OILS / opy / _regtest / src / sre_compile.py View on Github | oilshell.org

596 lines, 475 significant
1# -*- coding: utf-8 -*-
2#
3# Secret Labs' Regular Expression Engine
4#
5# convert template to internal format
6#
7# Copyright (c) 1997-2001 by Secret Labs AB. All rights reserved.
8#
9# See the sre.py file for information on usage and redistribution.
10#
11
12"""Internal support module for sre"""
13
14import _sre, sys
15import sre_parse
16from sre_constants import *
17
18assert _sre.MAGIC == MAGIC, "SRE module mismatch"
19
20if _sre.CODESIZE == 2:
21 MAXCODE = 65535
22else:
23 MAXCODE = 0xFFFFFFFFL
24
25_LITERAL_CODES = set([LITERAL, NOT_LITERAL])
26_REPEATING_CODES = set([REPEAT, MIN_REPEAT, MAX_REPEAT])
27_SUCCESS_CODES = set([SUCCESS, FAILURE])
28_ASSERT_CODES = set([ASSERT, ASSERT_NOT])
29
30# Sets of lowercase characters which have the same uppercase.
31_equivalences = (
32 # LATIN SMALL LETTER I, LATIN SMALL LETTER DOTLESS I
33 (0x69, 0x131), # iı
34 # LATIN SMALL LETTER S, LATIN SMALL LETTER LONG S
35 (0x73, 0x17f), # sſ
36 # MICRO SIGN, GREEK SMALL LETTER MU
37 (0xb5, 0x3bc), # µμ
38 # COMBINING GREEK YPOGEGRAMMENI, GREEK SMALL LETTER IOTA, GREEK PROSGEGRAMMENI
39 (0x345, 0x3b9, 0x1fbe), # \u0345ιι
40 # GREEK SMALL LETTER BETA, GREEK BETA SYMBOL
41 (0x3b2, 0x3d0), # βϐ
42 # GREEK SMALL LETTER EPSILON, GREEK LUNATE EPSILON SYMBOL
43 (0x3b5, 0x3f5), # εϵ
44 # GREEK SMALL LETTER THETA, GREEK THETA SYMBOL
45 (0x3b8, 0x3d1), # θϑ
46 # GREEK SMALL LETTER KAPPA, GREEK KAPPA SYMBOL
47 (0x3ba, 0x3f0), # κϰ
48 # GREEK SMALL LETTER PI, GREEK PI SYMBOL
49 (0x3c0, 0x3d6), # πϖ
50 # GREEK SMALL LETTER RHO, GREEK RHO SYMBOL
51 (0x3c1, 0x3f1), # ρϱ
52 # GREEK SMALL LETTER FINAL SIGMA, GREEK SMALL LETTER SIGMA
53 (0x3c2, 0x3c3), # ςσ
54 # GREEK SMALL LETTER PHI, GREEK PHI SYMBOL
55 (0x3c6, 0x3d5), # φϕ
56 # LATIN SMALL LETTER S WITH DOT ABOVE, LATIN SMALL LETTER LONG S WITH DOT ABOVE
57 (0x1e61, 0x1e9b), # ṡẛ
58)
59
60# Maps the lowercase code to lowercase codes which have the same uppercase.
