Wala shell

Advanced PHP File Manager

Server IP: 127.0.0.1 | PHP Version: 8.3.30 | Domains Hosted: 54

Name	Size	Last Modified	Actions
📁 ..	-	2026-03-21 15:56
📄 __init__.py # PLY package # Author: David Beazley (dave@dabeaz.com) __version__ = '3.11' __all__ = ['lex','yacc']	103 B	2018-02-15 18:44
📁 __pycache__	-	2025-02-27 15:06
📄 cpp.py # ----------------------------------------------------------------------------- # cpp.py # # Author: David Beazley (http://www.dabeaz.com) # Copyright (C) 2007 # All rights reserved # # This module implements an ANSI-C style lexical preprocessor for PLY. # ----------------------------------------------------------------------------- from __future__ import generators import sys # Some Python 3 compatibility shims if sys.version_info.major < 3: STRING_TYPES = (str, unicode) else: STRING_TYPES = str xrange = range # ----------------------------------------------------------------------------- # Default preprocessor lexer definitions. These tokens are enough to get # a basic preprocessor working. Other modules may import these if they want # ----------------------------------------------------------------------------- tokens = ( 'CPP_ID','CPP_INTEGER', 'CPP_FLOAT', 'CPP_STRING', 'CPP_CHAR', 'CPP_WS', 'CPP_COMMENT1', 'CPP_COMMENT2', 'CPP_POUND','CPP_DPOUND' ) literals = "+-/%\|&~^<>=!?()[]{}.,;:\\\'\"" # Whitespace def t_CPP_WS(t): r'\s+' t.lexer.lineno += t.value.count("\n") return t t_CPP_POUND = r'\#' t_CPP_DPOUND = r'\#\#' # Identifier t_CPP_ID = r'[A-Za-z_][\w_]' # Integer literal def CPP_INTEGER(t): r'(((((0x)\|(0X))[0-9a-fA-F]+)\|(\d+))([uU][lL]\|[lL][uU]\|[uU]\|[lL])?)' return t t_CPP_INTEGER = CPP_INTEGER # Floating literal t_CPP_FLOAT = r'((\d+)(\.\d+)(e(\+\|-)?(\d+))? \| (\d+)e(\+\|-)?(\d+))([lL]\|[fF])?' # String literal def t_CPP_STRING(t): r'\"([^\\\n]\|(\\(.\|\n)))?\"' t.lexer.lineno += t.value.count("\n") return t # Character constant 'c' or L'c' def t_CPP_CHAR(t): r'(L)?\'([^\\\n]\|(\\(.\|\n)))?\'' t.lexer.lineno += t.value.count("\n") return t # Comment def t_CPP_COMMENT1(t): r'(/\(.\|\n)?\/)' ncr = t.value.count("\n") t.lexer.lineno += ncr # replace with one space or a number of '\n' t.type = 'CPP_WS'; t.value = '\n' ncr if ncr else ' ' return t # Line comment def t_CPP_COMMENT2(t): r'(//.*?(\n\|$))' # replace with '/n' t.type = 'CPP_WS'; t.value = '\n' return t def t_error(t): t.type = t.value[0] t.value = t.value[0] t.lexer.skip(1) return t import re import copy import time import os.path # ----------------------------------------------------------------------------- # trigraph() # # Given an input string, this function replaces all trigraph sequences. # The following mapping is used: # # ??= # # ??/ \ # ??' ^ # ??( [ # ??) ] # ??! \| # ??< { # ??> } # ??- ~ # ----------------------------------------------------------------------------- _trigraph_pat = re.compile(r'''\?\?[=/\'\(\)\!<>\-]''') _trigraph_rep = { '=':'#', '/':'\\', "'":'^', '(':'[', ')':']', '!':'\|', '<':'{', '>':'}', '-':'~' } def trigraph(input): return _trigraph_pat.sub(lambda g: _trigraph_rep[g.group()[-1]],input) # ------------------------------------------------------------------ # Macro object # # This object holds information about preprocessor macros # # .name - Macro name (string) # .value - Macro value (a list of tokens) # .arglist - List of argument names # .variadic - Boolean indicating whether or not variadic macro # .vararg - Name of the variadic parameter # # When a macro is created, the macro replacement token sequence is # pre-scanned and used to create patch lists that are later used # during macro expansion # ------------------------------------------------------------------ class Macro(object): def __init__(self,name,value,arglist=None,variadic=False): self.name = name self.value = value self.arglist = arglist self.variadic = variadic if variadic: self.vararg = arglist[-1] self.source = None # ------------------------------------------------------------------ # Preprocessor object # # Object representing a preprocessor. Contains macro definitions, # include directories, and other information # ------------------------------------------------------------------ class Preprocessor(object): def __init__(self,lexer=None): if lexer is None: lexer = lex.lexer self.lexer = lexer self.macros = { } self.path = [] self.temp_path = [] # Probe the lexer for selected tokens self.lexprobe() tm = time.localtime() self.define("__DATE__ \"%s\"" % time.strftime("%b %d %Y",tm)) self.define("__TIME__ \"%s\"" % time.strftime("%H:%M:%S",tm)) self.parser = None # ----------------------------------------------------------------------------- # tokenize() # # Utility function. Given a string of text, tokenize into a list of tokens # ----------------------------------------------------------------------------- def tokenize(self,text): tokens = [] self.lexer.input(text) while True: tok = self.lexer.token() if not tok: break tokens.append(tok) return tokens # --------------------------------------------------------------------- # error() # # Report