ext/ply/ply/lex.py

   1 #-----------------------------------------------------------------------------
   2 # ply: lex.py
   3 #
   4 # Author: David M. Beazley (dave@dabeaz.com)
   5 #
   6 # Copyright (C) 2001-2007, David M. Beazley
   7 #
   8 # This library is free software; you can redistribute it and/or
   9 # modify it under the terms of the GNU Lesser General Public
  10 # License as published by the Free Software Foundation; either
  11 # version 2.1 of the License, or (at your option) any later version.
  12 #
  13 # This library is distributed in the hope that it will be useful,
  14 # but WITHOUT ANY WARRANTY; without even the implied warranty of
  15 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  16 # Lesser General Public License for more details.
  17 #
  18 # You should have received a copy of the GNU Lesser General Public
  19 # License along with this library; if not, write to the Free Software
  20 # Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
  21 #
  22 # See the file COPYING for a complete copy of the LGPL.
  23 #-----------------------------------------------------------------------------
  24
  25 __version__ = "2.3"
  26
  27 import re, sys, types
  28
  29 # Regular expression used to match valid token names
  30 _is_identifier = re.compile(r'^[a-zA-Z0-9_]+$')
  31
  32 # Available instance types.  This is used when lexers are defined by a class.
  33 # It's a little funky because I want to preserve backwards compatibility
  34 # with Python 2.0 where types.ObjectType is undefined.
  35
  36 try:
  37    _INSTANCETYPE = (types.InstanceType, types.ObjectType)
  38 except AttributeError:
  39    _INSTANCETYPE = types.InstanceType
  40    class object: pass       # Note: needed if no new-style classes present
  41
  42 # Exception thrown when invalid token encountered and no default error
  43 # handler is defined.
  44 class LexError(Exception):
  45     def __init__(self,message,s):
  46          self.args = (message,)
  47          self.text = s
  48
  49 # Token class
  50 class LexToken(object):
  51     def __str__(self):
  52         return "LexToken(%s,%r,%d,%d)" % (self.type,self.value,self.lineno,self.lexpos)
  53     def __repr__(self):
  54         return str(self)
  55     def skip(self,n):
  56         self.lexer.skip(n)
  57
  58 # -----------------------------------------------------------------------------
  59 # Lexer class
  60 #
  61 # This class encapsulates all of the methods and data associated with a lexer.
  62 #
  63 #    input()          -  Store a new string in the lexer
  64 #    token()          -  Get the next token
  65 # -----------------------------------------------------------------------------
  66
  67 class Lexer:
  68     def __init__(self):
  69         self.lexre = None             # Master regular expression. This is a list of
  70                                       # tuples (re,findex) where re is a compiled
  71                                       # regular expression and findex is a list
  72                                       # mapping regex group numbers to rules
  73         self.lexretext = None         # Current regular expression strings
  74         self.lexstatere = {}          # Dictionary mapping lexer states to master regexs
  75         self.lexstateretext = {}      # Dictionary mapping lexer states to regex strings
  76         self.lexstate = "INITIAL"     # Current lexer state
  77         self.lexstatestack = []       # Stack of lexer states
  78         self.lexstateinfo = None      # State information
  79         self.lexstateignore = {}      # Dictionary of ignored characters for each state
  80         self.lexstateerrorf = {}      # Dictionary of error functions for each state
  81         self.lexreflags = 0           # Optional re compile flags
  82         self.lexdata = None           # Actual input data (as a string)
  83         self.lexpos = 0               # Current position in input text
  84         self.lexlen = 0               # Length of the input text
  85         self.lexerrorf = None         # Error rule (if any)
  86         self.lextokens = None         # List of valid tokens
  87         self.lexignore = ""           # Ignored characters
  88         self.lexliterals = ""         # Literal characters that can be passed through
  89         self.lexmodule = None         # Module
  90         self.lineno = 1               # Current line number
  91         self.lexdebug = 0             # Debugging mode
  92         self.lexoptimize = 0          # Optimized mode
  93
  94     def clone(self,object=None):
  95         c = Lexer()
  96         c.lexstatere = self.lexstatere
  97         c.lexstateinfo = self.lexstateinfo
  98         c.lexstateretext = self.lexstateretext
  99         c.lexstate = self.lexstate
 100         c.lexstatestack = self.lexstatestack
 101         c.lexstateignore = self.lexstateignore
 102         c.lexstateerrorf = self.lexstateerrorf
 103         c.lexreflags = self.lexreflags
 104         c.lexdata = self.lexdata
 105         c.lexpos = self.lexpos
 106         c.lexlen = self.lexlen
 107         c.lextokens = self.lextokens
 108         c.lexdebug = self.lexdebug
 109         c.lineno = self.lineno
 110         c.lexoptimize = self.lexoptimize
 111         c.lexliterals = self.lexliterals
 112         c.lexmodule   = self.lexmodule
 113
 114         # If the object parameter has been supplied, it means we are attaching the
 115         # lexer to a new object.  In this case, we have to rebind all methods in
 116         # the lexstatere and lexstateerrorf tables.
