| Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 1 | # |
| 2 | # Secret Labs' Regular Expression Engine |
| 3 | # $Id$ |
| 4 | # |
| 5 | # convert template to internal format |
| 6 | # |
| 7 | # Copyright (c) 1997-2000 by Secret Labs AB. All rights reserved. |
| 8 | # |
| Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 9 | # Portions of this engine have been developed in cooperation with |
| 10 | # CNRI. Hewlett-Packard provided funding for 1.6 integration and |
| 11 | # other compatibility work. |
| 12 | # |
| 13 | |
| Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 14 | import array, string, sys |
| 15 | |
| 16 | import _sre |
| 17 | |
| 18 | from sre_constants import * |
| 19 | |
| 20 | # find an array type code that matches the engine's code size |
| 21 | for WORDSIZE in "BHil": |
| 22 | if len(array.array(WORDSIZE, [0]).tostring()) == _sre.getcodesize(): |
| Guido van Rossum | b81e70e | 2000-04-10 17:10:48 +0000 | [diff] [blame] | 23 | break |
| Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 24 | else: |
| 25 | raise RuntimeError, "cannot find a useable array type" |
| 26 | |
| 27 | # FIXME: <fl> should move some optimizations from the parser to here! |
| 28 | |
| 29 | class Code: |
| 30 | def __init__(self): |
| Guido van Rossum | b81e70e | 2000-04-10 17:10:48 +0000 | [diff] [blame] | 31 | self.data = [] |
| Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 32 | def __len__(self): |
| Guido van Rossum | b81e70e | 2000-04-10 17:10:48 +0000 | [diff] [blame] | 33 | return len(self.data) |
| Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 34 | def __getitem__(self, index): |
| Guido van Rossum | b81e70e | 2000-04-10 17:10:48 +0000 | [diff] [blame] | 35 | return self.data[index] |
| Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 36 | def __setitem__(self, index, code): |
| Guido van Rossum | b81e70e | 2000-04-10 17:10:48 +0000 | [diff] [blame] | 37 | self.data[index] = code |
| Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 38 | def append(self, code): |
| Guido van Rossum | b81e70e | 2000-04-10 17:10:48 +0000 | [diff] [blame] | 39 | self.data.append(code) |
| Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 40 | def todata(self): |
| Guido van Rossum | b81e70e | 2000-04-10 17:10:48 +0000 | [diff] [blame] | 41 | # print self.data |
| Jeremy Hylton | b1aa195 | 2000-06-01 17:39:12 +0000 | [diff] [blame] | 42 | try: |
| 43 | return array.array(WORDSIZE, self.data).tostring() |
| 44 | except OverflowError: |
| 45 | print self.data |
| 46 | raise |
| Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 47 | |
| Fredrik Lundh | 436c3d5 | 2000-06-29 08:58:44 +0000 | [diff] [blame] | 48 | def _compile(code, pattern, flags): |
| Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 49 | append = code.append |
| 50 | for op, av in pattern: |
| Guido van Rossum | b81e70e | 2000-04-10 17:10:48 +0000 | [diff] [blame] | 51 | if op is ANY: |
| Jeremy Hylton | b1aa195 | 2000-06-01 17:39:12 +0000 | [diff] [blame] | 52 | if flags & SRE_FLAG_DOTALL: |
| 53 | append(OPCODES[op]) # any character at all! |
| Guido van Rossum | b81e70e | 2000-04-10 17:10:48 +0000 | [diff] [blame] | 54 | else: |
| Jeremy Hylton | b1aa195 | 2000-06-01 17:39:12 +0000 | [diff] [blame] | 55 | append(OPCODES[CATEGORY]) |
