1 | n/a | # |
---|
2 | n/a | # Secret Labs' Regular Expression Engine |
---|
3 | n/a | # |
---|
4 | n/a | # various symbols used by the regular expression engine. |
---|
5 | n/a | # run this script to update the _sre include files! |
---|
6 | n/a | # |
---|
7 | n/a | # Copyright (c) 1998-2001 by Secret Labs AB. All rights reserved. |
---|
8 | n/a | # |
---|
9 | n/a | # See the sre.py file for information on usage and redistribution. |
---|
10 | n/a | # |
---|
11 | n/a | |
---|
12 | n/a | """Internal support module for sre""" |
---|
13 | n/a | |
---|
14 | n/a | # update when constants are added or removed |
---|
15 | n/a | |
---|
16 | n/a | MAGIC = 20140917 |
---|
17 | n/a | |
---|
18 | n/a | from _sre import MAXREPEAT, MAXGROUPS |
---|
19 | n/a | |
---|
20 | n/a | # SRE standard exception (access as sre.error) |
---|
21 | n/a | # should this really be here? |
---|
22 | n/a | |
---|
23 | n/a | class error(Exception): |
---|
24 | n/a | def __init__(self, msg, pattern=None, pos=None): |
---|
25 | n/a | self.msg = msg |
---|
26 | n/a | self.pattern = pattern |
---|
27 | n/a | self.pos = pos |
---|
28 | n/a | if pattern is not None and pos is not None: |
---|
29 | n/a | msg = '%s at position %d' % (msg, pos) |
---|
30 | n/a | if isinstance(pattern, str): |
---|
31 | n/a | newline = '\n' |
---|
32 | n/a | else: |
---|
33 | n/a | newline = b'\n' |
---|
34 | n/a | self.lineno = pattern.count(newline, 0, pos) + 1 |
---|
35 | n/a | self.colno = pos - pattern.rfind(newline, 0, pos) |
---|
36 | n/a | if newline in pattern: |
---|
37 | n/a | msg = '%s (line %d, column %d)' % (msg, self.lineno, self.colno) |
---|
38 | n/a | else: |
---|
39 | n/a | self.lineno = self.colno = None |
---|
40 | n/a | super().__init__(msg) |
---|
41 | n/a | |
---|
42 | n/a | |
---|
43 | n/a | class _NamedIntConstant(int): |
---|
44 | n/a | def __new__(cls, value, name): |
---|
45 | n/a | self = super(_NamedIntConstant, cls).__new__(cls, value) |
---|
46 | n/a | self.name = name |
---|
47 | n/a | return self |
---|
48 | n/a | |
---|
49 | n/a | def __str__(self): |
---|
50 | n/a | return self.name |
---|
51 | n/a | |
---|
52 | n/a | __repr__ = __str__ |
---|
53 | n/a | |
---|
54 | n/a | MAXREPEAT = _NamedIntConstant(MAXREPEAT, 'MAXREPEAT') |
---|
55 | n/a | |
---|
56 | n/a | def _makecodes(names): |
---|
57 | n/a | names = names.strip().split() |
---|
58 | n/a | items = [_NamedIntConstant(i, name) for i, name in enumerate(names)] |
---|
59 | n/a | globals().update({item.name: item for item in items}) |
---|
60 | n/a | return items |
---|
61 | n/a | |
---|
62 | n/a | # operators |
---|
63 | n/a | # failure=0 success=1 (just because it looks better that way :-) |
---|
64 | n/a | OPCODES = _makecodes(""" |
---|
65 | n/a | FAILURE SUCCESS |
---|
66 | n/a | |
---|
67 | n/a | ANY ANY_ALL |
---|
68 | n/a | ASSERT ASSERT_NOT |
---|
69 | n/a | AT |
---|
70 | n/a | BRANCH |
---|
71 | n/a | CALL |
---|
72 | n/a | CATEGORY |
---|
73 | n/a | CHARSET BIGCHARSET |
---|
74 | n/a | GROUPREF GROUPREF_EXISTS GROUPREF_IGNORE |
---|
75 | n/a | IN IN_IGNORE |
---|
76 | n/a | INFO |
---|
77 | n/a | JUMP |
---|
78 | n/a | LITERAL LITERAL_IGNORE |
---|
79 | n/a | MARK |
---|
80 | n/a | MAX_UNTIL |
---|
81 | n/a | MIN_UNTIL |
---|
82 | n/a | NOT_LITERAL NOT_LITERAL_IGNORE |
---|
