blob: 506661593740324f1822d6730f68902e411c546a [file] [log] [blame]
Guido van Rossum7627c0d2000-03-31 14:58:54 +00001#
2# Secret Labs' Regular Expression Engine
Guido van Rossum7627c0d2000-03-31 14:58:54 +00003#
Andrew M. Kuchling815d5b92000-06-09 14:08:07 +00004# convert re-style regular expression to sre pattern
Guido van Rossum7627c0d2000-03-31 14:58:54 +00005#
Fredrik Lundh770617b2001-01-14 15:06:11 +00006# Copyright (c) 1998-2001 by Secret Labs AB. All rights reserved.
Guido van Rossum7627c0d2000-03-31 14:58:54 +00007#
Fredrik Lundh29c4ba92000-08-01 18:20:07 +00008# See the sre.py file for information on usage and redistribution.
Guido van Rossum7627c0d2000-03-31 14:58:54 +00009#
10
Fred Drakeb8f22742001-09-04 19:10:20 +000011"""Internal support module for sre"""
12
Fredrik Lundh470ea5a2001-01-14 21:00:44 +000013# XXX: show string offset and offending character for all errors
14
Barry Warsaw8bee7612004-08-25 02:22:30 +000015import sys
Guido van Rossum7627c0d2000-03-31 14:58:54 +000016
17from sre_constants import *
18
19SPECIAL_CHARS = ".\\[{()*+?^$|"
Fredrik Lundh143328b2000-09-02 11:03:34 +000020REPEAT_CHARS = "*+?{"
Guido van Rossum7627c0d2000-03-31 14:58:54 +000021
Tim Peters17289422000-09-02 07:44:32 +000022DIGITS = tuple("0123456789")
Guido van Rossumb81e70e2000-04-10 17:10:48 +000023
Fredrik Lundh75f2d672000-06-29 11:34:28 +000024OCTDIGITS = tuple("01234567")
25HEXDIGITS = tuple("0123456789abcdefABCDEF")
Guido van Rossum7627c0d2000-03-31 14:58:54 +000026
Fredrik Lundh8a3ebf82000-07-23 21:46:17 +000027WHITESPACE = tuple(" \t\n\r\v\f")
Andrew M. Kuchling815d5b92000-06-09 14:08:07 +000028
Guido van Rossum7627c0d2000-03-31 14:58:54 +000029ESCAPES = {
Fredrik Lundhf2989b22001-02-18 12:05:16 +000030 r"\a": (LITERAL, ord("\a")),
31 r"\b": (LITERAL, ord("\b")),
32 r"\f": (LITERAL, ord("\f")),
33 r"\n": (LITERAL, ord("\n")),
34 r"\r": (LITERAL, ord("\r")),
35 r"\t": (LITERAL, ord("\t")),
36 r"\v": (LITERAL, ord("\v")),
Fredrik Lundh0640e112000-06-30 13:55:15 +000037 r"\\": (LITERAL, ord("\\"))
Guido van Rossum7627c0d2000-03-31 14:58:54 +000038}
39
40CATEGORIES = {
Fredrik Lundh770617b2001-01-14 15:06:11 +000041 r"\A": (AT, AT_BEGINNING_STRING), # start of string
Fredrik Lundh01016fe2000-06-30 00:27:46 +000042 r"\b": (AT, AT_BOUNDARY),
43 r"\B": (AT, AT_NON_BOUNDARY),
44 r"\d": (IN, [(CATEGORY, CATEGORY_DIGIT)]),
45 r"\D": (IN, [(CATEGORY, CATEGORY_NOT_DIGIT)]),
46 r"\s": (IN, [(CATEGORY, CATEGORY_SPACE)]),
47 r"\S": (IN, [(CATEGORY, CATEGORY_NOT_SPACE)]),
48 r"\w": (IN, [(CATEGORY, CATEGORY_WORD)]),
49 r"\W": (IN, [(CATEGORY, CATEGORY_NOT_WORD)]),
Fredrik Lundh770617b2001-01-14 15:06:11 +000050 r"\Z": (AT, AT_END_STRING), # end of string
Guido van Rossum7627c0d2000-03-31 14:58:54 +000051}
52
Andrew M. Kuchling815d5b92000-06-09 14:08:07 +000053FLAGS = {
Fredrik Lundh436c3d582000-06-29 08:58:44 +000054 # standard flags
Andrew M. Kuchling815d5b92000-06-09 14:08:07 +000055 "i": SRE_FLAG_IGNORECASE,
56 "L": SRE_FLAG_LOCALE,
57 "m": SRE_FLAG_MULTILINE,
58 "s": SRE_FLAG_DOTALL,
Andrew M. Kuchling815d5b92000-06-09 14:08:07 +000059 "x": SRE_FLAG_VERBOSE,
Fredrik Lundh436c3d582000-06-29 08:58:44 +000060 # extensions
61 "t": SRE_FLAG_TEMPLATE,
62 "u": SRE_FLAG_UNICODE,
Andrew M. Kuchling815d5b92000-06-09 14:08:07 +000063}
64
Fredrik Lundh8a3ebf82000-07-23 21:46:17 +000065class Pattern:
66 # master pattern object. keeps track of global attributes
Guido van Rossum7627c0d2000-03-31 14:58:54 +000067 def __init__(self):
Fredrik Lundh90a07912000-06-30 07:50:59 +000068 self.flags = 0
Fredrik Lundhebc37b22000-10-28 19:30:41 +000069 self.open = []
Fredrik Lundh90a07912000-06-30 07:50:59 +000070 self.groups = 1
71 self.groupdict = {}
Fredrik Lundhebc37b22000-10-28 19:30:41 +000072 def opengroup(self, name=None):
Fredrik Lundh90a07912000-06-30 07:50:59 +000073 gid = self.groups
74 self.groups = gid + 1
Raymond Hettingerf13eb552002-06-02 00:40:05 +000075 if name is not None:
Tim Peters75335872001-11-03 19:35:43 +000076 ogid = self.groupdict.get(name, None)
77 if ogid is not None:
Fredrik Lundh82b23072001-12-09 16:13:15 +000078 raise error, ("redefinition of group name %s as group %d; "
79 "was group %d" % (repr(name), gid, ogid))
Fredrik Lundh90a07912000-06-30 07:50:59 +000080 self.groupdict[name] = gid
Fredrik Lundhebc37b22000-10-28 19:30:41 +000081 self.open.append(gid)
