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