Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 1 | # |
| 2 | # Secret Labs' Regular Expression Engine |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 3 | # |
| 4 | # convert template to internal format |
| 5 | # |
Fredrik Lundh | 770617b | 2001-01-14 15:06:11 +0000 | [diff] [blame] | 6 | # Copyright (c) 1997-2001 by Secret Labs AB. All rights reserved. |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 7 | # |
Fredrik Lundh | 29c4ba9 | 2000-08-01 18:20:07 +0000 | [diff] [blame] | 8 | # See the sre.py file for information on usage and redistribution. |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 9 | # |
| 10 | |
Fred Drake | b8f2274 | 2001-09-04 19:10:20 +0000 | [diff] [blame] | 11 | """Internal support module for sre""" |
| 12 | |
Victor Stinner | 7fa767e | 2014-03-20 09:16:38 +0100 | [diff] [blame] | 13 | import _sre |
Christian Heimes | 5e69685 | 2008-04-09 08:37:03 +0000 | [diff] [blame] | 14 | import sre_parse |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 15 | from sre_constants import * |
Serhiy Storchaka | 70ca021 | 2013-02-16 16:47:47 +0200 | [diff] [blame] | 16 | from _sre import MAXREPEAT |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 17 | |
Fredrik Lundh | b35ffc0 | 2001-01-15 12:46:09 +0000 | [diff] [blame] | 18 | assert _sre.MAGIC == MAGIC, "SRE module mismatch" |
| 19 | |
Martin v. Löwis | 78e2f06 | 2003-04-19 12:56:08 +0000 | [diff] [blame] | 20 | if _sre.CODESIZE == 2: |
| 21 | MAXCODE = 65535 |
| 22 | else: |
Guido van Rossum | e2a383d | 2007-01-15 16:59:06 +0000 | [diff] [blame] | 23 | MAXCODE = 0xFFFFFFFF |
Fredrik Lundh | 3562f11 | 2000-07-02 12:00:07 +0000 | [diff] [blame] | 24 | |
Raymond Hettinger | 049ade2 | 2005-02-28 19:27:52 +0000 | [diff] [blame] | 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 | |
Serhiy Storchaka | 0c938f6 | 2014-11-10 12:37:16 +0200 | [diff] [blame] | 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 IOTA WITH DIALYTIKA AND TONOS, GREEK SMALL LETTER IOTA WITH DIALYTIKA AND OXIA |
| 41 | (0x390, 0x1fd3), # ΐΐ |
| 42 | # GREEK SMALL LETTER UPSILON WITH DIALYTIKA AND TONOS, GREEK SMALL LETTER UPSILON WITH DIALYTIKA AND OXIA |
| 43 | (0x3b0, 0x1fe3), # ΰΰ |
| 44 | # GREEK SMALL LETTER BETA, GREEK BETA SYMBOL |
| 45 | (0x3b2, 0x3d0), # βϐ |
| 46 | # GREEK SMALL LETTER EPSILON, GREEK LUNATE EPSILON SYMBOL |
| 47 | (0x3b5, 0x3f5), # εϵ |
| 48 | # GREEK SMALL LETTER THETA, GREEK THETA SYMBOL |
| 49 | (0x3b8, 0x3d1), # θϑ |
| 50 | # GREEK SMALL LETTER KAPPA, GREEK KAPPA SYMBOL |
| 51 | (0x3ba, 0x3f0), # κϰ |
| 52 | # GREEK SMALL LETTER PI, GREEK PI SYMBOL |
| 53 | (0x3c0, 0x3d6), # πϖ |
| 54 | # GREEK SMALL LETTER RHO, GREEK RHO SYMBOL |
| 55 | (0x3c1, 0x3f1), # ρϱ |
| 56 | # GREEK SMALL LETTER FINAL SIGMA, GREEK SMALL LETTER SIGMA |
| 57 | (0x3c2, 0x3c3), # ςσ |
| 58 | # GREEK SMALL LETTER PHI, GREEK PHI SYMBOL |
| 59 | (0x3c6, 0x3d5), # φϕ |
| 60 | # LATIN SMALL LETTER S WITH DOT ABOVE, LATIN SMALL LETTER LONG S WITH DOT ABOVE |
| 61 | (0x1e61, 0x1e9b), # ṡẛ |
| 62 | # LATIN SMALL LIGATURE LONG S T, LATIN SMALL LIGATURE ST |
| 63 | (0xfb05, 0xfb06), # ſtst |
| 64 | ) |
| 65 | |
| 66 | # Maps the lowercase code to lowercase codes which have the same uppercase. |
| 67 | _ignorecase_fixes = {i: tuple(j for j in t if i != j) |
| 68 | for t in _equivalences for i in t} |
| 69 | |
Fredrik Lundh | 7898c3e | 2000-08-07 20:59:04 +0000 | [diff] [blame] | 70 | def _compile(code, pattern, flags): |
| 71 | # internal: compile a (sub)pattern |
| 72 | emit = code.append |
Raymond Hettinger | 01c9f8c | 2004-03-26 11:16:55 +0000 | [diff] [blame] | 73 | _len = len |
Raymond Hettinger | 049ade2 | 2005-02-28 19:27:52 +0000 | [diff] [blame] | 74 | LITERAL_CODES = _LITERAL_CODES |
| 75 | REPEATING_CODES = _REPEATING_CODES |
| 76 | SUCCESS_CODES = _SUCCESS_CODES |
| 77 | ASSERT_CODES = _ASSERT_CODES |
Serhiy Storchaka | 0c938f6 | 2014-11-10 12:37:16 +0200 | [diff] [blame] | 78 | if (flags & SRE_FLAG_IGNORECASE and |
| 79 | not (flags & SRE_FLAG_LOCALE) and |
| 80 | flags & SRE_FLAG_UNICODE): |
| 81 | fixes = _ignorecase_fixes |
| 82 | else: |
| 83 | fixes = None |
Fredrik Lundh | 7898c3e | 2000-08-07 20:59:04 +0000 | [diff] [blame] | 84 | for op, av in pattern: |
Raymond Hettinger | 01c9f8c | 2004-03-26 11:16:55 +0000 | [diff] [blame] | 85 | if op in LITERAL_CODES: |