Fredrik Lundh90a07912000-06-30 07:50:59 +000082 return gid
Fredrik Lundhebc37b22000-10-28 19:30:41 +000083 def closegroup(self, gid):
84 self.open.remove(gid)
85 def checkgroup(self, gid):
86 return gid < self.groups and gid not in self.open
Guido van Rossum7627c0d2000-03-31 14:58:54 +000087
88class SubPattern:
89 # a subpattern, in intermediate form
90 def __init__(self, pattern, data=None):
Fredrik Lundh90a07912000-06-30 07:50:59 +000091 self.pattern = pattern
Raymond Hettingerf13eb552002-06-02 00:40:05 +000092 if data is None:
Fredrik Lundh90a07912000-06-30 07:50:59 +000093 data = []
94 self.data = data
95 self.width = None
Fredrik Lundh8a3ebf82000-07-23 21:46:17 +000096 def dump(self, level=0):
97 nl = 1
Raymond Hettinger968c56a2004-03-26 23:24:00 +000098 seqtypes = type(()), type([])
Fredrik Lundh8a3ebf82000-07-23 21:46:17 +000099 for op, av in self.data:
100 print level*" " + op,; nl = 0
101 if op == "in":
102 # member sublanguage
103 print; nl = 1
104 for op, a in av:
105 print (level+1)*" " + op, a
106 elif op == "branch":
107 print; nl = 1
108 i = 0
109 for a in av[1]:
110 if i > 0:
111 print level*" " + "or"
112 a.dump(level+1); nl = 1
113 i = i + 1
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000114 elif type(av) in seqtypes:
Fredrik Lundh8a3ebf82000-07-23 21:46:17 +0000115 for a in av:
116 if isinstance(a, SubPattern):
117 if not nl: print
118 a.dump(level+1); nl = 1
119 else:
120 print a, ; nl = 0
121 else:
122 print av, ; nl = 0
123 if not nl: print
Guido van Rossum7627c0d2000-03-31 14:58:54 +0000124 def __repr__(self):
Fredrik Lundh90a07912000-06-30 07:50:59 +0000125 return repr(self.data)
Guido van Rossum7627c0d2000-03-31 14:58:54 +0000126 def __len__(self):
Fredrik Lundh90a07912000-06-30 07:50:59 +0000127 return len(self.data)
Guido van Rossum7627c0d2000-03-31 14:58:54 +0000128 def __delitem__(self, index):
Fredrik Lundh90a07912000-06-30 07:50:59 +0000129 del self.data[index]
Guido van Rossum7627c0d2000-03-31 14:58:54 +0000130 def __getitem__(self, index):
Fredrik Lundh90a07912000-06-30 07:50:59 +0000131 return self.data[index]
Guido van Rossum7627c0d2000-03-31 14:58:54 +0000132 def __setitem__(self, index, code):
Fredrik Lundh90a07912000-06-30 07:50:59 +0000133 self.data[index] = code
Guido van Rossum7627c0d2000-03-31 14:58:54 +0000134 def __getslice__(self, start, stop):
Fredrik Lundh90a07912000-06-30 07:50:59 +0000135 return SubPattern(self.pattern, self.data[start:stop])
Guido van Rossum7627c0d2000-03-31 14:58:54 +0000136 def insert(self, index, code):
Fredrik Lundh90a07912000-06-30 07:50:59 +0000137 self.data.insert(index, code)
Guido van Rossum7627c0d2000-03-31 14:58:54 +0000138 def append(self, code):
Fredrik Lundh90a07912000-06-30 07:50:59 +0000139 self.data.append(code)
Guido van Rossum7627c0d2000-03-31 14:58:54 +0000140 def getwidth(self):
Fredrik Lundh90a07912000-06-30 07:50:59 +0000141 # determine the width (min, max) for this subpattern
142 if self.width:
143 return self.width
144 lo = hi = 0L
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000145 UNITCODES = (ANY, RANGE, IN, LITERAL, NOT_LITERAL, CATEGORY)
146 REPEATCODES = (MIN_REPEAT, MAX_REPEAT)
Fredrik Lundh90a07912000-06-30 07:50:59 +0000147 for op, av in self.data:
148 if op is BRANCH:
Fredrik Lundh2f2c67d2000-08-01 21:05:41 +0000149 i = sys.maxint
150 j = 0
Fredrik Lundh90a07912000-06-30 07:50:59 +0000151 for av in av[1]:
Fredrik Lundh2f2c67d2000-08-01 21:05:41 +0000152 l, h = av.getwidth()
153 i = min(i, l)
Fredrik Lundhe1869832000-08-01 22:47:49 +0000154 j = max(j, h)
Fredrik Lundh90a07912000-06-30 07:50:59 +0000155 lo = lo + i
156 hi = hi + j
157 elif op is CALL:
158 i, j = av.getwidth()
159 lo = lo + i
160 hi = hi + j
161 elif op is SUBPATTERN:
162 i, j = av[1].getwidth()
163 lo = lo + i
164 hi = hi + j
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000165 elif op in REPEATCODES:
Fredrik Lundh90a07912000-06-30 07:50:59 +0000166 i, j = av[2].getwidth()
167 lo = lo + long(i) * av[0]
168 hi = hi + long(j) * av[1]
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000169 elif op in UNITCODES:
Fredrik Lundh90a07912000-06-30 07:50:59 +0000170 lo = lo + 1
171 hi = hi + 1
172 elif op == SUCCESS:
173 break
174 self.width = int(min(lo, sys.maxint)), int(min(hi, sys.maxint))
175 return self.width
Guido van Rossum7627c0d2000-03-31 14:58:54 +0000176
177class Tokenizer:
