_ro = RefLog.path(self.rorepo.head) - assert os.path.isfile(rlp_master_ro) - - # simple read - reflog = RefLog.from_file(rlp_master_ro) - assert reflog._path is not None - assert isinstance(reflog, RefLog) - assert len(reflog) - - # iter_entries works with path and with stream - assert len(list(RefLog.iter_entries(open(rlp_master)))) - assert len(list(RefLog.iter_entries(rlp_master))) - - # raise on invalid revlog - # TODO: Try multiple corrupted ones ! - pp = 'reflog_invalid_' - for suffix in ('oldsha', 'newsha', 'email', 'date', 'sep'): - self.failUnlessRaises(ValueError, RefLog.from_file, fixture_path(pp+suffix)) - #END for each invalid file - - # cannot write an uninitialized reflog - self.failUnlessRaises(ValueError, RefLog().write) - - # test serialize and deserialize - results must match exactly - binsha = chr(255)*20 - msg = "my reflog message" - cr = self.rorepo.config_reader() - for rlp in (rlp_head, rlp_master): - reflog = RefLog.from_file(rlp) - tfile = os.path.join(tdir, os.path.basename(rlp)) - reflog.to_file(tfile) - assert reflog.write() is reflog - - # parsed result must match ... - treflog = RefLog.from_file(tfile) - assert treflog == reflog - - # ... as well as each bytes of the written stream - assert open(tfile).read() == open(rlp).read() - - # append an entry - entry = RefLog.append_entry(cr, tfile, IndexObject.NULL_BIN_SHA, binsha, msg) - assert entry.oldhexsha == IndexObject.NULL_HEX_SHA - assert entry.newhexsha == 'f'*40 - assert entry.message == msg - assert RefLog.from_file(tfile)[-1] == entry - - # index entry - # raises on invalid index - self.failUnlessRaises(IndexError, RefLog.entry_at, rlp, 10000) - - # indices can be positive ... - assert isinstance(RefLog.entry_at(rlp, 0), RefLogEntry) - RefLog.entry_at(rlp, 23) - - # ... and negative - for idx in (-1, -24): - RefLog.entry_at(rlp, idx) - #END for each index to read - # END for each reflog - - - # finally remove our temporary data - shutil.rmtree(tdir) + def test_reflogentry(self): + nullhexsha = IndexObject.NULL_HEX_SHA + hexsha = 'F' * 40 + actor = Actor('name', 'email') + msg = "message" + + self.failUnlessRaises(ValueError, RefLogEntry.new, nullhexsha, hexsha, 'noactor', 0, 0, "") + e = RefLogEntry.new(nullhexsha, hexsha, actor, 0, 1, msg) + + assert e.oldhexsha == nullhexsha + assert e.newhexsha == hexsha + assert e.actor == actor + assert e.time[0] == 0 + assert e.time[1] == 1 + assert e.message == msg + + # check representation (roughly) + assert repr(e).startswith(nullhexsha) + + def test_base(self): + rlp_head = fixture_path('reflog_HEAD') + rlp_master = fixture_path('reflog_master') + tdir = tempfile.mktemp(suffix="test_reflogs") + os.mkdir(tdir) + + rlp_master_ro = RefLog.path(self.rorepo.head) + assert os.path.isfile(rlp_master_ro) + + # simple read + reflog = RefLog.from_file(rlp_master_ro) + assert reflog._path is not None + assert isinstance(reflog, RefLog) + assert len(reflog) + + # iter_entries works with path and with stream + assert len(list(RefLog.iter_entries(open(rlp_master)))) + assert len(list(RefLog.iter_entries(rlp_master))) + + # raise on invalid revlog + # TODO: Try multiple corrupted ones ! + pp = 'reflog_invalid_' + for suffix in ('oldsha', 'newsha', 'email', 'date', 'sep'): + self.failUnlessRaises(ValueError, RefLog.from_file, fixture_path(pp+suffix)) + #END for each invalid file + + # cannot write an uninitialized reflog + self.failUnlessRaises(ValueError, RefLog().write) + + # test serialize and deserialize - results must match exactly + binsha = chr(255)*20 + msg = "my reflog message" + cr = self.rorepo.config_reader() + for rlp in (rlp_head, rlp_master): + reflog = RefLog.from_file(rlp) + tfile = os.path.join(tdir, os.path.basename(rlp)) + reflog.to_file(tfile) + assert reflog.write() is reflog + + # parsed result must match ... + treflog = RefLog.from_file(tfile) + assert treflog == reflog + + # ... as well as each bytes of the written stream + assert open(tfile).read() == open(rlp).read() + + # append an entry + entry = RefLog.append_entry(cr, tfile, IndexObject.NULL_BIN_SHA, binsha, msg) + assert entry.oldhexsha == IndexObject.NULL_HEX_SHA + assert entry.newhexsha == 'f'*40 + assert entry.message == msg + assert RefLog.from_file(tfile)[-1] == entry + + # index entry + # raises on invalid index + self.failUnlessRaises(IndexError, RefLog.entry_at, rlp, 10000) + + # indices can be positive ... + assert isinstance(RefLog.entry_at(rlp, 0), RefLogEntry) + RefLog.entry_at(rlp, 23) + + # ... and negative + for idx in (-1, -24): + RefLog.entry_at(rlp, idx) + #END for each index to read + # END for each reflog + + + # finally remove our temporary data + shutil.rmtree(tdir) diff --git a/git/test/refs/test_refs.py b/git/test/refs/test_refs.py index 81be173c..d3716cc4 100644 --- a/git/test/refs/test_refs.py +++ b/git/test/refs/test_refs.py @@ -20,521 +20,521 @@ from nose import SkipTest class TestRefs(TestBase): - def test_from_path(self): - # should be able to create any reference directly - for ref_type in (Reference, Head, TagReference, RemoteReference): - for name in ('rela_name', 'path/rela_name'): - full_path = ref_type.to_full_path(name) - instance = ref_type.from_path(self.rorepo, full_path) - assert isinstance(instance, ref_type) - # END for each name - # END for each type - - # invalid path - self.failUnlessRaises(ValueError, TagReference, self.rorepo, "refs/invalid/tag") - # works without path check - TagReference(self.rorepo, "refs/invalid/tag", check_path=False) - - def test_tag_base(self): - tag_object_refs = list() - for tag in TagReference.list_items(self.rorepo): - assert "refs/tags" in tag.path - assert tag.name - assert isinstance(tag.commit, tag.CommitCls) - if tag.tag is not None: - tag_object_refs.append(tag) - tagobj = tag.tag - # have no dict - self.failUnlessRaises(AttributeError, setattr, tagobj, 'someattr', 1) - assert isinstance(tagobj, TagObject) - assert tagobj.tag == tag.name - assert isinstance(tagobj.tagger, Actor) - assert isinstance(tagobj.tagged_date, int) - assert isinstance(tagobj.tagger_tz_offset, int) - assert tagobj.message - assert tag.object == tagobj - # can't assign the object - self.failUnlessRaises(AttributeError, setattr, tag, 'object', tagobj) - # END if we have a tag object - # END for tag in repo-tags - assert tag_object_refs - assert isinstance(TagReference.list_items(self.rorepo)['0.1.6'], TagReference) - - def test_tags(self): - # tag refs can point to tag objects or to commits - s = set() - ref_count = 0 - for ref in chain(TagReference.list_items(self.rorepo), Head.list_items(self.rorepo)): - ref_count += 1 - assert isinstance(ref, Reference) - assert str(ref) == ref.name - assert repr(ref) - assert ref == ref - assert not ref != ref - s.add(ref) - # END for each ref - assert len(s) == ref_count - assert len(s|s) == ref_count - - @with_rw_repo("0.1.6") - def test_heads(self, rw_repo): - for head in Head.iter_items(rw_repo): - assert head.name - assert head.path - assert "refs/heads" in head.path - prev_object = head.object - cur_object = head.object - assert prev_object == cur_object # represent the same git object - assert prev_object is not cur_object # but are different instances - - writer = head.config_writer() - tv = "testopt" - writer.set_value(tv, 1) - assert writer.get_value(tv) == 1 - del(writer) - assert head.config_reader().get_value(tv) == 1 - head.config_writer().remove_option(tv) - - # after the clone, we might still have a tracking branch setup - head.set_tracking_branch(None) - assert head.tracking_branch() is None - remote_ref = RemoteReference.list_items(rw_repo)[0] - assert head.set_tracking_branch(remote_ref) is head - assert head.tracking_branch() == remote_ref - head.set_tracking_branch(None) - assert head.tracking_branch() is None - # END for each head - - # verify REFLOG gets altered - head = HEAD(rw_repo) - cur_head = head.ref - cur_commit = cur_head.commit - pcommit = cur_head.commit.parents[0].parents[0] - hlog_len = len(head.log()) - blog_len = len(cur_head.log()) - assert head.set_reference(pcommit, 'detached head') is head - # one new log-entry - thlog = head.log() - assert len(thlog) == hlog_len + 1 - assert thlog[-1].oldhexsha == cur_commit.hexsha - assert thlog[-1].newhexsha == pcommit.hexsha - - # the ref didn't change though - assert len(cur_head.log()) == blog_len - - # head changes once again, cur_head doesn't change - head.set_reference(cur_head, 'reattach head') - assert len(head.log()) == hlog_len+2 - assert len(cur_head.log()) == blog_len - - # adjusting the head-ref also adjust the head, so both reflogs are - # altered - cur_head.set_commit(pcommit, 'changing commit') - assert len(cur_head.log()) == blog_len+1 - assert len(head.log()) == hlog_len+3 - - - # with automatic dereferencing - assert head.set_commit(cur_commit, 'change commit once again') is head - assert len(head.log()) == hlog_len+4 - assert len(cur_head.log()) == blog_len+2 - - # a new branch has just a single entry - other_head = Head.create(rw_repo, 'mynewhead', pcommit, logmsg='new head created') - log = other_head.log() - assert len(log) == 1 - assert log[0].oldhexsha == pcommit.NULL_HEX_SHA - assert log[0].newhexsha == pcommit.hexsha - - def test_refs(self): - types_found = set() - for ref in Reference.list_items(self.rorepo): - types_found.add(type(ref)) - assert len(types_found) >= 3 - - def test_is_valid(self): - assert Reference(self.rorepo, 'refs/doesnt/exist').is_valid() == False - assert HEAD(self.rorepo).is_valid() - assert HEAD(self.rorepo).reference.is_valid() - assert SymbolicReference(self.rorepo, 'hellothere').is_valid() == False - - def test_orig_head(self): - assert type(HEAD(self.rorepo).orig_head()) == SymbolicReference - - @with_rw_repo("0.1.6") - def test_head_reset(self, rw_repo): - cur_head = HEAD(rw_repo) - old_head_commit = cur_head.commit - new_head_commit = cur_head.ref.commit.parents[0] - - cur_head.reset(new_head_commit, index=True) # index only - assert cur_head.reference.commit == new_head_commit - - self.failUnlessRaises(ValueError, cur_head.reset, new_head_commit, index=False, working_tree=True) - new_head_commit = new_head_commit.parents[0] - cur_head.reset(new_head_commit, index=True, working_tree=True) # index + wt - assert cur_head.reference.commit == new_head_commit - - # paths - make sure we have something to do - rw_repo.index.reset(old_head_commit.parents[0]) - cur_head.reset(cur_head, paths = "test") - cur_head.reset(new_head_commit, paths = "lib") - # hard resets with paths don't work, its all or nothing - self.failUnlessRaises(GitCommandError, cur_head.reset, new_head_commit, working_tree=True, paths = "lib") - - # we can do a mixed reset, and then checkout from the index though - cur_head.reset(new_head_commit) - rw_repo.index.checkout(["lib"], force=True)# - - # now that we have a write write repo, change the HEAD reference - its - # like git-reset --soft - heads = Head.list_items(rw_repo) - assert heads - for head in heads: - cur_head.reference = head - assert cur_head.reference == head - assert isinstance(cur_head.reference, Head) - assert cur_head.commit == head.commit - assert not cur_head.is_detached - # END for each head - - # detach - active_head = heads[0] - curhead_commit = active_head.commit - cur_head.reference = curhead_commit - assert cur_head.commit == curhead_commit - assert cur_head.is_detached - self.failUnlessRaises(TypeError, getattr, cur_head, "reference") - - # tags are references, hence we can point to them - some_tag = TagReference.list_items(rw_repo)[0] - cur_head.reference = some_tag - assert not cur_head.is_detached - assert cur_head.commit == some_tag.commit - assert isinstance(cur_head.reference, TagReference) - - # put HEAD back to a real head, otherwise everything else fails - cur_head.reference = active_head - - # type check - self.failUnlessRaises(ValueError, setattr, cur_head, "reference", "that") - - # head handling - commit = 'HEAD' - prev_head_commit = cur_head.commit - for count, new_name in enumerate(("my_new_head", "feature/feature1")): - actual_commit = commit+"^"*count - new_head = Head.create(rw_repo, new_name, actual_commit) - assert new_head.is_detached - assert cur_head.commit == prev_head_commit - assert isinstance(new_head, Head) - # already exists, but has the same value, so its fine - Head.create(rw_repo, new_name, new_head.commit) - - # its not fine with a different value - self.failUnlessRaises(OSError, Head.create, rw_repo, new_name, new_head.commit.parents[0]) - - # force it - new_head = Head.create(rw_repo, new_name, actual_commit, force=True) - old_path = new_head.path - old_name = new_head.name - - assert new_head.rename("hello").name == "hello" - assert new_head.rename("hello/world").name == "hello/world" # yes, this must work - assert new_head.rename(old_name).name == old_name and new_head.path == old_path - - # rename with force - tmp_head = Head.create(rw_repo, "tmphead") - self.failUnlessRaises(GitCommandError, tmp_head.rename, new_head) - tmp_head.rename(new_head, force=True) - assert tmp_head == new_head and tmp_head.object == new_head.object - - logfile = RefLog.path(tmp_head) - assert os.path.isfile(logfile) - Head.delete(rw_repo, tmp_head) - # deletion removes the log as well - assert not os.path.isfile(logfile) - heads = rw_repo.heads - assert tmp_head not in heads and new_head not in heads - # force on deletion testing would be missing here, code looks okay though ;) - # END for each new head name - self.failUnlessRaises(TypeError, RemoteReference.create, rw_repo, "some_name") - - # tag ref - tag_name = "1.0.2" - light_tag = TagReference.create(rw_repo, tag_name) - self.failUnlessRaises(GitCommandError, TagReference.create, rw_repo, tag_name) - light_tag = TagReference.create(rw_repo, tag_name, "HEAD~1", force = True) - assert isinstance(light_tag, TagReference) - assert light_tag.name == tag_name - assert light_tag.commit == cur_head.commit.parents[0] - assert light_tag.tag is None - - # tag with tag object - other_tag_name = "releases/1.0.2RC" - msg = "my mighty tag\nsecond line" - obj_tag = TagReference.create(rw_repo, other_tag_name, message=msg) - assert isinstance(obj_tag, TagReference) - assert obj_tag.name == other_tag_name - assert obj_tag.commit == cur_head.commit - assert obj_tag.tag is not None - - TagReference.delete(rw_repo, light_tag, obj_tag) - tags = rw_repo.tags - assert light_tag not in tags and obj_tag not in tags - - # remote deletion - remote_refs_so_far = 0 - remotes = rw_repo.remotes - assert remotes - for remote in remotes: - refs = remote.refs - - # If a HEAD exists, it must be deleted first. Otherwise it might - # end up pointing to an invalid ref it the ref was deleted before. - remote_head_name = "HEAD" - if remote_head_name in refs: - RemoteReference.delete(rw_repo, refs[remote_head_name]) - del(refs[remote_head_name]) - #END handle HEAD deletion - - RemoteReference.delete(rw_repo, *refs) - remote_refs_so_far += len(refs) - for ref in refs: - assert ref.remote_name == remote.name - # END for each ref to delete - assert remote_refs_so_far - - for remote in remotes: - # remotes without references throw - self.failUnlessRaises(AssertionError, getattr, remote, 'refs') - # END for each remote - - # change where the active head points to - if cur_head.is_detached: - cur_head.reference = rw_repo.heads[0] - - head = cur_head.reference - old_commit = head.commit - head.commit = old_commit.parents[0] - assert head.commit == old_commit.parents[0] - assert head.commit == cur_head.commit - head.commit = old_commit - - # setting a non-commit as commit fails, but succeeds as object - head_tree = head.commit.tree - self.failUnlessRaises(ValueError, setattr, head, 'commit', head_tree) - assert head.commit == old_commit # and the ref did not change - # we allow heds to point to any object - head.object = head_tree - assert head.object == head_tree - # cannot query tree as commit - self.failUnlessRaises(TypeError, getattr, head, 'commit') - - # set the commit directly using the head. This would never detach the head - assert not cur_head.is_detached - head.object = old_commit - cur_head.reference = head.commit - assert cur_head.is_detached - parent_commit = head.commit.parents[0] - assert cur_head.is_detached - cur_head.commit = parent_commit - assert cur_head.is_detached and cur_head.commit == parent_commit - - cur_head.reference = head - assert not cur_head.is_detached - cur_head.commit = parent_commit - assert not cur_head.is_detached - assert head.commit == parent_commit - - # test checkout - active_branch = rw_repo.active_branch - for head in rw_repo.heads: - checked_out_head = head.checkout() - assert checked_out_head == head - # END for each head to checkout - - # checkout with branch creation - new_head = active_branch.checkout(b="new_head") - assert active_branch != rw_repo.active_branch - assert new_head == rw_repo.active_branch - - # checkout with force as we have a changed a file - # clear file - open(new_head.commit.tree.blobs[-1].abspath,'w').close() - assert len(new_head.commit.diff(None)) - - # create a new branch that is likely to touch the file we changed - far_away_head = rw_repo.create_head("far_head",'HEAD~100') - self.failUnlessRaises(GitCommandError, far_away_head.checkout) - assert active_branch == active_branch.checkout(force=True) - assert rw_repo.head.reference != far_away_head - - # test reference creation - partial_ref = 'sub/ref' - full_ref = 'refs/%s' % partial_ref - ref = Reference.create(rw_repo, partial_ref) - assert ref.path == full_ref - assert ref.object == rw_repo.head.commit - - self.failUnlessRaises(OSError, Reference.create, rw_repo, full_ref, 'HEAD~20') - # it works if it is at the same spot though and points to the same reference - assert Reference.create(rw_repo, full_ref, 'HEAD').path == full_ref - Reference.delete(rw_repo, full_ref) - - # recreate the reference using a full_ref - ref = Reference.create(rw_repo, full_ref) - assert ref.path == full_ref - assert ref.object == rw_repo.head.commit - - # recreate using force - ref = Reference.create(rw_repo, partial_ref, 'HEAD~1', force=True) - assert ref.path == full_ref - assert ref.object == rw_repo.head.commit.parents[0] - - # rename it - orig_obj = ref.object - for name in ('refs/absname', 'rela_name', 'feature/rela_name'): - ref_new_name = ref.rename(name) - assert isinstance(ref_new_name, Reference) - assert name in ref_new_name.path - assert ref_new_name.object == orig_obj - assert ref_new_name == ref - # END for each name type - - # References that don't exist trigger an error if we want to access them - self.failUnlessRaises(ValueError, getattr, Reference(rw_repo, "refs/doesntexist"), 'commit') - - # exists, fail unless we force - ex_ref_path = far_away_head.path - self.failUnlessRaises(OSError, ref.rename, ex_ref_path) - # if it points to the same commit it works - far_away_head.commit = ref.commit - ref.rename(ex_ref_path) - assert ref.path == ex_ref_path and ref.object == orig_obj - assert ref.rename(ref.path).path == ex_ref_path # rename to same name - - # create symbolic refs - symref_path = "symrefs/sym" - symref = SymbolicReference.create(rw_repo, symref_path, cur_head.reference) - assert symref.path == symref_path - assert symref.reference == cur_head.reference - - self.failUnlessRaises(OSError, SymbolicReference.create, rw_repo, symref_path, cur_head.reference.commit) - # it works if the new ref points to the same reference - SymbolicReference.create(rw_repo, symref.path, symref.reference).path == symref.path - SymbolicReference.delete(rw_repo, symref) - # would raise if the symref wouldn't have been deletedpbl - symref = SymbolicReference.create(rw_repo, symref_path, cur_head.reference) - - # test symbolic references which are not at default locations like HEAD - # or FETCH_HEAD - they may also be at spots in refs of course - symbol_ref_path = "refs/symbol_ref" - symref = SymbolicReference(rw_repo, symbol_ref_path) - assert symref.path == symbol_ref_path - - # set it - symref.reference = new_head - assert symref.reference == new_head - assert os.path.isfile(symref.abspath) - assert symref.commit == new_head.commit - - for name in ('absname','folder/rela_name'): - symref_new_name = symref.rename(name) - assert isinstance(symref_new_name, SymbolicReference) - assert name in symref_new_name.path - assert symref_new_name.reference == new_head - assert symref_new_name == symref - assert not symref.is_detached - # END for each ref - - # create a new non-head ref just to be sure we handle it even if packed - Reference.create(rw_repo, full_ref) - - # test ref listing - assure we have packed refs - rw_repo.git.pack_refs(all=True, prune=True) - heads = rw_repo.heads - assert heads - assert new_head in heads - assert active_branch in heads - assert rw_repo.tags - - # we should be able to iterate all symbolic refs as well - in that case - # we should expect only symbolic references to be returned - for symref in SymbolicReference.iter_items(rw_repo): - assert not symref.is_detached - - # when iterating references, we can get references and symrefs - # when deleting all refs, I'd expect them to be gone ! Even from - # the packed ones - # For this to work, we must not be on any branch - rw_repo.head.reference = rw_repo.head.commit - deleted_refs = set() - for ref in Reference.iter_items(rw_repo): - if ref.is_detached: - ref.delete(rw_repo, ref) - deleted_refs.add(ref) - # END delete ref - # END for each ref to iterate and to delete - assert deleted_refs - - for ref in Reference.iter_items(rw_repo): - if ref.is_detached: - assert ref not in deleted_refs - # END for each ref - - # reattach head - head will not be returned if it is not a symbolic - # ref - rw_repo.head.reference = Head.create(rw_repo, "master") - - # At least the head should still exist - assert os.path.isfile(rw_repo.head.abspath) - refs = list(SymbolicReference.iter_items(rw_repo)) - assert len(refs) == 1 - - - # test creation of new refs from scratch - for path in ("basename", "dir/somename", "dir2/subdir/basename"): - # REFERENCES - ############ - fpath = Reference.to_full_path(path) - ref_fp = Reference.from_path(rw_repo, fpath) - assert not ref_fp.is_valid() - ref = Reference(rw_repo, fpath) - assert ref == ref_fp - - # can be created by assigning a commit - ref.commit = rw_repo.head.commit - assert ref.is_valid() - - # if the assignment raises, the ref doesn't exist - Reference.delete(ref.repo, ref.path) - assert not ref.is_valid() - self.failUnlessRaises(ValueError, setattr, ref, 'commit', "nonsense") - assert not ref.is_valid() - - # I am sure I had my reason to make it a class method at first, but - # now it doesn't make so much sense anymore, want an instance method as well - # See http://byronimo.lighthouseapp.com/projects/51787-gitpython/tickets/27 - Reference.delete(ref.repo, ref.path) - assert not ref.is_valid() - - ref.object = rw_repo.head.commit - assert ref.is_valid() - - Reference.delete(ref.repo, ref.path) - assert not ref.is_valid() - self.failUnlessRaises(ValueError, setattr, ref, 'object', "nonsense") - assert not ref.is_valid() - - # END for each path - - def test_dereference_recursive(self): - # for now, just test the HEAD - assert SymbolicReference.dereference_recursive(self.rorepo, 'HEAD') - - def test_reflog(self): - assert isinstance(Head.list_items(self.rorepo).master.log(), RefLog) - - def test_pure_python_rename(self): - raise SkipTest("Pure python reference renames cannot properly handle refnames which become a directory after rename") + def test_from_path(self): + # should be able to create any reference directly + for ref_type in (Reference, Head, TagReference, RemoteReference): + for name in ('rela_name', 'path/rela_name'): + full_path = ref_type.to_full_path(name) + instance = ref_type.from_path(self.rorepo, full_path) + assert isinstance(instance, ref_type) + # END for each name + # END for each type + + # invalid path + self.failUnlessRaises(ValueError, TagReference, self.rorepo, "refs/invalid/tag") + # works without path check + TagReference(self.rorepo, "refs/invalid/tag", check_path=False) + + def test_tag_base(self): + tag_object_refs = list() + for tag in TagReference.list_items(self.rorepo): + assert "refs/tags" in tag.path + assert tag.name + assert isinstance(tag.commit, tag.CommitCls) + if tag.tag is not None: + tag_object_refs.append(tag) + tagobj = tag.tag + # have no dict + self.failUnlessRaises(AttributeError, setattr, tagobj, 'someattr', 1) + assert isinstance(tagobj, TagObject) + assert tagobj.tag == tag.name + assert isinstance(tagobj.tagger, Actor) + assert isinstance(tagobj.tagged_date, int) + assert isinstance(tagobj.tagger_tz_offset, int) + assert tagobj.message + assert tag.object == tagobj + # can't assign the object + self.failUnlessRaises(AttributeError, setattr, tag, 'object', tagobj) + # END if we have a tag object + # END for tag in repo-tags + assert tag_object_refs + assert isinstance(TagReference.list_items(self.rorepo)['0.1.6'], TagReference) + + def test_tags(self): + # tag refs can