gh-91575: Add a script for generating data for case-insensitive matching in re (GH-91660)

serhiy-storchaka · web-flow · commit f912cc0e413f · 2022-04-22T21:37:46.000+03:00
Also test that all extra cases are in BMP.
diff --git a/Lib/re/_casefix.py b/Lib/re/_casefix.py
@@ -0,0 +1,106 @@
+# Auto-generated by Tools/scripts/generate_re_casefix.py.
+
+# Maps the code of lowercased character to codes of different lowercased
+# characters which have the same uppercase.
+_EXTRA_CASES = {
+    # LATIN SMALL LETTER I: LATIN SMALL LETTER DOTLESS I
+    0x0069: (0x0131,), # 'i': 'ı'
+    # LATIN SMALL LETTER S: LATIN SMALL LETTER LONG S
+    0x0073: (0x017f,), # 's': 'ſ'
+    # MICRO SIGN: GREEK SMALL LETTER MU
+    0x00b5: (0x03bc,), # 'µ': 'μ'
+    # LATIN SMALL LETTER DOTLESS I: LATIN SMALL LETTER I
+    0x0131: (0x0069,), # 'ı': 'i'
+    # LATIN SMALL LETTER LONG S: LATIN SMALL LETTER S
+    0x017f: (0x0073,), # 'ſ': 's'
+    # COMBINING GREEK YPOGEGRAMMENI: GREEK SMALL LETTER IOTA, GREEK PROSGEGRAMMENI
+    0x0345: (0x03b9, 0x1fbe), # '\u0345': 'ιι'
+    # GREEK SMALL LETTER IOTA WITH DIALYTIKA AND TONOS: GREEK SMALL LETTER IOTA WITH DIALYTIKA AND OXIA
+    0x0390: (0x1fd3,), # 'ΐ': 'ΐ'
+    # GREEK SMALL LETTER UPSILON WITH DIALYTIKA AND TONOS: GREEK SMALL LETTER UPSILON WITH DIALYTIKA AND OXIA
+    0x03b0: (0x1fe3,), # 'ΰ': 'ΰ'
+    # GREEK SMALL LETTER BETA: GREEK BETA SYMBOL
+    0x03b2: (0x03d0,), # 'β': 'ϐ'
+    # GREEK SMALL LETTER EPSILON: GREEK LUNATE EPSILON SYMBOL
+    0x03b5: (0x03f5,), # 'ε': 'ϵ'
+    # GREEK SMALL LETTER THETA: GREEK THETA SYMBOL
+    0x03b8: (0x03d1,), # 'θ': 'ϑ'
+    # GREEK SMALL LETTER IOTA: COMBINING GREEK YPOGEGRAMMENI, GREEK PROSGEGRAMMENI
+    0x03b9: (0x0345, 0x1fbe), # 'ι': '\u0345ι'
+    # GREEK SMALL LETTER KAPPA: GREEK KAPPA SYMBOL
+    0x03ba: (0x03f0,), # 'κ': 'ϰ'
+    # GREEK SMALL LETTER MU: MICRO SIGN
+    0x03bc: (0x00b5,), # 'μ': 'µ'
+    # GREEK SMALL LETTER PI: GREEK PI SYMBOL
+    0x03c0: (0x03d6,), # 'π': 'ϖ'
+    # GREEK SMALL LETTER RHO: GREEK RHO SYMBOL
+    0x03c1: (0x03f1,), # 'ρ': 'ϱ'
+    # GREEK SMALL LETTER FINAL SIGMA: GREEK SMALL LETTER SIGMA
+    0x03c2: (0x03c3,), # 'ς': 'σ'
+    # GREEK SMALL LETTER SIGMA: GREEK SMALL LETTER FINAL SIGMA
+    0x03c3: (0x03c2,), # 'σ': 'ς'
+    # GREEK SMALL LETTER PHI: GREEK PHI SYMBOL
+    0x03c6: (0x03d5,), # 'φ': 'ϕ'
+    # GREEK BETA SYMBOL: GREEK SMALL LETTER BETA
+    0x03d0: (0x03b2,), # 'ϐ': 'β'
+    # GREEK THETA SYMBOL: GREEK SMALL LETTER THETA
+    0x03d1: (0x03b8,), # 'ϑ': 'θ'
