#

[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

# Copyright (c) 1998-2001 by Secret Labs AB. All rights reserved.

[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 = 20031017

[15] Fix | Delete

[16] Fix | Delete

try:

[17] Fix | Delete

from _sre import MAXREPEAT

[18] Fix | Delete

except ImportError:

[19] Fix | Delete

import _sre

[20] Fix | Delete

MAXREPEAT = _sre.MAXREPEAT = 65535

[21] Fix | Delete

[22] Fix | Delete

# SRE standard exception (access as sre.error)

[23] Fix | Delete

# should this really be here?

[24] Fix | Delete

[25] Fix | Delete

class error(Exception):

[26] Fix | Delete

pass

[27] Fix | Delete

[28] Fix | Delete

# operators

[29] Fix | Delete

[30] Fix | Delete

FAILURE = "failure"

[31] Fix | Delete

SUCCESS = "success"

[32] Fix | Delete

[33] Fix | Delete

ANY = "any"

[34] Fix | Delete

ANY_ALL = "any_all"

[35] Fix | Delete

ASSERT = "assert"

[36] Fix | Delete

ASSERT_NOT = "assert_not"

[37] Fix | Delete

AT = "at"

[38] Fix | Delete

BIGCHARSET = "bigcharset"

[39] Fix | Delete

BRANCH = "branch"

[40] Fix | Delete

CALL = "call"

[41] Fix | Delete

CATEGORY = "category"

[42] Fix | Delete

CHARSET = "charset"

[43] Fix | Delete

GROUPREF = "groupref"

[44] Fix | Delete

GROUPREF_IGNORE = "groupref_ignore"

[45] Fix | Delete

GROUPREF_EXISTS = "groupref_exists"

[46] Fix | Delete

IN = "in"

[47] Fix | Delete

IN_IGNORE = "in_ignore"

[48] Fix | Delete

INFO = "info"

[49] Fix | Delete

JUMP = "jump"

[50] Fix | Delete

LITERAL = "literal"

[51] Fix | Delete

LITERAL_IGNORE = "literal_ignore"

[52] Fix | Delete

MARK = "mark"

[53] Fix | Delete

MAX_REPEAT = "max_repeat"

[54] Fix | Delete

MAX_UNTIL = "max_until"

[55] Fix | Delete

MIN_REPEAT = "min_repeat"

[56] Fix | Delete

MIN_UNTIL = "min_until"

[57] Fix | Delete

NEGATE = "negate"

[58] Fix | Delete

NOT_LITERAL = "not_literal"

[59] Fix | Delete

NOT_LITERAL_IGNORE = "not_literal_ignore"

[60] Fix | Delete

RANGE = "range"

[61] Fix | Delete

REPEAT = "repeat"

[62] Fix | Delete

REPEAT_ONE = "repeat_one"

[63] Fix | Delete

SUBPATTERN = "subpattern"

[64] Fix | Delete

MIN_REPEAT_ONE = "min_repeat_one"

[65] Fix | Delete

[66] Fix | Delete

# positions

[67] Fix | Delete

AT_BEGINNING = "at_beginning"

[68] Fix | Delete

AT_BEGINNING_LINE = "at_beginning_line"

[69] Fix | Delete

AT_BEGINNING_STRING = "at_beginning_string"

[70] Fix | Delete

AT_BOUNDARY = "at_boundary"

[71] Fix | Delete

AT_NON_BOUNDARY = "at_non_boundary"

[72] Fix | Delete

AT_END = "at_end"

[73] Fix | Delete

AT_END_LINE = "at_end_line"

[74] Fix | Delete

AT_END_STRING = "at_end_string"

[75] Fix | Delete

AT_LOC_BOUNDARY = "at_loc_boundary"

[76] Fix | Delete

AT_LOC_NON_BOUNDARY = "at_loc_non_boundary"

[77] Fix | Delete

AT_UNI_BOUNDARY = "at_uni_boundary"

[78] Fix | Delete

AT_UNI_NON_BOUNDARY = "at_uni_non_boundary"