178 def __init__(self, string):
Fredrik Lundh90a07912000-06-30 07:50:59 +0000179 self.string = string
Fredrik Lundhc13222c2000-07-01 23:49:14 +0000180 self.index = 0
181 self.__next()
Guido van Rossum7627c0d2000-03-31 14:58:54 +0000182 def __next(self):
Fredrik Lundh90a07912000-06-30 07:50:59 +0000183 if self.index >= len(self.string):
Fredrik Lundhc13222c2000-07-01 23:49:14 +0000184 self.next = None
185 return
Fredrik Lundh90a07912000-06-30 07:50:59 +0000186 char = self.string[self.index]
187 if char[0] == "\\":
188 try:
189 c = self.string[self.index + 1]
190 except IndexError:
Fredrik Lundh8a0232d2001-11-02 13:59:51 +0000191 raise error, "bogus escape (end of line)"
Fredrik Lundh90a07912000-06-30 07:50:59 +0000192 char = char + c
193 self.index = self.index + len(char)
Fredrik Lundhc13222c2000-07-01 23:49:14 +0000194 self.next = char
Fredrik Lundh0c4fdba2000-08-31 22:57:55 +0000195 def match(self, char, skip=1):
Fredrik Lundh90a07912000-06-30 07:50:59 +0000196 if char == self.next:
Fredrik Lundh0c4fdba2000-08-31 22:57:55 +0000197 if skip:
198 self.__next()
Fredrik Lundh90a07912000-06-30 07:50:59 +0000199 return 1
200 return 0
Guido van Rossum7627c0d2000-03-31 14:58:54 +0000201 def get(self):
Fredrik Lundh90a07912000-06-30 07:50:59 +0000202 this = self.next
Fredrik Lundhc13222c2000-07-01 23:49:14 +0000203 self.__next()
Fredrik Lundh90a07912000-06-30 07:50:59 +0000204 return this
Fredrik Lundhc13222c2000-07-01 23:49:14 +0000205 def tell(self):
206 return self.index, self.next
207 def seek(self, index):
208 self.index, self.next = index
Guido van Rossum7627c0d2000-03-31 14:58:54 +0000209
Fredrik Lundh4781b072000-06-29 12:38:45 +0000210def isident(char):
211 return "a" <= char <= "z" or "A" <= char <= "Z" or char == "_"
212
213def isdigit(char):
214 return "0" <= char <= "9"
215
216def isname(name):
217 # check that group name is a valid string
Fredrik Lundh4781b072000-06-29 12:38:45 +0000218 if not isident(name[0]):
Guido van Rossum8ca162f2002-04-07 06:36:23 +0000219 return False
Gustavo Niemeyera01a2ee2004-09-03 17:06:10 +0000220 for char in name[1:]:
Fredrik Lundh90a07912000-06-30 07:50:59 +0000221 if not isident(char) and not isdigit(char):
Guido van Rossum8ca162f2002-04-07 06:36:23 +0000222 return False
223 return True
Fredrik Lundh4781b072000-06-29 12:38:45 +0000224
Andrew M. Kuchling815d5b92000-06-09 14:08:07 +0000225def _class_escape(source, escape):
226 # handle escape code inside character class
227 code = ESCAPES.get(escape)
Guido van Rossum7627c0d2000-03-31 14:58:54 +0000228 if code:
Fredrik Lundh90a07912000-06-30 07:50:59 +0000229 return code
Andrew M. Kuchling815d5b92000-06-09 14:08:07 +0000230 code = CATEGORIES.get(escape)
231 if code:
Fredrik Lundh90a07912000-06-30 07:50:59 +0000232 return code
Guido van Rossum7627c0d2000-03-31 14:58:54 +0000233 try:
Gustavo Niemeyera01a2ee2004-09-03 17:06:10 +0000234 c = escape[1:2]
235 if c == "x":
Fredrik Lundh0c4fdba2000-08-31 22:57:55 +0000236 # hexadecimal escape (exactly two digits)
237 while source.next in HEXDIGITS and len(escape) < 4:
Fredrik Lundh90a07912000-06-30 07:50:59 +0000238 escape = escape + source.get()
239 escape = escape[2:]
Fredrik Lundh0c4fdba2000-08-31 22:57:55 +0000240 if len(escape) != 2:
241 raise error, "bogus escape: %s" % repr("\\" + escape)
Barry Warsaw8bee7612004-08-25 02:22:30 +0000242 return LITERAL, int(escape, 16) & 0xff
Gustavo Niemeyera01a2ee2004-09-03 17:06:10 +0000243 elif c in OCTDIGITS:
Fredrik Lundh0c4fdba2000-08-31 22:57:55 +0000244 # octal escape (up to three digits)
Gustavo Niemeyera01a2ee2004-09-03 17:06:10 +0000245 while source.next in OCTDIGITS and len(escape) < 4:
Fredrik Lundh90a07912000-06-30 07:50:59 +0000246 escape = escape + source.get()
247 escape = escape[1:]
Barry Warsaw8bee7612004-08-25 02:22:30 +0000248 return LITERAL, int(escape, 8) & 0xff
Gustavo Niemeyera01a2ee2004-09-03 17:06:10 +0000249 elif c in DIGITS:
250 raise error, "bogus escape: %s" % repr(escape)
Fredrik Lundh90a07912000-06-30 07:50:59 +0000251 if len(escape) == 2:
Fredrik Lundh0640e112000-06-30 13:55:15 +0000252 return LITERAL, ord(escape[1])
Guido van Rossum7627c0d2000-03-31 14:58:54 +0000253 except ValueError:
Fredrik Lundh90a07912000-06-30 07:50:59 +0000254 pass
Fredrik Lundh436c3d582000-06-29 08:58:44 +0000255 raise error, "bogus escape: %s" % repr(escape)
