blob: 2d1cbb166563222eac4758e6f5cd224fe52fc121 [file] [log] [blame]
Guido van Rossum7627c0d2000-03-31 14:58:54 +00001#
2# Secret Labs' Regular Expression Engine
Guido van Rossum7627c0d2000-03-31 14:58:54 +00003#
4# convert template to internal format
5#
6# Copyright (c) 1997-2000 by Secret Labs AB. All rights reserved.
7#
Fredrik Lundh29c4ba92000-08-01 18:20:07 +00008# See the sre.py file for information on usage and redistribution.
Guido van Rossum7627c0d2000-03-31 14:58:54 +00009#
10
Guido van Rossum7627c0d2000-03-31 14:58:54 +000011import _sre
12
13from sre_constants import *
14
Fredrik Lundh3562f112000-07-02 12:00:07 +000015MAXCODE = 65535
16
Fredrik Lundh28552902000-07-05 21:14:16 +000017def _charset(charset, fixup=None):
Fredrik Lundh3562f112000-07-02 12:00:07 +000018 # internal: optimize character set
Fredrik Lundh28552902000-07-05 21:14:16 +000019 if not fixup:
20 fixup = lambda x: x
Fredrik Lundh3562f112000-07-02 12:00:07 +000021 out = []
22 charmap = [0]*256
23 try:
24 for op, av in charset:
25 if op is NEGATE:
26 out.append((op, av))
27 elif op is LITERAL:
28 charmap[fixup(av)] = 1
29 elif op is RANGE:
30 for i in range(fixup(av[0]), fixup(av[1])+1):
31 charmap[i] = 1
32 elif op is CATEGORY:
33 # FIXME: could append to charmap tail
34 return charset # cannot compress
35 except IndexError:
36 # unicode
37 return charset
38 # compress character map
39 i = p = n = 0
40 runs = []
41 for c in charmap:
42 if c:
43 if n == 0:
44 p = i
45 n = n + 1
46 elif n:
47 runs.append((p, n))
48 n = 0
49 i = i + 1
50 if n:
51 runs.append((p, n))
52 if len(runs) <= 2:
53 # use literal/range
54 for p, n in runs:
55 if n == 1:
56 out.append((LITERAL, p))
57 else:
58 out.append((RANGE, (p, p+n-1)))
59 if len(out) < len(charset):
60 return out
61 else:
62 # use bitmap
63 data = []
64 m = 1; v = 0
65 for c in charmap:
66 if c:
67 v = v + m
68 m = m << 1
69 if m > MAXCODE:
70 data.append(v)
71 m = 1; v = 0
72 out.append((CHARSET, data))
73 return out
74 return charset
75
Fredrik Lundh436c3d582000-06-29 08:58:44 +000076def _compile(code, pattern, flags):
Fredrik Lundh29c08be2000-06-29 23:33:12 +000077 # internal: compile a (sub)pattern
Fredrik Lundhbe2211e2000-06-29 16:57:40 +000078 emit = code.append
Guido van Rossum7627c0d2000-03-31 14:58:54 +000079 for op, av in pattern:
Fredrik Lundh43b3b492000-06-30 10:41:31 +000080 if op in (LITERAL, NOT_LITERAL):
Fredrik Lundh90a07912000-06-30 07:50:59 +000081 if flags & SRE_FLAG_IGNORECASE:
82 emit(OPCODES[OP_IGNORE[op]])
83 else:
84 emit(OPCODES[op])
Fredrik Lundh0640e112000-06-30 13:55:15 +000085 emit(av)
Fredrik Lundh90a07912000-06-30 07:50:59 +000086 elif op is IN:
87 if flags & SRE_FLAG_IGNORECASE:
88 emit(OPCODES[OP_IGNORE[op]])
89 def fixup(literal, flags=flags):
Fredrik Lundh0640e112000-06-30 13:55:15 +000090 return _sre.getlower(literal, flags)
Fredrik Lundh90a07912000-06-30 07:50:59 +000091 else:
92 emit(OPCODES[op])
Fredrik Lundh4ccea942000-06-30 18:39:20 +000093 fixup = lambda x: x
Fredrik Lundh90a07912000-06-30 07:50:59 +000094 skip = len(code); emit(0)
Fredrik Lundh3562f112000-07-02 12:00:07 +000095 for op, av in _charset(av, fixup):