Fredrik Lundh | 7898c3e | 2000-08-07 20:59:04 +0000 | [diff] [blame] | 86 | if flags & SRE_FLAG_IGNORECASE: |
Serhiy Storchaka | 0c938f6 | 2014-11-10 12:37:16 +0200 | [diff] [blame] | 87 | lo = _sre.getlower(av, flags) |
| 88 | if fixes and lo in fixes: |
| 89 | emit(OPCODES[IN_IGNORE]) |
| 90 | skip = _len(code); emit(0) |
| 91 | if op is NOT_LITERAL: |
| 92 | emit(OPCODES[NEGATE]) |
| 93 | for k in (lo,) + fixes[lo]: |
| 94 | emit(OPCODES[LITERAL]) |
| 95 | emit(k) |
| 96 | emit(OPCODES[FAILURE]) |
| 97 | code[skip] = _len(code) - skip |
| 98 | else: |
| 99 | emit(OPCODES[OP_IGNORE[op]]) |
| 100 | emit(lo) |
Fredrik Lundh | 7898c3e | 2000-08-07 20:59:04 +0000 | [diff] [blame] | 101 | else: |
| 102 | emit(OPCODES[op]) |
Fredrik Lundh | 2e24044 | 2001-01-15 18:28:14 +0000 | [diff] [blame] | 103 | emit(av) |
Fredrik Lundh | 7898c3e | 2000-08-07 20:59:04 +0000 | [diff] [blame] | 104 | elif op is IN: |
| 105 | if flags & SRE_FLAG_IGNORECASE: |
| 106 | emit(OPCODES[OP_IGNORE[op]]) |
| 107 | def fixup(literal, flags=flags): |
| 108 | return _sre.getlower(literal, flags) |
| 109 | else: |
| 110 | emit(OPCODES[op]) |
Serhiy Storchaka | b1847e7 | 2014-10-31 12:37:50 +0200 | [diff] [blame] | 111 | fixup = None |
Raymond Hettinger | 01c9f8c | 2004-03-26 11:16:55 +0000 | [diff] [blame] | 112 | skip = _len(code); emit(0) |
Serhiy Storchaka | 0c938f6 | 2014-11-10 12:37:16 +0200 | [diff] [blame] | 113 | _compile_charset(av, flags, code, fixup, fixes) |
Raymond Hettinger | 01c9f8c | 2004-03-26 11:16:55 +0000 | [diff] [blame] | 114 | code[skip] = _len(code) - skip |
Fredrik Lundh | 7898c3e | 2000-08-07 20:59:04 +0000 | [diff] [blame] | 115 | elif op is ANY: |
| 116 | if flags & SRE_FLAG_DOTALL: |
| 117 | emit(OPCODES[ANY_ALL]) |
| 118 | else: |
| 119 | emit(OPCODES[ANY]) |
Raymond Hettinger | 01c9f8c | 2004-03-26 11:16:55 +0000 | [diff] [blame] | 120 | elif op in REPEATING_CODES: |
Fredrik Lundh | 7898c3e | 2000-08-07 20:59:04 +0000 | [diff] [blame] | 121 | if flags & SRE_FLAG_TEMPLATE: |
Collin Winter | ce36ad8 | 2007-08-30 01:19:48 +0000 | [diff] [blame] | 122 | raise error("internal: unsupported template operator") |
Raymond Hettinger | 01c9f8c | 2004-03-26 11:16:55 +0000 | [diff] [blame] | 123 | elif _simple(av) and op is not REPEAT: |
| 124 | if op is MAX_REPEAT: |
Guido van Rossum | 41c99e7 | 2003-04-14 17:59:34 +0000 | [diff] [blame] | 125 | emit(OPCODES[REPEAT_ONE]) |
| 126 | else: |
| 127 | emit(OPCODES[MIN_REPEAT_ONE]) |
Raymond Hettinger | 01c9f8c | 2004-03-26 11:16:55 +0000 | [diff] [blame] | 128 | skip = _len(code); emit(0) |
Fredrik Lundh | 7898c3e | 2000-08-07 20:59:04 +0000 | [diff] [blame] | 129 | emit(av[0]) |
| 130 | emit(av[1]) |
| 131 | _compile(code, av[2], flags) |
| 132 | emit(OPCODES[SUCCESS]) |
Raymond Hettinger | 01c9f8c | 2004-03-26 11:16:55 +0000 | [diff] [blame] | 133 | code[skip] = _len(code) - skip |
Fredrik Lundh | 7898c3e | 2000-08-07 20:59:04 +0000 | [diff] [blame] | 134 | else: |
| 135 | emit(OPCODES[REPEAT]) |
Raymond Hettinger | 01c9f8c | 2004-03-26 11:16:55 +0000 | [diff] [blame] | 136 | skip = _len(code); emit(0) |
Fredrik Lundh | 7898c3e | 2000-08-07 20:59:04 +0000 | [diff] [blame] | 137 | emit(av[0]) |
| 138 | emit(av[1]) |
| 139 | _compile(code, av[2], flags) |
Raymond Hettinger | 01c9f8c | 2004-03-26 11:16:55 +0000 | [diff] [blame] | 140 | code[skip] = _len(code) - skip |
| 141 | if op is MAX_REPEAT: |
Fredrik Lundh | 7898c3e | 2000-08-07 20:59:04 +0000 | [diff] [blame] | 142 | emit(OPCODES[MAX_UNTIL]) |
| 143 | else: |
| 144 | emit(OPCODES[MIN_UNTIL]) |
| 145 | elif op is SUBPATTERN: |
| 146 | if av[0]: |
| 147 | emit(OPCODES[MARK]) |
| 148 | emit((av[0]-1)*2) |
| 149 | # _compile_info(code, av[1], flags) |
| 150 | _compile(code, av[1], flags) |
| 151 | if av[0]: |
| 152 | emit(OPCODES[MARK]) |
| 153 | emit((av[0]-1)*2+1) |
Raymond Hettinger | 01c9f8c | 2004-03-26 11:16:55 +0000 | [diff] [blame] | 154 | elif op in SUCCESS_CODES: |
Fredrik Lundh | 7898c3e | 2000-08-07 20:59:04 +0000 | [diff] [blame] | 155 | emit(OPCODES[op]) |
Raymond Hettinger | 01c9f8c | 2004-03-26 11:16:55 +0000 | [diff] [blame] | 156 | elif op in ASSERT_CODES: |
Fredrik Lundh | 7898c3e | 2000-08-07 20:59:04 +0000 | [diff] [blame] | 157 | emit(OPCODES[op]) |