Guido van Rossum7627c0d2000-03-31 14:58:54 +0000256
Andrew M. Kuchling815d5b92000-06-09 14:08:07 +0000257def _escape(source, escape, state):
258 # handle escape code in expression
259 code = CATEGORIES.get(escape)
260 if code:
Fredrik Lundh90a07912000-06-30 07:50:59 +0000261 return code
Andrew M. Kuchling815d5b92000-06-09 14:08:07 +0000262 code = ESCAPES.get(escape)
263 if code:
Fredrik Lundh90a07912000-06-30 07:50:59 +0000264 return code
Andrew M. Kuchling815d5b92000-06-09 14:08:07 +0000265 try:
Gustavo Niemeyera01a2ee2004-09-03 17:06:10 +0000266 c = escape[1:2]
267 if c == "x":
Fredrik Lundh0c4fdba2000-08-31 22:57:55 +0000268 # hexadecimal escape
269 while source.next in HEXDIGITS and len(escape) < 4:
Fredrik Lundh90a07912000-06-30 07:50:59 +0000270 escape = escape + source.get()
Fredrik Lundh143328b2000-09-02 11:03:34 +0000271 if len(escape) != 4:
272 raise ValueError
Barry Warsaw8bee7612004-08-25 02:22:30 +0000273 return LITERAL, int(escape[2:], 16) & 0xff
Gustavo Niemeyera01a2ee2004-09-03 17:06:10 +0000274 elif c == "0":
Fredrik Lundh0c4fdba2000-08-31 22:57:55 +0000275 # octal escape
Fredrik Lundh143328b2000-09-02 11:03:34 +0000276 while source.next in OCTDIGITS and len(escape) < 4:
Fredrik Lundh0c4fdba2000-08-31 22:57:55 +0000277 escape = escape + source.get()
Barry Warsaw8bee7612004-08-25 02:22:30 +0000278 return LITERAL, int(escape[1:], 8) & 0xff
Gustavo Niemeyera01a2ee2004-09-03 17:06:10 +0000279 elif c in DIGITS:
Fredrik Lundh0c4fdba2000-08-31 22:57:55 +0000280 # octal escape *or* decimal group reference (sigh)
Fredrik Lundh0c4fdba2000-08-31 22:57:55 +0000281 if source.next in DIGITS:
282 escape = escape + source.get()
Fredrik Lundh143328b2000-09-02 11:03:34 +0000283 if (escape[1] in OCTDIGITS and escape[2] in OCTDIGITS and
284 source.next in OCTDIGITS):
Fredrik Lundh0c4fdba2000-08-31 22:57:55 +0000285 # got three octal digits; this is an octal escape
Fredrik Lundh90a07912000-06-30 07:50:59 +0000286 escape = escape + source.get()
Barry Warsaw8bee7612004-08-25 02:22:30 +0000287 return LITERAL, int(escape[1:], 8) & 0xff
Gustavo Niemeyera01a2ee2004-09-03 17:06:10 +0000288 # not an octal escape, so this is a group reference
289 group = int(escape[1:])
290 if group < state.groups:
Fredrik Lundhebc37b22000-10-28 19:30:41 +0000291 if not state.checkgroup(group):
292 raise error, "cannot refer to open group"
Fredrik Lundh0c4fdba2000-08-31 22:57:55 +0000293 return GROUPREF, group
Fredrik Lundh143328b2000-09-02 11:03:34 +0000294 raise ValueError
Fredrik Lundh90a07912000-06-30 07:50:59 +0000295 if len(escape) == 2:
Fredrik Lundh0640e112000-06-30 13:55:15 +0000296 return LITERAL, ord(escape[1])
Andrew M. Kuchling815d5b92000-06-09 14:08:07 +0000297 except ValueError:
Fredrik Lundh90a07912000-06-30 07:50:59 +0000298 pass
Fredrik Lundh436c3d582000-06-29 08:58:44 +0000299 raise error, "bogus escape: %s" % repr(escape)
Andrew M. Kuchling815d5b92000-06-09 14:08:07 +0000300
Fredrik Lundh8a3ebf82000-07-23 21:46:17 +0000301def _parse_sub(source, state, nested=1):
302 # parse an alternation: a|b|c
Guido van Rossum7627c0d2000-03-31 14:58:54 +0000303
Fredrik Lundh8a3ebf82000-07-23 21:46:17 +0000304 items = []
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000305 itemsappend = items.append
306 sourcematch = source.match
Fredrik Lundh8a3ebf82000-07-23 21:46:17 +0000307 while 1:
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000308 itemsappend(_parse(source, state))
309 if sourcematch("|"):
Fredrik Lundh8a3ebf82000-07-23 21:46:17 +0000310 continue
311 if not nested:
312 break
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000313 if not source.next or sourcematch(")", 0):
Fredrik Lundh8a3ebf82000-07-23 21:46:17 +0000314 break
315 else:
316 raise error, "pattern not properly closed"
317
318 if len(items) == 1:
319 return items[0]
320
321 subpattern = SubPattern(state)
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000322 subpatternappend = subpattern.append
Andrew M. Kuchling815d5b92000-06-09 14:08:07 +0000323
Guido van Rossum7627c0d2000-03-31 14:58:54 +0000324 # check if all items share a common prefix
325 while 1:
Fredrik Lundh90a07912000-06-30 07:50:59 +0000326 prefix = None
327 for item in items:
328 if not item:
329 break
330 if prefix is None:
331 prefix = item[0]
332 elif item[0] != prefix:
333 break
334 else:
335 # all subitems start with a common "prefix".
