The usual

1 files changed, 233 insertions, 81 deletions

diff --git a/Lib/dos-8x3/sre_comp.py b/Lib/dos-8x3/sre_comp.py
index fa0cf37..97a57e2 100644
--- a/Lib/dos-8x3/sre_comp.py
+++ b/Lib/dos-8x3/sre_comp.py
@@ -5,22 +5,14 @@
 #
 # Copyright (c) 1997-2000 by Secret Labs AB.  All rights reserved.
 #
-# Portions of this engine have been developed in cooperation with
-# CNRI.  Hewlett-Packard provided funding for 1.6 integration and
-# other compatibility work.
+# See the sre.py file for information on usage and redistribution.
 #
 
-import array
 import _sre
 
 from sre_constants import *
 
-# find an array type code that matches the engine's code size
-for WORDSIZE in "BHil":
-    if len(array.array(WORDSIZE, [0]).tostring()) == _sre.getcodesize():
-        break
-else:
-    raise RuntimeError, "cannot find a usable array type"
+MAXCODE = 65535
 
 def _compile(code, pattern, flags):
     # internal: compile a (sub)pattern
@@ -41,34 +33,16 @@ def _compile(code, pattern, flags):
                 emit(OPCODES[op])
                 fixup = lambda x: x
             skip = len(code); emit(0)
-            for op, av in av:
-                emit(OPCODES[op])
-                if op is NEGATE:
-                    pass
-                elif op is LITERAL:
-                    emit(fixup(av))
-                elif op is RANGE:
-                    emit(fixup(av[0]))
-                    emit(fixup(av[1]))
-                elif op is CATEGORY:
-                    if flags & SRE_FLAG_LOCALE:
-                        emit(CHCODES[CH_LOCALE[av]])
-                    elif flags & SRE_FLAG_UNICODE:
-                        emit(CHCODES[CH_UNICODE[av]])
-                    else:
-                        emit(CHCODES[av])
-                else:
-                    raise error, "internal: unsupported set operator"
-            emit(OPCODES[FAILURE])
+            _compile_charset(av, flags, code, fixup)
             code[skip] = len(code) - skip
         elif op is ANY:
             if flags & SRE_FLAG_DOTALL:
-                emit(OPCODES[op])
+                emit(OPCODES[ANY_ALL])
             else:
-                emit(OPCODES[CATEGORY])
-                emit(CHCODES[CATEGORY_NOT_LINEBREAK])
+                emit(OPCODES[ANY])
         elif op in (REPEAT, MIN_REPEAT, MAX_REPEAT):
             if flags & SRE_FLAG_TEMPLATE:
+                raise error, "internal: unsupported template operator"
                 emit(OPCODES[REPEAT])
                 skip = len(code); emit(0)
                 emit(av[0])
@@ -76,40 +50,50 @@ def _compile(code, pattern, flags):
                 _compile(code, av[2], flags)
                 emit(OPCODES[SUCCESS])
                 code[skip] = len(code) - skip
+            elif _simple(av) and op == MAX_REPEAT:
+                emit(OPCODES[REPEAT_ONE])
+                skip = len(code); emit(0)
+                emit(av[0])
+                emit(av[1])
+                _compile(code, av[2], flags)
+                emit(OPCODES[SUCCESS])
+                code[skip] = len(code) - skip
             else:
-                lo, hi = av[2].getwidth()
-                if lo == 0:
-                    raise error, "nothing to repeat"
-                if 0 and lo == hi == 1 and op is MAX_REPEAT:
-                    # FIXME: <fl> need a better way to figure out when
-                    # it's safe to use this one (in the parser, probably)
-                    emit(OPCODES[MAX_REPEAT_ONE])
-                    skip = len(code); emit(0)
-                    emit(av[0])
-                    emit(av[1])
-                    _compile(code, av[2], flags)
-                    emit(OPCODES[SUCCESS])
-                    code[skip] = len(code) - skip
+                emit(OPCODES[REPEAT])
+                skip = len(code); emit(0)
+                emit(av[0])
+                emit(av[1])
+                _compile(code, av[2], flags)
+                code[skip] = len(code) - skip
+                if op == MAX_REPEAT:
+                    emit(OPCODES[MAX_UNTIL])
                 else:
-                    emit(OPCODES[op])
-                    skip = len(code); emit(0)
-                    emit(av[0])
-                    emit(av[1])
-                    _compile(code, av[2], flags)
-                    emit(OPCODES[SUCCESS])
-                    code[skip] = len(code) - skip
+                    emit(OPCODES[MIN_UNTIL])
         elif op is SUBPATTERN:
-            group = av[0]
-            if group:
+            if av[0]:
                 emit(OPCODES[MARK])
-                emit((group-1)*2)
+                emit((av[0]-1)*2)
+            # _compile_info(code, av[1], flags)
             _compile(code, av[1], flags)
-            if group:
+            if av[0]:
                 emit(OPCODES[MARK])
-                emit((group-1)*2+1)
+                emit((av[0]-1)*2+1)
         elif op in (SUCCESS, FAILURE):
             emit(OPCODES[op])
-        elif op in (ASSERT, ASSERT_NOT, CALL):
+        elif op in (ASSERT, ASSERT_NOT):
+            emit(OPCODES[op])
+            skip = len(code); emit(0)
+            if av[0] >= 0:
+                emit(0) # look ahead
+            else:
+                lo, hi = av[1].getwidth()
+                if lo != hi:
+                    raise error, "look-behind requires fixed-width pattern"
+                emit(lo) # look behind
+            _compile(code, av[1], flags)
+            emit(OPCODES[SUCCESS])
+            code[skip] = len(code) - skip
+        elif op is CALL:
             emit(OPCODES[op])
             skip = len(code); emit(0)
             _compile(code, av, flags)
@@ -126,6 +110,7 @@ def _compile(code, pattern, flags):
             tail = []
             for av in av[1]:
                 skip = len(code); emit(0)
+                # _compile_info(code, av, flags)
                 _compile(code, av, flags)
                 emit(OPCODES[JUMP])
                 tail.append(len(code)); emit(0)
@@ -141,51 +126,196 @@ def _compile(code, pattern, flags):
                 emit(CHCODES[CH_UNICODE[av]])
             else:
                 emit(CHCODES[av])
-        elif op is GROUP:
+        elif op is GROUPREF:
             if flags & SRE_FLAG_IGNORECASE:
                 emit(OPCODES[OP_IGNORE[op]])
             else:
                 emit(OPCODES[op])
             emit(av-1)
-        elif op is MARK:
-            emit(OPCODES[op])
-            emit(av)
         else:
             raise ValueError, ("unsupported operand type", op)
 
