Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 1 | # |
| 2 | # Secret Labs' Regular Expression Engine |
| 3 | # $Id$ |
| 4 | # |
| 5 | # various symbols used by the regular expression engine. |
| 6 | # run this script to update the _sre include files! |
| 7 | # |
| 8 | # Copyright (c) 1998-2000 by Secret Labs AB. All rights reserved. |
| 9 | # |
| 10 | # This code can only be used for 1.6 alpha testing. All other use |
| 11 | # require explicit permission from Secret Labs AB. |
| 12 | # |
| 13 | # Portions of this engine have been developed in cooperation with |
| 14 | # CNRI. Hewlett-Packard provided funding for 1.6 integration and |
| 15 | # other compatibility work. |
| 16 | # |
| 17 | |
| 18 | # operators |
| 19 | |
| 20 | FAILURE = "failure" |
| 21 | SUCCESS = "success" |
| 22 | |
| 23 | ANY = "any" |
| 24 | ASSERT = "assert" |
| 25 | AT = "at" |
| 26 | BRANCH = "branch" |
| 27 | CALL = "call" |
| 28 | CATEGORY = "category" |
| 29 | GROUP = "group" |
| 30 | GROUP_IGNORE = "group_ignore" |
| 31 | IN = "in" |
| 32 | IN_IGNORE = "in_ignore" |
| 33 | JUMP = "jump" |
| 34 | LITERAL = "literal" |
| 35 | LITERAL_IGNORE = "literal_ignore" |
| 36 | MARK = "mark" |
| 37 | MAX_REPEAT = "max_repeat" |
| 38 | MAX_REPEAT_ONE = "max_repeat_one" |
| 39 | MAX_UNTIL = "max_until" |
| 40 | MIN_REPEAT = "min_repeat" |
| 41 | MIN_UNTIL = "min_until" |
| 42 | NEGATE = "negate" |
| 43 | NOT_LITERAL = "not_literal" |
| 44 | NOT_LITERAL_IGNORE = "not_literal_ignore" |
| 45 | RANGE = "range" |
| 46 | REPEAT = "repeat" |
| 47 | SUBPATTERN = "subpattern" |
| 48 | |
| 49 | # positions |
| 50 | AT_BEGINNING = "at_beginning" |
Jeremy Hylton | b1aa195 | 2000-06-01 17:39:12 +0000 | [diff] [blame^] | 51 | AT_BEGINNING_LINE = "at_beginning_line" |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 52 | AT_BOUNDARY = "at_boundary" |
| 53 | AT_NON_BOUNDARY = "at_non_boundary" |
| 54 | AT_END = "at_end" |
Jeremy Hylton | b1aa195 | 2000-06-01 17:39:12 +0000 | [diff] [blame^] | 55 | AT_END_LINE = "at_end_line" |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 56 | |
| 57 | # categories |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 58 | CATEGORY_DIGIT = "category_digit" |
| 59 | CATEGORY_NOT_DIGIT = "category_not_digit" |
| 60 | CATEGORY_SPACE = "category_space" |
| 61 | CATEGORY_NOT_SPACE = "category_not_space" |
| 62 | CATEGORY_WORD = "category_word" |
| 63 | CATEGORY_NOT_WORD = "category_not_word" |
Jeremy Hylton | b1aa195 | 2000-06-01 17:39:12 +0000 | [diff] [blame^] | 64 | CATEGORY_LINEBREAK = "category_linebreak" |
| 65 | CATEGORY_NOT_LINEBREAK = "category_not_linebreak" |
| 66 | CATEGORY_LOC_DIGIT = "category_loc_digit" |
| 67 | CATEGORY_LOC_NOT_DIGIT = "category_loc_not_digit" |
| 68 | CATEGORY_LOC_SPACE = "category_loc_space" |
| 69 | CATEGORY_LOC_NOT_SPACE = "category_loc_not_space" |
| 70 | CATEGORY_LOC_WORD = "category_loc_word" |
| 71 | CATEGORY_LOC_NOT_WORD = "category_loc_not_word" |
| 72 | CATEGORY_LOC_LINEBREAK = "category_loc_linebreak" |
| 73 | CATEGORY_LOC_NOT_LINEBREAK = "category_loc_not_linebreak" |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 74 | |
Jeremy Hylton | b1aa195 | 2000-06-01 17:39:12 +0000 | [diff] [blame^] | 75 | OPCODES = [ |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 76 | |
| 77 | # failure=0 success=1 (just because it looks better that way :-) |
| 78 | FAILURE, SUCCESS, |
| 79 | |
| 80 | ANY, |
| 81 | ASSERT, |
| 82 | AT, |
| 83 | BRANCH, |
| 84 | CALL, |
| 85 | CATEGORY, |
| 86 | GROUP, GROUP_IGNORE, |
| 87 | IN, IN_IGNORE, |
| 88 | JUMP, |