336 # move it out of the branch
337 for item in items:
338 del item[0]
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000339 subpatternappend(prefix)
Fredrik Lundh90a07912000-06-30 07:50:59 +0000340 continue # check next one
341 break
Guido van Rossum7627c0d2000-03-31 14:58:54 +0000342
343 # check if the branch can be replaced by a character set
344 for item in items:
Fredrik Lundh90a07912000-06-30 07:50:59 +0000345 if len(item) != 1 or item[0][0] != LITERAL:
346 break
Guido van Rossum7627c0d2000-03-31 14:58:54 +0000347 else:
Fredrik Lundh90a07912000-06-30 07:50:59 +0000348 # we can store this as a character set instead of a
Fredrik Lundh8a3ebf82000-07-23 21:46:17 +0000349 # branch (the compiler may optimize this even more)
Fredrik Lundh90a07912000-06-30 07:50:59 +0000350 set = []
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000351 setappend = set.append
Fredrik Lundh90a07912000-06-30 07:50:59 +0000352 for item in items:
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000353 setappend(item[0])
354 subpatternappend((IN, set))
Fredrik Lundh90a07912000-06-30 07:50:59 +0000355 return subpattern
Guido van Rossum7627c0d2000-03-31 14:58:54 +0000356
357 subpattern.append((BRANCH, (None, items)))
Andrew M. Kuchling815d5b92000-06-09 14:08:07 +0000358 return subpattern
Guido van Rossum7627c0d2000-03-31 14:58:54 +0000359
Gustavo Niemeyerad3fc442003-10-17 22:13:16 +0000360def _parse_sub_cond(source, state, condgroup):
Tim Peters58eb11c2004-01-18 20:29:55 +0000361 item_yes = _parse(source, state)
Gustavo Niemeyerad3fc442003-10-17 22:13:16 +0000362 if source.match("|"):
Tim Peters58eb11c2004-01-18 20:29:55 +0000363 item_no = _parse(source, state)
Gustavo Niemeyerad3fc442003-10-17 22:13:16 +0000364 if source.match("|"):
365 raise error, "conditional backref with more than two branches"
366 else:
367 item_no = None
368 if source.next and not source.match(")", 0):
369 raise error, "pattern not properly closed"
370 subpattern = SubPattern(state)
371 subpattern.append((GROUPREF_EXISTS, (condgroup, item_yes, item_no)))
372 return subpattern
373
Fredrik Lundh55a4f4a2000-06-30 22:37:31 +0000374def _parse(source, state):
Fredrik Lundh8a3ebf82000-07-23 21:46:17 +0000375 # parse a simple pattern
Andrew M. Kuchling815d5b92000-06-09 14:08:07 +0000376 subpattern = SubPattern(state)
Guido van Rossum7627c0d2000-03-31 14:58:54 +0000377
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000378 # precompute constants into local variables
379 subpatternappend = subpattern.append
380 sourceget = source.get
381 sourcematch = source.match
382 _len = len
383 PATTERNENDERS = ("|", ")")
384 ASSERTCHARS = ("=", "!", "<")
385 LOOKBEHINDASSERTCHARS = ("=", "!")
386 REPEATCODES = (MIN_REPEAT, MAX_REPEAT)
387
Guido van Rossum7627c0d2000-03-31 14:58:54 +0000388 while 1:
389
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000390 if source.next in PATTERNENDERS:
Fredrik Lundh90a07912000-06-30 07:50:59 +0000391 break # end of subpattern
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000392 this = sourceget()
Fredrik Lundh90a07912000-06-30 07:50:59 +0000393 if this is None:
394 break # end of pattern
Guido van Rossum7627c0d2000-03-31 14:58:54 +0000395
Fredrik Lundh90a07912000-06-30 07:50:59 +0000396 if state.flags & SRE_FLAG_VERBOSE:
397 # skip whitespace and comments
398 if this in WHITESPACE:
399 continue
400 if this == "#":
401 while 1:
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000402 this = sourceget()
Fredrik Lundh90a07912000-06-30 07:50:59 +0000403 if this in (None, "\n"):
404 break
405 continue
Andrew M. Kuchling815d5b92000-06-09 14:08:07 +0000406
Fredrik Lundh90a07912000-06-30 07:50:59 +0000407 if this and this[0] not in SPECIAL_CHARS:
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000408 subpatternappend((LITERAL, ord(this)))
Guido van Rossum7627c0d2000-03-31 14:58:54 +0000409
Fredrik Lundh90a07912000-06-30 07:50:59 +0000410 elif this == "[":
411 # character set
412 set = []
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000413 setappend = set.append
414## if sourcematch(":"):
Fredrik Lundh90a07912000-06-30 07:50:59 +0000415## pass # handle character classes
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000416 if sourcematch("^"):
417 setappend((NEGATE, None))
Fredrik Lundh90a07912000-06-30 07:50:59 +0000418 # check remaining characters
419 start = set[:]
420 while 1:
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000421 this = sourceget()
Fredrik Lundh90a07912000-06-30 07:50:59 +0000422 if this == "]" and set != start:
423 break
424 elif this and this[0] == "\\":
425 code1 = _class_escape(source, this)
426 elif this:
Fredrik Lundh0640e112000-06-30 13:55:15 +0000427 code1 = LITERAL, ord(this)
Fredrik Lundh90a07912000-06-30 07:50:59 +0000428 else:
429 raise error, "unexpected end of regular expression"
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000430 if sourcematch("-"):
Fredrik Lundh90a07912000-06-30 07:50:59 +0000431 # potential range
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000432 this = sourceget()
Fredrik Lundh90a07912000-06-30 07:50:59 +0000433 if this == "]":
Fredrik Lundh025468d2000-10-07 10:16:19 +0000434 if code1[0] is IN:
435 code1 = code1[1][0]
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000436 setappend(code1)
437 setappend((LITERAL, ord("-")))
Fredrik Lundh90a07912000-06-30 07:50:59 +0000438 break
Guido van Rossum41c99e72003-04-14 17:59:34 +0000439 elif this:
Fredrik Lundh90a07912000-06-30 07:50:59 +0000440 if this[0] == "\\":
441 code2 = _class_escape(source, this)
442 else:
Fredrik Lundh0640e112000-06-30 13:55:15 +0000443 code2 = LITERAL, ord(this)
Fredrik Lundh90a07912000-06-30 07:50:59 +0000444 if code1[0] != LITERAL or code2[0] != LITERAL:
Fredrik Lundh470ea5a2001-01-14 21:00:44 +0000445 raise error, "bad character range"
Fredrik Lundh0c4fdba2000-08-31 22:57:55 +0000446 lo = code1[1]
447 hi = code2[1]
448 if hi < lo:
Fredrik Lundh470ea5a2001-01-14 21:00:44 +0000449 raise error, "bad character range"
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000450 setappend((RANGE, (lo, hi)))
Guido van Rossum41c99e72003-04-14 17:59:34 +0000451 else:
452 raise error, "unexpected end of regular expression"
Fredrik Lundh90a07912000-06-30 07:50:59 +0000453 else:
454 if code1[0] is IN:
455 code1 = code1[1][0]
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000456 setappend(code1)
Guido van Rossum7627c0d2000-03-31 14:58:54 +0000457
Fredrik Lundh770617b2001-01-14 15:06:11 +0000458 # XXX: <fl> should move set optimization to compiler!