point to tag objects or to commits + s = set() + ref_count = 0 + for ref in chain(TagReference.list_items(self.rorepo), Head.list_items(self.rorepo)): + ref_count += 1 + assert isinstance(ref, Reference) + assert str(ref) == ref.name + assert repr(ref) + assert ref == ref + assert not ref != ref + s.add(ref) + # END for each ref + assert len(s) == ref_count + assert len(s|s) == ref_count + + @with_rw_repo("0.1.6") + def test_heads(self, rw_repo): + for head in Head.iter_items(rw_repo): + assert head.name + assert head.path + assert "refs/heads" in head.path + prev_object = head.object + cur_object = head.object + assert prev_object == cur_object # represent the same git object + assert prev_object is not cur_object # but are different instances + + writer = head.config_writer() + tv = "testopt" + writer.set_value(tv, 1) + assert writer.get_value(tv) == 1 + del(writer) + assert head.config_reader().get_value(tv) == 1 + head.config_writer().remove_option(tv) + + # after the clone, we might still have a tracking branch setup + head.set_tracking_branch(None) + assert head.tracking_branch() is None + remote_ref = RemoteReference.list_items(rw_repo)[0] + assert head.set_tracking_branch(remote_ref) is head + assert head.tracking_branch() == remote_ref + head.set_tracking_branch(None) + assert head.tracking_branch() is None + # END for each head + + # verify REFLOG gets altered + head = HEAD(rw_repo) + cur_head = head.ref + cur_commit = cur_head.commit + pcommit = cur_head.commit.parents[0].parents[0] + hlog_len = len(head.log()) + blog_len = len(cur_head.log()) + assert head.set_reference(pcommit, 'detached head') is head + # one new log-entry + thlog = head.log() + assert len(thlog) == hlog_len + 1 + assert thlog[-1].oldhexsha == cur_commit.hexsha + assert thlog[-1].newhexsha == pcommit.hexsha + + # the ref didn't change though + assert len(cur_head.log()) == blog_len + + # head changes once again, cur_head doesn't change + head.set_reference(cur_head, 'reattach head') + assert len(head.log()) == hlog_len+2 + assert len(cur_head.log()) == blog_len + + # adjusting the head-ref also adjust the head, so both reflogs are + # altered + cur_head.set_commit(pcommit, 'changing commit') + assert len(cur_head.log()) == blog_len+1 + assert len(head.log()) == hlog_len+3 + + + # with automatic dereferencing + assert head.set_commit(cur_commit, 'change commit once again') is head + assert len(head.log()) == hlog_len+4 + assert len(cur_head.log()) == blog_len+2 + + # a new branch has just a single entry + other_head = Head.create(rw_repo, 'mynewhead', pcommit, logmsg='new head created') + log = other_head.log() + assert len(log) == 1 + assert log[0].oldhexsha == pcommit.NULL_HEX_SHA + assert log[0].newhexsha == pcommit.hexsha + + def test_refs(self): + types_found = set() + for ref in Reference.list_items(self.rorepo): + types_found.add(type(ref)) + assert len(types_found) >= 3 + + def test_is_valid(self): + assert Reference(self.rorepo, 'refs/doesnt/exist').is_valid() == False + assert HEAD(self.rorepo).is_valid() + assert HEAD(self.rorepo).reference.is_valid() + assert SymbolicReference(self.rorepo, 'hellothere').is_valid() == False + + def test_orig_head(self): + assert type(HEAD(self.rorepo).orig_head()) == SymbolicReference + + @with_rw_repo("0.1.6") + def test_head_reset(self, rw_repo): + cur_head = HEAD(rw_repo) + old_head_commit = cur_head.commit + new_head_commit = cur_head.ref.commit.parents[0] + + cur_head.reset(new_head_commit, index=True) # index only + assert cur_head.reference.commit == new_head_commit + + self.failUnlessRaises(ValueError, cur_head.reset, new_head_commit, index=False, working_tree=True) + new_head_commit = new_head_commit.parents[0] + cur_head.reset(new_head_commit, index=True, working_tree=True) # index + wt + assert cur_head.reference.commit == new_head_commit + + # paths - make sure we have something to do + rw_repo.index.reset(old_head_commit.parents[0]) + cur_head.reset(cur_head, paths = "test") + cur_head.reset(new_head_commit, paths = "lib") + # hard resets with paths don't work, its all or nothing + self.failUnlessRaises(GitCommandError, cur_head.reset, new_head_commit, working_tree=True, paths = "lib") + + # we can do a mixed reset, and then checkout from the index though + cur_head.reset(new_head_commit) + rw_repo.index.checkout(["lib"], force=True)# + + # now that we have a write write repo, change the HEAD reference - its + # like git-reset --soft + heads = Head.list_items(rw_repo) + assert heads + for head in heads: + cur_head.reference = head + assert cur_head.reference == head + assert isinstance(cur_head.reference, Head) + assert cur_head.commit == head.commit + assert not cur_head.is_detached + # END for each head + + # detach + active_head = heads[0] + curhead_commit = active_head.commit + cur_head.reference = curhead_commit + assert cur_head.commit == curhead_commit + assert cur_head.is_detached + self.failUnlessRaises(TypeError, getattr, cur_head, "reference") + + # tags are references, hence we can point to them + some_tag = TagReference.list_items(rw_repo)[0] + cur_head.reference = some_tag + assert not cur_head.is_detached + assert cur_head.commit == some_tag.commit + assert isinstance(cur_head.reference, TagReference) + + # put HEAD back to a real head, otherwise everything else fails + cur_head.reference = active_head + + # type check + self.failUnlessRaises(ValueError, setattr, cur_head, "reference", "that") + + # head handling + commit = 'HEAD' + prev_head_commit = cur_head.commit + for count, new_name in enumerate(("my_new_head", "feature/feature1")): + actual_commit = commit+"^"*count + new_head = Head.create(rw_repo, new_name, actual_commit) + assert new_head.is_detached + assert cur_head.commit == prev_head_commit + assert isinstance(new_head, Head) + # already exists, but has the same value, so its fine + Head.create(rw_repo, new_name, new_head.commit) + + # its not fine with a different value + self.failUnlessRaises(OSError, Head.create, rw_repo, new_name, new_head.commit.parents[0]) + + # force it + new_head = Head.create(rw_repo, new_name, actual_commit, force=True) + old_path = new_head.path + old_name = new_head.name + + assert new_head.rename("hello").name == "hello" + assert new_head.rename("hello/world").name == "hello/world" # yes, this must work + assert new_head.rename(old_name).name == old_name and new_head.path == old_path + + # rename with force + tmp_head = Head.create(rw_repo, "tmphead") + self.failUnlessRaises(GitCommandError, tmp_head.rename, new_head) + tmp_head.rename(new_head, force=True) + assert tmp_head == new_head and tmp_head.object == new_head.object + + logfile = RefLog.path(tmp_head) + assert os.path.isfile(logfile) + Head.delete(rw_repo, tmp_head) + # deletion removes the log as well + assert not os.path.isfile(logfile) + heads = rw_repo.heads + assert tmp_head not in heads and new_head not in heads + # force on deletion testing would be missing here, code looks okay though ;) + # END for each new head name + self.failUnlessRaises(TypeError, RemoteReference.create, rw_repo, "some_name") + + # tag ref + tag_name = "1.0.2" + light_tag = TagReference.create(rw_repo, tag_name) + self.failUnlessRaises(GitCommandError, TagReference.create, rw_repo, tag_name) + light_tag = TagReference.create(rw_repo, tag_name, "HEAD~1", force = True) + assert isinstance(light_tag, TagReference) + assert light_tag.name == tag_name + assert light_tag.commit == cur_head.commit.parents[0] + assert light_tag.tag is None + + # tag with tag object + other_tag_name = "releases/1.0.2RC" + msg = "my mighty tag\nsecond line" + obj_tag = TagReference.create(rw_repo, other_tag_name, message=msg) + assert isinstance(obj_tag, TagReference) + assert obj_tag.name == other_tag_name + assert obj_tag.commit == cur_head.commit + assert obj_tag.tag is not None + + TagReference.delete(rw_repo, light_tag, obj_tag) + tags = rw_repo.tags + assert light_tag not in tags and obj_tag not in tags + + # remote deletion + remote_refs_so_far = 0 + remotes = rw_repo.remotes + assert remotes + for remote in remotes: + refs = remote.refs + + # If a HEAD exists, it must be deleted first. Otherwise it might + # end up pointing to an invalid ref it the ref was deleted before. + remote_head_name = "HEAD" + if remote_head_name in refs: + RemoteReference.delete(rw_repo, refs[remote_head_name]) + del(refs[remote_head_name]) + #END handle HEAD deletion + + RemoteReference.delete(rw_repo, *refs) + remote_refs_so_far += len(refs) + for ref in refs: + assert ref.remote_name == remote.name + # END for each ref to delete + assert remote_refs_so_far + + for remote in remotes: + # remotes without references throw + self.failUnlessRaises(AssertionError, getattr, remote, 'refs') + # END for each remote + + # change where the active head points to + if cur_head.is_detached: + cur_head.reference = rw_repo.heads[0] + + head = cur_head.reference + old_commit = head.commit + head.commit = old_commit.parents[0] + assert head.commit == old_commit.parents[0] + assert head.commit == cur_head.commit + head.commit = old_commit + + # setting a non-commit as commit fails, but succeeds as object + head_tree = head.commit.tree + self.failUnlessRaises(ValueError, setattr, head, 'commit', head_tree) + assert head.commit == old_commit # and the ref did not change + # we allow heds to point to any object + head.object = head_tree + assert head.object == head_tree + # cannot query tree as commit + self.failUnlessRaises(TypeError, getattr, head, 'commit') + + # set the commit directly using the head. This would never detach the head + assert not cur_head.is_detached + head.object = old_commit + cur_head.reference = head.commit + assert cur_head.is_detached + parent_commit = head.commit.parents[0] + assert cur_head.is_detached + cur_head.commit = parent_commit + assert cur_head.is_detached and cur_head.commit == parent_commit + + cur_head.reference = head + assert not cur_head.is_detached + cur_head.commit = parent_commit + assert not cur_head.is_detached + assert head.commit == parent_commit + + # test checkout + active_branch = rw_repo.active_branch + for head in rw_repo.heads: + checked_out_head = head.checkout() + assert checked_out_head == head + # END for each head to checkout + + # checkout with branch creation + new_head = active_branch.checkout(b="new_head") + assert active_branch != rw_repo.active_branch + assert new_head == rw_repo.active_branch + + # checkout with force as we have a changed a file + # clear file + open(new_head.commit.tree.blobs[-1].abspath,'w').close() + assert len(new_head.commit.diff(None)) + + # create a new branch that is likely to touch the file we changed + far_away_head = rw_repo.create_head("far_head",'HEAD~100') + self.failUnlessRaises(GitCommandError, far_away_head.checkout) + assert active_branch == active_branch.checkout(force=True) + assert rw_repo.head.reference != far_away_head + + # test reference creation + partial_ref = 'sub/ref' + full_ref = 'refs/%s' % partial_ref + ref = Reference.create(rw_repo, partial_ref) + assert ref.path == full_ref + assert ref.object == rw_repo.head.commit + + self.failUnlessRaises(OSError, Reference.create, rw_repo, full_ref, 'HEAD~20') + # it works if it is at the same spot though and points to the same reference + assert Reference.create(rw_repo, full_ref, 'HEAD').path == full_ref + Reference.delete(rw_repo, full_ref) + + # recreate the reference using a full_ref + ref = Reference.create(rw_repo, full_ref) + assert ref.path == full_ref + assert ref.object == rw_repo.head.commit + + # recreate using force + ref = Reference.create(rw_repo, partial_ref, 'HEAD~1', force=True) + assert ref.path == full_ref + assert ref.object == rw_repo.head.commit.parents[0] + + # rename it + orig_obj = ref.object + for name in ('refs/absname', 'rela_name', 'feature/rela_name'): + ref_new_name = ref.rename(name) + assert isinstance(ref_new_name, Reference) + assert name in ref_new_name.path + assert ref_new_name.object == orig_obj + assert ref_new_name == ref + # END for each name type + + # References that don't exist trigger an error if we want to access them + self.failUnlessRaises(ValueError, getattr, Reference(rw_repo, "refs/doesntexist"), 'commit') + + # exists, fail unless we force + ex_ref_path = far_away_head.path + self.failUnlessRaises(OSError, ref.rename, ex_ref_path) + # if it points to the same commit it works + far_away_head.commit = ref.commit + ref.rename(ex_ref_path) + assert ref.path == ex_ref_path and ref.object == orig_obj + assert ref.rename(ref.path).path == ex_ref_path # rename to same name + + # create symbolic refs + symref_path = "symrefs/sym" + symref = SymbolicReference.create(rw_repo, symref_path, cur_head.reference) + assert symref.path == symref_path + assert symref.reference == cur_head.reference + + self.failUnlessRaises(OSError, SymbolicReference.create, rw_repo, symref_path, cur_head.reference.commit) + # it works if the new ref points to the same reference + SymbolicReference.create(rw_repo, symref.path, symref.reference).path == symref.path + SymbolicReference.delete(rw_repo, symref) + # would raise if the symref wouldn't have been deletedpbl + symref = SymbolicReference.create(rw_repo, symref_path, cur_head.reference) + + # test symbolic references which are not at default locations like HEAD + # or FETCH_HEAD - they may also be at spots in refs of course + symbol_ref_path = "refs/symbol_ref" + symref = SymbolicReference(rw_repo, symbol_ref_path) + assert symref.path == symbol_ref_path + + # set it + symref.reference = new_head + assert symref.reference == new_head + assert os.path.isfile(symref.abspath) + assert symref.commit == new_head.commit + + for name in ('absname','folder/rela_name'): + symref_new_name = symref.rename(name) + assert isinstance(symref_new_name, SymbolicReference) + assert name in symref_new_name.path + assert symref_new_name.reference == new_head + assert symref_new_name == symref + assert not symref.is_detached + # END for each ref + + # create a new non-head ref just to be sure we handle it even if packed + Reference.create(rw_repo, full_ref) + + # test ref listing - assure we have packed refs + rw_repo.git.pack_refs(all=True, prune=True) + heads = rw_repo.heads + assert heads + assert new_head in heads + assert active_branch in heads + assert rw_repo.tags + + # we should be able to iterate all symbolic refs as well - in that case + # we should expect only symbolic references to be returned + for symref in SymbolicReference.iter_items(rw_repo): + assert not symref.is_detached + + # when iterating references, we can get references and symrefs + # when deleting all refs, I'd expect them to be gone ! Even from + # the packed ones + # For this to work, we must not be on any branch + rw_repo.head.reference = rw_repo.head.commit + deleted_refs = set() + for ref in Reference.iter_items(rw_repo): + if ref.is_detached: + ref.delete(rw_repo, ref) + deleted_refs.add(ref) + # END delete ref + # END for each ref to iterate and to delete + assert deleted_refs + + for ref in Reference.iter_items(rw_repo): + if ref.is_detached: + assert ref not in deleted_refs + # END for each ref + + # reattach head - head will not be returned if it is not a symbolic + # ref + rw_repo.head.reference = Head.create(rw_repo, "master") + + # At least the head should still exist + assert os.path.isfile(rw_repo.head.abspath) + refs = list(SymbolicReference.iter_items(rw_repo)) + assert len(refs) == 1 + + + # test creation of new refs from scratch + for path in ("basename", "dir/somename", "dir2/subdir/basename"): + # REFERENCES + ############ + fpath = Reference.to_full_path(path) + ref_fp = Reference.from_path(rw_repo, fpath) + assert not ref_fp.is_valid() + ref = Reference(rw_repo, fpath) + assert ref == ref_fp + + # can be created by assigning a commit + ref.commit = rw_repo.head.commit + assert ref.is_valid() + + # if the assignment raises, the ref doesn't exist + Reference.delete(ref.repo, ref.path) + assert not ref.is_valid() + self.failUnlessRaises(ValueError, setattr, ref, 'commit', "nonsense") + assert not ref.is_valid() + + # I am sure I had my reason to make it a class method at first, but + # now it doesn't make so much sense anymore, want an instance method as well + # See http://byronimo.lighthouseapp.com/projects/51787-gitpython/tickets/27 + Reference.delete(ref.repo, ref.path) + assert not ref.is_valid() + + ref.object = rw_repo.head.commit + assert ref.is_valid() + + Reference.delete(ref.repo, ref.path) + assert not ref.is_valid() + self.failUnlessRaises(ValueError, setattr, ref, 'object', "nonsense") + assert not ref.is_valid() + + # END for each path + + def test_dereference_recursive(self): + # for now, just test the HEAD + assert SymbolicReference.dereference_recursive(self.rorepo, 'HEAD') + + def test_reflog(self): + assert isinstance(Head.list_items(self.rorepo).master.log(), RefLog) + + def test_pure_python_rename(self): + raise SkipTest("Pure python reference renames cannot properly handle refnames which become a directory after rename") diff --git a/git/test/test_base.py b/git/test/test_base.py index 7488ac6b..67f370d2 100644 --- a/git/test/test_base.py +++ b/git/test/test_base.py @@ -4,20 +4,20 @@ # This module is part of GitPython and is released under # the BSD License: http://www.opensource.org/licenses/bsd-license.php from lib import ( - TestBase, - with_rw_repo, - DummyStream, - DeriveTest, - with_rw_and_rw_remote_repo - ) + TestBase, + with_rw_repo, + DummyStream, + DeriveTest, + with_rw_and_rw_remote_repo + ) import git.objects.base as base from git.objects import ( - Blob, - Tree, - Commit, - TagObject - ) + Blob, + Tree, + Commit, + TagObject + ) import git.refs as refs @@ -29,184 +29,184 @@ import tempfile ################## from git.util import ( - NULL_BIN_SHA - ) + NULL_BIN_SHA + ) from git.typ import str_blob_type from git.base import ( - OInfo, - OPackInfo, - ODeltaPackInfo, - OStream, - OPackStream, - ODeltaPackStream, - IStream, - ) + OInfo, + OPackInfo, + ODeltaPackInfo, + OStream, + OPackStream, + ODeltaPackStream, + IStream, + ) import os class TestBase(TestBase): - - type_tuples = ( ("blob", "8741fc1d09d61f02ffd8cded15ff603eff1ec070", "blob.py"), - ("tree", "3a6a5e3eeed3723c09f1ef0399f81ed6b8d82e79", "directory"), - ("commit", "4251bd59fb8e11e40c40548cba38180a9536118c", None), - ("tag", "e56a60e8e9cd333cfba0140a77cd12b0d9398f10", None) ) - - def test_base_object(self): - # test interface of base object classes - types = (Blob, Tree, Commit, TagObject) - assert len(types) == len(self.type_tuples) - - s = set() - num_objs = 0 - num_index_objs = 0 - for obj_type, (typename, hexsha, path) in zip(types, self.type_tuples): - binsha = hex_to_bin(hexsha) - item = None - if path is None: - item = obj_type(self.rorepo,binsha) - else: - item = obj_type(self.rorepo,binsha, 0, path) - # END handle index objects - num_objs += 1 - assert item.hexsha == hexsha - assert item.type == typename - assert item.size - assert item == item - assert not item != item - assert str(item) == item.hexsha - assert repr(item) - s.add(item) - - if isinstance(item, base.IndexObject): - num_index_objs += 1 - if hasattr(item,'path'): # never runs here - assert not item.path.startswith("/") # must be relative - assert isinstance(item.mode, int) - # END index object check - - # read from stream - data_stream = item.data_stream - data = data_stream.read() - assert data - - tmpfile = os.tmpfile() - assert item == item.stream_data(tmpfile) - tmpfile.seek(0) - assert tmpfile.read() == data - # END stream to file directly - # END for each object type to create - - # each has a unique sha - assert len(s) == num_objs - assert len(s|s) == num_objs - assert num_index_objs == 2 - - def test_get_object_type_by_name(self): - for tname in base.Object.TYPES: - assert base.Object in get_object_type_by_name(tname).mro() - # END for each known type - - self.failUnlessRaises(ValueError, get_object_type_by_name, "doesntexist") + + type_tuples = ( ("blob", "8741fc1d09d61f02ffd8cded15ff603eff1ec070", "blob.py"), + ("tree", "3a6a5e3eeed3723c09f1ef0399f81ed6b8d82e79", "directory"), + ("commit", "4251bd59fb8e11e40c40548cba38180a9536118c", None), + ("tag", "e56a60e8e9cd333cfba0140a77cd12b0d9398f10", None) ) + + def test_base_object(self): + # test interface of base object classes + types = (Blob, Tree, Commit, TagObject) + assert len(types) == len(self.type_tuples) + + s = set() + num_objs = 0 + num_index_objs = 0 + for obj_type, (typename, hexsha, path) in zip(types, self.type_tuples): + binsha = hex_to_bin(hexsha) + item = None + if path is None: + item = obj_type(self.rorepo,binsha) + else: + item = obj_type(self.rorepo,binsha, 0, path) + # END handle index objects + num_objs += 1 + assert item.hexsha == hexsha + assert item.type == typename + assert item.size + assert item == item + assert not item != item + assert str(item) == item.hexsha + assert repr(item) + s.add(item) + + if isinstance(item, base.IndexObject): + num_index_objs += 1 + if hasattr(item,'path'): # never runs here + assert not item.path.startswith("/") # must be relative + assert isinstance(item.mode, int) + # END index object check + + # read from stream + data_stream = item.data_stream + data = data_stream.read() + assert data + + tmpfile = os.tmpfile() + assert item == item.stream_data(tmpfile) + tmpfile.seek(0) + assert tmpfile.read() == data + # END stream to file directly + # END for each object type to create + + # each has a unique sha + assert len(s) == num_objs + assert len(s|s) == num_objs + assert num_index_objs == 2 + + def test_get_object_type_by_name(self): + for tname in base.Object.TYPES: + assert base.Object in get_object_type_by_name(tname).mro() + # END for each known type + + self.failUnlessRaises(ValueError, get_object_type_by_name, "doesntexist") - def test_object_resolution(self): - # objects must be resolved to shas so they compare equal - assert self.rorepo.head.reference.object == self.rorepo.active_branch.object - - @with_rw_repo('HEAD', bare=True) - def test_with_bare_rw_repo(self, bare_rw_repo): - assert bare_rw_repo.config_reader("repository").getboolean("core", "bare") - assert os.path.isfile(os.path.join(bare_rw_repo.git_dir,'HEAD')) - - @with_rw_repo('0.1.6') - def test_with_rw_repo(self, rw_repo): - assert not rw_repo.config_reader("repository").getboolean("core", "bare") - assert os.path.isdir(os.path.join(rw_repo.working_tree_dir,'lib')) - - @with_rw_and_rw_remote_repo('0.1.6') - def test_with_rw_remote_and_rw_repo(self, rw_repo, rw_remote_repo): - assert not rw_repo.config_reader("repository").getboolean("core", "bare") - assert rw_remote_repo.config_reader("repository").getboolean("core", "bare") - assert os.path.isdir(os.path.join(rw_repo.working_tree_dir,'lib')) - - + def test_object_resolution(self): + # objects must be resolved to shas so they compare equal + assert self.rorepo.head.reference.object == self.rorepo.active_branch.object + + @with_rw_repo('HEAD', bare=True) + def test_with_bare_rw_repo(self, bare_rw_repo): + assert bare_rw_repo.config_reader("repository").getboolean("core", "bare") + assert os.path.isfile(os.path.join(bare_rw_repo.git_dir,'HEAD')) + + @with_rw_repo('0.1.6') + def test_with_rw_repo(self, rw_repo): + assert not rw_repo.config_reader("repository").getboolean("core", "bare") + assert os.path.isdir(os.path.join(rw_repo.working_tree_dir,'lib')) + + @with_rw_and_rw_remote_repo('0.1.6') + def test_with_rw_remote_and_rw_repo(self, rw_repo, rw_remote_repo): + assert not rw_repo.config_reader("repository").getboolean("core", "bare") + assert rw_remote_repo.config_reader("repository").getboolean("core", "bare") + assert os.path.isdir(os.path.join(rw_repo.working_tree_dir,'lib')) + + class TestBaseTypes(TestBase): - - def test_streams(self): - # test info - sha = NULL_BIN_SHA - s = 20 - blob_id = 3 - - info = OInfo(sha, str_blob_type, s) - assert info.binsha == sha - assert info.type == str_blob_type - assert info.type_id == blob_id - assert info.size == s - - # test pack info - # provides type_id - pinfo = OPackInfo(0, blob_id, s) - assert pinfo.type == str_blob_type - assert pinfo.type_id == blob_id - assert pinfo.pack_offset == 0 - - dpinfo = ODeltaPackInfo(0, blob_id, s, sha) - assert dpinfo.type == str_blob_type - assert dpinfo.type_id == blob_id - assert dpinfo.delta_info == sha - assert dpinfo.pack_offset == 0 - - - # test ostream - stream = DummyStream() - ostream = OStream(*(info + (stream, ))) - assert ostream.stream is stream - ostream.read(15) - stream._assert() - assert stream.bytes == 15 - ostream.read(20) - assert stream.bytes == 20 - - # test packstream - postream = OPackStream(*(pinfo + (stream, ))) - assert postream.stream is stream - postream.read(10) - stream._assert() - assert stream.bytes == 10 - - # test deltapackstream - dpostream = ODeltaPackStream(*(dpinfo + (stream, ))) - dpostream.stream is stream - dpostream.read(5) - stream._assert() - assert stream.bytes == 5 - - # derive with own args - DeriveTest(sha, str_blob_type, s, stream, 'mine',myarg = 3)._assert() - - # test istream - istream = IStream(str_blob_type, s, stream) - assert istream.binsha == None - istream.binsha = sha - assert istream.binsha == sha - - assert len(istream.binsha) == 20 - assert len(istream.hexsha) == 40 - - assert istream.size == s - istream.size = s * 2 - istream.size == s * 2 - assert istream.type == str_blob_type - istream.type = "something" - assert istream.type == "something" - assert istream.stream is stream - istream.stream = None - assert istream.stream is None - - assert istream.error is None - istream.error = Exception() - assert isinstance(istream.error, Exception) + + def test_streams(self): + # test info + sha = NULL_BIN_SHA + s = 20 + blob_id = 3 + + info = OInfo(sha, str_blob_type, s) + assert info.binsha == sha + assert info.type == str_blob_type + assert info.type_id == blob_id + assert info.size == s + + # test pack info + # provides type_id + pinfo = OPackInfo(0, blob_id, s) + assert pinfo.type == str_blob_type + assert pinfo.type_id == blob_id + assert pinfo.pack_offset == 0 + + dpinfo = ODeltaPackInfo(0, blob_id, s, sha) + assert dpinfo.type == str_blob_type + assert dpinfo.type_id == blob_id + assert dpinfo.delta_info == sha + assert dpinfo.pack_offset == 0 + + + # test ostream + stream = DummyStream() + ostream = OStream(*(info + (stream, ))) + assert ostream.stream is stream + ostream.read(15) + stream._assert() + assert stream.bytes == 15 + ostream.read(20) + assert stream.bytes == 20 + + # test packstream + postream = OPackStream(*(pinfo + (stream, ))) + assert postream.stream is stream + postream.read(10) + stream._assert() + assert stream.bytes == 10 + + # test deltapackstream + dpostream = ODeltaPackStream(*(dpinfo + (stream, ))) + dpostream.stream is stream + dpostream.read(5) + stream._assert() + assert stream.bytes == 5 + + # derive with own args + DeriveTest(sha, str_blob_type, s, stream, 'mine',myarg = 3)._assert() + + # test istream + istream = IStream(str_blob_type, s, stream) + assert istream.binsha == None + istream.binsha = sha + assert istream.binsha == sha + + assert len(istream.binsha) == 20 + assert len(istream.hexsha) == 40 + + assert istream.size == s + istream.size = s * 2 + istream.size == s * 2 + assert istream.type == str_blob_type + istream.type = "something" + assert istream.type == "something" + assert istream.stream is stream + istream.stream = None + assert istream.stream is None + + assert istream.error is None + istream.error = Exception() + assert isinstance(istream.error, Exception) diff --git a/git/test/test_cmd.py b/git/test/test_cmd.py index b5732339..983277be 100644 --- a/git/test/test_cmd.py +++ b/git/test/test_cmd.py @@ -6,127 +6,127 @@ import os, sys from git.test.lib import ( - TestBase, - patch_object, - raises, - assert_equal, - assert_true, - assert_match, - fixture_path - ) + TestBase, + patch_object, + raises, + assert_equal, + assert_true, + assert_match, + fixture_path + ) from git import Git, GitCommandError class TestGit(TestBase): - - @classmethod - def setUpAll(cls): - super(TestGit, cls).setUpAll() - cls.git = Git(cls.rorepo.working_dir) + + @classmethod + def setUpAll(cls): + super(TestGit, cls).setUpAll() + cls.git = Git(cls.rorepo.working_dir) - @patch_object(Git, 'execute') - def test_call_process_calls_execute(self, git): - git.return_value = '' - self.git.version() - assert_true(git.called) - assert_equal(git.call_args, ((['git', 'version'],), {})) + @patch_object(Git, 'execute') + def test_call_process_calls_execute(self, git): + git.return_value = '' + self.git.version() + assert_true(git.called) + assert_equal(git.call_args, ((['git', 'version'],), {})) - @raises(GitCommandError) - def test_it_raises_errors(self): - self.git.this_does_not_exist() + @raises(GitCommandError) + def test_it_raises_errors(self): + self.git.this_does_not_exist() - def test_it_transforms_kwargs_into_git_command_arguments(self): - assert_equal(["-s"], self.git.transform_kwargs(**{'s': True})) - assert_equal(["-s5"], self.git.transform_kwargs(**{'s': 5})) + def test_it_transforms_kwargs_into_git_command_arguments(self): + assert_equal(["-s"], self.git.transform_kwargs(**{'s': True})) + assert_equal(["-s5"], self.git.transform_kwargs(**{'s': 5})) - assert_equal(["--max-count"], self.git.transform_kwargs(**{'max_count': True})) - assert_equal(["--max-count=5"], self.git.transform_kwargs(**{'max_count': 5})) + assert_equal(["--max-count"], self.git.transform_kwargs(**{'max_count': True})) + assert_equal(["--max-count=5"], self.git.transform_kwargs(**{'max_count': 5})) - assert_equal(["-s", "-t"], self.git.transform_kwargs(**{'s': True, 't': True})) + assert_equal(["-s", "-t"], self.git.transform_kwargs(**{'s': True, 't': True})) - def test_it_executes_git_to_shell_and_returns_result(self): - assert_match('^git version [\d\.]{2}.*$', self.git.execute(["git","version"])) + def test_it_executes_git_to_shell_and_returns_result(self): + assert_match('^git version [\d\.]{2}.*$', self.git.execute(["git","version"])) - def test_it_accepts_stdin(self): - filename = fixture_path("cat_file_blob") - fh = open(filename, 'r') - assert_equal("70c379b63ffa0795fdbfbc128e5a2818397b7ef8", - self.git.hash_object(istream=fh, stdin=True)) - fh.close() + def test_it_accepts_stdin(self): + filename = fixture_path("cat_file_blob") + fh = open(filename, 'r') + assert_equal("70c379b63ffa0795fdbfbc128e5a2818397b7ef8", + self.git.hash_object(istream=fh, stdin=True)) + fh.close() - @patch_object(Git, 'execute') - def test_it_ignores_false_kwargs(self, git): - # this_should_not_be_ignored=False implies it *should* be ignored - output = self.git.version(pass_this_kwarg=False) - assert_true("pass_this_kwarg" not in git.call_args[1]) - - def test_persistent_cat_file_command(self): - # read header only - import subprocess as sp - hexsha = "b2339455342180c7cc1e9bba3e9f181f7baa5167" - g = self.git.cat_file(batch_check=True, istream=sp.PIPE,as_process=True) - g.stdin.write("b2339455342180c7cc1e9bba3e9f181f7baa5167\n") - g.stdin.flush() - obj_info = g.stdout.readline() - - # read header + data - g = self.git.cat_file(batch=True, istream=sp.PIPE,as_process=True) - g.stdin.write("b2339455342180c7cc1e9bba3e9f181f7baa5167\n") - g.stdin.flush() - obj_info_two = g.stdout.readline() - assert obj_info == obj_info_two - - # read data - have to read it in one large chunk - size = int(obj_info.split()[2]) - data = g.stdout.read(size) - terminating_newline = g.stdout.read(1) - - # now we should be able to read a new object - g.stdin.write("b2339455342180c7cc1e9bba3e9f181f7baa5167\n") - g.stdin.flush() - assert g.stdout.readline() == obj_info - - - # same can be achived using the respective command functions - hexsha, typename, size = self.git.get_object_header(hexsha) - hexsha, typename_two, size_two, data = self.git.get_object_data(hexsha) - assert typename == typename_two and size == size_two - - def test_version(self): - v = self.git.version_info - assert isinstance(v, tuple) - for n in v: - assert isinstance(n, int) - #END verify number types - - def test_cmd_override(self): - prev_cmd = self.git.GIT_PYTHON_GIT_EXECUTABLE - try: - # set it to something that doens't exist, assure it raises - type(self.git).GIT_PYTHON_GIT_EXECUTABLE = os.path.join("some", "path", "which", "doesn't", "exist", "gitbinary") - self.failUnlessRaises(OSError, self.git.version) - finally: - type(self.git).GIT_PYTHON_GIT_EXECUTABLE = prev_cmd - #END undo adjustment - - def test_output_strip(self): - import subprocess as sp - hexsha = "b2339455342180c7cc1e9bba3e9f181f7baa5167" + @patch_object(Git, 'execute') + def test_it_ignores_false_kwargs(self, git): + # this_should_not_be_ignored=False implies it *should* be ignored + output = self.git.version(pass_this_kwarg=False) + assert_true("pass_this_kwarg" not in git.call_args[1]) + + def test_persistent_cat_file_command(self): + # read header only + import subprocess as sp + hexsha = "b2339455342180c7cc1e9bba3e9f181f7baa5167" + g = self.git.cat_file(batch_check=True, istream=sp.PIPE,as_process=True) + g.stdin.write("b2339455342180c7cc1e9bba3e9f181f7baa5167\n") + g.stdin.flush() + obj_info = g.stdout.readline() + + # read header + data + g = self.git.cat_file(batch=True, istream=sp.PIPE,as_process=True) + g.stdin.write("b2339455342180c7cc1e9bba3e9f181f7baa5167\n") + g.stdin.flush() + obj_info_two = g.stdout.readline() + assert obj_info == obj_info_two + + # read data - have to read it in one large chunk + size = int(obj_info.split()[2]) + data = g.stdout.read(size) + terminating_newline = g.stdout.read(1) + + # now we should be able to read a new object + g.stdin.write("b2339455342180c7cc1e9bba3e9f181f7baa5167\n") + g.stdin.flush() + assert g.stdout.readline() == obj_info + + + # same can be achived using the respective command functions + hexsha, typename, size = self.git.get_object_header(hexsha) + hexsha, typename_two, size_two, data = self.git.get_object_data(hexsha) + assert typename == typename_two and size == size_two + + def test_version(self): + v = self.git.version_info + assert isinstance(v, tuple) + for n in v: + assert isinstance(n, int) + #END verify number types + + def test_cmd_override(self): + prev_cmd = self.git.GIT_PYTHON_GIT_EXECUTABLE + try: + # set it to something that doens't exist, assure it raises + type(self.git).GIT_PYTHON_GIT_EXECUTABLE = os.path.join("some", "path", "which", "doesn't", "exist", "gitbinary") + self.failUnlessRaises(OSError, self.git.version) + finally: + type(self.git).GIT_PYTHON_GIT_EXECUTABLE = prev_cmd + #END undo adjustment + + def test_output_strip(self): + import subprocess as sp + hexsha = "b2339455342180c7cc1e9bba3e9f181f7baa5167" - # Verify that a trailing newline is stripped from the output of a git - # command. - content = self.git.cat_file('blob', hexsha) - g = self.git.hash_object(istream=sp.PIPE, as_process=True, stdin=True) - g.stdin.write(content) - g.stdin.close() - newsha = g.stdout.readline().strip() - self.assertNotEquals(newsha, hexsha) + # Verify that a trailing newline is stripped from the output of a git + # command. + content = self.git.cat_file('blob', hexsha) + g = self.git.hash_object(istream=sp.PIPE, as_process=True, stdin=True) + g.stdin.write(content) + g.stdin.close() + newsha = g.stdout.readline().strip() + self.assertNotEquals(newsha, hexsha) - # Verify that output of a git command which ends with an empty - # line is not modified when the output_strip flag is cleared. - content = self.git.cat_file('blob', hexsha, output_strip=False) - g = self.git.hash_object(istream=sp.PIPE, as_process=True, stdin=True) - g.stdin.write(content) - g.stdin.close() - newsha = g.stdout.readline().strip() - self.assertEquals(newsha, hexsha) + # Verify that output of a git command which ends with an empty + # line is not modified when the output_strip flag is cleared. + content = self.git.cat_file('blob', hexsha, output_strip=False) + g = self.git.hash_object(istream=sp.PIPE, as_process=True, stdin=True) + g.stdin.write(content) + g.stdin.close() + newsha = g.stdout.readline().strip() + self.assertEquals(newsha, hexsha) diff --git a/git/test/test_diff.py b/git/test/test_diff.py index 55ad3379..98e72d6c 100644 --- a/git/test/test_diff.py +++ b/git/test/test_diff.py @@ -5,132 +5,132 @@ # the BSD License: http://www.opensource.org/licenses/bsd-license.php from git.test.lib import ( - TestBase, - StringProcessAdapter, - fixture, - assert_equal, - assert_true - ) + TestBase, + StringProcessAdapter, + fixture, + assert_equal, + assert_true + ) from git.diff import * class TestDiff(TestBase): - - def _assert_diff_format(self, diffs): - # verify that the format of the diff is sane - for diff in diffs: - if diff.a_mode: - assert isinstance(diff.a_mode, int) - if diff.b_mode: - assert isinstance(diff.b_mode, int) - - if diff.a_blob: - assert not diff.a_blob.path.endswith('\n') - if diff.b_blob: - assert not diff.b_blob.path.endswith('\n') - # END for each diff - return diffs - - def test_list_from_string_new_mode(self): - output = StringProcessAdapter(fixture('diff_new_mode')) - diffs = Diff._index_from_patch_format(self.rorepo, output.stdout) - self._assert_diff_format(diffs) - - assert_equal(1, len(diffs)) - assert_equal(10, len(diffs[0].diff.splitlines())) + + def _assert_diff_format(self, diffs): + # verify that the format of the diff is sane + for diff in diffs: + if diff.a_mode: + assert isinstance(diff.a_mode, int) + if diff.b_mode: + assert isinstance(diff.b_mode, int) + + if diff.a_blob: + assert not diff.a_blob.path.endswith('\n') + if diff.b_blob: + assert not diff.b_blob.path.endswith('\n') + # END for each diff + return diffs + + def test_list_from_string_new_mode(self): + output = StringProcessAdapter(fixture('diff_new_mode')) + diffs = Diff._index_from_patch_format(self.rorepo, output.stdout) + self._assert_diff_format(diffs) + + assert_equal(1, len(diffs)) + assert_equal(10, len(diffs[0].diff.splitlines())) - def test_diff_with_rename(self): - output = StringProcessAdapter(fixture('diff_rename')) - diffs = Diff._index_from_patch_format(self.rorepo, output.stdout) - self._assert_diff_format(diffs) - - assert_equal(1, len(diffs)) + def test_diff_with_rename(self): + output = StringProcessAdapter(fixture('diff_rename')) + diffs = Diff._index_from_patch_format(self.rorepo, output.stdout) + self._assert_diff_format(diffs) + + assert_equal(1, len(diffs)) - diff = diffs[0] - assert_true(diff.renamed) - assert_equal(diff.rename_from, 'AUTHORS') - assert_equal(diff.rename_to, 'CONTRIBUTORS') + diff = diffs[0] + assert_true(diff.renamed) + assert_equal(diff.rename_from, 'AUTHORS') + assert_equal(diff.rename_to, 'CONTRIBUTORS') - def test_diff_with_rename_raw(self): - output = StringProcessAdapter(fixture('diff_rename_raw')) - diffs = Diff._index_from_raw_format(self.rorepo, output.stdout) - self._assert_diff_format(diffs) + def test_diff_with_rename_raw(self): + output = StringProcessAdapter(fixture('diff_rename_raw')) + diffs = Diff._index_from_raw_format(self.rorepo, output.stdout) + self._assert_diff_format(diffs) - diffs = filter(lambda d: d.renamed, diffs) - assert_equal(3, len(diffs)) + diffs = filter(lambda d: d.renamed, diffs) + assert_equal(3, len(diffs)) - diff = diffs[0] - assert_true(diff.renamed) - assert_equal(diff.rename_from, 'git/test/test_reflog.py') - assert_equal(diff.rename_to, 'git/test/refs/test_reflog.py') + diff = diffs[0] + assert_true(diff.renamed) + assert_equal(diff.rename_from, 'git/test/test_reflog.py') + assert_equal(diff.rename_to, 'git/test/refs/test_reflog.py') - def test_diff_patch_format(self): - # test all of the 'old' format diffs for completness - it should at least - # be able to deal with it - fixtures = ("diff_2", "diff_2f", "diff_f", "diff_i", "diff_mode_only", - "diff_new_mode", "diff_numstat", "diff_p", "diff_rename", - "diff_tree_numstat_root" ) - - for fixture_name in fixtures: - diff_proc = StringProcessAdapter(fixture(fixture_name)) - diffs = Diff._index_from_patch_format(self.rorepo, diff_proc.stdout) - # END for each fixture + def test_diff_patch_format(self): + # test all of the 'old' format diffs for completness - it should at least + # be able to deal with it + fixtures = ("diff_2", "diff_2f", "diff_f", "diff_i", "diff_mode_only", + "diff_new_mode", "diff_numstat", "diff_p", "diff_rename", + "diff_tree_numstat_root" ) + + for fixture_name in fixtures: + diff_proc = StringProcessAdapter(fixture(fixture_name)) + diffs = Diff._index_from_patch_format(self.rorepo, diff_proc.stdout) + # END for each fixture - def test_diff_interface(self): - # test a few variations of the main diff routine - assertion_map = dict() - for i, commit in enumerate(self.rorepo.iter_commits('0.1.6', max_count=2)): - diff_item = commit - if i%2 == 0: - diff_item = commit.tree - # END use tree every second item - - for other in (None, commit.Index, commit.parents[0]): - for paths in (None, "CHANGES", ("CHANGES", "lib")): - for create_patch in range(2): - diff_index = diff_item.diff(other, paths, create_patch) - assert isinstance(diff_index, DiffIndex) - - if diff_index: - self._assert_diff_format(diff_index) - for ct in DiffIndex.change_type: - key = 'ct_%s'%ct - assertion_map.setdefault(key, 0) - assertion_map[key] = assertion_map[key]+len(list(diff_index.iter_change_type(ct))) - # END for each changetype - - # check entries - diff_set = set() - diff_set.add(diff_index[0]) - diff_set.add(diff_index[0]) - assert len(diff_set) == 1 - assert diff_index[0] == diff_index[0] - assert not (diff_index[0] != diff_index[0]) - # END diff index checking - # END for each patch option - # END for each path option - # END for each other side - # END for each commit + def test_diff_interface(self): + # test a few variations of the main diff routine + assertion_map = dict() + for i, commit in enumerate(self.rorepo.iter_commits('0.1.6', max_count=2)): + diff_item = commit + if i%2 == 0: + diff_item = commit.tree + # END use tree every second item + + for other in (None, commit.Index, commit.parents[0]): + for paths in (None, "CHANGES", ("CHANGES", "lib")): + for create_patch in range(2): + diff_index = diff_item.diff(other, paths, create_patch) + assert isinstance(diff_index, DiffIndex) + + if diff_index: + self._assert_diff_format(diff_index) + for ct in DiffIndex.change_type: + key = 'ct_%s'%ct + assertion_map.setdefault(key, 0) + assertion_map[key] = assertion_map[key]+len(list(diff_index.iter_change_type(ct))) + # END for each changetype + + # check entries + diff_set = set() + diff_set.add(diff_index[0]) + diff_set.add(diff_index[0]) + assert len(diff_set) == 1 + assert diff_index[0] == diff_index[0] + assert not (diff_index[0] != diff_index[0]) + # END diff index checking + # END for each patch option + # END for each path option + # END for each other side + # END for each commit - # test rename detection - rename_commit = self.rorepo.rev_parse('4772fe0') - rename_diffs = rename_commit.parents[0].diff(rename_commit, M=True) - rename_diffs = filter(lambda d: d.renamed, rename_diffs) - assert len(rename_diffs) == 3 - assert rename_diffs[0].rename_from == rename_diffs[0].a_blob.path - assert rename_diffs[0].rename_to == rename_diffs[0].b_blob.path - - # assert we could always find at least one instance of the members we - # can iterate in the diff index - if not this indicates its not working correctly - # or our test does not span the whole range of possibilities - for key,value in assertion_map.items(): - assert value, "Did not find diff for %s" % key - # END for each iteration type - - # test path not existing in the index - should be ignored - c = self.rorepo.head.commit - cp = c.parents[0] - diff_index = c.diff(cp, ["does/not/exist"]) - assert len(diff_index) == 0 - - + # test rename detection + rename_commit = self.rorepo.rev_parse('4772fe0') + rename_diffs = rename_commit.parents[0].diff(rename_commit, M=True) + rename_diffs = filter(lambda d: d.renamed, rename_diffs) + assert len(rename_diffs) == 3 + assert rename_diffs[0].rename_from == rename_diffs[0].a_blob.path + assert rename_diffs[0].rename_to == rename_diffs[0].b_blob.path + + # assert we could always find at least one instance of the members we + # can iterate in the diff index - if not this indicates its not working correctly + # or our test does not span the whole range of possibilities + for key,value in assertion_map.items(): + assert value, "Did not find diff for %s" % key + # END for each iteration type + + # test path not existing in the index - should be ignored + c = self.rorepo.head.commit + cp = c.parents[0] + diff_index = c.diff(cp, ["does/not/exist"]) + assert len(diff_index) == 0 + + diff --git a/git/test/test_example.py b/git/test/test_example.py index dbab3118..1fd87b3f 100644 --- a/git/test/test_example.py +++ b/git/test/test_example.py @@ -7,58 +7,58 @@ from lib import TestBase, fixture_path from git.base import IStream from git.db.py.loose import PureLooseObjectODB from git.util import pool - + from cStringIO import StringIO from async import IteratorReader - + class TestExamples(TestBase): - - def test_base(self): - ldb = PureLooseObjectODB(fixture_path("../../../.git/objects")) - - for sha1 in ldb.sha_iter(): - oinfo = ldb.info(sha1) - ostream = ldb.stream(sha1) - assert oinfo[:3] == ostream[:3] - - assert len(ostream.read()) == ostream.size - assert ldb.has_object(oinfo.binsha) - # END for each sha in database - # assure we close all files - try: - del(ostream) - del(oinfo) - except UnboundLocalError: - pass - # END ignore exception if there are no loose objects - - data = "my data" - istream = IStream("blob", len(data), StringIO(data)) - - # the object does not yet have a sha - assert istream.binsha is None - ldb.store(istream) - # now the sha is set - assert len(istream.binsha) == 20 - assert ldb.has_object(istream.binsha) - - - # async operation - # Create a reader from an iterator - reader = IteratorReader(ldb.sha_iter()) - - # get reader for object streams - info_reader = ldb.stream_async(reader) - - # read one - info = info_reader.read(1)[0] - - # read all the rest until depletion - ostreams = info_reader.read() - - # set the pool to use two threads - pool.set_size(2) - - # synchronize the mode of operation - pool.set_size(0) + + def test_base(self): + ldb = PureLooseObjectODB(fixture_path("../../../.git/objects")) + + for sha1 in ldb.sha_iter(): + oinfo = ldb.info(sha1) + ostream = ldb.stream(sha1) + assert oinfo[:3] == ostream[:3] + + assert len(ostream.read()) == ostream.size + assert ldb.has_object(oinfo.binsha) + # END for each sha in database + # assure we close all files + try: + del(ostream) + del(oinfo) + except UnboundLocalError: + pass + # END ignore exception if there are no loose objects + + data = "my data" + istream = IStream("blob", len(data), StringIO(data)) + + # the object does not yet have a sha + assert istream.binsha is None + ldb.store(istream) + # now the sha is set + assert len(istream.binsha) == 20 + assert ldb.has_object(istream.binsha) + + + # async operation + # Create a reader from an iterator + reader = IteratorReader(ldb.sha_iter()) + + # get reader for object streams + info_reader = ldb.stream_async(reader) + + # read one + info = info_reader.read(1)[0] + + # read all the rest until depletion + ostreams = info_reader.read() + + # set the pool to use two threads + pool.set_size(2) + + # synchronize the mode of operation + pool.set_size(0) diff --git a/git/test/test_fun.py b/git/test/test_fun.py index ed069912..15bc20ed 100644 --- a/git/test/test_fun.py +++ b/git/test/test_fun.py @@ -1,251 +1,251 @@ from git.test.lib import TestBase, with_rw_repo from git.objects.fun import ( - traverse_tree_recursive, - traverse_trees_recursive, - tree_to_stream - ) + traverse_tree_recursive, + traverse_trees_recursive, + tree_to_stream + ) from git.index.fun import ( - aggressive_tree_merge - ) + aggressive_tree_merge + ) from git.util import bin_to_hex from git.base import IStream from git.typ import str_tree_type from stat import ( - S_IFDIR, - S_IFREG, - S_IFLNK - ) + S_IFDIR, + S_IFREG, + S_IFLNK + ) from git.index import IndexFile from cStringIO import StringIO class TestFun(TestBase): - - def _assert_index_entries(self, entries, trees): - index = IndexFile.from_tree(self.rorepo, *[self.rorepo.tree(bin_to_hex(t)) for t in trees]) - assert entries - assert len(index.entries) == len(entries) - for entry in entries: - assert (entry.path, entry.stage) in index.entries - # END assert entry matches fully - - def test_aggressive_tree_merge(self): - # head tree with additions, removals and modification compared to its predecessor - odb = self.rorepo.odb - HC = self.rorepo.commit("6c1faef799095f3990e9970bc2cb10aa0221cf9c") - H = HC.tree - B = HC.parents[0].tree - - # entries from single tree - trees = [H.binsha] - self._assert_index_entries(aggressive_tree_merge(odb, trees), trees) - - # from multiple trees - trees = [B.binsha, H.binsha] - self._assert_index_entries(aggressive_tree_merge(odb, trees), trees) - - # three way, no conflict - tree = self.rorepo.tree - B = tree("35a09c0534e89b2d43ec4101a5fb54576b577905") - H = tree("4fe5cfa0e063a8d51a1eb6f014e2aaa994e5e7d4") - M = tree("1f2b19de3301e76ab3a6187a49c9c93ff78bafbd") - trees = [B.binsha, H.binsha, M.binsha] - self._assert_index_entries(aggressive_tree_merge(odb, trees), trees) - - # three-way, conflict in at least one file, both modified - B = tree("a7a4388eeaa4b6b94192dce67257a34c4a6cbd26") - H = tree("f9cec00938d9059882bb8eabdaf2f775943e00e5") - M = tree("44a601a068f4f543f73fd9c49e264c931b1e1652") - trees = [B.binsha, H.binsha, M.binsha] - self._assert_index_entries(aggressive_tree_merge(odb, trees), trees) - - # too many trees - self.failUnlessRaises(ValueError, aggressive_tree_merge, odb, trees*2) + + def _assert_index_entries(self, entries, trees): + index = IndexFile.from_tree(self.rorepo, *[self.rorepo.tree(bin_to_hex(t)) for t in trees]) + assert entries + assert len(index.entries) == len(entries) + for entry in entries: + assert (entry.path, entry.stage) in index.entries + # END assert entry matches fully + + def test_aggressive_tree_merge(self): + # head tree with additions, removals and modification compared to its predecessor + odb = self.rorepo.odb + HC = self.rorepo.commit("6c1faef799095f3990e9970bc2cb10aa0221cf9c") + H = HC.tree + B = HC.parents[0].tree + + # entries from single tree + trees = [H.binsha] + self._assert_index_entries(aggressive_tree_merge(odb, trees), trees) + + # from multiple trees + trees = [B.binsha, H.binsha] + self._assert_index_entries(aggressive_tree_merge(odb, trees), trees) + + # three way, no conflict + tree = self.rorepo.tree + B = tree("35a09c0534e89b2d43ec4101a5fb54576b577905") + H = tree("4fe5cfa0e063a8d51a1eb6f014e2aaa994e5e7d4") + M = tree("1f2b19de3301e76ab3a6187a49c9c93ff78bafbd") + trees = [B.binsha, H.binsha, M.binsha] + self._assert_index_entries(aggressive_tree_merge(odb, trees), trees) + + # three-way, conflict in at least one file, both modified + B = tree("a7a4388eeaa4b6b94192dce67257a34c4a6cbd26") + H = tree("f9cec00938d9059882bb8eabdaf2f775943e00e5") + M = tree("44a601a068f4f543f73fd9c49e264c931b1e1652") + trees = [B.binsha, H.binsha, M.binsha] + self._assert_index_entries(aggressive_tree_merge(odb, trees), trees) + + # too many trees + self.failUnlessRaises(ValueError, aggressive_tree_merge, odb, trees*2) - def mktree(self, odb, entries): - """create a tree from the given tree entries and safe it to the database""" - sio = StringIO() - tree_to_stream(entries, sio.write) - sio.seek(0) - istream = odb.store(IStream(str_tree_type, len(sio.getvalue()), sio)) - return istream.binsha - - @with_rw_repo('0.1.6') - def test_three_way_merge(self, rwrepo): - def mkfile(name, sha, executable=0): - return (sha, S_IFREG | 0644 | executable*0111, name) - def mkcommit(name, sha): - return (sha, S_IFDIR | S_IFLNK, name) - def assert_entries(entries, num_entries, has_conflict=False): - assert len(entries) == num_entries - assert has_conflict == (len([e for e in entries if e.stage != 0]) > 0) - mktree = self.mktree - - shaa = "\1"*20 - shab = "\2"*20 - shac = "\3"*20 - - odb = rwrepo.odb - - # base tree - bfn = 'basefile' - fbase = mkfile(bfn, shaa) - tb = mktree(odb, [fbase]) - - # non-conflicting new files, same data - fa = mkfile('1', shab) - th = mktree(odb, [fbase, fa]) - fb = mkfile('2', shac) - tm = mktree(odb, [fbase, fb]) - - # two new files, same base file - trees = [tb, th, tm] - assert_entries(aggressive_tree_merge(odb, trees), 3) - - # both delete same file, add own one - fa = mkfile('1', shab) - th = mktree(odb, [fa]) - fb = mkfile('2', shac) - tm = mktree(odb, [fb]) - - # two new files - trees = [tb, th, tm] - assert_entries(aggressive_tree_merge(odb, trees), 2) - - # same file added in both, differently - fa = mkfile('1', shab) - th = mktree(odb, [fa]) - fb = mkfile('1', shac) - tm = mktree(odb, [fb]) - - # expect conflict - trees = [tb, th, tm] - assert_entries(aggressive_tree_merge(odb, trees), 2, True) - - # same file added, different mode - fa = mkfile('1', shab) - th = mktree(odb, [fa]) - fb = mkcommit('1', shab) - tm = mktree(odb, [fb]) - - # expect conflict - trees = [tb, th, tm] - assert_entries(aggressive_tree_merge(odb, trees), 2, True) - - # same file added in both - fa = mkfile('1', shab) - th = mktree(odb, [fa]) - fb = mkfile('1', shab) - tm = mktree(odb, [fb]) - - # expect conflict - trees = [tb, th, tm] - assert_entries(aggressive_tree_merge(odb, trees), 1) - - # modify same base file, differently - fa = mkfile(bfn, shab) - th = mktree(odb, [fa]) - fb = mkfile(bfn, shac) - tm = mktree(odb, [fb]) - - # conflict, 3 versions on 3 stages - trees = [tb, th, tm] - assert_entries(aggressive_tree_merge(odb, trees), 3, True) - - - # change mode on same base file, by making one a commit, the other executable - # no content change ( this is totally unlikely to happen in the real world ) - fa = mkcommit(bfn, shaa) - th = mktree(odb, [fa]) - fb = mkfile(bfn, shaa, executable=1) - tm = mktree(odb, [fb]) - - # conflict, 3 versions on 3 stages, because of different mode - trees = [tb, th, tm] - assert_entries(aggressive_tree_merge(odb, trees), 3, True) - - for is_them in range(2): - # only we/they change contents - fa = mkfile(bfn, shab) - th = mktree(odb, [fa]) - - trees = [tb, th, tb] - if is_them: - trees = [tb, tb, th] - entries = aggressive_tree_merge(odb, trees) - assert len(entries) == 1 and entries[0].binsha == shab - - # only we/they change the mode - fa = mkcommit(bfn, shaa) - th = mktree(odb, [fa]) - - trees = [tb, th, tb] - if is_them: - trees = [tb, tb, th] - entries = aggressive_tree_merge(odb, trees) - assert len(entries) == 1 and entries[0].binsha == shaa and entries[0].mode == fa[1] - - # one side deletes, the other changes = conflict - fa = mkfile(bfn, shab) - th = mktree(odb, [fa]) - tm = mktree(odb, []) - trees = [tb, th, tm] - if is_them: - trees = [tb, tm, th] - # as one is deleted, there are only 2 entries - assert_entries(aggressive_tree_merge(odb, trees), 2, True) - # END handle ours, theirs - - def _assert_tree_entries(self, entries, num_trees): - for entry in entries: - assert len(entry) == num_trees - paths = set(e[2] for e in entry if e) - - # only one path per set of entries - assert len(paths) == 1 - # END verify entry - - def test_tree_traversal(self): - # low level tree tarversal - odb = self.rorepo.odb - H = self.rorepo.tree('29eb123beb1c55e5db4aa652d843adccbd09ae18') # head tree - M = self.rorepo.tree('e14e3f143e7260de9581aee27e5a9b2645db72de') # merge tree - B = self.rorepo.tree('f606937a7a21237c866efafcad33675e6539c103') # base tree - B_old = self.rorepo.tree('1f66cfbbce58b4b552b041707a12d437cc5f400a') # old base tree - - # two very different trees - entries = traverse_trees_recursive(odb, [B_old.binsha, H.binsha], '') - self._assert_tree_entries(entries, 2) - - oentries = traverse_trees_recursive(odb, [H.binsha, B_old.binsha], '') - assert len(oentries) == len(entries) - self._assert_tree_entries(oentries, 2) - - # single tree - is_no_tree = lambda i, d: i.type != 'tree' - entries = traverse_trees_recursive(odb, [B.binsha], '') - assert len(entries) == len(list(B.traverse(predicate=is_no_tree))) - self._assert_tree_entries(entries, 1) - - # two trees - entries = traverse_trees_recursive(odb, [B.binsha, H.binsha], '') - self._assert_tree_entries(entries, 2) - - # tree trees - entries = traverse_trees_recursive(odb, [B.binsha, H.binsha, M.binsha], '') - self._assert_tree_entries(entries, 3) - - def test_tree_traversal_single(self): - max_count = 50 - count = 0 - odb = self.rorepo.odb - for commit in self.rorepo.commit("29eb123beb1c55e5db4aa652d843adccbd09ae18").traverse(): - if count >= max_count: - break - count += 1 - entries = traverse_tree_recursive(odb, commit.tree.binsha, '') - assert entries - # END for each commit + def mktree(self, odb, entries): + """create a tree from the given tree entries and safe it to the database""" + sio = StringIO() + tree_to_stream(entries, sio.write) + sio.seek(0) + istream = odb.store(IStream(str_tree_type, len(sio.getvalue()), sio)) + return istream.binsha + + @with_rw_repo('0.1.6') + def test_three_way_merge(self, rwrepo): + def mkfile(name, sha, executable=0): + return (sha, S_IFREG | 0644 | executable*0111, name) + def mkcommit(name, sha): + return (sha, S_IFDIR | S_IFLNK, name) + def assert_entries(entries, num_entries, has_conflict=False): + assert len(entries) == num_entries + assert has_conflict == (len([e for e in entries if e.stage != 0]) > 0) + mktree = self.mktree + + shaa = "\1"*20 + shab = "\2"*20 + shac = "\3"*20 + + odb = rwrepo.odb + + # base tree + bfn = 'basefile' + fbase = mkfile(bfn, shaa) + tb = mktree(odb, [fbase]) + + # non-conflicting new files, same data + fa = mkfile('1', shab) + th = mktree(odb, [fbase, fa]) + fb = mkfile('2', shac) + tm = mktree(odb, [fbase, fb]) + + # two new files, same base file + trees = [tb, th, tm] + assert_entries(aggressive_tree_merge(odb, trees), 3) + + # both delete same file, add own one + fa = mkfile('1', shab) + th = mktree(odb, [fa]) + fb = mkfile('2', shac) + tm = mktree(odb, [fb]) + + # two new files + trees = [tb, th, tm] + assert_entries(aggressive_tree_merge(odb, trees), 2) + + # same file added in both, differently + fa = mkfile('1', shab) + th = mktree(odb, [fa]) + fb = mkfile('1', shac) + tm = mktree(odb, [fb]) + + # expect conflict + trees = [tb, th, tm] + assert_entries(aggressive_tree_merge(odb, trees), 2, True) + + # same file added, different mode + fa = mkfile('1', shab) + th = mktree(odb, [fa]) + fb = mkcommit('1', shab) + tm = mktree(odb, [fb]) + + # expect conflict + trees = [tb, th, tm] + assert_entries(aggressive_tree_merge(odb, trees), 2, True) + + # same file added in both + fa = mkfile('1', shab) + th = mktree(odb, [fa]) + fb = mkfile('1', shab) + tm = mktree(odb, [fb]) + + # expect conflict + trees = [tb, th, tm] + assert_entries(aggressive_tree_merge(odb, trees), 1) + + # modify same base file, differently + fa = mkfile(bfn, shab) + th = mktree(odb, [fa]) + fb = mkfile(bfn, shac) + tm = mktree(odb, [fb]) + + # conflict, 3 versions on 3 stages + trees = [tb, th, tm] + assert_entries(aggressive_tree_merge(odb, trees), 3, True) + + + # change mode on same base file, by making one a commit, the other executable + # no content change ( this is totally unlikely to happen in the real world ) + fa = mkcommit(bfn, shaa) + th = mktree(odb, [fa]) + fb = mkfile(bfn, shaa, executable=1) + tm = mktree(odb, [fb]) + + # conflict, 3 versions on 3 stages, because of different mode + trees = [tb, th, tm] + assert_entries(aggressive_tree_merge(odb, trees), 3, True) + + for is_them in range(2): + # only we/they change contents + fa = mkfile(bfn, shab) + th = mktree(odb, [fa]) + + trees = [tb, th, tb] + if is_them: + trees = [tb, tb, th] + entries = aggressive_tree_merge(odb, trees) + assert len(entries) == 1 and entries[0].binsha == shab + + # only we/they change the mode + fa = mkcommit(bfn, shaa) + th = mktree(odb, [fa]) + + trees = [tb, th, tb] + if is_them: + trees = [tb, tb, th] + entries = aggressive_tree_merge(odb, trees) + assert len(entries) == 1 and entries[0].binsha == shaa and entries[0].mode == fa[1] + + # one side deletes, the other changes = conflict + fa = mkfile(bfn, shab) + th = mktree(odb, [fa]) + tm = mktree(odb, []) + trees = [tb, th, tm] + if is_them: + trees = [tb, tm, th] + # as one is deleted, there are only 2 entries + assert_entries(aggressive_tree_merge(odb, trees), 2, True) + # END handle ours, theirs + + def _assert_tree_entries(self, entries, num_trees): + for entry in entries: + assert len(entry) == num_trees + paths = set(e[2] for e in entry if e) + + # only one path per set of entries + assert len(paths) == 1 + # END verify entry + + def test_tree_traversal(self): + # low level tree tarversal + odb = self.rorepo.odb + H = self.rorepo.tree('29eb123beb1c55e5db4aa652d843adccbd09ae18') # head tree + M = self.rorepo.tree('e14e3f143e7260de9581aee27e5a9b2645db72de') # merge tree + B = self.rorepo.tree('f606937a7a21237c866efafcad33675e6539c103') # base tree + B_old = self.rorepo.tree('1f66cfbbce58b4b552b041707a12d437cc5f400a') # old base tree + + # two very different trees + entries = traverse_trees_recursive(odb, [B_old.binsha, H.binsha], '') + self._assert_tree_entries(entries, 2) + + oentries = traverse_trees_recursive(odb, [H.binsha, B_old.binsha], '') + assert len(oentries) == len(entries) + self._assert_tree_entries(oentries, 2) + + # single tree + is_no_tree = lambda i, d: i.type != 'tree' + entries = traverse_trees_recursive(odb, [B.binsha], '') + assert len(entries) == len(list(B.traverse(predicate=is_no_tree))) + self._assert_tree_entries(entries, 1) + + # two trees + entries = traverse_trees_recursive(odb, [B.binsha, H.binsha], '') + self._assert_tree_entries(entries, 2) + + # tree trees + entries = traverse_trees_recursive(odb, [B.binsha, H.binsha, M.binsha], '') + self._assert_tree_entries(entries, 3) + + def test_tree_traversal_single(self): + max_count = 50 + count = 0 + odb = self.rorepo.odb + for commit in self.rorepo.commit("29eb123beb1c55e5db4aa652d843adccbd09ae18").traverse(): + if count >= max_count: + break + count += 1 + entries = traverse_tree_recursive(odb, commit.tree.binsha, '') + assert entries + # END for each commit diff --git a/git/test/test_import.py b/git/test/test_import.py index a5a1d11b..606d4b03 100644 --- a/git/test/test_import.py +++ b/git/test/test_import.py @@ -11,48 +11,48 @@ import os from git import * def import_all(topdir, topmodule='git', skip = "test"): - base = os.path.basename - join = os.path.join - init_script = '__init__.py' - prev_cwd = os.getcwd() - try: - os.chdir(os.path.dirname(topdir)) - for root, dirs, files in os.walk(base(topdir)): - if init_script not in files: - del(dirs[:]) - continue - #END ignore non-packages - - if skip in root: - continue - #END handle ignores - - for relafile in files: - if not relafile.endswith('.py'): - continue - if relafile == init_script: - continue - module_path = join(root, os.path.splitext(relafile)[0]).replace("/", ".").replace("\\", ".") - - m = __import__(module_path, globals(), locals(), [""]) - try: - attrlist = m.__all__ - for attr in attrlist: - assert hasattr(m, attr), "Invalid item in %s.__all__: %s" % (module_path, attr) - #END veriy - except AttributeError: - pass - # END try each listed attribute - #END for each file in dir - #END for each item - finally: - os.chdir(prev_cwd) - #END handle previous currentdir - - + base = os.path.basename + join = os.path.join + init_script = '__init__.py' + prev_cwd = os.getcwd() + try: + os.chdir(os.path.dirname(topdir)) + for root, dirs, files in os.walk(base(topdir)): + if init_script not in files: + del(dirs[:]) + continue + #END ignore non-packages + + if skip in root: + continue + #END handle ignores + + for relafile in files: + if not relafile.endswith('.py'): + continue + if relafile == init_script: + continue + module_path = join(root, os.path.splitext(relafile)[0]).replace("/", ".").replace("\\", ".") + + m = __import__(module_path, globals(), locals(), [""]) + try: + attrlist = m.__all__ + for attr in attrlist: + assert hasattr(m, attr), "Invalid item in %s.__all__: %s" % (module_path, attr) + #END veriy + except AttributeError: + pass + # END try each listed attribute + #END for each file in dir + #END for each item + finally: + os.chdir(prev_cwd) + #END handle previous currentdir + + class TestDummy(object): - def test_base(self): - dn = os.path.dirname - # NOTE: i don't think this is working, as the __all__ variable is not used in this case - import_all(dn(dn(__file__))) + def test_base(self): + dn = os.path.dirname + # NOTE: i don't think this is working, as the __all__ variable is not used in this case + import_all(dn(dn(__file__))) diff --git a/git/test/test_index.py b/git/test/test_index.py index 7d65cb9b..029c961b 100644 --- a/git/test/test_index.py +++ b/git/test/test_index.py @@ -5,11 +5,11 @@ # the BSD License: http://www.opensource.org/licenses/bsd-license.php from git.test.lib import ( - TestBase, - with_rw_repo, - fixture_path, - fixture - ) + TestBase, + with_rw_repo, + fixture_path, + fixture + ) from git import * import inspect import os @@ -21,655 +21,655 @@ import time from stat import * class TestIndex(TestBase): - - def __init__(self, *args): - super(TestIndex, self).__init__(*args) - self._reset_progress() - - def _assert_fprogress(self, entries): - assert len(entries) == len(self._fprogress_map) - for path, call_count in self._fprogress_map.iteritems(): - assert call_count == 2 - # END for each item in progress map - self._reset_progress() + + def __init__(self, *args): + super(TestIndex, self).__init__(*args) + self._reset_progress() + + def _assert_fprogress(self, entries): + assert len(entries) == len(self._fprogress_map) + for path, call_count in self._fprogress_map.iteritems(): + assert call_count == 2 + # END for each item in progress map + self._reset_progress() - def _fprogress(self, path, done, item): - self._fprogress_map.setdefault(path, 0) - curval = self._fprogress_map[path] - if curval == 0: - assert not done - if curval == 1: - assert done - self._fprogress_map[path] = curval + 1 - - def _fprogress_add(self, path, done, item): - """Called as progress func - we keep track of the proper - call order""" - assert item is not None - self._fprogress(path, done, item) - - def _reset_progress(self): - # maps paths to the count of calls - self._fprogress_map = dict() - - def _assert_entries(self, entries): - for entry in entries: - assert isinstance(entry, BaseIndexEntry) - assert not os.path.isabs(entry.path) - assert not "\\" in entry.path - # END for each entry - - def test_index_file_base(self): - # read from file - index = IndexFile(self.rorepo, fixture_path("index")) - assert index.entries - assert index.version > 0 - - # test entry - last_val = None - entry = index.entries.itervalues().next() - for attr in ("path","ctime","mtime","dev","inode","mode","uid", - "gid","size","binsha", "hexsha", "stage"): - val = getattr(entry, attr) - # END for each method - - # test update - entries = index.entries - assert isinstance(index.update(), IndexFile) - assert entries is not index.entries - - # test stage - index_merge = IndexFile(self.rorepo, fixture_path("index_merge")) - assert len(index_merge.entries) == 106 - assert len(list(e for e in index_merge.entries.itervalues() if e.stage != 0 )) - - # write the data - it must match the original - tmpfile = tempfile.mktemp() - index_merge.write(tmpfile) - fp = open(tmpfile, 'rb') - assert fp.read() == fixture("index_merge") - fp.close() - os.remove(tmpfile) - - def _cmp_tree_index(self, tree, index): - # fail unless both objects contain the same paths and blobs - if isinstance(tree, str): - tree = self.rorepo.commit(tree).tree - - num_blobs = 0 - blist = list() - for blob in tree.traverse(predicate = lambda e,d: e.type == "blob", branch_first=False): - assert (blob.path,0) in index.entries - blist.append(blob) - # END for each blob in tree - if len(blist) != len(index.entries): - iset = set(k[0] for k in index.entries.keys()) - bset = set(b.path for b in blist) - raise AssertionError( "CMP Failed: Missing entries in index: %s, missing in tree: %s" % (bset-iset, iset-bset) ) - # END assertion message - - @with_rw_repo('0.1.6') - def test_index_file_from_tree(self, rw_repo): - common_ancestor_sha = "5117c9c8a4d3af19a9958677e45cda9269de1541" - cur_sha = "4b43ca7ff72d5f535134241e7c797ddc9c7a3573" - other_sha = "39f85c4358b7346fee22169da9cad93901ea9eb9" - - # simple index from tree - base_index = IndexFile.from_tree(rw_repo, common_ancestor_sha) - assert base_index.entries - self._cmp_tree_index(common_ancestor_sha, base_index) - - # merge two trees - its like a fast-forward - two_way_index = IndexFile.from_tree(rw_repo, common_ancestor_sha, cur_sha) - assert two_way_index.entries - self._cmp_tree_index(cur_sha, two_way_index) - - # merge three trees - here we have a merge conflict - three_way_index = IndexFile.from_tree(rw_repo, common_ancestor_sha, cur_sha, other_sha) - assert len(list(e for e in three_way_index.entries.values() if e.stage != 0)) - - - # ITERATE BLOBS - merge_required = lambda t: t[0] != 0 - merge_blobs = list(three_way_index.iter_blobs(merge_required)) - assert merge_blobs - assert merge_blobs[0][0] in (1,2,3) - assert isinstance(merge_blobs[0][1], Blob) - - # test BlobFilter - prefix = 'lib/git' - for stage, blob in base_index.iter_blobs(BlobFilter([prefix])): - assert blob.path.startswith(prefix) - - - # writing a tree should fail with an unmerged index - self.failUnlessRaises(UnmergedEntriesError, three_way_index.write_tree) - - # removed unmerged entries - unmerged_blob_map = three_way_index.unmerged_blobs() - assert unmerged_blob_map - - # pick the first blob at the first stage we find and use it as resolved version - three_way_index.resolve_blobs( l[0][1] for l in unmerged_blob_map.itervalues() ) - tree = three_way_index.write_tree() - assert isinstance(tree, Tree) - num_blobs = 0 - for blob in tree.traverse(predicate=lambda item,d: item.type == "blob"): - assert (blob.path,0) in three_way_index.entries - num_blobs += 1 - # END for each blob - assert num_blobs == len(three_way_index.entries) - - @with_rw_repo('0.1.6') - def test_index_merge_tree(self, rw_repo): - # A bit out of place, but we need a different repo for this: - assert self.rorepo != rw_repo and not (self.rorepo == rw_repo) - assert len(set((self.rorepo, self.rorepo, rw_repo, rw_repo))) == 2 - - # SINGLE TREE MERGE - # current index is at the (virtual) cur_commit - next_commit = "4c39f9da792792d4e73fc3a5effde66576ae128c" - parent_commit = rw_repo.head.commit.parents[0] - manifest_key = IndexFile.entry_key('MANIFEST.in', 0) - manifest_entry = rw_repo.index.entries[manifest_key] - rw_repo.index.merge_tree(next_commit) - # only one change should be recorded - assert manifest_entry.binsha != rw_repo.index.entries[manifest_key].binsha - - rw_repo.index.reset(rw_repo.head) - assert rw_repo.index.entries[manifest_key].binsha == manifest_entry.binsha - - # FAKE MERGE - ############# - # Add a change with a NULL sha that should conflict with next_commit. We - # pretend there was a change, but we do not even bother adding a proper - # sha for it ( which makes things faster of course ) - manifest_fake_entry = BaseIndexEntry((manifest_entry[0], "\0"*20, 0, manifest_entry[3])) - # try write flag - self._assert_entries(rw_repo.index.add([manifest_fake_entry], write=False)) - # add actually resolves the null-hex-sha for us as a feature, but we can - # edit the index manually - assert rw_repo.index.entries[manifest_key].binsha != Object.NULL_BIN_SHA - # must operate on the same index for this ! Its a bit problematic as - # it might confuse people - index = rw_repo.index - index.entries[manifest_key] = IndexEntry.from_base(manifest_fake_entry) - index.write() - assert rw_repo.index.entries[manifest_key].hexsha == Diff.NULL_HEX_SHA - - # write an unchanged index ( just for the fun of it ) - rw_repo.index.write() - - # a three way merge would result in a conflict and fails as the command will - # not overwrite any entries in our index and hence leave them unmerged. This is - # mainly a protection feature as the current index is not yet in a tree - self.failUnlessRaises(GitCommandError, index.merge_tree, next_commit, base=parent_commit) - - # the only way to get the merged entries is to safe the current index away into a tree, - # which is like a temporary commit for us. This fails as well as the NULL sha deos not - # have a corresponding object - # NOTE: missing_ok is not a kwarg anymore, missing_ok is always true - # self.failUnlessRaises(GitCommandError, index.write_tree) - - # if missing objects are okay, this would work though ( they are always okay now ) - tree = index.write_tree() - - # now make a proper three way merge with unmerged entries - unmerged_tree = IndexFile.from_tree(rw_repo, parent_commit, tree, next_commit) - unmerged_blobs = unmerged_tree.unmerged_blobs() - assert len(unmerged_blobs) == 1 and unmerged_blobs.keys()[0] == manifest_key[0] - - - @with_rw_repo('0.1.6') - def test_index_file_diffing(self, rw_repo): - # default Index instance points to our index - index = IndexFile(rw_repo) - assert index.path is not None - assert len(index.entries) - - # write the file back - index.write() - - # could sha it, or check stats - - # test diff - # resetting the head will leave the index in a different state, and the - # diff will yield a few changes - cur_head_commit = rw_repo.head.reference.commit - ref = rw_repo.head.reset('HEAD~6', index=True, working_tree=False) - - # diff against same index is 0 - diff = index.diff() - assert len(diff) == 0 - - # against HEAD as string, must be the same as it matches index - diff = index.diff('HEAD') - assert len(diff) == 0 - - # against previous head, there must be a difference - diff = index.diff(cur_head_commit) - assert len(diff) - - # we reverse the result - adiff = index.diff(str(cur_head_commit), R=True) - odiff = index.diff(cur_head_commit, R=False) # now its not reversed anymore - assert adiff != odiff - assert odiff == diff # both unreversed diffs against HEAD - - # against working copy - its still at cur_commit - wdiff = index.diff(None) - assert wdiff != adiff - assert wdiff != odiff - - # against something unusual - self.failUnlessRaises(ValueError, index.diff, int) - - # adjust the index to match an old revision - cur_branch = rw_repo.active_branch - cur_commit = cur_branch.commit - rev_head_parent = 'HEAD~1' - assert index.reset(rev_head_parent) is index - - assert cur_branch == rw_repo.active_branch - assert cur_commit == rw_repo.head.commit - - # there must be differences towards the working tree which is in the 'future' - assert index.diff(None) - - # reset the working copy as well to current head,to pull 'back' as well - new_data = "will be reverted" - file_path = os.path.join(rw_repo.working_tree_dir, "CHANGES") - fp = open(file_path, "wb") - fp.write(new_data) - fp.close() - index.reset(rev_head_parent, working_tree=True) - assert