# ----------------------------------------------------------------------------- # ply: yacc.py # # Copyright (C) 2001-2009, # David M. Beazley (Dabeaz LLC) # All rights reserved. # # Redistribution and use in source and binary forms, with or without # modification, are permitted provided that the following conditions are # met: # # * Redistributions of source code must retain the above copyright notice, # this list of conditions and the following disclaimer. # * Redistributions in binary form must reproduce the above copyright notice, # this list of conditions and the following disclaimer in the documentation # and/or other materials provided with the distribution. # * Neither the name of the David Beazley or Dabeaz LLC may be used to # endorse or promote products derived from this software without # specific prior written permission. # # THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS # "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT # LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR # A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT # OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, # SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT # LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, # DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY # THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT # (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE # OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. # ----------------------------------------------------------------------------- # # This implements an LR parser that is constructed from grammar rules defined # as Python functions. The grammer is specified by supplying the BNF inside # Python documentation strings. The inspiration for this technique was borrowed # from John Aycock's Spark parsing system. PLY might be viewed as cross between # Spark and the GNU bison utility. # # The current implementation is only somewhat object-oriented. The # LR parser itself is defined in terms of an object (which allows multiple # parsers to co-exist). However, most of the variables used during table # construction are defined in terms of global variables. Users shouldn't # notice unless they are trying to define multiple parsers at the same # time using threads (in which case they should have their head examined). # # This implementation supports both SLR and LALR(1) parsing. LALR(1) # support was originally implemented by Elias Ioup (ezioup@alumni.uchicago.edu), # using the algorithm found in Aho, Sethi, and Ullman "Compilers: Principles, # Techniques, and Tools" (The Dragon Book). LALR(1) has since been replaced # by the more efficient DeRemer and Pennello algorithm. # # :::::::: WARNING ::::::: # # Construction of LR parsing tables is fairly complicated and expensive. # To make this module run fast, a *LOT* of work has been put into # optimization---often at the expensive of readability and what might # consider to be good Python "coding style." Modify the code at your # own risk! # ---------------------------------------------------------------------------- __version__ = "3.3" __tabversion__ = "3.2" # Table version #----------------------------------------------------------------------------- # === User configurable parameters === # # Change these to modify the default behavior of yacc (if you wish) #----------------------------------------------------------------------------- yaccdebug = 0 # Debugging mode. If set, yacc generates a # a 'parser.out' file in the current directory debug_file = 'parser.out' # Default name of the debugging file tab_module = 'parsetab' # Default name of the table module default_lr = 'LALR' # Default LR table generation method error_count = 3 # Number of symbols that must be shifted to leave recovery mode yaccdevel = 0 # Set to True if developing yacc. This turns off optimized # implementations of certain functions. resultlimit = 40 # Size limit of results when running in debug mode. pickle_protocol = 0 # Protocol to use when writing pickle files import re, types, sys, os.path # Compatibility function for python 2.6/3.0 if sys.version_info[0] < 3: def func_code(f): return f.func_code else: def func_code(f): return f.__code__ # Compatibility try: MAXINT = sys.maxint except AttributeError: MAXINT = sys.maxsize # Python 2.x/3.0 compatibility. def load_ply_lex(): if sys.version_info[0] < 3: import lex else: import ply.lex as lex return lex # This object is a stand-in for a logging object created by the # logging module. PLY will use this by default to create things # such as the parser.out file. If a user wants more detailed # information, they can create their own logging object and pass # it into PLY. class PlyLogger(object): def __init__(self,f): self.f = f def debug(self,msg,*args,**kwargs): self.f.write((msg % args) + "\n") info = debug def warning(self,msg,*args,**kwargs): self.f.write("WARNING: "+ (msg % args) + "\n") def error(self,msg,*args,**kwargs): self.f.write("ERROR: " + (msg % args) + "\n") critical = debug # Null logger is used when no output is generated. Does nothing. class NullLogger(object): def __getattribute__(self,name): return self def __call__(self,*args,**kwargs): return self # Exception raised for yacc-related errors class YaccError(Exception): pass # Format the result message that the parser produces when running in debug mode. def format_result(r): repr_str = repr(r) if '\n' in repr_str: repr_str = repr(repr_str) if len(repr_str) > resultlimit: repr_str = repr_str[:resultlimit]+" ..." result = "<%s @ 0x%x> (%s)" % (type(r).