Raymond Hettinger | 01c9f8c | 2004-03-26 11:16:55 +0000 | [diff] [blame] | 158 | skip = _len(code); emit(0) |
Fredrik Lundh | 7898c3e | 2000-08-07 20:59:04 +0000 | [diff] [blame] | 159 | if av[0] >= 0: |
| 160 | emit(0) # look ahead |
| 161 | else: |
| 162 | lo, hi = av[1].getwidth() |
| 163 | if lo != hi: |
Collin Winter | ce36ad8 | 2007-08-30 01:19:48 +0000 | [diff] [blame] | 164 | raise error("look-behind requires fixed-width pattern") |
Fredrik Lundh | 7898c3e | 2000-08-07 20:59:04 +0000 | [diff] [blame] | 165 | emit(lo) # look behind |
| 166 | _compile(code, av[1], flags) |
| 167 | emit(OPCODES[SUCCESS]) |
Raymond Hettinger | 01c9f8c | 2004-03-26 11:16:55 +0000 | [diff] [blame] | 168 | code[skip] = _len(code) - skip |
Fredrik Lundh | 7898c3e | 2000-08-07 20:59:04 +0000 | [diff] [blame] | 169 | elif op is CALL: |
| 170 | emit(OPCODES[op]) |
Raymond Hettinger | 01c9f8c | 2004-03-26 11:16:55 +0000 | [diff] [blame] | 171 | skip = _len(code); emit(0) |
Fredrik Lundh | 7898c3e | 2000-08-07 20:59:04 +0000 | [diff] [blame] | 172 | _compile(code, av, flags) |
| 173 | emit(OPCODES[SUCCESS]) |
Raymond Hettinger | 01c9f8c | 2004-03-26 11:16:55 +0000 | [diff] [blame] | 174 | code[skip] = _len(code) - skip |
Fredrik Lundh | 7898c3e | 2000-08-07 20:59:04 +0000 | [diff] [blame] | 175 | elif op is AT: |
| 176 | emit(OPCODES[op]) |
| 177 | if flags & SRE_FLAG_MULTILINE: |
Fredrik Lundh | b25e1ad | 2001-03-22 15:50:10 +0000 | [diff] [blame] | 178 | av = AT_MULTILINE.get(av, av) |
| 179 | if flags & SRE_FLAG_LOCALE: |
| 180 | av = AT_LOCALE.get(av, av) |
| 181 | elif flags & SRE_FLAG_UNICODE: |
| 182 | av = AT_UNICODE.get(av, av) |
| 183 | emit(ATCODES[av]) |
Fredrik Lundh | 7898c3e | 2000-08-07 20:59:04 +0000 | [diff] [blame] | 184 | elif op is BRANCH: |
| 185 | emit(OPCODES[op]) |
| 186 | tail = [] |
Raymond Hettinger | 01c9f8c | 2004-03-26 11:16:55 +0000 | [diff] [blame] | 187 | tailappend = tail.append |
Fredrik Lundh | 7898c3e | 2000-08-07 20:59:04 +0000 | [diff] [blame] | 188 | for av in av[1]: |
Raymond Hettinger | 01c9f8c | 2004-03-26 11:16:55 +0000 | [diff] [blame] | 189 | skip = _len(code); emit(0) |
Fredrik Lundh | 7898c3e | 2000-08-07 20:59:04 +0000 | [diff] [blame] | 190 | # _compile_info(code, av, flags) |
| 191 | _compile(code, av, flags) |
| 192 | emit(OPCODES[JUMP]) |
Raymond Hettinger | 01c9f8c | 2004-03-26 11:16:55 +0000 | [diff] [blame] | 193 | tailappend(_len(code)); emit(0) |
| 194 | code[skip] = _len(code) - skip |
Fredrik Lundh | 7898c3e | 2000-08-07 20:59:04 +0000 | [diff] [blame] | 195 | emit(0) # end of branch |
| 196 | for tail in tail: |
Raymond Hettinger | 01c9f8c | 2004-03-26 11:16:55 +0000 | [diff] [blame] | 197 | code[tail] = _len(code) - tail |
Fredrik Lundh | 7898c3e | 2000-08-07 20:59:04 +0000 | [diff] [blame] | 198 | elif op is CATEGORY: |
| 199 | emit(OPCODES[op]) |
| 200 | if flags & SRE_FLAG_LOCALE: |
Fredrik Lundh | b25e1ad | 2001-03-22 15:50:10 +0000 | [diff] [blame] | 201 | av = CH_LOCALE[av] |
Fredrik Lundh | 7898c3e | 2000-08-07 20:59:04 +0000 | [diff] [blame] | 202 | elif flags & SRE_FLAG_UNICODE: |
Fredrik Lundh | b25e1ad | 2001-03-22 15:50:10 +0000 | [diff] [blame] | 203 | av = CH_UNICODE[av] |
| 204 | emit(CHCODES[av]) |
Fredrik Lundh | 7898c3e | 2000-08-07 20:59:04 +0000 | [diff] [blame] | 205 | elif op is GROUPREF: |
| 206 | if flags & SRE_FLAG_IGNORECASE: |
| 207 | emit(OPCODES[OP_IGNORE[op]]) |
| 208 | else: |
| 209 | emit(OPCODES[op]) |
| 210 | emit(av-1) |
Gustavo Niemeyer | ad3fc44 | 2003-10-17 22:13:16 +0000 | [diff] [blame] | 211 | elif op is GROUPREF_EXISTS: |
| 212 | emit(OPCODES[op]) |
Andrew M. Kuchling | c30faa8 | 2005-06-02 13:35:52 +0000 | [diff] [blame] | 213 | emit(av[0]-1) |
Raymond Hettinger | 01c9f8c | 2004-03-26 11:16:55 +0000 | [diff] [blame] | 214 | skipyes = _len(code); emit(0) |
Gustavo Niemeyer | ad3fc44 | 2003-10-17 22:13:16 +0000 | [diff] [blame] | 215 | _compile(code, av[1], flags) |
| 216 | if av[2]: |
| 217 | emit(OPCODES[JUMP]) |
Raymond Hettinger | 01c9f8c | 2004-03-26 11:16:55 +0000 | [diff] [blame] | 218 | skipno = _len(code); emit(0) |
| 219 | code[skipyes] = _len(code) - skipyes + 1 |
Gustavo Niemeyer | ad3fc44 | 2003-10-17 22:13:16 +0000 | [diff] [blame] | 220 | _compile(code, av[2], flags) |