 117
 118         if object:
 119             newtab = { }
 120             for key, ritem in self.lexstatere.items():
 121                 newre = []
 122                 for cre, findex in ritem:
 123                      newfindex = []
 124                      for f in findex:
 125                          if not f or not f[0]:
 126                              newfindex.append(f)
 127                              continue
 128                          newfindex.append((getattr(object,f[0].__name__),f[1]))
 129                 newre.append((cre,newfindex))
 130                 newtab[key] = newre
 131             c.lexstatere = newtab
 132             c.lexstateerrorf = { }
 133             for key, ef in self.lexstateerrorf.items():
 134                 c.lexstateerrorf[key] = getattr(object,ef.__name__)
 135             c.lexmodule = object
 136
 137         # Set up other attributes
 138         c.begin(c.lexstate)
 139         return c
 140
 141     # ------------------------------------------------------------
 142     # writetab() - Write lexer information to a table file
 143     # ------------------------------------------------------------
 144     def writetab(self,tabfile):
 145         tf = open(tabfile+".py","w")
 146         tf.write("# %s.py. This file automatically created by PLY (version %s). Don't edit!\n" % (tabfile,__version__))
 147         tf.write("_lextokens    = %s\n" % repr(self.lextokens))
 148         tf.write("_lexreflags   = %s\n" % repr(self.lexreflags))
 149         tf.write("_lexliterals  = %s\n" % repr(self.lexliterals))
 150         tf.write("_lexstateinfo = %s\n" % repr(self.lexstateinfo))
 151
 152         tabre = { }
 153         for key, lre in self.lexstatere.items():
 154              titem = []
 155              for i in range(len(lre)):
 156                   titem.append((self.lexstateretext[key][i],_funcs_to_names(lre[i][1])))
 157              tabre[key] = titem
 158
 159         tf.write("_lexstatere   = %s\n" % repr(tabre))
 160         tf.write("_lexstateignore = %s\n" % repr(self.lexstateignore))
 161
 162         taberr = { }
 163         for key, ef in self.lexstateerrorf.items():
 164              if ef:
 165                   taberr[key] = ef.__name__
 166              else:
 167                   taberr[key] = None
 168         tf.write("_lexstateerrorf = %s\n" % repr(taberr))
 169         tf.close()
 170
 171     # ------------------------------------------------------------
 172     # readtab() - Read lexer information from a tab file
 173     # ------------------------------------------------------------
 174     def readtab(self,tabfile,fdict):
 175         exec "import %s as lextab" % tabfile
 176         self.lextokens      = lextab._lextokens
 177         self.lexreflags     = lextab._lexreflags
 178         self.lexliterals    = lextab._lexliterals
 179         self.lexstateinfo   = lextab._lexstateinfo
 180         self.lexstateignore = lextab._lexstateignore
 181         self.lexstatere     = { }
 182         self.lexstateretext = { }
 183         for key,lre in lextab._lexstatere.items():
 184              titem = []
 185              txtitem = []
 186              for i in range(len(lre)):
 187                   titem.append((re.compile(lre[i][0],lextab._lexreflags),_names_to_funcs(lre[i][1],fdict)))
 188                   txtitem.append(lre[i][0])
 189              self.lexstatere[key] = titem
 190              self.lexstateretext[key] = txtitem
 191         self.lexstateerrorf = { }
 192         for key,ef in lextab._lexstateerrorf.items():
 193              self.lexstateerrorf[key] = fdict[ef]
 194         self.begin('INITIAL')
 195
 196     # ------------------------------------------------------------
 197     # input() - Push a new string into the lexer
 198     # ------------------------------------------------------------
 199     def input(self,s):
 200         if not (isinstance(s,types.StringType) or isinstance(s,types.UnicodeType)):
 201             raise ValueError, "Expected a string"
 202         self.lexdata = s
 203         self.lexpos = 0
 204         self.lexlen = len(s)
 205
 206     # ------------------------------------------------------------
 207     # begin() - Changes the lexing state
 208     # ------------------------------------------------------------
 209     def begin(self,state):
 210         if not self.lexstatere.has_key(state):
 211             raise ValueError, "Undefined state"
 212         self.lexre = self.lexstatere[state]
 213         self.lexretext = self.lexstateretext[state]
 214         self.lexignore = self.lexstateignore.get(state,"")
 215         self.lexerrorf = self.lexstateerrorf.get(state,None)