__name__,id(r),repr_str) return result # Format stack entries when the parser is running in debug mode def format_stack_entry(r): repr_str = repr(r) if '\n' in repr_str: repr_str = repr(repr_str) if len(repr_str) < 16: return repr_str else: return "<%s @ 0x%x>" % (type(r).__name__,id(r)) #----------------------------------------------------------------------------- # === LR Parsing Engine === # # The following classes are used for the LR parser itself. These are not # used during table construction and are independent of the actual LR # table generation algorithm #----------------------------------------------------------------------------- # This class is used to hold non-terminal grammar symbols during parsing. # It normally has the following attributes set: # .type = Grammar symbol type # .value = Symbol value # .lineno = Starting line number # .endlineno = Ending line number (optional, set automatically) # .lexpos = Starting lex position # .endlexpos = Ending lex position (optional, set automatically) class YaccSymbol: def __str__(self): return self.type def __repr__(self): return str(self) # This class is a wrapper around the objects actually passed to each # grammar rule. Index lookup and assignment actually assign the # .value attribute of the underlying YaccSymbol object. # The lineno() method returns the line number of a given # item (or 0 if not defined). The linespan() method returns # a tuple of (startline,endline) representing the range of lines # for a symbol. The lexspan() method returns a tuple (lexpos,endlexpos) # representing the range of positional information for a symbol. class YaccProduction: def __init__(self,s,stack=None): self.slice = s self.stack = stack self.lexer = None self.parser= None def __getitem__(self,n): if n >= 0: return self.slice[n].value else: return self.stack[n].value def __setitem__(self,n,v): self.slice[n].value = v def __getslice__(self,i,j): return [s.value for s in self.slice[i:j]] def __len__(self): return len(self.slice) def lineno(self,n): return getattr(self.slice[n],"lineno",0) def set_lineno(self,n,lineno): self.slice[n].lineno = lineno def linespan(self,n): startline = getattr(self.slice[n],"lineno",0) endline = getattr(self.slice[n],"endlineno",startline) return startline,endline def lexpos(self,n): return getattr(self.slice[n],"lexpos",0) def lexspan(self,n): startpos = getattr(self.slice[n],"lexpos",0) endpos = getattr(self.slice[n],"endlexpos",startpos) return startpos,endpos def error(self): raise SyntaxError # ----------------------------------------------------------------------------- # == LRParser == # # The LR Parsing engine. # ----------------------------------------------------------------------------- class LRParser: def __init__(self,lrtab,errorf): self.productions = lrtab.lr_productions self.action = lrtab.lr_action self.goto = lrtab.lr_goto self.errorfunc = errorf def errok(self): self.errorok = 1 def restart(self): del self.statestack[:] del self.symstack[:] sym = YaccSymbol() sym.type = '$end' self.symstack.append(sym) self.statestack.append(0) def parse(self,input=None,lexer=None,debug=0,tracking=0,tokenfunc=None): if debug or yaccdevel: if isinstance(debug,int): debug = PlyLogger(sys.stderr) return self.parsedebug(input,lexer,debug,tracking,tokenfunc) elif tracking: return self.parseopt(input,lexer,debug,tracking,tokenfunc) else: return self.parseopt_notrack(input,lexer,debug,tracking,tokenfunc) # !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! # parsedebug(). # # This is the debugging enabled version of parse(). All changes made to the # parsing engine should be made here. For the non-debugging version, # copy this code to a method parseopt() and delete all of the sections # enclosed in: # # #--! DEBUG # statements # #--! DEBUG # # !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! def parsedebug(self,input=None,lexer=None,debug=None,tracking=0,tokenfunc=None): lookahead = None # Current lookahead symbol lookaheadstack = [ ] # Stack of lookahead symbols actions = self.action # Local reference to action table (to avoid lookup on self.) goto = self.goto # Local reference to goto table (to avoid lookup on self.) prod = self.productions # Local reference to production list (to avoid lookup on self.) pslice = YaccProduction(None) # Production object passed to grammar rules errorcount = 0 # Used during error recovery # --! DEBUG debug.info("PLY: PARSE DEBUG START") # --! DEBUG # If no lexer was given, we will try to use the lex module if not lexer: lex = load_ply_lex() lexer = lex.lexer # Set up the lexer and parser objects on pslice pslice.lexer = lexer pslice.parser = self # If input was supplied, pass to lexer if input is not None: lexer.input(input) if tokenfunc is None: # Tokenize function get_token = lexer.token else: get_token = tokenfunc # Set up the state and symbol stacks statestack = [ ] # Stack of parsing states self.statestack = statestack symstack = [ ] # Stack of grammar symbols self.symstack = symstack pslice.stack = symstack # Put in the production errtoken = None # Err token # The start state is assumed to be (0,$end) statestack.append(0) sym = YaccSymbol() sym.type = "$end" symstack.append(sym) state = 0 while 1: # Get the next symbol on the input. If a lookahead symbol # is already set, we just use that. Otherwise, we'll pull # the next token off of the lookaheadstack or from the lexer # --! DEBUG debug.debug('') debug.debug('State : %s', state) # --! DEBUG if not lookahead: if not lookaheadstack: lookahead = get_token() # Get the next token else: lookahead = lookaheadstack.pop() if not lookahead: lookahead = YaccSymbol() lookahead.type = "$end" # --! DEBUG debug.debug('Stack : %s', ("%s . %s" % (" ".join([xx.type for xx in