Raymond Hettinger | 01c9f8c | 2004-03-26 11:16:55 +0000 | [diff] [blame] | 221 | code[skipno] = _len(code) - skipno |
Gustavo Niemeyer | ad3fc44 | 2003-10-17 22:13:16 +0000 | [diff] [blame] | 222 | else: |
Raymond Hettinger | 01c9f8c | 2004-03-26 11:16:55 +0000 | [diff] [blame] | 223 | code[skipyes] = _len(code) - skipyes + 1 |
Fredrik Lundh | 7898c3e | 2000-08-07 20:59:04 +0000 | [diff] [blame] | 224 | else: |
Collin Winter | ce36ad8 | 2007-08-30 01:19:48 +0000 | [diff] [blame] | 225 | raise ValueError("unsupported operand type", op) |
Fredrik Lundh | 7898c3e | 2000-08-07 20:59:04 +0000 | [diff] [blame] | 226 | |
Serhiy Storchaka | 0c938f6 | 2014-11-10 12:37:16 +0200 | [diff] [blame] | 227 | def _compile_charset(charset, flags, code, fixup=None, fixes=None): |
Fredrik Lundh | 7898c3e | 2000-08-07 20:59:04 +0000 | [diff] [blame] | 228 | # compile charset subprogram |
| 229 | emit = code.append |
Serhiy Storchaka | 0c938f6 | 2014-11-10 12:37:16 +0200 | [diff] [blame] | 230 | for op, av in _optimize_charset(charset, fixup, fixes, |
| 231 | flags & SRE_FLAG_UNICODE): |
Fredrik Lundh | 7898c3e | 2000-08-07 20:59:04 +0000 | [diff] [blame] | 232 | emit(OPCODES[op]) |
| 233 | if op is NEGATE: |
| 234 | pass |
| 235 | elif op is LITERAL: |
Serhiy Storchaka | b1847e7 | 2014-10-31 12:37:50 +0200 | [diff] [blame] | 236 | emit(av) |
Fredrik Lundh | 7898c3e | 2000-08-07 20:59:04 +0000 | [diff] [blame] | 237 | elif op is RANGE: |
Serhiy Storchaka | b1847e7 | 2014-10-31 12:37:50 +0200 | [diff] [blame] | 238 | emit(av[0]) |
| 239 | emit(av[1]) |
Fredrik Lundh | 7898c3e | 2000-08-07 20:59:04 +0000 | [diff] [blame] | 240 | elif op is CHARSET: |
| 241 | code.extend(av) |
Fredrik Lundh | 19af43d | 2001-07-02 16:58:38 +0000 | [diff] [blame] | 242 | elif op is BIGCHARSET: |
| 243 | code.extend(av) |
Fredrik Lundh | 7898c3e | 2000-08-07 20:59:04 +0000 | [diff] [blame] | 244 | elif op is CATEGORY: |
| 245 | if flags & SRE_FLAG_LOCALE: |
| 246 | emit(CHCODES[CH_LOCALE[av]]) |
| 247 | elif flags & SRE_FLAG_UNICODE: |
| 248 | emit(CHCODES[CH_UNICODE[av]]) |
| 249 | else: |
| 250 | emit(CHCODES[av]) |
| 251 | else: |
Collin Winter | ce36ad8 | 2007-08-30 01:19:48 +0000 | [diff] [blame] | 252 | raise error("internal: unsupported set operator") |
Fredrik Lundh | 7898c3e | 2000-08-07 20:59:04 +0000 | [diff] [blame] | 253 | emit(OPCODES[FAILURE]) |
| 254 | |
Serhiy Storchaka | 0c938f6 | 2014-11-10 12:37:16 +0200 | [diff] [blame] | 255 | def _optimize_charset(charset, fixup, fixes, isunicode): |
Fredrik Lundh | 7898c3e | 2000-08-07 20:59:04 +0000 | [diff] [blame] | 256 | # internal: optimize character set |
Fredrik Lundh | 3562f11 | 2000-07-02 12:00:07 +0000 | [diff] [blame] | 257 | out = [] |
Serhiy Storchaka | 68457be | 2013-10-27 08:20:29 +0200 | [diff] [blame] | 258 | tail = [] |
| 259 | charmap = bytearray(256) |
| 260 | for op, av in charset: |
| 261 | while True: |
| 262 | try: |
| 263 | if op is LITERAL: |
Serhiy Storchaka | b1847e7 | 2014-10-31 12:37:50 +0200 | [diff] [blame] | 264 | if fixup: |
Serhiy Storchaka | 0c938f6 | 2014-11-10 12:37:16 +0200 | [diff] [blame] | 265 | i = fixup(av) |
| 266 | charmap[i] = 1 |
| 267 | if fixes and i in fixes: |
| 268 | for k in fixes[i]: |
| 269 | charmap[k] = 1 |
| 270 | else: |
| 271 | charmap[av] = 1 |
Serhiy Storchaka | 68457be | 2013-10-27 08:20:29 +0200 | [diff] [blame] | 272 | elif op is RANGE: |
Serhiy Storchaka | b1847e7 | 2014-10-31 12:37:50 +0200 | [diff] [blame] | 273 | r = range(av[0], av[1]+1) |
| 274 | if fixup: |
| 275 | r = map(fixup, r) |
Serhiy Storchaka | 0c938f6 | 2014-11-10 12:37:16 +0200 | [diff] [blame] | 276 | if fixup and fixes: |
| 277 | for i in r: |
| 278 | charmap[i] = 1 |
| 279 | if i in fixes: |
| 280 | for k in fixes[i]: |
| 281 | charmap[k] = 1 |
| 282 | else: |
| 283 | for i in r: |
| 284 | charmap[i] = 1 |
Serhiy Storchaka | 68457be | 2013-10-27 08:20:29 +0200 | [diff] [blame] | 285 | elif op is NEGATE: |
| 286 | out.append((op, av)) |
| 287 | else: |
| 288 | tail.append((op, av)) |
| 289 | except IndexError: |
| 290 | if len(charmap) == 256: |
| 291 | # character set contains non-UCS1 character codes |
| 292 | charmap += b'\0' * 0xff00 |
| 293 | continue |
| 294 | # character set contains non-BMP character codes |
