Add CatalogRepository class

2 files changed, 91 insertions, 2 deletions

diff --git a/sphinx/util/i18n.py b/sphinx/util/i18n.py
index 9196e6fb6..6aa285f5d 100644
--- a/sphinx/util/i18n.py
+++ b/sphinx/util/i18n.py
@@ -24,14 +24,14 @@ from sphinx.errors import SphinxError
 from sphinx.locale import __
 from sphinx.util import logging
 from sphinx.util.matching import Matcher
-from sphinx.util.osutil import SEP, relpath
+from sphinx.util.osutil import SEP, canon_path, relpath
 
 
 logger = logging.getLogger(__name__)
 
 if False:
     # For type annotation
-    from typing import Callable, List, Set  # NOQA
+    from typing import Callable, Generator, List, Set, Tuple  # NOQA
     from sphinx.environment import BuildEnvironment  # NOQA
 
 LocaleFileInfoBase = namedtuple('CatalogInfo', 'base_dir,domain,charset')
@@ -81,6 +81,51 @@ class CatalogInfo(LocaleFileInfoBase):
                 logger.warning(__('writing error: %s, %s'), self.mo_path, exc)
 
 
+class CatalogRepository:
+    """A repository for message catalogs."""
+
+    def __init__(self, basedir, locale_dirs, language, encoding):
+        # type: (str, List[str], str, str) -> None
+        self.basedir = basedir
+        self._locale_dirs = locale_dirs
+        self.language = language
+        self.encoding = encoding
+
+    @property
+    def locale_dirs(self):
+        # type: () -> Generator[str, None, None]
+        if not self.language:
+            return
+
+        for locale_dir in self._locale_dirs:
+            locale_dir = path.join(self.basedir, locale_dir)
+            if path.exists(path.join(locale_dir, self.language, 'LC_MESSAGES')):
+                yield locale_dir
+
+    @property
+    def pofiles(self):
+        # type: () -> Generator[Tuple[str, str], None, None]
+        for locale_dir in self.locale_dirs:
+            basedir = path.join(locale_dir, self.language, 'LC_MESSAGES')
+            for root, dirnames, filenames in os.walk(basedir):
+                # skip dot-directories
+                for dirname in dirnames:
+                    if dirname.startswith('.'):
+                        dirnames.remove(dirname)
+
+                for filename in filenames:
+                    if filename.endswith('.po'):
+                        fullpath = path.join(root, filename)
+                        yield basedir, relpath(fullpath, basedir)
+
+    @property
+    def catalogs(self):
+        # type: () -> Generator[CatalogInfo, None, None]
+        for basedir, filename in self.pofiles:
+            domain = canon_path(path.splitext(filename)[0])
+            yield CatalogInfo(basedir, domain, self.encoding)
+
+
 def find_catalog(docname, compaction):
     # type: (str, bool) -> str
     if compaction:
diff --git a/tests/test_util_i18n.py b/tests/test_util_i18n.py
index b25e29575..5208689e8 100644
--- a/tests/test_util_i18n.py
+++ b/tests/test_util_i18n.py
@@ -255,3 +255,47 @@ def test_get_filename_for_language(app):
     app.env.config.figure_language_filename = '{root}.{invalid}{ext}'
     with pytest.raises(SphinxError):
         i18n.get_image_filename_for_language('foo.png', app.env)
+
+
+def test_CatalogRepository(tempdir):
+    (tempdir / 'loc1' / 'xx' / 'LC_MESSAGES').makedirs()
+    (tempdir / 'loc1' / 'xx' / 'LC_MESSAGES' / 'test1.po').write_text('#')
+    (tempdir / 'loc1' / 'xx' / 'LC_MESSAGES' / 'test2.po').write_text('#')
+    (tempdir / 'loc1' / 'xx' / 'LC_MESSAGES' / 'sub').makedirs()
+    (tempdir / 'loc1' / 'xx' / 'LC_MESSAGES' / 'sub' / 'test3.po').write_text('#')
+    (tempdir / 'loc1' / 'xx' / 'LC_MESSAGES' / 'sub' / 'test4.po').write_text('#')
+    (tempdir / 'loc1' / 'xx' / 'LC_MESSAGES' / '.dotdir').makedirs()
+    (tempdir / 'loc1' / 'xx' / 'LC_MESSAGES' / '.dotdir' / 'test5.po').write_text('#')
+    (tempdir / 'loc1' / 'yy' / 'LC_MESSAGES').makedirs()
+    (tempdir / 'loc1' / 'yy' / 'LC_MESSAGES' / 'test6.po').write_text('#')
+    (tempdir / 'loc2' / 'xx' / 'LC_MESSAGES').makedirs()
+    (tempdir / 'loc2' / 'xx' / 'LC_MESSAGES' / 'test1.po').write_text('#')
+    (tempdir / 'loc2' / 'xx' / 'LC_MESSAGES' / 'test7.po').write_text('#')
+
+    # for language xx
+    repo = i18n.CatalogRepository(tempdir, ['loc1', 'loc2'], 'xx', 'utf-8')
+    assert list(repo.locale_dirs) == [str(tempdir / 'loc1'),
+                                      str(tempdir / 'loc2')]
+    assert all(isinstance(c, i18n.CatalogInfo) for c in repo.catalogs)
+    assert sorted(c.domain for c in repo.catalogs) == ['sub/test3', 'sub/test4',
+                                                       'test1', 'test1', 'test2', 'test7']
+
+    # for language yy
+    repo = i18n.CatalogRepository(tempdir, ['loc1', 'loc2'], 'yy', 'utf-8')
+    assert sorted(c.domain for c in repo.catalogs) == ['test6']
+
+    # unknown languages
+    repo = i18n.CatalogRepository(tempdir, ['loc1', 'loc2'], 'zz', 'utf-8')
+    assert sorted(c.domain for c in repo.catalogs) == []
+
+    # no languages
+    repo = i18n.CatalogRepository(tempdir, ['loc1', 'loc2'], None, 'utf-8')
+    assert sorted(c.domain for c in repo.catalogs) == []
+
+    # unknown locale_dirs
+    repo = i18n.CatalogRepository(tempdir, ['loc3'], None, 'utf-8')
+    assert sorted(c.domain for c in repo.catalogs) == []
+
+    # no locale_dirs
+    repo = i18n.CatalogRepository(tempdir, [], None, 'utf-8')
+    assert sorted(c.domain for c in repo.catalogs) == []