Re-organised files.

Updated to Unicode 17.0.0.

Author: mrabarnett

MANIFEST.in

@@ -1,7 +1,15 @@
-include regex_3/*.c
-include regex_3/*.h
-include regex_3/*.py
-include docs/*.*
-include tools/*.py
+exclude *.*
+include docs/Features.html
+include docs/UnicodeProperties.rst
 include LICENSE.txt
 include pyproject.toml
+include README.rst
+include regex/__init__.py
+include regex/_main.py
+include regex/_regex_core.py
+include regex/test_regex.py
+include src/_regex.c
+include src/_regex.h
+include src/_regex_unicode.c
+include src/_regex_unicode.h
+include tools/build_regex_unicode.py

README.rst

@@ -21,7 +21,7 @@ The regex module releases the GIL during matching on instances of the built-in (
 Unicode
 -------
 
-This module supports Unicode 16.0.0. Full Unicode case-folding is supported.
+This module supports Unicode 17.0.0. Full Unicode case-folding is supported.
 
 Flags
 -----

docs/Features.html

@@ -382,7 +382,7 @@ <h1>Multithreading</h1>
 </div>
 <div class="section" id="unicode">
 <h1>Unicode</h1>
-<p>This module supports Unicode 16.0.0. Full Unicode case-folding is supported.</p>
+<p>This module supports Unicode 17.0.0. Full Unicode case-folding is supported.</p>
 </div>
 <div class="section" id="flags">
 <h1>Flags</h1>

pyproject.toml

@@ -35,10 +35,10 @@ requires-python = ">= 3.9"
 Homepage = "https://github.com/mrabarnett/mrab-regex"
 
 [tool.setuptools]
-package-dir = {regex = "regex_3"}
+package-dir = {"regex" = "regex"}
 py-modules = [
     "regex.__init__",
-    "regex.regex",
+    "regex._main",
     "regex._regex_core",
     "regex.test_regex",
 ]

regex/__init__.py

@@ -0,0 +1,3 @@
+import regex._main
+from regex._main import *
+__all__ = regex._main.__all__

regex_3/regex.py renamed to regex/_main.py

@@ -241,7 +241,7 @@
   "VERSION1", "X", "VERBOSE", "W", "WORD", "error", "Regex", "__version__",
   "__doc__", "RegexFlag"]
 
-__version__ = "2.5.162"
+__version__ = "2025.9.18"
 
 # --------------------------------------------------------------------
 # Public interface.
@@ -414,8 +414,8 @@ def escape(pattern, special_only=True, literal_spaces=False):
 # --------------------------------------------------------------------
 # Internals.
 
-import regex._regex_core as _regex_core
-import regex._regex as _regex
+from regex import _regex_core
+from regex import _regex
 from threading import RLock as _RLock
 from locale import getpreferredencoding as _getpreferredencoding
 from regex._regex_core import *
@@ -429,7 +429,7 @@ def escape(pattern, special_only=True, literal_spaces=False):
 # Version 0 is the old behaviour, compatible with the original 're' module.
 # Version 1 is the new behaviour, which differs slightly.
 
-DEFAULT_VERSION = VERSION0
+DEFAULT_VERSION = RegexFlag.VERSION0
 
 _METACHARS = frozenset("()[]{}?*+|^$\\.-#&~")
 

regex_3/_regex_core.py renamed to regex/_regex_core.py

@@ -18,7 +18,7 @@
 import unicodedata
 from collections import defaultdict
 
-import regex._regex as _regex
+from regex import _regex
 
 __all__ = ["A", "ASCII", "B", "BESTMATCH", "D", "DEBUG", "E", "ENHANCEMATCH",
   "F", "FULLCASE", "I", "IGNORECASE", "L", "LOCALE", "M", "MULTILINE", "P",
@@ -121,7 +121,42 @@ def __repr__(self):
 
     __str__ = object.__str__
 
-globals().update(RegexFlag.__members__)
+# Put the flags into the module namespace. Being explicit here helps tools like
+# linters and IDEs understand the code better.
+ASCII = RegexFlag.ASCII
+BESTMATCH = RegexFlag.BESTMATCH
+DEBUG = RegexFlag.DEBUG
+DOTALL = RegexFlag.DOTALL
+ENHANCEMATCH = RegexFlag.ENHANCEMATCH
+FULLCASE = RegexFlag.FULLCASE
+IGNORECASE = RegexFlag.IGNORECASE
+LOCALE = RegexFlag.LOCALE
+MULTILINE = RegexFlag.MULTILINE
+POSIX = RegexFlag.POSIX
+REVERSE = RegexFlag.REVERSE
+TEMPLATE = RegexFlag.TEMPLATE
+UNICODE = RegexFlag.UNICODE
+VERBOSE = RegexFlag.VERBOSE
+VERSION0 = RegexFlag.VERSION0
+VERSION1 = RegexFlag.VERSION1
+WORD = RegexFlag.WORD
+A = RegexFlag.A
+B = RegexFlag.B
+D = RegexFlag.D
+E = RegexFlag.E
+F = RegexFlag.F
+I = RegexFlag.I
+L = RegexFlag.L
+M = RegexFlag.M
+P = RegexFlag.P
+R = RegexFlag.R
+S = RegexFlag.S
+U = RegexFlag.U
+V0 = RegexFlag.V0
+V1 = RegexFlag.V1
+W = RegexFlag.W
+X = RegexFlag.X
+T = RegexFlag.T
 
 DEFAULT_VERSION = VERSION1
 
@@ -2488,7 +2523,7 @@ def fix_groups(self, pattern, reverse, fuzzy):
         self._key = self.__class__, self.group
 
     def remove_captures(self):
-        raise error("group reference not allowed", pattern, self.position)
+        raise error("group reference not allowed", self.pattern, self.position)
 
     def _compile(self, reverse, fuzzy):
         return [(OP.GROUP_CALL, self.call_ref)]
@@ -3058,7 +3093,7 @@ def _compile(self, reverse, fuzzy):
     def dump(self, indent, reverse):
         group = self.group
         if group < 0:
-            group = private_groups[group]
+            group = self.info.private_groups[group]
         print("{}GROUP {}".format(INDENT * indent, group))
         self.subpattern.dump(indent + 1, reverse)
 
@@ -3413,7 +3448,7 @@ def fix_groups(self, pattern, reverse, fuzzy):
         self._key = self.__class__, self.group, self.case_flags
 
     def remove_captures(self):
-        raise error("group reference not allowed", pattern, self.position)
+        raise error("group reference not allowed", self.pattern, self.position)
 
     def _compile(self, reverse, fuzzy):
         flags = 0

regex_3/test_regex.py renamed to regex/test_regex.py

@@ -883,7 +883,7 @@ def test_bug_817234(self):
     def test_empty_array(self):
         # SF buf 1647541.
         import array
-        for typecode in 'bBuhHiIlLfd':
+        for typecode in 'bBhHiIlLfd':
             a = array.array(typecode)
             self.assertEqual(regex.compile(b"bla").match(a), None)
             self.assertEqual(regex.compile(b"").match(a)[1 : ], ())

regex_3/__init__.py

@@ -11,7 +11,7 @@
     macros.append(("Py_GIL_DISABLED", "1"))
 
 setup(
-    ext_modules=[Extension('regex._regex', [join('regex_3', '_regex.c'),
-      join('regex_3', '_regex_unicode.c')])],
+    ext_modules=[Extension('regex._regex', ['src/_regex.c',
+      'src/_regex_unicode.c'])],
       define_macros=macros,
 )