]>
Commit | Line | Data |
---|---|---|
3257aa99 DM |
1 | #\r |
2 | # Secret Labs' Regular Expression Engine\r | |
3 | #\r | |
4 | # various symbols used by the regular expression engine.\r | |
5 | # run this script to update the _sre include files!\r | |
6 | #\r | |
7 | # Copyright (c) 1998-2001 by Secret Labs AB. All rights reserved.\r | |
8 | #\r | |
9 | # See the sre.py file for information on usage and redistribution.\r | |
10 | #\r | |
11 | \r | |
12 | """Internal support module for sre"""\r | |
13 | \r | |
14 | # update when constants are added or removed\r | |
15 | \r | |
16 | MAGIC = 20031017\r | |
17 | \r | |
18 | try:\r | |
19 | from _sre import MAXREPEAT\r | |
20 | except ImportError:\r | |
21 | import _sre\r | |
22 | MAXREPEAT = _sre.MAXREPEAT = 65535\r | |
23 | \r | |
24 | # SRE standard exception (access as sre.error)\r | |
25 | # should this really be here?\r | |
26 | \r | |
27 | class error(Exception):\r | |
28 | pass\r | |
29 | \r | |
30 | # operators\r | |
31 | \r | |
32 | FAILURE = "failure"\r | |
33 | SUCCESS = "success"\r | |
34 | \r | |
35 | ANY = "any"\r | |
36 | ANY_ALL = "any_all"\r | |
37 | ASSERT = "assert"\r | |
38 | ASSERT_NOT = "assert_not"\r | |
39 | AT = "at"\r | |
40 | BIGCHARSET = "bigcharset"\r | |
41 | BRANCH = "branch"\r | |
42 | CALL = "call"\r | |
43 | CATEGORY = "category"\r | |
44 | CHARSET = "charset"\r | |
45 | GROUPREF = "groupref"\r | |
46 | GROUPREF_IGNORE = "groupref_ignore"\r | |
47 | GROUPREF_EXISTS = "groupref_exists"\r | |
48 | IN = "in"\r | |
49 | IN_IGNORE = "in_ignore"\r | |
50 | INFO = "info"\r | |
51 | JUMP = "jump"\r | |
52 | LITERAL = "literal"\r | |
53 | LITERAL_IGNORE = "literal_ignore"\r | |
54 | MARK = "mark"\r | |
55 | MAX_REPEAT = "max_repeat"\r | |
56 | MAX_UNTIL = "max_until"\r | |
57 | MIN_REPEAT = "min_repeat"\r | |
58 | MIN_UNTIL = "min_until"\r | |
59 | NEGATE = "negate"\r | |
60 | NOT_LITERAL = "not_literal"\r | |
61 | NOT_LITERAL_IGNORE = "not_literal_ignore"\r | |
62 | RANGE = "range"\r | |
63 | REPEAT = "repeat"\r | |
64 | REPEAT_ONE = "repeat_one"\r | |
65 | SUBPATTERN = "subpattern"\r | |
66 | MIN_REPEAT_ONE = "min_repeat_one"\r | |
67 | \r | |
68 | # positions\r | |
69 | AT_BEGINNING = "at_beginning"\r | |
70 | AT_BEGINNING_LINE = "at_beginning_line"\r | |
71 | AT_BEGINNING_STRING = "at_beginning_string"\r | |
72 | AT_BOUNDARY = "at_boundary"\r | |
73 | AT_NON_BOUNDARY = "at_non_boundary"\r | |
74 | AT_END = "at_end"\r | |
75 | AT_END_LINE = "at_end_line"\r | |