 216         self.lexstate = state
 217
 218     # ------------------------------------------------------------
 219     # push_state() - Changes the lexing state and saves old on stack
 220     # ------------------------------------------------------------
 221     def push_state(self,state):
 222         self.lexstatestack.append(self.lexstate)
 223         self.begin(state)
 224
 225     # ------------------------------------------------------------
 226     # pop_state() - Restores the previous state
 227     # ------------------------------------------------------------
 228     def pop_state(self):
 229         self.begin(self.lexstatestack.pop())
 230
 231     # ------------------------------------------------------------
 232     # current_state() - Returns the current lexing state
 233     # ------------------------------------------------------------
 234     def current_state(self):
 235         return self.lexstate
 236
 237     # ------------------------------------------------------------
 238     # skip() - Skip ahead n characters
 239     # ------------------------------------------------------------
 240     def skip(self,n):
 241         self.lexpos += n
 242
 243     # ------------------------------------------------------------
 244     # token() - Return the next token from the Lexer
 245     #
 246     # Note: This function has been carefully implemented to be as fast
 247     # as possible.  Don't make changes unless you really know what
 248     # you are doing
 249     # ------------------------------------------------------------
 250     def token(self):
 251         # Make local copies of frequently referenced attributes
 252         lexpos    = self.lexpos
 253         lexlen    = self.lexlen
 254         lexignore = self.lexignore
 255         lexdata   = self.lexdata
 256
 257         while lexpos < lexlen:
 258             # This code provides some short-circuit code for whitespace, tabs, and other ignored characters
 259             if lexdata[lexpos] in lexignore:
 260                 lexpos += 1
 261                 continue
 262
 263             # Look for a regular expression match
 264             for lexre,lexindexfunc in self.lexre:
 265                 m = lexre.match(lexdata,lexpos)
 266                 if not m: continue
 267
 268                 # Set last match in lexer so that rules can access it if they want
 269                 self.lexmatch = m
 270
 271                 # Create a token for return
 272                 tok = LexToken()
 273                 tok.value = m.group()
 274                 tok.lineno = self.lineno
 275                 tok.lexpos = lexpos
 276                 tok.lexer = self
 277
 278                 lexpos = m.end()
 279                 i = m.lastindex
 280                 func,tok.type = lexindexfunc[i]
 281                 self.lexpos = lexpos
 282
 283                 if not func:
 284                    # If no token type was set, it's an ignored token
 285                    if tok.type: return tok
 286                    break
 287
 288                 # if func not callable, it means it's an ignored token
 289                 if not callable(func):
 290                    break
 291
 292                 # If token is processed by a function, call it
 293                 newtok = func(tok)
 294
 295                 # Every function must return a token, if nothing, we just move to next token
 296                 if not newtok:
 297                     lexpos = self.lexpos        # This is here in case user has updated lexpos.
 298                     break
 299
 300                 # Verify type of the token.  If not in the token map, raise an error
 301                 if not self.lexoptimize:
 302                     if not self.lextokens.has_key(newtok.type):
 303                         raise LexError, ("%s:%d: Rule '%s' returned an unknown token type '%s'" % (
 304                             func.func_code.co_filename, func.func_code.co_firstlineno,
 305                             func.__name__, newtok.type),lexdata[lexpos:])
 306
 307                 return newtok
 308             else:
 309                 # No match, see if in literals
 310                 if lexdata[lexpos] in self.lexliterals:
 311                     tok = LexToken()
 312                     tok.value = lexdata[lexpos]
 313                     tok.lineno = self.lineno
 314                     tok.lexer = self
 315                     tok.type = tok.value
 316                     tok.lexpos = lexpos
 317                     self.lexpos = lexpos + 1
 318                     return tok
 319
 320                 # No match. Call t_error() if defined.