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000459 if _len(set)==1 and set[0][0] is LITERAL:
460 subpatternappend(set[0]) # optimization
461 elif _len(set)==2 and set[0][0] is NEGATE and set[1][0] is LITERAL:
462 subpatternappend((NOT_LITERAL, set[1][1])) # optimization
Fredrik Lundh90a07912000-06-30 07:50:59 +0000463 else:
Fredrik Lundh770617b2001-01-14 15:06:11 +0000464 # XXX: <fl> should add charmap optimization here
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000465 subpatternappend((IN, set))
Guido van Rossum7627c0d2000-03-31 14:58:54 +0000466
Fredrik Lundh90a07912000-06-30 07:50:59 +0000467 elif this and this[0] in REPEAT_CHARS:
468 # repeat previous item
469 if this == "?":
470 min, max = 0, 1
471 elif this == "*":
472 min, max = 0, MAXREPEAT
Fredrik Lundhc0c7ee32001-02-18 21:04:48 +0000473
Fredrik Lundh90a07912000-06-30 07:50:59 +0000474 elif this == "+":
475 min, max = 1, MAXREPEAT
476 elif this == "{":
Fredrik Lundhc13222c2000-07-01 23:49:14 +0000477 here = source.tell()
Fredrik Lundh90a07912000-06-30 07:50:59 +0000478 min, max = 0, MAXREPEAT
479 lo = hi = ""
480 while source.next in DIGITS:
481 lo = lo + source.get()
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000482 if sourcematch(","):
Fredrik Lundh90a07912000-06-30 07:50:59 +0000483 while source.next in DIGITS:
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000484 hi = hi + sourceget()
Fredrik Lundh90a07912000-06-30 07:50:59 +0000485 else:
486 hi = lo
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000487 if not sourcematch("}"):
488 subpatternappend((LITERAL, ord(this)))
Fredrik Lundhc13222c2000-07-01 23:49:14 +0000489 source.seek(here)
490 continue
Fredrik Lundh90a07912000-06-30 07:50:59 +0000491 if lo:
Barry Warsaw8bee7612004-08-25 02:22:30 +0000492 min = int(lo)
Fredrik Lundh90a07912000-06-30 07:50:59 +0000493 if hi:
Barry Warsaw8bee7612004-08-25 02:22:30 +0000494 max = int(hi)
Fredrik Lundh470ea5a2001-01-14 21:00:44 +0000495 if max < min:
496 raise error, "bad repeat interval"
Fredrik Lundh90a07912000-06-30 07:50:59 +0000497 else:
498 raise error, "not supported"
499 # figure out which item to repeat
500 if subpattern:
501 item = subpattern[-1:]
502 else:
Fredrik Lundhc0c7ee32001-02-18 21:04:48 +0000503 item = None
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000504 if not item or (_len(item) == 1 and item[0][0] == AT):
Fredrik Lundh90a07912000-06-30 07:50:59 +0000505 raise error, "nothing to repeat"
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000506 if item[0][0] in REPEATCODES:
Fredrik Lundh470ea5a2001-01-14 21:00:44 +0000507 raise error, "multiple repeat"
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000508 if sourcematch("?"):
Fredrik Lundh90a07912000-06-30 07:50:59 +0000509 subpattern[-1] = (MIN_REPEAT, (min, max, item))
510 else:
511 subpattern[-1] = (MAX_REPEAT, (min, max, item))
Andrew M. Kuchling815d5b92000-06-09 14:08:07 +0000512
Fredrik Lundh90a07912000-06-30 07:50:59 +0000513 elif this == ".":
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000514 subpatternappend((ANY, None))
Andrew M. Kuchling815d5b92000-06-09 14:08:07 +0000515
Fredrik Lundh90a07912000-06-30 07:50:59 +0000516 elif this == "(":
517 group = 1
518 name = None
Gustavo Niemeyerad3fc442003-10-17 22:13:16 +0000519 condgroup = None
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000520 if sourcematch("?"):
Fredrik Lundh90a07912000-06-30 07:50:59 +0000521 group = 0
522 # options
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000523 if sourcematch("P"):
Fredrik Lundh90a07912000-06-30 07:50:59 +0000524 # python extensions
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000525 if sourcematch("<"):
Fredrik Lundh90a07912000-06-30 07:50:59 +0000526 # named group: skip forward to end of name
527 name = ""
528 while 1:
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000529 char = sourceget()
Fredrik Lundh90a07912000-06-30 07:50:59 +0000530 if char is None:
531 raise error, "unterminated name"
532 if char == ">":
533 break
534 name = name + char
535 group = 1
536 if not isname(name):
Fredrik Lundh470ea5a2001-01-14 21:00:44 +0000537 raise error, "bad character in group name"
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000538 elif sourcematch("="):
Fredrik Lundh90a07912000-06-30 07:50:59 +0000539 # named backreference
Fredrik Lundhb71624e2000-06-30 09:13:06 +0000540 name = ""
541 while 1:
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000542 char = sourceget()
Fredrik Lundhb71624e2000-06-30 09:13:06 +0000543 if char is None:
544 raise error, "unterminated name"
545 if char == ")":
546 break
547 name = name + char
548 if not isname(name):
Fredrik Lundh470ea5a2001-01-14 21:00:44 +0000549 raise error, "bad character in group name"
Fredrik Lundhb71624e2000-06-30 09:13:06 +0000550 gid = state.groupdict.get(name)
551 if gid is None:
552 raise error, "unknown group name"
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000553 subpatternappend((GROUPREF, gid))
Fredrik Lundh7cafe4d2000-07-02 17:33:27 +0000554 continue
Fredrik Lundh90a07912000-06-30 07:50:59 +0000555 else:
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000556 char = sourceget()
Fredrik Lundh90a07912000-06-30 07:50:59 +0000557 if char is None:
558 raise error, "unexpected end of pattern"
559 raise error, "unknown specifier: ?P%s" % char
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000560 elif sourcematch(":"):
Fredrik Lundh90a07912000-06-30 07:50:59 +0000561 # non-capturing group
562 group = 2
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000563 elif sourcematch("#"):
Fredrik Lundh90a07912000-06-30 07:50:59 +0000564 # comment
565 while 1:
566 if source.next is None or source.next == ")":
567 break
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000568 sourceget()
