ci: タイムスタンプだけの更新を取り込まない

Author: methane

.github/workflows/update.yml

@@ -49,6 +49,8 @@ jobs:
           sudo apt-get update -y && sudo apt-get install gettext -y
           pip install --upgrade pip
           pip install -r requirements.txt -r cpython/Doc/requirements.txt
+      - name: Test translation tooling
+        run: python -B -m unittest discover -s tests
       - name: Update translations
         run: |
           sh scripts/update.sh
@@ -58,6 +60,8 @@ jobs:
       - name: Wrap catalog message files
         run: |
           powrap --modified
+      - name: Ignore timestamp-only changes
+        run: python scripts/translations.py restore-timestamp-only .
       - name: Commit and push changes
         if: github.repository == 'python/python-docs-ja'
         run: |
@@ -113,6 +117,8 @@ jobs:
       - name: Wrap catalog message files
         run: |
           powrap --modified -C "${{ matrix.branch }}"
+      - name: Ignore timestamp-only changes
+        run: python "${{ env.CPYTHON_BRANCH }}/scripts/translations.py" restore-timestamp-only "${{ matrix.branch }}"
 
       - name: Commit and push changes
         if: github.repository == 'python/python-docs-ja'

scripts/translations.py

@@ -4,10 +4,55 @@
 import configparser
 import os
 from pathlib import Path, PurePosixPath
+import re
 import shutil
 import subprocess
 
 
+def without_po_timestamps(content):
+    """Ignore date fields in the PO header, preserving every other byte."""
+    header = re.search(
+        rb'^msgid ""\r?\nmsgstr "[^\r\n]*"\r?\n(?:"[^\r\n]*"\r?\n)*',
+        content, re.MULTILINE,
+    )
+    if header is None:
+        return content
+    normalized = re.sub(
+        rb'^"(?:POT-Creation-Date|PO-Revision-Date):[^"\r\n]*\\n"\r?\n',
+        b'', header.group(), flags=re.MULTILINE,
+    )
+    return content[:header.start()] + normalized + content[header.end():]
+
+
+def restore_timestamp_only_changes(root):
+    """Restore PO files whose unstaged changes only affect header dates.
+
+    Run after wrapping, so formatting differences in downloaded or merged files
+    do not prevent timestamp-only changes from being recognized.
+    """
+    result = subprocess.run(
+        ["git", "-C", str(root), "diff", "--name-only", "-z", "--", "*.po"],
+        check=True, stdout=subprocess.PIPE,
+    )
+    for name in result.stdout.split(b"\0"):
+        if not name:
+            continue
+        path = os.fsdecode(name)
+        target = root / path
+        if not target.is_file():
+            continue
+        original = subprocess.run(
+            ["git", "-C", str(root), "show", f":{path}"],
+            check=True, stdout=subprocess.PIPE,
+        ).stdout
+        if (
+            without_po_timestamps(target.read_bytes())
+            == without_po_timestamps(original)
+        ):
+            target.write_bytes(original)
+            print(f"Ignored timestamp-only change: {path}")
+
+
 def active_catalogs(config_path):
     config = configparser.ConfigParser(interpolation=None)
     with config_path.open(encoding="utf-8") as stream:
@@ -89,11 +134,15 @@ def main():
     merge_parser = commands.add_parser("merge")
     merge_parser.add_argument("source", type=Path)
     merge_parser.add_argument("destination", type=Path)
+    restore_parser = commands.add_parser("restore-timestamp-only")
+    restore_parser.add_argument("root", type=Path)
     args = parser.parse_args()
     if args.command == "sync":
         sync(args.source, args.destination, args.language)
-    else:
+    elif args.command == "merge":
         merge(args.source, args.destination)
+    else:
+        restore_timestamp_only_changes(args.root)
 
 
 if __name__ == "__main__":

tests/test_translations.py

@@ -31,6 +31,21 @@ def write_po(root, name, translation):
     return path
 
 
+DATED_PO = (
+    '# Catalog\n'
+    'msgid ""\n'
+    'msgstr ""\n'
+    '"Content-Type: text/plain; charset=UTF-8\\n"\n'
+    '"POT-Creation-Date: 2026-09-29 18:58+0000\\n"\n'
+    '"PO-Revision-Date: 2025-09-16 00:01+0000\\n"\n'
+    '"Last-Translator: Translator\\n"\n'
+    '\n'
+    '#: example.rst:10\n'
+    'msgid "example"\n'
+    'msgstr "翻訳"\n'
+).encode('utf-8')
+
+
 class TranslationTests(unittest.TestCase):
     def setUp(self):
         temporary = tempfile.TemporaryDirectory()
@@ -49,6 +64,89 @@ def track(self):
             ["git", "-C", str(self.destination), "add", "."], check=True
         )
 