 321                 if self.lexerrorf:
 322                     tok = LexToken()
 323                     tok.value = self.lexdata[lexpos:]
 324                     tok.lineno = self.lineno
 325                     tok.type = "error"
 326                     tok.lexer = self
 327                     tok.lexpos = lexpos
 328                     self.lexpos = lexpos
 329                     newtok = self.lexerrorf(tok)
 330                     if lexpos == self.lexpos:
 331                         # Error method didn't change text position at all. This is an error.
 332                         raise LexError, ("Scanning error. Illegal character '%s'" % (lexdata[lexpos]), lexdata[lexpos:])
 333                     lexpos = self.lexpos
 334                     if not newtok: continue
 335                     return newtok
 336
 337                 self.lexpos = lexpos
 338                 raise LexError, ("Illegal character '%s' at index %d" % (lexdata[lexpos],lexpos), lexdata[lexpos:])
 339
 340         self.lexpos = lexpos + 1
 341         if self.lexdata is None:
 342              raise RuntimeError, "No input string given with input()"
 343         return None
 344
 345 # -----------------------------------------------------------------------------
 346 # _validate_file()
 347 #
 348 # This checks to see if there are duplicated t_rulename() functions or strings
 349 # in the parser input file.  This is done using a simple regular expression
 350 # match on each line in the filename.
 351 # -----------------------------------------------------------------------------
 352
 353 def _validate_file(filename):
 354     import os.path
 355     base,ext = os.path.splitext(filename)
 356     if ext != '.py': return 1        # No idea what the file is. Return OK
 357
 358     try:
 359         f = open(filename)
 360         lines = f.readlines()
 361         f.close()
 362     except IOError:
 363         return 1                       # Oh well
 364
 365     fre = re.compile(r'\s*def\s+(t_[a-zA-Z_0-9]*)\(')
 366     sre = re.compile(r'\s*(t_[a-zA-Z_0-9]*)\s*=')
 367     counthash = { }
 368     linen = 1
 369     noerror = 1
 370     for l in lines:
 371         m = fre.match(l)
 372         if not m:
 373             m = sre.match(l)
 374         if m:
 375             name = m.group(1)
 376             prev = counthash.get(name)
 377             if not prev:
 378                 counthash[name] = linen
 379             else:
 380                 print >>sys.stderr, "%s:%d: Rule %s redefined. Previously defined on line %d" % (filename,linen,name,prev)
 381                 noerror = 0
 382         linen += 1
 383     return noerror
 384
 385 # -----------------------------------------------------------------------------
 386 # _funcs_to_names()
 387 #
 388 # Given a list of regular expression functions, this converts it to a list
 389 # suitable for output to a table file
 390 # -----------------------------------------------------------------------------
 391
 392 def _funcs_to_names(funclist):
 393     result = []
 394     for f in funclist:
 395          if f and f[0]:
 396              result.append((f[0].__name__,f[1]))
 397          else:
 398              result.append(f)
 399     return result
 400
 401 # -----------------------------------------------------------------------------
 402 # _names_to_funcs()
 403 #
 404 # Given a list of regular expression function names, this converts it back to
 405 # functions.
 406 # -----------------------------------------------------------------------------
 407
 408 def _names_to_funcs(namelist,fdict):
 409      result = []
 410      for n in namelist:
 411           if n and n[0]:
 412               result.append((fdict[n[0]],n[1]))
 413           else:
 414               result.append(n)
 415      return result
 416
 417 # -----------------------------------------------------------------------------
 418 # _form_master_re()
 419 #
 420 # This function takes a list of all of the regex components and attempts to
 421 # form the master regular expression.  Given limitations in the Python re
 422 # module, it may be necessary to break the master regex into separate expressions.