symstack][1:]), str(lookahead))).lstrip()) # --! DEBUG # Check the action table ltype = lookahead.type t = actions[state].get(ltype) if t is not None: if t > 0: # shift a symbol on the stack statestack.append(t) state = t # --! DEBUG debug.debug("Action : Shift and goto state %s", t) # --! DEBUG symstack.append(lookahead) lookahead = None # Decrease error count on successful shift if errorcount: errorcount -=1 continue if t < 0: # reduce a symbol on the stack, emit a production p = prod[-t] pname = p.name plen = p.len # Get production function sym = YaccSymbol() sym.type = pname # Production name sym.value = None # --! DEBUG if plen: debug.info("Action : Reduce rule [%s] with %s and goto state %d", p.str, "["+",".join([format_stack_entry(_v.value) for _v in symstack[-plen:]])+"]",-t) else: debug.info("Action : Reduce rule [%s] with %s and goto state %d", p.str, [],-t) # --! DEBUG if plen: targ = symstack[-plen-1:] targ[0] = sym # --! TRACKING if tracking: t1 = targ[1] sym.lineno = t1.lineno sym.lexpos = t1.lexpos t1 = targ[-1] sym.endlineno = getattr(t1,"endlineno",t1.lineno) sym.endlexpos = getattr(t1,"endlexpos",t1.lexpos) # --! TRACKING # !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! # The code enclosed in this section is duplicated # below as a performance optimization. Make sure # changes get made in both locations. pslice.slice = targ try: # Call the grammar rule with our special slice object del symstack[-plen:] del statestack[-plen:] p.callable(pslice) # --! DEBUG debug.info("Result : %s", format_result(pslice[0])) # --! DEBUG symstack.append(sym) state = goto[statestack[-1]][pname] statestack.append(state) except SyntaxError: # If an error was set. Enter error recovery state lookaheadstack.append(lookahead) symstack.pop() statestack.pop() state = statestack[-1] sym.type = 'error' lookahead = sym errorcount = error_count self.errorok = 0 continue # !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! else: # --! TRACKING if tracking: sym.lineno = lexer.lineno sym.lexpos = lexer.lexpos # --! TRACKING targ = [ sym ] # !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! # The code enclosed in this section is duplicated # above as a performance optimization. Make sure # changes get made in both locations. pslice.slice = targ try: # Call the grammar rule with our special slice object p.callable(pslice) # --! DEBUG debug.info("Result : %s", format_result(pslice[0])) # --! DEBUG symstack.append(sym) state = goto[statestack[-1]][pname] statestack.append(state) except SyntaxError: # If an error was set. Enter error recovery state lookaheadstack.append(lookahead) symstack.pop() statestack.pop() state = statestack[-1] sym.type = 'error' lookahead = sym errorcount = error_count self.errorok = 0 continue # !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! if t == 0: n = symstack[-1] result = getattr(n,"value",None) # --! DEBUG debug.info("Done : Returning %s", format_result(result)) debug.info("PLY: PARSE DEBUG END") # --! DEBUG return result if t == None: # --! DEBUG debug.error('Error : %s', ("%s . %s" % (" ".join([xx.type for xx in symstack][1:]), str(lookahead))).lstrip()) # --! DEBUG # We have some kind of parsing error here. To handle # this, we are going to push the current token onto # the tokenstack and replace it with an 'error' token. # If there are any synchronization rules, they may # catch it. # # In addition to pushing the error token, we call call # the user defined p_error() function if this is the # first syntax error. This function is only called if # errorcount == 0. if errorcount == 0 or self.errorok: errorcount = error_count self.errorok = 0 errtoken = lookahead if errtoken.type == "$end": errtoken = None # End of file! if self.errorfunc: global errok,token,restart errok = self.errok # Set some special functions available in error recovery token = get_token restart = self.restart if errtoken and not hasattr(errtoken,'lexer'): errtoken.lexer = lexer tok = self.errorfunc(errtoken) del errok, token, restart # Delete special functions if self.errorok: # User must have done some kind of panic # mode recovery on their own. The # returned token is the next lookahead lookahead = tok errtoken = None continue else: if errtoken: if hasattr(errtoken,"lineno"): lineno = lookahead.lineno else: lineno = 0 if lineno: sys.stderr.write("yacc: Syntax error at line %d, token=%s\n" % (lineno, errtoken.type)) else: sys.stderr.write("yacc: Syntax error, token=%s" % errtoken.type) else: sys.stderr.write("yacc: Parse error in input. EOF\n") return else: errorcount = error_count # case 1: the statestack only has 1 entry on it. If we're in this state, the # entire parse has been rolled back and we're completely hosed. The token is # discarded and we just keep going. if len(statestack) <= 1 and lookahead.type != "$end": lookahead = None errtoken = None state = 0 # Nuke the pushback stack del lookaheadstack[:] continue # case 2: the statestack has a couple of entries on it, but we're # at the end of the file. nuke the top entry and generate an error token # Start nuking entries on the stack if lookahead.type == "$end": # Whoa. We're really hosed here. Bail out return if lookahead.type != 'error': sym = symstack[-1] if sym.type == 'error': # Hmmm. Error is on top of stack, we'll just nuke input # symbol and continue lookahead = None continue t = YaccSymbol() t.type = 'error' if hasattr(lookahead,"lineno"): t.lineno = lookahead.lineno t.value = lookahead lookaheadstack.append(lookahead) lookahead = t else: symstack.pop() statestack.pop() state = statestack[-1] # Potential bug fix continue # Call an error function here raise RuntimeError("yacc: internal parser error!!!