Fix multi-line copyright when SOURCE_DATE_EPOCH is set (#11524)

Co-authored-by: Bénédikt Tran <10796600+picnixz@users.noreply.github.com>

CHANGES

@@ -16,6 +16,9 @@ Features added
 Bugs fixed
 ----------
 
+* #11514: Fix ``SOURCE_DATE_EPOCH`` in multi-line copyright footer.
+  Patch by Bénédikt Tran.
+
 Testing
 -------
 

sphinx/config.py

@@ -2,7 +2,7 @@
 
 from __future__ import annotations
 
-import re
+import time
 import traceback
 import types
 from os import getenv, path
@@ -11,7 +11,6 @@
 from sphinx.errors import ConfigError, ExtensionError
 from sphinx.locale import _, __
 from sphinx.util import logging
-from sphinx.util.i18n import format_date
 from sphinx.util.osutil import fs_encoding
 from sphinx.util.tags import Tags
 from sphinx.util.typing import NoneType
@@ -22,14 +21,15 @@
     from sphinx.util.osutil import _chdir as chdir
 
 if TYPE_CHECKING:
+    from collections.abc import Sequence
+
     from sphinx.application import Sphinx
     from sphinx.environment import BuildEnvironment
 
 logger = logging.getLogger(__name__)
 
 CONFIG_FILENAME = 'conf.py'
 UNSERIALIZABLE_TYPES = (type, types.ModuleType, types.FunctionType)
-copyright_year_re = re.compile(r'^((\d{4}-)?)(\d{4})(?=[ ,])')
 
 
 class ConfigValue(NamedTuple):
@@ -417,17 +417,52 @@ def init_numfig_format(app: Sphinx, config: Config) -> None:
     config.numfig_format = numfig_format  # type: ignore
 
 
-def correct_copyright_year(app: Sphinx, config: Config) -> None:
+def correct_copyright_year(_app: Sphinx, config: Config) -> None:
     """Correct values of copyright year that are not coherent with
     the SOURCE_DATE_EPOCH environment variable (if set)
 
     See https://reproducible-builds.org/specs/source-date-epoch/
     """
-    if getenv('SOURCE_DATE_EPOCH') is not None:
-        for k in ('copyright', 'epub_copyright'):
-            if k in config:
-                replace = r'\g<1>%s' % format_date('%Y', language='en')
-                config[k] = copyright_year_re.sub(replace, config[k])
+    if (source_date_epoch := getenv('SOURCE_DATE_EPOCH')) is None:
+        return
+
+    source_date_epoch_year = str(time.gmtime(int(source_date_epoch)).tm_year)
+
+    for k in ('copyright', 'epub_copyright'):
+        if k in config:
+            value: str | Sequence[str] = config[k]
+            if isinstance(value, str):
+                config[k] = _substitute_copyright_year(value, source_date_epoch_year)
+            else:
+                items = (_substitute_copyright_year(x, source_date_epoch_year) for x in value)
+                config[k] = type(value)(items)  # type: ignore[call-arg]
+
+
+def _substitute_copyright_year(copyright_line: str, replace_year: str) -> str:
+    """Replace the year in a single copyright line.
+
+    Legal formats are:
+
+    * ``YYYY,``
+    * ``YYYY ``
+    * ``YYYY-YYYY,``
+    * ``YYYY-YYYY ``
+
+    The final year in the string is replaced with ``replace_year``.
+    """
+    if not copyright_line[:4].isdigit():
+        return copyright_line
+
+    if copyright_line[4] in ' ,':
+        return replace_year + copyright_line[4:]
+
+    if copyright_line[4] != '-':
+        return copyright_line
+
+    if copyright_line[5:9].isdigit() and copyright_line[9] in ' ,':
+        return copyright_line[:5] + replace_year + copyright_line[9:]
+
+    return copyright_line
 
 
 def check_confval_types(app: Sphinx | None, config: Config) -> None:

tests/test_config.py

@@ -1,5 +1,6 @@
 """Test the sphinx.config.Config class."""
 
