Preserve nonconsecutive repeated comment lines

babel/messages/catalog.py

@@ -18,6 +18,7 @@
 from difflib import SequenceMatcher
 from email import message_from_string
 from heapq import nlargest
+from itertools import groupby
 from string import Formatter
 from typing import TYPE_CHECKING, TypedDict
 
@@ -181,8 +182,9 @@ def __init__(
             self.flags.add('python-brace-format')
         else:
             self.flags.discard('python-brace-format')
-        self.auto_comments = list(dict.fromkeys(auto_comments)) if auto_comments else []
-        self.user_comments = list(dict.fromkeys(user_comments)) if user_comments else []
+        # Collapse consecutive repeats without losing repeated wrapped fragments.
+        self.auto_comments = [comment for comment, _ in groupby(auto_comments)] if auto_comments else []
+        self.user_comments = [comment for comment, _ in groupby(user_comments)] if user_comments else []
         if previous_id:
             if isinstance(previous_id, str):
                 self.previous_id = [previous_id]

tests/messages/test_catalog.py

@@ -15,6 +15,8 @@
 import pickle
 from io import StringIO
 
+import pytest
+
 from babel.dates import UTC, format_datetime
 from babel.messages import catalog, pofile
 from babel.util import FixedOffsetTimezone
@@ -66,6 +68,23 @@ def test_message_translator_comments():
     assert mess.auto_comments == ['Comment 1 About `foo`', 'Comment 2 About `foo`']
 
 
+@pytest.mark.parametrize('attribute', ['auto_comments', 'user_comments'])
+@pytest.mark.parametrize(
+    ('comments', 'expected'),
+    [
+        ([], []),
+        (['KEY:', 'KEY:', 'KEY:'], ['KEY:']),
+        (['KEY:', 'first', 'KEY:', 'second'], ['KEY:', 'first', 'KEY:', 'second']),
+        (['KEY:', 'KEY:', 'first', 'KEY:', 'KEY:'], ['KEY:', 'first', 'KEY:']),
+        (['', '', 'text', ''], ['', 'text', '']),
+    ],
+)
+def test_message_collapses_only_consecutive_comment_lines(attribute, comments, expected):
+    message = catalog.Message('foo', **{attribute: iter(comments)})
+    assert getattr(message, attribute) == expected
+    assert getattr(message.clone(), attribute) == expected
+
+
 def test_message_clone_message_object():
     msg = catalog.Message('foo', locations=[('foo.py', 42)])
     clone = msg.clone()

tests/messages/test_pofile.py

@@ -20,6 +20,37 @@
 from babel.messages.pofile import _enclose_filename_if_necessary, _extract_locations
 
 
+@pytest.mark.parametrize(
+    ('prefix', 'attribute'),
+    [('#.', 'auto_comments'), ('#', 'user_comments')],
+)
+@pytest.mark.parametrize('obsolete', [False, True])
+def test_read_po_preserves_repeated_comment_lines(prefix, attribute, obsolete):
+    comments = ['KEY:', 'first.long.key', 'KEY:', 'second.long.key']
+    source = '\n'.join(f'{prefix} {line}' for line in comments) + '\n'
+    message_prefix = '#~ ' if obsolete else ''
+    source += f'{message_prefix}msgid "Example"\n{message_prefix}msgstr ""\n'
+
+    catalog = pofile.read_po(StringIO(source))
+    message = catalog.obsolete['Example'] if obsolete else catalog['Example']
+    assert getattr(message, attribute) == comments
+
+
+@pytest.mark.parametrize('attribute', ['auto_comments', 'user_comments'])
+def test_wrapped_comments_roundtrip(attribute):
+    catalog = Catalog()
+    comments = ['KEY: ' + 'foo.' * 20, 'KEY: ' + 'bar.' * 20]
+    catalog.add('Example', **{attribute: comments})
+    first = BytesIO()
+    pofile.write_po(first, catalog, omit_header=True)
+    first.seek(0)
+
+    restored = pofile.read_po(first)
+    second = BytesIO()
+    pofile.write_po(second, restored, omit_header=True)
+    assert second.getvalue() == first.getvalue()
+
+
 def test_enclosed_filenames_in_location_comment():
     catalog = Catalog()
     catalog.add("foo", lineno=2, locations=[("main 1.py", 1)], string="")