| 56 | append(CHCODES[CATEGORY_NOT_LINEBREAK]) |
| Guido van Rossum | b81e70e | 2000-04-10 17:10:48 +0000 | [diff] [blame] | 57 | elif op in (SUCCESS, FAILURE): |
| Jeremy Hylton | b1aa195 | 2000-06-01 17:39:12 +0000 | [diff] [blame] | 58 | append(OPCODES[op]) |
| Guido van Rossum | b81e70e | 2000-04-10 17:10:48 +0000 | [diff] [blame] | 59 | elif op is AT: |
| Jeremy Hylton | b1aa195 | 2000-06-01 17:39:12 +0000 | [diff] [blame] | 60 | append(OPCODES[op]) |
| 61 | if flags & SRE_FLAG_MULTILINE: |
| 62 | append(ATCODES[AT_MULTILINE[av]]) |
| 63 | else: |
| 64 | append(ATCODES[av]) |
| Guido van Rossum | b81e70e | 2000-04-10 17:10:48 +0000 | [diff] [blame] | 65 | elif op is BRANCH: |
| Jeremy Hylton | b1aa195 | 2000-06-01 17:39:12 +0000 | [diff] [blame] | 66 | append(OPCODES[op]) |
| Guido van Rossum | b81e70e | 2000-04-10 17:10:48 +0000 | [diff] [blame] | 67 | tail = [] |
| 68 | for av in av[1]: |
| 69 | skip = len(code); append(0) |
| Fredrik Lundh | 436c3d5 | 2000-06-29 08:58:44 +0000 | [diff] [blame] | 70 | _compile(code, av, flags) |
| 71 | ## append(OPCODES[SUCCESS]) |
| 72 | append(OPCODES[JUMP]) |
| 73 | tail.append(len(code)); append(0) |
| Guido van Rossum | b81e70e | 2000-04-10 17:10:48 +0000 | [diff] [blame] | 74 | code[skip] = len(code) - skip |
| 75 | append(0) # end of branch |
| Fredrik Lundh | 436c3d5 | 2000-06-29 08:58:44 +0000 | [diff] [blame] | 76 | for tail in tail: |
| Guido van Rossum | b81e70e | 2000-04-10 17:10:48 +0000 | [diff] [blame] | 77 | code[tail] = len(code) - tail |
| 78 | elif op is CALL: |
| Jeremy Hylton | b1aa195 | 2000-06-01 17:39:12 +0000 | [diff] [blame] | 79 | append(OPCODES[op]) |
| Guido van Rossum | b81e70e | 2000-04-10 17:10:48 +0000 | [diff] [blame] | 80 | skip = len(code); append(0) |
| Fredrik Lundh | 436c3d5 | 2000-06-29 08:58:44 +0000 | [diff] [blame] | 81 | _compile(code, av, flags) |
| Jeremy Hylton | b1aa195 | 2000-06-01 17:39:12 +0000 | [diff] [blame] | 82 | append(OPCODES[SUCCESS]) |
| Guido van Rossum | b81e70e | 2000-04-10 17:10:48 +0000 | [diff] [blame] | 83 | code[skip] = len(code) - skip |
| Fredrik Lundh | 436c3d5 | 2000-06-29 08:58:44 +0000 | [diff] [blame] | 84 | elif op is CATEGORY: |
| Jeremy Hylton | b1aa195 | 2000-06-01 17:39:12 +0000 | [diff] [blame] | 85 | append(OPCODES[op]) |
| 86 | if flags & SRE_FLAG_LOCALE: |
| 87 | append(CH_LOCALE[CHCODES[av]]) |
| Fredrik Lundh | 436c3d5 | 2000-06-29 08:58:44 +0000 | [diff] [blame] | 88 | elif flags & SRE_FLAG_UNICODE: |
| 89 | append(CH_UNICODE[CHCODES[av]]) |
| Jeremy Hylton | b1aa195 | 2000-06-01 17:39:12 +0000 | [diff] [blame] | 90 | else: |
| 91 | append(CHCODES[av]) |
| Guido van Rossum | b81e70e | 2000-04-10 17:10:48 +0000 | [diff] [blame] | 92 | elif op is GROUP: |
| Jeremy Hylton | b1aa195 | 2000-06-01 17:39:12 +0000 | [diff] [blame] | 93 | if flags & SRE_FLAG_IGNORECASE: |
| 94 | append(OPCODES[OP_IGNORE[op]]) |
| Guido van Rossum | b81e70e | 2000-04-10 17:10:48 +0000 | [diff] [blame] | 95 | else: |
| Jeremy Hylton | b1aa195 | 2000-06-01 17:39:12 +0000 | [diff] [blame] | 96 | append(OPCODES[op]) |
| 97 | append(av-1) |
| Guido van Rossum | b81e70e | 2000-04-10 17:10:48 +0000 | [diff] [blame] | 98 | elif op is IN: |