83 | n/a | NEGATE |
---|
84 | n/a | RANGE |
---|
85 | n/a | REPEAT |
---|
86 | n/a | REPEAT_ONE |
---|
87 | n/a | SUBPATTERN |
---|
88 | n/a | MIN_REPEAT_ONE |
---|
89 | n/a | RANGE_IGNORE |
---|
90 | n/a | |
---|
91 | n/a | MIN_REPEAT MAX_REPEAT |
---|
92 | n/a | """) |
---|
93 | n/a | del OPCODES[-2:] # remove MIN_REPEAT and MAX_REPEAT |
---|
94 | n/a | |
---|
95 | n/a | # positions |
---|
96 | n/a | ATCODES = _makecodes(""" |
---|
97 | n/a | AT_BEGINNING AT_BEGINNING_LINE AT_BEGINNING_STRING |
---|
98 | n/a | AT_BOUNDARY AT_NON_BOUNDARY |
---|
99 | n/a | AT_END AT_END_LINE AT_END_STRING |
---|
100 | n/a | AT_LOC_BOUNDARY AT_LOC_NON_BOUNDARY |
---|
101 | n/a | AT_UNI_BOUNDARY AT_UNI_NON_BOUNDARY |
---|
102 | n/a | """) |
---|
103 | n/a | |
---|
104 | n/a | # categories |
---|
105 | n/a | CHCODES = _makecodes(""" |
---|
106 | n/a | CATEGORY_DIGIT CATEGORY_NOT_DIGIT |
---|
107 | n/a | CATEGORY_SPACE CATEGORY_NOT_SPACE |
---|
108 | n/a | CATEGORY_WORD CATEGORY_NOT_WORD |
---|
109 | n/a | CATEGORY_LINEBREAK CATEGORY_NOT_LINEBREAK |
---|
110 | n/a | CATEGORY_LOC_WORD CATEGORY_LOC_NOT_WORD |
---|
111 | n/a | CATEGORY_UNI_DIGIT CATEGORY_UNI_NOT_DIGIT |
---|
112 | n/a | CATEGORY_UNI_SPACE CATEGORY_UNI_NOT_SPACE |
---|
113 | n/a | CATEGORY_UNI_WORD CATEGORY_UNI_NOT_WORD |
---|
114 | n/a | CATEGORY_UNI_LINEBREAK CATEGORY_UNI_NOT_LINEBREAK |
---|
115 | n/a | """) |
---|
116 | n/a | |
---|
117 | n/a | |
---|
118 | n/a | # replacement operations for "ignore case" mode |
---|
119 | n/a | OP_IGNORE = { |
---|
120 | n/a | GROUPREF: GROUPREF_IGNORE, |
---|
121 | n/a | IN: IN_IGNORE, |
---|
122 | n/a | LITERAL: LITERAL_IGNORE, |
---|
123 | n/a | NOT_LITERAL: NOT_LITERAL_IGNORE, |
---|
124 | n/a | RANGE: RANGE_IGNORE, |
---|
125 | n/a | } |
---|
126 | n/a | |
---|
127 | n/a | AT_MULTILINE = { |
---|
128 | n/a | AT_BEGINNING: AT_BEGINNING_LINE, |
---|
129 | n/a | AT_END: AT_END_LINE |
---|
130 | n/a | } |
---|
131 | n/a | |
---|
132 | n/a | AT_LOCALE = { |
---|
133 | n/a | AT_BOUNDARY: AT_LOC_BOUNDARY, |
---|
134 | n/a | AT_NON_BOUNDARY: AT_LOC_NON_BOUNDARY |
---|
135 | n/a | } |
---|
136 | n/a | |
---|
137 | n/a | AT_UNICODE = { |
---|
138 | n/a | AT_BOUNDARY: AT_UNI_BOUNDARY, |
---|
139 | n/a | AT_NON_BOUNDARY: AT_UNI_NON_BOUNDARY |
---|
140 | n/a | } |
---|
141 | n/a | |
---|
142 | n/a | CH_LOCALE = { |
---|
143 | n/a | CATEGORY_DIGIT: CATEGORY_DIGIT, |
---|
144 | n/a | CATEGORY_NOT_DIGIT: CATEGORY_NOT_DIGIT, |
---|
145 | n/a | CATEGORY_SPACE: CATEGORY_SPACE, |
---|
146 | n/a | CATEGORY_NOT_SPACE: CATEGORY_NOT_SPACE, |
---|
147 | n/a | CATEGORY_WORD: CATEGORY_LOC_WORD, |
---|
148 | n/a | CATEGORY_NOT_WORD: CATEGORY_LOC_NOT_WORD, |
---|
149 | n/a | CATEGORY_LINEBREAK: CATEGORY_LINEBREAK, |
---|
150 | n/a | CATEGORY_NOT_LINEBREAK: CATEGORY_NOT_LINEBREAK |
---|
151 | n/a | } |
---|
152 | n/a | |
---|
153 | n/a | CH_UNICODE = { |
---|
154 | n/a | CATEGORY_DIGIT: CATEGORY_UNI_DIGIT, |
---|
155 | n/a | CATEGORY_NOT_DIGIT: CATEGORY_UNI_NOT_DIGIT, |
---|
156 | n/a | CATEGORY_SPACE: CATEGORY_UNI_SPACE, |
---|
157 | n/a | CATEGORY_NOT_SPACE: CATEGORY_UNI_NOT_SPACE, |
---|
158 | n/a | CATEGORY_WORD: CATEGORY_UNI_WORD, |
---|
159 | n/a | CATEGORY_NOT_WORD: CATEGORY_UNI_NOT_WORD, |
---|
160 | n/a | CATEGORY_LINEBREAK: CATEGORY_UNI_LINEBREAK, |