[79] Fix | Delete

[80] Fix | Delete

# categories

[81] Fix | Delete

CATEGORY_DIGIT = "category_digit"

[82] Fix | Delete

CATEGORY_NOT_DIGIT = "category_not_digit"

[83] Fix | Delete

CATEGORY_SPACE = "category_space"

[84] Fix | Delete

CATEGORY_NOT_SPACE = "category_not_space"

[85] Fix | Delete

CATEGORY_WORD = "category_word"

[86] Fix | Delete

CATEGORY_NOT_WORD = "category_not_word"

[87] Fix | Delete

CATEGORY_LINEBREAK = "category_linebreak"

[88] Fix | Delete

CATEGORY_NOT_LINEBREAK = "category_not_linebreak"

[89] Fix | Delete

CATEGORY_LOC_WORD = "category_loc_word"

[90] Fix | Delete

CATEGORY_LOC_NOT_WORD = "category_loc_not_word"

[91] Fix | Delete

CATEGORY_UNI_DIGIT = "category_uni_digit"

[92] Fix | Delete

CATEGORY_UNI_NOT_DIGIT = "category_uni_not_digit"

[93] Fix | Delete

CATEGORY_UNI_SPACE = "category_uni_space"

[94] Fix | Delete

CATEGORY_UNI_NOT_SPACE = "category_uni_not_space"

[95] Fix | Delete

CATEGORY_UNI_WORD = "category_uni_word"

[96] Fix | Delete

CATEGORY_UNI_NOT_WORD = "category_uni_not_word"

[97] Fix | Delete

CATEGORY_UNI_LINEBREAK = "category_uni_linebreak"

[98] Fix | Delete

CATEGORY_UNI_NOT_LINEBREAK = "category_uni_not_linebreak"

[99] Fix | Delete

[100] Fix | Delete

OPCODES = [

[101] Fix | Delete

[102] Fix | Delete

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

[103] Fix | Delete

FAILURE, SUCCESS,

[104] Fix | Delete

[105] Fix | Delete

ANY, ANY_ALL,

[106] Fix | Delete

ASSERT, ASSERT_NOT,

[107] Fix | Delete

AT,

[108] Fix | Delete

BRANCH,

[109] Fix | Delete

CALL,

[110] Fix | Delete

CATEGORY,

[111] Fix | Delete

CHARSET, BIGCHARSET,

[112] Fix | Delete

GROUPREF, GROUPREF_EXISTS, GROUPREF_IGNORE,

[113] Fix | Delete

IN, IN_IGNORE,

[114] Fix | Delete

INFO,

[115] Fix | Delete

JUMP,

[116] Fix | Delete

LITERAL, LITERAL_IGNORE,

[117] Fix | Delete

MARK,

[118] Fix | Delete

MAX_UNTIL,

[119] Fix | Delete

MIN_UNTIL,

[120] Fix | Delete

NOT_LITERAL, NOT_LITERAL_IGNORE,

[121] Fix | Delete

NEGATE,

[122] Fix | Delete

RANGE,

[123] Fix | Delete

REPEAT,

[124] Fix | Delete

REPEAT_ONE,

[125] Fix | Delete

SUBPATTERN,

[126] Fix | Delete

MIN_REPEAT_ONE

[127] Fix | Delete

[128] Fix | Delete

]

[129] Fix | Delete

[130] Fix | Delete

ATCODES = [

[131] Fix | Delete

AT_BEGINNING, AT_BEGINNING_LINE, AT_BEGINNING_STRING, AT_BOUNDARY,

[132] Fix | Delete

AT_NON_BOUNDARY, AT_END, AT_END_LINE, AT_END_STRING,

[133] Fix | Delete

AT_LOC_BOUNDARY, AT_LOC_NON_BOUNDARY, AT_UNI_BOUNDARY,

[134] Fix | Delete

AT_UNI_NON_BOUNDARY

[135] Fix | Delete

]

