blob: c0423750e3b9073e58c072f5ac262772484a2c4b [file] [log] [blame]
Guido van Rossumaad67612000-05-08 17:31:04 +00001#
2# Secret Labs' Regular Expression Engine
Guido van Rossumaad67612000-05-08 17:31:04 +00003#
4# convert template to internal format
5#
6# Copyright (c) 1997-2000 by Secret Labs AB. All rights reserved.
7#
Guido van Rossumaad67612000-05-08 17:31:04 +00008# Portions of this engine have been developed in cooperation with
9# CNRI. Hewlett-Packard provided funding for 1.6 integration and
10# other compatibility work.
11#
12
Guido van Rossum3e06ab12000-06-29 19:35:29 +000013import array
Guido van Rossumaad67612000-05-08 17:31:04 +000014import _sre
15
16from sre_constants import *
17
18# find an array type code that matches the engine's code size
19for WORDSIZE in "BHil":
20 if len(array.array(WORDSIZE, [0]).tostring()) == _sre.getcodesize():
21 break
22else:
23 raise RuntimeError, "cannot find a useable array type"
24
Guido van Rossumaad67612000-05-08 17:31:04 +000025def _compile(code, pattern, flags):
Guido van Rossum3e06ab12000-06-29 19:35:29 +000026 emit = code.append
Guido van Rossumaad67612000-05-08 17:31:04 +000027 for op, av in pattern:
28 if op is ANY:
Guido van Rossum3e06ab12000-06-29 19:35:29 +000029 if flags & SRE_FLAG_DOTALL:
30 emit(OPCODES[op])
Guido van Rossumaad67612000-05-08 17:31:04 +000031 else:
Guido van Rossum3e06ab12000-06-29 19:35:29 +000032 emit(OPCODES[CATEGORY])
33 emit(CHCODES[CATEGORY_NOT_LINEBREAK])
Guido van Rossumaad67612000-05-08 17:31:04 +000034 elif op in (SUCCESS, FAILURE):
Guido van Rossum3e06ab12000-06-29 19:35:29 +000035 emit(OPCODES[op])
Guido van Rossumaad67612000-05-08 17:31:04 +000036 elif op is AT:
Guido van Rossum3e06ab12000-06-29 19:35:29 +000037 emit(OPCODES[op])
38 if flags & SRE_FLAG_MULTILINE:
39 emit(ATCODES[AT_MULTILINE[av]])
40 else:
41 emit(ATCODES[av])
Guido van Rossumaad67612000-05-08 17:31:04 +000042 elif op is BRANCH:
Guido van Rossum3e06ab12000-06-29 19:35:29 +000043 emit(OPCODES[op])
Guido van Rossumaad67612000-05-08 17:31:04 +000044 tail = []
45 for av in av[1]:
Guido van Rossum3e06ab12000-06-29 19:35:29 +000046 skip = len(code); emit(0)
Guido van Rossumaad67612000-05-08 17:31:04 +000047 _compile(code, av, flags)
Guido van Rossum3e06ab12000-06-29 19:35:29 +000048 emit(OPCODES[JUMP])
49 tail.append(len(code)); emit(0)
Guido van Rossumaad67612000-05-08 17:31:04 +000050 code[skip] = len(code) - skip
Guido van Rossum3e06ab12000-06-29 19:35:29 +000051 emit(0) # end of branch
52 for tail in tail:
Guido van Rossumaad67612000-05-08 17:31:04 +000053 code[tail] = len(code) - tail
54 elif op is CALL:
Guido van Rossum3e06ab12000-06-29 19:35:29 +000055 emit(OPCODES[op])
56 skip = len(code); emit(0)
Guido van Rossumaad67612000-05-08 17:31:04 +000057 _compile(code, av, flags)
Guido van Rossum3e06ab12000-06-29 19:35:29 +000058 emit(OPCODES[SUCCESS])
Guido van Rossumaad67612000-05-08 17:31:04 +000059 code[skip] = len(code) - skip
Guido van Rossum3e06ab12000-06-29 19:35:29 +000060 elif op is CATEGORY:
61 emit(OPCODES[op])
62 if flags & SRE_FLAG_LOCALE:
63 emit(CH_LOCALE[CHCODES[av]])
64 elif flags & SRE_FLAG_UNICODE:
65 emit(CH_UNICODE[CHCODES[av]])
66 else:
67 emit(CHCODES[av])
Guido van Rossumaad67612000-05-08 17:31:04 +000068 elif op is GROUP:
Guido van Rossum3e06ab12000-06-29 19:35:29 +000069 if flags & SRE_FLAG_IGNORECASE:
70 emit(OPCODES[OP_IGNORE[op]])
Guido van Rossumaad67612000-05-08 17:31:04 +000071 else:
Guido van Rossum3e06ab12000-06-29 19:35:29 +000072 emit(OPCODES[op])
73 emit(av-1)
Guido van Rossumaad67612000-05-08 17:31:04 +000074 elif op is IN:
Guido van Rossum3e06ab12000-06-29 19:35:29 +000075 if flags & SRE_FLAG_IGNORECASE:
76 emit(OPCODES[OP_IGNORE[op]])
77 def fixup(literal, flags=flags):
78 return _sre.getlower(ord(literal), flags)
Guido van Rossumaad67612000-05-08 17:31:04 +000079 else:
Guido van Rossum3e06ab12000-06-29 19:35:29 +000080 emit(OPCODES[op])
Guido van Rossumaad67612000-05-08 17:31:04 +000081 fixup = ord
Guido van Rossum3e06ab12000-06-29 19:35:29 +000082 skip = len(code); emit(0)