| Jeremy Hylton | b1aa195 | 2000-06-01 17:39:12 +0000 | [diff] [blame] | 99 | if flags & SRE_FLAG_IGNORECASE: |
| 100 | append(OPCODES[OP_IGNORE[op]]) |
| Fredrik Lundh | 436c3d5 | 2000-06-29 08:58:44 +0000 | [diff] [blame] | 101 | def fixup(literal, flags=flags): |
| 102 | return _sre.getlower(ord(literal), flags) |
| Guido van Rossum | b81e70e | 2000-04-10 17:10:48 +0000 | [diff] [blame] | 103 | else: |
| Jeremy Hylton | b1aa195 | 2000-06-01 17:39:12 +0000 | [diff] [blame] | 104 | append(OPCODES[op]) |
| Guido van Rossum | b81e70e | 2000-04-10 17:10:48 +0000 | [diff] [blame] | 105 | fixup = ord |
| 106 | skip = len(code); append(0) |
| 107 | for op, av in av: |
| Jeremy Hylton | b1aa195 | 2000-06-01 17:39:12 +0000 | [diff] [blame] | 108 | append(OPCODES[op]) |
| Guido van Rossum | b81e70e | 2000-04-10 17:10:48 +0000 | [diff] [blame] | 109 | if op is NEGATE: |
| 110 | pass |
| 111 | elif op is LITERAL: |
| 112 | append(fixup(av)) |
| 113 | elif op is RANGE: |
| 114 | append(fixup(av[0])) |
| 115 | append(fixup(av[1])) |
| 116 | elif op is CATEGORY: |
| Jeremy Hylton | b1aa195 | 2000-06-01 17:39:12 +0000 | [diff] [blame] | 117 | if flags & SRE_FLAG_LOCALE: |
| 118 | append(CH_LOCALE[CHCODES[av]]) |
| Fredrik Lundh | 436c3d5 | 2000-06-29 08:58:44 +0000 | [diff] [blame] | 119 | elif flags & SRE_FLAG_UNICODE: |
| 120 | append(CH_UNICODE[CHCODES[av]]) |
| Jeremy Hylton | b1aa195 | 2000-06-01 17:39:12 +0000 | [diff] [blame] | 121 | else: |
| 122 | append(CHCODES[av]) |
| Guido van Rossum | b81e70e | 2000-04-10 17:10:48 +0000 | [diff] [blame] | 123 | else: |
| 124 | raise ValueError, "unsupported set operator" |
| Jeremy Hylton | b1aa195 | 2000-06-01 17:39:12 +0000 | [diff] [blame] | 125 | append(OPCODES[FAILURE]) |
| Guido van Rossum | b81e70e | 2000-04-10 17:10:48 +0000 | [diff] [blame] | 126 | code[skip] = len(code) - skip |
| 127 | elif op in (LITERAL, NOT_LITERAL): |
| Jeremy Hylton | b1aa195 | 2000-06-01 17:39:12 +0000 | [diff] [blame] | 128 | if flags & SRE_FLAG_IGNORECASE: |
| 129 | append(OPCODES[OP_IGNORE[op]]) |
| Guido van Rossum | b81e70e | 2000-04-10 17:10:48 +0000 | [diff] [blame] | 130 | else: |
| Jeremy Hylton | b1aa195 | 2000-06-01 17:39:12 +0000 | [diff] [blame] | 131 | append(OPCODES[op]) |
| Fredrik Lundh | 436c3d5 | 2000-06-29 08:58:44 +0000 | [diff] [blame] | 132 | append(ord(av)) |
| Guido van Rossum | b81e70e | 2000-04-10 17:10:48 +0000 | [diff] [blame] | 133 | elif op is MARK: |
| Jeremy Hylton | b1aa195 | 2000-06-01 17:39:12 +0000 | [diff] [blame] | 134 | append(OPCODES[op]) |
| Guido van Rossum | b81e70e | 2000-04-10 17:10:48 +0000 | [diff] [blame] | 135 | append(av) |
| 136 | elif op in (REPEAT, MIN_REPEAT, MAX_REPEAT): |
| Fredrik Lundh | 436c3d5 | 2000-06-29 08:58:44 +0000 | [diff] [blame] | 137 | if flags & SRE_FLAG_TEMPLATE: |
| 138 | append(OPCODES[REPEAT]) |
| Guido van Rossum | b81e70e | 2000-04-10 17:10:48 +0000 | [diff] [blame] | 139 | skip = len(code); append(0) |
| 140 | append(av[0]) |
| 141 | append(av[1]) |
| Fredrik Lundh | 436c3d5 | 2000-06-29 08:58:44 +0000 | [diff] [blame] | 142 | _compile(code, av[2], flags) |
| Jeremy Hylton | b1aa195 | 2000-06-01 17:39:12 +0000 | [diff] [blame] | 143 | append(OPCODES[SUCCESS]) |
| Guido van Rossum | b81e70e | 2000-04-10 17:10:48 +0000 | [diff] [blame] | 144 | code[skip] = len(code) - skip |
