blob: b0164312d0357e66608710a97705fbc6824eeb43 [file] [log] [blame]
Guido van Rossum7627c0d2000-03-31 14:58:54 +00001#
2# Secret Labs' Regular Expression Engine
Guido van Rossum7627c0d2000-03-31 14:58:54 +00003#
4# various symbols used by the regular expression engine.
5# run this script to update the _sre include files!
6#
Fredrik Lundh770617b2001-01-14 15:06:11 +00007# Copyright (c) 1998-2001 by Secret Labs AB. All rights reserved.
Guido van Rossum7627c0d2000-03-31 14:58:54 +00008#
Fredrik Lundh29c4ba92000-08-01 18:20:07 +00009# See the sre.py file for information on usage and redistribution.
Guido van Rossum7627c0d2000-03-31 14:58:54 +000010#
11
Fred Drakeb8f22742001-09-04 19:10:20 +000012"""Internal support module for sre"""
13
Fredrik Lundhb35ffc02001-01-15 12:46:09 +000014# update when constants are added or removed
15
Serhiy Storchaka898ff032017-05-05 08:53:40 +030016MAGIC = 20170530
Fredrik Lundhb35ffc02001-01-15 12:46:09 +000017
Serhiy Storchaka9baa5b22014-09-29 22:49:23 +030018from _sre import MAXREPEAT, MAXGROUPS
Serhiy Storchaka9acb9bc2013-02-18 11:14:21 +020019
Fredrik Lundhb35ffc02001-01-15 12:46:09 +000020# SRE standard exception (access as sre.error)
Fredrik Lundh436c3d582000-06-29 08:58:44 +000021# should this really be here?
22
23class error(Exception):
Serhiy Storchakaad446d52014-11-10 13:49:00 +020024 def __init__(self, msg, pattern=None, pos=None):
25 self.msg = msg
26 self.pattern = pattern
27 self.pos = pos
28 if pattern is not None and pos is not None:
29 msg = '%s at position %d' % (msg, pos)
30 if isinstance(pattern, str):
31 newline = '\n'
32 else:
33 newline = b'\n'
34 self.lineno = pattern.count(newline, 0, pos) + 1
35 self.colno = pos - pattern.rfind(newline, 0, pos)
36 if newline in pattern:
37 msg = '%s (line %d, column %d)' % (msg, self.lineno, self.colno)
38 else:
39 self.lineno = self.colno = None
40 super().__init__(msg)
41
Serhiy Storchakac7f7d382014-11-09 20:48:36 +020042
43class _NamedIntConstant(int):
44 def __new__(cls, value, name):
45 self = super(_NamedIntConstant, cls).__new__(cls, value)
46 self.name = name
47 return self
48
49 def __str__(self):
50 return self.name
51
52 __repr__ = __str__
53
54MAXREPEAT = _NamedIntConstant(MAXREPEAT, 'MAXREPEAT')
55
56def _makecodes(names):
57 names = names.strip().split()
58 items = [_NamedIntConstant(i, name) for i, name in enumerate(names)]
59 globals().update({item.name: item for item in items})
60 return items
61
Guido van Rossum7627c0d2000-03-31 14:58:54 +000062# operators
Serhiy Storchakac7f7d382014-11-09 20:48:36 +020063# failure=0 success=1 (just because it looks better that way :-)
64OPCODES = _makecodes("""
65 FAILURE SUCCESS
Guido van Rossum7627c0d2000-03-31 14:58:54 +000066
Serhiy Storchakac7f7d382014-11-09 20:48:36 +020067 ANY ANY_ALL
68 ASSERT ASSERT_NOT
69 AT
70 BRANCH
71 CALL
72 CATEGORY
73 CHARSET BIGCHARSET
74 GROUPREF GROUPREF_EXISTS GROUPREF_IGNORE
75 IN IN_IGNORE
76 INFO
77 JUMP
78 LITERAL LITERAL_IGNORE
79 MARK
80 MAX_UNTIL
81 MIN_UNTIL
82 NOT_LITERAL NOT_LITERAL_IGNORE
83 NEGATE
84 RANGE
85 REPEAT
86 REPEAT_ONE
87 SUBPATTERN
88 MIN_REPEAT_ONE
89 RANGE_IGNORE
Serhiy Storchaka898ff032017-05-05 08:53:40 +030090 LITERAL_LOC_IGNORE
91 NOT_LITERAL_LOC_IGNORE
92 IN_LOC_IGNORE
Guido van Rossum7627c0d2000-03-31 14:58:54 +000093