not index.diff(None) - assert cur_branch == rw_repo.active_branch - assert cur_commit == rw_repo.head.commit - fp = open(file_path,'rb') - try: - assert fp.read() != new_data - finally: - fp.close() - - # test full checkout - test_file = os.path.join(rw_repo.working_tree_dir, "CHANGES") - open(test_file, 'ab').write("some data") - rval = index.checkout(None, force=True, fprogress=self._fprogress) - assert 'CHANGES' in list(rval) - self._assert_fprogress([None]) - assert os.path.isfile(test_file) - - os.remove(test_file) - rval = index.checkout(None, force=False, fprogress=self._fprogress) - assert 'CHANGES' in list(rval) - self._assert_fprogress([None]) - assert os.path.isfile(test_file) - - # individual file - os.remove(test_file) - rval = index.checkout(test_file, fprogress=self._fprogress) - assert list(rval)[0] == 'CHANGES' - self._assert_fprogress([test_file]) - assert os.path.exists(test_file) - - # checking out non-existing file throws - self.failUnlessRaises(CheckoutError, index.checkout, "doesnt_exist_ever.txt.that") - self.failUnlessRaises(CheckoutError, index.checkout, paths=["doesnt/exist"]) - - # checkout file with modifications - append_data = "hello" - fp = open(test_file, "ab") - fp.write(append_data) - fp.close() - try: - index.checkout(test_file) - except CheckoutError, e: - assert len(e.failed_files) == 1 and e.failed_files[0] == os.path.basename(test_file) - assert (len(e.failed_files) == len(e.failed_reasons)) and isinstance(e.failed_reasons[0], basestring) - assert len(e.valid_files) == 0 - assert open(test_file).read().endswith(append_data) - else: - raise AssertionError("Exception CheckoutError not thrown") - - # if we force it it should work - index.checkout(test_file, force=True) - assert not open(test_file).read().endswith(append_data) - - # checkout directory - shutil.rmtree(os.path.join(rw_repo.working_tree_dir, "lib")) - rval = index.checkout('lib') - assert len(list(rval)) > 1 - - def _count_existing(self, repo, files): - """ - Returns count of files that actually exist in the repository directory. - """ - existing = 0 - basedir = repo.working_tree_dir - for f in files: - existing += os.path.isfile(os.path.join(basedir, f)) - # END for each deleted file - return existing - # END num existing helper - - @with_rw_repo('0.1.6') - def test_index_mutation(self, rw_repo): - index = rw_repo.index - num_entries = len(index.entries) - cur_head = rw_repo.head - - uname = "Some Developer" - umail = "sd@company.com" - rw_repo.config_writer().set_value("user", "name", uname) - rw_repo.config_writer().set_value("user", "email", umail) - - # remove all of the files, provide a wild mix of paths, BaseIndexEntries, - # IndexEntries - def mixed_iterator(): - count = 0 - for entry in index.entries.itervalues(): - type_id = count % 4 - if type_id == 0: # path - yield entry.path - elif type_id == 1: # blob - yield Blob(rw_repo, entry.binsha, entry.mode, entry.path) - elif type_id == 2: # BaseIndexEntry - yield BaseIndexEntry(entry[:4]) - elif type_id == 3: # IndexEntry - yield entry - else: - raise AssertionError("Invalid Type") - count += 1 - # END for each entry - # END mixed iterator - deleted_files = index.remove(mixed_iterator(), working_tree=False) - assert deleted_files - assert self._count_existing(rw_repo, deleted_files) == len(deleted_files) - assert len(index.entries) == 0 - - # reset the index to undo our changes - index.reset() - assert len(index.entries) == num_entries - - # remove with working copy - deleted_files = index.remove(mixed_iterator(), working_tree=True) - assert deleted_files - assert self._count_existing(rw_repo, deleted_files) == 0 - - # reset everything - index.reset(working_tree=True) - assert self._count_existing(rw_repo, deleted_files) == len(deleted_files) - - # invalid type - self.failUnlessRaises(TypeError, index.remove, [1]) - - # absolute path - deleted_files = index.remove([os.path.join(rw_repo.working_tree_dir,"lib")], r=True) - assert len(deleted_files) > 1 - self.failUnlessRaises(ValueError, index.remove, ["/doesnt/exists"]) - - # TEST COMMITTING - # commit changed index - cur_commit = cur_head.commit - commit_message = "commit default head" - - new_commit = index.commit(commit_message, head=False) - assert cur_commit != new_commit - assert new_commit.author.name == uname - assert new_commit.author.email == umail - assert new_commit.committer.name == uname - assert new_commit.committer.email == umail - assert new_commit.message == commit_message - assert new_commit.parents[0] == cur_commit - assert len(new_commit.parents) == 1 - assert cur_head.commit == cur_commit - - # same index, no parents - commit_message = "index without parents" - commit_no_parents = index.commit(commit_message, parent_commits=list(), head=True) - assert commit_no_parents.message == commit_message - assert len(commit_no_parents.parents) == 0 - assert cur_head.commit == commit_no_parents - - # same index, multiple parents - commit_message = "Index with multiple parents\n commit with another line" - commit_multi_parent = index.commit(commit_message,parent_commits=(commit_no_parents, new_commit)) - assert commit_multi_parent.message == commit_message - assert len(commit_multi_parent.parents) == 2 - assert commit_multi_parent.parents[0] == commit_no_parents - assert commit_multi_parent.parents[1] == new_commit - assert cur_head.commit == commit_multi_parent - - # re-add all files in lib - # get the lib folder back on disk, but get an index without it - index.reset(new_commit.parents[0], working_tree=True).reset(new_commit, working_tree=False) - lib_file_path = os.path.join("lib", "git", "__init__.py") - assert (lib_file_path, 0) not in index.entries - assert os.path.isfile(os.path.join(rw_repo.working_tree_dir, lib_file_path)) - - # directory - entries = index.add(['lib'], fprogress=self._fprogress_add) - self._assert_entries(entries) - self._assert_fprogress(entries) - assert len(entries)>1 - - # glob - entries = index.reset(new_commit).add([os.path.join('lib', 'git', '*.py')], fprogress=self._fprogress_add) - self._assert_entries(entries) - self._assert_fprogress(entries) - assert len(entries) == 14 - - # same file - entries = index.reset(new_commit).add([os.path.abspath(os.path.join('lib', 'git', 'head.py'))]*2, fprogress=self._fprogress_add) - self._assert_entries(entries) - assert entries[0].mode & 0644 == 0644 - # would fail, test is too primitive to handle this case - # self._assert_fprogress(entries) - self._reset_progress() - assert len(entries) == 2 - - # missing path - self.failUnlessRaises(OSError, index.reset(new_commit).add, ['doesnt/exist/must/raise']) - - # blob from older revision overrides current index revision - old_blob = new_commit.parents[0].tree.blobs[0] - entries = index.reset(new_commit).add([old_blob], fprogress=self._fprogress_add) - self._assert_entries(entries) - self._assert_fprogress(entries) - assert index.entries[(old_blob.path,0)].hexsha == old_blob.hexsha and len(entries) == 1 - - # mode 0 not allowed - null_hex_sha = Diff.NULL_HEX_SHA - null_bin_sha = "\0" * 20 - self.failUnlessRaises(ValueError, index.reset(new_commit).add, [BaseIndexEntry((0, null_bin_sha,0,"doesntmatter"))]) - - # add new file - new_file_relapath = "my_new_file" - new_file_path = self._make_file(new_file_relapath, "hello world", rw_repo) - entries = index.reset(new_commit).add([BaseIndexEntry((010644, null_bin_sha, 0, new_file_relapath))], fprogress=self._fprogress_add) - self._assert_entries(entries) - self._assert_fprogress(entries) - assert len(entries) == 1 and entries[0].hexsha != null_hex_sha - - # add symlink - if sys.platform != "win32": - basename = "my_real_symlink" - target = "/etc/that" - link_file = os.path.join(rw_repo.working_tree_dir, basename) - os.symlink(target, link_file) - entries = index.reset(new_commit).add([link_file], fprogress=self._fprogress_add) - self._assert_entries(entries) - self._assert_fprogress(entries) - assert len(entries) == 1 and S_ISLNK(entries[0].mode) - assert S_ISLNK(index.entries[index.entry_key("my_real_symlink", 0)].mode) - - # we expect only the target to be written - assert index.repo.odb.stream(entries[0].binsha).read() == target - # END real symlink test - - # add fake symlink and assure it checks-our as symlink - fake_symlink_relapath = "my_fake_symlink" - link_target = "/etc/that" - fake_symlink_path = self._make_file(fake_symlink_relapath, link_target, rw_repo) - fake_entry = BaseIndexEntry((0120000, null_bin_sha, 0, fake_symlink_relapath)) - entries = index.reset(new_commit).add([fake_entry], fprogress=self._fprogress_add) - self._assert_entries(entries) - self._assert_fprogress(entries) - assert entries[0].hexsha != null_hex_sha - assert len(entries) == 1 and S_ISLNK(entries[0].mode) - - # assure this also works with an alternate method - full_index_entry = IndexEntry.from_base(BaseIndexEntry((0120000, entries[0].binsha, 0, entries[0].path))) - entry_key = index.entry_key(full_index_entry) - index.reset(new_commit) - - assert entry_key not in index.entries - index.entries[entry_key] = full_index_entry - index.write() - index.update() # force reread of entries - new_entry = index.entries[entry_key] - assert S_ISLNK(new_entry.mode) - - # a tree created from this should contain the symlink - tree = index.write_tree() - assert fake_symlink_relapath in tree - index.write() # flush our changes for the checkout - - # checkout the fakelink, should be a link then - assert not S_ISLNK(os.stat(fake_symlink_path)[ST_MODE]) - os.remove(fake_symlink_path) - index.checkout(fake_symlink_path) - - # on windows we will never get symlinks - if os.name == 'nt': - # simlinks should contain the link as text ( which is what a - # symlink actually is ) - open(fake_symlink_path,'rb').read() == link_target - else: - assert S_ISLNK(os.lstat(fake_symlink_path)[ST_MODE]) - - # TEST RENAMING - def assert_mv_rval(rval): - for source, dest in rval: - assert not os.path.exists(source) and os.path.exists(dest) - # END for each renamed item - # END move assertion utility - - self.failUnlessRaises(ValueError, index.move, ['just_one_path']) - # file onto existing file - files = ['AUTHORS', 'LICENSE'] - self.failUnlessRaises(GitCommandError, index.move, files) - - # again, with force - assert_mv_rval(index.move(files, f=True)) - - # files into directory - dry run - paths = ['LICENSE', 'VERSION', 'doc'] - rval = index.move(paths, dry_run=True) - assert len(rval) == 2 - assert os.path.exists(paths[0]) - - # again, no dry run - rval = index.move(paths) - assert_mv_rval(rval) - - # dir into dir - rval = index.move(['doc', 'test']) - assert_mv_rval(rval) - - - # TEST PATH REWRITING - ###################### - count = [0] - def rewriter(entry): - rval = str(count[0]) - count[0] += 1 - return rval - # END rewriter - - def make_paths(): - # two existing ones, one new one - yield 'CHANGES' - yield 'ez_setup.py' - yield index.entries[index.entry_key('README', 0)] - yield index.entries[index.entry_key('.gitignore', 0)] - - for fid in range(3): - fname = 'newfile%i' % fid - open(fname, 'wb').write("abcd") - yield Blob(rw_repo, Blob.NULL_BIN_SHA, 0100644, fname) - # END for each new file - # END path producer - paths = list(make_paths()) - self._assert_entries(index.add(paths, path_rewriter=rewriter)) - - for filenum in range(len(paths)): - assert index.entry_key(str(filenum), 0) in index.entries - - - # TEST RESET ON PATHS - ###################### - arela = "aa" - brela = "bb" - afile = self._make_file(arela, "adata", rw_repo) - bfile = self._make_file(brela, "bdata", rw_repo) - akey = index.entry_key(arela, 0) - bkey = index.entry_key(brela, 0) - keys = (akey, bkey) - absfiles = (afile, bfile) - files = (arela, brela) - - for fkey in keys: - assert not fkey in index.entries - - index.add(files, write=True) - nc = index.commit("2 files committed", head=False) - - for fkey in keys: - assert fkey in index.entries - - # just the index - index.reset(paths=(arela, afile)) - assert not akey in index.entries - assert bkey in index.entries - - # now with working tree - files on disk as well as entries must be recreated - rw_repo.head.commit = nc - for absfile in absfiles: - os.remove(absfile) - - index.reset(working_tree=True, paths=files) - - for fkey in keys: - assert fkey in index.entries - for absfile in absfiles: - assert os.path.isfile(absfile) - - - @with_rw_repo('HEAD') - def test_compare_write_tree(self, rw_repo): - # write all trees and compare them - # its important to have a few submodules in there too - max_count = 25 - count = 0 - for commit in rw_repo.head.commit.traverse(): - if count >= max_count: - break - count += 1 - index = rw_repo.index.reset(commit) - orig_tree = commit.tree - assert index.write_tree() == orig_tree - # END for each commit - - def test_index_new(self): - B = self.rorepo.tree("6d9b1f4f9fa8c9f030e3207e7deacc5d5f8bba4e") - H = self.rorepo.tree("25dca42bac17d511b7e2ebdd9d1d679e7626db5f") - M = self.rorepo.tree("e746f96bcc29238b79118123028ca170adc4ff0f") - - for args in ((B,), (B,H), (B,H,M)): - index = IndexFile.new(self.rorepo, *args) - assert isinstance(index, IndexFile) - # END for each arg tuple - - + def _fprogress(self, path, done, item): + self._fprogress_map.setdefault(path, 0) + curval = self._fprogress_map[path] + if curval == 0: + assert not done + if curval == 1: + assert done + self._fprogress_map[path] = curval + 1 + + def _fprogress_add(self, path, done, item): + """Called as progress func - we keep track of the proper + call order""" + assert item is not None + self._fprogress(path, done, item) + + def _reset_progress(self): + # maps paths to the count of calls + self._fprogress_map = dict() + + def _assert_entries(self, entries): + for entry in entries: + assert isinstance(entry, BaseIndexEntry) + assert not os.path.isabs(entry.path) + assert not "\\" in entry.path + # END for each entry + + def test_index_file_base(self): + # read from file + index = IndexFile(self.rorepo, fixture_path("index")) + assert index.entries + assert index.version > 0 + + # test entry + last_val = None + entry = index.entries.itervalues().next() + for attr in ("path","ctime","mtime","dev","inode","mode","uid", + "gid","size","binsha", "hexsha", "stage"): + val = getattr(entry, attr) + # END for each method + + # test update + entries = index.entries + assert isinstance(index.update(), IndexFile) + assert entries is not index.entries + + # test stage + index_merge = IndexFile(self.rorepo, fixture_path("index_merge")) + assert len(index_merge.entries) == 106 + assert len(list(e for e in index_merge.entries.itervalues() if e.stage != 0 )) + + # write the data - it must match the original + tmpfile = tempfile.mktemp() + index_merge.write(tmpfile) + fp = open(tmpfile, 'rb') + assert fp.read() == fixture("index_merge") + fp.close() + os.remove(tmpfile) + + def _cmp_tree_index(self, tree, index): + # fail unless both objects contain the same paths and blobs + if isinstance(tree, str): + tree = self.rorepo.commit(tree).tree + + num_blobs = 0 + blist = list() + for blob in tree.traverse(predicate = lambda e,d: e.type == "blob", branch_first=False): + assert (blob.path,0) in index.entries + blist.append(blob) + # END for each blob in tree + if len(blist) != len(index.entries): + iset = set(k[0] for k in index.entries.keys()) + bset = set(b.path for b in blist) + raise AssertionError( "CMP Failed: Missing entries in index: %s, missing in tree: %s" % (bset-iset, iset-bset) ) + # END assertion message + + @with_rw_repo('0.1.6') + def test_index_file_from_tree(self, rw_repo): + common_ancestor_sha = "5117c9c8a4d3af19a9958677e45cda9269de1541" + cur_sha = "4b43ca7ff72d5f535134241e7c797ddc9c7a3573" + other_sha = "39f85c4358b7346fee22169da9cad93901ea9eb9" + + # simple index from tree + base_index = IndexFile.from_tree(rw_repo, common_ancestor_sha) + assert base_index.entries + self._cmp_tree_index(common_ancestor_sha, base_index) + + # merge two trees - its like a fast-forward + two_way_index = IndexFile.from_tree(rw_repo, common_ancestor_sha, cur_sha) + assert two_way_index.entries + self._cmp_tree_index(cur_sha, two_way_index) + + # merge three trees - here we have a merge conflict + three_way_index = IndexFile.from_tree(rw_repo, common_ancestor_sha, cur_sha, other_sha) + assert len(list(e for e in three_way_index.entries.values() if e.stage != 0)) + + + # ITERATE BLOBS + merge_required = lambda t: t[0] != 0 + merge_blobs = list(three_way_index.iter_blobs(merge_required)) + assert merge_blobs + assert merge_blobs[0][0] in (1,2,3) + assert isinstance(merge_blobs[0][1], Blob) + + # test BlobFilter + prefix = 'lib/git' + for stage, blob in base_index.iter_blobs(BlobFilter([prefix])): + assert blob.path.startswith(prefix) + + + # writing a tree should fail with an unmerged index + self.failUnlessRaises(UnmergedEntriesError, three_way_index.write_tree) + + # removed unmerged entries + unmerged_blob_map = three_way_index.unmerged_blobs() + assert unmerged_blob_map + + # pick the first blob at the first stage we find and use it as resolved version + three_way_index.resolve_blobs( l[0][1] for l in unmerged_blob_map.itervalues() ) + tree = three_way_index.write_tree() + assert isinstance(tree, Tree) + num_blobs = 0 + for blob in tree.traverse(predicate=lambda item,d: item.type == "blob"): + assert (blob.path,0) in three_way_index.entries + num_blobs += 1 + # END for each blob + assert num_blobs == len(three_way_index.entries) + + @with_rw_repo('0.1.6') + def test_index_merge_tree(self, rw_repo): + # A bit out of place, but we need a different repo for this: + assert self.rorepo != rw_repo and not (self.rorepo == rw_repo) + assert len(set((self.rorepo, self.rorepo, rw_repo, rw_repo))) == 2 + + # SINGLE TREE MERGE + # current index is at the (virtual) cur_commit + next_commit = "4c39f9da792792d4e73fc3a5effde66576ae128c" + parent_commit = rw_repo.head.commit.parents[0] + manifest_key = IndexFile.entry_key('MANIFEST.in', 0) + manifest_entry = rw_repo.index.entries[manifest_key] + rw_repo.index.merge_tree(next_commit) + # only one change should be recorded + assert manifest_entry.binsha != rw_repo.index.entries[manifest_key].binsha + + rw_repo.index.reset(rw_repo.head) + assert rw_repo.index.entries[manifest_key].binsha == manifest_entry.binsha + + # FAKE MERGE + ############# + # Add a change with a NULL sha that should conflict with next_commit. We + # pretend there was a change, but we do not even bother adding a proper + # sha for it ( which makes things faster of course ) + manifest_fake_entry = BaseIndexEntry((manifest_entry[0], "\0"*20, 0, manifest_entry[3])) + # try write flag + self._assert_entries(rw_repo.index.add([manifest_fake_entry], write=False)) + # add actually resolves the null-hex-sha for us as a feature, but we can + # edit the index manually + assert rw_repo.index.entries[manifest_key].binsha != Object.NULL_BIN_SHA + # must operate on the same index for this ! Its a bit problematic as + # it might confuse people + index = rw_repo.index + index.entries[manifest_key] = IndexEntry.from_base(manifest_fake_entry) + index.write() + assert rw_repo.index.entries[manifest_key].hexsha == Diff.NULL_HEX_SHA + + # write an unchanged index ( just for the fun of it ) + rw_repo.index.write() + + # a three way merge would result in a conflict and fails as the command will + # not overwrite any entries in our index and hence leave them unmerged. This is + # mainly a protection feature as the current index is not yet in a tree + self.failUnlessRaises(GitCommandError, index.merge_tree, next_commit, base=parent_commit) + + # the only way to get the merged entries is to safe the current index away into a tree, + # which is like a temporary commit for us. This fails as well as the NULL sha deos not + # have a corresponding object + # NOTE: missing_ok is not a kwarg anymore, missing_ok is always true + # self.failUnlessRaises(GitCommandError, index.write_tree) + + # if missing objects are okay, this would work though ( they are always okay now ) + tree = index.write_tree() + + # now make a proper three way merge with unmerged entries + unmerged_tree = IndexFile.from_tree(rw_repo, parent_commit, tree, next_commit) + unmerged_blobs = unmerged_tree.unmerged_blobs() + assert len(unmerged_blobs) == 1 and unmerged_blobs.keys()[0] == manifest_key[0] + + + @with_rw_repo('0.1.6') + def test_index_file_diffing(self, rw_repo): + # default Index instance points to our index + index = IndexFile(rw_repo) + assert index.path is not None + assert len(index.entries) + + # write the file back + index.write() + + # could sha it, or check stats + + # test diff + # resetting the head will leave the index in a different state, and the + # diff will yield a few changes + cur_head_commit = rw_repo.head.reference.commit + ref = rw_repo.head.reset('HEAD~6', index=True, working_tree=False) + + # diff against same index is 0 + diff = index.diff() + assert len(diff) == 0 + + # against HEAD as string, must be the same as it matches index + diff = index.diff('HEAD') + assert len(diff) == 0 + + # against previous head, there must be a difference + diff = index.diff(cur_head_commit) + assert len(diff) + + # we reverse the result + adiff = index.diff(str(cur_head_commit), R=True) + odiff = index.diff(cur_head_commit, R=False) # now its not reversed anymore + assert adiff != odiff + assert odiff == diff # both unreversed diffs against HEAD + + # against working copy - its still at cur_commit + wdiff = index.diff(None) + assert wdiff != adiff + assert wdiff != odiff + + # against something unusual + self.failUnlessRaises(ValueError, index.diff, int) + + # adjust the index to match an old revision + cur_branch = rw_repo.active_branch + cur_commit = cur_branch.commit + rev_head_parent = 'HEAD~1' + assert index.reset(rev_head_parent) is index + + assert cur_branch == rw_repo.active_branch + assert cur_commit == rw_repo.head.commit + + # there must be differences towards the working tree which is in the 'future' + assert index.diff(None) + + # reset the working copy as well to current head,to pull 'back' as well + new_data = "will be reverted" + file_path = os.path.join(rw_repo.working_tree_dir, "CHANGES") + fp = open(file_path, "wb") + fp.write(new_data) + fp.close() + index.reset(rev_head_parent, working_tree=True) + assert not index.diff(None) + assert cur_branch == rw_repo.active_branch + assert cur_commit == rw_repo.head.commit + fp = open(file_path,'rb') + try: + assert fp.read() != new_data + finally: + fp.close() + + # test full checkout + test_file = os.path.join(rw_repo.working_tree_dir, "CHANGES") + open(test_file, 'ab').write("some data") + rval = index.checkout(None, force=True, fprogress=self._fprogress) + assert 'CHANGES' in list(rval) + self._assert_fprogress([None]) + assert os.path.isfile(test_file) + + os.remove(test_file) + rval = index.checkout(None, force=False, fprogress=self._fprogress) + assert 'CHANGES' in list(rval) + self._assert_fprogress([None]) + assert os.path.isfile(test_file) + + # individual file + os.remove(test_file) + rval = index.checkout(test_file, fprogress=self._fprogress) + assert list(rval)[0] == 'CHANGES' + self._assert_fprogress([test_file]) + assert os.path.exists(test_file) + + # checking out non-existing file throws + self.failUnlessRaises(CheckoutError, index.checkout, "doesnt_exist_ever.txt.that") + self.failUnlessRaises(CheckoutError, index.checkout, paths=["doesnt/exist"]) + + # checkout file with modifications + append_data = "hello" + fp = open(test_file, "ab") + fp.write(append_data) + fp.close() + try: + index.checkout(test_file) + except CheckoutError, e: + assert len(e.failed_files) == 1 and e.failed_files[0] == os.path.basename(test_file) + assert (len(e.failed_files) == len(e.failed_reasons)) and isinstance(e.failed_reasons[0], basestring) + assert len(e.valid_files) == 0 + assert open(test_file).read().endswith(append_data) + else: + raise AssertionError("Exception CheckoutError not thrown") + + # if we force it it should work + index.checkout(test_file, force=True) + assert not open(test_file).read().endswith(append_data) + + # checkout directory + shutil.rmtree(os.path.join(rw_repo.working_tree_dir, "lib")) + rval = index.checkout('lib') + assert len(list(rval)) > 1 + + def _count_existing(self, repo, files): + """ + Returns count of files that actually exist in the repository directory. + """ + existing = 0 + basedir = repo.working_tree_dir + for f in files: + existing += os.path.isfile(os.path.join(basedir, f)) + # END for each deleted file + return existing + # END num existing helper + + @with_rw_repo('0.1.6') + def test_index_mutation(self, rw_repo): + index = rw_repo.index + num_entries = len(index.entries) + cur_head = rw_repo.head + + uname = "Some Developer" + umail = "sd@company.com" + rw_repo.config_writer().set_value("user", "name", uname) + rw_repo.config_writer().set_value("user", "email", umail) + + # remove all of the files, provide a wild mix of paths, BaseIndexEntries, + # IndexEntries + def mixed_iterator(): + count = 0 + for