 423 # -----------------------------------------------------------------------------
 424
 425 def _form_master_re(relist,reflags,ldict,toknames):
 426     if not relist: return []
 427     regex = "|".join(relist)
 428     try:
 429         lexre = re.compile(regex,re.VERBOSE | reflags)
 430
 431         # Build the index to function map for the matching engine
 432         lexindexfunc = [ None ] * (max(lexre.groupindex.values())+1)
 433         for f,i in lexre.groupindex.items():
 434             handle = ldict.get(f,None)
 435             if type(handle) in (types.FunctionType, types.MethodType):
 436                 lexindexfunc[i] = (handle,toknames[handle.__name__])
 437             elif handle is not None:
 438                 # If rule was specified as a string, we build an anonymous
 439                 # callback function to carry out the action
 440                 if f.find("ignore_") > 0:
 441                     lexindexfunc[i] = (None,None)
 442                 else:
 443                     lexindexfunc[i] = (None, toknames[f])
 444
 445         return [(lexre,lexindexfunc)],[regex]
 446     except Exception,e:
 447         m = int(len(relist)/2)
 448         if m == 0: m = 1
 449         llist, lre = _form_master_re(relist[:m],reflags,ldict,toknames)
 450         rlist, rre = _form_master_re(relist[m:],reflags,ldict,toknames)
 451         return llist+rlist, lre+rre
 452
 453 # -----------------------------------------------------------------------------
 454 # def _statetoken(s,names)
 455 #
 456 # Given a declaration name s of the form "t_" and a dictionary whose keys are
 457 # state names, this function returns a tuple (states,tokenname) where states
 458 # is a tuple of state names and tokenname is the name of the token.  For example,
 459 # calling this with s = "t_foo_bar_SPAM" might return (('foo','bar'),'SPAM')
 460 # -----------------------------------------------------------------------------
 461
 462 def _statetoken(s,names):
 463     nonstate = 1
 464     parts = s.split("_")
 465     for i in range(1,len(parts)):
 466          if not names.has_key(parts[i]) and parts[i] != 'ANY': break
 467     if i > 1:
 468        states = tuple(parts[1:i])
 469     else:
 470        states = ('INITIAL',)
 471
 472     if 'ANY' in states:
 473        states = tuple(names.keys())
 474
 475     tokenname = "_".join(parts[i:])
 476     return (states,tokenname)
 477
 478 # -----------------------------------------------------------------------------
 479 # lex(module)
 480 #
 481 # Build all of the regular expression rules from definitions in the supplied module
 482 # -----------------------------------------------------------------------------
 483 def lex(module=None,object=None,debug=0,optimize=0,lextab="lextab",reflags=0,nowarn=0):
 484     global lexer
 485     ldict = None
 486     stateinfo  = { 'INITIAL' : 'inclusive'}
 487     error = 0
 488     files = { }
 489     lexobj = Lexer()
 490     lexobj.lexdebug = debug
 491     lexobj.lexoptimize = optimize
 492     global token,input
 493
 494     if nowarn: warn = 0
 495     else: warn = 1
 496
 497     if object: module = object
 498
 499     if module:
 500         # User supplied a module object.
 501         if isinstance(module, types.ModuleType):
 502             ldict = module.__dict__
 503         elif isinstance(module, _INSTANCETYPE):
 504             _items = [(k,getattr(module,k)) for k in dir(module)]
 505             ldict = { }
 506             for (i,v) in _items:
 507                 ldict[i] = v
 508         else:
 509             raise ValueError,"Expected a module or instance"
 510         lexobj.lexmodule = module
 511
 512     else:
 513         # No module given.  We might be able to get information from the caller.
 514         try:
 515             raise RuntimeError
 516         except RuntimeError:
 517             e,b,t = sys.exc_info()
 518             f = t.tb_frame
 519             f = f.f_back           # Walk out to our calling function
 520             ldict = f.f_globals    # Grab its globals dictionary
 521
 522     if optimize and lextab:
 523         try:
 524             lexobj.readtab(lextab,ldict)
 525             token = lexobj.token
 526             input = lexobj.input
 527             lexer = lexobj
 528             return lexobj
 529
 530         except ImportError:
 531             pass
 532
 533     # Get the tokens, states, and literals variables (if any)
 534     if (module and isinstance(module,_INSTANCETYPE)):
 535         tokens   = getattr(module,"tokens",None)
 536         states   = getattr(module,"states",None)
 537         literals = getattr(module,"literals","")
 538     else:
 539         tokens   = ldict.get("tokens",None)
 540         states   = ldict.get("states",None)
 541         literals = ldict.get("literals","")
 542
 543     if not tokens:
 544         raise SyntaxError,"lex: module does not define 'tokens'"
 545     if not (isinstance(tokens,types.ListType) or isinstance(tokens,types.TupleType)):
 546         raise SyntaxError,"lex: tokens must be a list or tuple."
