Edit File by line

[0] Fix | Delete

# Secret Labs' Regular Expression Engine

[1] Fix | Delete

[2] Fix | Delete

# various symbols used by the regular expression engine.

[3] Fix | Delete

# run this script to update the _sre include files!

[4] Fix | Delete

[5] Fix | Delete

[6] Fix | Delete

[7] Fix | Delete

# See the sre.py file for information on usage and redistribution.

[8] Fix | Delete

[9] Fix | Delete

[10] Fix | Delete

"""Internal support module for sre"""

[11] Fix | Delete

[12] Fix | Delete

# update when constants are added or removed

[13] Fix | Delete

[14] Fix | Delete

MAGIC = 20140917

[15] Fix | Delete

[16] Fix | Delete

from _sre import MAXREPEAT, MAXGROUPS

[17] Fix | Delete

[18] Fix | Delete

# SRE standard exception (access as sre.error)

[19] Fix | Delete

# should this really be here?

[20] Fix | Delete

[21] Fix | Delete

class error(Exception):

[22] Fix | Delete

"""Exception raised for invalid regular expressions.

[23] Fix | Delete

[24] Fix | Delete

Attributes:

[25] Fix | Delete

[26] Fix | Delete

msg: The unformatted error message

[27] Fix | Delete

pattern: The regular expression pattern

[28] Fix | Delete

pos: The index in the pattern where compilation failed (may be None)

[29] Fix | Delete

lineno: The line corresponding to pos (may be None)

[30] Fix | Delete

colno: The column corresponding to pos (may be None)

[31] Fix | Delete

"""

[32] Fix | Delete

[33] Fix | Delete

def __init__(self, msg, pattern=None, pos=None):

[34] Fix | Delete

self.msg = msg

[35] Fix | Delete

self.pattern = pattern

[36] Fix | Delete

self.pos = pos

[37] Fix | Delete

if pattern is not None and pos is not None:

[38] Fix | Delete

msg = '%s at position %d' % (msg, pos)

[39] Fix | Delete

if isinstance(pattern, str):

[40] Fix | Delete

newline = '\n'

[41] Fix | Delete

else:

[42] Fix | Delete

newline = b'\n'

[43] Fix | Delete

self.lineno = pattern.count(newline, 0, pos) + 1

[44] Fix | Delete

self.colno = pos - pattern.rfind(newline, 0, pos)

[45] Fix | Delete

if newline in pattern:

[46] Fix | Delete

msg = '%s (line %d, column %d)' % (msg, self.lineno, self.colno)

[47] Fix | Delete

else:

[48] Fix | Delete

self.lineno = self.colno = None

[49] Fix | Delete

super().__init__(msg)

[50] Fix | Delete

[51] Fix | Delete

[52] Fix | Delete

class _NamedIntConstant(int):

[53] Fix | Delete

def __new__(cls, value, name):

[54] Fix | Delete

self = super(_NamedIntConstant, cls).__new__(cls, value)

[55] Fix | Delete

self.name = name

[56] Fix | Delete

return self

[57] Fix | Delete

[58] Fix | Delete

def __str__(self):

[59] Fix | Delete

return self.name

[60] Fix | Delete

[61] Fix | Delete

__repr__ = __str__

[62] Fix | Delete

[63] Fix | Delete

MAXREPEAT = _NamedIntConstant(MAXREPEAT, 'MAXREPEAT')

[64] Fix | Delete

[65] Fix | Delete

def _makecodes(names):

[66] Fix | Delete

names = names.strip().split()

[67] Fix | Delete

items = [_NamedIntConstant(i, name) for i, name in enumerate(names)]

[68] Fix | Delete

globals().update({item.name: item for item in items})

[69] Fix | Delete

return items

[70] Fix | Delete

[71] Fix | Delete

# operators

[72] Fix | Delete

# failure=0 success=1 (just because it looks better that way :-)

[73] Fix | Delete

OPCODES = _makecodes("""

[74] Fix | Delete

FAILURE SUCCESS

[75] Fix | Delete

[76] Fix | Delete

ANY ANY_ALL

[77] Fix | Delete

ASSERT ASSERT_NOT

[78] Fix | Delete

[79] Fix | Delete

BRANCH

[80] Fix | Delete

CALL

[81] Fix | Delete

CATEGORY

[82] Fix | Delete

CHARSET BIGCHARSET

[83] Fix | Delete

GROUPREF GROUPREF_EXISTS GROUPREF_IGNORE

[84] Fix | Delete

IN IN_IGNORE

[85] Fix | Delete

INFO

[86] Fix | Delete

JUMP

[87] Fix | Delete

LITERAL LITERAL_IGNORE

[88] Fix | Delete

MARK

[89] Fix | Delete

MAX_UNTIL

[90] Fix | Delete

MIN_UNTIL

[91] Fix | Delete

NOT_LITERAL NOT_LITERAL_IGNORE

[92] Fix | Delete

NEGATE

[93] Fix | Delete

RANGE

[94] Fix | Delete

REPEAT

[95] Fix | Delete

REPEAT_ONE

[96] Fix | Delete

SUBPATTERN

[97] Fix | Delete

MIN_REPEAT_ONE

[98] Fix | Delete

RANGE_IGNORE

[99] Fix | Delete

[100] Fix | Delete

MIN_REPEAT MAX_REPEAT

[101] Fix | Delete

""")