[136] Fix | Delete

[137] Fix | Delete

CHCODES = [

[138] Fix | Delete

CATEGORY_DIGIT, CATEGORY_NOT_DIGIT, CATEGORY_SPACE,

[139] Fix | Delete

CATEGORY_NOT_SPACE, CATEGORY_WORD, CATEGORY_NOT_WORD,

[140] Fix | Delete

CATEGORY_LINEBREAK, CATEGORY_NOT_LINEBREAK, CATEGORY_LOC_WORD,

[141] Fix | Delete

CATEGORY_LOC_NOT_WORD, CATEGORY_UNI_DIGIT, CATEGORY_UNI_NOT_DIGIT,

[142] Fix | Delete

CATEGORY_UNI_SPACE, CATEGORY_UNI_NOT_SPACE, CATEGORY_UNI_WORD,

[143] Fix | Delete

CATEGORY_UNI_NOT_WORD, CATEGORY_UNI_LINEBREAK,

[144] Fix | Delete

CATEGORY_UNI_NOT_LINEBREAK

[145] Fix | Delete

]

[146] Fix | Delete

[147] Fix | Delete

def makedict(list):

[148] Fix | Delete

d = {}

[149] Fix | Delete

i = 0

[150] Fix | Delete

for item in list:

[151] Fix | Delete

d[item] = i

[152] Fix | Delete

i = i + 1

[153] Fix | Delete

return d

[154] Fix | Delete

[155] Fix | Delete

OPCODES = makedict(OPCODES)

[156] Fix | Delete

ATCODES = makedict(ATCODES)

[157] Fix | Delete

CHCODES = makedict(CHCODES)