Serhiy Storchakac7f7d382014-11-09 20:48:36 +020094 MIN_REPEAT MAX_REPEAT
95""")
96del OPCODES[-2:] # remove MIN_REPEAT and MAX_REPEAT
Guido van Rossum7627c0d2000-03-31 14:58:54 +000097
98# positions
Serhiy Storchakac7f7d382014-11-09 20:48:36 +020099ATCODES = _makecodes("""
100 AT_BEGINNING AT_BEGINNING_LINE AT_BEGINNING_STRING
101 AT_BOUNDARY AT_NON_BOUNDARY
102 AT_END AT_END_LINE AT_END_STRING
103 AT_LOC_BOUNDARY AT_LOC_NON_BOUNDARY
104 AT_UNI_BOUNDARY AT_UNI_NON_BOUNDARY
105""")
Guido van Rossum7627c0d2000-03-31 14:58:54 +0000106
107# categories
Serhiy Storchakac7f7d382014-11-09 20:48:36 +0200108CHCODES = _makecodes("""
109 CATEGORY_DIGIT CATEGORY_NOT_DIGIT
110 CATEGORY_SPACE CATEGORY_NOT_SPACE
111 CATEGORY_WORD CATEGORY_NOT_WORD
112 CATEGORY_LINEBREAK CATEGORY_NOT_LINEBREAK
113 CATEGORY_LOC_WORD CATEGORY_LOC_NOT_WORD
114 CATEGORY_UNI_DIGIT CATEGORY_UNI_NOT_DIGIT
115 CATEGORY_UNI_SPACE CATEGORY_UNI_NOT_SPACE
116 CATEGORY_UNI_WORD CATEGORY_UNI_NOT_WORD
117 CATEGORY_UNI_LINEBREAK CATEGORY_UNI_NOT_LINEBREAK
118""")
Guido van Rossum7627c0d2000-03-31 14:58:54 +0000119
Guido van Rossum7627c0d2000-03-31 14:58:54 +0000120
121# replacement operations for "ignore case" mode
Jeremy Hyltonb1aa1952000-06-01 17:39:12 +0000122OP_IGNORE = {
Fredrik Lundh72b82ba2000-07-03 21:31:48 +0000123 GROUPREF: GROUPREF_IGNORE,
Guido van Rossum7627c0d2000-03-31 14:58:54 +0000124 IN: IN_IGNORE,
125 LITERAL: LITERAL_IGNORE,
Serhiy Storchaka4b8f8942014-10-31 12:36:56 +0200126 NOT_LITERAL: NOT_LITERAL_IGNORE,
127 RANGE: RANGE_IGNORE,
Guido van Rossum7627c0d2000-03-31 14:58:54 +0000128}
129
Serhiy Storchaka898ff032017-05-05 08:53:40 +0300130OP_LOC_IGNORE = {
131 LITERAL: LITERAL_LOC_IGNORE,
132 NOT_LITERAL: NOT_LITERAL_LOC_IGNORE,
133}
134
Jeremy Hyltonb1aa1952000-06-01 17:39:12 +0000135AT_MULTILINE = {
136 AT_BEGINNING: AT_BEGINNING_LINE,
137 AT_END: AT_END_LINE
Guido van Rossum7627c0d2000-03-31 14:58:54 +0000138}
139
Fredrik Lundhb25e1ad2001-03-22 15:50:10 +0000140AT_LOCALE = {
141 AT_BOUNDARY: AT_LOC_BOUNDARY,
142 AT_NON_BOUNDARY: AT_LOC_NON_BOUNDARY
143}
144
145AT_UNICODE = {
146 AT_BOUNDARY: AT_UNI_BOUNDARY,
147 AT_NON_BOUNDARY: AT_UNI_NON_BOUNDARY
148}
149
Jeremy Hyltonb1aa1952000-06-01 17:39:12 +0000150CH_LOCALE = {
Fredrik Lundh436c3d582000-06-29 08:58:44 +0000151 CATEGORY_DIGIT: CATEGORY_DIGIT,
152 CATEGORY_NOT_DIGIT: CATEGORY_NOT_DIGIT,
153 CATEGORY_SPACE: CATEGORY_SPACE,
154 CATEGORY_NOT_SPACE: CATEGORY_NOT_SPACE,
Jeremy Hyltonb1aa1952000-06-01 17:39:12 +0000155 CATEGORY_WORD: CATEGORY_LOC_WORD,
156 CATEGORY_NOT_WORD: CATEGORY_LOC_NOT_WORD,
Fredrik Lundh436c3d582000-06-29 08:58:44 +0000157 CATEGORY_LINEBREAK: CATEGORY_LINEBREAK,
158 CATEGORY_NOT_LINEBREAK: CATEGORY_NOT_LINEBREAK
159}
160
161CH_UNICODE = {
162 CATEGORY_DIGIT: CATEGORY_UNI_DIGIT,
163 CATEGORY_NOT_DIGIT: CATEGORY_UNI_NOT_DIGIT,
164 CATEGORY_SPACE: CATEGORY_UNI_SPACE,
165 CATEGORY_NOT_SPACE: CATEGORY_UNI_NOT_SPACE,
166 CATEGORY_WORD: CATEGORY_UNI_WORD,
167 CATEGORY_NOT_WORD: CATEGORY_UNI_NOT_WORD,
168 CATEGORY_LINEBREAK: CATEGORY_UNI_LINEBREAK,
169 CATEGORY_NOT_LINEBREAK: CATEGORY_UNI_NOT_LINEBREAK
Guido van Rossum7627c0d2000-03-31 14:58:54 +0000170}
171
Jeremy Hyltonb1aa1952000-06-01 17:39:12 +0000172# flags
Fredrik Lundh3562f112000-07-02 12:00:07 +0000173SRE_FLAG_TEMPLATE = 1 # template mode (disable backtracking)
