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 | # various symbols used by the regular expression engine. |
| 5 | # run this script to update the _sre include files! |
| 6 | # |
Fredrik Lundh | 770617b | 2001-01-14 15:06:11 +0000 | [diff] [blame] | 7 | # Copyright (c) 1998-2001 by Secret Labs AB. All rights reserved. |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 8 | # |
Fredrik Lundh | 29c4ba9 | 2000-08-01 18:20:07 +0000 | [diff] [blame] | 9 | # See the sre.py file for information on usage and redistribution. |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 10 | # |
| 11 | |
Fred Drake | b8f2274 | 2001-09-04 19:10:20 +0000 | [diff] [blame] | 12 | """Internal support module for sre""" |
| 13 | |
Fredrik Lundh | b35ffc0 | 2001-01-15 12:46:09 +0000 | [diff] [blame] | 14 | # update when constants are added or removed |
| 15 | |
Serhiy Storchaka | 898ff03 | 2017-05-05 08:53:40 +0300 | [diff] [blame] | 16 | MAGIC = 20170530 |
Fredrik Lundh | b35ffc0 | 2001-01-15 12:46:09 +0000 | [diff] [blame] | 17 | |
Serhiy Storchaka | 9baa5b2 | 2014-09-29 22:49:23 +0300 | [diff] [blame] | 18 | from _sre import MAXREPEAT, MAXGROUPS |
Serhiy Storchaka | 9acb9bc | 2013-02-18 11:14:21 +0200 | [diff] [blame] | 19 | |
Fredrik Lundh | b35ffc0 | 2001-01-15 12:46:09 +0000 | [diff] [blame] | 20 | # SRE standard exception (access as sre.error) |
Fredrik Lundh | 436c3d58 | 2000-06-29 08:58:44 +0000 | [diff] [blame] | 21 | # should this really be here? |
| 22 | |
| 23 | class error(Exception): |
Serhiy Storchaka | 12d6b5d | 2017-05-27 16:12:48 +0300 | [diff] [blame] | 24 | """Exception raised for invalid regular expressions. |
| 25 | |
| 26 | Attributes: |
| 27 | |
| 28 | msg: The unformatted error message |
| 29 | pattern: The regular expression pattern |
| 30 | pos: The index in the pattern where compilation failed (may be None) |
| 31 | lineno: The line corresponding to pos (may be None) |
| 32 | colno: The column corresponding to pos (may be None) |
| 33 | """ |
| 34 | |
Serhiy Storchaka | ad446d5 | 2014-11-10 13:49:00 +0200 | [diff] [blame] | 35 | def __init__(self, msg, pattern=None, pos=None): |
| 36 | self.msg = msg |
| 37 | self.pattern = pattern |
| 38 | self.pos = pos |
| 39 | if pattern is not None and pos is not None: |
| 40 | msg = '%s at position %d' % (msg, pos) |
| 41 | if isinstance(pattern, str): |
| 42 | newline = '\n' |
| 43 | else: |
| 44 | newline = b'\n' |
| 45 | self.lineno = pattern.count(newline, 0, pos) + 1 |
| 46 | self.colno = pos - pattern.rfind(newline, 0, pos) |
| 47 | if newline in pattern: |
| 48 | msg = '%s (line %d, column %d)' % (msg, self.lineno, self.colno) |
| 49 | else: |
| 50 | self.lineno = self.colno = None |
| 51 | super().__init__(msg) |
| 52 | |
Serhiy Storchaka | c7f7d38 | 2014-11-09 20:48:36 +0200 | [diff] [blame] | 53 | |
| 54 | class _NamedIntConstant(int): |
| 55 | def __new__(cls, value, name): |
| 56 | self = super(_NamedIntConstant, cls).__new__(cls, value) |
| 57 | self.name = name |
| 58 | return self |
| 59 | |
| 60 | def __str__(self): |
| 61 | return self.name |
| 62 | |
| 63 | __repr__ = __str__ |
| 64 | |
| 65 | MAXREPEAT = _NamedIntConstant(MAXREPEAT, 'MAXREPEAT') |
| 66 | |
| 67 | def _makecodes(names): |
| 68 | names = names.strip().split() |