Fredrik Lundh90a07912000-06-30 07:50:59 +000096 emit(OPCODES[op])
97 if op is NEGATE:
98 pass
99 elif op is LITERAL:
100 emit(fixup(av))
101 elif op is RANGE:
102 emit(fixup(av[0]))
103 emit(fixup(av[1]))
Fredrik Lundh3562f112000-07-02 12:00:07 +0000104 elif op is CHARSET:
105 code.extend(av)
Fredrik Lundh90a07912000-06-30 07:50:59 +0000106 elif op is CATEGORY:
107 if flags & SRE_FLAG_LOCALE:
108 emit(CHCODES[CH_LOCALE[av]])
109 elif flags & SRE_FLAG_UNICODE:
110 emit(CHCODES[CH_UNICODE[av]])
111 else:
112 emit(CHCODES[av])
113 else:
114 raise error, "internal: unsupported set operator"
115 emit(OPCODES[FAILURE])
116 code[skip] = len(code) - skip
Fredrik Lundh43b3b492000-06-30 10:41:31 +0000117 elif op is ANY:
118 if flags & SRE_FLAG_DOTALL:
Fredrik Lundh90a07912000-06-30 07:50:59 +0000119 emit(OPCODES[op])
Fredrik Lundh43b3b492000-06-30 10:41:31 +0000120 else:
121 emit(OPCODES[CATEGORY])
122 emit(CHCODES[CATEGORY_NOT_LINEBREAK])
Fredrik Lundh90a07912000-06-30 07:50:59 +0000123 elif op in (REPEAT, MIN_REPEAT, MAX_REPEAT):
124 if flags & SRE_FLAG_TEMPLATE:
Fredrik Lundh29c4ba92000-08-01 18:20:07 +0000125 raise error, "internal: unsupported template operator"
Fredrik Lundh90a07912000-06-30 07:50:59 +0000126 emit(OPCODES[REPEAT])
127 skip = len(code); emit(0)
128 emit(av[0])
129 emit(av[1])
130 _compile(code, av[2], flags)
131 emit(OPCODES[SUCCESS])
132 code[skip] = len(code) - skip
133 else:
134 lo, hi = av[2].getwidth()
135 if lo == 0:
136 raise error, "nothing to repeat"
137 if 0 and lo == hi == 1 and op is MAX_REPEAT:
Fredrik Lundh29c4ba92000-08-01 18:20:07 +0000138 # FIXME: <fl> fast and wrong (but we'll fix that)
139 emit(OPCODES[REPEAT_ONE])
Fredrik Lundh90a07912000-06-30 07:50:59 +0000140 skip = len(code); emit(0)
141 emit(av[0])
142 emit(av[1])
143 _compile(code, av[2], flags)
144 emit(OPCODES[SUCCESS])
145 code[skip] = len(code) - skip
146 else:
Fredrik Lundh29c4ba92000-08-01 18:20:07 +0000147 emit(OPCODES[REPEAT])
Fredrik Lundh90a07912000-06-30 07:50:59 +0000148 skip = len(code); emit(0)
149 emit(av[0])
150 emit(av[1])
151 _compile(code, av[2], flags)
Fredrik Lundh90a07912000-06-30 07:50:59 +0000152 code[skip] = len(code) - skip
Fredrik Lundh29c4ba92000-08-01 18:20:07 +0000153 if op == MAX_REPEAT:
154 emit(OPCODES[MAX_UNTIL])
155 else:
156 emit(OPCODES[MIN_UNTIL])
Fredrik Lundh90a07912000-06-30 07:50:59 +0000157 elif op is SUBPATTERN:
Fredrik Lundh29c4ba92000-08-01 18:20:07 +0000158 if av[0]:
Fredrik Lundh90a07912000-06-30 07:50:59 +0000159 emit(OPCODES[MARK])
Fredrik Lundh29c4ba92000-08-01 18:20:07 +0000160 emit((av[0]-1)*2)
Fredrik Lundh90a07912000-06-30 07:50:59 +0000161 _compile(code, av[1], flags)
Fredrik Lundh29c4ba92000-08-01 18:20:07 +0000162 if av[0]:
Fredrik Lundh90a07912000-06-30 07:50:59 +0000163 emit(OPCODES[MARK])
Fredrik Lundh29c4ba92000-08-01 18:20:07 +0000164 emit((av[0]-1)*2+1)
Fredrik Lundh43b3b492000-06-30 10:41:31 +0000165 elif op in (SUCCESS, FAILURE):
166 emit(OPCODES[op])
Fredrik Lundh6f013982000-07-03 18:44:21 +0000167 elif op in (ASSERT, ASSERT_NOT):
168 emit(OPCODES[op])
169 skip = len(code); emit(0)
