Lib/sre_parse.py

   1 #
   2 # Secret Labs' Regular Expression Engine
   3 #
   4 # convert re-style regular expression to sre pattern
   5 #
   6 # Copyright (c) 1998-2001 by Secret Labs AB.  All rights reserved.
   7 #
   8 # See the sre.py file for information on usage and redistribution.
   9 #
  10
  11 """Internal support module for sre"""
  12
  13 # XXX: show string offset and offending character for all errors
  14
  15 # this module works under 1.5.2 and later.  don't use string methods
  16 import string, sys
  17
  18 from sre_constants import *
  19
  20 SPECIAL_CHARS = ".\\[{()*+?^$|"
  21 REPEAT_CHARS = "*+?{"
  22
  23 DIGITS = tuple("0123456789")
  24
  25 OCTDIGITS = tuple("01234567")
  26 HEXDIGITS = tuple("0123456789abcdefABCDEF")
  27
  28 WHITESPACE = tuple(" \t\n\r\v\f")
  29
  30 ESCAPES = {
  31     r"\a": (LITERAL, ord("\a")),
  32     r"\b": (LITERAL, ord("\b")),
  33     r"\f": (LITERAL, ord("\f")),
  34     r"\n": (LITERAL, ord("\n")),
  35     r"\r": (LITERAL, ord("\r")),
  36     r"\t": (LITERAL, ord("\t")),
  37     r"\v": (LITERAL, ord("\v")),
  38     r"\\": (LITERAL, ord("\\"))
  39 }
  40
  41 CATEGORIES = {
  42     r"\A": (AT, AT_BEGINNING_STRING), # start of string
  43     r"\b": (AT, AT_BOUNDARY),
  44     r"\B": (AT, AT_NON_BOUNDARY),
  45     r"\d": (IN, [(CATEGORY, CATEGORY_DIGIT)]),
  46     r"\D": (IN, [(CATEGORY, CATEGORY_NOT_DIGIT)]),
  47     r"\s": (IN, [(CATEGORY, CATEGORY_SPACE)]),
  48     r"\S": (IN, [(CATEGORY, CATEGORY_NOT_SPACE)]),
  49     r"\w": (IN, [(CATEGORY, CATEGORY_WORD)]),
  50     r"\W": (IN, [(CATEGORY, CATEGORY_NOT_WORD)]),
  51     r"\Z": (AT, AT_END_STRING), # end of string
  52 }
  53
  54 FLAGS = {
  55     # standard flags
  56     "i": SRE_FLAG_IGNORECASE,
  57     "L": SRE_FLAG_LOCALE,
  58     "m": SRE_FLAG_MULTILINE,
  59     "s": SRE_FLAG_DOTALL,
  60     "x": SRE_FLAG_VERBOSE,
  61     # extensions
  62     "t": SRE_FLAG_TEMPLATE,
  63     "u": SRE_FLAG_UNICODE,
  64 }
  65
  66 # figure out best way to convert hex/octal numbers to integers
  67 try:
  68     int("10", 8)
  69     atoi = int # 2.0 and later
  70 except TypeError:
  71     atoi = string.atoi # 1.5.2
  72
  73 class Pattern:
  74     # master pattern object.  keeps track of global attributes
  75     def __init__(self):
  76         self.flags = 0
  77         self.open = []
  78         self.groups = 1
  79         self.groupdict = {}
  80     def opengroup(self, name=None):
  81         gid = self.groups
  82         self.groups = gid + 1
  83         if name is not None:
  84             ogid = self.groupdict.get(name, None)
  85             if ogid is not None:
  86                 raise error, ("redefinition of group name %s as group %d; "
  87                               "was group %d" % (repr(name), gid,  ogid))
  88             self.groupdict[name] = gid
  89         self.open.append(gid)
  90         return gid
  91     def closegroup(self, gid):
  92         self.open.remove(gid)
  93     def checkgroup(self, gid):
  94         return gid < self.groups and gid not in self.open
  95
  96 class SubPattern:
  97     # a subpattern, in intermediate form
  98     def __init__(self, pattern, data=None):
  99         self.pattern = pattern
 100         if data is None:
 101             data = []
 102         self.data = data
 103         self.width = None
 104     def dump(self, level=0):
 105         nl = 1
 106         for op, av in self.data:
 107             print level*"  " + op,; nl = 0
 108             if op == "in":
 109                 # member sublanguage
 110                 print; nl = 1
 111                 for op, a in av:
 112                     print (level+1)*"  " + op, a
 113             elif op == "branch":
 114                 print; nl = 1
 115                 i = 0
 116                 for a in av[1]:
 117                     if i > 0:
 118                         print level*"  " + "or"
 119                     a.dump(level+1); nl = 1
 120                     i = i + 1
 121             elif type(av) in (type(()), type([])):
 122                 for a in av:
 123                     if isinstance(a, SubPattern):
 124                         if not nl: print
 125                         a.dump(level+1); nl = 1
 126                     else:
 127                         print a, ; nl = 0
 128             else:
 129                 print av, ; nl = 0
 130             if not nl: print
 131     def __repr__(self):
 132         return repr(self.data)
 133     def __len__(self):
 134         return len(self.data)
 135     def __delitem__(self, index):
 136         del self.data[index]
 137     def __getitem__(self, index):
 138         return self.data[index]
 139     def __setitem__(self, index, code):
 140         self.data[index] = code
 141     def __getslice__(self, start, stop):
 142         return SubPattern(self.pattern, self.data[start:stop])
 143     def insert(self, index, code):
 144         self.data.insert(index, code)
 145     def append(self, code):
 146         self.data.append(code)
 147     def getwidth(self):
 148         # determine the width (min, max) for this subpattern
 149         if self.width:
 150             return self.width
 151         lo = hi = 0L
 152         for op, av in self.data:
 153             if op is BRANCH:
 154                 i = sys.maxint
 155                 j = 0
 156                 for av in av[1]:
 157                     l, h = av.getwidth()
 158                     i = min(i, l)
 159                     j = max(j, h)
 160                 lo = lo + i
 161                 hi = hi + j
 162             elif op is CALL:
 163                 i, j = av.getwidth()
 164                 lo = lo + i
 165                 hi = hi + j
 166             elif op is SUBPATTERN:
 167                 i, j = av[1].getwidth()
 168                 lo = lo + i
 169                 hi = hi + j
 170             elif op in (MIN_REPEAT, MAX_REPEAT):
 171                 i, j = av[2].getwidth()
 172                 lo = lo + long(i) * av[0]
 173                 hi = hi + long(j) * av[1]
 174             elif op in (ANY, RANGE, IN, LITERAL, NOT_LITERAL, CATEGORY):
 175                 lo = lo + 1
 176                 hi = hi + 1
 177             elif op == SUCCESS:
 178                 break
 179         self.width = int(min(lo, sys.maxint)), int(min(hi, sys.maxint))
 180         return self.width
 181
 182 class Tokenizer:
 183     def __init__(self, string):
 184         self.string = string
 185         self.index = 0
 186         self.__next()
 187     def __next(self):
 188         if self.index >= len(self.string):
 189             self.next = None
 190             return
 191         char = self.string[self.index]
 192         if char[0] == "\\":
 193             try:
 194                 c = self.string[self.index + 1]
 195             except IndexError:
 196                 raise error, "bogus escape (end of line)"
 197             char = char + c
 198         self.index = self.index + len(char)
 199         self.next = char
 200     def match(self, char, skip=1):
 201         if char == self.next:
 202             if skip:
 203                 self.__next()
 204             return 1
 205         return 0
 206     def get(self):
 207         this = self.next
 208         self.__next()
 209         return this
 210     def tell(self):
 211         return self.index, self.next
 212     def seek(self, index):
 213         self.index, self.next = index
 214
 215 def isident(char):
 216     return "a" <= char <= "z" or "A" <= char <= "Z" or char == "_"
 217
 218 def isdigit(char):
 219     return "0" <= char <= "9"
 220
 221 def isname(name):
 222     # check that group name is a valid string
 223     if not isident(name[0]):
 224         return False
 225     for char in name:
 226         if not isident(char) and not isdigit(char):
 227             return False
 228     return True
 229
 230 def _group(escape, groups):
 231     # check if the escape string represents a valid group
 232     try:
 233         gid = atoi(escape[1:])
 234         if gid and gid < groups:
 235             return gid
 236     except ValueError:
 237         pass
 238     return None # not a valid group