+import time
 from unittest import mock
 
 import pytest
@@ -444,23 +445,67 @@ def test_conf_py_nitpick_ignore_list(tempdir):
     assert cfg.nitpick_ignore_regex == []
 
 
+@pytest.fixture(params=[
+    # test with SOURCE_DATE_EPOCH unset: no modification
+    None,
+    # test with SOURCE_DATE_EPOCH set: copyright year should be updated
+    1293840000,
+    1293839999,
+])
+def source_date_year(request, monkeypatch):
+    sde = request.param
+    with monkeypatch.context() as m:
+        if sde:
+            m.setenv('SOURCE_DATE_EPOCH', sde)
+            yield time.gmtime(sde).tm_year
+        else:
+            m.delenv('SOURCE_DATE_EPOCH', raising=False)
+            yield None
+
+
 @pytest.mark.sphinx(testroot='copyright-multiline')
-def test_multi_line_copyright(app, status, warning):
+def test_multi_line_copyright(source_date_year, app, monkeypatch):
     app.builder.build_all()
 
     content = (app.outdir / 'index.html').read_text(encoding='utf-8')
 
-    assert '      &#169; Copyright 2006-2009, Alice.<br/>' in content
-    assert '      &#169; Copyright 2010-2013, Bob.<br/>' in content
-    assert '      &#169; Copyright 2014-2017, Charlie.<br/>' in content
-    assert '      &#169; Copyright 2018-2021, David.<br/>' in content
-    assert '      &#169; Copyright 2022-2025, Eve.' in content
-
-    lines = (
-        '      &#169; Copyright 2006-2009, Alice.<br/>\n    \n'
-        '      &#169; Copyright 2010-2013, Bob.<br/>\n    \n'
-        '      &#169; Copyright 2014-2017, Charlie.<br/>\n    \n'
-        '      &#169; Copyright 2018-2021, David.<br/>\n    \n'
-        '      &#169; Copyright 2022-2025, Eve.\n    \n'
-    )
-    assert lines in content
+    if source_date_year is None:
+        # check the copyright footer line by line (empty lines ignored)
+        assert '  &#169; Copyright 2006-2009, Alice.<br/>\n' in content
+        assert '  &#169; Copyright 2010-2013, Bob.<br/>\n' in content
+        assert '  &#169; Copyright 2014-2017, Charlie.<br/>\n' in content
+        assert '  &#169; Copyright 2018-2021, David.<br/>\n' in content
+        assert '  &#169; Copyright 2022-2025, Eve.' in content
+
+        # check the raw copyright footer block (empty lines included)
+        assert (
+            '      &#169; Copyright 2006-2009, Alice.<br/>\n'
+            '    \n'
+            '      &#169; Copyright 2010-2013, Bob.<br/>\n'
+            '    \n'
+            '      &#169; Copyright 2014-2017, Charlie.<br/>\n'
+            '    \n'
+            '      &#169; Copyright 2018-2021, David.<br/>\n'
+            '    \n'
+            '      &#169; Copyright 2022-2025, Eve.'
+        ) in content
+    else:
+        # check the copyright footer line by line (empty lines ignored)
+        assert f'  &#169; Copyright 2006-{source_date_year}, Alice.<br/>\n' in content
+        assert f'  &#169; Copyright 2010-{source_date_year}, Bob.<br/>\n' in content
+        assert f'  &#169; Copyright 2014-{source_date_year}, Charlie.<br/>\n' in content
+        assert f'  &#169; Copyright 2018-{source_date_year}, David.<br/>\n' in content
+        assert f'  &#169; Copyright 2022-{source_date_year}, Eve.' in content
+
+        # check the raw copyright footer block (empty lines included)
+        assert (
+            f'      &#169; Copyright 2006-{source_date_year}, Alice.<br/>\n'
+            f'    \n'
+            f'      &#169; Copyright 2010-{source_date_year}, Bob.<br/>\n'
+            f'    \n'
+            f'      &#169; Copyright 2014-{source_date_year}, Charlie.<br/>\n'
+            f'    \n'
+            f'      &#169; Copyright 2018-{source_date_year}, David.<br/>\n'
+            f'    \n'
+            f'      &#169; Copyright 2022-{source_date_year}, Eve.'
+        ) in content