\n") # !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! # parseopt(). # # Optimized version of parse() method. DO NOT EDIT THIS CODE DIRECTLY. # Edit the debug version above, then copy any modifications to the method # below while removing #--! DEBUG sections. # !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! def parseopt(self,input=None,lexer=None,debug=0,tracking=0,tokenfunc=None): lookahead = None # Current lookahead symbol lookaheadstack = [ ] # Stack of lookahead symbols actions = self.action # Local reference to action table (to avoid lookup on self.) goto = self.goto # Local reference to goto table (to avoid lookup on self.) prod = self.productions # Local reference to production list (to avoid lookup on self.) pslice = YaccProduction(None) # Production object passed to grammar rules errorcount = 0 # Used during error recovery # If no lexer was given, we will try to use the lex module if not lexer: lex = load_ply_lex() lexer = lex.lexer # Set up the lexer and parser objects on pslice pslice.lexer = lexer pslice.parser = self # If input was supplied, pass to lexer if input is not None: lexer.input(input) if tokenfunc is None: # Tokenize function get_token = lexer.token else: get_token = tokenfunc # Set up the state and symbol stacks statestack = [ ] # Stack of parsing states self.statestack = statestack symstack = [ ] # Stack of grammar symbols self.symstack = symstack pslice.stack = symstack # Put in the production errtoken = None # Err token # The start state is assumed to be (0,$end) statestack.append(0) sym = YaccSymbol() sym.type = '$end' symstack.append(sym) state = 0 while 1: # Get the next symbol on the input. If a lookahead symbol # is already set, we just use that. Otherwise, we'll pull # the next token off of the lookaheadstack or from the lexer if not lookahead: if not lookaheadstack: lookahead = get_token() # Get the next token else: lookahead = lookaheadstack.pop() if not lookahead: lookahead = YaccSymbol() lookahead.type = '$end' # Check the action table ltype = lookahead.type t = actions[state].get(ltype) if t is not None: if t > 0: # shift a symbol on the stack statestack.append(t) state = t symstack.append(lookahead) lookahead = None # Decrease error count on successful shift if errorcount: errorcount -=1 continue if t < 0: # reduce a symbol on the stack, emit a production p = prod[-t] pname = p.name plen = p.len # Get production function sym = YaccSymbol() sym.type = pname # Production name sym.value = None if plen: targ = symstack[-plen-1:] targ[0] = sym # --! TRACKING if tracking: t1 = targ[1] sym.lineno = t1.lineno sym.lexpos = t1.lexpos t1 = targ[-1] sym.endlineno = getattr(t1,"endlineno",t1.lineno) sym.endlexpos = getattr(t1,"endlexpos",t1.lexpos) # --! TRACKING # !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! # The code enclosed in this section is duplicated # below as a performance optimization. Make sure # changes get made in both locations. pslice.slice = targ try: # Call the grammar rule with our special slice object del symstack[-plen:] del statestack[-plen:] p.callable(pslice) symstack.append(sym) state = goto[statestack[-1]][pname] statestack.append(state) except SyntaxError: # If an error was set. Enter error recovery state lookaheadstack.append(lookahead) symstack.pop() statestack.pop() state = statestack[-1] sym.type = 'error' lookahead = sym errorcount = error_count self.errorok = 0 continue # !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! else: # --! TRACKING if tracking: sym.lineno = lexer.lineno sym.lexpos = lexer.lexpos # --! TRACKING targ = [ sym ] # !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! # The code enclosed in this section is duplicated # above as a performance optimization. Make sure # changes get made in both locations. pslice.slice = targ try: # Call the grammar rule with our special slice object p.callable(pslice) symstack.append(sym) state = goto[statestack[-1]][pname] statestack.append(state) except SyntaxError: # If an error was set. Enter error recovery state lookaheadstack.append(lookahead) symstack.pop() statestack.pop() state = statestack[-1] sym.type = 'error' lookahead = sym errorcount = error_count self.errorok = 0 continue # !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! if t == 0: n = symstack[-1] return getattr(n,"value",None) if t == None: # We have some kind of parsing error here. To handle # this, we are going to push the current token onto # the tokenstack and replace it with an 'error' token. # If there are any synchronization rules, they may # catch it. # # In addition to pushing the error token, we call call # the user defined p_error() function if this is the # first syntax error. This function is only called if # errorcount == 0. if errorcount == 0 or self.errorok: errorcount = error_count self.errorok = 0 errtoken = lookahead if errtoken.type == '$end': errtoken = None # End of file! if self.errorfunc: global errok,token,restart errok = self.errok # Set some special functions available in error recovery token = get_token restart = self.restart if errtoken and not hasattr(errtoken,'lexer'): errtoken.lexer = lexer tok = self.errorfunc(errtoken) del errok, token, restart # Delete special functions if self.errorok: # User must have done some kind of panic # mode recovery on their own. The # returned token is the next lookahead lookahead = tok errtoken = None continue else: if errtoken: if hasattr(errtoken,"lineno"): lineno = lookahead.lineno else: lineno = 0 if lineno: sys.stderr.write("yacc: Syntax error at line %d, token=%s\n" % (lineno, errtoken.type)) else: sys.stderr.write("yacc: Syntax