test: pin hand-duplicated regex/table copies to their config source

The layering rule forbids the pipeline/_render from importing
nameparser.config directly, so several patterns are copied by hand
with "keep in sync by hand" comments -- nothing previously checked
that promise. Tests may legally import both sides (test_layering.py's
own convention), so add sync tests for every duplicated
pattern/table, including the documented "initial" trailing-"?"
deviation between config and the pipeline copies (review finding on
#288).

Author: derek73

tests/v2/test_regex_sync.py

@@ -0,0 +1,73 @@
+"""Pins the pipeline's hand-duplicated regex/table copies to their
+nameparser.config.regexes source of truth.
+
+The 2.0 layering rule forbids nameparser._pipeline/_render importing
+nameparser.config directly, so several patterns and tables are copied
+by hand into the modules that need them, each with a "keep in sync by
+hand" comment. Nothing previously enforced that promise: if
+config/regexes.py changed, the copies would silently diverge with no
+CI signal. Tests may legally import both sides (test_layering.py's own
+convention), so this module is where the promise gets checked.
+"""
+from nameparser.config import regexes as _config
+from nameparser._pipeline import _assign, _post_rules, _tokenize, _vocab
+from nameparser import _render
+
+
+def test_emoji_ranges_match_config() -> None:
+    assert _tokenize._EMOJI_RANGES == _config._EMOJI_RANGES
+
+
+def test_bidi_pattern_matches_config() -> None:
+    assert _tokenize._BIDI.pattern == _config.re_bidi.pattern
+    assert _tokenize._BIDI.flags == _config.re_bidi.flags
+
+
+def test_period_not_at_end_matches_config() -> None:
+    source = _config.REGEXES["period_not_at_end"]
+    assert _vocab._PERIOD_NOT_AT_END.pattern == source.pattern
+    assert _vocab._PERIOD_NOT_AT_END.flags == source.flags
+
+
+def test_period_abbreviation_matches_config() -> None:
+    source = _config.REGEXES["period_abbreviation"]
+    assert _assign._PERIOD_ABBREV.pattern == source.pattern
+    assert _assign._PERIOD_ABBREV.flags == source.flags
+
+
+def test_roman_numeral_matches_config() -> None:
+    source = _config.REGEXES["roman_numeral"]
+    assert _assign._ROMAN.pattern == source.pattern
+    assert _assign._ROMAN.flags == source.flags
+
+
+def test_patronymic_patterns_match_config() -> None:
+    pairs = (
+        (_post_rules._EAST_SLAVIC, "east_slavic_patronymic"),
+        (_post_rules._EAST_SLAVIC_CYR, "east_slavic_patronymic_cyrillic"),
+        (_post_rules._TURKIC, "turkic_patronymic_marker"),
+        (_post_rules._TURKIC_CYR, "turkic_patronymic_marker_cyrillic"),
+    )
+    for copy, key in pairs:
+        source = _config.REGEXES[key]
+        assert copy.pattern == source.pattern, key
+        assert copy.flags == source.flags, key
+
+
+def test_initial_copies_agree_with_each_other_and_config() -> None:
+    # _vocab._INITIAL and _render._INITIAL are both v1's "initial"
+    # pattern minus its trailing "?" (documented at _render.py's
+    # _INITIAL definition: the two call sites always fullmatch a
+    # non-empty token, so the empty-string alternative is dropped on
+    # purpose). Assert both the internal-copy agreement and the exact,
+    # documented relationship to the config source, so a future edit to
+    # either side that breaks the relationship fails loudly here
+    # instead of silently drifting.
+    assert _vocab._INITIAL.pattern == _render._INITIAL.pattern
+    source = _config.REGEXES["initial"]
+    # config's pattern is the pipeline copy with an extra "?" spliced in
+    # just before the trailing "$", making the whole group optional.
+    trimmed = _vocab._INITIAL.pattern
+    reconstructed = trimmed[:-1] + "?" + trimmed[-1:]
+    assert source.pattern == reconstructed
+    assert source.flags == _vocab._INITIAL.flags