[158] Fix | Delete

[159] Fix | Delete

# replacement operations for "ignore case" mode

[160] Fix | Delete

OP_IGNORE = {

[161] Fix | Delete

GROUPREF: GROUPREF_IGNORE,

[162] Fix | Delete

IN: IN_IGNORE,

[163] Fix | Delete

LITERAL: LITERAL_IGNORE,

[164] Fix | Delete

NOT_LITERAL: NOT_LITERAL_IGNORE

[165] Fix | Delete

}

[166] Fix | Delete

[167] Fix | Delete

AT_MULTILINE = {

[168] Fix | Delete

AT_BEGINNING: AT_BEGINNING_LINE,

[169] Fix | Delete

AT_END: AT_END_LINE

[170] Fix | Delete

}

[171] Fix | Delete

[172] Fix | Delete

AT_LOCALE = {

[173] Fix | Delete

AT_BOUNDARY: AT_LOC_BOUNDARY,

[174] Fix | Delete

AT_NON_BOUNDARY: AT_LOC_NON_BOUNDARY

[175] Fix | Delete

}

[176] Fix | Delete

[177] Fix | Delete

AT_UNICODE = {

[178] Fix | Delete

AT_BOUNDARY: AT_UNI_BOUNDARY,

[179] Fix | Delete

AT_NON_BOUNDARY: AT_UNI_NON_BOUNDARY

[180] Fix | Delete

}

[181] Fix | Delete

[182] Fix | Delete

CH_LOCALE = {

[183] Fix | Delete

CATEGORY_DIGIT: CATEGORY_DIGIT,

[184] Fix | Delete

CATEGORY_NOT_DIGIT: CATEGORY_NOT_DIGIT,

[185] Fix | Delete

CATEGORY_SPACE: CATEGORY_SPACE,

[186] Fix | Delete

CATEGORY_NOT_SPACE: CATEGORY_NOT_SPACE,

[187] Fix | Delete

CATEGORY_WORD: CATEGORY_LOC_WORD,

[188] Fix | Delete

CATEGORY_NOT_WORD: CATEGORY_LOC_NOT_WORD,

[189] Fix | Delete

CATEGORY_LINEBREAK: CATEGORY_LINEBREAK,

[190] Fix | Delete

CATEGORY_NOT_LINEBREAK: CATEGORY_NOT_LINEBREAK

[191] Fix | Delete

}

[192] Fix | Delete

[193] Fix | Delete

CH_UNICODE = {

[194] Fix | Delete

CATEGORY_DIGIT: CATEGORY_UNI_DIGIT,

[195] Fix | Delete

CATEGORY_NOT_DIGIT: CATEGORY_UNI_NOT_DIGIT,

[196] Fix | Delete

CATEGORY_SPACE: CATEGORY_UNI_SPACE,

[197] Fix | Delete

CATEGORY_NOT_SPACE: CATEGORY_UNI_NOT_SPACE,

[198] Fix | Delete

CATEGORY_WORD: CATEGORY_UNI_WORD,

[199] Fix | Delete

CATEGORY_NOT_WORD: CATEGORY_UNI_NOT_WORD,

[200] Fix | Delete

CATEGORY_LINEBREAK: CATEGORY_UNI_LINEBREAK,

[201] Fix | Delete

CATEGORY_NOT_LINEBREAK: CATEGORY_UNI_NOT_LINEBREAK

[202] Fix | Delete

}

[203] Fix | Delete

[204] Fix | Delete

# flags

[205] Fix | Delete

SRE_FLAG_TEMPLATE = 1 # template mode (disable backtracking)

[206] Fix | Delete

SRE_FLAG_IGNORECASE = 2 # case insensitive

[207] Fix | Delete

SRE_FLAG_LOCALE = 4 # honour system locale

[208] Fix | Delete

SRE_FLAG_MULTILINE = 8 # treat target as multiline string

[209] Fix | Delete

SRE_FLAG_DOTALL = 16 # treat target as a single string

[210] Fix | Delete

SRE_FLAG_UNICODE = 32 # use unicode locale

[211] Fix | Delete

SRE_FLAG_VERBOSE = 64 # ignore whitespace and comments

[212] Fix | Delete

SRE_FLAG_DEBUG = 128 # debugging

[213] Fix | Delete

[214] Fix | Delete

# flags for INFO primitive

[215] Fix | Delete

SRE_INFO_PREFIX = 1 # has prefix

[216] Fix | Delete

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

[217] Fix | Delete

SRE_INFO_CHARSET = 4 # pattern starts with character from given set

[218] Fix | Delete

[219] Fix | Delete

if __name__ == "__main__":

[220] Fix | Delete

def dump(f, d, prefix):

[221] Fix | Delete

items = d.items()

[222] Fix | Delete

items.sort(key=lambda a: a[1])

[223] Fix | Delete

for k, v in items:

[224] Fix | Delete

f.write("#define %s_%s %s\n" % (prefix, k.upper(), v))

[225] Fix | Delete

f = open("sre_constants.h", "w")

[226] Fix | Delete

f.write("""\

[227] Fix | Delete

/*

[228] Fix | Delete

* Secret Labs' Regular Expression Engine

[229] Fix | Delete

*

[230] Fix | Delete

* regular expression matching engine

[231] Fix | Delete

*

[232] Fix | Delete

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

[233] Fix | Delete

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

[234] Fix | Delete

*

[235] Fix | Delete

* Copyright (c) 1997-2001 by Secret Labs AB. All rights reserved.

[236] Fix | Delete

*

[237] Fix | Delete

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

[238] Fix | Delete

*/

[239] Fix | Delete

[240] Fix | Delete

""")

[241] Fix | Delete

[242] Fix | Delete

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

[243] Fix | Delete

[244] Fix | Delete

dump(f, OPCODES, "SRE_OP")

[245] Fix | Delete

dump(f, ATCODES, "SRE")

[246] Fix | Delete

dump(f, CHCODES, "SRE")

[247] Fix | Delete

[248] Fix | Delete

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

[249] Fix | Delete

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

[250] Fix | Delete

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

[251] Fix | Delete

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

[252] Fix | Delete

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

[253] Fix | Delete

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

[254] Fix | Delete

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

[255] Fix | Delete

[256] Fix | Delete

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

[257] Fix | Delete

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

[258] Fix | Delete

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

[259] Fix | Delete

[260] Fix | Delete

f.close()

[261] Fix | Delete

print "done"

[262] Fix | Delete

[263] Fix | Delete