 239
 240 def _class_escape(source, escape):
 241     # handle escape code inside character class
 242     code = ESCAPES.get(escape)
 243     if code:
 244         return code
 245     code = CATEGORIES.get(escape)
 246     if code:
 247         return code
 248     try:
 249         if escape[1:2] == "x":
 250             # hexadecimal escape (exactly two digits)
 251             while source.next in HEXDIGITS and len(escape) < 4:
 252                 escape = escape + source.get()
 253             escape = escape[2:]
 254             if len(escape) != 2:
 255                 raise error, "bogus escape: %s" % repr("\\" + escape)
 256             return LITERAL, atoi(escape, 16) & 0xff
 257         elif str(escape[1:2]) in OCTDIGITS:
 258             # octal escape (up to three digits)
 259             while source.next in OCTDIGITS and len(escape) < 5:
 260                 escape = escape + source.get()
 261             escape = escape[1:]
 262             return LITERAL, atoi(escape, 8) & 0xff
 263         if len(escape) == 2:
 264             return LITERAL, ord(escape[1])
 265     except ValueError:
 266         pass
 267     raise error, "bogus escape: %s" % repr(escape)
 268
 269 def _escape(source, escape, state):
 270     # handle escape code in expression
 271     code = CATEGORIES.get(escape)
 272     if code:
 273         return code
 274     code = ESCAPES.get(escape)
 275     if code:
 276         return code
 277     try:
 278         if escape[1:2] == "x":
 279             # hexadecimal escape
 280             while source.next in HEXDIGITS and len(escape) < 4:
 281                 escape = escape + source.get()
 282             if len(escape) != 4:
 283                 raise ValueError
 284             return LITERAL, atoi(escape[2:], 16) & 0xff
 285         elif escape[1:2] == "0":
 286             # octal escape
 287             while source.next in OCTDIGITS and len(escape) < 4:
 288                 escape = escape + source.get()
 289             return LITERAL, atoi(escape[1:], 8) & 0xff
 290         elif escape[1:2] in DIGITS:
 291             # octal escape *or* decimal group reference (sigh)
 292             if source.next in DIGITS:
 293                 escape = escape + source.get()
 294                 if (escape[1] in OCTDIGITS and escape[2] in OCTDIGITS and
 295                     source.next in OCTDIGITS):
 296                     # got three octal digits; this is an octal escape
 297                     escape = escape + source.get()
 298                     return LITERAL, atoi(escape[1:], 8) & 0xff
 299             # got at least one decimal digit; this is a group reference
 300             group = _group(escape, state.groups)
 301             if group:
 302                 if not state.checkgroup(group):
 303                     raise error, "cannot refer to open group"
 304                 return GROUPREF, group
 305             raise ValueError
 306         if len(escape) == 2:
 307             return LITERAL, ord(escape[1])
 308     except ValueError:
 309         pass
 310     raise error, "bogus escape: %s" % repr(escape)
 311
 312 def _parse_sub(source, state, nested=1):
 313     # parse an alternation: a|b|c
 314
 315     items = []
 316     while 1:
 317         items.append(_parse(source, state))
 318         if source.match("|"):
 319             continue
 320         if not nested:
 321             break
 322         if not source.next or source.match(")", 0):
 323             break
 324         else:
 325             raise error, "pattern not properly closed"
 326
 327     if len(items) == 1:
 328         return items[0]
 329
 330     subpattern = SubPattern(state)
 331
 332     # check if all items share a common prefix
 333     while 1:
 334         prefix = None
 335         for item in items:
 336             if not item:
 337                 break
 338             if prefix is None:
 339                 prefix = item[0]
 340             elif item[0] != prefix:
 341                 break
 342         else:
 343             # all subitems start with a common "prefix".
