unix-ffi: re: convert to PCRE2

PCRE is marked as EOL and won't receive any new security update.

Convert the re module to PCRE2 API to enforce security.
Additional dependency is now needed with uctypes due to changes in how
PCRE2 return the match_data in a pointer and require special handling.

The converted module is tested with the test_re.py with no regression.

Signed-off-by: Christian Marangi <[email protected]>

Author: Ansuel

unix-ffi/re/re.py

@@ -1,36 +1,49 @@
 import sys
-import ffilib
 import array
+import ffilib
+import uctypes
 
+pcre2 = ffilib.open("libpcre2-8")
 
-pcre = ffilib.open("libpcre")
+#       pcre2_code *pcre2_compile(PCRE2_SPTR pattern, PCRE2_SIZE length,
+#           uint32_t options, int *errorcode, PCRE2_SIZE *erroroffset,
+#           pcre2_compile_context *ccontext);
+pcre2_compile = pcre2.func("p", "pcre2_compile_8", "siippp")
 
-#       pcre *pcre_compile(const char *pattern, int options,
-#            const char **errptr, int *erroffset,
-#            const unsigned char *tableptr);
-pcre_compile = pcre.func("p", "pcre_compile", "sipps")
+#       int pcre2_match(const pcre2_code *code, PCRE2_SPTR subject,
+#           PCRE2_SIZE length, PCRE2_SIZE startoffset, uint32_t options,
+#           pcre2_match_data *match_data, pcre2_match_context *mcontext);
+pcre2_match = pcre2.func("i", "pcre2_match_8", "Psiiipp")
 
-#       int pcre_exec(const pcre *code, const pcre_extra *extra,
-#            const char *subject, int length, int startoffset,
-#            int options, int *ovector, int ovecsize);
-pcre_exec = pcre.func("i", "pcre_exec", "PPsiiipi")
+#       int pcre2_pattern_info(const pcre2_code *code, uint32_t what,
+#           void *where);
+pcre2_pattern_info = pcre2.func("i", "pcre2_pattern_info_8", "Pip")
 
-#       int pcre_fullinfo(const pcre *code, const pcre_extra *extra,
-#            int what, void *where);
-pcre_fullinfo = pcre.func("i", "pcre_fullinfo", "PPip")
+#       PCRE2_SIZE *pcre2_get_ovector_pointer(pcre2_match_data *match_data);
+pcre2_get_ovector_pointer = pcre2.func("p", "pcre2_get_ovector_pointer_8", "p")
 
+#       pcre2_match_data *pcre2_match_data_create_from_pattern(const pcre2_code *code,
+#           pcre2_general_context *gcontext);
+pcre2_match_data_create_from_pattern = pcre2.func(
+    "p", "pcre2_match_data_create_from_pattern_8", "Pp"
+)
 
-IGNORECASE = I = 1
-MULTILINE = M = 2
-DOTALL = S = 4
+IGNORECASE = I = 0x8
+MULTILINE = M = 0x400
+DOTALL = S = 0x20
 VERBOSE = X = 8
-PCRE_ANCHORED = 0x10
+PCRE2_ANCHORED = 0x80000000
+
+# Real value in pcre2.h is 0xFFFFFFFF for 32bit and
+# 0x0xFFFFFFFFFFFFFFFF for 64bit that is equivalent
+# to -1
+PCRE2_ZERO_TERMINATED = -1
 
 # TODO. Note that Python3 has unicode by default
 ASCII = A = 0
 UNICODE = U = 0
 
-PCRE_INFO_CAPTURECOUNT = 2
+PCRE2_INFO_CAPTURECOUNT = 4
 
 
 class PCREMatch:
@@ -67,19 +80,26 @@ def __init__(self, compiled_ptn):
     def search(self, s, pos=0, endpos=-1, _flags=0):
         assert endpos == -1, "pos: %d, endpos: %d" % (pos, endpos)
         buf = array.array("i", [0])
-        pcre_fullinfo(self.obj, None, PCRE_INFO_CAPTURECOUNT, buf)
+        pcre2_pattern_info(self.obj, PCRE2_INFO_CAPTURECOUNT, buf)
         cap_count = buf[0]
-        ov = array.array("i", [0, 0, 0] * (cap_count + 1))
-        num = pcre_exec(self.obj, None, s, len(s), pos, _flags, ov, len(ov))
+        match_data = pcre2_match_data_create_from_pattern(self.obj, None)
+        num = pcre2_match(self.obj, s, len(s), pos, _flags, match_data, None)
         if num == -1:
             # No match
             return None
+        ov_ptr = pcre2_get_ovector_pointer(match_data)
+        # pcre2_get_ovector_pointer return PCRE2_SIZE that is of type
+        # size_t. Use ULONG as type to support both 32bit and 64bit.
+        ov_buf = uctypes.bytearray_at(
+            ov_ptr, uctypes.sizeof({"field": 0 | uctypes.ULONG}) * (cap_count + 1) * 2
+        )
+        ov = array.array("L", ov_buf)
         # We don't care how many matching subexpressions we got, we
         # care only about total # of capturing ones (including empty)
         return PCREMatch(s, cap_count + 1, ov)
 
     def match(self, s, pos=0, endpos=-1):
-        return self.search(s, pos, endpos, PCRE_ANCHORED)
+        return self.search(s, pos, endpos, PCRE2_ANCHORED)
 
     def sub(self, repl, s, count=0):
         if not callable(repl):
@@ -141,9 +161,9 @@ def findall(self, s):
 
 
 def compile(pattern, flags=0):
-    errptr = bytes(4)
+    errcode = bytes(4)
     erroffset = bytes(4)
-    regex = pcre_compile(pattern, flags, errptr, erroffset, None)
+    regex = pcre2_compile(pattern, PCRE2_ZERO_TERMINATED, flags, errcode, erroffset, None)
     assert regex
     return PCREPattern(regex)
 
@@ -154,7 +174,7 @@ def search(pattern, string, flags=0):
 
 
 def match(pattern, string, flags=0):
-    r = compile(pattern, flags | PCRE_ANCHORED)
+    r = compile(pattern, flags | PCRE2_ANCHORED)
     return r.search(string)