entry in index.entries.itervalues(): + type_id = count % 4 + if type_id == 0: # path + yield entry.path + elif type_id == 1: # blob + yield Blob(rw_repo, entry.binsha, entry.mode, entry.path) + elif type_id == 2: # BaseIndexEntry + yield BaseIndexEntry(entry[:4]) + elif type_id == 3: # IndexEntry + yield entry + else: + raise AssertionError("Invalid Type") + count += 1 + # END for each entry + # END mixed iterator + deleted_files = index.remove(mixed_iterator(), working_tree=False) + assert deleted_files + assert self._count_existing(rw_repo, deleted_files) == len(deleted_files) + assert len(index.entries) == 0 + + # reset the index to undo our changes + index.reset() + assert len(index.entries) == num_entries + + # remove with working copy + deleted_files = index.remove(mixed_iterator(), working_tree=True) + assert deleted_files + assert self._count_existing(rw_repo, deleted_files) == 0 + + # reset everything + index.reset(working_tree=True) + assert self._count_existing(rw_repo, deleted_files) == len(deleted_files) + + # invalid type + self.failUnlessRaises(TypeError, index.remove, [1]) + + # absolute path + deleted_files = index.remove([os.path.join(rw_repo.working_tree_dir,"lib")], r=True) + assert len(deleted_files) > 1 + self.failUnlessRaises(ValueError, index.remove, ["/doesnt/exists"]) + + # TEST COMMITTING + # commit changed index + cur_commit = cur_head.commit + commit_message = "commit default head" + + new_commit = index.commit(commit_message, head=False) + assert cur_commit != new_commit + assert new_commit.author.name == uname + assert new_commit.author.email == umail + assert new_commit.committer.name == uname + assert new_commit.committer.email == umail + assert new_commit.message == commit_message + assert new_commit.parents[0] == cur_commit + assert len(new_commit.parents) == 1 + assert cur_head.commit == cur_commit + + # same index, no parents + commit_message = "index without parents" + commit_no_parents = index.commit(commit_message, parent_commits=list(), head=True) + assert commit_no_parents.message == commit_message + assert len(commit_no_parents.parents) == 0 + assert cur_head.commit == commit_no_parents + + # same index, multiple parents + commit_message = "Index with multiple parents\n commit with another line" + commit_multi_parent = index.commit(commit_message,parent_commits=(commit_no_parents, new_commit)) + assert commit_multi_parent.message == commit_message + assert len(commit_multi_parent.parents) == 2 + assert commit_multi_parent.parents[0] == commit_no_parents + assert commit_multi_parent.parents[1] == new_commit + assert cur_head.commit == commit_multi_parent + + # re-add all files in lib + # get the lib folder back on disk, but get an index without it + index.reset(new_commit.parents[0], working_tree=True).reset(new_commit, working_tree=False) + lib_file_path = os.path.join("lib", "git", "__init__.py") + assert (lib_file_path, 0) not in index.entries + assert os.path.isfile(os.path.join(rw_repo.working_tree_dir, lib_file_path)) + + # directory + entries = index.add(['lib'], fprogress=self._fprogress_add) + self._assert_entries(entries) + self._assert_fprogress(entries) + assert len(entries)>1 + + # glob + entries = index.reset(new_commit).add([os.path.join('lib', 'git', '*.py')], fprogress=self._fprogress_add) + self._assert_entries(entries) + self._assert_fprogress(entries) + assert len(entries) == 14 + + # same file + entries = index.reset(new_commit).add([os.path.abspath(os.path.join('lib', 'git', 'head.py'))]*2, fprogress=self._fprogress_add) + self._assert_entries(entries) + assert entries[0].mode & 0644 == 0644 + # would fail, test is too primitive to handle this case + # self._assert_fprogress(entries) + self._reset_progress() + assert len(entries) == 2 + + # missing path + self.failUnlessRaises(OSError, index.reset(new_commit).add, ['doesnt/exist/must/raise']) + + # blob from older revision overrides current index revision + old_blob = new_commit.parents[0].tree.blobs[0] + entries = index.reset(new_commit).add([old_blob], fprogress=self._fprogress_add) + self._assert_entries(entries) + self._assert_fprogress(entries) + assert index.entries[(old_blob.path,0)].hexsha == old_blob.hexsha and len(entries) == 1 + + # mode 0 not allowed + null_hex_sha = Diff.NULL_HEX_SHA + null_bin_sha = "\0" * 20 + self.failUnlessRaises(ValueError, index.reset(new_commit).add, [BaseIndexEntry((0, null_bin_sha,0,"doesntmatter"))]) + + # add new file + new_file_relapath = "my_new_file" + new_file_path = self._make_file(new_file_relapath, "hello world", rw_repo) + entries = index.reset(new_commit).add([BaseIndexEntry((010644, null_bin_sha, 0, new_file_relapath))], fprogress=self._fprogress_add) + self._assert_entries(entries) + self._assert_fprogress(entries) + assert len(entries) == 1 and entries[0].hexsha != null_hex_sha + + # add symlink + if sys.platform != "win32": + basename = "my_real_symlink" + target = "/etc/that" + link_file = os.path.join(rw_repo.working_tree_dir, basename) + os.symlink(target, link_file) + entries = index.reset(new_commit).add([link_file], fprogress=self._fprogress_add) + self._assert_entries(entries) + self._assert_fprogress(entries) + assert len(entries) == 1 and S_ISLNK(entries[0].mode) + assert S_ISLNK(index.entries[index.entry_key("my_real_symlink", 0)].mode) + + # we expect only the target to be written + assert index.repo.odb.stream(entries[0].binsha).read() == target + # END real symlink test + + # add fake symlink and assure it checks-our as symlink + fake_symlink_relapath = "my_fake_symlink" + link_target = "/etc/that" + fake_symlink_path = self._make_file(fake_symlink_relapath, link_target, rw_repo) + fake_entry = BaseIndexEntry((0120000, null_bin_sha, 0, fake_symlink_relapath)) + entries = index.reset(new_commit).add([fake_entry], fprogress=self._fprogress_add) + self._assert_entries(entries) + self._assert_fprogress(entries) + assert entries[0].hexsha != null_hex_sha + assert len(entries) == 1 and S_ISLNK(entries[0].mode) + + # assure this also works with an alternate method + full_index_entry = IndexEntry.from_base(BaseIndexEntry((0120000, entries[0].binsha, 0, entries[0].path))) + entry_key = index.entry_key(full_index_entry) + index.reset(new_commit) + + assert entry_key not in index.entries + index.entries[entry_key] = full_index_entry + index.write() + index.update() # force reread of entries + new_entry = index.entries[entry_key] + assert S_ISLNK(new_entry.mode) + + # a tree created from this should contain the symlink + tree = index.write_tree() + assert fake_symlink_relapath in tree + index.write() # flush our changes for the checkout + + # checkout the fakelink, should be a link then + assert not S_ISLNK(os.stat(fake_symlink_path)[ST_MODE]) + os.remove(fake_symlink_path) + index.checkout(fake_symlink_path) + + # on windows we will never get symlinks + if os.name == 'nt': + # simlinks should contain the link as text ( which is what a + # symlink actually is ) + open(fake_symlink_path,'rb').read() == link_target + else: + assert S_ISLNK(os.lstat(fake_symlink_path)[ST_MODE]) + + # TEST RENAMING + def assert_mv_rval(rval): + for source, dest in rval: + assert not os.path.exists(source) and os.path.exists(dest) + # END for each renamed item + # END move assertion utility + + self.failUnlessRaises(ValueError, index.move, ['just_one_path']) + # file onto existing file + files = ['AUTHORS', 'LICENSE'] + self.failUnlessRaises(GitCommandError, index.move, files) + + # again, with force + assert_mv_rval(index.move(files, f=True)) + + # files into directory - dry run + paths = ['LICENSE', 'VERSION', 'doc'] + rval = index.move(paths, dry_run=True) + assert len(rval) == 2 + assert os.path.exists(paths[0]) + + # again, no dry run + rval = index.move(paths) + assert_mv_rval(rval) + + # dir into dir + rval = index.move(['doc', 'test']) + assert_mv_rval(rval) + + + # TEST PATH REWRITING + ###################### + count = [0] + def rewriter(entry): + rval = str(count[0]) + count[0] += 1 + return rval + # END rewriter + + def make_paths(): + # two existing ones, one new one + yield 'CHANGES' + yield 'ez_setup.py' + yield index.entries[index.entry_key('README', 0)] + yield index.entries[index.entry_key('.gitignore', 0)] + + for fid in range(3): + fname = 'newfile%i' % fid + open(fname, 'wb').write("abcd") + yield Blob(rw_repo, Blob.NULL_BIN_SHA, 0100644, fname) + # END for each new file + # END path producer + paths = list(make_paths()) + self._assert_entries(index.add(paths, path_rewriter=rewriter)) + + for filenum in range(len(paths)): + assert index.entry_key(str(filenum), 0) in index.entries + + + # TEST RESET ON PATHS + ###################### + arela = "aa" + brela = "bb" + afile = self._make_file(arela, "adata", rw_repo) + bfile = self._make_file(brela, "bdata", rw_repo) + akey = index.entry_key(arela, 0) + bkey = index.entry_key(brela, 0) + keys = (akey, bkey) + absfiles = (afile, bfile) + files = (arela, brela) + + for fkey in keys: + assert not fkey in index.entries + + index.add(files, write=True) + nc = index.commit("2 files committed", head=False) + + for fkey in keys: + assert fkey in index.entries + + # just the index + index.reset(paths=(arela, afile)) + assert not akey in index.entries + assert bkey in index.entries + + # now with working tree - files on disk as well as entries must be recreated + rw_repo.head.commit = nc + for absfile in absfiles: + os.remove(absfile) + + index.reset(working_tree=True, paths=files) + + for fkey in keys: + assert fkey in index.entries + for absfile in absfiles: + assert os.path.isfile(absfile) + + + @with_rw_repo('HEAD') + def test_compare_write_tree(self, rw_repo): + # write all trees and compare them + # its important to have a few submodules in there too + max_count = 25 + count = 0 + for commit in rw_repo.head.commit.traverse(): + if count >= max_count: + break + count += 1 + index = rw_repo.index.reset(commit) + orig_tree = commit.tree + assert index.write_tree() == orig_tree + # END for each commit + + def test_index_new(self): + B = self.rorepo.tree("6d9b1f4f9fa8c9f030e3207e7deacc5d5f8bba4e") + H = self.rorepo.tree("25dca42bac17d511b7e2ebdd9d1d679e7626db5f") + M = self.rorepo.tree("e746f96bcc29238b79118123028ca170adc4ff0f") + + for args in ((B,), (B,H), (B,H,M)): + index = IndexFile.new(self.rorepo, *args) + assert isinstance(index, IndexFile) + # END for each arg tuple + + diff --git a/git/test/test_pack.py b/git/test/test_pack.py index c398fc56..1c308689 100644 --- a/git/test/test_pack.py +++ b/git/test/test_pack.py @@ -4,23 +4,23 @@ # the New BSD License: http://www.opensource.org/licenses/bsd-license.php """Test everything about packs reading and writing""" from lib import ( - TestBase, - with_rw_directory, - with_packs_rw, - fixture_path - ) + TestBase, + with_rw_directory, + with_packs_rw, + fixture_path + ) from git.stream import DeltaApplyReader from git.pack import ( - PackEntity, - PackIndexFile, - PackFile - ) + PackEntity, + PackIndexFile, + PackFile + ) from git.base import ( - OInfo, - OStream, - ) + OInfo, + OStream, + ) from git.fun import delta_types from git.exc import UnsupportedOperation @@ -35,213 +35,213 @@ import tempfile #{ Utilities def bin_sha_from_filename(filename): - return to_bin_sha(os.path.splitext(os.path.basename(filename))[0][5:]) + return to_bin_sha(os.path.splitext(os.path.basename(filename))[0][5:]) #} END utilities class TestPack(TestBase): - - packindexfile_v1 = (fixture_path('packs/pack-c0438c19fb16422b6bbcce24387b3264416d485b.idx'), 1, 67) - packindexfile_v2 = (fixture_path('packs/pack-11fdfa9e156ab73caae3b6da867192221f2089c2.idx'), 2, 30) - packindexfile_v2_3_ascii = (fixture_path('packs/pack-a2bf8e71d8c18879e499335762dd95119d93d9f1.idx'), 2, 42) - packfile_v2_1 = (fixture_path('packs/pack-c0438c19fb16422b6bbcce24387b3264416d485b.pack'), 2, packindexfile_v1[2]) - packfile_v2_2 = (fixture_path('packs/pack-11fdfa9e156ab73caae3b6da867192221f2089c2.pack'), 2, packindexfile_v2[2]) - packfile_v2_3_ascii = (fixture_path('packs/pack-a2bf8e71d8c18879e499335762dd95119d93d9f1.pack'), 2, packindexfile_v2_3_ascii[2]) - - - def _assert_index_file(self, index, version, size): - assert index.packfile_checksum() != index.indexfile_checksum() - assert len(index.packfile_checksum()) == 20 - assert len(index.indexfile_checksum()) == 20 - assert index.version() == version - assert index.size() == size - assert len(index.offsets()) == size - - # get all data of all objects - for oidx in xrange(index.size()): - sha = index.sha(oidx) - assert oidx == index.sha_to_index(sha) - - entry = index.entry(oidx) - assert len(entry) == 3 - - assert entry[0] == index.offset(oidx) - assert entry[1] == sha - assert entry[2] == index.crc(oidx) - - # verify partial sha - for l in (4,8,11,17,20): - assert index.partial_sha_to_index(sha[:l], l*2) == oidx - - # END for each object index in indexfile - self.failUnlessRaises(ValueError, index.partial_sha_to_index, "\0", 2) - - - def _assert_pack_file(self, pack, version, size): - assert pack.version() == 2 - assert pack.size() == size - assert len(pack.checksum()) == 20 - - num_obj = 0 - for obj in pack.stream_iter(): - num_obj += 1 - info = pack.info(obj.pack_offset) - stream = pack.stream(obj.pack_offset) - - assert info.pack_offset == stream.pack_offset - assert info.type_id == stream.type_id - assert hasattr(stream, 'read') - - # it should be possible to read from both streams - assert obj.read() == stream.read() - - streams = pack.collect_streams(obj.pack_offset) - assert streams - - # read the stream - try: - dstream = DeltaApplyReader.new(streams) - except ValueError: - # ignore these, old git versions use only ref deltas, - # which we havent resolved ( as we are without an index ) - # Also ignore non-delta streams - continue - # END get deltastream - - # read all - data = dstream.read() - assert len(data) == dstream.size - - # test seek - dstream.seek(0) - assert dstream.read() == data - - - # read chunks - # NOTE: the current implementation is safe, it basically transfers - # all calls to the underlying memory map - - # END for each object - assert num_obj == size - - - def test_pack_index(self): - # check version 1 and 2 - for indexfile, version, size in (self.packindexfile_v1, self.packindexfile_v2): - index = PackIndexFile(indexfile) - self._assert_index_file(index, version, size) - # END run tests - - def test_pack(self): - # there is this special version 3, but apparently its like 2 ... - for packfile, version, size in (self.packfile_v2_3_ascii, self.packfile_v2_1, self.packfile_v2_2): - pack = PackFile(packfile) - self._assert_pack_file(pack, version, size) - # END for each pack to test - - @with_rw_directory - def test_pack_entity(self, rw_dir): - pack_objs = list() - for packinfo, indexinfo in ( (self.packfile_v2_1, self.packindexfile_v1), - (self.packfile_v2_2, self.packindexfile_v2), - (self.packfile_v2_3_ascii, self.packindexfile_v2_3_ascii)): - packfile, version, size = packinfo - indexfile, version, size = indexinfo - entity = PackEntity(packfile) - assert entity.pack().path() == packfile - assert entity.index().path() == indexfile - pack_objs.extend(entity.stream_iter()) - - count = 0 - for info, stream in izip(entity.info_iter(), entity.stream_iter()): - count += 1 - assert info.binsha == stream.binsha - assert len(info.binsha) == 20 - assert info.type_id == stream.type_id - assert info.size == stream.size - - # we return fully resolved items, which is implied by the sha centric access - assert not info.type_id in delta_types - - # try all calls - assert len(entity.collect_streams(info.binsha)) - oinfo = entity.info(info.binsha) - assert isinstance(oinfo, OInfo) - assert oinfo.binsha is not None - ostream = entity.stream(info.binsha) - assert isinstance(ostream, OStream) - assert ostream.binsha is not None - - # verify the stream - try: - assert entity.is_valid_stream(info.binsha, use_crc=True) - except UnsupportedOperation: - pass - # END ignore version issues - assert entity.is_valid_stream(info.binsha, use_crc=False) - # END for each info, stream tuple - assert count == size - - # END for each entity - - # pack writing - write all packs into one - # index path can be None - pack_path = tempfile.mktemp('', "pack", rw_dir) - index_path = tempfile.mktemp('', 'index', rw_dir) - iteration = 0 - def rewind_streams(): - for obj in pack_objs: - obj.stream.seek(0) - #END utility - for ppath, ipath, num_obj in zip((pack_path, )*2, (index_path, None), (len(pack_objs), None)): - pfile = open(ppath, 'wb') - iwrite = None - if ipath: - ifile = open(ipath, 'wb') - iwrite = ifile.write - #END handle ip - - # make sure we rewind the streams ... we work on the same objects over and over again - if iteration > 0: - rewind_streams() - #END rewind streams - iteration += 1 - - pack_sha, index_sha = PackEntity.write_pack(pack_objs, pfile.write, iwrite, object_count=num_obj) - pfile.close() - assert os.path.getsize(ppath) > 100 - - # verify pack - pf = PackFile(ppath) - assert pf.size() == len(pack_objs) - assert pf.version() == PackFile.pack_version_default - assert pf.checksum() == pack_sha - - # verify index - if ipath is not None: - ifile.close() - assert os.path.getsize(ipath) > 100 - idx = PackIndexFile(ipath) - assert idx.version() == PackIndexFile.index_version_default - assert idx.packfile_checksum() == pack_sha - assert idx.indexfile_checksum() == index_sha - assert idx.size() == len(pack_objs) - #END verify files exist - #END for each packpath, indexpath pair - - # verify the packs throughly - rewind_streams() - entity = PackEntity.create(pack_objs, rw_dir) - count = 0 - for info in entity.info_iter(): - count += 1 - for use_crc in range(2): - assert entity.is_valid_stream(info.binsha, use_crc) - # END for each crc mode - #END for each info - assert count == len(pack_objs) - - - def test_pack_64(self): - # TODO: hex-edit a pack helping us to verify that we can handle 64 byte offsets - # of course without really needing such a huge pack - raise SkipTest() + + packindexfile_v1 = (fixture_path('packs/pack-c0438c19fb16422b6bbcce24387b3264416d485b.idx'), 1, 67) + packindexfile_v2 = (fixture_path('packs/pack-11fdfa9e156ab73caae3b6da867192221f2089c2.idx'), 2, 30) + packindexfile_v2_3_ascii = (fixture_path('packs/pack-a2bf8e71d8c18879e499335762dd95119d93d9f1.idx'), 2, 42) + packfile_v2_1 = (fixture_path('packs/pack-c0438c19fb16422b6bbcce24387b3264416d485b.pack'), 2, packindexfile_v1[2]) + packfile_v2_2 = (fixture_path('packs/pack-11fdfa9e156ab73caae3b6da867192221f2089c2.pack'), 2, packindexfile_v2[2]) + packfile_v2_3_ascii = (fixture_path('packs/pack-a2bf8e71d8c18879e499335762dd95119d93d9f1.pack'), 2, packindexfile_v2_3_ascii[2]) + + + def _assert_index_file(self, index, version, size): + assert index.packfile_checksum() != index.indexfile_checksum() + assert len(index.packfile_checksum()) == 20 + assert len(index.indexfile_checksum()) == 20 + assert index.version() == version + assert index.size() == size + assert len(index.offsets()) == size + + # get all data of all objects + for oidx in xrange(index.size()): + sha = index.sha(oidx) + assert oidx == index.sha_to_index(sha) + + entry = index.entry(oidx) + assert len(entry) == 3 + + assert entry[0] == index.offset(oidx) + assert entry[1] == sha + assert entry[2] == index.crc(oidx) + + # verify partial sha + for l in (4,8,11,17,20): + assert index.partial_sha_to_index(sha[:l], l*2) == oidx + + # END for each object index in indexfile + self.failUnlessRaises(ValueError, index.partial_sha_to_index, "\0", 2) + + + def _assert_pack_file(self, pack, version, size): + assert pack.version() == 2 + assert pack.size() == size + assert len(pack.checksum()) == 20 + + num_obj = 0 + for obj in pack.stream_iter(): + num_obj += 1 + info = pack.info(obj.pack_offset) + stream = pack.stream(obj.pack_offset) + + assert info.pack_offset == stream.pack_offset + assert info.type_id == stream.type_id + assert hasattr(stream, 'read') + + # it should be possible to read from both streams + assert obj.read() == stream.read() + + streams = pack.collect_streams(obj.pack_offset) + assert streams + + # read the stream + try: + dstream = DeltaApplyReader.new(streams) + except ValueError: + # ignore these, old git versions use only ref deltas, + # which we havent resolved ( as we are without an index ) + # Also ignore non-delta streams + continue + # END get deltastream + + # read all + data = dstream.read() + assert len(data) == dstream.size + + # test seek + dstream.seek(0) + assert dstream.read() == data + + + # read chunks + # NOTE: the current implementation is safe, it basically transfers + # all calls to the underlying memory map + + # END for each object + assert num_obj == size + + + def test_pack_index(self): + # check version 1 and 2 + for indexfile, version, size in (self.packindexfile_v1, self.packindexfile_v2): + index = PackIndexFile(indexfile) + self._assert_index_file(index, version, size) + # END run tests + + def test_pack(self): + # there is this special version 3, but apparently its like 2 ... + for packfile, version, size in (self.packfile_v2_3_ascii, self.packfile_v2_1, self.packfile_v2_2): + pack = PackFile(packfile) + self._assert_pack_file(pack, version, size) + # END for each pack to test + + @with_rw_directory + def test_pack_entity(self, rw_dir): + pack_objs = list() + for packinfo, indexinfo in ( (self.packfile_v2_1, self.packindexfile_v1), + (self.packfile_v2_2, self.packindexfile_v2), + (self.packfile_v2_3_ascii, self.packindexfile_v2_3_ascii)): + packfile, version, size = packinfo + indexfile, version, size = indexinfo + entity = PackEntity(packfile) + assert entity.pack().path() == packfile + assert entity.index().path() == indexfile + pack_objs.extend(entity.stream_iter()) + + count = 0 + for info, stream in izip(entity.info_iter(), entity.stream_iter()): + count += 1 + assert info.binsha == stream.binsha + assert len(info.binsha) == 20 + assert info.type_id == stream.type_id + assert info.size == stream.size + + # we return fully resolved items, which is implied by the sha centric access + assert not info.type_id in delta_types + + # try all calls + assert len(entity.collect_streams(info.binsha)) + oinfo = entity.info(info.binsha) + assert isinstance(oinfo, OInfo) + assert oinfo.binsha is not None + ostream = entity.stream(info.binsha) + assert isinstance(ostream, OStream) + assert ostream.binsha is not None + + # verify the stream + try: + assert entity.is_valid_stream(info.binsha, use_crc=True) + except UnsupportedOperation: + pass + # END ignore version issues + assert entity.is_valid_stream(info.binsha, use_crc=False) + # END for each info, stream tuple + assert count == size + + # END for each entity + + # pack writing - write all packs into one + # index path can be None + pack_path = tempfile.mktemp('', "pack", rw_dir) + index_path = tempfile.mktemp('', 'index', rw_dir) + iteration = 0 + def rewind_streams(): + for obj in pack_objs: + obj.stream.seek(0) + #END utility + for ppath, ipath, num_obj in zip((pack_path, )*2, (index_path, None), (len(pack_objs), None)): + pfile = open(ppath, 'wb') + iwrite = None + if ipath: + ifile = open(ipath, 'wb') + iwrite = ifile.write + #END handle ip + + # make sure we rewind the streams ... we work on the same objects over and over again + if iteration > 0: + rewind_streams() + #END rewind streams + iteration += 1 + + pack_sha, index_sha = PackEntity.write_pack(pack_objs, pfile.write, iwrite, object_count=num_obj) + pfile.close() + assert os.path.getsize(ppath) > 100 + + # verify pack + pf = PackFile(ppath) + assert pf.size() == len(pack_objs) + assert pf.version() == PackFile.pack_version_default + assert pf.checksum() == pack_sha + + # verify index + if ipath is not None: + ifile.close() + assert os.path.getsize(ipath) > 100 + idx = PackIndexFile(ipath) + assert idx.version() == PackIndexFile.index_version_default + assert idx.packfile_checksum() == pack_sha + assert idx.indexfile_checksum() == index_sha + assert idx.size() == len(pack_objs) + #END verify files exist + #END for each packpath, indexpath pair + + # verify the packs throughly + rewind_streams() + entity = PackEntity.create(pack_objs, rw_dir) + count = 0 + for info in entity.info_iter(): + count += 1 + for use_crc in range(2): + assert entity.is_valid_stream(info.binsha, use_crc) + # END for each crc mode + #END for each info + assert count == len(pack_objs) + + + def test_pack_64(self): + # TODO: hex-edit a pack helping us to verify that we can handle 64 byte offsets + # of course without really needing such a huge pack + raise SkipTest() diff --git a/git/test/test_remote.py b/git/test/test_remote.py index 30bd1232..87fcd7fe 100644 --- a/git/test/test_remote.py +++ b/git/test/test_remote.py @@ -5,21 +5,21 @@ # the BSD License: http://www.opensource.org/licenses/bsd-license.php from git.test.lib import ( - TestBase, - with_rw_and_rw_remote_repo, - with_rw_repo, - ) + TestBase, + with_rw_and_rw_remote_repo, + with_rw_repo, + ) from git.util import IterableList from git.db.interface import PushInfo, FetchInfo, RemoteProgress from git.remote import * from git.exc import GitCommandError from git.refs import ( - Reference, - TagReference, - RemoteReference, - Head, - SymbolicReference - ) + Reference, + TagReference, + RemoteReference, + Head, + SymbolicReference + ) from nose import SkipTest @@ -32,427 +32,427 @@ import random random.seed(0) class TestRemoteProgress(RemoteProgress): - __slots__ = ( "_seen_lines", "_stages_per_op", '_num_progress_messages') - def __init__(self): - super(TestRemoteProgress, self).