 344             # move it out of the branch
 345             for item in items:
 346                 del item[0]
 347             subpattern.append(prefix)
 348             continue # check next one
 349         break
 350
 351     # check if the branch can be replaced by a character set
 352     for item in items:
 353         if len(item) != 1 or item[0][0] != LITERAL:
 354             break
 355     else:
 356         # we can store this as a character set instead of a
 357         # branch (the compiler may optimize this even more)
 358         set = []
 359         for item in items:
 360             set.append(item[0])
 361         subpattern.append((IN, set))
 362         return subpattern
 363
 364     subpattern.append((BRANCH, (None, items)))
 365     return subpattern
 366
 367 def _parse(source, state):
 368     # parse a simple pattern
 369
 370     subpattern = SubPattern(state)
 371
 372     while 1:
 373
 374         if source.next in ("|", ")"):
 375             break # end of subpattern
 376         this = source.get()
 377         if this is None:
 378             break # end of pattern
 379
 380         if state.flags & SRE_FLAG_VERBOSE:
 381             # skip whitespace and comments
 382             if this in WHITESPACE:
 383                 continue
 384             if this == "#":
 385                 while 1:
 386                     this = source.get()
 387                     if this in (None, "\n"):
 388                         break
 389                 continue
 390
 391         if this and this[0] not in SPECIAL_CHARS:
 392             subpattern.append((LITERAL, ord(this)))
 393
 394         elif this == "[":
 395             # character set
 396             set = []
 397 ##          if source.match(":"):
 398 ##              pass # handle character classes
 399             if source.match("^"):
 400                 set.append((NEGATE, None))
 401             # check remaining characters
 402             start = set[:]
 403             while 1:
 404                 this = source.get()
 405                 if this == "]" and set != start:
 406                     break
 407                 elif this and this[0] == "\\":
 408                     code1 = _class_escape(source, this)
 409                 elif this:
 410                     code1 = LITERAL, ord(this)
 411                 else:
 412                     raise error, "unexpected end of regular expression"
 413                 if source.match("-"):
 414                     # potential range
 415                     this = source.get()
 416                     if this == "]":
 417                         if code1[0] is IN:
 418                             code1 = code1[1][0]
 419                         set.append(code1)
 420                         set.append((LITERAL, ord("-")))
 421                         break
 422                     else:
 423                         if this[0] == "\\":
 424                             code2 = _class_escape(source, this)
 425                         else:
 426                             code2 = LITERAL, ord(this)
 427                         if code1[0] != LITERAL or code2[0] != LITERAL:
 428                             raise error, "bad character range"
 429                         lo = code1[1]
 430                         hi = code2[1]
 431                         if hi < lo:
 432                             raise error, "bad character range"
 433                         set.append((RANGE, (lo, hi)))
 434                 else:
 435                     if code1[0] is IN:
 436                         code1 = code1[1][0]
 437                     set.append(code1)
 438
 439             # XXX: <fl> should move set optimization to compiler!
 440             if len(set)==1 and set[0][0] is LITERAL:
 441                 subpattern.append(set[0]) # optimization
 442             elif len(set)==2 and set[0][0] is NEGATE and set[1][0] is LITERAL:
 443                 subpattern.append((NOT_LITERAL, set[1][1])) # optimization
 444             else:
 445                 # XXX: <fl> should add charmap optimization here
 446                 subpattern.append((IN, set))
 447
 448         elif this and this[0] in REPEAT_CHARS:
 449             # repeat previous item
 450             if this == "?":
 451                 min, max = 0, 1
 452             elif this == "*":
 453                 min, max = 0, MAXREPEAT
 454
 455             elif this == "+":
 456                 min, max = 1, MAXREPEAT
 457             elif this == "{":
 458                 here = source.tell()
 459                 min, max = 0, MAXREPEAT
 460                 lo = hi = ""
 461                 while source.next in DIGITS:
 462                     lo = lo + source.get()
 463                 if source.match(","):
 464                     while source.next in DIGITS:
 465                         hi = hi + source.get()
 466                 else:
 467                     hi = lo
 468                 if not source.match("}"):
 469                     subpattern.append((LITERAL, ord(this)))
 470                     source.seek(here)
 471                     continue
 472                 if lo:
 473                     min = atoi(lo)
 474                 if hi:
 475                     max = atoi(hi)
 476                 if max < min:
 477                     raise error, "bad repeat interval"
 478             else:
 479                 raise error, "not supported"
 480             # figure out which item to repeat
 481             if subpattern:
 482                 item = subpattern[-1:]
 483             else:
 484                 item = None
 485             if not item or (len(item) == 1 and item[0][0] == AT):
 486                 raise error, "nothing to repeat"
 487             if item[0][0] in (MIN_REPEAT, MAX_REPEAT):
 488                 raise error, "multiple repeat"
 489             if source.match("?"):
 490                 subpattern[-1] = (MIN_REPEAT, (min, max, item))
 491             else:
 492                 subpattern[-1] = (MAX_REPEAT, (min, max, item))
 493
 494         elif this == ".":
