Home | History | Annotate | Download | only in Lib
      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 
     16 MAGIC = 20140917
     17 
     18 from _sre import MAXREPEAT, MAXGROUPS
     19 
     20 # SRE standard exception (access as sre.error)
     21 # should this really be here?
     22 
     23 class error(Exception):
     24     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 
     42 
     43 class _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 
     54 MAXREPEAT = _NamedIntConstant(MAXREPEAT, 'MAXREPEAT')
     55 
     56 def _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 
     62 # operators
     63 # failure=0 success=1 (just because it looks better that way :-)
     64 OPCODES = _makecodes("""
     65     FAILURE SUCCESS
     66 
     67     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
     90 
     91     MIN_REPEAT MAX_REPEAT
     92 """)
     93 del OPCODES[-2:] # remove MIN_REPEAT and MAX_REPEAT
     94 
     95 # positions
     96 ATCODES = _makecodes("""
     97     AT_BEGINNING AT_BEGINNING_LINE AT_BEGINNING_STRING
     98     AT_BOUNDARY AT_NON_BOUNDARY
     99     AT_END AT_END_LINE AT_END_STRING
    100     AT_LOC_BOUNDARY AT_LOC_NON_BOUNDARY
    101     AT_UNI_BOUNDARY AT_UNI_NON_BOUNDARY
    102 """)
    103 
    104 # categories
    105 CHCODES = _makecodes("""
    106     CATEGORY_DIGIT CATEGORY_NOT_DIGIT
    107     CATEGORY_SPACE CATEGORY_NOT_SPACE
    108     CATEGORY_WORD CATEGORY_NOT_WORD
    109     CATEGORY_LINEBREAK CATEGORY_NOT_LINEBREAK
    110     CATEGORY_LOC_WORD CATEGORY_LOC_NOT_WORD
    111     CATEGORY_UNI_DIGIT CATEGORY_UNI_NOT_DIGIT
    112     CATEGORY_UNI_SPACE CATEGORY_UNI_NOT_SPACE
    113     CATEGORY_UNI_WORD CATEGORY_UNI_NOT_WORD
    114     CATEGORY_UNI_LINEBREAK CATEGORY_UNI_NOT_LINEBREAK
    115 """)
    116 
    117 
    118 # replacement operations for "ignore case" mode
    119 OP_IGNORE = {
    120     GROUPREF: GROUPREF_IGNORE,
    121     IN: IN_IGNORE,
    122     LITERAL: LITERAL_IGNORE,
    123     NOT_LITERAL: NOT_LITERAL_IGNORE,
    124     RANGE: RANGE_IGNORE,
    125 }
    126 
    127 AT_MULTILINE = {
    128     AT_BEGINNING: AT_BEGINNING_LINE,
    129     AT_END: AT_END_LINE
    130 }
    131 
    132 AT_LOCALE = {
    133     AT_BOUNDARY: AT_LOC_BOUNDARY,
    134     AT_NON_BOUNDARY: AT_LOC_NON_BOUNDARY
    135 }
    136 
    137 AT_UNICODE = {
    138     AT_BOUNDARY: AT_UNI_BOUNDARY,
    139     AT_NON_BOUNDARY: AT_UNI_NON_BOUNDARY
    140 }
    141 
    142 CH_LOCALE = {
    143     CATEGORY_DIGIT: CATEGORY_DIGIT,
    144     CATEGORY_NOT_DIGIT: CATEGORY_NOT_DIGIT,
    145     CATEGORY_SPACE: CATEGORY_SPACE,
    146     CATEGORY_NOT_SPACE: CATEGORY_NOT_SPACE,
    147     CATEGORY_WORD: CATEGORY_LOC_WORD,
    148     CATEGORY_NOT_WORD: CATEGORY_LOC_NOT_WORD,
    149     CATEGORY_LINEBREAK: CATEGORY_LINEBREAK,
    150     CATEGORY_NOT_LINEBREAK: CATEGORY_NOT_LINEBREAK
    151 }
    152 
    153 CH_UNICODE = {
    154     CATEGORY_DIGIT: CATEGORY_UNI_DIGIT,
    155     CATEGORY_NOT_DIGIT: CATEGORY_UNI_NOT_DIGIT,