61_ignorecase_fixes = {i: tuple(j for j in t if i != j)
62 for t in _equivalences for i in t}
63
64def _compile(code, pattern, flags):
65 # internal: compile a (sub)pattern
66 emit = code.append
67 _len = len
68 LITERAL_CODES = _LITERAL_CODES
69 REPEATING_CODES = _REPEATING_CODES
70 SUCCESS_CODES = _SUCCESS_CODES
71 ASSERT_CODES = _ASSERT_CODES
72 if (flags & SRE_FLAG_IGNORECASE and
73 not (flags & SRE_FLAG_LOCALE) and
74 flags & SRE_FLAG_UNICODE):
75 fixes = _ignorecase_fixes
76 else:
77 fixes = None
78 for op, av in pattern:
79 if op in LITERAL_CODES:
80 if flags & SRE_FLAG_IGNORECASE:
81 lo = _sre.getlower(av, flags)
82 if fixes and lo in fixes:
83 emit(OPCODES[IN_IGNORE])
84 skip = _len(code); emit(0)
85 if op is NOT_LITERAL:
86 emit(OPCODES[NEGATE])
87 for k in (lo,) + fixes[lo]:
88 emit(OPCODES[LITERAL])
89 emit(k)
90 emit(OPCODES[FAILURE])
91 code[skip] = _len(code) - skip
92 else:
93 emit(OPCODES[OP_IGNORE[op]])
94 emit(lo)
95 else:
96 emit(OPCODES[op])
97 emit(av)
98 elif op is IN:
99 if flags & SRE_FLAG_IGNORECASE:
100 emit(OPCODES[OP_IGNORE[op]])
101 def fixup(literal, flags=flags):
102 return _sre.getlower(literal, flags)
103 else:
104 emit(OPCODES[op])
105 fixup = None
106 skip = _len(code); emit(0)
107 _compile_charset(av, flags, code, fixup, fixes)
108 code[skip] = _len(code) - skip
109 elif op is ANY:
110 if flags & SRE_FLAG_DOTALL:
111 emit(OPCODES[ANY_ALL])
112 else:
113 emit(OPCODES[ANY])
114 elif op in REPEATING_CODES:
115 if flags & SRE_FLAG_TEMPLATE:
116 raise error, "internal: unsupported template operator"
117 emit(OPCODES[REPEAT])
118 skip = _len(code); emit(0)
119 emit(av[0])
120 emit(av[1])
121 _compile(code, av[2], flags)
122 emit(OPCODES[SUCCESS])
123 code[skip] = _len(code) - skip
124 elif _simple(av) and op is not REPEAT:
125 if op is MAX_REPEAT:
126 emit(OPCODES[REPEAT_ONE])
127 else:
128 emit(OPCODES[MIN_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
135 else:
136 emit(OPCODES[REPEAT])
137 skip = _len(code); emit(0)
138 emit(av[0])
139 emit(av[1])
140 _compile(code, av[2], flags)
141 code[skip] = _len(code) - skip
142 if op is MAX_REPEAT:
143 emit(OPCODES[MAX_UNTIL])
144 else:
145 emit(OPCODES[MIN_UNTIL])
146 elif op is SUBPATTERN:
147 if av[0]:
148 emit(OPCODES[MARK])
149 emit((av[0]-1)*2)
150 # _compile_info(code, av[1], flags)
151 _compile(code, av[1], flags)
152 if av[0]:
153 emit(OPCODES[MARK])
154 emit((av[0]-1)*2+1)
155 elif op in SUCCESS_CODES:
156 emit(OPCODES[op])
157 elif op in ASSERT_CODES:
158 emit(OPCODES[op])
159 skip = _len(code); emit(0)
160 if av[0] >= 0:
161 emit(0) # look ahead
162 else:
163 lo, hi = av[1].getwidth()
164 if lo != hi:
165 raise error, "look-behind requires fixed-width pattern"
166 emit(lo) # look behind
167 _compile(code, av[1], flags)
168 emit(OPCODES[SUCCESS])
169 code[skip] = _len(code) - skip
170 elif op is CALL:
171 emit(OPCODES[op])
172 skip = _len(code); emit(0)
173 _compile(code, av, flags)
174 emit(OPCODES[SUCCESS])
175 code[skip] = _len(code) - skip
176 elif op is AT:
177 emit(OPCODES[op])
178 if flags & SRE_FLAG_MULTILINE:
179 av = AT_MULTILINE.get(av, av)
180 if flags & SRE_FLAG_LOCALE:
