Improved configurability of import_project command.

User can now pass regexp used to parse matches.

Issue #760
Signed-off-by: Michal Čihař <michal@cihar.com>

docs/admin/management.rst

@@ -106,6 +106,12 @@ You can also specify file format to use (see :ref:`formats`) by the
 In case you need to specify version control system to use, you can do this using
 ``--vcs`` parameter. The default version control is Git.
 
+You can override parsing of component name from matched files by
+``--component-regexp``. This is a regular expression which will be matched
+against file name (as matched by `<filemask>`) and has to contain named group
+`name`. This can be also used for excluding files in case they do not match
+this expression. For example: ``.*/(?P<name>[^-]*)\.po``
+
 To give you some examples, let's try importing two projects.
 
 As first we import The Debian Handbook translations, where each language has

docs/changes.rst

@@ -8,6 +8,7 @@ Released on ? 2015.
 
 * Improved support for PHP files.
 * Ability to add ACL to anonymous user.
+* Improved configurability of import_project command.
 
 weblate 2.3
 -----------

weblate/trans/management/commands/import_project.py

@@ -50,6 +50,13 @@ class Command(BaseCommand):
                 'match to a project name'
             )
         ),
+        make_option(
+            '--component-regexp',
+            default=None,
+            help=(
+                'Regular expression to match component out of filename'
+            )
+        ),
         make_option(
             '--base-file-template',
             default='',
@@ -73,6 +80,7 @@ class Command(BaseCommand):
     def __init__(self, *args, **kwargs):
         super(Command, self).__init__(*args, **kwargs)
         self.filemask = None
+        self.component_re = None
         self.file_format = None
         self.name_template = None
         self.base_file_template = None
@@ -93,6 +101,9 @@ class Command(BaseCommand):
         Returns file name from patch based on filemask.
         """
         matches = self.match_regexp.match(path)
+        if matches is None:
+            self.logger.warning('Skipping {0}'.format(path))
+            return None
         return matches.group('name')
 
     @property
@@ -100,6 +111,8 @@ class Command(BaseCommand):
         '''
         Returns regexp for file matching
         '''
+        if self.component_re is not None:
+            return self.component_re
         if self._mask_regexp is None:
             match = fnmatch.translate(self.filemask)
             match = match.replace('.*.*', '(?P<name>.*)')
@@ -140,7 +153,9 @@ class Command(BaseCommand):
         # Parse subproject names out of them
         names = set()
         for match in matches:
-            names.add(self.get_name(match))
+            name = self.get_name(match)
+            if name:
+                names.add(name)
         self.logger.info('Found %d subprojects', len(names))
         return names
 
@@ -166,6 +181,22 @@ class Command(BaseCommand):
         self.file_format = options['file_format']
         self.name_template = options['name_template']
         self.base_file_template = options['base_file_template']
+        if options['component_regexp']:
+            try:
+                self.component_re = re.compile(
+                    options['component_regexp'],
+                    re.MULTILINE | re.DOTALL
+                )
+            except re.error as error:
+                raise CommandError(
+                    'Failed to compile regullar expression "{0}": {1}'.format(
+                        options['component_regexp'], error
+                    )
+                )
+            if 'name' not in self.component_re.groupindex:
+                raise CommandError(
+                    'Component regullar expression lacks named group "name"'
+                )
 
         # Try to get project
         try:

weblate/trans/tests/test_commands.py

@@ -48,6 +48,42 @@ class ImportProjectTest(RepoTestCase):
         # We should have loaded four subprojects
         self.assertEqual(project.subproject_set.count(), 4)
 
+    def test_import_re(self):
+        project = self.create_project()
+        call_command(
+            'import_project',
+            'test',
+            self.git_repo_path,
+            'master',
+            '**/*.po',
+            component_regexp=r'(?P<name>[^/-]*)/.*\.po'
+        )
+        self.assertEqual(project.subproject_set.count(), 1)
+
+    def test_import_re_missing(self):
+        self.assertRaises(
+            CommandError,
+            call_command,
+            'import_project',
+            'test',
+            self.git_repo_path,
+            'master',
+            '**/*.po',
+            component_regexp=r'(?P<xname>[^/-]*)/.*\.po'
+        )
+
+    def test_import_re_wrong(self):
+        self.assertRaises(
+            CommandError,
+            call_command,
+            'import_project',
+            'test',
+            self.git_repo_path,
+            'master',
+            '**/*.po',
+            component_regexp=r'(?P<xname>[^/-]*'
+        )
+
     def test_import_po(self):
         project = self.create_project()
         call_command(