error, token=%s" % errtoken.type) else: sys.stderr.write("yacc: Parse error in input. EOF\n") return else: errorcount = error_count # case 1: the statestack only has 1 entry on it. If we're in this state, the # entire parse has been rolled back and we're completely hosed. The token is # discarded and we just keep going. if len(statestack) <= 1 and lookahead.type != '$end': lookahead = None errtoken = None state = 0 # Nuke the pushback stack del lookaheadstack[:] continue # case 2: the statestack has a couple of entries on it, but we're # at the end of the file. nuke the top entry and generate an error token # Start nuking entries on the stack if lookahead.type == '$end': # Whoa. We're really hosed here. Bail out return if lookahead.type != 'error': sym = symstack[-1] if sym.type == 'error': # Hmmm. Error is on top of stack, we'll just nuke input # symbol and continue lookahead = None continue t = YaccSymbol() t.type = 'error' if hasattr(lookahead,"lineno"): t.lineno = lookahead.lineno t.value = lookahead lookaheadstack.append(lookahead) lookahead = t else: symstack.pop() statestack.pop() state = statestack[-1] # Potential bug fix continue # Call an error function here raise RuntimeError("yacc: internal parser error!!!\n") # !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! # parseopt_notrack(). # # Optimized version of parseopt() with line number tracking removed. # DO NOT EDIT THIS CODE DIRECTLY. Copy the optimized version and remove # code in the #--! TRACKING sections # !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! def parseopt_notrack(self,input=None,lexer=None,debug=0,tracking=0,tokenfunc=None): lookahead = None # Current lookahead symbol lookaheadstack = [ ] # Stack of lookahead symbols actions = self.action # Local reference to action table (to avoid lookup on self.) goto = self.goto # Local reference to goto table (to avoid lookup on self.) prod = self.productions # Local reference to production list (to avoid lookup on self.) pslice = YaccProduction(None) # Production object passed to grammar rules errorcount = 0 # Used during error recovery # If no lexer was given, we will try to use the lex module if not lexer: lex = load_ply_lex() lexer = lex.lexer # Set up the lexer and parser objects on pslice pslice.lexer = lexer pslice.parser = self # If input was supplied, pass to lexer if input is not None: lexer.input(input) if tokenfunc is None: # Tokenize function get_token = lexer.token else: get_token = tokenfunc # Set up the state and symbol stacks statestack = [ ] # Stack of parsing states self.statestack = statestack symstack = [ ] # Stack of grammar symbols self.symstack = symstack pslice.stack = symstack # Put in the production errtoken = None # Err token # The start state is assumed to be (0,$end) statestack.append(0) sym = YaccSymbol() sym.type = '$end' symstack.append(sym) state = 0 while 1: # Get the next symbol on the input. If a lookahead symbol # is already set, we just use that. Otherwise, we'll pull # the next token off of the lookaheadstack or from the lexer if not lookahead: if not lookaheadstack: lookahead = get_token() # Get the next token else: lookahead = lookaheadstack.pop() if not lookahead: lookahead = YaccSymbol() lookahead.type = '$end' # Check the action table ltype = lookahead.type t = actions[state].get(ltype) if t is not None: if t > 0: # shift a symbol on the stack statestack.append(t) state = t symstack.append(lookahead) lookahead = None # Decrease error count on successful shift if errorcount: errorcount -=1 continue if t < 0: # reduce a symbol on the stack, emit a production p = prod[-t] pname = p.name plen = p.len # Get production function sym = YaccSymbol() sym.type = pname # Production name sym.value = None if plen: targ = symstack[-plen-1:] targ[0] = sym # !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! # The code enclosed in this section is duplicated # below as a performance optimization. Make sure # changes get made in both locations. pslice.slice = targ try: # Call the grammar rule with our special slice object del symstack[-plen:] del statestack[-plen:] p.callable(pslice) symstack.append(sym) state = goto[statestack[-1]][pname] statestack.append(state) except SyntaxError: # If an error was set. Enter error recovery state lookaheadstack.append(lookahead) symstack.pop() statestack.pop() state = statestack[-1] sym.type = 'error' lookahead = sym errorcount = error_count self.errorok = 0 continue # !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! else: targ = [ sym ] # !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! # The code enclosed in this section is duplicated # above as a performance optimization. Make sure # changes get made in both locations. pslice.slice = targ try: # Call the grammar rule with our special slice object p.callable(pslice) symstack.append(sym) state = goto[statestack[-1]][pname] statestack.append(state) except SyntaxError: # If an error was set. Enter error recovery state lookaheadstack.append(lookahead) symstack.pop() statestack.pop() state = statestack[-1] sym.type = 'error' lookahead = sym errorcount = error_count self.errorok = 0 continue # !