[102] Fix | Delete

del OPCODES[-2:] # remove MIN_REPEAT and MAX_REPEAT

[103] Fix | Delete

[104] Fix | Delete

# positions

[105] Fix | Delete

ATCODES = _makecodes("""

[106] Fix | Delete

AT_BEGINNING AT_BEGINNING_LINE AT_BEGINNING_STRING

[107] Fix | Delete

AT_BOUNDARY AT_NON_BOUNDARY

[108] Fix | Delete

AT_END AT_END_LINE AT_END_STRING

[109] Fix | Delete

AT_LOC_BOUNDARY AT_LOC_NON_BOUNDARY

[110] Fix | Delete

AT_UNI_BOUNDARY AT_UNI_NON_BOUNDARY

[111] Fix | Delete

""")

[112] Fix | Delete

[113] Fix | Delete

# categories

[114] Fix | Delete

CHCODES = _makecodes("""

[115] Fix | Delete

CATEGORY_DIGIT CATEGORY_NOT_DIGIT

[116] Fix | Delete

CATEGORY_SPACE CATEGORY_NOT_SPACE

[117] Fix | Delete

CATEGORY_WORD CATEGORY_NOT_WORD

[118] Fix | Delete

CATEGORY_LINEBREAK CATEGORY_NOT_LINEBREAK

[119] Fix | Delete

CATEGORY_LOC_WORD CATEGORY_LOC_NOT_WORD

[120] Fix | Delete

CATEGORY_UNI_DIGIT CATEGORY_UNI_NOT_DIGIT

[121] Fix | Delete

CATEGORY_UNI_SPACE CATEGORY_UNI_NOT_SPACE

[122] Fix | Delete

CATEGORY_UNI_WORD CATEGORY_UNI_NOT_WORD

[123] Fix | Delete

CATEGORY_UNI_LINEBREAK CATEGORY_UNI_NOT_LINEBREAK

[124] Fix | Delete

""")

[125] Fix | Delete

[126] Fix | Delete

[127] Fix | Delete

# replacement operations for "ignore case" mode

[128] Fix | Delete

OP_IGNORE = {

[129] Fix | Delete

GROUPREF: GROUPREF_IGNORE,

[130] Fix | Delete

IN: IN_IGNORE,

[131] Fix | Delete

LITERAL: LITERAL_IGNORE,

[132] Fix | Delete

NOT_LITERAL: NOT_LITERAL_IGNORE,

[133] Fix | Delete

RANGE: RANGE_IGNORE,

[134] Fix | Delete

}

[135] Fix | Delete

[136] Fix | Delete

AT_MULTILINE = {

[137] Fix | Delete

AT_BEGINNING: AT_BEGINNING_LINE,

[138] Fix | Delete

AT_END: AT_END_LINE

[139] Fix | Delete

}

[140] Fix | Delete

[141] Fix | Delete

AT_LOCALE = {

[142] Fix | Delete

AT_BOUNDARY: AT_LOC_BOUNDARY,

[143] Fix | Delete

AT_NON_BOUNDARY: AT_LOC_NON_BOUNDARY

[144] Fix | Delete

}

[145] Fix | Delete

[146] Fix | Delete

AT_UNICODE = {

[147] Fix | Delete

AT_BOUNDARY: AT_UNI_BOUNDARY,

[148] Fix | Delete

AT_NON_BOUNDARY: AT_UNI_NON_BOUNDARY

[149] Fix | Delete

}

[150] Fix | Delete

[151] Fix | Delete

CH_LOCALE = {

[152] Fix | Delete

CATEGORY_DIGIT: CATEGORY_DIGIT,

[153] Fix | Delete

CATEGORY_NOT_DIGIT: CATEGORY_NOT_DIGIT,

[154] Fix | Delete

CATEGORY_SPACE: CATEGORY_SPACE,

[155] Fix | Delete

CATEGORY_NOT_SPACE: CATEGORY_NOT_SPACE,

[156] Fix | Delete

CATEGORY_WORD: CATEGORY_LOC_WORD,

[157] Fix | Delete

CATEGORY_NOT_WORD: CATEGORY_LOC_NOT_WORD,

[158] Fix | Delete

CATEGORY_LINEBREAK: CATEGORY_LINEBREAK,

[159] Fix | Delete

CATEGORY_NOT_LINEBREAK: CATEGORY_NOT_LINEBREAK

[160] Fix | Delete

}

[161] Fix | Delete

[162] Fix | Delete

CH_UNICODE = {

[163] Fix | Delete

CATEGORY_DIGIT: CATEGORY_UNI_DIGIT,

[164] Fix | Delete

CATEGORY_NOT_DIGIT: CATEGORY_UNI_NOT_DIGIT,

[165] Fix | Delete

CATEGORY_SPACE: CATEGORY_UNI_SPACE,

[166] Fix | Delete

CATEGORY_NOT_SPACE: CATEGORY_UNI_NOT_SPACE,

[167] Fix | Delete

CATEGORY_WORD: CATEGORY_UNI_WORD,

[168] Fix | Delete

CATEGORY_NOT_WORD: CATEGORY_UNI_NOT_WORD,

[169] Fix | Delete

CATEGORY_LINEBREAK: CATEGORY_UNI_LINEBREAK,

[170] Fix | Delete

CATEGORY_NOT_LINEBREAK: CATEGORY_UNI_NOT_LINEBREAK

[171] Fix | Delete

}

[172] Fix | Delete

[173] Fix | Delete

# flags

[174] Fix | Delete

SRE_FLAG_TEMPLATE = 1 # template mode (disable backtracking)

