changelog: ignore entries with wrong day padding (#508)

changelog: ignore entries with wrong day padding

This first version of this PR was completely implemented by Claude Code and Ambient Code.
After Nikola's amazing review, I fixed the problems myself, especially tinkering with padding fixups and updating the regex that parses day of month to properly account for double-digit days (and don't stop at the first digit - thanks Claude for that one).
Fixes #216
RELEASE NOTES BEGIN
Changelog entries that have incorrect padding set for a day of a month are now fixed if possible or ignored.
RELEASE NOTES END

Reviewed-by: gemini-code-assist[bot]
Reviewed-by: Nikola Forró
Reviewed-by: Tomas Tomecek <tomas@tomecek.net>

Author: centosinfra-prod-github-app[bot]

specfile/changelog.py

@@ -167,25 +167,33 @@ def extended_timestamp(self) -> bool:
         return m is not None
 
     @property
-    def day_of_month_padding(self) -> str:
-        """Padding of day of month in the entry header timestamp"""
+    def sanitized_day_of_month_padding(self) -> str:
+        """
+        Padding of day of month in the entry header timestamp. Nonsensical padding
+        (i.e. multiple spaces before double-digit day of month) is corrected.
+        """
         weekdays = "|".join(WEEKDAYS)
         months = "|".join(MONTHS)
+
         m = re.search(
             rf"""
             ({weekdays})                 # weekday
             [ ]
             ({months})                   # month
             [ ]
             (?P<wsp>[ ]*)                # optional whitespace padding
-            ((?P<zp>0)?\d|[12]\d|3[01])  # possibly zero-padded day of month
+            (?P<zp>0)?                   # optional zero-padding
+            (?P<day>[12]\d|3[01]|[1-9])  # day of month (1-31)
             """,
             self.header,
             re.VERBOSE,
         )
         if not m:
             return ""
-        return m.group("wsp") + (m.group("zp") or "")
+        # remove extra padding for double-digit day numbers
+        whitespace_padding = "" if len(m.group("day")) == 2 else m.group("wsp")
+        zero_padding = m.group("zp") or ""
+        return whitespace_padding + zero_padding
 
     @property
     def style(self) -> ChangelogStyle:

specfile/specfile.py

@@ -641,9 +641,13 @@ def add_changelog_entry(
                     elif email is not None:
                         author += f" <{email}>"
                     if changelog:
-                        # try to preserve padding of day of month
+                        # compute day of month padding from the longest valid entry
+                        # incorrect padding is stripped; invalid entries are ignored
                         padding = max(
-                            (e.day_of_month_padding for e in reversed(changelog)),
+                            (
+                                e.sanitized_day_of_month_padding
+                                for e in reversed(changelog)
+                            ),
                             key=len,
                         )
                     else:

tests/integration/test_specfile.py

@@ -226,6 +226,33 @@ def test_add_changelog_entry(
         assert sections.changelog[: len(result)] == result
 
 
+def test_add_changelog_entry_ignores_invalid_padding(specfile_factory, spec_minimal):
+    """Test that add_changelog_entry ignores entries with invalid day padding (issue #216)."""
+    spec = specfile_factory(spec_minimal)
+
+    # Simulate scipy scenario: malformed entry with space before double-digit day
+    with spec.sections() as sections:
+        sections.changelog.data = [
+            "* Mon Dec  11 2021 Author <email> - 0.5.1-5",  # INVALID
+            "- Initial version",
+            "",
+            "* Tue May 04 2020 Another <email> - 0.1-0",  # VALID (zero-padded)
+            "- First version",
+        ]
+
+    # Add new entry with single-digit day - should use zero-padding from valid entry
+    flexmock(specfile.specfile).should_receive("guess_packager").and_return(
+        "Test User <test@example.com>"
+    ).once()
+    spec.add_changelog_entry("test entry", timestamp=datetime.date(2024, 1, 5))
+
+    with spec.sections() as sections:
+        new_entry = sections.changelog[0]
+        # Should use zero-padding style "Jan 05", not space-padding "Jan  5"
+        assert "Jan 05" in new_entry
+        assert "Jan  5" not in new_entry
+
+
 @pytest.mark.parametrize(
     "version, release",
     [

tests/unit/test_changelog.py

@@ -92,12 +92,17 @@ def test_entry_has_extended_timestamp(header, extended):
         ),
         (
             "* Mon Oct  18 12:34:45 CEST 2021 Nikola Forró <nforro@redhat.com> - 0.2-1",
-            " ",
+            "",  # Invalid: double-digit day with space padding, returns ""
         ),
+        (
+            "* Mon Dec  11 2006 Author <email> - 1.0",
+            "",
+        ),  # Invalid: space before double-digit
+        ("* Invalid header", ""),  # Invalid: unparsable
     ],
 )
 def test_entry_day_of_month_padding(header, padding):
-    assert ChangelogEntry(header, [""]).day_of_month_padding == padding
+    assert ChangelogEntry(header, [""]).sanitized_day_of_month_padding == padding
 
 
 @pytest.mark.parametrize(