| 69 | items = [_NamedIntConstant(i, name) for i, name in enumerate(names)] |
| 70 | globals().update({item.name: item for item in items}) |
| 71 | return items |
| 72 | |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 73 | # operators |
Serhiy Storchaka | c7f7d38 | 2014-11-09 20:48:36 +0200 | [diff] [blame] | 74 | # failure=0 success=1 (just because it looks better that way :-) |
| 75 | OPCODES = _makecodes(""" |
| 76 | FAILURE SUCCESS |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 77 | |
Serhiy Storchaka | c7f7d38 | 2014-11-09 20:48:36 +0200 | [diff] [blame] | 78 | ANY ANY_ALL |
| 79 | ASSERT ASSERT_NOT |
| 80 | AT |
| 81 | BRANCH |
| 82 | CALL |
| 83 | CATEGORY |
| 84 | CHARSET BIGCHARSET |
| 85 | GROUPREF GROUPREF_EXISTS GROUPREF_IGNORE |
| 86 | IN IN_IGNORE |
| 87 | INFO |
| 88 | JUMP |
| 89 | LITERAL LITERAL_IGNORE |
| 90 | MARK |
| 91 | MAX_UNTIL |
| 92 | MIN_UNTIL |
| 93 | NOT_LITERAL NOT_LITERAL_IGNORE |
| 94 | NEGATE |
| 95 | RANGE |
| 96 | REPEAT |
| 97 | REPEAT_ONE |
| 98 | SUBPATTERN |
| 99 | MIN_REPEAT_ONE |
| 100 | RANGE_IGNORE |
Serhiy Storchaka | 898ff03 | 2017-05-05 08:53:40 +0300 | [diff] [blame] | 101 | LITERAL_LOC_IGNORE |
| 102 | NOT_LITERAL_LOC_IGNORE |
| 103 | IN_LOC_IGNORE |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 104 | |
Serhiy Storchaka | c7f7d38 | 2014-11-09 20:48:36 +0200 | [diff] [blame] | 105 | MIN_REPEAT MAX_REPEAT |
| 106 | """) |
| 107 | del OPCODES[-2:] # remove MIN_REPEAT and MAX_REPEAT |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 108 | |
| 109 | # positions |
Serhiy Storchaka | c7f7d38 | 2014-11-09 20:48:36 +0200 | [diff] [blame] | 110 | ATCODES = _makecodes(""" |
| 111 | AT_BEGINNING AT_BEGINNING_LINE AT_BEGINNING_STRING |
| 112 | AT_BOUNDARY AT_NON_BOUNDARY |
| 113 | AT_END AT_END_LINE AT_END_STRING |
| 114 | AT_LOC_BOUNDARY AT_LOC_NON_BOUNDARY |
| 115 | AT_UNI_BOUNDARY AT_UNI_NON_BOUNDARY |
| 116 | """) |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 117 | |
| 118 | # categories |
Serhiy Storchaka | c7f7d38 | 2014-11-09 20:48:36 +0200 | [diff] [blame] | 119 | CHCODES = _makecodes(""" |
| 120 | CATEGORY_DIGIT CATEGORY_NOT_DIGIT |
| 121 | CATEGORY_SPACE CATEGORY_NOT_SPACE |
| 122 | CATEGORY_WORD CATEGORY_NOT_WORD |
| 123 | CATEGORY_LINEBREAK CATEGORY_NOT_LINEBREAK |
| 124 | CATEGORY_LOC_WORD CATEGORY_LOC_NOT_WORD |
| 125 | CATEGORY_UNI_DIGIT CATEGORY_UNI_NOT_DIGIT |
| 126 | CATEGORY_UNI_SPACE CATEGORY_UNI_NOT_SPACE |
| 127 | CATEGORY_UNI_WORD CATEGORY_UNI_NOT_WORD |
| 128 | CATEGORY_UNI_LINEBREAK CATEGORY_UNI_NOT_LINEBREAK |
| 129 | """) |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 130 | |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 131 | |
| 132 | # replacement operations for "ignore case" mode |
Jeremy Hylton | b1aa195 | 2000-06-01 17:39:12 +0000 | [diff] [blame] | 133 | OP_IGNORE = { |
Fredrik Lundh | 72b82ba | 2000-07-03 21:31:48 +0000 | [diff] [blame] | 134 | GROUPREF: GROUPREF_IGNORE, |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 135 | IN: IN_IGNORE, |
| 136 | LITERAL: LITERAL_IGNORE, |
Serhiy Storchaka | 4b8f894 | 2014-10-31 12:36:56 +0200 | [diff] [blame] | 137 | NOT_LITERAL: NOT_LITERAL_IGNORE, |
| 138 | RANGE: RANGE_IGNORE, |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 139 | } |