Serhiy Storchaka | b1847e7 | 2014-10-31 12:37:50 +0200 | [diff] [blame] | 295 | if fixup and isunicode and op is RANGE: |
| 296 | lo, hi = av |
| 297 | ranges = [av] |
| 298 | # There are only two ranges of cased astral characters: |
| 299 | # 10400-1044F (Deseret) and 118A0-118DF (Warang Citi). |
| 300 | _fixup_range(max(0x10000, lo), min(0x11fff, hi), |
| 301 | ranges, fixup) |
| 302 | for lo, hi in ranges: |
| 303 | if lo == hi: |
| 304 | tail.append((LITERAL, hi)) |
| 305 | else: |
| 306 | tail.append((RANGE, (lo, hi))) |
| 307 | else: |
| 308 | tail.append((op, av)) |
Serhiy Storchaka | 68457be | 2013-10-27 08:20:29 +0200 | [diff] [blame] | 309 | break |
| 310 | |
Fredrik Lundh | 3562f11 | 2000-07-02 12:00:07 +0000 | [diff] [blame] | 311 | # compress character map |
Fredrik Lundh | 3562f11 | 2000-07-02 12:00:07 +0000 | [diff] [blame] | 312 | runs = [] |
Serhiy Storchaka | 68457be | 2013-10-27 08:20:29 +0200 | [diff] [blame] | 313 | q = 0 |
| 314 | while True: |
| 315 | p = charmap.find(1, q) |
| 316 | if p < 0: |
| 317 | break |
| 318 | if len(runs) >= 2: |
| 319 | runs = None |
| 320 | break |
| 321 | q = charmap.find(0, p) |
| 322 | if q < 0: |
| 323 | runs.append((p, len(charmap))) |
| 324 | break |
| 325 | runs.append((p, q)) |
| 326 | if runs is not None: |
Fredrik Lundh | 3562f11 | 2000-07-02 12:00:07 +0000 | [diff] [blame] | 327 | # use literal/range |
Serhiy Storchaka | 68457be | 2013-10-27 08:20:29 +0200 | [diff] [blame] | 328 | for p, q in runs: |
| 329 | if q - p == 1: |
| 330 | out.append((LITERAL, p)) |
Fredrik Lundh | 3562f11 | 2000-07-02 12:00:07 +0000 | [diff] [blame] | 331 | else: |
Serhiy Storchaka | 68457be | 2013-10-27 08:20:29 +0200 | [diff] [blame] | 332 | out.append((RANGE, (p, q - 1))) |
| 333 | out += tail |
Serhiy Storchaka | b1847e7 | 2014-10-31 12:37:50 +0200 | [diff] [blame] | 334 | # if the case was changed or new representation is more compact |
| 335 | if fixup or len(out) < len(charset): |
Fredrik Lundh | 3562f11 | 2000-07-02 12:00:07 +0000 | [diff] [blame] | 336 | return out |
Serhiy Storchaka | b1847e7 | 2014-10-31 12:37:50 +0200 | [diff] [blame] | 337 | # else original character set is good enough |
Serhiy Storchaka | 68457be | 2013-10-27 08:20:29 +0200 | [diff] [blame] | 338 | return charset |
| 339 | |
| 340 | # use bitmap |
| 341 | if len(charmap) == 256: |
Fredrik Lundh | 19af43d | 2001-07-02 16:58:38 +0000 | [diff] [blame] | 342 | data = _mk_bitmap(charmap) |
Serhiy Storchaka | 68457be | 2013-10-27 08:20:29 +0200 | [diff] [blame] | 343 | out.append((CHARSET, data)) |
| 344 | out += tail |
Fredrik Lundh | 3562f11 | 2000-07-02 12:00:07 +0000 | [diff] [blame] | 345 | return out |
Fredrik Lundh | 3562f11 | 2000-07-02 12:00:07 +0000 | [diff] [blame] | 346 | |
Serhiy Storchaka | 68457be | 2013-10-27 08:20:29 +0200 | [diff] [blame] | 347 | # To represent a big charset, first a bitmap of all characters in the |
| 348 | # set is constructed. Then, this bitmap is sliced into chunks of 256 |
| 349 | # characters, duplicate chunks are eliminated, and each chunk is |
| 350 | # given a number. In the compiled expression, the charset is |
| 351 | # represented by a 32-bit word sequence, consisting of one word for |
| 352 | # the number of different chunks, a sequence of 256 bytes (64 words) |
| 353 | # of chunk numbers indexed by their original chunk position, and a |
| 354 | # sequence of 256-bit chunks (8 words each). |
Fredrik Lundh | 19af43d | 2001-07-02 16:58:38 +0000 | [diff] [blame] | 355 | |
Serhiy Storchaka | 68457be | 2013-10-27 08:20:29 +0200 | [diff] [blame] | 356 | # Compression is normally good: in a typical charset, large ranges of |
| 357 | # Unicode will be either completely excluded (e.g. if only cyrillic |
| 358 | # letters are to be matched), or completely included (e.g. if large |
| 359 | # subranges of Kanji match). These ranges will be represented by |
| 360 | # chunks of all one-bits or all zero-bits. |
Fredrik Lundh | 19af43d | 2001-07-02 16:58:38 +0000 | [diff] [blame] | 361 | |
Serhiy Storchaka | 68457be | 2013-10-27 08:20:29 +0200 | [diff] [blame] | 362 | # Matching can be also done efficiently: the more significant byte of |
| 363 | # the Unicode character is an index into the chunk number, and the |
| 364 | # less significant byte is a bit index in the chunk (just like the |
| 365 | # CHARSET matching). |
