1# 2# Secret Labs' Regular Expression Engine 3# 4# various symbols used by the regular expression engine. 5# run this script to update the _sre include files! 6# 7# Copyright (c) 1998-2001 by Secret Labs AB. All rights reserved. 8# 9# See the __init__.py file for information on usage and redistribution. 10# 11 12"""Internal support module for sre""" 13 14# update when constants are added or removed 15 16MAGIC = 20220615 17 18from _sre import MAXREPEAT, MAXGROUPS 19 20# SRE standard exception (access as sre.error) 21# should this really be here? 22 23class error(Exception): 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 35 __module__ = 're' 36 37 def __init__(self, msg, pattern=None, pos=None): 38 self.msg = msg 39 self.pattern = pattern 40 self.pos = pos 41 if pattern is not None and pos is not None: 42 msg = '%s at position %d' % (msg, pos) 43 if isinstance(pattern, str): 44 newline = '\n' 45 else: 46 newline = b'\n' 47 self.lineno = pattern.count(newline, 0, pos) + 1 48 self.colno = pos - pattern.rfind(newline, 0, pos) 49 if newline in pattern: 50 msg = '%s (line %d, column %d)' % (msg, self.lineno, self.colno) 51 else: 52 self.lineno = self.colno = None 53 super().__init__(msg) 54 55 56class _NamedIntConstant(int): 57 def __new__(cls, value, name): 58 self = super(_NamedIntConstant, cls).__new__(cls, value) 59 self.name = name 60 return self 61 62 def __repr__(self): 63 return self.name 64 65 __reduce__ = None 66 67MAXREPEAT = _NamedIntConstant(MAXREPEAT, 'MAXREPEAT') 68 69def _makecodes(*names): 70 items = [_NamedIntConstant(i, name) for i, name in enumerate(names)] 71 globals().update({item.name: item for item in items}) 72 return items 73 74# operators 75OPCODES = _makecodes( 76 # failure=0 success=1 (just because it looks better that way :-) 77 'FAILURE', 'SUCCESS', 78 79 'ANY', 'ANY_ALL', 80 'ASSERT', 'ASSERT_NOT', 81 'AT', 82 'BRANCH', 83 'CATEGORY', 84 'CHARSET', 'BIGCHARSET', 85 'GROUPREF', 'GROUPREF_EXISTS', 86 'IN', 87 'INFO', 88 'JUMP', 89 'LITERAL', 90 'MARK', 91 'MAX_UNTIL', 92 'MIN_UNTIL', 93 'NOT_LITERAL', 94 'NEGATE', 95 'RANGE', 96 'REPEAT', 97 'REPEAT_ONE', 98 'SUBPATTERN', 99 'MIN_REPEAT_ONE', 100 'ATOMIC_GROUP', 101 'POSSESSIVE_REPEAT', 102 'POSSESSIVE_REPEAT_ONE', 103 104 'GROUPREF_IGNORE', 105 'IN_IGNORE', 106 'LITERAL_IGNORE', 107 'NOT_LITERAL_IGNORE', 108 109 'GROUPREF_LOC_IGNORE', 110 'IN_LOC_IGNORE', 111 'LITERAL_LOC_IGNORE', 112 'NOT_LITERAL_LOC_IGNORE', 113 114 'GROUPREF_UNI_IGNORE', 115 'IN_UNI_IGNORE', 116 'LITERAL_UNI_IGNORE', 117 'NOT_LITERAL_UNI_IGNORE', 118 'RANGE_UNI_IGNORE', 119 120 # The following opcodes are only occurred in the parser output, 121 # but not in the compiled code. 122 'MIN_REPEAT', 'MAX_REPEAT', 123) 124del OPCODES[-2:] # remove MIN_REPEAT and MAX_REPEAT 125 126# positions 127ATCODES = _makecodes( 128 'AT_BEGINNING', 