170 if av[0] >= 0:
171 emit(0) # look ahead
172 else:
173 lo, hi = av[1].getwidth()
174 if lo != hi:
175 raise error, "look-behind requires fixed-width pattern"
176 emit(lo) # look behind
177 _compile(code, av[1], flags)
178 emit(OPCODES[SUCCESS])
179 code[skip] = len(code) - skip
180 elif op is CALL:
Fredrik Lundh43b3b492000-06-30 10:41:31 +0000181 emit(OPCODES[op])
182 skip = len(code); emit(0)
183 _compile(code, av, flags)
184 emit(OPCODES[SUCCESS])
185 code[skip] = len(code) - skip
186 elif op is AT:
187 emit(OPCODES[op])
188 if flags & SRE_FLAG_MULTILINE:
Fredrik Lundh55a4f4a2000-06-30 22:37:31 +0000189 emit(ATCODES[AT_MULTILINE.get(av, av)])
Fredrik Lundh43b3b492000-06-30 10:41:31 +0000190 else:
191 emit(ATCODES[av])
192 elif op is BRANCH:
Fredrik Lundh29c4ba92000-08-01 18:20:07 +0000193 emit(OPCODES[op])
Fredrik Lundh43b3b492000-06-30 10:41:31 +0000194 tail = []
195 for av in av[1]:
196 skip = len(code); emit(0)
197 _compile(code, av, flags)
198 emit(OPCODES[JUMP])
199 tail.append(len(code)); emit(0)
200 code[skip] = len(code) - skip
201 emit(0) # end of branch
202 for tail in tail:
203 code[tail] = len(code) - tail
204 elif op is CATEGORY:
205 emit(OPCODES[op])
206 if flags & SRE_FLAG_LOCALE:
207 emit(CHCODES[CH_LOCALE[av]])
208 elif flags & SRE_FLAG_UNICODE:
209 emit(CHCODES[CH_UNICODE[av]])
210 else:
211 emit(CHCODES[av])
Fredrik Lundh72b82ba2000-07-03 21:31:48 +0000212 elif op is GROUPREF:
Fredrik Lundh43b3b492000-06-30 10:41:31 +0000213 if flags & SRE_FLAG_IGNORECASE:
214 emit(OPCODES[OP_IGNORE[op]])
215 else:
216 emit(OPCODES[op])
217 emit(av-1)
Fredrik Lundh90a07912000-06-30 07:50:59 +0000218 else:
219 raise ValueError, ("unsupported operand type", op)
Guido van Rossum7627c0d2000-03-31 14:58:54 +0000220
Fredrik Lundh29c08be2000-06-29 23:33:12 +0000221def _compile_info(code, pattern, flags):
222 # internal: compile an info block. in the current version,
Fredrik Lundh3562f112000-07-02 12:00:07 +0000223 # this contains min/max pattern width, and an optional literal
224 # prefix or a character map
Fredrik Lundh29c08be2000-06-29 23:33:12 +0000225 lo, hi = pattern.getwidth()
226 if lo == 0:
Fredrik Lundh90a07912000-06-30 07:50:59 +0000227 return # not worth it
Fredrik Lundh29c08be2000-06-29 23:33:12 +0000228 # look for a literal prefix
229 prefix = []
Fredrik Lundh3562f112000-07-02 12:00:07 +0000230 charset = [] # not used
Fredrik Lundh29c08be2000-06-29 23:33:12 +0000231 if not (flags & SRE_FLAG_IGNORECASE):
Fredrik Lundh90a07912000-06-30 07:50:59 +0000232 for op, av in pattern.data:
233 if op is LITERAL:
Fredrik Lundh0640e112000-06-30 13:55:15 +0000234 prefix.append(av)
Fredrik Lundh90a07912000-06-30 07:50:59 +0000235 else:
236 break
Fredrik Lundh29c08be2000-06-29 23:33:12 +0000237 # add an info block
238 emit = code.append
239 emit(OPCODES[INFO])
240 skip = len(code); emit(0)
241 # literal flag
242 mask = 0
Fredrik Lundh3562f112000-07-02 12:00:07 +0000243 if prefix:
244 mask = SRE_INFO_PREFIX
245 if len(prefix) == len(pattern.data):
246 mask = mask + SRE_INFO_LITERAL
247 elif charset:
248 mask = mask + SRE_INFO_CHARSET
