[pypy-commit] pypy default: Support for re.fullpatch(), for Python >= 3.4

[arigo]

Author: Armin Rigo <arigo at tunes.org>
Branch: 
Changeset: r86692:6d29597bf1ac
Date: 2016-08-29 14:36 +0200
http://bitbucket.org/pypy/pypy/changeset/6d29597bf1ac/

Log:	Support for re.fullpatch(), for Python >= 3.4

diff --git a/rpython/rlib/rsre/rsre_char.py b/rpython/rlib/rsre/rsre_char.py
--- a/rpython/rlib/rsre/rsre_char.py
+++ b/rpython/rlib/rsre/rsre_char.py
@@ -52,6 +52,7 @@
 SRE_INFO_CHARSET = 4
 SRE_FLAG_LOCALE = 4 # honour system locale
 SRE_FLAG_UNICODE = 32 # use unicode locale
+SRE_FLAG_FULLMATCH = 0x4000   # PyPy extension, for CPython >= 3.4
 
 
 def getlower(char_ord, flags):
diff --git a/rpython/rlib/rsre/rsre_core.py b/rpython/rlib/rsre/rsre_core.py
--- a/rpython/rlib/rsre/rsre_core.py
+++ b/rpython/rlib/rsre/rsre_core.py
@@ -526,9 +526,16 @@
         if op == OPCODE_FAILURE:
             return
 
-        if (op == OPCODE_SUCCESS or
-            op == OPCODE_MAX_UNTIL or
-            op == OPCODE_MIN_UNTIL):
+        elif op == OPCODE_SUCCESS:
+            if ctx.flags & rsre_char.SRE_FLAG_FULLMATCH:
+                if ptr != ctx.end:
+                    return     # not a full match
+            ctx.match_end = ptr
+            ctx.match_marks = marks
+            return MATCHED_OK
+
+        elif (op == OPCODE_MAX_UNTIL or
+              op == OPCODE_MIN_UNTIL):
             ctx.match_end = ptr
             ctx.match_marks = marks
             return MATCHED_OK
@@ -1007,6 +1014,10 @@
     else:
         return None
 
+def fullmatch(pattern, string, start=0, end=sys.maxint, flags=0):
+    return match(pattern, string, start, end,
+                 flags | rsre_char.SRE_FLAG_FULLMATCH)
+
 def search(pattern, string, start=0, end=sys.maxint, flags=0):
     start, end = _adjust(start, end, len(string))
     ctx = StrMatchContext(pattern, string, start, end, flags)
diff --git a/rpython/rlib/rsre/test/test_match.py b/rpython/rlib/rsre/test/test_match.py
--- a/rpython/rlib/rsre/test/test_match.py
+++ b/rpython/rlib/rsre/test/test_match.py
@@ -272,3 +272,24 @@
         r = get_code("\\{\\{((?:.*?)+)\\}\\}")
         match = rsre_core.match(r, "{{a}}{{b}}")
         assert match.group(1) == "a"
+
+    def test_fullmatch_1(self):
+        r = get_code(r"ab*c")
+        assert not rsre_core.fullmatch(r, "abbbcdef")
+        assert rsre_core.fullmatch(r, "abbbc")
+
+    def test_fullmatch_2(self):
+        r = get_code(r"a(b*?)")
+        match = rsre_core.fullmatch(r, "abbb")
+        assert match.group(1) == "bbb"
+        assert not rsre_core.fullmatch(r, "abbbc")
+
+    def test_fullmatch_3(self):
+        r = get_code(r"a((bp)*?)c")
+        match = rsre_core.fullmatch(r, "abpbpbpc")
+        assert match.group(1) == "bpbpbp"
+
+    def test_fullmatch_4(self):
+        r = get_code(r"a((bp)*)c")
+        match = rsre_core.fullmatch(r, "abpbpbpc")
+        assert match.group(1) == "bpbpbp"