feat: replay changes from @otetard

andreasvc#51 - thank you!

Author: tyteen4a03

pyproject.toml

@@ -5,7 +5,7 @@ requires = [
     "Cython>=0.20",
     "pybind11>=2.12",
     "ninja; sys_platform != 'Windows'",
-    "cmake>=3.18",
+    "cmake (>=3.18,<4.0.0)",
 ]
 
 build-backend = "setuptools.build_meta"

src/compile.pxi

@@ -42,14 +42,13 @@ def _compile(object pattern, int flags=0, int max_mem=8388608):
         return fallback(original_pattern, flags, "re.LOCALE not supported")
     pattern = unicode_to_bytes(pattern, &encoded, -1)
     newflags = flags
-    if not PY2:
-        if not encoded and flags & _U:  # re.UNICODE
-            pass  # can use UNICODE with bytes pattern, but assumes valid UTF-8
-            # raise ValueError("can't use UNICODE flag with a bytes pattern")
-        elif encoded and not (flags & ASCII):  # re.ASCII (not in Python 2)
-            newflags = flags | _U  # re.UNICODE
-        elif encoded and flags & ASCII:
-            newflags = flags & ~_U  # re.UNICODE
+    if not encoded and flags & _U:  # re.UNICODE
+        pass  # can use UNICODE with bytes pattern, but assumes valid UTF-8
+        # raise ValueError("can't use UNICODE flag with a bytes pattern")
+    elif encoded and not (flags & ASCII):  # re.ASCII (not in Python 2)
+        newflags = flags | _U  # re.UNICODE
+    elif encoded and flags & ASCII:
+        newflags = flags & ~_U  # re.UNICODE
     try:
         pattern = _prepare_pattern(pattern, newflags)
     except BackreferencesException:

src/includes.pxi

@@ -12,11 +12,6 @@ cdef extern from *:
     cdef void emit_endif "#endif //" ()
 
 
-cdef extern from "Python.h":
-    int PyObject_CheckReadBuffer(object)
-    int PyObject_AsReadBuffer(object, const void **, Py_ssize_t *)
-
-
 cdef extern from "re2/stringpiece.h" namespace "re2":
     cdef cppclass StringPiece:
         StringPiece()

src/match.pxi

@@ -115,12 +115,12 @@ cdef class Match:
         """Expand a template with groups."""
         cdef bytearray result = bytearray()
         if isinstance(template, unicode):
-            if not PY2 and not self.encoded:
+            if not self.encoded:
                 raise ValueError(
                         'cannot expand unicode template on bytes pattern')
             templ = template.encode('utf8')
         else:
-            if not PY2 and self.encoded:
+            if self.encoded:
                 raise ValueError(
                         'cannot expand bytes template on unicode pattern')
             templ = bytes(template)

src/pattern.pxi

@@ -417,7 +417,7 @@ cdef class Pattern:
         if not repl_encoded and not isinstance(repl, bytes):
             repl_b = bytes(repl)  # coerce buffer to bytes object
 
-        if count > 1 or (b'\\' if PY2 else <char>b'\\') in repl_b:
+        if count > 1 or <char>b'\\' in repl_b:
             # Limit on number of substitutions or replacement string contains
             # escape sequences; handle with Match.expand() implementation.
             # RE2 does support simple numeric group references \1, \2,

src/re2.pyx

@@ -151,7 +151,6 @@ VERSION_HEX = 0x000217
 
 cdef int _I = I, _M = M, _S = S, _U = U, _X = X, _L = L
 cdef int current_notification = FALLBACK_QUIETLY
-cdef bint PY2 = PY_MAJOR_VERSION == 2
 
 # Type of compiled re object from Python stdlib
 SREPattern = type(re.compile(''))
@@ -252,7 +251,7 @@ def escape(pattern):
     """Escape all non-alphanumeric characters in pattern."""
     cdef bint uni = isinstance(pattern, unicode)
     cdef list s
-    if PY2 or uni:
+    if uni:
         s = list(pattern)
     else:
         s = [bytes([c]) for c in pattern]
@@ -350,9 +349,9 @@ cdef inline unicode_to_bytes(object pystring, int * encoded,
         encoded[0] = 1 if origlen == len(pystring) else 2
     else:
         encoded[0] = 0
-    if not PY2 and checkotherencoding > 0 and not encoded[0]:
+    if checkotherencoding > 0 and not encoded[0]:
         raise TypeError("can't use a string pattern on a bytes-like object")
-    elif not PY2 and checkotherencoding == 0 and encoded[0]:
+    elif checkotherencoding == 0 and encoded[0]:
         raise TypeError("can't use a bytes pattern on a string-like object")
     return pystring
 