+    # GREEK PHI SYMBOL: GREEK SMALL LETTER PHI
+    0x03d5: (0x03c6,), # 'ϕ': 'φ'
+    # GREEK PI SYMBOL: GREEK SMALL LETTER PI
+    0x03d6: (0x03c0,), # 'ϖ': 'π'
+    # GREEK KAPPA SYMBOL: GREEK SMALL LETTER KAPPA
+    0x03f0: (0x03ba,), # 'ϰ': 'κ'
+    # GREEK RHO SYMBOL: GREEK SMALL LETTER RHO
+    0x03f1: (0x03c1,), # 'ϱ': 'ρ'
+    # GREEK LUNATE EPSILON SYMBOL: GREEK SMALL LETTER EPSILON
+    0x03f5: (0x03b5,), # 'ϵ': 'ε'
+    # CYRILLIC SMALL LETTER VE: CYRILLIC SMALL LETTER ROUNDED VE
+    0x0432: (0x1c80,), # 'в': 'ᲀ'
+    # CYRILLIC SMALL LETTER DE: CYRILLIC SMALL LETTER LONG-LEGGED DE
+    0x0434: (0x1c81,), # 'д': 'ᲁ'
+    # CYRILLIC SMALL LETTER O: CYRILLIC SMALL LETTER NARROW O
+    0x043e: (0x1c82,), # 'о': 'ᲂ'
+    # CYRILLIC SMALL LETTER ES: CYRILLIC SMALL LETTER WIDE ES
+    0x0441: (0x1c83,), # 'с': 'ᲃ'
+    # CYRILLIC SMALL LETTER TE: CYRILLIC SMALL LETTER TALL TE, CYRILLIC SMALL LETTER THREE-LEGGED TE
+    0x0442: (0x1c84, 0x1c85), # 'т': 'ᲄᲅ'
+    # CYRILLIC SMALL LETTER HARD SIGN: CYRILLIC SMALL LETTER TALL HARD SIGN
+    0x044a: (0x1c86,), # 'ъ': 'ᲆ'
+    # CYRILLIC SMALL LETTER YAT: CYRILLIC SMALL LETTER TALL YAT
+    0x0463: (0x1c87,), # 'ѣ': 'ᲇ'
+    # CYRILLIC SMALL LETTER ROUNDED VE: CYRILLIC SMALL LETTER VE
+    0x1c80: (0x0432,), # 'ᲀ': 'в'
+    # CYRILLIC SMALL LETTER LONG-LEGGED DE: CYRILLIC SMALL LETTER DE
+    0x1c81: (0x0434,), # 'ᲁ': 'д'
+    # CYRILLIC SMALL LETTER NARROW O: CYRILLIC SMALL LETTER O
+    0x1c82: (0x043e,), # 'ᲂ': 'о'
+    # CYRILLIC SMALL LETTER WIDE ES: CYRILLIC SMALL LETTER ES
+    0x1c83: (0x0441,), # 'ᲃ': 'с'
+    # CYRILLIC SMALL LETTER TALL TE: CYRILLIC SMALL LETTER TE, CYRILLIC SMALL LETTER THREE-LEGGED TE
+    0x1c84: (0x0442, 0x1c85), # 'ᲄ': 'тᲅ'
+    # CYRILLIC SMALL LETTER THREE-LEGGED TE: CYRILLIC SMALL LETTER TE, CYRILLIC SMALL LETTER TALL TE
+    0x1c85: (0x0442, 0x1c84), # 'ᲅ': 'тᲄ'
+    # CYRILLIC SMALL LETTER TALL HARD SIGN: CYRILLIC SMALL LETTER HARD SIGN
+    0x1c86: (0x044a,), # 'ᲆ': 'ъ'
+    # CYRILLIC SMALL LETTER TALL YAT: CYRILLIC SMALL LETTER YAT
+    0x1c87: (0x0463,), # 'ᲇ': 'ѣ'
+    # CYRILLIC SMALL LETTER UNBLENDED UK: CYRILLIC SMALL LETTER MONOGRAPH UK
+    0x1c88: (0xa64b,), # 'ᲈ': 'ꙋ'
+    # LATIN SMALL LETTER S WITH DOT ABOVE: LATIN SMALL LETTER LONG S WITH DOT ABOVE
+    0x1e61: (0x1e9b,), # 'ṡ': 'ẛ'
+    # LATIN SMALL LETTER LONG S WITH DOT ABOVE: LATIN SMALL LETTER S WITH DOT ABOVE
+    0x1e9b: (0x1e61,), # 'ẛ': 'ṡ'
+    # GREEK PROSGEGRAMMENI: COMBINING GREEK YPOGEGRAMMENI, GREEK SMALL LETTER IOTA