569 if not sourcematch(")"):
Fredrik Lundh0c4fdba2000-08-31 22:57:55 +0000570 raise error, "unbalanced parenthesis"
571 continue
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000572 elif source.next in ASSERTCHARS:
Fredrik Lundh43b3b492000-06-30 10:41:31 +0000573 # lookahead assertions
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000574 char = sourceget()
Fredrik Lundh6f013982000-07-03 18:44:21 +0000575 dir = 1
576 if char == "<":
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000577 if source.next not in LOOKBEHINDASSERTCHARS:
Fredrik Lundh6f013982000-07-03 18:44:21 +0000578 raise error, "syntax error"
579 dir = -1 # lookbehind
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000580 char = sourceget()
Fredrik Lundh8a3ebf82000-07-23 21:46:17 +0000581 p = _parse_sub(source, state)
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000582 if not sourcematch(")"):
Fredrik Lundh0c4fdba2000-08-31 22:57:55 +0000583 raise error, "unbalanced parenthesis"
Fredrik Lundh8a3ebf82000-07-23 21:46:17 +0000584 if char == "=":
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000585 subpatternappend((ASSERT, (dir, p)))
Fredrik Lundh8a3ebf82000-07-23 21:46:17 +0000586 else:
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000587 subpatternappend((ASSERT_NOT, (dir, p)))
Fredrik Lundh8a3ebf82000-07-23 21:46:17 +0000588 continue
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000589 elif sourcematch("("):
Gustavo Niemeyerad3fc442003-10-17 22:13:16 +0000590 # conditional backreference group
591 condname = ""
592 while 1:
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000593 char = sourceget()
Gustavo Niemeyerad3fc442003-10-17 22:13:16 +0000594 if char is None:
595 raise error, "unterminated name"
596 if char == ")":
597 break
598 condname = condname + char
599 group = 2
600 if isname(condname):
601 condgroup = state.groupdict.get(condname)
602 if condgroup is None:
603 raise error, "unknown group name"
604 else:
605 try:
Barry Warsaw8bee7612004-08-25 02:22:30 +0000606 condgroup = int(condname)
Gustavo Niemeyerad3fc442003-10-17 22:13:16 +0000607 except ValueError:
608 raise error, "bad character in group name"
Fredrik Lundh90a07912000-06-30 07:50:59 +0000609 else:
610 # flags
Raymond Hettinger54f02222002-06-01 14:18:47 +0000611 if not source.next in FLAGS:
Fredrik Lundh470ea5a2001-01-14 21:00:44 +0000612 raise error, "unexpected end of pattern"
Raymond Hettinger54f02222002-06-01 14:18:47 +0000613 while source.next in FLAGS:
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000614 state.flags = state.flags | FLAGS[sourceget()]
Fredrik Lundh90a07912000-06-30 07:50:59 +0000615 if group:
616 # parse group contents
Fredrik Lundh90a07912000-06-30 07:50:59 +0000617 if group == 2:
618 # anonymous group
619 group = None
620 else:
Fredrik Lundhebc37b22000-10-28 19:30:41 +0000621 group = state.opengroup(name)
Gustavo Niemeyerad3fc442003-10-17 22:13:16 +0000622 if condgroup:
623 p = _parse_sub_cond(source, state, condgroup)
624 else:
625 p = _parse_sub(source, state)
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000626 if not sourcematch(")"):
Fredrik Lundh0c4fdba2000-08-31 22:57:55 +0000627 raise error, "unbalanced parenthesis"
Fredrik Lundhebc37b22000-10-28 19:30:41 +0000628 if group is not None:
629 state.closegroup(group)
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000630 subpatternappend((SUBPATTERN, (group, p)))
Fredrik Lundh90a07912000-06-30 07:50:59 +0000631 else:
632 while 1:
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000633 char = sourceget()
Fredrik Lundh470ea5a2001-01-14 21:00:44 +0000634 if char is None:
635 raise error, "unexpected end of pattern"
636 if char == ")":
Fredrik Lundh90a07912000-06-30 07:50:59 +0000637 break
638 raise error, "unknown extension"
Guido van Rossum7627c0d2000-03-31 14:58:54 +0000639
Fredrik Lundh90a07912000-06-30 07:50:59 +0000640 elif this == "^":
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000641 subpatternappend((AT, AT_BEGINNING))
Guido van Rossum7627c0d2000-03-31 14:58:54 +0000642
Fredrik Lundh90a07912000-06-30 07:50:59 +0000643 elif this == "$":
644 subpattern.append((AT, AT_END))
Guido van Rossum7627c0d2000-03-31 14:58:54 +0000645
Fredrik Lundh90a07912000-06-30 07:50:59 +0000646 elif this and this[0] == "\\":
647 code = _escape(source, this, state)
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000648 subpatternappend(code)
Guido van Rossum7627c0d2000-03-31 14:58:54 +0000649
Fredrik Lundh90a07912000-06-30 07:50:59 +0000650 else:
651 raise error, "parser error"
Guido van Rossum7627c0d2000-03-31 14:58:54 +0000652
653 return subpattern
654
Fredrik Lundh7898c3e2000-08-07 20:59:04 +0000655def parse(str, flags=0, pattern=None):
Andrew M. Kuchling815d5b92000-06-09 14:08:07 +0000656 # parse 're' pattern into list of (opcode, argument) tuples
Fredrik Lundh8a3ebf82000-07-23 21:46:17 +0000657
658 source = Tokenizer(str)
659
Fredrik Lundh7898c3e2000-08-07 20:59:04 +0000660 if pattern is None:
661 pattern = Pattern()
Fredrik Lundh8a3ebf82000-07-23 21:46:17 +0000662 pattern.flags = flags
Fredrik Lundh470ea5a2001-01-14 21:00:44 +0000663 pattern.str = str
Fredrik Lundh8a3ebf82000-07-23 21:46:17 +0000664
665 p = _parse_sub(source, pattern, 0)
666
667 tail = source.get()
668 if tail == ")":
669 raise error, "unbalanced parenthesis"
670 elif tail:
671 raise error, "bogus characters at end of regular expression"
672
Fredrik Lundh770617b2001-01-14 15:06:11 +0000673 if flags & SRE_FLAG_DEBUG:
674 p.dump()
Fredrik Lundh8a3ebf82000-07-23 21:46:17 +0000675
Fredrik Lundhd11b5e52000-10-03 19:22:26 +0000676 if not (flags & SRE_FLAG_VERBOSE) and p.pattern.flags & SRE_FLAG_VERBOSE:
677 # the VERBOSE flag was switched on inside the pattern. to be
678 # on the safe side, we'll parse the whole thing again...