Fredrik Lundh | 19af43d | 2001-07-02 16:58:38 +0000 | [diff] [blame] | 366 | |
Serhiy Storchaka | 68457be | 2013-10-27 08:20:29 +0200 | [diff] [blame] | 367 | charmap = bytes(charmap) # should be hashable |
Fredrik Lundh | 19af43d | 2001-07-02 16:58:38 +0000 | [diff] [blame] | 368 | comps = {} |
Serhiy Storchaka | 68457be | 2013-10-27 08:20:29 +0200 | [diff] [blame] | 369 | mapping = bytearray(256) |
Fredrik Lundh | 19af43d | 2001-07-02 16:58:38 +0000 | [diff] [blame] | 370 | block = 0 |
Serhiy Storchaka | 68457be | 2013-10-27 08:20:29 +0200 | [diff] [blame] | 371 | data = bytearray() |
| 372 | for i in range(0, 65536, 256): |
| 373 | chunk = charmap[i: i + 256] |
| 374 | if chunk in comps: |
| 375 | mapping[i // 256] = comps[chunk] |
| 376 | else: |
| 377 | mapping[i // 256] = comps[chunk] = block |
| 378 | block += 1 |
| 379 | data += chunk |
| 380 | data = _mk_bitmap(data) |
| 381 | data[0:0] = [block] + _bytes_to_codes(mapping) |
| 382 | out.append((BIGCHARSET, data)) |
| 383 | out += tail |
| 384 | return out |
| 385 | |
Serhiy Storchaka | b1847e7 | 2014-10-31 12:37:50 +0200 | [diff] [blame] | 386 | def _fixup_range(lo, hi, ranges, fixup): |
| 387 | for i in map(fixup, range(lo, hi+1)): |
| 388 | for k, (lo, hi) in enumerate(ranges): |
| 389 | if i < lo: |
| 390 | if l == lo - 1: |
| 391 | ranges[k] = (i, hi) |
| 392 | else: |
| 393 | ranges.insert(k, (i, i)) |
| 394 | break |
| 395 | elif i > hi: |
| 396 | if i == hi + 1: |
| 397 | ranges[k] = (lo, i) |
| 398 | break |
| 399 | else: |
| 400 | break |
| 401 | else: |
| 402 | ranges.append((i, i)) |
| 403 | |
Serhiy Storchaka | 68457be | 2013-10-27 08:20:29 +0200 | [diff] [blame] | 404 | _CODEBITS = _sre.CODESIZE * 8 |
| 405 | _BITS_TRANS = b'0' + b'1' * 255 |
| 406 | def _mk_bitmap(bits, _CODEBITS=_CODEBITS, _int=int): |
| 407 | s = bits.translate(_BITS_TRANS)[::-1] |
| 408 | return [_int(s[i - _CODEBITS: i], 2) |
| 409 | for i in range(len(s), 0, -_CODEBITS)] |
| 410 | |
| 411 | def _bytes_to_codes(b): |
| 412 | # Convert block indices to word array |
Serhiy Storchaka | 19e9158 | 2014-11-10 13:24:47 +0200 | [diff] [blame] | 413 | a = memoryview(b).cast('I') |
Serhiy Storchaka | 68457be | 2013-10-27 08:20:29 +0200 | [diff] [blame] | 414 | assert a.itemsize == _sre.CODESIZE |
| 415 | assert len(a) * a.itemsize == len(b) |
| 416 | return a.tolist() |
Fredrik Lundh | 19af43d | 2001-07-02 16:58:38 +0000 | [diff] [blame] | 417 | |
Fredrik Lundh | e186983 | 2000-08-01 22:47:49 +0000 | [diff] [blame] | 418 | def _simple(av): |
| 419 | # check if av is a "simple" operator |
| 420 | lo, hi = av[2].getwidth() |
Fredrik Lundh | e186983 | 2000-08-01 22:47:49 +0000 | [diff] [blame] | 421 | return lo == hi == 1 and av[2][0][0] != SUBPATTERN |
| 422 | |
Antoine Pitrou | 79aa68d | 2013-10-25 21:36:10 +0200 | [diff] [blame] | 423 | def _generate_overlap_table(prefix): |
| 424 | """ |
| 425 | Generate an overlap table for the following prefix. |
| 426 | An overlap table is a table of the same size as the prefix which |
| 427 | informs about the potential self-overlap for each index in the prefix: |
| 428 | - if overlap[i] == 0, prefix[i:] can't overlap prefix[0:...] |
| 429 | - if overlap[i] == k with 0 < k <= i, prefix[i-k+1:i+1] overlaps with |
| 430 | prefix[0:k] |
| 431 | """ |
| 432 | table = [0] * len(prefix) |
| 433 | for i in range(1, len(prefix)): |
| 434 | idx = table[i - 1] |
| 435 | while prefix[i] != prefix[idx]: |
| 436 | if idx == 0: |
| 437 | table[i] = 0 |
| 438 | break |
| 439 | idx = table[idx - 1] |
| 440 | else: |
| 441 | table[i] = idx + 1 |
| 442 | return table |
| 443 | |
Fredrik Lundh | 29c08be | 2000-06-29 23:33:12 +0000 | [diff] [blame] | 444 | def _compile_info(code, pattern, flags): |
| 445 | # internal: compile an info block. in the current version, |
Fredrik Lundh | 3562f11 | 2000-07-02 12:00:07 +0000 | [diff] [blame] | 446 | # this contains min/max pattern width, and an optional literal |
| 447 | # prefix or a character map |
Fredrik Lundh | 29c08be | 2000-06-29 23:33:12 +0000 | [diff] [blame] | 448 | lo, hi = pattern.getwidth() |
| 449 | if lo == 0: |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 450 | return # not worth it |
Fredrik Lundh | 29c08be | 2000-06-29 23:33:12 +0000 | [diff] [blame] | 451 | # look for a literal prefix |