+    0x1fbe: (0x0345, 0x03b9), # 'ι': '\u0345ι'
+    # GREEK SMALL LETTER IOTA WITH DIALYTIKA AND OXIA: GREEK SMALL LETTER IOTA WITH DIALYTIKA AND TONOS
+    0x1fd3: (0x0390,), # 'ΐ': 'ΐ'
+    # GREEK SMALL LETTER UPSILON WITH DIALYTIKA AND OXIA: GREEK SMALL LETTER UPSILON WITH DIALYTIKA AND TONOS
+    0x1fe3: (0x03b0,), # 'ΰ': 'ΰ'
+    # CYRILLIC SMALL LETTER MONOGRAPH UK: CYRILLIC SMALL LETTER UNBLENDED UK
+    0xa64b: (0x1c88,), # 'ꙋ': 'ᲈ'
+    # LATIN SMALL LIGATURE LONG S T: LATIN SMALL LIGATURE ST
+    0xfb05: (0xfb06,), # 'ﬅ': 'ﬆ'
+    # LATIN SMALL LIGATURE ST: LATIN SMALL LIGATURE LONG S T
+    0xfb06: (0xfb05,), # 'ﬆ': 'ﬅ'
+}
diff --git a/Lib/re/_compiler.py b/Lib/re/_compiler.py
@@ -13,6 +13,7 @@
 import _sre
 from . import _parser
 from ._constants import *
+from ._casefix import _EXTRA_CASES
 
 assert _sre.MAGIC == MAGIC, "SRE module mismatch"
 
@@ -27,62 +28,6 @@
     POSSESSIVE_REPEAT: (POSSESSIVE_REPEAT, SUCCESS, POSSESSIVE_REPEAT_ONE),
 }
 
-# Sets of lowercase characters which have the same uppercase.
-_equivalences = (
-    # LATIN SMALL LETTER I, LATIN SMALL LETTER DOTLESS I
-    (0x69, 0x131), # iı
-    # LATIN SMALL LETTER S, LATIN SMALL LETTER LONG S
-    (0x73, 0x17f), # sſ
-    # MICRO SIGN, GREEK SMALL LETTER MU
-    (0xb5, 0x3bc), # µμ
-    # COMBINING GREEK YPOGEGRAMMENI, GREEK SMALL LETTER IOTA, GREEK PROSGEGRAMMENI
-    (0x345, 0x3b9, 0x1fbe), # \u0345ιι
-    # GREEK SMALL LETTER IOTA WITH DIALYTIKA AND TONOS, GREEK SMALL LETTER IOTA WITH DIALYTIKA AND OXIA
-    (0x390, 0x1fd3), # ΐΐ
-    # GREEK SMALL LETTER UPSILON WITH DIALYTIKA AND TONOS, GREEK SMALL LETTER UPSILON WITH DIALYTIKA AND OXIA
-    (0x3b0, 0x1fe3), # ΰΰ
-    # GREEK SMALL LETTER BETA, GREEK BETA SYMBOL
-    (0x3b2, 0x3d0), # βϐ
-    # GREEK SMALL LETTER EPSILON, GREEK LUNATE EPSILON SYMBOL
-    (0x3b5, 0x3f5), # εϵ
-    # GREEK SMALL LETTER THETA, GREEK THETA SYMBOL
-    (0x3b8, 0x3d1), # θϑ
-    # GREEK SMALL LETTER KAPPA, GREEK KAPPA SYMBOL
-    (0x3ba, 0x3f0), # κϰ
-    # GREEK SMALL LETTER PI, GREEK PI SYMBOL
-    (0x3c0, 0x3d6), # πϖ
-    # GREEK SMALL LETTER RHO, GREEK RHO SYMBOL
-    (0x3c1, 0x3f1), # ρϱ
-    # GREEK SMALL LETTER FINAL SIGMA, GREEK SMALL LETTER SIGMA
-    (0x3c2, 0x3c3), # ςσ
-    # GREEK SMALL LETTER PHI, GREEK PHI SYMBOL
-    (0x3c6, 0x3d5), # φϕ
-    # CYRILLIC SMALL LETTER VE, CYRILLIC SMALL LETTER ROUNDED VE
-    (0x432, 0x1c80), # вᲀ
-    # CYRILLIC SMALL LETTER DE, CYRILLIC SMALL LETTER LONG-LEGGED DE
-    (0x434, 0x1c81), # дᲁ