+def _compile_charset(charset, flags, code, fixup=None):
+    # compile charset subprogram
+    emit = code.append
+    if not fixup:
+        fixup = lambda x: x
+    for op, av in _optimize_charset(charset, fixup):
+        emit(OPCODES[op])
+        if op is NEGATE:
+            pass
+        elif op is LITERAL:
+            emit(fixup(av))
+        elif op is RANGE:
+            emit(fixup(av[0]))
+            emit(fixup(av[1]))
+        elif op is CHARSET:
+            code.extend(av)
+        elif op is CATEGORY:
+            if flags & SRE_FLAG_LOCALE:
+                emit(CHCODES[CH_LOCALE[av]])
+            elif flags & SRE_FLAG_UNICODE:
+                emit(CHCODES[CH_UNICODE[av]])
+            else:
+                emit(CHCODES[av])
+        else:
+            raise error, "internal: unsupported set operator"
+    emit(OPCODES[FAILURE])
+
+def _optimize_charset(charset, fixup):
+    # internal: optimize character set
+    out = []
+    charmap = [0]*256
+    try:
+        for op, av in charset:
+            if op is NEGATE:
+                out.append((op, av))
+            elif op is LITERAL:
+                charmap[fixup(av)] = 1
+            elif op is RANGE:
+                for i in range(fixup(av[0]), fixup(av[1])+1):
+                    charmap[i] = 1
+            elif op is CATEGORY:
+                # FIXME: could append to charmap tail
+                return charset # cannot compress
+    except IndexError:
+        # character set contains unicode characters
+        return charset
+    # compress character map
+    i = p = n = 0
+    runs = []
+    for c in charmap:
+        if c:
+            if n == 0:
+                p = i
+            n = n + 1
+        elif n:
+            runs.append((p, n))
+            n = 0
+        i = i + 1
+    if n:
+        runs.append((p, n))
+    if len(runs) <= 2:
+        # use literal/range
+        for p, n in runs:
+            if n == 1:
+                out.append((LITERAL, p))
+            else:
+                out.append((RANGE, (p, p+n-1)))
+        if len(out) < len(charset):
+            return out
+    else:
+        # use bitmap
+        data = []
+        m = 1; v = 0
+        for c in charmap:
+            if c:
+                v = v + m
+            m = m << 1
+            if m > MAXCODE:
+                data.append(v)
+                m = 1; v = 0
+        out.append((CHARSET, data))
+        return out
+    return charset
+
+def _simple(av):
+    # check if av is a "simple" operator
+    lo, hi = av[2].getwidth()
+    if lo == 0:
+        raise error, "nothing to repeat"
+    return lo == hi == 1 and av[2][0][0] != SUBPATTERN
+
 def _compile_info(code, pattern, flags):
     # internal: compile an info block.  in the current version,
-    # this contains min/max pattern width and a literal prefix,
-    # if any
+    # this contains min/max pattern width, and an optional literal
+    # prefix or a character map
     lo, hi = pattern.getwidth()
     if lo == 0:
         return # not worth it
     # look for a literal prefix
     prefix = []
+    prefix_skip = 0
+    charset = [] # not used
     if not (flags & SRE_FLAG_IGNORECASE):
+        # look for literal prefix
         for op, av in pattern.data:
             if op is LITERAL:
+                if len(prefix) == prefix_skip:
+                    prefix_skip = prefix_skip + 1
                 prefix.append(av)
+            elif op is SUBPATTERN and len(av[1]) == 1:
+                op, av = av[1][0]
+                if op is LITERAL:
+                    prefix.append(av)
+                else:
+                    break
             else:
                 break
+        # if no prefix, look for charset prefix
+        if not prefix and pattern.data:
+            op, av = pattern.data[0]
+            if op is SUBPATTERN and av[1]:
+                op, av = av[1][0]
+                if op is LITERAL:
+                    charset.append((op, av))
+                elif op is BRANCH:
+                    c = []
+                    for p in av[1]:
+                        if not p:
+                            break
+                        op, av = p[0]
+                        if op is LITERAL:
+                            c.append((op, av))
+                        else:
+                            break
+                    else:
+                        charset = c
+            elif op is BRANCH:
+                c = []
+                for p in av[1]:
+                    if not p:
+                        break
+                    op, av = p[0]
+                    if op is LITERAL:
+                        c.append((op, av))
+                    else:
+                        break
+                else:
+                    charset = c
+            elif op is IN:
+                charset = av
+##     if prefix:
+##         print "*** PREFIX", prefix, prefix_skip
+##     if charset:
+##         print "*** CHARSET", charset
     # add an info block
     emit = code.append
     emit(OPCODES[INFO])
     skip = len(code); emit(0)
     # literal flag
     mask = 0
-    if len(prefix) == len(pattern.data):
-        mask = 1
+    if prefix:
+        mask = SRE_INFO_PREFIX
+        if len(prefix) == prefix_skip == len(pattern.data):
+            mask = mask + SRE_INFO_LITERAL
+    elif charset:
+        mask = mask + SRE_INFO_CHARSET
     emit(mask)
     # pattern length
-    emit(lo)
-    if hi < 32768:
+    if lo < MAXCODE:
+        emit(lo)
+    else:
+        emit(MAXCODE)
+        prefix = prefix[:MAXCODE]
+    if hi < MAXCODE:
         emit(hi)
     else:
         emit(0)
     # add literal prefix
-    emit(len(prefix))
     if prefix:
+        emit(len(prefix)) # length
+        emit(prefix_skip) # skip
         code.extend(prefix)
         # generate overlap table
         table = [-1] + ([0]*len(prefix))
@@ -194,18 +324,18 @@ def _compile_info(code, pattern, flags):
             while table[i+1] > 0 and prefix[i] != prefix[table[i+1]-1]:
                 table[i+1] = table[table[i+1]-1]+1
         code.extend(table[1:]) # don't store first entry
+    elif charset:
+        _compile_charset(charset, 0, code)
     code[skip] = len(code) - skip
 