Guido van Rossumaad67612000-05-08 17:31:04 +000083 for op, av in av:
Guido van Rossum3e06ab12000-06-29 19:35:29 +000084 emit(OPCODES[op])
Guido van Rossumaad67612000-05-08 17:31:04 +000085 if op is NEGATE:
86 pass
87 elif op is LITERAL:
Guido van Rossum3e06ab12000-06-29 19:35:29 +000088 emit(fixup(av))
Guido van Rossumaad67612000-05-08 17:31:04 +000089 elif op is RANGE:
Guido van Rossum3e06ab12000-06-29 19:35:29 +000090 emit(fixup(av[0]))
91 emit(fixup(av[1]))
Guido van Rossumaad67612000-05-08 17:31:04 +000092 elif op is CATEGORY:
Guido van Rossum3e06ab12000-06-29 19:35:29 +000093 if flags & SRE_FLAG_LOCALE:
94 emit(CH_LOCALE[CHCODES[av]])
95 elif flags & SRE_FLAG_UNICODE:
96 emit(CH_UNICODE[CHCODES[av]])
97 else:
98 emit(CHCODES[av])
Guido van Rossumaad67612000-05-08 17:31:04 +000099 else:
Guido van Rossum3e06ab12000-06-29 19:35:29 +0000100 raise error, "internal: unsupported set operator"
101 emit(OPCODES[FAILURE])
Guido van Rossumaad67612000-05-08 17:31:04 +0000102 code[skip] = len(code) - skip
103 elif op in (LITERAL, NOT_LITERAL):
Guido van Rossum3e06ab12000-06-29 19:35:29 +0000104 if flags & SRE_FLAG_IGNORECASE:
105 emit(OPCODES[OP_IGNORE[op]])
Guido van Rossumaad67612000-05-08 17:31:04 +0000106 else:
Guido van Rossum3e06ab12000-06-29 19:35:29 +0000107 emit(OPCODES[op])
108 emit(ord(av))
Guido van Rossumaad67612000-05-08 17:31:04 +0000109 elif op is MARK:
Guido van Rossum3e06ab12000-06-29 19:35:29 +0000110 emit(OPCODES[op])
111 emit(av)
Guido van Rossumaad67612000-05-08 17:31:04 +0000112 elif op in (REPEAT, MIN_REPEAT, MAX_REPEAT):
Guido van Rossum3e06ab12000-06-29 19:35:29 +0000113 if flags & SRE_FLAG_TEMPLATE:
114 emit(OPCODES[REPEAT])
115 skip = len(code); emit(0)
116 emit(av[0])
117 emit(av[1])
Guido van Rossumaad67612000-05-08 17:31:04 +0000118 _compile(code, av[2], flags)
Guido van Rossum3e06ab12000-06-29 19:35:29 +0000119 emit(OPCODES[SUCCESS])
Guido van Rossumaad67612000-05-08 17:31:04 +0000120 code[skip] = len(code) - skip
121 else:
Guido van Rossum3e06ab12000-06-29 19:35:29 +0000122 lo, hi = av[2].getwidth()
123 if lo == 0:
124 raise error, "nothing to repeat"
125 if 0 and lo == hi == 1 and op is MAX_REPEAT:
126 # FIXME: <fl> need a better way to figure out when
127 # it's safe to use this one (in the parser, probably)
128 emit(OPCODES[MAX_REPEAT_ONE])
129 skip = len(code); emit(0)
130 emit(av[0])
131 emit(av[1])
132 _compile(code, av[2], flags)
133 emit(OPCODES[SUCCESS])
134 code[skip] = len(code) - skip
Guido van Rossumaad67612000-05-08 17:31:04 +0000135 else:
Guido van Rossum3e06ab12000-06-29 19:35:29 +0000136 emit(OPCODES[op])
137 skip = len(code); emit(0)
138 emit(av[0])
139 emit(av[1])
140 _compile(code, av[2], flags)
141 emit(OPCODES[SUCCESS])
142 code[skip] = len(code) - skip
Guido van Rossumaad67612000-05-08 17:31:04 +0000143 elif op is SUBPATTERN:
Guido van Rossum3e06ab12000-06-29 19:35:29 +0000144 group = av[0]
145 if group:
146 emit(OPCODES[MARK])
147 emit((group-1)*2)
Guido van Rossumaad67612000-05-08 17:31:04 +0000148 _compile(code, av[1], flags)
Guido van Rossum3e06ab12000-06-29 19:35:29 +0000149 if group:
150 emit(OPCODES[MARK])
151 emit((group-1)*2+1)
Guido van Rossumaad67612000-05-08 17:31:04 +0000152 else:
153 raise ValueError, ("unsupported operand type", op)
154
Guido van Rossum3e06ab12000-06-29 19:35:29 +0000155def compile(p, flags=0):
156 # internal: convert pattern list to internal format
Guido van Rossumaad67612000-05-08 17:31:04 +0000157 if type(p) in (type(""), type(u"")):
158 import sre_parse
159 pattern = p
160 p = sre_parse.parse(p)
161 else:
162 pattern = None
Guido van Rossum3e06ab12000-06-29 19:35:29 +0000163 flags = p.pattern.flags | flags
164 code = []
165 _compile(code, p.data, flags)
166 code.append(OPCODES[SUCCESS])
167 # FIXME: <fl> get rid of this limitation
168 assert p.pattern.groups <= 100,\
169 "sorry, but this version only supports 100 named groups"
170 return _sre.compile(
171 pattern, flags,
172 array.array(WORDSIZE, code).tostring(),
173 p.pattern.groups-1, p.pattern.groupdict
174 )