    156     CATEGORY_SPACE: CATEGORY_UNI_SPACE,
    157     CATEGORY_NOT_SPACE: CATEGORY_UNI_NOT_SPACE,
    158     CATEGORY_WORD: CATEGORY_UNI_WORD,
    159     CATEGORY_NOT_WORD: CATEGORY_UNI_NOT_WORD,
    160     CATEGORY_LINEBREAK: CATEGORY_UNI_LINEBREAK,
    161     CATEGORY_NOT_LINEBREAK: CATEGORY_UNI_NOT_LINEBREAK
    162 }
    163 
    164 # flags
    165 SRE_FLAG_TEMPLATE = 1 # template mode (disable backtracking)
    166 SRE_FLAG_IGNORECASE = 2 # case insensitive
    167 SRE_FLAG_LOCALE = 4 # honour system locale
    168 SRE_FLAG_MULTILINE = 8 # treat target as multiline string
    169 SRE_FLAG_DOTALL = 16 # treat target as a single string
    170 SRE_FLAG_UNICODE = 32 # use unicode "locale"
    171 SRE_FLAG_VERBOSE = 64 # ignore whitespace and comments
    172 SRE_FLAG_DEBUG = 128 # debugging
    173 SRE_FLAG_ASCII = 256 # use ascii "locale"
    174 
    175 # flags for INFO primitive
    176 SRE_INFO_PREFIX = 1 # has prefix
    177 SRE_INFO_LITERAL = 2 # entire pattern is literal (given by prefix)
    178 SRE_INFO_CHARSET = 4 # pattern starts with character from given set
    179 
    180 if __name__ == "__main__":
    181     def dump(f, d, prefix):
    182         items = sorted(d)
    183         for item in items:
    184             f.write("#define %s_%s %d\n" % (prefix, item, item))
    185     with open("sre_constants.h", "w") as f:
    186         f.write("""\
    187 /*
    188  * Secret Labs' Regular Expression Engine
    189  *
    190  * regular expression matching engine
    191  *
    192  * NOTE: This file is generated by sre_constants.py.  If you need
    193  * to change anything in here, edit sre_constants.py and run it.
    194  *
    195  * Copyright (c) 1997-2001 by Secret Labs AB.  All rights reserved.
    196  *
    197  * See the _sre.c file for information on usage and redistribution.
    198  */
    199 
    200 """)
    201 
    202         f.write("#define SRE_MAGIC %d\n" % MAGIC)
    203 
    204         dump(f, OPCODES, "SRE_OP")
    205         dump(f, ATCODES, "SRE")
    206         dump(f, CHCODES, "SRE")
    207 
    208         f.write("#define SRE_FLAG_TEMPLATE %d\n" % SRE_FLAG_TEMPLATE)
    209         f.write("#define SRE_FLAG_IGNORECASE %d\n" % SRE_FLAG_IGNORECASE)
    210         f.write("#define SRE_FLAG_LOCALE %d\n" % SRE_FLAG_LOCALE)
    211         f.write("#define SRE_FLAG_MULTILINE %d\n" % SRE_FLAG_MULTILINE)
    212         f.write("#define SRE_FLAG_DOTALL %d\n" % SRE_FLAG_DOTALL)
    213         f.write("#define SRE_FLAG_UNICODE %d\n" % SRE_FLAG_UNICODE)
    214         f.write("#define SRE_FLAG_VERBOSE %d\n" % SRE_FLAG_VERBOSE)
    215         f.write("#define SRE_FLAG_DEBUG %d\n" % SRE_FLAG_DEBUG)
    216         f.write("#define SRE_FLAG_ASCII %d\n" % SRE_FLAG_ASCII)
    217 
    218         f.write("#define SRE_INFO_PREFIX %d\n" % SRE_INFO_PREFIX)
    219         f.write("#define SRE_INFO_LITERAL %d\n" % SRE_INFO_LITERAL)
    220         f.write("#define SRE_INFO_CHARSET %d\n" % SRE_INFO_CHARSET)
    221 
    222     print("done")
    223