 547
 548     # Build a dictionary of valid token names
 549     lexobj.lextokens = { }
 550     if not optimize:
 551         for n in tokens:
 552             if not _is_identifier.match(n):
 553                 print >>sys.stderr, "lex: Bad token name '%s'" % n
 554                 error = 1
 555             if warn and lexobj.lextokens.has_key(n):
 556                 print >>sys.stderr, "lex: Warning. Token '%s' multiply defined." % n
 557             lexobj.lextokens[n] = None
 558     else:
 559         for n in tokens: lexobj.lextokens[n] = None
 560
 561     if debug:
 562         print "lex: tokens = '%s'" % lexobj.lextokens.keys()
 563
 564     try:
 565          for c in literals:
 566                if not (isinstance(c,types.StringType) or isinstance(c,types.UnicodeType)) or len(c) > 1:
 567                     print >>sys.stderr, "lex: Invalid literal %s. Must be a single character" % repr(c)
 568                     error = 1
 569                     continue
 570
 571     except TypeError:
 572          print >>sys.stderr, "lex: Invalid literals specification. literals must be a sequence of characters."
 573          error = 1
 574
 575     lexobj.lexliterals = literals
 576
 577     # Build statemap
 578     if states:
 579          if not (isinstance(states,types.TupleType) or isinstance(states,types.ListType)):
 580               print >>sys.stderr, "lex: states must be defined as a tuple or list."
 581               error = 1
 582          else:
 583               for s in states:
 584                     if not isinstance(s,types.TupleType) or len(s) != 2:
 585                            print >>sys.stderr, "lex: invalid state specifier %s. Must be a tuple (statename,'exclusive|inclusive')" % repr(s)
 586                            error = 1
 587                            continue
 588                     name, statetype = s
 589                     if not isinstance(name,types.StringType):
 590                            print >>sys.stderr, "lex: state name %s must be a string" % repr(name)
 591                            error = 1
 592                            continue
 593                     if not (statetype == 'inclusive' or statetype == 'exclusive'):
 594                            print >>sys.stderr, "lex: state type for state %s must be 'inclusive' or 'exclusive'" % name
 595                            error = 1
 596                            continue
 597                     if stateinfo.has_key(name):
 598                            print >>sys.stderr, "lex: state '%s' already defined." % name
 599                            error = 1
 600                            continue
 601                     stateinfo[name] = statetype
 602
 603     # Get a list of symbols with the t_ or s_ prefix
 604     tsymbols = [f for f in ldict.keys() if f[:2] == 't_' ]
 605
 606     # Now build up a list of functions and a list of strings
 607
 608     funcsym =  { }        # Symbols defined as functions
 609     strsym =   { }        # Symbols defined as strings
 610     toknames = { }        # Mapping of symbols to token names
 611
 612     for s in stateinfo.keys():
 613          funcsym[s] = []
 614          strsym[s] = []
 615
 616     ignore   = { }        # Ignore strings by state
 617     errorf   = { }        # Error functions by state
 618
 619     if len(tsymbols) == 0:
 620         raise SyntaxError,"lex: no rules of the form t_rulename are defined."