'AT_BEGINNING_LINE', 'AT_BEGINNING_STRING', 129 'AT_BOUNDARY', 'AT_NON_BOUNDARY', 130 'AT_END', 'AT_END_LINE', 'AT_END_STRING', 131 132 'AT_LOC_BOUNDARY', 'AT_LOC_NON_BOUNDARY', 133 134 'AT_UNI_BOUNDARY', 'AT_UNI_NON_BOUNDARY', 135) 136 137# categories 138CHCODES = _makecodes( 139 'CATEGORY_DIGIT', 'CATEGORY_NOT_DIGIT', 140 'CATEGORY_SPACE', 'CATEGORY_NOT_SPACE', 141 'CATEGORY_WORD', 'CATEGORY_NOT_WORD', 142 'CATEGORY_LINEBREAK', 'CATEGORY_NOT_LINEBREAK', 143 144 'CATEGORY_LOC_WORD', 'CATEGORY_LOC_NOT_WORD', 145 146 'CATEGORY_UNI_DIGIT', 'CATEGORY_UNI_NOT_DIGIT', 147 'CATEGORY_UNI_SPACE', 'CATEGORY_UNI_NOT_SPACE', 148 'CATEGORY_UNI_WORD', 'CATEGORY_UNI_NOT_WORD', 149 'CATEGORY_UNI_LINEBREAK', 'CATEGORY_UNI_NOT_LINEBREAK', 150) 151 152 153# replacement operations for "ignore case" mode 154OP_IGNORE = { 155 LITERAL: LITERAL_IGNORE, 156 NOT_LITERAL: NOT_LITERAL_IGNORE, 157} 158 159OP_LOCALE_IGNORE = { 160 LITERAL: LITERAL_LOC_IGNORE, 161 NOT_LITERAL: NOT_LITERAL_LOC_IGNORE, 162} 163 164OP_UNICODE_IGNORE = { 165 LITERAL: LITERAL_UNI_IGNORE, 166 NOT_LITERAL: NOT_LITERAL_UNI_IGNORE, 167} 168 169AT_MULTILINE = { 170 AT_BEGINNING: AT_BEGINNING_LINE, 171 AT_END: AT_END_LINE 172} 173 174AT_LOCALE = { 175 AT_BOUNDARY: AT_LOC_BOUNDARY, 176 AT_NON_BOUNDARY: AT_LOC_NON_BOUNDARY 177} 178 179AT_UNICODE = { 180 AT_BOUNDARY: AT_UNI_BOUNDARY, 181 AT_NON_BOUNDARY: AT_UNI_NON_BOUNDARY 182} 183 184CH_LOCALE = { 185 CATEGORY_DIGIT: CATEGORY_DIGIT, 186 CATEGORY_NOT_DIGIT: CATEGORY_NOT_DIGIT, 187 CATEGORY_SPACE: CATEGORY_SPACE, 188 CATEGORY_NOT_SPACE: CATEGORY_NOT_SPACE, 189 CATEGORY_WORD: CATEGORY_LOC_WORD, 190 CATEGORY_NOT_WORD: CATEGORY_LOC_NOT_WORD, 191 CATEGORY_LINEBREAK: CATEGORY_LINEBREAK, 192 CATEGORY_NOT_LINEBREAK: CATEGORY_NOT_LINEBREAK 193} 194 195CH_UNICODE = { 196 CATEGORY_DIGIT: CATEGORY_UNI_DIGIT, 197 CATEGORY_NOT_DIGIT: CATEGORY_UNI_NOT_DIGIT, 198 CATEGORY_SPACE: CATEGORY_UNI_SPACE, 199 CATEGORY_NOT_SPACE: CATEGORY_UNI_NOT_SPACE, 200 CATEGORY_WORD: CATEGORY_UNI_WORD, 201 CATEGORY_NOT_WORD: CATEGORY_UNI_NOT_WORD, 202 CATEGORY_LINEBREAK: CATEGORY_UNI_LINEBREAK, 203 CATEGORY_NOT_LINEBREAK: CATEGORY_UNI_NOT_LINEBREAK 204} 205 206# flags 207SRE_FLAG_TEMPLATE = 1 # template mode (unknown purpose, deprecated) 208SRE_FLAG_IGNORECASE = 2 # case insensitive 209SRE_FLAG_LOCALE = 4 # honour system locale 210SRE_FLAG_MULTILINE = 8 # treat target as multiline string 211SRE_FLAG_DOTALL = 16 # treat target as a single string 212SRE_FLAG_UNICODE = 32 # use unicode "locale" 213SRE_FLAG_VERBOSE = 64 # ignore whitespace and comments 214SRE_FLAG_DEBUG = 128 # debugging 215SRE_FLAG_ASCII = 256 # use ascii "locale" 216 217# flags for INFO primitive 218SRE_INFO_PREFIX = 1 # has prefix 219SRE_INFO_LITERAL = 2 # entire pattern is literal (given by prefix) 220SRE_INFO_CHARSET = 4 # pattern starts with character from given set 221