181 av = AT_LOCALE.get(av, av)
182 elif flags & SRE_FLAG_UNICODE:
183 av = AT_UNICODE.get(av, av)
184 emit(ATCODES[av])
185 elif op is BRANCH:
186 emit(OPCODES[op])
187 tail = []
188 tailappend = tail.append
189 for av in av[1]:
190 skip = _len(code); emit(0)
191 # _compile_info(code, av, flags)
192 _compile(code, av, flags)
193 emit(OPCODES[JUMP])
194 tailappend(_len(code)); emit(0)
195 code[skip] = _len(code) - skip
196 emit(0) # end of branch
197 for tail in tail:
198 code[tail] = _len(code) - tail
199 elif op is CATEGORY:
200 emit(OPCODES[op])
201 if flags & SRE_FLAG_LOCALE:
202 av = CH_LOCALE[av]
203 elif flags & SRE_FLAG_UNICODE:
204 av = CH_UNICODE[av]
205 emit(CHCODES[av])
206 elif op is GROUPREF:
207 if flags & SRE_FLAG_IGNORECASE:
208 emit(OPCODES[OP_IGNORE[op]])
209 else:
210 emit(OPCODES[op])
211 emit(av-1)
212 elif op is GROUPREF_EXISTS:
213 emit(OPCODES[op])
214 emit(av[0]-1)
215 skipyes = _len(code); emit(0)
216 _compile(code, av[1], flags)
217 if av[2]:
218 emit(OPCODES[JUMP])
219 skipno = _len(code); emit(0)
220 code[skipyes] = _len(code) - skipyes + 1
221 _compile(code, av[2], flags)
222 code[skipno] = _len(code) - skipno
223 else:
224 code[skipyes] = _len(code) - skipyes + 1
225 else:
226 raise ValueError, ("unsupported operand type", op)
227
228def _compile_charset(charset, flags, code, fixup=None, fixes=None):
229 # compile charset subprogram
230 emit = code.append
231 for op, av in _optimize_charset(charset, fixup, fixes,
232 flags & SRE_FLAG_UNICODE):
233 emit(OPCODES[op])
234 if op is NEGATE:
235 pass
236 elif op is LITERAL:
237 emit(av)
238 elif op is RANGE:
239 emit(av[0])
240 emit(av[1])
241 elif op is CHARSET:
242 code.extend(av)
243 elif op is BIGCHARSET:
244 code.extend(av)
245 elif op is CATEGORY:
246 if flags & SRE_FLAG_LOCALE:
247 emit(CHCODES[CH_LOCALE[av]])
248 elif flags & SRE_FLAG_UNICODE:
249 emit(CHCODES[CH_UNICODE[av]])
250 else:
251 emit(CHCODES[av])
252 else:
253 raise error, "internal: unsupported set operator"
254 emit(OPCODES[FAILURE])
255
256def _optimize_charset(charset, fixup, fixes, isunicode):
257 # internal: optimize character set
258 out = []
259 tail = []
260 charmap = bytearray(256)
261 for op, av in charset:
262 while True:
263 try:
264 if op is LITERAL:
265 if fixup:
266 i = fixup(av)
267 charmap[i] = 1
268 if fixes and i in fixes:
269 for k in fixes[i]:
270 charmap[k] = 1
271 else:
272 charmap[av] = 1
273 elif op is RANGE:
274 r = range(av[0], av[1]+1)
275 if fixup:
276 r = map(fixup, r)
277 if fixup and fixes:
278 for i in r:
279 charmap[i] = 1
280 if i in fixes:
281 for k in fixes[i]:
282 charmap[k] = 1
283 else:
284 for i in r:
285 charmap[i] = 1
286 elif op is NEGATE:
287 out.append((op, av))
288 else:
289 tail.append((op, av))
290 except IndexError:
291 if len(charmap) == 256:
292 # character set contains non-UCS1 character codes
293 charmap += b'\0' * 0xff00
294 continue
295 # character set contains non-BMP character codes
296 if fixup and isunicode and op is RANGE:
297 lo, hi = av
298 ranges = [av]
299 # There are only two ranges of cased astral characters:
300 # 10400-1044F (Deseret) and 118A0-118DF (Warang Citi).