-def compile(p, flags=0):
-    # internal: convert pattern list to internal format
+STRING_TYPES = [type("")]
 
-    # compile, as necessary
-    if type(p) in (type(""), type(u"")):
-        import sre_parse
-        pattern = p
-        p = sre_parse.parse(p, flags)
-    else:
-        pattern = None
+try:
+    STRING_TYPES.append(type(unicode("")))
+except NameError:
+    pass
+
+def _code(p, flags):
 
     flags = p.pattern.flags | flags
     code = []
@@ -218,12 +348,34 @@ def compile(p, flags=0):
 
     code.append(OPCODES[SUCCESS])
 
+    return code
+
+def compile(p, flags=0):
+    # internal: convert pattern list to internal format
+
+    if type(p) in STRING_TYPES:
+        import sre_parse
+        pattern = p
+        p = sre_parse.parse(p, flags)
+    else:
+        pattern = None
+
+    code = _code(p, flags)
+
+    # print code
+
     # FIXME: <fl> get rid of this limitation!
     assert p.pattern.groups <= 100,\
            "sorry, but this version only supports 100 named groups"
 
+    # map in either direction
+    groupindex = p.pattern.groupdict
+    indexgroup = [None] * p.pattern.groups
+    for k, i in groupindex.items():
+        indexgroup[i] = k
+
     return _sre.compile(
-        pattern, flags,
-        array.array(WORDSIZE, code).tostring(),
-        p.pattern.groups-1, p.pattern.groupdict
+        pattern, flags, code,
+        p.pattern.groups-1,
+        groupindex, indexgroup
         )