| 140 | |
Serhiy Storchaka | 898ff03 | 2017-05-05 08:53:40 +0300 | [diff] [blame] | 141 | OP_LOC_IGNORE = { |
| 142 | LITERAL: LITERAL_LOC_IGNORE, |
| 143 | NOT_LITERAL: NOT_LITERAL_LOC_IGNORE, |
| 144 | } |
| 145 | |
Jeremy Hylton | b1aa195 | 2000-06-01 17:39:12 +0000 | [diff] [blame] | 146 | AT_MULTILINE = { |
| 147 | AT_BEGINNING: AT_BEGINNING_LINE, |
| 148 | AT_END: AT_END_LINE |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 149 | } |
| 150 | |
Fredrik Lundh | b25e1ad | 2001-03-22 15:50:10 +0000 | [diff] [blame] | 151 | AT_LOCALE = { |
| 152 | AT_BOUNDARY: AT_LOC_BOUNDARY, |
| 153 | AT_NON_BOUNDARY: AT_LOC_NON_BOUNDARY |
| 154 | } |
| 155 | |
| 156 | AT_UNICODE = { |
| 157 | AT_BOUNDARY: AT_UNI_BOUNDARY, |
| 158 | AT_NON_BOUNDARY: AT_UNI_NON_BOUNDARY |
| 159 | } |
| 160 | |
Jeremy Hylton | b1aa195 | 2000-06-01 17:39:12 +0000 | [diff] [blame] | 161 | CH_LOCALE = { |
Fredrik Lundh | 436c3d58 | 2000-06-29 08:58:44 +0000 | [diff] [blame] | 162 | CATEGORY_DIGIT: CATEGORY_DIGIT, |
| 163 | CATEGORY_NOT_DIGIT: CATEGORY_NOT_DIGIT, |
| 164 | CATEGORY_SPACE: CATEGORY_SPACE, |
| 165 | CATEGORY_NOT_SPACE: CATEGORY_NOT_SPACE, |
Jeremy Hylton | b1aa195 | 2000-06-01 17:39:12 +0000 | [diff] [blame] | 166 | CATEGORY_WORD: CATEGORY_LOC_WORD, |
| 167 | CATEGORY_NOT_WORD: CATEGORY_LOC_NOT_WORD, |
Fredrik Lundh | 436c3d58 | 2000-06-29 08:58:44 +0000 | [diff] [blame] | 168 | CATEGORY_LINEBREAK: CATEGORY_LINEBREAK, |
| 169 | CATEGORY_NOT_LINEBREAK: CATEGORY_NOT_LINEBREAK |
| 170 | } |
| 171 | |
| 172 | CH_UNICODE = { |
| 173 | CATEGORY_DIGIT: CATEGORY_UNI_DIGIT, |
| 174 | CATEGORY_NOT_DIGIT: CATEGORY_UNI_NOT_DIGIT, |
| 175 | CATEGORY_SPACE: CATEGORY_UNI_SPACE, |
| 176 | CATEGORY_NOT_SPACE: CATEGORY_UNI_NOT_SPACE, |
| 177 | CATEGORY_WORD: CATEGORY_UNI_WORD, |
| 178 | CATEGORY_NOT_WORD: CATEGORY_UNI_NOT_WORD, |
| 179 | CATEGORY_LINEBREAK: CATEGORY_UNI_LINEBREAK, |
| 180 | CATEGORY_NOT_LINEBREAK: CATEGORY_UNI_NOT_LINEBREAK |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 181 | } |
| 182 | |
Jeremy Hylton | b1aa195 | 2000-06-01 17:39:12 +0000 | [diff] [blame] | 183 | # flags |
Fredrik Lundh | 3562f11 | 2000-07-02 12:00:07 +0000 | [diff] [blame] | 184 | SRE_FLAG_TEMPLATE = 1 # template mode (disable backtracking) |
| 185 | SRE_FLAG_IGNORECASE = 2 # case insensitive |
Fredrik Lundh | 8a3ebf8 | 2000-07-23 21:46:17 +0000 | [diff] [blame] | 186 | SRE_FLAG_LOCALE = 4 # honour system locale |
Fredrik Lundh | 3562f11 | 2000-07-02 12:00:07 +0000 | [diff] [blame] | 187 | SRE_FLAG_MULTILINE = 8 # treat target as multiline string |
| 188 | SRE_FLAG_DOTALL = 16 # treat target as a single string |
Antoine Pitrou | fd03645 | 2008-08-19 17:56:33 +0000 | [diff] [blame] | 189 | SRE_FLAG_UNICODE = 32 # use unicode "locale" |
Fredrik Lundh | 3562f11 | 2000-07-02 12:00:07 +0000 | [diff] [blame] | 190 | SRE_FLAG_VERBOSE = 64 # ignore whitespace and comments |
Fredrik Lundh | 770617b | 2001-01-14 15:06:11 +0000 | [diff] [blame] | 191 | SRE_FLAG_DEBUG = 128 # debugging |
Antoine Pitrou | fd03645 | 2008-08-19 17:56:33 +0000 | [diff] [blame] | 192 | SRE_FLAG_ASCII = 256 # use ascii "locale" |
Fredrik Lundh | 3562f11 | 2000-07-02 12:00:07 +0000 | [diff] [blame] | 193 | |
| 194 | # flags for INFO primitive |