301 _fixup_range(max(0x10000, lo), min(0x11fff, hi),
302 ranges, fixup)
303 for lo, hi in ranges:
304 if lo == hi:
305 tail.append((LITERAL, hi))
306 else:
307 tail.append((RANGE, (lo, hi)))
308 else:
309 tail.append((op, av))
310 break
311
312 # compress character map
313 runs = []
314 q = 0
315 while True:
316 p = charmap.find(b'\1', q)
317 if p < 0:
318 break
319 if len(runs) >= 2:
320 runs = None
321 break
322 q = charmap.find(b'\0', p)
323 if q < 0:
324 runs.append((p, len(charmap)))
325 break
326 runs.append((p, q))
327 if runs is not None:
328 # use literal/range
329 for p, q in runs:
330 if q - p == 1:
331 out.append((LITERAL, p))
332 else:
333 out.append((RANGE, (p, q - 1)))
334 out += tail
335 # if the case was changed or new representation is more compact
336 if fixup or len(out) < len(charset):
337 return out
338 # else original character set is good enough
339 return charset
340
341 # use bitmap
342 if len(charmap) == 256:
343 data = _mk_bitmap(charmap)
344 out.append((CHARSET, data))
345 out += tail
346 return out
347
348 # To represent a big charset, first a bitmap of all characters in the
349 # set is constructed. Then, this bitmap is sliced into chunks of 256
350 # characters, duplicate chunks are eliminated, and each chunk is
351 # given a number. In the compiled expression, the charset is
352 # represented by a 32-bit word sequence, consisting of one word for
353 # the number of different chunks, a sequence of 256 bytes (64 words)
354 # of chunk numbers indexed by their original chunk position, and a
355 # sequence of 256-bit chunks (8 words each).
356
357 # Compression is normally good: in a typical charset, large ranges of
358 # Unicode will be either completely excluded (e.g. if only cyrillic
359 # letters are to be matched), or completely included (e.g. if large
360 # subranges of Kanji match). These ranges will be represented by
361 # chunks of all one-bits or all zero-bits.
362
363 # Matching can be also done efficiently: the more significant byte of
364 # the Unicode character is an index into the chunk number, and the
365 # less significant byte is a bit index in the chunk (just like the
366 # CHARSET matching).
367
368 # In UCS-4 mode, the BIGCHARSET opcode still supports only subsets
369 # of the basic multilingual plane; an efficient representation
370 # for all of Unicode has not yet been developed.
371
372 charmap = bytes(charmap) # should be hashable
373 comps = {}
374 mapping = bytearray(256)
375 block = 0
376 data = bytearray()
377 for i in range(0, 65536, 256):
378 chunk = charmap[i: i + 256]
379 if chunk in comps:
380 mapping[i // 256] = comps[chunk]
381 else:
382 mapping[i // 256] = comps[chunk] = block
383 block += 1
384 data += chunk
385 data = _mk_bitmap(data)
386 data[0:0] = [block] + _bytes_to_codes(mapping)
387 out.append((BIGCHARSET, data))
388 out += tail
389 return out
390
391def _fixup_range(lo, hi, ranges, fixup):
392 for i in map(fixup, range(lo, hi+1)):
393 for k, (lo, hi) in enumerate(ranges):
394 if i < lo:
395 if l == lo - 1:
396 ranges[k] = (i, hi)
397 else:
398 ranges.insert(k, (i, i))
399 break
400 elif i > hi:
401 if i == hi + 1:
402 ranges[k] = (lo, i)
403 break
404 else:
405 break
406 else:
407 ranges.append((i, i))
408
409_CODEBITS = _sre.CODESIZE * 8
410_BITS_TRANS = b'0' + b'1' * 255
411def _mk_bitmap(bits, _CODEBITS=_CODEBITS, _int=int):
412 s = bytes(bits).translate(_BITS_TRANS)[::-1]
413 return [_int(s[i - _CODEBITS: i], 2)
414 for i in range(len(s), 0, -_CODEBITS)]
415
416def _bytes_to_codes(b):
417 # Convert block indices to word array
418 import array
419 if _sre.CODESIZE == 2:
420 code = 'H'
421 else:
422 code = 'I'
423 a = array.array(code, bytes(b))
424 assert a.itemsize == _sre.CODESIZE
425 assert len(a) * a.itemsize == len(b)
426 return a.tolist()
427
428def _simple(av):
429 # check if av is a "simple" operator
430 lo, hi = av[2].getwidth()
