Add incremental blame support

This adds a sibling method to Repo's blame method:

    Repo.blame_incremental(rev, path, **kwargs)

This can alternatively be called using:

    Repo.blame(rev, path, incremental=True)

The main difference is that blame incremental is a bit more efficient
and does not return the full file's contents, just the commits and the
line number ranges.  The parser is a bit more straight-forward and
faster since the incremental output format is defined a little stricter.

1 files changed, 24 insertions, 0 deletions

diff --git a/git/test/test_repo.py b/git/test/test_repo.py
index 177aa176..ab6c502f 100644
--- a/git/test/test_repo.py
+++ b/git/test/test_repo.py
@@ -50,6 +50,16 @@ from io import BytesIO
 from nose import SkipTest
 
 
+def iter_flatten(lol):
+    for items in lol:
+        for item in items:
+            yield item
+
+
+def flatten(lol):
+    return list(iter_flatten(lol))
+
+
 class TestRepo(TestBase):
 
     @raises(InvalidGitRepositoryError)
@@ -324,6 +334,20 @@ class TestRepo(TestBase):
         assert nml, "There should at least be one blame commit that contains multiple lines"
 
     @patch.object(Git, '_call_process')
+    def test_blame_incremental(self, git):
+        git.return_value = fixture('blame_incremental')
+        blame_output = self.rorepo.blame_incremental('9debf6b0aafb6f7781ea9d1383c86939a1aacde3', 'AUTHORS')
+        blame_output = list(blame_output)
+        assert len(blame_output) == 5
+
+        # Check all outputted line numbers
+        ranges = flatten([line_numbers for _, line_numbers in blame_output])
+        assert ranges == flatten([range(2, 3), range(14, 15), range(1, 2), range(3, 14), range(15, 17)]), str(ranges)
+
+        commits = [c.hexsha[:7] for c, _ in blame_output]
+        assert commits == ['82b8902', '82b8902', 'c76852d', 'c76852d', 'c76852d'], str(commits)
+
+    @patch.object(Git, '_call_process')
     def test_blame_complex_revision(self, git):
         git.return_value = fixture('blame_complex_revision')
         res = self.rorepo.blame("HEAD~10..HEAD", "README.md")