76 | AT_END_STRING = "at_end_string"\r | |
77 | AT_LOC_BOUNDARY = "at_loc_boundary"\r | |
78 | AT_LOC_NON_BOUNDARY = "at_loc_non_boundary"\r | |
79 | AT_UNI_BOUNDARY = "at_uni_boundary"\r | |
80 | AT_UNI_NON_BOUNDARY = "at_uni_non_boundary"\r | |
81 | \r | |
82 | # categories\r | |
83 | CATEGORY_DIGIT = "category_digit"\r | |
84 | CATEGORY_NOT_DIGIT = "category_not_digit"\r | |
85 | CATEGORY_SPACE = "category_space"\r | |
86 | CATEGORY_NOT_SPACE = "category_not_space"\r | |
87 | CATEGORY_WORD = "category_word"\r | |
88 | CATEGORY_NOT_WORD = "category_not_word"\r | |
89 | CATEGORY_LINEBREAK = "category_linebreak"\r | |
90 | CATEGORY_NOT_LINEBREAK = "category_not_linebreak"\r | |
91 | CATEGORY_LOC_WORD = "category_loc_word"\r | |
92 | CATEGORY_LOC_NOT_WORD = "category_loc_not_word"\r | |
93 | CATEGORY_UNI_DIGIT = "category_uni_digit"\r | |
94 | CATEGORY_UNI_NOT_DIGIT = "category_uni_not_digit"\r | |
95 | CATEGORY_UNI_SPACE = "category_uni_space"\r | |
96 | CATEGORY_UNI_NOT_SPACE = "category_uni_not_space"\r | |
97 | CATEGORY_UNI_WORD = "category_uni_word"\r | |
98 | CATEGORY_UNI_NOT_WORD = "category_uni_not_word"\r | |
99 | CATEGORY_UNI_LINEBREAK = "category_uni_linebreak"\r | |
100 | CATEGORY_UNI_NOT_LINEBREAK = "category_uni_not_linebreak"\r | |
101 | \r | |
102 | OPCODES = [\r | |
103 | \r | |
104 | # failure=0 success=1 (just because it looks better that way :-)\r | |
105 | FAILURE, SUCCESS,\r | |
106 | \r | |
107 | ANY, ANY_ALL,\r | |
108 | ASSERT, ASSERT_NOT,\r | |
109 | AT,\r | |
110 | BRANCH,\r | |
111 | CALL,\r | |
112 | CATEGORY,\r | |
113 | CHARSET, BIGCHARSET,\r | |
114 | GROUPREF, GROUPREF_EXISTS, GROUPREF_IGNORE,\r | |
115 | IN, IN_IGNORE,\r | |
116 | INFO,\r | |
117 | JUMP,\r | |
118 | LITERAL, LITERAL_IGNORE,\r | |
119 | MARK,\r | |
120 | MAX_UNTIL,\r | |
121 | MIN_UNTIL,\r | |
122 | NOT_LITERAL, NOT_LITERAL_IGNORE,\r | |
123 | NEGATE,\r | |
124 | RANGE,\r | |
125 | REPEAT,\r | |
126 | REPEAT_ONE,\r | |
127 | SUBPATTERN,\r | |
128 | MIN_REPEAT_ONE\r | |
129 | \r | |
130 | ]\r | |
131 | \r | |
132 | ATCODES = [\r | |
133 | AT_BEGINNING, AT_BEGINNING_LINE, AT_BEGINNING_STRING, AT_BOUNDARY,\r | |
134 | AT_NON_BOUNDARY, AT_END, AT_END_LINE, AT_END_STRING,\r | |
135 | AT_LOC_BOUNDARY, AT_LOC_NON_BOUNDARY, AT_UNI_BOUNDARY,\r | |
136 | AT_UNI_NON_BOUNDARY\r | |
137 | ]\r | |
138 | \r | |
139 | CHCODES = [\r | |
140 | CATEGORY_DIGIT, CATEGORY_NOT_DIGIT, CATEGORY_SPACE,\r | |