| 452 | prefix = [] |
Raymond Hettinger | 01c9f8c | 2004-03-26 11:16:55 +0000 | [diff] [blame] | 453 | prefixappend = prefix.append |
Fredrik Lundh | 7898c3e | 2000-08-07 20:59:04 +0000 | [diff] [blame] | 454 | prefix_skip = 0 |
Fredrik Lundh | 3562f11 | 2000-07-02 12:00:07 +0000 | [diff] [blame] | 455 | charset = [] # not used |
Raymond Hettinger | 01c9f8c | 2004-03-26 11:16:55 +0000 | [diff] [blame] | 456 | charsetappend = charset.append |
Fredrik Lundh | 29c08be | 2000-06-29 23:33:12 +0000 | [diff] [blame] | 457 | if not (flags & SRE_FLAG_IGNORECASE): |
Fredrik Lundh | 7898c3e | 2000-08-07 20:59:04 +0000 | [diff] [blame] | 458 | # look for literal prefix |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 459 | for op, av in pattern.data: |
| 460 | if op is LITERAL: |
Fredrik Lundh | 7898c3e | 2000-08-07 20:59:04 +0000 | [diff] [blame] | 461 | if len(prefix) == prefix_skip: |
| 462 | prefix_skip = prefix_skip + 1 |
Raymond Hettinger | 01c9f8c | 2004-03-26 11:16:55 +0000 | [diff] [blame] | 463 | prefixappend(av) |
Fredrik Lundh | 7898c3e | 2000-08-07 20:59:04 +0000 | [diff] [blame] | 464 | elif op is SUBPATTERN and len(av[1]) == 1: |
| 465 | op, av = av[1][0] |
| 466 | if op is LITERAL: |
Raymond Hettinger | 01c9f8c | 2004-03-26 11:16:55 +0000 | [diff] [blame] | 467 | prefixappend(av) |
Fredrik Lundh | 7898c3e | 2000-08-07 20:59:04 +0000 | [diff] [blame] | 468 | else: |
| 469 | break |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 470 | else: |
| 471 | break |
Fredrik Lundh | 7898c3e | 2000-08-07 20:59:04 +0000 | [diff] [blame] | 472 | # if no prefix, look for charset prefix |
| 473 | if not prefix and pattern.data: |
| 474 | op, av = pattern.data[0] |
| 475 | if op is SUBPATTERN and av[1]: |
| 476 | op, av = av[1][0] |
| 477 | if op is LITERAL: |
Raymond Hettinger | 01c9f8c | 2004-03-26 11:16:55 +0000 | [diff] [blame] | 478 | charsetappend((op, av)) |
Fredrik Lundh | 7898c3e | 2000-08-07 20:59:04 +0000 | [diff] [blame] | 479 | elif op is BRANCH: |
| 480 | c = [] |
Raymond Hettinger | 01c9f8c | 2004-03-26 11:16:55 +0000 | [diff] [blame] | 481 | cappend = c.append |
Fredrik Lundh | 7898c3e | 2000-08-07 20:59:04 +0000 | [diff] [blame] | 482 | for p in av[1]: |
| 483 | if not p: |
| 484 | break |
| 485 | op, av = p[0] |
| 486 | if op is LITERAL: |
Raymond Hettinger | 01c9f8c | 2004-03-26 11:16:55 +0000 | [diff] [blame] | 487 | cappend((op, av)) |
Fredrik Lundh | 7898c3e | 2000-08-07 20:59:04 +0000 | [diff] [blame] | 488 | else: |
| 489 | break |
| 490 | else: |
| 491 | charset = c |
| 492 | elif op is BRANCH: |
| 493 | c = [] |
Raymond Hettinger | 01c9f8c | 2004-03-26 11:16:55 +0000 | [diff] [blame] | 494 | cappend = c.append |
Fredrik Lundh | 7898c3e | 2000-08-07 20:59:04 +0000 | [diff] [blame] | 495 | for p in av[1]: |
| 496 | if not p: |
| 497 | break |
| 498 | op, av = p[0] |
| 499 | if op is LITERAL: |
Raymond Hettinger | 01c9f8c | 2004-03-26 11:16:55 +0000 | [diff] [blame] | 500 | cappend((op, av)) |
Fredrik Lundh | 7898c3e | 2000-08-07 20:59:04 +0000 | [diff] [blame] | 501 | else: |
| 502 | break |
| 503 | else: |
| 504 | charset = c |
| 505 | elif op is IN: |
| 506 | charset = av |
| 507 | ## if prefix: |
| 508 | ## print "*** PREFIX", prefix, prefix_skip |
| 509 | ## if charset: |
| 510 | ## print "*** CHARSET", charset |
Fredrik Lundh | 29c08be | 2000-06-29 23:33:12 +0000 | [diff] [blame] | 511 | # add an info block |
| 512 | emit = code.append |
| 513 | emit(OPCODES[INFO]) |
| 514 | skip = len(code); emit(0) |
| 515 | # literal flag |
| 516 | mask = 0 |
Fredrik Lundh | 3562f11 | 2000-07-02 12:00:07 +0000 | [diff] [blame] | 517 | if prefix: |
| 518 | mask = SRE_INFO_PREFIX |
Fredrik Lundh | 7898c3e | 2000-08-07 20:59:04 +0000 | [diff] [blame] | 519 | if len(prefix) == prefix_skip == len(pattern.data): |
Fredrik Lundh | 3562f11 | 2000-07-02 12:00:07 +0000 | [diff] [blame] | 520 | mask = mask + SRE_INFO_LITERAL |
| 521 | elif charset: |
| 522 | mask = mask + SRE_INFO_CHARSET |
Fredrik Lundh | 29c08be | 2000-06-29 23:33:12 +0000 | [diff] [blame] | 523 | emit(mask) |
| 524 | # pattern length |
Fredrik Lundh | 3562f11 | 2000-07-02 12:00:07 +0000 | [diff] [blame] | 525 | if lo < MAXCODE: |
| 526 | emit(lo) |
| 527 | else: |
| 528 | emit(MAXCODE) |