Fredrik Lundh29c08be2000-06-29 23:33:12 +0000249 emit(mask)
250 # pattern length
Fredrik Lundh3562f112000-07-02 12:00:07 +0000251 if lo < MAXCODE:
252 emit(lo)
253 else:
254 emit(MAXCODE)
255 prefix = prefix[:MAXCODE]
256 if hi < MAXCODE:
Fredrik Lundh90a07912000-06-30 07:50:59 +0000257 emit(hi)
Fredrik Lundh29c08be2000-06-29 23:33:12 +0000258 else:
Fredrik Lundh90a07912000-06-30 07:50:59 +0000259 emit(0)
Fredrik Lundh29c08be2000-06-29 23:33:12 +0000260 # add literal prefix
Fredrik Lundh29c08be2000-06-29 23:33:12 +0000261 if prefix:
Fredrik Lundh3562f112000-07-02 12:00:07 +0000262 emit(len(prefix))
263 if prefix:
264 code.extend(prefix)
265 # generate overlap table
266 table = [-1] + ([0]*len(prefix))
267 for i in range(len(prefix)):
268 table[i+1] = table[i]+1
269 while table[i+1] > 0 and prefix[i] != prefix[table[i+1]-1]:
270 table[i+1] = table[table[i+1]-1]+1
271 code.extend(table[1:]) # don't store first entry
272 elif charset:
273 for char in charset:
274 emit(OPCODES[LITERAL])
275 emit(char)
276 emit(OPCODES[FAILURE])
Fredrik Lundh29c08be2000-06-29 23:33:12 +0000277 code[skip] = len(code) - skip
278
Fredrik Lundh8a3ebf82000-07-23 21:46:17 +0000279STRING_TYPES = [type("")]
280
281try:
282 STRING_TYPES.append(type(unicode("")))
283except NameError:
284 pass
285
Fredrik Lundh29c4ba92000-08-01 18:20:07 +0000286def _compile1(p, flags):
Fredrik Lundh29c08be2000-06-29 23:33:12 +0000287
Jeremy Hyltonb1aa1952000-06-01 17:39:12 +0000288 flags = p.pattern.flags | flags
Fredrik Lundhbe2211e2000-06-29 16:57:40 +0000289 code = []
Fredrik Lundh29c08be2000-06-29 23:33:12 +0000290
291 # compile info block
292 _compile_info(code, p, flags)
293
294 # compile the pattern
Jeremy Hyltonb1aa1952000-06-01 17:39:12 +0000295 _compile(code, p.data, flags)
Fredrik Lundh29c08be2000-06-29 23:33:12 +0000296
Jeremy Hyltonb1aa1952000-06-01 17:39:12 +0000297 code.append(OPCODES[SUCCESS])
Fredrik Lundh29c08be2000-06-29 23:33:12 +0000298
Fredrik Lundh29c4ba92000-08-01 18:20:07 +0000299 return code
300
301def compile(p, flags=0):
302 # internal: convert pattern list to internal format
303
304 if type(p) in STRING_TYPES:
305 import sre_parse
306 pattern = p
307 p = sre_parse.parse(p, flags)
308 else:
309 pattern = None
310
311 code = _compile1(p, flags)
312
Fredrik Lundh8a3ebf82000-07-23 21:46:17 +0000313 # print code
314
Fredrik Lundh29c08be2000-06-29 23:33:12 +0000315 # FIXME: <fl> get rid of this limitation!
Fredrik Lundhbe2211e2000-06-29 16:57:40 +0000316 assert p.pattern.groups <= 100,\
Fredrik Lundh90a07912000-06-30 07:50:59 +0000317 "sorry, but this version only supports 100 named groups"
Fredrik Lundh29c08be2000-06-29 23:33:12 +0000318
Fredrik Lundhc2301732000-07-02 22:25:39 +0000319 # map in either direction
320 groupindex = p.pattern.groupdict
321 indexgroup = [None] * p.pattern.groups
322 for k, i in groupindex.items():
323 indexgroup[i] = k
324
Jeremy Hyltonb1aa1952000-06-01 17:39:12 +0000325 return _sre.compile(
Fredrik Lundh6f013982000-07-03 18:44:21 +0000326 pattern, flags, code,
327 p.pattern.groups-1,
328 groupindex, indexgroup
Fredrik Lundh90a07912000-06-30 07:50:59 +0000329 )