+    def restore_catalog(self, before, incoming):
+        target = self.destination / "example.po"
+        target.write_bytes(before)
+        self.track()
+        target.write_bytes(incoming)
+        translations.restore_timestamp_only_changes(self.destination)
+        return target
+
+    def test_restore_timestamp_only_changes(self):
+        for newline in (b'\n', b'\r\n'):
+            before = DATED_PO.replace(b'\n', newline)
+            changes = [
+                before,
+                before.replace(b'2026-09-29', b'2026-10-19'),
+                before.replace(b'2025-09-16', b'2026-10-20'),
+                before.replace(b'2026-09-29', b'2026-10-19').replace(
+                    b'2025-09-16', b'2026-10-20'
+                ),
+            ]
+            for incoming in changes:
+                with self.subTest(newline=newline, incoming=incoming):
+                    target = self.restore_catalog(before, incoming)
+                    self.assertEqual(target.read_bytes(), before)
+
+    def test_restore_keeps_other_changes_and_their_new_timestamps(self):
+        updated = DATED_PO.replace(b'2026-09-29', b'2026-10-19')
+        replacements = [
+            ('msgstr "翻訳"', 'msgstr "新しい訳"'),
+            ('msgid "example"', 'msgid "changed"'),
+            ('#: example.rst:10', '#: example.rst:20'),
+            ('# Catalog', '# Updated comment'),
+            ('msgid "example"', '#, fuzzy\nmsgid "example"'),
+            ('Last-Translator: Translator', 'Last-Translator: New Translator'),
+            ('msgstr "翻訳"', 'msgstr ""\n"翻訳"'),
+        ]
+        for old, new in replacements:
+            with self.subTest(change=new):
+                incoming = updated.replace(old.encode(), new.encode())
+                target = self.restore_catalog(DATED_PO, incoming)
+                self.assertEqual(target.read_bytes(), incoming)
+
+    def test_restore_does_not_ignore_timestamp_text_in_translation(self):
+        before = DATED_PO.replace(
+            'msgstr "翻訳"\n'.encode(),
+            b'msgstr ""\n"POT-Creation-Date: 2026-09-29 18:58+0000\\n"\n',
+        )
+        incoming = before.replace(b'2026-09-29', b'2026-10-19')
+        target = self.restore_catalog(before, incoming)
+        self.assertEqual(target.read_bytes(), incoming)
+
+    def test_restore_preserves_new_files_and_deletions(self):
+        deleted = self.destination / "deleted.po"
+        deleted.write_bytes(DATED_PO)
+        self.track()
+        deleted.unlink()
+        new = self.destination / "new.po"
+        new.write_bytes(DATED_PO)
+        translations.restore_timestamp_only_changes(self.destination)
+        self.assertFalse(deleted.exists())
+        self.assertEqual(new.read_bytes(), DATED_PO)
+
+    def test_restore_command_handles_multiple_files_and_explicit_root(self):
+        catalogs = ["unchanged.po", "translated.po", "library/with space.po"]
+        for path in catalogs:
+            target = self.destination / path
+            target.parent.mkdir(parents=True, exist_ok=True)
+            target.write_bytes(DATED_PO)
+        self.track()
+        updated = DATED_PO.replace(b'2026-09-29', b'2026-10-19')
+        translated = updated.replace('翻訳'.encode(), '新しい訳'.encode())
+        (self.destination / "unchanged.po").write_bytes(updated)
+        (self.destination / "translated.po").write_bytes(translated)
+        (self.destination / "library/with space.po").write_bytes(updated)
+        subprocess.run(
+            [sys.executable, translations.__file__, "restore-timestamp-only",
+             str(self.destination)],
+            cwd=self.source, check=True, stdout=subprocess.PIPE,
+        )
+
+        self.assertEqual((self.destination / "unchanged.po").read_bytes(), DATED_PO)
+        self.assertEqual((self.destination / "translated.po").read_bytes(), translated)
+        self.assertEqual((self.destination / "library/with space.po").read_bytes(), DATED_PO)
+
     def test_sync_removes_old_path_and_preserves_untracked_files(self):
         catalogs = ["builtins/functions.po", "library/builtins.po", "bugs.po"]
         write_config(self.source, catalogs)