141 | CATEGORY_NOT_SPACE, CATEGORY_WORD, CATEGORY_NOT_WORD,\r | |
142 | CATEGORY_LINEBREAK, CATEGORY_NOT_LINEBREAK, CATEGORY_LOC_WORD,\r | |
143 | CATEGORY_LOC_NOT_WORD, CATEGORY_UNI_DIGIT, CATEGORY_UNI_NOT_DIGIT,\r | |
144 | CATEGORY_UNI_SPACE, CATEGORY_UNI_NOT_SPACE, CATEGORY_UNI_WORD,\r | |
145 | CATEGORY_UNI_NOT_WORD, CATEGORY_UNI_LINEBREAK,\r | |
146 | CATEGORY_UNI_NOT_LINEBREAK\r | |
147 | ]\r | |
148 | \r | |
149 | def makedict(list):\r | |
150 | d = {}\r | |
151 | i = 0\r | |
152 | for item in list:\r | |
153 | d[item] = i\r | |
154 | i = i + 1\r | |
155 | return d\r | |
156 | \r | |
157 | OPCODES = makedict(OPCODES)\r | |
158 | ATCODES = makedict(ATCODES)\r | |
159 | CHCODES = makedict(CHCODES)\r | |
160 | \r | |
161 | # replacement operations for "ignore case" mode\r | |
162 | OP_IGNORE = {\r | |
163 | GROUPREF: GROUPREF_IGNORE,\r | |
164 | IN: IN_IGNORE,\r | |
165 | LITERAL: LITERAL_IGNORE,\r | |
166 | NOT_LITERAL: NOT_LITERAL_IGNORE\r | |
167 | }\r | |
168 | \r | |
169 | AT_MULTILINE = {\r | |
170 | AT_BEGINNING: AT_BEGINNING_LINE,\r | |
171 | AT_END: AT_END_LINE\r | |
172 | }\r | |
173 | \r | |
174 | AT_LOCALE = {\r | |
175 | AT_BOUNDARY: AT_LOC_BOUNDARY,\r | |
176 | AT_NON_BOUNDARY: AT_LOC_NON_BOUNDARY\r | |
177 | }\r | |
178 | \r | |
179 | AT_UNICODE = {\r | |
180 | AT_BOUNDARY: AT_UNI_BOUNDARY,\r | |
181 | AT_NON_BOUNDARY: AT_UNI_NON_BOUNDARY\r | |
182 | }\r | |
183 | \r | |
184 | CH_LOCALE = {\r | |
185 | CATEGORY_DIGIT: CATEGORY_DIGIT,\r | |
186 | CATEGORY_NOT_DIGIT: CATEGORY_NOT_DIGIT,\r | |
187 | CATEGORY_SPACE: CATEGORY_SPACE,\r | |
188 | CATEGORY_NOT_SPACE: CATEGORY_NOT_SPACE,\r | |
189 | CATEGORY_WORD: CATEGORY_LOC_WORD,\r | |
190 | CATEGORY_NOT_WORD: CATEGORY_LOC_NOT_WORD,\r | |
191 | CATEGORY_LINEBREAK: CATEGORY_LINEBREAK,\r | |
192 | CATEGORY_NOT_LINEBREAK: CATEGORY_NOT_LINEBREAK\r | |
193 | }\r | |
194 | \r | |
195 | CH_UNICODE = {\r | |
196 | CATEGORY_DIGIT: CATEGORY_UNI_DIGIT,\r | |
197 | CATEGORY_NOT_DIGIT: CATEGORY_UNI_NOT_DIGIT,\r | |
198 | CATEGORY_SPACE: CATEGORY_UNI_SPACE,\r | |
199 | CATEGORY_NOT_SPACE: CATEGORY_UNI_NOT_SPACE,\r | |
200 | CATEGORY_WORD: CATEGORY_UNI_WORD,\r | |
201 | CATEGORY_NOT_WORD: CATEGORY_UNI_NOT_WORD,\r | |
202 | CATEGORY_LINEBREAK: CATEGORY_UNI_LINEBREAK,\r | |
203 | CATEGORY_NOT_LINEBREAK: CATEGORY_UNI_NOT_LINEBREAK\r | |
204 | }\r | |
205 | \r | |
206 | # flags\r | |
207 | SRE_FLAG_TEMPLATE = 1 # template mode (disable backtracking)\r | |
208 | SRE_FLAG_IGNORECASE = 2 # case insensitive\r | |