@@ -366,14 +365,7 @@ cdef inline int pystring_to_cstring(
     cdef int result = -1
     cstring[0] = NULL
     size[0] = 0
-    if PY2:
-        # Although the new-style buffer interface was backported to Python 2.6,
-        # some modules, notably mmap, only support the old buffer interface.
-        # Cf. http://bugs.python.org/issue9229
-        if PyObject_CheckReadBuffer(pystring) == 1:
-            result = PyObject_AsReadBuffer(
-                    pystring, <const void **>cstring, size)
-    elif PyObject_CheckBuffer(pystring) == 1:  # new-style Buffer interface
+    if PyObject_CheckBuffer(pystring) == 1:  # new-style Buffer interface
         result = PyObject_GetBuffer(pystring, buf, PyBUF_SIMPLE)
         if result == 0:
             cstring[0] = <char *>buf.buf
@@ -383,8 +375,7 @@ cdef inline int pystring_to_cstring(
 
 cdef inline void release_cstring(Py_buffer *buf):
     """Release buffer if necessary."""
-    if not PY2:
-        PyBuffer_Release(buf)
+    PyBuffer_Release(buf)
 
 
 cdef utf8indices(char * cstring, Py_ssize_t size, Py_ssize_t *pos, Py_ssize_t *endpos):

tests/test_charliterals.py

@@ -0,0 +1,40 @@
+import re2 as re
+import warnings
+
+warnings.filterwarnings('ignore', category=DeprecationWarning)
+
+import unittest
+
+class TestCharLiterals(unittest.TestCase):
+    def test_character_literals(self):
+        i = 126
+
+        assert re.compile(r"\%03o" % i) == re.compile('\\176')
+        assert re.compile(r"\%03o" % i)._dump_pattern() == '\\176'
+        assert (re.match(r"\%03o" % i, chr(i)) is None) == False
+        assert (re.match(r"\%03o0" % i, chr(i) + "0") is None) == False
+        assert (re.match(r"\%03o8" % i, chr(i) + "8") is None) == False
+        assert (re.match(r"\x%02x" % i, chr(i)) is None) == False
+        assert (re.match(r"\x%02x0" % i, chr(i) + "0") is None) == False
+        assert (re.match(r"\x%02xz" % i, chr(i) + "z") is None) == False
+
+        try:
+            re.match("\911", "")
+        except Exception as exp:
+            assert exp.msg == "invalid group reference 91 at position 1"
+
+
+    def test_character_class_literals(self):
+        i = 126
+
+        assert (re.match(r"[\%03o]" % i, chr(i)) is None) == False
+        assert (re.match(r"[\%03o0]" % i, chr(i) + "0") is None) == False
+        assert (re.match(r"[\%03o8]" % i, chr(i) + "8") is None) == False
+        assert (re.match(r"[\x%02x]" % i, chr(i)) is None) == False
+        assert (re.match(r"[\x%02x0]" % i, chr(i) + "0") is None) == False
+        assert (re.match(r"[\x%02xz]" % i, chr(i) + "z") is None) == False
+
+        try:
+            re.match("[\911]", "")
+        except Exception as exp:
+            assert exp.msg == "invalid escape sequence: \9"

tests/test_charliterals.txt

@@ -1,12 +1,11 @@
 [tox]
-envlist = py3{7,8,9,10,11,12,13}
+envlist = py3{8,9,10,11,12,13}
 skip_missing_interpreters = true
 isolated_build = true
 skipsdist=True
 
 [gh-actions]
 python =
-    3.7: py37
     3.8: py38
     3.9: py39
     3.10: py310
@@ -167,7 +166,7 @@ deps =
     pip>=20.0.1
 
 commands =
-    pip install pyre2-updated --force-reinstall --prefer-binary -f dist/
+    pip install pyre2-updated --force-reinstall --prefer-binary --no-index -f dist/
     python -m unittest discover -f -s .
 
 [testenv:style]