 495             subpattern.append((ANY, None))
 496
 497         elif this == "(":
 498             group = 1
 499             name = None
 500             if source.match("?"):
 501                 group = 0
 502                 # options
 503                 if source.match("P"):
 504                     # python extensions
 505                     if source.match("<"):
 506                         # named group: skip forward to end of name
 507                         name = ""
 508                         while 1:
 509                             char = source.get()
 510                             if char is None:
 511                                 raise error, "unterminated name"
 512                             if char == ">":
 513                                 break
 514                             name = name + char
 515                         group = 1
 516                         if not isname(name):
 517                             raise error, "bad character in group name"
 518                     elif source.match("="):
 519                         # named backreference
 520                         name = ""
 521                         while 1:
 522                             char = source.get()
 523                             if char is None:
 524                                 raise error, "unterminated name"
 525                             if char == ")":
 526                                 break
 527                             name = name + char
 528                         if not isname(name):
 529                             raise error, "bad character in group name"
 530                         gid = state.groupdict.get(name)
 531                         if gid is None:
 532                             raise error, "unknown group name"
 533                         subpattern.append((GROUPREF, gid))
 534                         continue
 535                     else:
 536                         char = source.get()
 537                         if char is None:
 538                             raise error, "unexpected end of pattern"
 539                         raise error, "unknown specifier: ?P%s" % char
 540                 elif source.match(":"):
 541                     # non-capturing group
 542                     group = 2
 543                 elif source.match("#"):
 544                     # comment
 545                     while 1:
 546                         if source.next is None or source.next == ")":
 547                             break
 548                         source.get()
 549                     if not source.match(")"):
 550                         raise error, "unbalanced parenthesis"
 551                     continue
 552                 elif source.next in ("=", "!", "<"):
 553                     # lookahead assertions
 554                     char = source.get()
 555                     dir = 1
 556                     if char == "<":
 557                         if source.next not in ("=", "!"):
 558                             raise error, "syntax error"
 559                         dir = -1 # lookbehind
 560                         char = source.get()
 561                     p = _parse_sub(source, state)
 562                     if not source.match(")"):
 563                         raise error, "unbalanced parenthesis"
 564                     if char == "=":
 565                         subpattern.append((ASSERT, (dir, p)))
 566                     else:
 567                         subpattern.append((ASSERT_NOT, (dir, p)))
 568                     continue
 569                 else:
 570                     # flags
 571                     if not source.next in FLAGS:
 572                         raise error, "unexpected end of pattern"
 573                     while source.next in FLAGS:
 574                         state.flags = state.flags | FLAGS[source.get()]
 575             if group:
 576                 # parse group contents
 577                 if group == 2:
 578                     # anonymous group
 579                     group = None
 580                 else:
 581                     group = state.opengroup(name)
 582                 p = _parse_sub(source, state)
 583                 if not source.match(")"):
 584                     raise error, "unbalanced parenthesis"
 585                 if group is not None:
 586                     state.closegroup(group)
 587                 subpattern.append((SUBPATTERN, (group, p)))
 588             else:
 589                 while 1:
 590                     char = source.get()
 591                     if char is None:
 592                         raise error, "unexpected end of pattern"
 593                     if char == ")":
 594                         break
 595                     raise error, "unknown extension"
 596
 597         elif this == "^":
 598             subpattern.append((AT, AT_BEGINNING))
 599
 600         elif this == "$":
 601             subpattern.append((AT, AT_END))
 602
 603         elif this and this[0] == "\\":
 604             code = _escape(source, this, state)
 605             subpattern.append(code)
 606
 607         else:
 608             raise error, "parser error"
 609
 610     return subpattern
 611
 612 def parse(str, flags=0, pattern=None):
 613     # parse 're' pattern into list of (opcode, argument) tuples
 614
 615     source = Tokenizer(str)
 616
 617     if pattern is None:
 618         pattern = Pattern()
 619     pattern.flags = flags
 620     pattern.str = str
 621
 622     p = _parse_sub(source, pattern, 0)
 623
 624     tail = source.get()
 625     if tail == ")":
 626         raise error, "unbalanced parenthesis"
 627     elif tail:
 628         raise error, "bogus characters at end of regular expression"
 629
 630     if flags & SRE_FLAG_DEBUG:
 631         p.dump()
 632
 633     if not (flags & SRE_FLAG_VERBOSE) and p.pattern.flags & SRE_FLAG_VERBOSE:
 634         # the VERBOSE flag was switched on inside the pattern.  to be
 635         # on the safe side, we'll parse the whole thing again...