| 89 | LITERAL, LITERAL_IGNORE, |
| 90 | MARK, |
| 91 | MAX_REPEAT, MAX_UNTIL, |
| 92 | MAX_REPEAT_ONE, |
| 93 | MIN_REPEAT, MIN_UNTIL, |
| 94 | NOT_LITERAL, NOT_LITERAL_IGNORE, |
| 95 | NEGATE, |
| 96 | RANGE, |
| 97 | REPEAT |
| 98 | |
| 99 | ] |
| 100 | |
Jeremy Hylton | b1aa195 | 2000-06-01 17:39:12 +0000 | [diff] [blame^] | 101 | ATCODES = [ |
| 102 | AT_BEGINNING, AT_BEGINNING_LINE, AT_BOUNDARY, |
| 103 | AT_NON_BOUNDARY, AT_END, AT_END_LINE |
| 104 | ] |
| 105 | |
| 106 | CHCODES = [ |
| 107 | CATEGORY_DIGIT, CATEGORY_NOT_DIGIT, CATEGORY_SPACE, |
| 108 | CATEGORY_NOT_SPACE, CATEGORY_WORD, CATEGORY_NOT_WORD, |
| 109 | CATEGORY_LINEBREAK, CATEGORY_NOT_LINEBREAK, CATEGORY_LOC_DIGIT, |
| 110 | CATEGORY_LOC_NOT_DIGIT, CATEGORY_LOC_SPACE, |
| 111 | CATEGORY_LOC_NOT_SPACE, CATEGORY_LOC_WORD, CATEGORY_LOC_NOT_WORD, |
| 112 | CATEGORY_LOC_LINEBREAK, CATEGORY_LOC_NOT_LINEBREAK |
| 113 | ] |
| 114 | |
| 115 | def makedict(list): |
| 116 | d = {} |
| 117 | i = 0 |
| 118 | for item in list: |
| 119 | d[item] = i |
| 120 | i = i + 1 |
| 121 | return d |
| 122 | |
| 123 | OPCODES = makedict(OPCODES) |
| 124 | ATCODES = makedict(ATCODES) |
| 125 | CHCODES = makedict(CHCODES) |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 126 | |
| 127 | # replacement operations for "ignore case" mode |
Jeremy Hylton | b1aa195 | 2000-06-01 17:39:12 +0000 | [diff] [blame^] | 128 | OP_IGNORE = { |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 129 | GROUP: GROUP_IGNORE, |
| 130 | IN: IN_IGNORE, |
| 131 | LITERAL: LITERAL_IGNORE, |
| 132 | NOT_LITERAL: NOT_LITERAL_IGNORE |
| 133 | } |
| 134 | |
Jeremy Hylton | b1aa195 | 2000-06-01 17:39:12 +0000 | [diff] [blame^] | 135 | AT_MULTILINE = { |
| 136 | AT_BEGINNING: AT_BEGINNING_LINE, |
| 137 | AT_END: AT_END_LINE |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 138 | } |
| 139 | |
Jeremy Hylton | b1aa195 | 2000-06-01 17:39:12 +0000 | [diff] [blame^] | 140 | CH_LOCALE = { |
| 141 | CATEGORY_DIGIT: CATEGORY_LOC_DIGIT, |
| 142 | CATEGORY_NOT_DIGIT: CATEGORY_LOC_NOT_DIGIT, |
| 143 | CATEGORY_SPACE: CATEGORY_LOC_SPACE, |
| 144 | CATEGORY_NOT_SPACE: CATEGORY_LOC_NOT_SPACE, |
| 145 | CATEGORY_WORD: CATEGORY_LOC_WORD, |
| 146 | CATEGORY_NOT_WORD: CATEGORY_LOC_NOT_WORD, |
| 147 | CATEGORY_LINEBREAK: CATEGORY_LOC_LINEBREAK, |
| 148 | CATEGORY_NOT_LINEBREAK: CATEGORY_LOC_NOT_LINEBREAK |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 149 | } |
| 150 | |
Jeremy Hylton | b1aa195 | 2000-06-01 17:39:12 +0000 | [diff] [blame^] | 151 | # flags |
| 152 | SRE_FLAG_TEMPLATE = 1 # NYI |
| 153 | SRE_FLAG_IGNORECASE = 2 |
| 154 | SRE_FLAG_LOCALE = 4 |
| 155 | SRE_FLAG_MULTILINE = 8 |
| 156 | SRE_FLAG_DOTALL = 16 |
| 157 | SRE_FLAG_VERBOSE = 32 |
| 158 | |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 159 | if __name__ == "__main__": |
| 160 | import string |
Jeremy Hylton | b1aa195 | 2000-06-01 17:39:12 +0000 | [diff] [blame^] | 161 | def dump(f, d, prefix): |
| 162 | items = d.items() |
| 163 | items.sort(lambda a, b: cmp(a[1], b[1])) |
| 164 | for k, v in items: |
| 165 | f.write("#define %s_%s %s\n" % (prefix, string.upper(k), v)) |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 166 | f = open("sre_constants.h", "w") |
Jeremy Hylton | b1aa195 | 2000-06-01 17:39:12 +0000 | [diff] [blame^] | 167 | f.write("/* generated from sre_constants.py */\n") |
| 168 | dump(f, OPCODES, "SRE_OP") |
| 169 | dump(f, ATCODES, "SRE") |
| 170 | dump(f, CHCODES, "SRE") |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 171 | f.close() |
| 172 | print "done" |