431 return lo == hi == 1 and av[2][0][0] != SUBPATTERN
432
433def _compile_info(code, pattern, flags):
434 # internal: compile an info block. in the current version,
435 # this contains min/max pattern width, and an optional literal
436 # prefix or a character map
437 lo, hi = pattern.getwidth()
438 if lo == 0:
439 return # not worth it
440 # look for a literal prefix
441 prefix = []
442 prefixappend = prefix.append
443 prefix_skip = 0
444 charset = [] # not used
445 charsetappend = charset.append
446 if not (flags & SRE_FLAG_IGNORECASE):
447 # look for literal prefix
448 for op, av in pattern.data:
449 if op is LITERAL:
450 if len(prefix) == prefix_skip:
451 prefix_skip = prefix_skip + 1
452 prefixappend(av)
453 elif op is SUBPATTERN and len(av[1]) == 1:
454 op, av = av[1][0]
455 if op is LITERAL:
456 prefixappend(av)
457 else:
458 break
459 else:
460 break
461 # if no prefix, look for charset prefix
462 if not prefix and pattern.data:
463 op, av = pattern.data[0]
464 if op is SUBPATTERN and av[1]:
465 op, av = av[1][0]
466 if op is LITERAL:
467 charsetappend((op, av))
468 elif op is BRANCH:
469 c = []
470 cappend = c.append
471 for p in av[1]:
472 if not p:
473 break
474 op, av = p[0]
475 if op is LITERAL:
476 cappend((op, av))
477 else:
478 break
479 else:
480 charset = c
481 elif op is BRANCH:
482 c = []
483 cappend = c.append
484 for p in av[1]:
485 if not p:
486 break
487 op, av = p[0]
488 if op is LITERAL:
489 cappend((op, av))
490 else:
491 break
492 else:
493 charset = c
494 elif op is IN:
495 charset = av
496## if prefix:
497## print "*** PREFIX", prefix, prefix_skip
498## if charset:
499## print "*** CHARSET", charset
500 # add an info block
501 emit = code.append
502 emit(OPCODES[INFO])
503 skip = len(code); emit(0)
504 # literal flag
505 mask = 0
506 if prefix:
507 mask = SRE_INFO_PREFIX
508 if len(prefix) == prefix_skip == len(pattern.data):
509 mask = mask + SRE_INFO_LITERAL
510 elif charset:
511 mask = mask + SRE_INFO_CHARSET
512 emit(mask)
513 # pattern length
514 if lo < MAXCODE:
515 emit(lo)
516 else:
517 emit(MAXCODE)
518 prefix = prefix[:MAXCODE]
519 if hi < MAXCODE:
520 emit(hi)
521 else:
522 emit(0)
523 # add literal prefix
524 if prefix:
525 emit(len(prefix)) # length
526 emit(prefix_skip) # skip
527 code.extend(prefix)
528 # generate overlap table
529 table = [-1] + ([0]*len(prefix))
530 for i in xrange(len(prefix)):
531 table[i+1] = table[i]+1
532 while table[i+1] > 0 and prefix[i] != prefix[table[i+1]-1]:
533 table[i+1] = table[table[i+1]-1]+1
534 code.extend(table[1:]) # don't store first entry
535 elif charset:
536 _compile_charset(charset, flags, code)
537 code[skip] = len(code) - skip
538
539try:
540 unicode
541except NameError:
542 STRING_TYPES = (type(""),)
543else:
544 STRING_TYPES = (type(""), type(unicode("")))
545
546def isstring(obj):
547 for tp in STRING_TYPES:
548 if isinstance(obj, tp):
549 return 1
550 return 0
551
552def _code(p, flags):
553
554 flags = p.pattern.flags | flags
555 code = []
556
557 # compile info block
558 _compile_info(code, p, flags)
559
560 # compile the pattern
561 _compile(code, p.data, flags)
562
563 code.append(OPCODES[SUCCESS])
564
565 return code
566
567def compile(p, flags=0):
568 # internal: convert pattern list to internal format
569
570 if isstring(p):
571 pattern = p
572 p = sre_parse.parse(p, flags)
573 else:
574 pattern = None
575
576 code = _code(p, flags)
577
578 # print code
579
580 # XXX: <fl> get rid of this limitation!
581 if p.pattern.groups > 100:
582 raise AssertionError(
583 "sorry, but this version only supports 100 named groups"
584 )
585
586 # map in either direction
587 groupindex = p.pattern.groupdict
588 indexgroup = [None] * p.pattern.groups
589 for k, i in groupindex.items():
590 indexgroup[i] = k
591
592 return _sre.compile(
593 pattern, flags | p.pattern.flags, code,
594 p.pattern.groups-1,
595 groupindex, indexgroup
596 )