---|
161 | n/a | CATEGORY_NOT_LINEBREAK: CATEGORY_UNI_NOT_LINEBREAK |
---|
162 | n/a | } |
---|
163 | n/a | |
---|
164 | n/a | # flags |
---|
165 | n/a | SRE_FLAG_TEMPLATE = 1 # template mode (disable backtracking) |
---|
166 | n/a | SRE_FLAG_IGNORECASE = 2 # case insensitive |
---|
167 | n/a | SRE_FLAG_LOCALE = 4 # honour system locale |
---|
168 | n/a | SRE_FLAG_MULTILINE = 8 # treat target as multiline string |
---|
169 | n/a | SRE_FLAG_DOTALL = 16 # treat target as a single string |
---|
170 | n/a | SRE_FLAG_UNICODE = 32 # use unicode "locale" |
---|
171 | n/a | SRE_FLAG_VERBOSE = 64 # ignore whitespace and comments |
---|
172 | n/a | SRE_FLAG_DEBUG = 128 # debugging |
---|
173 | n/a | SRE_FLAG_ASCII = 256 # use ascii "locale" |
---|
174 | n/a | |
---|
175 | n/a | # flags for INFO primitive |
---|
176 | n/a | SRE_INFO_PREFIX = 1 # has prefix |
---|
177 | n/a | SRE_INFO_LITERAL = 2 # entire pattern is literal (given by prefix) |
---|
178 | n/a | SRE_INFO_CHARSET = 4 # pattern starts with character from given set |
---|
179 | n/a | |
---|
180 | n/a | if __name__ == "__main__": |
---|
181 | n/a | def dump(f, d, prefix): |
---|
182 | n/a | items = sorted(d) |
---|
183 | n/a | for item in items: |
---|
184 | n/a | f.write("#define %s_%s %d\n" % (prefix, item, item)) |
---|
185 | n/a | with open("sre_constants.h", "w") as f: |
---|
186 | n/a | f.write("""\ |
---|
187 | n/a | /* |
---|
188 | n/a | * Secret Labs' Regular Expression Engine |
---|
189 | n/a | * |
---|
190 | n/a | * regular expression matching engine |
---|
191 | n/a | * |
---|
192 | n/a | * NOTE: This file is generated by sre_constants.py. If you need |
---|
193 | n/a | * to change anything in here, edit sre_constants.py and run it. |
---|
194 | n/a | * |
---|
195 | n/a | * Copyright (c) 1997-2001 by Secret Labs AB. All rights reserved. |
---|
196 | n/a | * |
---|
197 | n/a | * See the _sre.c file for information on usage and redistribution. |
---|
198 | n/a | */ |
---|
199 | n/a | |
---|
200 | n/a | """) |
---|
201 | n/a | |
---|
202 | n/a | f.write("#define SRE_MAGIC %d\n" % MAGIC) |
---|
203 | n/a | |
---|
204 | n/a | dump(f, OPCODES, "SRE_OP") |
---|
205 | n/a | dump(f, ATCODES, "SRE") |
---|
206 | n/a | dump(f, CHCODES, "SRE") |
---|
207 | n/a | |
---|
208 | n/a | f.write("#define SRE_FLAG_TEMPLATE %d\n" % SRE_FLAG_TEMPLATE) |
---|
209 | n/a | f.write("#define SRE_FLAG_IGNORECASE %d\n" % SRE_FLAG_IGNORECASE) |
---|
210 | n/a | f.write("#define SRE_FLAG_LOCALE %d\n" % SRE_FLAG_LOCALE) |
---|
211 | n/a | f.write("#define SRE_FLAG_MULTILINE %d\n" % SRE_FLAG_MULTILINE) |
---|
212 | n/a | f.write("#define SRE_FLAG_DOTALL %d\n" % SRE_FLAG_DOTALL) |
---|
213 | n/a | f.write("#define SRE_FLAG_UNICODE %d\n" % SRE_FLAG_UNICODE) |
---|
214 | n/a | f.write("#define SRE_FLAG_VERBOSE %d\n" % SRE_FLAG_VERBOSE) |
---|
215 | n/a | f.write("#define SRE_FLAG_DEBUG %d\n" % SRE_FLAG_DEBUG) |
---|
216 | n/a | f.write("#define SRE_FLAG_ASCII %d\n" % SRE_FLAG_ASCII) |
---|
217 | n/a | |
---|
218 | n/a | f.write("#define SRE_INFO_PREFIX %d\n" % SRE_INFO_PREFIX) |
---|
219 | n/a | f.write("#define SRE_INFO_LITERAL %d\n" % SRE_INFO_LITERAL) |
---|
220 | n/a | f.write("#define SRE_INFO_CHARSET %d\n" % SRE_INFO_CHARSET) |
---|
221 | n/a | |
---|
222 | n/a | print("done") |
---|