• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
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 sre.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 = 20171005
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 __str__(self):
63        return self.name
64
65    __repr__ = __str__
66
67MAXREPEAT = _NamedIntConstant(MAXREPEAT, 'MAXREPEAT')
68
69def _makecodes(names):
70    names = names.strip().split()
71    items = [_NamedIntConstant(i, name) for i, name in enumerate(names)]
72    globals().update({item.name: item for item in items})
73    return items
74
75# operators
76# failure=0 success=1 (just because it looks better that way :-)
77OPCODES = _makecodes("""
78    FAILURE SUCCESS
79
80    ANY ANY_ALL
81    ASSERT ASSERT_NOT
82    AT
83    BRANCH
84    CALL
85    CATEGORY
86    CHARSET BIGCHARSET
87    GROUPREF GROUPREF_EXISTS
88    IN
89    INFO
90    JUMP
91    LITERAL
92    MARK
93    MAX_UNTIL
94    MIN_UNTIL
95    NOT_LITERAL
96    NEGATE
97    RANGE
98    REPEAT
99    REPEAT_ONE
100    SUBPATTERN
101    MIN_REPEAT_ONE
102
103    GROUPREF_IGNORE
104    IN_IGNORE
105    LITERAL_IGNORE
106    NOT_LITERAL_IGNORE
107
108    GROUPREF_LOC_IGNORE
109    IN_LOC_IGNORE
110    LITERAL_LOC_IGNORE
111    NOT_LITERAL_LOC_IGNORE
112
113    GROUPREF_UNI_IGNORE
114    IN_UNI_IGNORE
115    LITERAL_UNI_IGNORE
116    NOT_LITERAL_UNI_IGNORE
117    RANGE_UNI_IGNORE
118
119    MIN_REPEAT MAX_REPEAT
120""")
121del OPCODES[-2:] # remove MIN_REPEAT and MAX_REPEAT
122
123# positions
124ATCODES = _makecodes("""
125    AT_BEGINNING AT_BEGINNING_LINE AT_BEGINNING_STRING
126    AT_BOUNDARY AT_NON_BOUNDARY
127    AT_END AT_END_LINE AT_END_STRING
128
129    AT_LOC_BOUNDARY AT_LOC_NON_BOUNDARY
130
131    AT_UNI_BOUNDARY AT_UNI_NON_BOUNDARY
132""")
133
134# categories
135CHCODES = _makecodes("""
136    CATEGORY_DIGIT CATEGORY_NOT_DIGIT
137    CATEGORY_SPACE CATEGORY_NOT_SPACE
138    CATEGORY_WORD CATEGORY_NOT_WORD
139    CATEGORY_LINEBREAK CATEGORY_NOT_LINEBREAK
140
141    CATEGORY_LOC_WORD CATEGORY_LOC_NOT_WORD
142
143    CATEGORY_UNI_DIGIT CATEGORY_UNI_NOT_DIGIT
144    CATEGORY_UNI_SPACE CATEGORY_UNI_NOT_SPACE
145    CATEGORY_UNI_WORD CATEGORY_UNI_NOT_WORD
146    CATEGORY_UNI_LINEBREAK CATEGORY_UNI_NOT_LINEBREAK
147""")
148
149
150# replacement operations for "ignore case" mode
151OP_IGNORE = {
152    LITERAL: LITERAL_IGNORE,
153    NOT_LITERAL: NOT_LITERAL_IGNORE,
154}
155
156OP_LOCALE_IGNORE = {
157    LITERAL: LITERAL_LOC_IGNORE,
158    NOT_LITERAL: NOT_LITERAL_LOC_IGNORE,
159}
160
161OP_UNICODE_IGNORE = {
162    LITERAL: LITERAL_UNI_IGNORE,
163    NOT_LITERAL: NOT_LITERAL_UNI_IGNORE,
164}
165
166AT_MULTILINE = {
167    AT_BEGINNING: AT_BEGINNING_LINE,
168    AT_END: AT_END_LINE
169}
170
171AT_LOCALE = {
172    AT_BOUNDARY: AT_LOC_BOUNDARY,
173    AT_NON_BOUNDARY: AT_LOC_NON_BOUNDARY
174}
175
176AT_UNICODE = {
177    AT_BOUNDARY: AT_UNI_BOUNDARY,
178    AT_NON_BOUNDARY: AT_UNI_NON_BOUNDARY
179}
180
181CH_LOCALE = {
182    CATEGORY_DIGIT: CATEGORY_DIGIT,
183    CATEGORY_NOT_DIGIT: CATEGORY_NOT_DIGIT,
184    CATEGORY_SPACE: CATEGORY_SPACE,