| 529 | prefix = prefix[:MAXCODE] |
| 530 | if hi < MAXCODE: |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 531 | emit(hi) |
Fredrik Lundh | 29c08be | 2000-06-29 23:33:12 +0000 | [diff] [blame] | 532 | else: |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 533 | emit(0) |
Fredrik Lundh | 29c08be | 2000-06-29 23:33:12 +0000 | [diff] [blame] | 534 | # add literal prefix |
Fredrik Lundh | 29c08be | 2000-06-29 23:33:12 +0000 | [diff] [blame] | 535 | if prefix: |
Fredrik Lundh | 7898c3e | 2000-08-07 20:59:04 +0000 | [diff] [blame] | 536 | emit(len(prefix)) # length |
| 537 | emit(prefix_skip) # skip |
| 538 | code.extend(prefix) |
| 539 | # generate overlap table |
Antoine Pitrou | 79aa68d | 2013-10-25 21:36:10 +0200 | [diff] [blame] | 540 | code.extend(_generate_overlap_table(prefix)) |
Fredrik Lundh | 3562f11 | 2000-07-02 12:00:07 +0000 | [diff] [blame] | 541 | elif charset: |
Guido van Rossum | 577fb5a | 2003-02-24 01:18:35 +0000 | [diff] [blame] | 542 | _compile_charset(charset, flags, code) |
Fredrik Lundh | 29c08be | 2000-06-29 23:33:12 +0000 | [diff] [blame] | 543 | code[skip] = len(code) - skip |
| 544 | |
Just van Rossum | 7490250 | 2003-07-02 21:37:16 +0000 | [diff] [blame] | 545 | def isstring(obj): |
Thomas Wouters | 40a088d | 2008-03-18 20:19:54 +0000 | [diff] [blame] | 546 | return isinstance(obj, (str, bytes)) |
Just van Rossum | 7490250 | 2003-07-02 21:37:16 +0000 | [diff] [blame] | 547 | |
Fredrik Lundh | 2f2c67d | 2000-08-01 21:05:41 +0000 | [diff] [blame] | 548 | def _code(p, flags): |
Fredrik Lundh | 29c08be | 2000-06-29 23:33:12 +0000 | [diff] [blame] | 549 | |
Jeremy Hylton | b1aa195 | 2000-06-01 17:39:12 +0000 | [diff] [blame] | 550 | flags = p.pattern.flags | flags |
Fredrik Lundh | be2211e | 2000-06-29 16:57:40 +0000 | [diff] [blame] | 551 | code = [] |
Fredrik Lundh | 29c08be | 2000-06-29 23:33:12 +0000 | [diff] [blame] | 552 | |
| 553 | # compile info block |
| 554 | _compile_info(code, p, flags) |
| 555 | |
| 556 | # compile the pattern |
Jeremy Hylton | b1aa195 | 2000-06-01 17:39:12 +0000 | [diff] [blame] | 557 | _compile(code, p.data, flags) |
Fredrik Lundh | 29c08be | 2000-06-29 23:33:12 +0000 | [diff] [blame] | 558 | |
Jeremy Hylton | b1aa195 | 2000-06-01 17:39:12 +0000 | [diff] [blame] | 559 | code.append(OPCODES[SUCCESS]) |
Fredrik Lundh | 29c08be | 2000-06-29 23:33:12 +0000 | [diff] [blame] | 560 | |
Fredrik Lundh | 29c4ba9 | 2000-08-01 18:20:07 +0000 | [diff] [blame] | 561 | return code |
| 562 | |
| 563 | def compile(p, flags=0): |
| 564 | # internal: convert pattern list to internal format |
| 565 | |
Just van Rossum | 7490250 | 2003-07-02 21:37:16 +0000 | [diff] [blame] | 566 | if isstring(p): |
Fredrik Lundh | 29c4ba9 | 2000-08-01 18:20:07 +0000 | [diff] [blame] | 567 | pattern = p |
| 568 | p = sre_parse.parse(p, flags) |
| 569 | else: |
| 570 | pattern = None |
| 571 | |
Fredrik Lundh | 2f2c67d | 2000-08-01 21:05:41 +0000 | [diff] [blame] | 572 | code = _code(p, flags) |
Fredrik Lundh | 29c4ba9 | 2000-08-01 18:20:07 +0000 | [diff] [blame] | 573 | |
Fredrik Lundh | 8a3ebf8 | 2000-07-23 21:46:17 +0000 | [diff] [blame] | 574 | # print code |
| 575 | |
Fredrik Lundh | 770617b | 2001-01-14 15:06:11 +0000 | [diff] [blame] | 576 | # XXX: <fl> get rid of this limitation! |
Fredrik Lundh | 5e7d51b | 2004-10-15 06:15:08 +0000 | [diff] [blame] | 577 | if p.pattern.groups > 100: |
| 578 | raise AssertionError( |
| 579 | "sorry, but this version only supports 100 named groups" |
| 580 | ) |
Fredrik Lundh | 29c08be | 2000-06-29 23:33:12 +0000 | [diff] [blame] | 581 | |
Fredrik Lundh | c230173 | 2000-07-02 22:25:39 +0000 | [diff] [blame] | 582 | # map in either direction |
| 583 | groupindex = p.pattern.groupdict |
| 584 | indexgroup = [None] * p.pattern.groups |
| 585 | for k, i in groupindex.items(): |
| 586 | indexgroup[i] = k |
| 587 | |
Jeremy Hylton | b1aa195 | 2000-06-01 17:39:12 +0000 | [diff] [blame] | 588 | return _sre.compile( |
Christian Heimes | 072c0f1 | 2008-01-03 23:01:04 +0000 | [diff] [blame] | 589 | pattern, flags | p.pattern.flags, code, |
Fredrik Lundh | 6f01398 | 2000-07-03 18:44:21 +0000 | [diff] [blame] | 590 | p.pattern.groups-1, |
| 591 | groupindex, indexgroup |
Fredrik Lundh | 90a0791 | 2000-06-30 07:50:59 +0000 | [diff] [blame] | 592 | ) |