 636         return parse(str, p.pattern.flags)
 637
 638     return p
 639
 640 def parse_template(source, pattern):
 641     # parse 're' replacement string into list of literals and
 642     # group references
 643     s = Tokenizer(source)
 644     p = []
 645     a = p.append
 646     def literal(literal, p=p):
 647         if p and p[-1][0] is LITERAL:
 648             p[-1] = LITERAL, p[-1][1] + literal
 649         else:
 650             p.append((LITERAL, literal))
 651     sep = source[:0]
 652     if type(sep) is type(""):
 653         makechar = chr
 654     else:
 655         makechar = unichr
 656     while 1:
 657         this = s.get()
 658         if this is None:
 659             break # end of replacement string
 660         if this and this[0] == "\\":
 661             # group
 662             if this == "\\g":
 663                 name = ""
 664                 if s.match("<"):
 665                     while 1:
 666                         char = s.get()
 667                         if char is None:
 668                             raise error, "unterminated group name"
 669                         if char == ">":
 670                             break
 671                         name = name + char
 672                 if not name:
 673                     raise error, "bad group name"
 674                 try:
 675                     index = atoi(name)
 676                 except ValueError:
 677                     if not isname(name):
 678                         raise error, "bad character in group name"
 679                     try:
 680                         index = pattern.groupindex[name]
 681                     except KeyError:
 682                         raise IndexError, "unknown group name"
 683                 a((MARK, index))
 684             elif len(this) > 1 and this[1] in DIGITS:
 685                 code = None
 686                 while 1:
 687                     group = _group(this, pattern.groups+1)
 688                     if group:
 689                         if (s.next not in DIGITS or
 690                             not _group(this + s.next, pattern.groups+1)):
 691                             code = MARK, group
 692                             break
 693                     elif s.next in OCTDIGITS:
 694                         this = this + s.get()
 695                     else:
 696                         break
 697                 if not code:
 698                     this = this[1:]
 699                     code = LITERAL, makechar(atoi(this[-6:], 8) & 0xff)
 700                 if code[0] is LITERAL:
 701                     literal(code[1])
 702                 else:
 703                     a(code)
 704             else:
 705                 try:
 706                     this = makechar(ESCAPES[this][1])
 707                 except KeyError:
 708                     pass
 709                 literal(this)
 710         else:
 711             literal(this)
 712     # convert template to groups and literals lists
 713     i = 0
 714     groups = []
 715     literals = []
 716     for c, s in p:
 717         if c is MARK:
 718             groups.append((i, s))
 719             literals.append(None)
 720         else:
 721             literals.append(s)
 722         i = i + 1
 723     return groups, literals
 724
 725 def expand_template(template, match):
 726     g = match.group
 727     sep = match.string[:0]
 728     groups, literals = template
 729     literals = literals[:]
 730     try:
 731         for index, group in groups:
 732             literals[index] = s = g(group)
 733             if s is None:
 734                 raise IndexError
 735     except IndexError:
 736         raise error, "empty group"
 737     return string.join(literals, sep)