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! if t == 0: n = symstack[-1] return getattr(n,"value",None) if t == None: # We have some kind of parsing error here. To handle # this, we are going to push the current token onto # the tokenstack and replace it with an 'error' token. # If there are any synchronization rules, they may # catch it. # # In addition to pushing the error token, we call call # the user defined p_error() function if this is the # first syntax error. This function is only called if # errorcount == 0. if errorcount == 0 or self.errorok: errorcount = error_count self.errorok = 0 errtoken = lookahead if errtoken.type == '$end': errtoken = None # End of file! if self.errorfunc: global errok,token,restart errok = self.errok # Set some special functions available in error recovery token = get_token restart = self.restart if errtoken and not hasattr(errtoken,'lexer'): errtoken.lexer = lexer tok = self.errorfunc(errtoken) del errok, token, restart # Delete special functions if self.errorok: # User must have done some kind of panic # mode recovery on their own. The # returned token is the next lookahead lookahead = tok errtoken = None continue else: if errtoken: if hasattr(errtoken,"lineno"): lineno = lookahead.lineno else: lineno = 0 if lineno: sys.stderr.write("yacc: Syntax error at line %d, token=%s\n" % (lineno, errtoken.type)) else: sys.stderr.write("yacc: Syntax error, token=%s" % errtoken.type) else: sys.stderr.write("yacc: Parse error in input. EOF\n") return else: errorcount = error_count # case 1: the statestack only has 1 entry on it. If we're in this state, the # entire parse has been rolled back and we're completely hosed. The token is # discarded and we just keep going. if len(statestack) <= 1 and lookahead.type != '$end': lookahead = None errtoken = None state = 0 # Nuke the pushback stack del lookaheadstack[:] continue # case 2: the statestack has a couple of entries on it, but we're # at the end of the file. nuke the top entry and generate an error token # Start nuking entries on the stack if lookahead.type == '$end': # Whoa. We're really hosed here. Bail out return if lookahead.type != 'error': sym = symstack[-1] if sym.type == 'error': # Hmmm. Error is on top of stack, we'll just nuke input # symbol and continue lookahead = None continue t = YaccSymbol() t.type = 'error' if hasattr(lookahead,"lineno"): t.lineno = lookahead.lineno t.value = lookahead lookaheadstack.append(lookahead) lookahead = t else: symstack.pop() statestack.pop() state = statestack[-1] # Potential bug fix continue # Call an error function here raise RuntimeError("yacc: internal parser error!!!\n") # ----------------------------------------------------------------------------- # === Grammar Representation === # # The following functions, classes, and variables are used to represent and # manipulate the rules that make up a grammar. # ----------------------------------------------------------------------------- import re # regex matching identifiers _is_identifier = re.compile(r'^[a-zA-Z0-9_-]+$') # ----------------------------------------------------------------------------- # class Production: # # This class stores the raw information about a single production or grammar rule. # A grammar rule refers to a specification such as this: # # expr : expr PLUS term # # Here are the basic attributes defined on all productions # # name - Name of the production. For example 'expr' # prod - A list of symbols on the right side ['expr','PLUS','term'] # prec - Production precedence level # number - Production number. # func - Function that executes on reduce # file - File where production function is defined # lineno - Line number where production function is defined # # The following attributes are defined or optional. # # len - Length of the production (number of symbols on right hand side) # usyms - Set of unique symbols found in the production # ----------------------------------------------------------------------------- class Production(object): reduced = 0 def __init__(self,number,name,prod,precedence=('right',0),func=None,file='',line=0): self.name = name self.prod = tuple(prod) self.number = number self.func = func self.callable = None self.file = file self.line = line self.prec = precedence # Internal settings used during table construction self.len = len(self.prod) # Length of the production # Create a list of unique production symbols used in the production self.usyms = [ ] for s in self.prod: if s not in self.usyms: self.usyms.append(s) # List of all LR items for the production self.lr_items = [] self.lr_next = None # Create a string representation if self.prod: self.str = "%s -> %s" % (self.name," ".join(self.prod)) else: self.str = "%s -> " % self.name def __str__(self): return self.str def __repr__(self): return "Production("+str(self)+")" def __len__(self): return len(self.prod) def __nonzero__(self): return 1 def __getitem__(self,index): return self.prod[index] # Return the nth lr_item from the production (or None if at the end) def lr_item(self,n): if n > len(self.prod): return None p = LRItem(self,n) # Precompute the list of productions immediately following. Hack. Remove later try: p.lr_after = Prodnames[p.prod[n+1]] except (IndexError,KeyError): p.lr_after = [] try: p.lr_before = p.prod[n-1] except IndexError: p.lr_before = None return p # Bind the production function name to a callable def bind(self,pdict): if self.func: self.callable = pdict[self.func] # This class serves as a minimal standin for Production objects when # reading table data from files. It only contains information # actually used by the LR parsing engine, plus some additional # debugging information. class MiniProduction(object): def __init__(self,str,name,len,func,file,line): self.name = name self.len = len self.func = func self.callable = None self.file = file self.line = line self.str = str def __str__(self): return self.str def __repr__(self): return "MiniProduction(%s)" % self.str # Bind the production function name to a callable def bind(self,pdict): if self.func: self.callable = pdict[self.func] # ----------------------------------------------------------------------------- # class