Add option extract_from_dir and frontend to follow symbolic links when traversing a directory

Author: jmoraleda

babel/messages/extract.py

@@ -126,6 +126,7 @@ def extract_from_dir(
     callback: Callable[[str, str, dict[str, Any]], object] | None = None,
     strip_comment_tags: bool = False,
     directory_filter: Callable[[str], bool] | None = None,
+    follow_links: bool = False,
 ) -> Generator[_FileExtractionResult, None, None]:
     """Extract messages from any source files found in the given directory.
 
@@ -194,6 +195,8 @@ def extract_from_dir(
     :param directory_filter: a callback to determine whether a directory should
                              be recursed into. Receives the full directory path;
                              should return True if the directory is valid.
+    :param follow_links: Whether symbolic links should be followed in OS's that
+                     support them. By default they are not followed.
     :see: `pathmatch`
     """
     if dirname is None:
@@ -204,7 +207,7 @@ def extract_from_dir(
         directory_filter = default_directory_filter
 
     absname = os.path.abspath(dirname)
-    for root, dirnames, filenames in os.walk(absname):
+    for root, dirnames, filenames in os.walk(absname, followlinks=follow_links):
         dirnames[:] = [
             subdir for subdir in dirnames
             if directory_filter(os.path.join(root, subdir))

babel/messages/frontend.py

@@ -347,10 +347,12 @@ class ExtractMessages(CommandMixin):
          'header comment for the catalog'),
         ('last-translator=', None,
          'set the name and email of the last translator in output'),
+        ('follow-links', 'l',
+         'follow symbolic links when traversing directories'),
     ]
     boolean_options = [
         'no-default-keywords', 'no-location', 'omit-header', 'no-wrap',
-        'sort-output', 'sort-by-file', 'strip-comments',
+        'sort-output', 'sort-by-file', 'strip-comments', 'follow-links'
     ]
     as_args = 'input-paths'
     multiple_value_options = (
@@ -390,6 +392,7 @@ def initialize_options(self):
         self.version = None
         self.add_comments = None
         self.strip_comments = False
+        self.follow_links = False
         self.include_lineno = True
         self.ignore_dirs = None
         self.header_comment = None
@@ -518,6 +521,7 @@ def run(self):
                         callback=callback,
                         strip_comment_tags=self.strip_comments,
                         directory_filter=self.directory_filter,
+                        follow_links=self.follow_links,
                     )
                 for filename, lineno, message, comments, context in extracted:
                     if os.path.isfile(path):

docs/cmdline.rst

@@ -129,6 +129,8 @@ a collection of source files::
                             Patterns for directories to ignore when scanning for
                             messages. Separate multiple patterns with spaces
                             (default ".* ._")
+      -f, --follow-links
+                            follow symbolic links when traversing directories   
       --header-comment=HEADER_COMMENT
                             header comment for the catalog
 

tests/messages/test_frontend.py

@@ -1608,6 +1608,14 @@ def test_extract_cli_knows_dash_s():
     assert cmdinst.strip_comments
 
 
+
+def test_extract_cli_knows_follow_links():
+    # This tests the follow-links command line argument
+    cmdinst = configure_cli_command("extract --follow-links -o foo babel")
+    assert isinstance(cmdinst, ExtractMessages)
+    assert cmdinst.follow_links
+    
+
 def test_extract_cli_knows_dash_dash_last_dash_translator():
     cmdinst = configure_cli_command('extract --last-translator "FULL NAME EMAIL@ADDRESS" -o foo babel')
     assert isinstance(cmdinst, ExtractMessages)