| 145 | else: |
| Fredrik Lundh | 436c3d5 | 2000-06-29 08:58:44 +0000 | [diff] [blame] | 146 | lo, hi = av[2].getwidth() |
| 147 | if lo == 0: |
| 148 | raise error, "nothing to repeat" |
| 149 | if 0 and lo == hi == 1 and op is MAX_REPEAT: |
| 150 | # FIXME: <fl> need a better way to figure out when |
| 151 | # it's safe to use this one (in the parser, probably) |
| 152 | append(OPCODES[MAX_REPEAT_ONE]) |
| 153 | skip = len(code); append(0) |
| 154 | append(av[0]) |
| 155 | append(av[1]) |
| 156 | _compile(code, av[2], flags) |
| 157 | append(OPCODES[SUCCESS]) |
| 158 | code[skip] = len(code) - skip |
| Guido van Rossum | b81e70e | 2000-04-10 17:10:48 +0000 | [diff] [blame] | 159 | else: |
| Fredrik Lundh | 436c3d5 | 2000-06-29 08:58:44 +0000 | [diff] [blame] | 160 | append(OPCODES[op]) |
| 161 | skip = len(code); append(0) |
| 162 | append(av[0]) |
| 163 | append(av[1]) |
| 164 | _compile(code, av[2], flags) |
| 165 | append(OPCODES[SUCCESS]) |
| 166 | code[skip] = len(code) - skip |
| Guido van Rossum | b81e70e | 2000-04-10 17:10:48 +0000 | [diff] [blame] | 167 | elif op is SUBPATTERN: |
| Jeremy Hylton | b1aa195 | 2000-06-01 17:39:12 +0000 | [diff] [blame] | 168 | group = av[0] |
| 169 | if group: |
| 170 | append(OPCODES[MARK]) |
| 171 | append((group-1)*2) |
| Fredrik Lundh | 436c3d5 | 2000-06-29 08:58:44 +0000 | [diff] [blame] | 172 | _compile(code, av[1], flags) |
| Jeremy Hylton | b1aa195 | 2000-06-01 17:39:12 +0000 | [diff] [blame] | 173 | if group: |
| 174 | append(OPCODES[MARK]) |
| 175 | append((group-1)*2+1) |
| Guido van Rossum | b81e70e | 2000-04-10 17:10:48 +0000 | [diff] [blame] | 176 | else: |
| 177 | raise ValueError, ("unsupported operand type", op) |
| Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 178 | |
| Jeremy Hylton | b1aa195 | 2000-06-01 17:39:12 +0000 | [diff] [blame] | 179 | def compile(p, flags=0): |
| Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 180 | # convert pattern list to internal format |
| Guido van Rossum | b81e70e | 2000-04-10 17:10:48 +0000 | [diff] [blame] | 181 | if type(p) in (type(""), type(u"")): |
| 182 | import sre_parse |
| 183 | pattern = p |
| 184 | p = sre_parse.parse(p) |
| Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 185 | else: |
| Guido van Rossum | b81e70e | 2000-04-10 17:10:48 +0000 | [diff] [blame] | 186 | pattern = None |
| Jeremy Hylton | b1aa195 | 2000-06-01 17:39:12 +0000 | [diff] [blame] | 187 | flags = p.pattern.flags | flags |
| Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 188 | code = Code() |
| Jeremy Hylton | b1aa195 | 2000-06-01 17:39:12 +0000 | [diff] [blame] | 189 | _compile(code, p.data, flags) |
| 190 | code.append(OPCODES[SUCCESS]) |
| Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 191 | data = code.todata() |
| 192 | if 0: # debugging |
| Guido van Rossum | b81e70e | 2000-04-10 17:10:48 +0000 | [diff] [blame] | 193 | print |
| 194 | print "-" * 68 |
| 195 | import sre_disasm |
| 196 | sre_disasm.disasm(data) |
| 197 | print "-" * 68 |
| Jeremy Hylton | b1aa195 | 2000-06-01 17:39:12 +0000 | [diff] [blame] | 198 | return _sre.compile( |
| 199 | pattern, flags, |
| 200 | data, |
| 201 | p.pattern.groups-1, p.pattern.groupdict |
| 202 | ) |