__init__() - self._seen_lines = list() - self._stages_per_op = dict() - self._seen_ops = set() - self._num_progress_messages = 0 - - def line_dropped(self, line): - try: - self._seen_lines.remove(line) - except ValueError: - pass - - def __call__(self, message, input=''): - pass - - def update(self, op_code, cur_count, max_count=None, message='', input=''): - # check each stage only comes once - if input: - self._seen_lines.append(input) - #END handle input - op_id = op_code & self.OP_MASK - assert op_id in (self.COUNTING, self.COMPRESSING, self.WRITING) - - self._stages_per_op.setdefault(op_id, 0) - self._stages_per_op[ op_id ] = self._stages_per_op[ op_id ] | (op_code & self.STAGE_MASK) - - if op_code & (self.WRITING|self.END) == (self.WRITING|self.END): - assert message - # END check we get message - - self._num_progress_messages += 1 - - - def make_assertion(self): - # we don't always receive messages - if not self._seen_lines: - return - - # sometimes objects are not compressed which is okay - assert len(self._stages_per_op.keys()) in (2,3) - assert self._stages_per_op - - # must have seen all stages - for op, stages in self._stages_per_op.items(): - assert stages & self.STAGE_MASK == self.STAGE_MASK - # END for each op/stage + __slots__ = ( "_seen_lines", "_stages_per_op", '_num_progress_messages') + def __init__(self): + super(TestRemoteProgress, self).__init__() + self._seen_lines = list() + self._stages_per_op = dict() + self._seen_ops = set() + self._num_progress_messages = 0 + + def line_dropped(self, line): + try: + self._seen_lines.remove(line) + except ValueError: + pass + + def __call__(self, message, input=''): + pass + + def update(self, op_code, cur_count, max_count=None, message='', input=''): + # check each stage only comes once + if input: + self._seen_lines.append(input) + #END handle input + op_id = op_code & self.OP_MASK + assert op_id in (self.COUNTING, self.COMPRESSING, self.WRITING) + + self._stages_per_op.setdefault(op_id, 0) + self._stages_per_op[ op_id ] = self._stages_per_op[ op_id ] | (op_code & self.STAGE_MASK) + + if op_code & (self.WRITING|self.END) == (self.WRITING|self.END): + assert message + # END check we get message + + self._num_progress_messages += 1 + + + def make_assertion(self): + # we don't always receive messages + if not self._seen_lines: + return + + # sometimes objects are not compressed which is okay + assert len(self._stages_per_op.keys()) in (2,3) + assert self._stages_per_op + + # must have seen all stages + for op, stages in self._stages_per_op.items(): + assert stages & self.STAGE_MASK == self.STAGE_MASK + # END for each op/stage - def assert_received_message(self): - assert self._num_progress_messages - + def assert_received_message(self): + assert self._num_progress_messages + class TestRemote(TestBase): - - def _print_fetchhead(self, repo): - fp = open(os.path.join(repo.git_dir, "FETCH_HEAD")) - fp.close() - - - def _do_test_fetch_result(self, results, remote): - # self._print_fetchhead(remote.repo) - assert len(results) > 0 and isinstance(results[0], FetchInfo) - for info in results: - assert isinstance(info.note, basestring) - if isinstance(info.ref, Reference): - assert info.flags != 0 - # END reference type flags handling - assert isinstance(info.ref, (SymbolicReference, Reference)) - if info.flags & (info.FORCED_UPDATE|info.FAST_FORWARD): - assert isinstance(info.old_commit_binsha, str) and len(info.old_commit_binsha) == 20 - else: - assert info.old_commit_binsha is None - # END forced update checking - # END for each info - - def _do_test_push_result(self, results, remote): - assert len(results) > 0 and isinstance(results[0], PushInfo) - for info in results: - assert info.flags - assert isinstance(info.summary, basestring) - if info.old_commit_binsha is not None: - assert isinstance(info.old_commit_binsha, str) and len(info.old_commit_binsha) == 20 - if info.flags & info.ERROR: - has_one = False - for bitflag in (info.REJECTED, info.REMOTE_REJECTED, info.REMOTE_FAILURE): - has_one |= bool(info.flags & bitflag) - # END for each bitflag - assert has_one - else: - # there must be a remote commit - if info.flags & info.DELETED == 0: - assert isinstance(info.local_ref, Reference) - else: - assert info.local_ref is None - assert type(info.remote_ref) in (TagReference, RemoteReference) - # END error checking - # END for each info - - def _commit_random_file(self, repo): - #Create a file with a random name and random data and commit it to repo. - # Return the commited absolute file path - index = repo.index - new_file = self._make_file(os.path.basename(tempfile.mktemp()),str(random.random()), repo) - index.add([new_file]) - index.commit("Committing %s" % new_file) - return new_file - - def _do_test_fetch(self,remote, rw_repo, remote_repo): - def fetch_and_test(remote, **kwargs): - progress = TestRemoteProgress() - kwargs['progress'] = progress - res = remote.fetch(**kwargs) - progress.make_assertion() - self._do_test_fetch_result(res, remote) - return res - # END fetch and check - - def get_info(res, remote, name): - return res["%s/%s"%(remote,name)] - - # put remote head to master as it is garantueed to exist - remote_repo.head.reference = remote_repo.heads.master - - res = fetch_and_test(remote) - # all uptodate - for info in res: - assert info.flags & info.HEAD_UPTODATE - - # rewind remote head to trigger rejection - # index must be false as remote is a bare repo - rhead = remote_repo.head - remote_commit = rhead.commit - rhead.reset("HEAD~2", index=False) - res = fetch_and_test(remote) - mkey = "%s/%s"%(remote,'master') - master_info = res[mkey] - assert master_info.flags & FetchInfo.FORCED_UPDATE and master_info.note is not None - - # normal fast forward - set head back to previous one - rhead.commit = remote_commit - res = fetch_and_test(remote) - assert res[mkey].flags & FetchInfo.FAST_FORWARD - - # new remote branch - new_remote_branch = Head.create(remote_repo, "new_branch") - res = fetch_and_test(remote) - new_branch_info = get_info(res, remote, new_remote_branch) - assert new_branch_info.flags & FetchInfo.NEW_HEAD - - # remote branch rename ( causes creation of a new one locally ) - new_remote_branch.rename("other_branch_name") - res = fetch_and_test(remote) - other_branch_info = get_info(res, remote, new_remote_branch) - assert other_branch_info.ref.commit == new_branch_info.ref.commit - - # remove new branch - Head.delete(new_remote_branch.repo, new_remote_branch) - res = fetch_and_test(remote) - # deleted remote will not be fetched - self.failUnlessRaises(IndexError, get_info, res, remote, new_remote_branch) - - # prune stale tracking branches - stale_refs = remote.stale_refs - assert len(stale_refs) == 2 and isinstance(stale_refs[0], RemoteReference) - RemoteReference.delete(rw_repo, *stale_refs) - - # test single branch fetch with refspec including target remote - res = fetch_and_test(remote, refspec="master:refs/remotes/%s/master"%remote) - assert len(res) == 1 and get_info(res, remote, 'master') - - # ... with respec and no target - res = fetch_and_test(remote, refspec='master') - assert len(res) == 1 - - # add new tag reference - rtag = TagReference.create(remote_repo, "1.0-RV_hello.there") - res = fetch_and_test(remote, tags=True) - tinfo = res[str(rtag)] - assert isinstance(tinfo.ref, TagReference) and tinfo.ref.commit == rtag.commit - assert tinfo.flags & tinfo.NEW_TAG - - # adjust tag commit - Reference.set_object(rtag, rhead.commit.parents[0].parents[0]) - res = fetch_and_test(remote, tags=True) - tinfo = res[str(rtag)] - assert tinfo.commit == rtag.commit - assert tinfo.flags & tinfo.TAG_UPDATE - - # delete remote tag - local one will stay - TagReference.delete(remote_repo, rtag) - res = fetch_and_test(remote, tags=True) - self.failUnlessRaises(IndexError, get_info, res, remote, str(rtag)) - - # provoke to receive actual objects to see what kind of output we have to - # expect. For that we need a remote transport protocol - # Create a new UN-shared repo and fetch into it after we pushed a change - # to the shared repo - other_repo_dir = tempfile.mktemp("other_repo") - # must clone with a local path for the repo implementation not to freak out - # as it wants local paths only ( which I can understand ) - other_repo = remote_repo.clone(other_repo_dir, shared=False) - remote_repo_url = "git://localhost%s"%remote_repo.git_dir - - # put origin to git-url - other_origin = other_repo.remotes.origin - other_origin.config_writer.set("url", remote_repo_url) - # it automatically creates alternates as remote_repo is shared as well. - # It will use the transport though and ignore alternates when fetching - # assert not other_repo.alternates # this would fail - - # assure we are in the right state - rw_repo.head.reset(remote.refs.master, working_tree=True) - try: - self._commit_random_file(rw_repo) - remote.push(rw_repo.head.reference) - - # here I would expect to see remote-information about packing - # objects and so on. Unfortunately, this does not happen - # if we are redirecting the output - git explicitly checks for this - # and only provides progress information to ttys - res = fetch_and_test(other_origin) - finally: - shutil.rmtree(other_repo_dir) - # END test and cleanup - - def _verify_push_and_pull(self,remote, rw_repo, remote_repo): - # push our changes - lhead = rw_repo.head - lindex = rw_repo.index - # assure we are on master and it is checked out where the remote is - try: - lhead.reference = rw_repo.heads.master - except AttributeError: - # if the author is on a non-master branch, the clones might not have - # a local master yet. We simply create it - lhead.reference = rw_repo.create_head('master') - # END master handling - lhead.reset(remote.refs.master, working_tree=True) - - # push without spec should fail ( without further configuration ) - # well, works nicely - # self.failUnlessRaises(GitCommandError, remote.push) - - # simple file push - self._commit_random_file(rw_repo) - progress = TestRemoteProgress() - res = remote.push(lhead.reference, progress) - assert isinstance(res, IterableList) - self._do_test_push_result(res, remote) - progress.make_assertion() - - # rejected - undo last commit - lhead.reset("HEAD~1") - res = remote.push(lhead.reference) - assert res[0].flags & PushInfo.ERROR - assert res[0].flags & PushInfo.REJECTED - self._do_test_push_result(res, remote) - - # force rejected pull - res = remote.push('+%s' % lhead.reference) - assert res[0].flags & PushInfo.ERROR == 0 - assert res[0].flags & PushInfo.FORCED_UPDATE - self._do_test_push_result(res, remote) - - # invalid refspec - res = remote.push("hellothere") - assert len(res) == 0 - - # push new tags - progress = TestRemoteProgress() - to_be_updated = "my_tag.1.0RV" - new_tag = TagReference.create(rw_repo, to_be_updated) - other_tag = TagReference.create(rw_repo, "my_obj_tag.2.1aRV", message="my message") - res = remote.push(progress=progress, tags=True) - assert res[-1].flags & PushInfo.NEW_TAG - progress.make_assertion() - self._do_test_push_result(res, remote) - - # update push new tags - # Rejection is default - new_tag = TagReference.create(rw_repo, to_be_updated, ref='HEAD~1', force=True) - res = remote.push(tags=True) - self._do_test_push_result(res, remote) - assert res[-1].flags & PushInfo.REJECTED and res[-1].flags & PushInfo.ERROR - - # push force this tag - res = remote.push("+%s" % new_tag.path) - assert res[-1].flags & PushInfo.ERROR == 0 and res[-1].flags & PushInfo.FORCED_UPDATE - - # delete tag - have to do it using refspec - res = remote.push(":%s" % new_tag.path) - self._do_test_push_result(res, remote) - assert res[0].flags & PushInfo.DELETED - # Currently progress is not properly transferred, especially not using - # the git daemon - # progress.assert_received_message() - - # push new branch - new_head = Head.create(rw_repo, "my_new_branch") - progress = TestRemoteProgress() - res = remote.push(new_head, progress) - assert res[0].flags & PushInfo.NEW_HEAD - progress.make_assertion() - self._do_test_push_result(res, remote) - - # delete new branch on the remote end and locally - res = remote.push(":%s" % new_head.path) - self._do_test_push_result(res, remote) - Head.delete(rw_repo, new_head) - assert res[-1].flags & PushInfo.DELETED - - # --all - res = remote.push(all=True) - self._do_test_push_result(res, remote) - - remote.pull('master') - - # cleanup - delete created tags and branches as we are in an innerloop on - # the same repository - TagReference.delete(rw_repo, new_tag, other_tag) - remote.push(":%s" % other_tag.path) + + def _print_fetchhead(self, repo): + fp = open(os.path.join(repo.git_dir, "FETCH_HEAD")) + fp.close() + + + def _do_test_fetch_result(self, results, remote): + # self._print_fetchhead(remote.repo) + assert len(results) > 0 and isinstance(results[0], FetchInfo) + for info in results: + assert isinstance(info.note, basestring) + if isinstance(info.ref, Reference): + assert info.flags != 0 + # END reference type flags handling + assert isinstance(info.ref, (SymbolicReference, Reference)) + if info.flags & (info.FORCED_UPDATE|info.FAST_FORWARD): + assert isinstance(info.old_commit_binsha, str) and len(info.old_commit_binsha) == 20 + else: + assert info.old_commit_binsha is None + # END forced update checking + # END for each info + + def _do_test_push_result(self, results, remote): + assert len(results) > 0 and isinstance(results[0], PushInfo) + for info in results: + assert info.flags + assert isinstance(info.summary, basestring) + if info.old_commit_binsha is not None: + assert isinstance(info.old_commit_binsha, str) and len(info.old_commit_binsha) == 20 + if info.flags & info.ERROR: + has_one = False + for bitflag in (info.REJECTED, info.REMOTE_REJECTED, info.REMOTE_FAILURE): + has_one |= bool(info.flags & bitflag) + # END for each bitflag + assert has_one + else: + # there must be a remote commit + if info.flags & info.DELETED == 0: + assert isinstance(info.local_ref, Reference) + else: + assert info.local_ref is None + assert type(info.remote_ref) in (TagReference, RemoteReference) + # END error checking + # END for each info + + def _commit_random_file(self, repo): + #Create a file with a random name and random data and commit it to repo. + # Return the commited absolute file path + index = repo.index + new_file = self._make_file(os.path.basename(tempfile.mktemp()),str(random.random()), repo) + index.add([new_file]) + index.commit("Committing %s" % new_file) + return new_file + + def _do_test_fetch(self,remote, rw_repo, remote_repo): + def fetch_and_test(remote, **kwargs): + progress = TestRemoteProgress() + kwargs['progress'] = progress + res = remote.fetch(**kwargs) + progress.make_assertion() + self._do_test_fetch_result(res, remote) + return res + # END fetch and check + + def get_info(res, remote, name): + return res["%s/%s"%(remote,name)] + + # put remote head to master as it is garantueed to exist + remote_repo.head.reference = remote_repo.heads.master + + res = fetch_and_test(remote) + # all uptodate + for info in res: + assert info.flags & info.HEAD_UPTODATE + + # rewind remote head to trigger rejection + # index must be false as remote is a bare repo + rhead = remote_repo.head + remote_commit = rhead.commit + rhead.reset("HEAD~2", index=False) + res = fetch_and_test(remote) + mkey = "%s/%s"%(remote,'master') + master_info = res[mkey] + assert master_info.flags & FetchInfo.FORCED_UPDATE and master_info.note is not None + + # normal fast forward - set head back to previous one + rhead.commit = remote_commit + res = fetch_and_test(remote) + assert res[mkey].flags & FetchInfo.FAST_FORWARD + + # new remote branch + new_remote_branch = Head.create(remote_repo, "new_branch") + res = fetch_and_test(remote) + new_branch_info = get_info(res, remote, new_remote_branch) + assert new_branch_info.flags & FetchInfo.NEW_HEAD + + # remote branch rename ( causes creation of a new one locally ) + new_remote_branch.rename("other_branch_name") + res = fetch_and_test(remote) + other_branch_info = get_info(res, remote, new_remote_branch) + assert other_branch_info.ref.commit == new_branch_info.ref.commit + + # remove new branch + Head.delete(new_remote_branch.repo, new_remote_branch) + res = fetch_and_test(remote) + # deleted remote will not be fetched + self.failUnlessRaises(IndexError, get_info, res, remote, new_remote_branch) + + # prune stale tracking branches + stale_refs = remote.stale_refs + assert len(stale_refs) == 2 and isinstance(stale_refs[0], RemoteReference) + RemoteReference.delete(rw_repo, *stale_refs) + + # test single branch fetch with refspec including target remote + res = fetch_and_test(remote, refspec="master:refs/remotes/%s/master"%remote) + assert len(res) == 1 and get_info(res, remote, 'master') + + # ... with respec and no target + res = fetch_and_test(remote, refspec='master') + assert len(res) == 1 + + # add new tag reference + rtag = TagReference.create(remote_repo, "1.0-RV_hello.there") + res = fetch_and_test(remote, tags=True) + tinfo = res[str(rtag)] + assert isinstance(tinfo.ref, TagReference) and tinfo.ref.commit == rtag.commit + assert tinfo.flags & tinfo.NEW_TAG + + # adjust tag commit + Reference.set_object(rtag, rhead.commit.parents[0].parents[0]) + res = fetch_and_test(remote, tags=True) + tinfo = res[str(rtag)] + assert tinfo.commit == rtag.commit + assert tinfo.flags & tinfo.TAG_UPDATE + + # delete remote tag - local one will stay + TagReference.delete(remote_repo, rtag) + res = fetch_and_test(remote, tags=True) + self.failUnlessRaises(IndexError, get_info, res, remote, str(rtag)) + + # provoke to receive actual objects to see what kind of output we have to + # expect. For that we need a remote transport protocol + # Create a new UN-shared repo and fetch into it after we pushed a change + # to the shared repo + other_repo_dir = tempfile.mktemp("other_repo") + # must clone with a local path for the repo implementation not to freak out + # as it wants local paths only ( which I can understand ) + other_repo = remote_repo.clone(other_repo_dir, shared=False) + remote_repo_url = "git://localhost%s"%remote_repo.git_dir + + # put origin to git-url + other_origin = other_repo.remotes.origin + other_origin.config_writer.set("url", remote_repo_url) + # it automatically creates alternates as remote_repo is shared as well. + # It will use the transport though and ignore alternates when fetching + # assert not other_repo.alternates # this would fail + + # assure we are in the right state + rw_repo.head.reset(remote.refs.master, working_tree=True) + try: + self._commit_random_file(rw_repo) + remote.push(rw_repo.head.reference) + + # here I would expect to see remote-information about packing + # objects and so on. Unfortunately, this does not happen + # if we are redirecting the output - git explicitly checks for this + # and only provides progress information to ttys + res = fetch_and_test(other_origin) + finally: + shutil.rmtree(other_repo_dir) + # END test and cleanup + + def _verify_push_and_pull(self,remote, rw_repo, remote_repo): + # push our changes + lhead = rw_repo.head + lindex = rw_repo.index + # assure we are on master and it is checked out where the remote is + try: + lhead.reference = rw_repo.heads.master + except AttributeError: + # if the author is on a non-master branch, the clones might not have + # a local master yet. We simply create it + lhead.reference = rw_repo.create_head('master') + # END master handling + lhead.reset(remote.refs.master, working_tree=True) + + # push without spec should fail ( without further configuration ) + # well, works nicely + # self.failUnlessRaises(GitCommandError, remote.push) + + # simple file push + self._commit_random_file(rw_repo) + progress = TestRemoteProgress() + res = remote.push(lhead.reference, progress) + assert isinstance(res, IterableList) + self._do_test_push_result(res, remote) + progress.make_assertion() + + # rejected - undo last commit + lhead.reset("HEAD~1") + res = remote.push(lhead.reference) + assert res[0].flags & PushInfo.ERROR + assert res[0].flags & PushInfo.REJECTED + self._do_test_push_result(res, remote) + + # force rejected pull + res = remote.push('+%s' % lhead.reference) + assert res[0].flags & PushInfo.ERROR == 0 + assert res[0].flags & PushInfo.FORCED_UPDATE + self._do_test_push_result(res, remote) + + # invalid refspec + res = remote.push("hellothere") + assert len(res) == 0 + + # push new tags + progress = TestRemoteProgress() + to_be_updated = "my_tag.1.0RV" + new_tag = TagReference.create(rw_repo, to_be_updated) + other_tag = TagReference.create(rw_repo, "my_obj_tag.2.1aRV", message="my message") + res = remote.push(progress=progress, tags=True) + assert res[-1].flags & PushInfo.NEW_TAG + progress.make_assertion() + self._do_test_push_result(res, remote) + + # update push new tags + # Rejection is default + new_tag = TagReference.create(rw_repo, to_be_updated, ref='HEAD~1', force=True) + res = remote.push(tags=True) + self._do_test_push_result(res, remote) + assert res[-1].flags & PushInfo.REJECTED and res[-1].flags & PushInfo.ERROR + + # push force this tag + res = remote.push("+%s" % new_tag.path) + assert res[-1].flags & PushInfo.ERROR == 0 and res[-1].flags & PushInfo.FORCED_UPDATE + + # delete tag - have to do it using refspec + res = remote.push(":%s" % new_tag.path) + self._do_test_push_result(res, remote) + assert res[0].flags & PushInfo.DELETED + # Currently progress is not properly transferred, especially not using + # the git daemon + # progress.assert_received_message() + + # push new branch + new_head = Head.create(rw_repo, "my_new_branch") + progress = TestRemoteProgress() + res = remote.push(new_head, progress) + assert res[0].flags & PushInfo.NEW_HEAD + progress.make_assertion() + self._do_test_push_result(res, remote) + + # delete new branch on the remote end and locally + res = remote.push(":%s" % new_head.path) + self._do_test_push_result(res, remote) + Head.delete(rw_repo, new_head) + assert res[-1].flags & PushInfo.DELETED + + # --all + res = remote.push(all=True) + self._do_test_push_result(res, remote) + + remote.pull('master') + + # cleanup - delete created tags and branches as we are in an innerloop on + # the same repository + TagReference.delete(rw_repo, new_tag, other_tag) + remote.push(":%s" % other_tag.path) - def test_todo(self): - # If you see this, plesase remind yourself, that all this needs to be run - # per repository type ! - raise SkipTest("todo") - + def test_todo(self): + # If you see this, plesase remind yourself, that all this needs to be run + # per repository type ! + raise SkipTest("todo") + - @with_rw_and_rw_remote_repo('0.1.6') - def test_base(self, rw_repo, remote_repo): - num_remotes = 0 - remote_set = set() - ran_fetch_test = False - - for remote in rw_repo.remotes: - num_remotes += 1 - assert remote == remote - assert str(remote) != repr(remote) - remote_set.add(remote) - remote_set.add(remote) # should already exist - - # REFS - refs = remote.refs - assert refs - for ref in refs: - assert ref.remote_name == remote.name - assert ref.remote_head - # END for each ref - - # OPTIONS - # cannot use 'fetch' key anymore as it is now a method - for opt in ("url", ): - val = getattr(remote, opt) - reader = remote.config_reader - assert reader.get(opt) == val - assert reader.get_value(opt, None) == val - - # unable to write with a reader - self.failUnlessRaises(IOError, reader.set, opt, "test") - - # change value - writer = remote.config_writer - new_val = "myval" - writer.set(opt, new_val) - assert writer.get(opt) == new_val - writer.set(opt, val) - assert writer.get(opt) == val - del(writer) - assert getattr(remote, opt) == val - # END for each default option key - - # RENAME - other_name = "totally_other_name" - prev_name = remote.name - assert remote.rename(other_name) == remote - assert prev_name != remote.name - # multiple times - for time in range(2): - assert remote.rename(prev_name).name == prev_name - # END for each rename ( back to prev_name ) - - # PUSH/PULL TESTING - self._verify_push_and_pull(remote, rw_repo, remote_repo) - - # FETCH TESTING - # Only for remotes - local cases are the same or less complicated - # as additional progress information will never be emitted - if remote.name == "daemon_origin": - self._do_test_fetch(remote, rw_repo, remote_repo) - ran_fetch_test = True - # END fetch test - - remote.update() - # END for each remote - - assert ran_fetch_test - assert num_remotes - assert num_remotes == len(remote_set) - - origin = rw_repo.remote('origin') - assert origin == rw_repo.remotes.origin - - @with_rw_repo('HEAD', bare=True) - def test_creation_and_removal(self, bare_rw_repo): - new_name = "test_new_one" - arg_list = (new_name, "git@server:hello.git") - remote = Remote.create(bare_rw_repo, *arg_list ) - assert remote.name == "test_new_one" - assert remote in bare_rw_repo.remotes - - # create same one again - self.failUnlessRaises(GitCommandError, Remote.create, bare_rw_repo, *arg_list) - - Remote.remove(bare_rw_repo, new_name) - - for remote in bare_rw_repo.remotes: - if remote.name == new_name: - raise AssertionError("Remote removal failed") - # END if deleted remote matches existing remote's name - # END for each remote - - - + @with_rw_and_rw_remote_repo('0.1.6') + def test_base(self, rw_repo, remote_repo): + num_remotes = 0 + remote_set = set() + ran_fetch_test = False + + for remote in rw_repo.remotes: + num_remotes += 1 + assert remote == remote + assert str(remote) != repr(remote) + remote_set.add(remote) + remote_set.add(remote) # should already exist + + # REFS + refs = remote.refs + assert refs + for ref in refs: + assert ref.remote_name == remote.name + assert ref.remote_head + # END for each ref + + # OPTIONS + # cannot use 'fetch' key anymore as it is now a method + for opt in ("url", ): + val = getattr(remote, opt) + reader = remote.config_reader + assert reader.get(opt) == val + assert reader.get_value(opt, None) == val + + # unable to write with a reader + self.failUnlessRaises(IOError, reader.set, opt, "test") + + # change value + writer = remote.config_writer + new_val = "myval" + writer.set(opt, new_val) + assert writer.get(opt) == new_val + writer.set(opt, val) + assert writer.get(opt) == val + del(writer) + assert getattr(remote, opt) == val + # END for each default option key + + # RENAME + other_name = "totally_other_name" + prev_name = remote.name + assert remote.rename(other_name) == remote + assert prev_name != remote.name + # multiple times + for time in range(2): + assert remote.rename(prev_name).name == prev_name + # END for each rename ( back to prev_name ) + + # PUSH/PULL TESTING + self._verify_push_and_pull(remote, rw_repo, remote_repo) + + # FETCH TESTING + # Only for remotes - local cases are the same or less complicated + # as additional progress information will never be emitted + if remote.name == "daemon_origin": + self._do_test_fetch(remote, rw_repo, remote_repo) + ran_fetch_test = True + # END fetch test + + remote.update() + # END for each remote + + assert ran_fetch_test + assert num_remotes + assert num_remotes == len(remote_set) + + origin = rw_repo.remote('origin') + assert origin == rw_repo.remotes.origin + + @with_rw_repo('HEAD', bare=True) + def test_creation_and_removal(self, bare_rw_repo): + new_name = "test_new_one" + arg_list = (new_name, "git@server:hello.git") + remote = Remote.create(bare_rw_repo, *arg_list ) + assert remote.name == "test_new_one" + assert remote in bare_rw_repo.remotes + + # create same one again + self.failUnlessRaises(GitCommandError, Remote.create, bare_rw_repo, *arg_list) + + Remote.remove(bare_rw_repo, new_name) + + for remote in bare_rw_repo.remotes: + if remote.name == new_name: + raise AssertionError("Remote removal failed") + # END if deleted remote matches existing remote's name + # END for each remote + + + diff --git a/git/test/test_stats.py b/git/test/test_stats.py index 27be6a77..5210e7bc 100644 --- a/git/test/test_stats.py +++ b/git/test/test_stats.py @@ -5,10 +5,10 @@ # the BSD License: http://www.opensource.org/licenses/bsd-license.php from git.test.lib import ( - TestBase, - fixture, - assert_equal - ) + TestBase, + fixture, + assert_equal + ) from git.util import Stats class TestStats(TestBase): diff --git a/git/test/test_stream.py b/git/test/test_stream.py index 8d7a5f9a..7af652b7 100644 --- a/git/test/test_stream.py +++ b/git/test/test_stream.py @@ -4,24 +4,24 @@ # the New BSD License: http://www.opensource.org/licenses/bsd-license.php """Test for object db""" from lib import ( - TestBase, - DummyStream, - Sha1Writer, - make_bytes, - make_object, - fixture_path - ) + TestBase, + DummyStream, + Sha1Writer, + make_bytes, + make_object, + fixture_path + ) from git.stream import * from git.util import ( - NULL_HEX_SHA, - hex_to_bin - ) + NULL_HEX_SHA, + hex_to_bin + ) from git.util import zlib from git.typ import ( - str_blob_type - ) + str_blob_type + ) from git.db.py.loose import PureLooseObjectODB import time @@ -32,124 +32,124 @@ import os class TestStream(TestBase): - """Test stream classes""" - - data_sizes = (15, 10000, 1000*1024+512) - - def _assert_stream_reader(self, stream, cdata, rewind_stream=lambda s: None): - """Make stream tests - the orig_stream is seekable, allowing it to be - rewound and reused - :param cdata: the data we expect to read from stream, the contents - :param rewind_stream: function called to rewind the stream to make it ready - for reuse""" - ns = 10 - assert len(cdata) > ns-1, "Data must be larger than %i, was %i" % (ns, len(cdata)) - - # read in small steps - ss = len(cdata) / ns - for i in range(ns): - data = stream.read(ss) - chunk = cdata[i*ss:(i+1)*ss] - assert data == chunk - # END for each step - rest = stream.read() - if rest: - assert rest == cdata[-len(rest):] - # END handle rest - - if isinstance(stream, DecompressMemMapReader): - assert len(stream.data()) == stream.compressed_bytes_read() - # END handle special type - - rewind_stream(stream) - - # read everything - rdata = stream.read() - assert rdata == cdata - - if isinstance(stream, DecompressMemMapReader): - assert len(stream.data()) == stream.compressed_bytes_read() - # END handle special type - - def test_decompress_reader(self): - for close_on_deletion in range(2): - for with_size in range(2): - for ds in self.data_sizes: - cdata = make_bytes(ds, randomize=False) - - # zdata = zipped actual data - # cdata = original content data - - # create reader - if with_size: - # need object data - zdata = zlib.compress(make_object(str_blob_type, cdata)) - type, size, reader = DecompressMemMapReader.new(zdata, close_on_deletion) - assert size == len(cdata) - assert type == str_blob_type - - # even if we don't set the size, it will be set automatically on first read - test_reader = DecompressMemMapReader(zdata, close_on_deletion=False) - assert test_reader._s == len(cdata) - else: - # here we need content data - zdata = zlib.compress(cdata) - reader = DecompressMemMapReader(zdata, close_on_deletion, len(cdata)) - assert reader._s == len(cdata) - # END get reader - - self._assert_stream_reader(reader, cdata, lambda r: r.seek(0)) - - # put in a dummy stream for closing - dummy = DummyStream() - reader._m = dummy - - assert not dummy.closed - del(reader) - assert dummy.closed == close_on_deletion - # END for each datasize - # END whether size should be used - # END whether stream should be closed when deleted - - def test_sha_writer(self): - writer = Sha1Writer() - assert 2 == writer.write("hi") - assert len(writer.sha(as_hex=1)) == 40 - assert len(writer.sha(as_hex=0)) == 20 - - # make sure it does something ;) - prev_sha = writer.sha() - writer.write("hi again") - assert writer.sha() != prev_sha - - def test_compressed_writer(self): - for ds in self.data_sizes: - fd, path = tempfile.mkstemp() - ostream = FDCompressedSha1Writer(fd) - data = make_bytes(ds, randomize=False) - - # for now, just a single write, code doesn't care about chunking - assert len(data) == ostream.write(data) - ostream.close() - - # its closed already - self.failUnlessRaises(OSError, os.close, fd) - - # read everything back, compare to data we zip - fd = os.open(path, os.O_RDONLY|getattr(os, 'O_BINARY', 0)) - written_data = os.read(fd, os.path.getsize(path)) - assert len(written_data) == os.path.getsize(path) - os.close(fd) - assert written_data == zlib.compress(data, 1) # best speed - - os.remove(path) - # END for each os - - def test_decompress_reader_special_case(self): - odb = PureLooseObjectODB(fixture_path('objects')) - ostream = odb.stream(hex_to_bin('7bb839852ed5e3a069966281bb08d50012fb309b')) - - # if there is a bug, we will be missing one byte exactly ! - data = ostream.read() - assert len(data) == ostream.size - + """Test stream classes""" + + data_sizes = (15, 10000, 1000*1024+512) + + def _assert_stream_reader(self, stream, cdata, rewind_stream=lambda s: None): + """Make stream tests - the orig_stream is seekable, allowing it to be + rewound and reused + :param cdata: the data we expect to read from stream, the contents + :param rewind_stream: function called to rewind the stream to make it ready + for reuse""" + ns = 10 + assert len(cdata) > ns-1, "Data must be larger than %i, was %i" % (ns, len(cdata)) + + # read in small steps + ss = len(cdata) / ns + for i in range(ns): + data = stream.read(ss) + chunk = cdata[i*ss:(i+1)*ss] + assert data == chunk + # END for each step + rest = stream.read() + if rest: + assert rest == cdata[-len(rest):] + # END handle rest + + if isinstance(stream, DecompressMemMapReader): + assert len(stream.data()) == stream.compressed_bytes_read() + # END handle special type + + rewind_stream(stream) + + # read everything + rdata = stream.read() + assert rdata == cdata + + if isinstance(stream, DecompressMemMapReader): + assert len(stream.data()) == stream.compressed_bytes_read() + # END handle special type + + def test_decompress_reader(self): + for close_on_deletion in range(2): + for with_size in range(2): + for ds in self.data_sizes: + cdata = make_bytes(ds, randomize=False) + + # zdata = zipped actual data + # cdata = original content data + + # create reader + if with_size: + # need object data + zdata = zlib.compress(make_object(str_blob_type, cdata)) + type, size, reader = DecompressMemMapReader.new(zdata, close_on_deletion) + assert size == len(cdata) + assert type == str_blob_type + + # even if we don't set the size, it will be set automatically on first read + test_reader = DecompressMemMapReader(zdata, close_on_deletion=False) + assert test_reader._s == len(cdata) + else: + # here we need content data + zdata = zlib.compress(cdata) + reader = DecompressMemMapReader(zdata, close_on_deletion, len(cdata)) + assert reader._s == len(cdata) + # END get reader + + self._assert_stream_reader(reader, cdata, lambda r: r.seek(0)) + + # put in a dummy stream for closing + dummy = DummyStream() + reader._m = dummy + + assert not dummy.closed + del(reader) + assert dummy.closed == close_on_deletion + # END for each datasize + # END whether size should be used + # END whether stream should be closed when deleted + + def test_sha_writer(self): + writer = Sha1Writer() + assert 2 == writer.write("hi") + assert len(writer.sha(as_hex=1)) == 40 + assert len(writer.sha(as_hex=0)) == 20 + + # make sure it does something ;) + prev_sha = writer.sha() + writer.write("hi again") + assert writer.sha() != prev_sha + + def test_compressed_writer(self): + for ds in self.data_sizes: + fd, path = tempfile.mkstemp() + ostream = FDCompressedSha1Writer(fd) + data = make_bytes(ds, randomize=False) + + # for now, just a single write, code doesn't care about chunking + assert len(data) == ostream.write(data) + ostream.close() + + # its closed already + self.failUnlessRaises(OSError, os.close, fd) + + # read everything back, compare to data we zip + fd = os.open(path, os.O_RDONLY|getattr(os, 'O_BINARY', 0)) + written_data = os.read(fd, os.path.getsize(path)) + assert len(written_data) == os.path.getsize(path) + os.close(fd) + assert written_data == zlib.compress(data, 1) # best speed + + os.remove(path) + # END for each os + + def test_decompress_reader_special_case(self): + odb = PureLooseObjectODB(fixture_path('objects')) + ostream = odb.stream(hex_to_bin('7bb839852ed5e3a069966281bb08d50012fb309b')) + + # if there is a bug, we will be missing one byte exactly ! + data = ostream.read() + assert len(data) == ostream.size + diff --git a/git/test/test_util.py b/git/test/test_util.py index 7cfcad3f..d2ca8bf2 100644 --- a/git/test/test_util.py +++ b/git/test/test_util.py @@ -16,250 +16,250 @@ from git.cmd import dashify import time from git.util import ( - to_hex_sha, - to_bin_sha, - NULL_HEX_SHA, - LockedFD, - Actor, - IterableList - ) + to_hex_sha, + to_bin_sha, + NULL_HEX_SHA, + LockedFD, + Actor, + IterableList + ) class TestIterableMember(object): - """A member of an iterable list""" - __slots__ = ("name", "prefix_name") - - def __init__(self, name): - self.name = name - self.prefix_name = name - + """A member of an iterable list""" + __slots__ = ("name", "prefix_name") + + def __init__(self, name): + self.name = name + self.prefix_name = name + class TestUtils(TestBase): - def setup(self): - self.testdict = { - "string": "42", - "int": 42, - "array": [ 42 ], - } + def setup(self): + self.testdict = { + "string": "42", + "int": 42, + "array": [ 42 ], + } - def test_it_should_dashify(self): - assert 'this-is-my-argument' == dashify('this_is_my_argument') - assert 'foo' == dashify('foo') - - - def test_lock_file(self): - my_file = tempfile.mktemp() - lock_file = LockFile(my_file) - assert not lock_file._has_lock() - # release lock we don't have - fine - lock_file._release_lock() - - # get lock - lock_file._obtain_lock_or_raise() - assert lock_file._has_lock() - - # concurrent access - other_lock_file = LockFile(my_file) - assert not other_lock_file._has_lock() - self.failUnlessRaises(IOError, other_lock_file._obtain_lock_or_raise) - - lock_file._release_lock() - assert not lock_file._has_lock() - - other_lock_file._obtain_lock_or_raise() - self.failUnlessRaises(IOError, lock_file._obtain_lock_or_raise) - - # auto-release on destruction - del(other_lock_file) - lock_file._obtain_lock_or_raise() - lock_file._release_lock() - - def test_blocking_lock_file(self): - my_file = tempfile.mktemp() - lock_file = BlockingLockFile(my_file) - lock_file._obtain_lock() - - # next one waits for the lock - start = time.time() - wait_time = 0.1 - wait_lock = BlockingLockFile(my_file, 0.05, wait_time) - self.failUnlessRaises(IOError, wait_lock._obtain_lock) - elapsed = time.time() - start - assert elapsed <= wait_time + 0.02 # some extra time it may cost - - def test_user_id(self): - assert '@' in get_user_id() - - def test_parse_date(self): - # test all supported formats - def assert_rval(rval, veri_time, offset=0): - assert len(rval) == 2 - assert isinstance(rval[0], int) and isinstance(rval[1], int) - assert rval[0] == veri_time - assert rval[1] == offset - - # now that we are here, test our conversion functions as well - utctz = altz_to_utctz_str(offset) - assert isinstance(utctz, basestring) - assert utctz_to_altz(verify_utctz(utctz)) == offset - # END assert rval utility - - rfc = ("Thu, 07 Apr 2005 22:13:11 +0000", 0) - iso = ("2005-04-07T22:13:11 -0200", 7200) - iso2 = ("2005-04-07 22:13:11 +0400", -14400) - iso3 = ("2005.04.07 22:13:11 -0000", 0) - alt = ("04/07/2005 22:13:11", 0) - alt2 = ("07.04.2005 22:13:11", 0) - veri_time = 1112904791 # the time this represents - for date, offset in (rfc, iso, iso2, iso3, alt, alt2): - assert_rval(parse_date(date), veri_time, offset) - # END for each date type - - # and failure - self.failUnlessRaises(ValueError, parse_date, 'invalid format') - self.failUnlessRaises(ValueError, parse_date, '123456789 -02000') - self.failUnlessRaises(ValueError, parse_date, ' 123456789 -0200') - - def test_actor(self): - for cr in (None, self.rorepo.config_reader()): - assert isinstance(Actor.committer(cr), Actor) - assert isinstance(Actor.author(cr), Actor) - #END assure config reader is handled - - def test_basics(self): - assert to_hex_sha(NULL_HEX_SHA) == NULL_HEX_SHA - assert len(to_bin_sha(NULL_HEX_SHA)) == 20 - assert to_hex_sha(to_bin_sha(NULL_HEX_SHA)) == NULL_HEX_SHA - - def _cmp_contents(self, file_path, data): - # raise if data from file at file_path - # does not match data string - fp = open(file_path, "rb") - try: - assert fp.read() == data - finally: - fp.close() - - def test_lockedfd(self): - my_file = tempfile.mktemp() - orig_data = "hello" - new_data = "world" - my_file_fp = open(my_file, "wb") - my_file_fp.write(orig_data) - my_file_fp.close() - - try: - lfd = LockedFD(my_file) - lockfilepath = lfd._lockfilepath() - - # cannot end before it was started - self.failUnlessRaises(AssertionError, lfd.rollback) - self.failUnlessRaises(AssertionError, lfd.commit) - - # open for writing - assert not os.path.isfile(lockfilepath) - wfd = lfd.open(write=True) - assert lfd._fd is wfd - assert os.path.isfile(lockfilepath) - - # write data and fail - os.write(wfd, new_data) - lfd.rollback() - assert lfd._fd is None - self._cmp_contents(my_file, orig_data) - assert not os.path.isfile(lockfilepath) - - # additional call doesnt fail - lfd.commit() - lfd.rollback() - - # test reading - lfd = LockedFD(my_file) - rfd = lfd.open(write=False) - assert os.read(rfd, len(orig_data)) == orig_data - - assert os.path.isfile(lockfilepath) - # deletion rolls back - del(lfd) - assert not os.path.isfile(lockfilepath) - - - # write data - concurrently - lfd = LockedFD(my_file) - olfd = LockedFD(my_file) - assert not os.path.isfile(lockfilepath) - wfdstream = lfd.open(write=True, stream=True) # this time as stream - assert os.path.isfile(lockfilepath) - # another one fails - self.failUnlessRaises(IOError, olfd.open) - - wfdstream.write(new_data) - lfd.commit() - assert not os.path.isfile(lockfilepath) - self._cmp_contents(my_file, new_data) - - # could test automatic _end_writing on destruction - finally: - os.remove(my_file) - # END final cleanup - - # try non-existing file for reading - lfd = LockedFD(tempfile.mktemp()) - try: - lfd.open(write=False) - except OSError: - assert not os.path.exists(lfd._lockfilepath()) - else: - self.fail("expected OSError") - # END handle exceptions + def test_it_should_dashify(self): + assert 'this-is-my-argument' == dashify('this_is_my_argument') + assert 'foo' == dashify('foo') + + + def test_lock_file(self): + my_file = tempfile.mktemp() + lock_file = LockFile(my_file) + assert not lock_file._has_lock() + # release lock we don't have - fine + lock_file._release_lock() + + # get lock + lock_file._obtain_lock_or_raise() + assert lock_file._has_lock() + + # concurrent access + other_lock_file = LockFile(my_file) + assert not other_lock_file._has_lock() + self.failUnlessRaises(IOError, other_lock_file._obtain_lock_or_raise) + + lock_file._release_lock() + assert not lock_file._has_lock() + + other_lock_file._obtain_lock_or_raise() + self.failUnlessRaises(IOError, lock_file._obtain_lock_or_raise) + + # auto-release on destruction + del(other_lock_file) + lock_file._obtain_lock_or_raise() + lock_file._release_lock() + + def test_blocking_lock_file(self): + my_file = tempfile.mktemp() + lock_file = BlockingLockFile(my_file) + lock_file._obtain_lock() + + # next one waits for the lock + start = time.time() + wait_time = 0.1 + wait_lock = BlockingLockFile(my_file, 0.05, wait_time) + self.failUnlessRaises(IOError, wait_lock._obtain_lock) + elapsed = time.time() - start + assert elapsed <= wait_time + 0.02 # some extra time it may cost + + def test_user_id(self): + assert '@' in get_user_id() + + def test_parse_date(self): + # test all supported formats + def assert_rval(rval, veri_time, offset=0): + assert len(rval) == 2 + assert isinstance(rval[0], int) and isinstance(rval[1], int) + assert rval[0] == veri_time + assert rval[1] == offset + + # now that we are here, test our conversion functions as well + utctz = altz_to_utctz_str(offset) + assert isinstance(utctz, basestring) + assert utctz_to_altz(verify_utctz(utctz)) == offset + # END assert rval utility + + rfc = ("Thu, 07 Apr 2005 22:13:11 +0000", 0) + iso = ("2005-04-07T22:13:11 -0200", 7200) + iso2 = ("2005-04-07 22:13:11 +0400", -14400) + iso3 = ("2005.04.07 22:13:11 -0000", 0) + alt = ("04/07/2005 22:13:11", 0) + alt2 = ("07.04.2005 22:13:11", 0) + veri_time = 1112904791 # the time this represents + for date, offset in (rfc, iso, iso2, iso3, alt, alt2): + assert_rval(parse_date(date), veri_time, offset) + # END for each date type + + # and failure + self.failUnlessRaises(ValueError, parse_date, 'invalid format') + self.failUnlessRaises(ValueError, parse_date, '123456789 -02000') + self.failUnlessRaises(ValueError, parse_date, ' 123456789 -0200') + + def test_actor(self): + for cr in (None, self.rorepo.config_reader()): + assert isinstance(Actor.committer(cr), Actor) + assert isinstance(Actor.author(cr), Actor) + #END assure config reader is handled + + def test_basics(self): + assert to_hex_sha(NULL_HEX_SHA) == NULL_HEX_SHA + assert len(to_bin_sha(NULL_HEX_SHA)) == 20 + assert to_hex_sha(to_bin_sha(NULL_HEX_SHA)) == NULL_HEX_SHA + + def _cmp_contents(self, file_path, data): + # raise if data from file at file_path + # does not match data string + fp = open(file_path, "rb") + try: + assert fp.read() == data + finally: + fp.close() + + def test_lockedfd(self): + my_file = tempfile.mktemp() + orig_data = "hello" + new_data = "world" + my_file_fp = open(my_file, "wb") + my_file_fp.write(orig_data) + my_file_fp.close() + + try: + lfd = LockedFD(my_file) + lockfilepath = lfd._lockfilepath() + + # cannot end before it was started + self.failUnlessRaises(AssertionError, lfd.rollback) + self.failUnlessRaises(AssertionError, lfd.commit) + + # open for writing + assert not os.path.isfile(lockfilepath) + wfd = lfd.open(write=True) + assert lfd._fd is wfd + assert os.path.isfile(lockfilepath) + + # write data and fail + os.write(wfd, new_data) + lfd.rollback() + assert lfd._fd is None + self._cmp_contents(my_file, orig_data) + assert not os.path.isfile(lockfilepath) + + # additional call doesnt fail + lfd.commit() + lfd.rollback() + + # test reading + lfd = LockedFD(my_file) + rfd = lfd.open(write=False) + assert os.read(rfd, len(orig_data)) == orig_data + + assert os.path.isfile(lockfilepath) + # deletion rolls back + del(lfd) + assert not os.path.isfile(lockfilepath) + + + # write data - concurrently + lfd = LockedFD(my_file) + olfd = LockedFD(my_file) + assert not os.path.isfile(lockfilepath) + wfdstream = lfd.open(write=True, stream=True) # this time as stream + assert os.path.isfile(lockfilepath) + # another one fails + self.failUnlessRaises(IOError, olfd.open) + + wfdstream.write(new_data) + lfd.commit() + assert not os.path.isfile(lockfilepath) + self._cmp_contents(my_file, new_data) + + # could test automatic _end_writing on destruction + finally: + os.remove(my_file) + # END final cleanup + + # try non-existing file for reading + lfd = LockedFD(tempfile.mktemp()) + try: + lfd.open(write=False) + except OSError: + assert not os.path.exists(lfd._lockfilepath()) + else: + self.fail("expected OSError") + # END handle exceptions - def test_iterable_list(self): - for args in (('name',), ('name', 'prefix_')): - l = IterableList('name') - - m1 = TestIterableMember('one') - m2 = TestIterableMember('two') - - l.extend((m1, m2)) - - assert len(l) == 2 - - # contains works with name and identity - assert m1.name in l - assert m2.name in l - assert m2 in l - assert m2 in l - assert 'invalid' not in l - - # with string index - assert l[m1.name] is m1 - assert l[m2.name] is m2 - - # with int index - assert l[0] is m1 - assert l[1] is m2 - - # with getattr - assert l.one is m1 - assert l.two is m2 - - # test exceptions - self.failUnlessRaises(AttributeError, getattr, l, 'something') - self.failUnlessRaises(IndexError, l.__getitem__, 'something') - - # delete by name and index - self.failUnlessRaises(IndexError, l.__delitem__, 'something') - del(l[m2.name]) - assert len(l) == 1 - assert m2.name not in l and m1.name in l - del(l[0]) - assert m1.name not in l - assert len(l) == 0 - - self.failUnlessRaises(IndexError, l.__delitem__, 0) - self.failUnlessRaises(IndexError, l.__delitem__, 'something') - #END for each possible mode - + def test_iterable_list(self): + for args in (('name',), ('name', 'prefix_')): + l = IterableList('name') + + m1 = TestIterableMember('one') + m2 = TestIterableMember('two') + + l.extend((m1, m2)) + + assert len(l) == 2 + + # contains works with name and identity + assert m1.name in l + assert m2.name in l + assert m2 in l + assert m2 in l + assert 'invalid' not in l + + # with string index + assert l[m1.name] is m1 + assert l[m2.name] is m2 + + # with int index + assert l[0] is m1 + assert l[1] is m2 + + # with getattr + assert l.one is m1 + assert l.two is m2 + + # test exceptions + self.failUnlessRaises(AttributeError, getattr, l, 'something') + self.failUnlessRaises(IndexError, l.__getitem__, 'something') + + # delete by name and index + self.failUnlessRaises(IndexError, l.__delitem__, 'something') + del(l[m2.name]) + assert len(l) == 1 + assert m2.name not in l and m1.name in l + del(l[0]) + assert m1.name not in l + assert len(l) == 0 + + self.failUnlessRaises(IndexError, l.__delitem__, 0) + self.failUnlessRaises(IndexError, l.__delitem__, 'something') + #END for each possible mode + class TestActor(TestBase): def test_from_string_should_separate_name_and_email(self): |