 621
 622     for f in tsymbols:
 623         t = ldict[f]
 624         states, tokname = _statetoken(f,stateinfo)
 625         toknames[f] = tokname
 626
 627         if callable(t):
 628             for s in states: funcsym[s].append((f,t))
 629         elif (isinstance(t, types.StringType) or isinstance(t,types.UnicodeType)):
 630             for s in states: strsym[s].append((f,t))
 631         else:
 632             print >>sys.stderr, "lex: %s not defined as a function or string" % f
 633             error = 1
 634
 635     # Sort the functions by line number
 636     for f in funcsym.values():
 637         f.sort(lambda x,y: cmp(x[1].func_code.co_firstlineno,y[1].func_code.co_firstlineno))
 638
 639     # Sort the strings by regular expression length
 640     for s in strsym.values():
 641         s.sort(lambda x,y: (len(x[1]) < len(y[1])) - (len(x[1]) > len(y[1])))
 642
 643     regexs = { }
 644
 645     # Build the master regular expressions
 646     for state in stateinfo.keys():
 647         regex_list = []
 648
 649         # Add rules defined by functions first
 650         for fname, f in funcsym[state]:
 651             line = f.func_code.co_firstlineno
 652             file = f.func_code.co_filename
 653             files[file] = None
 654             tokname = toknames[fname]
 655
 656             ismethod = isinstance(f, types.MethodType)
 657
 658             if not optimize:
 659                 nargs = f.func_code.co_argcount
 660                 if ismethod:
 661                     reqargs = 2
 662                 else:
 663                     reqargs = 1
 664                 if nargs > reqargs:
 665                     print >>sys.stderr, "%s:%d: Rule '%s' has too many arguments." % (file,line,f.__name__)
 666                     error = 1
 667                     continue
 668
 669                 if nargs < reqargs:
 670                     print >>sys.stderr, "%s:%d: Rule '%s' requires an argument." % (file,line,f.__name__)
 671                     error = 1
 672                     continue
 673
 674                 if tokname == 'ignore':
 675                     print >>sys.stderr, "%s:%d: Rule '%s' must be defined as a string." % (file,line,f.__name__)
 676                     error = 1
 677                     continue
 678
 679             if tokname == 'error':
 680                 errorf[state] = f
 681                 continue
 682
 683             if f.__doc__:
 684                 if not optimize:
 685                     try:
 686                         c = re.compile("(?P<%s>%s)" % (f.__name__,f.__doc__), re.VERBOSE | reflags)
 687                         if c.match(""):
 688                              print >>sys.stderr, "%s:%d: Regular expression for rule '%s' matches empty string." % (file,line,f.__name__)
 689                              error = 1
 690                              continue
 691                     except re.error,e:
 692                         print >>sys.stderr, "%s:%d: Invalid regular expression for rule '%s'. %s" % (file,line,f.__name__,e)
 693                         if '#' in f.__doc__:
 694                              print >>sys.stderr, "%s:%d. Make sure '#' in rule '%s' is escaped with '\\#'." % (file,line, f.__name__)
 695                         error = 1
 696                         continue
 697
 698                     if debug:
 699                         print "lex: Adding rule %s -> '%s' (state '%s')" % (f.__name__,f.__doc__, state)
 700
 701                 # Okay. The regular expression seemed okay.  Let's append it to the master regular
 702                 # expression we're building
 703
 704                 regex_list.append("(?P<%s>%s)" % (f.__name__,f.__doc__))
 705             else:
 706                 print >>sys.stderr, "%s:%d: No regular expression defined for rule '%s'" % (file,line,f.__name__)
 707
 708         # Now add all of the simple rules
 709         for name,r in strsym[state]:
 710             tokname = toknames[name]
 711
 712             if tokname == 'ignore':
 713                  if "\\" in r:
 714                       print >>sys.stderr, "lex: Warning. %s contains a literal backslash '\\'" % name
 715                  ignore[state] = r
 716                  continue
 717
 718             if not optimize:
 719                 if tokname == 'error':
 720                     raise SyntaxError,"lex: Rule '%s' must be defined as a function" % name
 721                     error = 1
 722                     continue
 723
 724                 if not lexobj.lextokens.has_key(tokname) and tokname.find("ignore_") < 0:
 725                     print >>sys.stderr, "lex: Rule '%s' defined for an unspecified token %s." % (name,tokname)
 726                     error = 1
 727                     continue
 728                 try:
 729                     c = re.compile("(?P<%s>%s)" % (name,r),re.VERBOSE | reflags)
 730                     if (c.match("")):
 731                          print >>sys.stderr, "lex: Regular expression for rule '%s' matches empty string." % name
 732                          error = 1
 733                          continue
 734                 except re.error,e:
 735                     print >>sys.stderr, "lex: Invalid regular expression for rule '%s'. %s" % (name,e)
 736                     if '#' in r:
 737                          print >>sys.stderr, "lex: Make sure '#' in rule '%s' is escaped with '\\#'." % name
 738
 739                     error = 1
 740                     continue
 741                 if debug:
 742                     print "lex: Adding rule %s -> '%s' (state '%s')" % (name,r,state)
 743
 744             regex_list.append("(?P<%s>%s)" % (name,r))
 745
 746         if not regex_list:
 747              print >>sys.stderr, "lex: No rules defined for state '%s'" % state
 748              error = 1
 749
 750         regexs[state] = regex_list
 751
 752
 753     if not optimize:
 754         for f in files.keys():
 755            if not _validate_file(f):
 756                 error = 1
 757
 758     if error:
 759         raise SyntaxError,"lex: Unable to build lexer."