| 195 | SRE_INFO_PREFIX = 1 # has prefix |
| 196 | SRE_INFO_LITERAL = 2 # entire pattern is literal (given by prefix) |
| 197 | SRE_INFO_CHARSET = 4 # pattern starts with character from given set |
Jeremy Hylton | b1aa195 | 2000-06-01 17:39:12 +0000 | [diff] [blame] | 198 | |
Guido van Rossum | 7627c0d | 2000-03-31 14:58:54 +0000 | [diff] [blame] | 199 | if __name__ == "__main__": |
Jeremy Hylton | b1aa195 | 2000-06-01 17:39:12 +0000 | [diff] [blame] | 200 | def dump(f, d, prefix): |
Serhiy Storchaka | c7f7d38 | 2014-11-09 20:48:36 +0200 | [diff] [blame] | 201 | items = sorted(d) |
| 202 | for item in items: |
| 203 | f.write("#define %s_%s %d\n" % (prefix, item, item)) |
Serhiy Storchaka | 46ba6c8 | 2015-04-04 11:01:02 +0300 | [diff] [blame] | 204 | with open("sre_constants.h", "w") as f: |
| 205 | f.write("""\ |
Fredrik Lundh | 43b3b49 | 2000-06-30 10:41:31 +0000 | [diff] [blame] | 206 | /* |
| 207 | * Secret Labs' Regular Expression Engine |
| 208 | * |
| 209 | * regular expression matching engine |
| 210 | * |
| 211 | * NOTE: This file is generated by sre_constants.py. If you need |
| 212 | * to change anything in here, edit sre_constants.py and run it. |
| 213 | * |
Fredrik Lundh | 770617b | 2001-01-14 15:06:11 +0000 | [diff] [blame] | 214 | * Copyright (c) 1997-2001 by Secret Labs AB. All rights reserved. |
Fredrik Lundh | 43b3b49 | 2000-06-30 10:41:31 +0000 | [diff] [blame] | 215 | * |
| 216 | * See the _sre.c file for information on usage and redistribution. |
| 217 | */ |
| 218 | |
| 219 | """) |
| 220 | |
Serhiy Storchaka | 46ba6c8 | 2015-04-04 11:01:02 +0300 | [diff] [blame] | 221 | f.write("#define SRE_MAGIC %d\n" % MAGIC) |
Fredrik Lundh | b35ffc0 | 2001-01-15 12:46:09 +0000 | [diff] [blame] | 222 | |
Serhiy Storchaka | 46ba6c8 | 2015-04-04 11:01:02 +0300 | [diff] [blame] | 223 | dump(f, OPCODES, "SRE_OP") |
| 224 | dump(f, ATCODES, "SRE") |
| 225 | dump(f, CHCODES, "SRE") |
Fredrik Lundh | 3562f11 | 2000-07-02 12:00:07 +0000 | [diff] [blame] | 226 | |
Serhiy Storchaka | 46ba6c8 | 2015-04-04 11:01:02 +0300 | [diff] [blame] | 227 | f.write("#define SRE_FLAG_TEMPLATE %d\n" % SRE_FLAG_TEMPLATE) |
| 228 | f.write("#define SRE_FLAG_IGNORECASE %d\n" % SRE_FLAG_IGNORECASE) |
| 229 | f.write("#define SRE_FLAG_LOCALE %d\n" % SRE_FLAG_LOCALE) |
| 230 | f.write("#define SRE_FLAG_MULTILINE %d\n" % SRE_FLAG_MULTILINE) |
| 231 | f.write("#define SRE_FLAG_DOTALL %d\n" % SRE_FLAG_DOTALL) |
| 232 | f.write("#define SRE_FLAG_UNICODE %d\n" % SRE_FLAG_UNICODE) |
| 233 | f.write("#define SRE_FLAG_VERBOSE %d\n" % SRE_FLAG_VERBOSE) |
| 234 | f.write("#define SRE_FLAG_DEBUG %d\n" % SRE_FLAG_DEBUG) |
| 235 | f.write("#define SRE_FLAG_ASCII %d\n" % SRE_FLAG_ASCII) |
Fredrik Lundh | 3562f11 | 2000-07-02 12:00:07 +0000 | [diff] [blame] | 236 | |
Serhiy Storchaka | 46ba6c8 | 2015-04-04 11:01:02 +0300 | [diff] [blame] | 237 | f.write("#define SRE_INFO_PREFIX %d\n" % SRE_INFO_PREFIX) |
| 238 | f.write("#define SRE_INFO_LITERAL %d\n" % SRE_INFO_LITERAL) |
| 239 | f.write("#define SRE_INFO_CHARSET %d\n" % SRE_INFO_CHARSET) |
Fredrik Lundh | 3562f11 | 2000-07-02 12:00:07 +0000 | [diff] [blame] | 240 | |
Guido van Rossum | be19ed7 | 2007-02-09 05:37:30 +0000 | [diff] [blame] | 241 | print("done") |