185    CATEGORY_NOT_SPACE: CATEGORY_NOT_SPACE,
186    CATEGORY_WORD: CATEGORY_LOC_WORD,
187    CATEGORY_NOT_WORD: CATEGORY_LOC_NOT_WORD,
188    CATEGORY_LINEBREAK: CATEGORY_LINEBREAK,
189    CATEGORY_NOT_LINEBREAK: CATEGORY_NOT_LINEBREAK
190}
191
192CH_UNICODE = {
193    CATEGORY_DIGIT: CATEGORY_UNI_DIGIT,
194    CATEGORY_NOT_DIGIT: CATEGORY_UNI_NOT_DIGIT,
195    CATEGORY_SPACE: CATEGORY_UNI_SPACE,
196    CATEGORY_NOT_SPACE: CATEGORY_UNI_NOT_SPACE,
197    CATEGORY_WORD: CATEGORY_UNI_WORD,
198    CATEGORY_NOT_WORD: CATEGORY_UNI_NOT_WORD,
199    CATEGORY_LINEBREAK: CATEGORY_UNI_LINEBREAK,
200    CATEGORY_NOT_LINEBREAK: CATEGORY_UNI_NOT_LINEBREAK
201}
202
203# flags
204SRE_FLAG_TEMPLATE = 1 # template mode (disable backtracking)
205SRE_FLAG_IGNORECASE = 2 # case insensitive
206SRE_FLAG_LOCALE = 4 # honour system locale
207SRE_FLAG_MULTILINE = 8 # treat target as multiline string
208SRE_FLAG_DOTALL = 16 # treat target as a single string
209SRE_FLAG_UNICODE = 32 # use unicode "locale"
210SRE_FLAG_VERBOSE = 64 # ignore whitespace and comments
211SRE_FLAG_DEBUG = 128 # debugging
212SRE_FLAG_ASCII = 256 # use ascii "locale"
213
214# flags for INFO primitive
215SRE_INFO_PREFIX = 1 # has prefix
216SRE_INFO_LITERAL = 2 # entire pattern is literal (given by prefix)
217SRE_INFO_CHARSET = 4 # pattern starts with character from given set
218
219if __name__ == "__main__":
220    def dump(f, d, prefix):
221        items = sorted(d)
222        for item in items:
223            f.write("#define %s_%s %d\n" % (prefix, item, item))
224    with open("sre_constants.h", "w") as f:
225        f.write("""\
226/*
227 * Secret Labs' Regular Expression Engine
228 *
229 * regular expression matching engine
230 *
231 * NOTE: This file is generated by sre_constants.py.  If you need
232 * to change anything in here, edit sre_constants.py and run it.
233 *
234 * Copyright (c) 1997-2001 by Secret Labs AB.  All rights reserved.
235 *
236 * See the _sre.c file for information on usage and redistribution.
237 */
238
239""")
240
241        f.write("#define SRE_MAGIC %d\n" % MAGIC)
242
243        dump(f, OPCODES, "SRE_OP")
244        dump(f, ATCODES, "SRE")
245        dump(f, CHCODES, "SRE")
246
247        f.write("#define SRE_FLAG_TEMPLATE %d\n" % SRE_FLAG_TEMPLATE)
248        f.write("#define SRE_FLAG_IGNORECASE %d\n" % SRE_FLAG_IGNORECASE)
249        f.write("#define SRE_FLAG_LOCALE %d\n" % SRE_FLAG_LOCALE)
250        f.write("#define SRE_FLAG_MULTILINE %d\n" % SRE_FLAG_MULTILINE)
251        f.write("#define SRE_FLAG_DOTALL %d\n" % SRE_FLAG_DOTALL)
252        f.write("#define SRE_FLAG_UNICODE %d\n" % SRE_FLAG_UNICODE)
253        f.write("#define SRE_FLAG_VERBOSE %d\n" % SRE_FLAG_VERBOSE)
254        f.write("#define SRE_FLAG_DEBUG %d\n" % SRE_FLAG_DEBUG)
255        f.write("#define SRE_FLAG_ASCII %d\n" % SRE_FLAG_ASCII)
256
257        f.write("#define SRE_INFO_PREFIX %d\n" % SRE_INFO_PREFIX)
258        f.write("#define SRE_INFO_LITERAL %d\n" % SRE_INFO_LITERAL)
259        f.write("#define SRE_INFO_CHARSET %d\n" % SRE_INFO_CHARSET)
260
261    print("done")
262