[175] Fix | Delete

SRE_FLAG_IGNORECASE = 2 # case insensitive

[176] Fix | Delete

SRE_FLAG_LOCALE = 4 # honour system locale

[177] Fix | Delete

SRE_FLAG_MULTILINE = 8 # treat target as multiline string

[178] Fix | Delete

SRE_FLAG_DOTALL = 16 # treat target as a single string

[179] Fix | Delete

SRE_FLAG_UNICODE = 32 # use unicode "locale"

[180] Fix | Delete

SRE_FLAG_VERBOSE = 64 # ignore whitespace and comments

[181] Fix | Delete

SRE_FLAG_DEBUG = 128 # debugging

[182] Fix | Delete

SRE_FLAG_ASCII = 256 # use ascii "locale"

[183] Fix | Delete

[184] Fix | Delete

# flags for INFO primitive

[185] Fix | Delete

SRE_INFO_PREFIX = 1 # has prefix

[186] Fix | Delete

SRE_INFO_LITERAL = 2 # entire pattern is literal (given by prefix)

[187] Fix | Delete

SRE_INFO_CHARSET = 4 # pattern starts with character from given set

[188] Fix | Delete

[189] Fix | Delete

if __name__ == "__main__":

[190] Fix | Delete

def dump(f, d, prefix):

[191] Fix | Delete

items = sorted(d)

[192] Fix | Delete

for item in items:

[193] Fix | Delete

f.write("#define %s_%s %d\n" % (prefix, item, item))

[194] Fix | Delete

with open("sre_constants.h", "w") as f:

[195] Fix | Delete

f.write("""\

[196] Fix | Delete

[197] Fix | Delete

* Secret Labs' Regular Expression Engine

[198] Fix | Delete

[199] Fix | Delete

* regular expression matching engine

[200] Fix | Delete

[201] Fix | Delete

* NOTE: This file is generated by sre_constants.py. If you need

[202] Fix | Delete

* to change anything in here, edit sre_constants.py and run it.

[203] Fix | Delete

[204] Fix | Delete

[205] Fix | Delete

[206] Fix | Delete

* See the _sre.c file for information on usage and redistribution.

[207] Fix | Delete

[208] Fix | Delete

[209] Fix | Delete

""")

[210] Fix | Delete

[211] Fix | Delete

f.write("#define SRE_MAGIC %d\n" % MAGIC)

[212] Fix | Delete

[213] Fix | Delete

dump(f, OPCODES, "SRE_OP")

[214] Fix | Delete

dump(f, ATCODES, "SRE")

[215] Fix | Delete

dump(f, CHCODES, "SRE")

[216] Fix | Delete

[217] Fix | Delete

f.write("#define SRE_FLAG_TEMPLATE %d\n" % SRE_FLAG_TEMPLATE)

[218] Fix | Delete

f.write("#define SRE_FLAG_IGNORECASE %d\n" % SRE_FLAG_IGNORECASE)

[219] Fix | Delete

f.write("#define SRE_FLAG_LOCALE %d\n" % SRE_FLAG_LOCALE)

[220] Fix | Delete

f.write("#define SRE_FLAG_MULTILINE %d\n" % SRE_FLAG_MULTILINE)

[221] Fix | Delete

f.write("#define SRE_FLAG_DOTALL %d\n" % SRE_FLAG_DOTALL)

[222] Fix | Delete

f.write("#define SRE_FLAG_UNICODE %d\n" % SRE_FLAG_UNICODE)

[223] Fix | Delete

f.write("#define SRE_FLAG_VERBOSE %d\n" % SRE_FLAG_VERBOSE)

[224] Fix | Delete

f.write("#define SRE_FLAG_DEBUG %d\n" % SRE_FLAG_DEBUG)

[225] Fix | Delete

f.write("#define SRE_FLAG_ASCII %d\n" % SRE_FLAG_ASCII)

[226] Fix | Delete

[227] Fix | Delete

f.write("#define SRE_INFO_PREFIX %d\n" % SRE_INFO_PREFIX)

[228] Fix | Delete

f.write("#define SRE_INFO_LITERAL %d\n" % SRE_INFO_LITERAL)

[229] Fix | Delete

f.write("#define SRE_INFO_CHARSET %d\n" % SRE_INFO_CHARSET)

[230] Fix | Delete

[231] Fix | Delete

print("done")

[232] Fix | Delete

[233] Fix | Delete