LRItem # # This class represents a specific stage of parsing a production rule. For # example: # # expr : expr . PLUS term # # In the above, the "." represents the current location of the parse. Here # basic attributes: # # name - Name of the production. For example 'expr' # prod - A list of symbols on the right side ['expr','.', 'PLUS','term'] # number - Production number. # # lr_next Next LR item. Example, if we are ' expr -> expr . PLUS term' # then lr_next refers to 'expr -> expr PLUS . term' # lr_index - LR item index (location of the ".") in the prod list. # lookaheads - LALR lookahead symbols for this item # len - Length of the production (number of symbols on right hand side) # lr_after - List of all productions that immediately follow # lr_before - Grammar symbol immediately before # ----------------------------------------------------------------------------- class LRItem(object): def __init__(self,p,n): self.name = p.name self.prod = list(p.prod) self.number = p.number self.lr_index = n self.lookaheads = { } self.prod.insert(n,".") self.prod = tuple(self.prod) self.len = len(self.prod) self.usyms = p.usyms def __str__(self): if self.prod: s = "%s -> %s" % (self.name," ".join(self.prod)) else: s = "%s -> " % self.name return s def __repr__(self): return "LRItem("+str(self)+")" # ----------------------------------------------------------------------------- # rightmost_terminal() # # Return the rightmost terminal from a list of symbols. Used in add_production() # ----------------------------------------------------------------------------- def rightmost_terminal(symbols, terminals): i = len(symbols) - 1 while i >= 0: if symbols[i] in terminals: return symbols[i] i -= 1 return None # ----------------------------------------------------------------------------- # === GRAMMAR CLASS === # # The following class represents the contents of the specified grammar along # with various computed properties such as first sets, follow sets, LR items, etc. # This data is used for critical parts of the table generation process later. # ----------------------------------------------------------------------------- class GrammarError(YaccError): pass class Grammar(object): def __init__(self,terminals): self.Productions = [None] # A list of all of the productions. The first # entry is always reserved for the purpose of # building an augmented grammar self.Prodnames = { } # A dictionary mapping the names of nonterminals to a list of all # productions of that nonterminal. self.Prodmap = { } # A dictionary that is only used to detect duplicate # productions. self.Terminals = { } # A dictionary mapping the names of terminal symbols to a # list of the rules where they are used. for term in terminals: self.Terminals[term] = [] self.Terminals['error'] = [] self.Nonterminals = { } # A dictionary mapping names of nonterminals to a list # of rule numbers where they are used. self.First = { } # A dictionary of precomputed FIRST(x) symbols self.Follow = { } # A dictionary of precomputed FOLLOW(x) symbols self.Precedence = { } # Precedence rules for each terminal. Contains tuples of the # form ('right',level) or ('nonassoc', level) or ('left',level) self.UsedPrecedence = { } # Precedence rules that were actually used by the grammer. # This is only used to provide error checking and to generate # a warning about unused precedence rules. self.Start = None # Starting symbol for the grammar def __len__(self): return len(self.Productions) def __getitem__(self,index): return self.Productions[index] # ----------------------------------------------------------------------------- # set_precedence() # # Sets the precedence for a given terminal. assoc is the associativity such as # 'left','right', or 'nonassoc'. level is a numeric level. # # ----------------------------------------------------------------------------- def set_precedence(self,term,assoc,level): assert self.Productions == [None],"Must call set_precedence() before add_production()" if term in self.Precedence: raise GrammarError("Precedence already specified for terminal '%s'" % term) if assoc not in ['left','right','nonassoc']: raise GrammarError("Associativity must be one of 'left','right', or 'nonassoc'") self.Precedence[term] = (assoc,level) # ----------------------------------------------------------------------------- # add_production() # # Given an action function, this function assembles a production rule and # computes its precedence level. # # The production rule is supplied as a list of symbols. For example, # a rule such as 'expr : expr PLUS term' has a production name of 'expr' and # symbols ['expr','PLUS','term']. # # Precedence is determined by the precedence of the right-most non-terminal # or the precedence of a terminal specified by %prec. # # A variety of error checks are performed to make sure production symbols # are valid and that %prec is used correctly. # ----------------------------------------------------------------------------- def add_production(self,prodname,syms,func=None,file='',line=0): if prodname in self.Terminals: raise GrammarError("%s:%d: Illegal rule name '%s'. Already defined as a token" % (file,line,prodname)) if prodname == 'error': raise GrammarError("%s:%d: Illegal rule name '%s'. error is a reserved word" % (file,line,prodname)) if not _is_identifier.match(prodname): raise GrammarError("%s:%d: Illegal rule name '%s'" % (file,line,prodname)) # Look for literal tokens for n,s in enumerate(syms): if s[0] in "'\"": try: c = eval(s) if (len(c) > 1): raise GrammarError("%s:%d: Literal token %s in rule '%s' may only be a single character" % (file,line,s, prodname)) if not c