-    # CYRILLIC SMALL LETTER O, CYRILLIC SMALL LETTER NARROW O
-    (0x43e, 0x1c82), # оᲂ
-    # CYRILLIC SMALL LETTER ES, CYRILLIC SMALL LETTER WIDE ES
-    (0x441, 0x1c83), # сᲃ
-    # CYRILLIC SMALL LETTER TE, CYRILLIC SMALL LETTER TALL TE, CYRILLIC SMALL LETTER THREE-LEGGED TE
-    (0x442, 0x1c84, 0x1c85), # тᲄᲅ
-    # CYRILLIC SMALL LETTER HARD SIGN, CYRILLIC SMALL LETTER TALL HARD SIGN
-    (0x44a, 0x1c86), # ъᲆ
-    # CYRILLIC SMALL LETTER YAT, CYRILLIC SMALL LETTER TALL YAT
-    (0x463, 0x1c87), # ѣᲇ
-    # CYRILLIC SMALL LETTER UNBLENDED UK, CYRILLIC SMALL LETTER MONOGRAPH UK
-    (0x1c88, 0xa64b), # ᲈꙋ
-    # LATIN SMALL LETTER S WITH DOT ABOVE, LATIN SMALL LETTER LONG S WITH DOT ABOVE
-    (0x1e61, 0x1e9b), # ṡẛ
-    # LATIN SMALL LIGATURE LONG S T, LATIN SMALL LIGATURE ST
-    (0xfb05, 0xfb06), # ﬅﬆ
-)
-
-# Maps the lowercase code to lowercase codes which have the same uppercase.
-_ignorecase_fixes = {i: tuple(j for j in t if i != j)
-                     for t in _equivalences for i in t}
-
 class _CompileData:
     __slots__ = ('code', 'repeat_count')
     def __init__(self):
@@ -111,7 +56,7 @@ def _compile(data, pattern, flags):
         if flags & SRE_FLAG_UNICODE:
             iscased = _sre.unicode_iscased
             tolower = _sre.unicode_tolower
-            fixes = _ignorecase_fixes
+            fixes = _EXTRA_CASES
         else:
             iscased = _sre.ascii_iscased
             tolower = _sre.ascii_tolower
diff --git a/Makefile.pre.in b/Makefile.pre.in
@@ -948,6 +948,12 @@ regen-test-frozenmain: $(BUILDPYTHON)
 	# using Programs/freeze_test_frozenmain.py
 	$(RUNSHARED) ./$(BUILDPYTHON) $(srcdir)/Programs/freeze_test_frozenmain.py Programs/test_frozenmain.h
 
+.PHONY: regen-re
+regen-re: $(BUILDPYTHON)
+	# Regenerate Lib/re/_casefix.py
+	# using Tools/scripts/generate_re_casefix.py
+	$(RUNSHARED) ./$(BUILDPYTHON) $(srcdir)/Tools/scripts/generate_re_casefix.py $(srcdir)/Lib/re/_casefix.py
+
 Programs/_testembed: Programs/_testembed.o $(LINK_PYTHON_DEPS)
 	$(LINKCC) $(PY_CORE_LDFLAGS) $(LINKFORSHARED) -o $@ Programs/_testembed.o $(LINK_PYTHON_OBJS) $(LIBS) $(MODLIBS) $(SYSLIBS)
 
diff --git a/Misc/NEWS.d/next/Tools-Demos/2022-04-18-12-52-16.gh-issue-91575.fK1TEh.rst b/Misc/NEWS.d/next/Tools-Demos/2022-04-18-12-52-16.gh-issue-91575.fK1TEh.rst
@@ -0,0 +1,3 @@
+Add script ``Tools/scripts/generate_re_casefix.py`` and the make target
+``regen-re`` for generating additional data for case-insensitive matching
+according to the current Unicode version.