679 return parse(str, p.pattern.flags)
680
Guido van Rossum7627c0d2000-03-31 14:58:54 +0000681 return p
682
Fredrik Lundh436c3d582000-06-29 08:58:44 +0000683def parse_template(source, pattern):
Andrew M. Kuchling815d5b92000-06-09 14:08:07 +0000684 # parse 're' replacement string into list of literals and
685 # group references
686 s = Tokenizer(source)
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000687 sget = s.get
Andrew M. Kuchling815d5b92000-06-09 14:08:07 +0000688 p = []
689 a = p.append
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000690 def literal(literal, p=p, pappend=a):
Fredrik Lundhb25e1ad2001-03-22 15:50:10 +0000691 if p and p[-1][0] is LITERAL:
692 p[-1] = LITERAL, p[-1][1] + literal
693 else:
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000694 pappend((LITERAL, literal))
Fredrik Lundhb25e1ad2001-03-22 15:50:10 +0000695 sep = source[:0]
696 if type(sep) is type(""):
Fredrik Lundh59b68652001-09-18 20:55:24 +0000697 makechar = chr
Fredrik Lundhb25e1ad2001-03-22 15:50:10 +0000698 else:
Fredrik Lundh59b68652001-09-18 20:55:24 +0000699 makechar = unichr
Andrew M. Kuchling815d5b92000-06-09 14:08:07 +0000700 while 1:
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000701 this = sget()
Fredrik Lundh90a07912000-06-30 07:50:59 +0000702 if this is None:
703 break # end of replacement string
704 if this and this[0] == "\\":
705 # group
Gustavo Niemeyera01a2ee2004-09-03 17:06:10 +0000706 c = this[1:2]
707 if c == "g":
Fredrik Lundh90a07912000-06-30 07:50:59 +0000708 name = ""
709 if s.match("<"):
710 while 1:
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000711 char = sget()
Fredrik Lundh90a07912000-06-30 07:50:59 +0000712 if char is None:
713 raise error, "unterminated group name"
714 if char == ">":
715 break
716 name = name + char
717 if not name:
718 raise error, "bad group name"
719 try:
Barry Warsaw8bee7612004-08-25 02:22:30 +0000720 index = int(name)
Gustavo Niemeyera01a2ee2004-09-03 17:06:10 +0000721 if index < 0:
722 raise error, "negative group number"
Fredrik Lundh90a07912000-06-30 07:50:59 +0000723 except ValueError:
724 if not isname(name):
Fredrik Lundh470ea5a2001-01-14 21:00:44 +0000725 raise error, "bad character in group name"
Fredrik Lundh90a07912000-06-30 07:50:59 +0000726 try:
727 index = pattern.groupindex[name]
728 except KeyError:
729 raise IndexError, "unknown group name"
730 a((MARK, index))
Gustavo Niemeyera01a2ee2004-09-03 17:06:10 +0000731 elif c == "0":
732 if s.next in OCTDIGITS:
733 this = this + sget()
734 if s.next in OCTDIGITS:
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000735 this = this + sget()
Gustavo Niemeyera01a2ee2004-09-03 17:06:10 +0000736 literal(makechar(int(this[1:], 8) & 0xff))
737 elif c in DIGITS:
738 isoctal = False
739 if s.next in DIGITS:
740 this = this + sget()
Gustavo Niemeyerf5a15992004-09-03 20:15:56 +0000741 if (c in OCTDIGITS and this[2] in OCTDIGITS and
742 s.next in OCTDIGITS):
Gustavo Niemeyera01a2ee2004-09-03 17:06:10 +0000743 this = this + sget()
744 isoctal = True
745 literal(makechar(int(this[1:], 8) & 0xff))
746 if not isoctal:
747 a((MARK, int(this[1:])))
Fredrik Lundh90a07912000-06-30 07:50:59 +0000748 else:
749 try:
Fredrik Lundh59b68652001-09-18 20:55:24 +0000750 this = makechar(ESCAPES[this][1])
Fredrik Lundh90a07912000-06-30 07:50:59 +0000751 except KeyError:
Fredrik Lundhb25e1ad2001-03-22 15:50:10 +0000752 pass
753 literal(this)
Fredrik Lundh90a07912000-06-30 07:50:59 +0000754 else:
Fredrik Lundhb25e1ad2001-03-22 15:50:10 +0000755 literal(this)
756 # convert template to groups and literals lists
757 i = 0
758 groups = []
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000759 groupsappend = groups.append
760 literals = [None] * len(p)
Fredrik Lundhb25e1ad2001-03-22 15:50:10 +0000761 for c, s in p:
762 if c is MARK:
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000763 groupsappend((i, s))
764 # literal[i] is already None
Fredrik Lundhb25e1ad2001-03-22 15:50:10 +0000765 else:
Raymond Hettinger968c56a2004-03-26 23:24:00 +0000766 literals[i] = s
Fredrik Lundhb25e1ad2001-03-22 15:50:10 +0000767 i = i + 1
768 return groups, literals
Andrew M. Kuchling815d5b92000-06-09 14:08:07 +0000769
Fredrik Lundh436c3d582000-06-29 08:58:44 +0000770def expand_template(template, match):
Fredrik Lundhb25e1ad2001-03-22 15:50:10 +0000771 g = match.group
Fredrik Lundh0640e112000-06-30 13:55:15 +0000772 sep = match.string[:0]
Fredrik Lundhb25e1ad2001-03-22 15:50:10 +0000773 groups, literals = template
774 literals = literals[:]
775 try:
776 for index, group in groups:
777 literals[index] = s = g(group)
Fredrik Lundh90a07912000-06-30 07:50:59 +0000778 if s is None:
Gustavo Niemeyera01a2ee2004-09-03 17:06:10 +0000779 raise error, "unmatched group"
Fredrik Lundhb25e1ad2001-03-22 15:50:10 +0000780 except IndexError:
Gustavo Niemeyera01a2ee2004-09-03 17:06:10 +0000781 raise error, "invalid group reference"
Barry Warsaw8bee7612004-08-25 02:22:30 +0000782 return sep.join(literals)