174SRE_FLAG_IGNORECASE = 2 # case insensitive
Fredrik Lundh8a3ebf82000-07-23 21:46:17 +0000175SRE_FLAG_LOCALE = 4 # honour system locale
Fredrik Lundh3562f112000-07-02 12:00:07 +0000176SRE_FLAG_MULTILINE = 8 # treat target as multiline string
177SRE_FLAG_DOTALL = 16 # treat target as a single string
Antoine Pitroufd036452008-08-19 17:56:33 +0000178SRE_FLAG_UNICODE = 32 # use unicode "locale"
Fredrik Lundh3562f112000-07-02 12:00:07 +0000179SRE_FLAG_VERBOSE = 64 # ignore whitespace and comments
Fredrik Lundh770617b2001-01-14 15:06:11 +0000180SRE_FLAG_DEBUG = 128 # debugging
Antoine Pitroufd036452008-08-19 17:56:33 +0000181SRE_FLAG_ASCII = 256 # use ascii "locale"
Fredrik Lundh3562f112000-07-02 12:00:07 +0000182
183# flags for INFO primitive
184SRE_INFO_PREFIX = 1 # has prefix
185SRE_INFO_LITERAL = 2 # entire pattern is literal (given by prefix)
186SRE_INFO_CHARSET = 4 # pattern starts with character from given set
Jeremy Hyltonb1aa1952000-06-01 17:39:12 +0000187
Guido van Rossum7627c0d2000-03-31 14:58:54 +0000188if __name__ == "__main__":
Jeremy Hyltonb1aa1952000-06-01 17:39:12 +0000189 def dump(f, d, prefix):
Serhiy Storchakac7f7d382014-11-09 20:48:36 +0200190 items = sorted(d)
191 for item in items:
192 f.write("#define %s_%s %d\n" % (prefix, item, item))
Serhiy Storchaka46ba6c82015-04-04 11:01:02 +0300193 with open("sre_constants.h", "w") as f:
194 f.write("""\
Fredrik Lundh43b3b492000-06-30 10:41:31 +0000195/*
196 * Secret Labs' Regular Expression Engine
197 *
198 * regular expression matching engine
199 *
200 * NOTE: This file is generated by sre_constants.py. If you need
201 * to change anything in here, edit sre_constants.py and run it.
202 *
Fredrik Lundh770617b2001-01-14 15:06:11 +0000203 * Copyright (c) 1997-2001 by Secret Labs AB. All rights reserved.
Fredrik Lundh43b3b492000-06-30 10:41:31 +0000204 *
205 * See the _sre.c file for information on usage and redistribution.
206 */
207
208""")
209
Serhiy Storchaka46ba6c82015-04-04 11:01:02 +0300210 f.write("#define SRE_MAGIC %d\n" % MAGIC)
Fredrik Lundhb35ffc02001-01-15 12:46:09 +0000211
Serhiy Storchaka46ba6c82015-04-04 11:01:02 +0300212 dump(f, OPCODES, "SRE_OP")
213 dump(f, ATCODES, "SRE")
214 dump(f, CHCODES, "SRE")
Fredrik Lundh3562f112000-07-02 12:00:07 +0000215
Serhiy Storchaka46ba6c82015-04-04 11:01:02 +0300216 f.write("#define SRE_FLAG_TEMPLATE %d\n" % SRE_FLAG_TEMPLATE)
217 f.write("#define SRE_FLAG_IGNORECASE %d\n" % SRE_FLAG_IGNORECASE)
218 f.write("#define SRE_FLAG_LOCALE %d\n" % SRE_FLAG_LOCALE)
219 f.write("#define SRE_FLAG_MULTILINE %d\n" % SRE_FLAG_MULTILINE)
220 f.write("#define SRE_FLAG_DOTALL %d\n" % SRE_FLAG_DOTALL)
221 f.write("#define SRE_FLAG_UNICODE %d\n" % SRE_FLAG_UNICODE)
222 f.write("#define SRE_FLAG_VERBOSE %d\n" % SRE_FLAG_VERBOSE)
223 f.write("#define SRE_FLAG_DEBUG %d\n" % SRE_FLAG_DEBUG)
224 f.write("#define SRE_FLAG_ASCII %d\n" % SRE_FLAG_ASCII)
Fredrik Lundh3562f112000-07-02 12:00:07 +0000225
Serhiy Storchaka46ba6c82015-04-04 11:01:02 +0300226 f.write("#define SRE_INFO_PREFIX %d\n" % SRE_INFO_PREFIX)
227 f.write("#define SRE_INFO_LITERAL %d\n" % SRE_INFO_LITERAL)
228 f.write("#define SRE_INFO_CHARSET %d\n" % SRE_INFO_CHARSET)
Fredrik Lundh3562f112000-07-02 12:00:07 +0000229
Guido van Rossumbe19ed72007-02-09 05:37:30 +0000230 print("done")