in self.Terminals: self.Terminals[c] = [] syms[n] = c continue except SyntaxError: pass if not _is_identifier.match(s) and s != '%prec': raise GrammarError("%s:%d: Illegal name '%s' in rule '%s'" % (file,line,s, prodname)) # Determine the precedence level if '%prec' in syms: if syms[-1] == '%prec': raise GrammarError("%s:%d: Syntax error. Nothing follows %%prec" % (file,line)) if syms[-2] != '%prec': raise GrammarError("%s:%d: Syntax error. %%prec can only appear at the end of a grammar rule" % (file,line)) precname = syms[-1] prodprec = self.Precedence.get(precname,None) if not prodprec: raise GrammarError("%s:%d: Nothing known about the precedence of '%s'" % (file,line,precname)) else: self.UsedPrecedence[precname] = 1 del syms[-2:] # Drop %prec from the rule else: # If no %prec, precedence is determined by the rightmost terminal symbol precname = rightmost_terminal(syms,self.Terminals) prodprec = self.Precedence.get(precname,('right',0)) # See if the rule is already in the rulemap map = "%s -> %s" % (prodname,syms) if map in self.Prodmap: m = self.Prodmap[map] raise GrammarError("%s:%d: Duplicate rule %s. " % (file,line, m) + "Previous definition at %s:%d" % (m.file, m.line)) # From this point on, everything is valid. Create a new Production instance pnumber = len(self.Productions) if not prodname in self.Nonterminals: self.Nonterminals[prodname] = [ ] # Add the production number to Terminals and Nonterminals for t in syms: if t in self.Terminals: self.Terminals[t].append(pnumber) else: if not t in self.Nonterminals: self.Nonterminals[t] = [ ] self.Nonterminals[t].append(pnumber) # Create a production and add it to the list of productions p = Production(pnumber,prodname,syms,prodprec,func,file,line) self.Productions.append(p) self.Prodmap[map] = p # Add to the global productions list try: self.Prodnames[prodname].append(p) except KeyError: self.Prodnames[prodname] = [ p ] return 0 # ----------------------------------------------------------------------------- # set_start() # # Sets the starting symbol and creates the augmented grammar. Production # rule 0 is S' -> start where start is the start symbol. # ----------------------------------------------------------------------------- def set_start(self,start=None): if not start: start = self.Productions[1].name if start not in self.Nonterminals: raise GrammarError("start symbol %s undefined" % start) self.Productions[0] = Production(0,"S'",[start]) self.Nonterminals[start].append(0) self.Start = start # ----------------------------------------------------------------------------- # find_unreachable() # # Find all of the nonterminal symbols that can't be reached from the starting # symbol. Returns a list of nonterminals that can't be reached. # ----------------------------------------------------------------------------- def find_unreachable(self): # Mark all symbols that are reachable from a symbol s def mark_reachable_from(s): if reachable[s]: # We've already reached symbol s. return reachable[s] = 1 for p in self.Prodnames.get(s,[]): for r in p.prod: mark_reachable_from(r) reachable = { } for s in list(self.Terminals) + list(self.Nonterminals): reachable[s] = 0 mark_reachable_from( self.Productions[0].prod[0] ) return [s for s in list(self.Nonterminals) if not reachable[s]] # ----------------------------------------------------------------------------- # infinite_cycles() # # This function looks at the various parsing rules and tries to detect # infinite recursion cycles (grammar rules where there is no possible way # to derive a string of only terminals). # ----------------------------------------------------------------------------- def infinite_cycles(self): terminates = {} # Terminals: for t in self.Terminals: terminates[t] = 1 terminates['$end'] = 1 # Nonterminals: # Initialize to false: for n in self.Nonterminals: terminates[n] = 0 # Then propagate termination until no change: while 1: some_change = 0 for (n,pl) in self.Prodnames.items(): # Nonterminal n terminates iff any of its productions terminates. for p in pl: # Production p terminates iff all of its rhs symbols terminate. for s in p.prod: if not terminates[s]: # The symbol s does not terminate, # so production p does not terminate. p_terminates = 0 break else: # didn't break from the loop, # so every symbol s terminates # so production p terminates. p_terminates = 1 if p_terminates: # symbol n terminates! if not terminates[n]: terminates[n] = 1 some_change = 1 # Don't need to consider any more productions for this n. break if not some_change: break infinite = [] for (s,term) in terminates.items(): if not term: if not s in self.Prodnames and not s in self.Terminals and s != 'error': # s is used-but-not-defined, and we've already warned of that, # so it would be overkill to say that it's also non-terminating. pass else: infinite.append(s) return infinite # ----------------------------------------------------------------------------- # undefined_symbols() # # Find all symbols that were used the grammar, but not defined as tokens or # grammar rules. Returns a list of tuples (sym, prod) where sym in the symbol # and prod is the production where the symbol was used. # ----------------------------------------------------------------------------- def undefined_symbols(self): result = [] for p in self.Productions: if not p: continue for s in p.prod: if not s in self.Prodnames and not s in self.Terminals and s != 'error': result.append((s,p)) return result # ----------------------------------------------------------------------------- # unused_terminals() # # Find all terminals that were defined, but not used by the grammar. Returns # a list of a