 760
 761     # From this point forward, we're reasonably confident that we can build the lexer.
 762     # No more errors will be generated, but there might be some warning messages.
 763
 764     # Build the master regular expressions
 765
 766     for state in regexs.keys():
 767         lexre, re_text = _form_master_re(regexs[state],reflags,ldict,toknames)
 768         lexobj.lexstatere[state] = lexre
 769         lexobj.lexstateretext[state] = re_text
 770         if debug:
 771             for i in range(len(re_text)):
 772                  print "lex: state '%s'. regex[%d] = '%s'" % (state, i, re_text[i])
 773
 774     # For inclusive states, we need to add the INITIAL state
 775     for state,type in stateinfo.items():
 776         if state != "INITIAL" and type == 'inclusive':
 777              lexobj.lexstatere[state].extend(lexobj.lexstatere['INITIAL'])
 778              lexobj.lexstateretext[state].extend(lexobj.lexstateretext['INITIAL'])
 779
 780     lexobj.lexstateinfo = stateinfo
 781     lexobj.lexre = lexobj.lexstatere["INITIAL"]
 782     lexobj.lexretext = lexobj.lexstateretext["INITIAL"]
 783
 784     # Set up ignore variables
 785     lexobj.lexstateignore = ignore
 786     lexobj.lexignore = lexobj.lexstateignore.get("INITIAL","")
 787
 788     # Set up error functions
 789     lexobj.lexstateerrorf = errorf
 790     lexobj.lexerrorf = errorf.get("INITIAL",None)
 791     if warn and not lexobj.lexerrorf:
 792         print >>sys.stderr, "lex: Warning. no t_error rule is defined."
 793
 794     # Check state information for ignore and error rules
 795     for s,stype in stateinfo.items():
 796         if stype == 'exclusive':
 797               if warn and not errorf.has_key(s):
 798                    print >>sys.stderr, "lex: Warning. no error rule is defined for exclusive state '%s'" % s
 799               if warn and not ignore.has_key(s) and lexobj.lexignore:
 800                    print >>sys.stderr, "lex: Warning. no ignore rule is defined for exclusive state '%s'" % s
 801         elif stype == 'inclusive':
 802               if not errorf.has_key(s):
 803                    errorf[s] = errorf.get("INITIAL",None)
 804               if not ignore.has_key(s):
 805                    ignore[s] = ignore.get("INITIAL","")
 806
 807
 808     # Create global versions of the token() and input() functions
 809     token = lexobj.token
 810     input = lexobj.input
 811     lexer = lexobj
 812
 813     # If in optimize mode, we write the lextab
 814     if lextab and optimize:
 815         lexobj.writetab(lextab)
 816
 817     return lexobj
 818
 819 # -----------------------------------------------------------------------------
 820 # runmain()
 821 #
 822 # This runs the lexer as a main program
 823 # -----------------------------------------------------------------------------
 824
 825 def runmain(lexer=None,data=None):
 826     if not data:
 827         try:
 828             filename = sys.argv[1]
 829             f = open(filename)
 830             data = f.read()
 831             f.close()
 832         except IndexError:
 833             print "Reading from standard input (type EOF to end):"
 834             data = sys.stdin.read()
 835
 836     if lexer:
 837         _input = lexer.input
 838     else:
 839         _input = input
 840     _input(data)
 841     if lexer:
 842         _token = lexer.token
 843     else:
 844         _token = token
 845
 846     while 1:
 847         tok = _token()
 848         if not tok: break
 849         print "(%s,%r,%d,%d)" % (tok.type, tok.value, tok.lineno,tok.lexpos)
 850
 851
 852 # -----------------------------------------------------------------------------
 853 # @TOKEN(regex)
 854 #
 855 # This decorator function can be used to set the regex expression on a function
 856 # when its docstring might need to be set in an alternative way
 857 # -----------------------------------------------------------------------------
 858
 859 def TOKEN(r):
 860     def set_doc(f):
 861         f.__doc__ = r
 862         return f
 863     return set_doc
 864
 865 # Alternative spelling of the TOKEN decorator
 866 Token = TOKEN
 867