diff --git a/Tools/scripts/generate_re_casefix.py b/Tools/scripts/generate_re_casefix.py
@@ -0,0 +1,95 @@
+#! /usr/bin/env python3
+# This script generates Lib/re/_casefix.py.
+
+import collections
+import re
+import sys
+import unicodedata
+
+def update_file(file, content):
+    try:
+        with open(file, 'r', encoding='utf-8') as fobj:
+            if fobj.read() == content:
+                return False
+    except (OSError, ValueError):
+        pass
+    with open(file, 'w', encoding='utf-8') as fobj:
+        fobj.write(content)
+    return True
+
+re_casefix_template = """\
+# Auto-generated by Tools/scripts/generate_re_casefix.py.
+
+# Maps the code of lowercased character to codes of different lowercased
+# characters which have the same uppercase.
+_EXTRA_CASES = {
+%s
+}
+"""
+
+def uname(i):
+    return unicodedata.name(chr(i), r'U+%04X' % i)
+
+class hexint(int):
+    def __repr__(self):
+        return '%#06x' % self
+
+def alpha(i):
+    c = chr(i)
+    return c if c.isalpha() else ascii(c)[1:-1]
+
+
+def main(outfile='Lib/re/_casefix.py'):
+    # Find sets of characters which have the same uppercase.
+    equivalent_chars = collections.defaultdict(str)
+    for c in map(chr, range(sys.maxunicode + 1)):
+        equivalent_chars[c.upper()] += c
+    equivalent_chars = [t for t in equivalent_chars.values() if len(t) > 1]
+
+    # List of codes of lowercased characters which have the same uppercase.
+    equivalent_lower_codes = [sorted(t)
+                              for s in equivalent_chars
+                              for t in [set(ord(c.lower()) for c in s)]
+                              if len(t) > 1]
+
+    bad_codes = []
+    for t in equivalent_lower_codes:
+        for i in t:
+            if i > 0xffff:
+                bad_codes.extend(t)
+                try:
+                    bad_codes.append(ord(chr(i).upper()))
+                except (ValueError, TypeError):
+                    pass
+                break
+    if bad_codes:
+        print('Case-insensitive matching may not work correctly for character:',
+              file=sys.stderr)
+        for i in sorted(bad_codes):
+            print("  '%s' (U+%04x, %s)" % (alpha(i), i, uname(i)),
+                  file=sys.stderr)
+        sys.exit(1)
+
+    mapping = {i: tuple(j for j in t if i != j)
+               for t in equivalent_lower_codes
+               for i in t}
+
+    items = []
+    for i, t in sorted(mapping.items()):
+        items.append('    # %s: %s' % (
+            uname(i),
+            ', '.join(map(uname, t)),
+        ))
+        items.append("    %r: %r, # '%s': '%s'" % (
+            hexint(i),
+            tuple(map(hexint, t)),
+            alpha(i),
+            ''.join(map(alpha, t)),
+        ))
+
+    update_file(outfile, re_casefix_template % '\n'.join(items))
+
+
+if __name__ == '__main__':
+    import sys
+    main(*sys.argv[1:])

Original file line number	Diff line number	Diff line change
`@@ -0,0 +1,3 @@`
	`1`	+Add script ``Tools/scripts/generate_re_casefix.py`` and the make target
	`2`	+``regen-re`` for generating additional data for case-insensitive matching
	`3`	`+according to the current Unicode version.`