209 | SRE_FLAG_LOCALE = 4 # honour system locale\r | |
210 | SRE_FLAG_MULTILINE = 8 # treat target as multiline string\r | |
211 | SRE_FLAG_DOTALL = 16 # treat target as a single string\r | |
212 | SRE_FLAG_UNICODE = 32 # use unicode locale\r | |
213 | SRE_FLAG_VERBOSE = 64 # ignore whitespace and comments\r | |
214 | SRE_FLAG_DEBUG = 128 # debugging\r | |
215 | \r | |
216 | # flags for INFO primitive\r | |
217 | SRE_INFO_PREFIX = 1 # has prefix\r | |
218 | SRE_INFO_LITERAL = 2 # entire pattern is literal (given by prefix)\r | |
219 | SRE_INFO_CHARSET = 4 # pattern starts with character from given set\r | |
220 | \r | |
221 | if __name__ == "__main__":\r | |
222 | def dump(f, d, prefix):\r | |
223 | items = d.items()\r | |
224 | items.sort(key=lambda a: a[1])\r | |
225 | for k, v in items:\r | |
226 | f.write("#define %s_%s %s\n" % (prefix, k.upper(), v))\r | |
227 | f = open("sre_constants.h", "w")\r | |
228 | f.write("""\\r | |
229 | /*\r | |
230 | * Secret Labs' Regular Expression Engine\r | |
231 | *\r | |
232 | * regular expression matching engine\r | |
233 | *\r | |
234 | * NOTE: This file is generated by sre_constants.py. If you need\r | |
235 | * to change anything in here, edit sre_constants.py and run it.\r | |
236 | *\r | |
237 | * Copyright (c) 1997-2001 by Secret Labs AB. All rights reserved.\r | |
238 | *\r | |
239 | * See the _sre.c file for information on usage and redistribution.\r | |
240 | */\r | |
241 | \r | |
242 | """)\r | |
243 | \r | |
244 | f.write("#define SRE_MAGIC %d\n" % MAGIC)\r | |
245 | \r | |
246 | dump(f, OPCODES, "SRE_OP")\r | |
247 | dump(f, ATCODES, "SRE")\r | |
248 | dump(f, CHCODES, "SRE")\r | |
249 | \r | |
250 | f.write("#define SRE_FLAG_TEMPLATE %d\n" % SRE_FLAG_TEMPLATE)\r | |
251 | f.write("#define SRE_FLAG_IGNORECASE %d\n" % SRE_FLAG_IGNORECASE)\r | |
252 | f.write("#define SRE_FLAG_LOCALE %d\n" % SRE_FLAG_LOCALE)\r | |
253 | f.write("#define SRE_FLAG_MULTILINE %d\n" % SRE_FLAG_MULTILINE)\r | |
254 | f.write("#define SRE_FLAG_DOTALL %d\n" % SRE_FLAG_DOTALL)\r | |
255 | f.write("#define SRE_FLAG_UNICODE %d\n" % SRE_FLAG_UNICODE)\r | |
256 | f.write("#define SRE_FLAG_VERBOSE %d\n" % SRE_FLAG_VERBOSE)\r | |
257 | \r | |
258 | f.write("#define SRE_INFO_PREFIX %d\n" % SRE_INFO_PREFIX)\r | |
259 | f.write("#define SRE_INFO_LITERAL %d\n" % SRE_INFO_LITERAL)\r | |
260 | f.write("#define SRE_INFO_CHARSET %d\n" % SRE_INFO_CHARSET)\r | |
261 | \r | |
262 | f.close()\r | |
263 | print "done"\r |