a preprocessor error/warning of some kind # ---------------------------------------------------------------------- def error(self,file,line,msg): print("%s:%d %s" % (file,line,msg)) # ---------------------------------------------------------------------- # lexprobe() # # This method probes the preprocessor lexer object to discover # the token types of symbols that are important to the preprocessor. # If this works right, the preprocessor will simply "work" # with any suitable lexer regardless of how tokens have been named. # ---------------------------------------------------------------------- def lexprobe(self): # Determine the token type for identifiers self.lexer.input("identifier") tok = self.lexer.token() if not tok or tok.value != "identifier": print("Couldn't determine identifier type") else: self.t_ID = tok.type # Determine the token type for integers self.lexer.input("12345") tok = self.lexer.token() if not tok or int(tok.value) != 12345: print("Couldn't determine integer type") else: self.t_INTEGER = tok.type self.t_INTEGER_TYPE = type(tok.value) # Determine the token type for strings enclosed in double quotes self.lexer.input("\"filename\"") tok = self.lexer.token() if not tok or tok.value != "\"filename\"": print("Couldn't determine string type") else: self.t_STRING = tok.type # Determine the token type for whitespace--if any self.lexer.input(" ") tok = self.lexer.token() if not tok or tok.value != " ": self.t_SPACE = None else: self.t_SPACE = tok.type # Determine the token type for newlines self.lexer.input("\n") tok = self.lexer.token() if not tok or tok.value != "\n": self.t_NEWLINE = None print("Couldn't determine token for newlines") else: self.t_NEWLINE = tok.type self.t_WS = (self.t_SPACE, self.t_NEWLINE) # Check for other characters used by the preprocessor chars = [ '<','>','#','##','\\','(',')',',','.'] for c in chars: self.lexer.input(c) tok = self.lexer.token() if not tok or tok.value != c: print("Unable to lex '%s' required for preprocessor" % c) # ---------------------------------------------------------------------- # add_path() # # Adds a search path to the preprocessor. # ---------------------------------------------------------------------- def add_path(self,path): self.path.append(path) # ---------------------------------------------------------------------- # group_lines() # # Given an input string, this function splits it into lines. Trailing whitespace # is removed. Any line ending with \ is grouped with the next line. This # function forms the lowest level of the preprocessor---grouping into text into # a line-by-line format. # ---------------------------------------------------------------------- def group_lines(self,input): lex = self.lexer.clone() lines = [x.rstrip() for x in input.splitlines()] for i in xrange(len(lines)): j = i+1 while lines[i].endswith('\\') and (j < len(lines)): lines[i] = lines[i][:-1]+lines[j] lines[j] = "" j += 1 input = "\n".join(lines) lex.input(input) lex.lineno = 1 current_line = [] while True: tok = lex.token() if not tok: break current_line.append(tok) if tok.type in self.t_WS and '\n' in tok.value: yield current_line current_line = [] if current_line: yield current_line # ---------------------------------------------------------------------- # tokenstrip() # # Remove leading/trailing whitespace tokens from a token list # ---------------------------------------------------------------------- def tokenstrip(self,tokens): i = 0 while i < len(tokens) and tokens[i].type in self.t_WS: i += 1 del tokens[:i] i = len(tokens)-1 while i >= 0 and tokens[i].type in self.t_WS: i -= 1 del tokens[i+1:] return tokens # ---------------------------------------------------------------------- # collect_args() # # Collects comma separated arguments from a list of tokens. The arguments # must be enclosed in parenthesis. Returns a tuple (tokencount,args,positions) # where tokencount is the number of tokens consumed, args is a list of arguments, # and positions is a list of integers containing the starting index of each # argument. Each argument is represented by a list of tokens. # # When collecting arguments, leading and trailing whitespace is removed # from each argument. # # This function properly handles nested parenthesis and commas---these do not # define new arguments. # ---------------------------------------------------------------------- def collect_args(self,tokenlist): args = [] positions = [] current_arg = [] nesting = 1 tokenlen = len(tokenlist) # Search for the opening '('. i = 0 while (i < tokenlen) and (tokenlist[i].type in self.t_WS): i += 1 if (i < tokenlen) and (tokenlist[i].value == '('): positions.append(i+1) else: self.error(self.source,tokenlist[0].lineno,"Missing '(' in macro arguments") return 0, [], [] i += 1 while i < tokenlen: t = tokenlist[i] if t.value == '(': current_arg.append(t) nesting += 1 elif t.value == ')': nesting -= 1 if nesting == 0: if current_arg: args.append(self.tokenstrip(current_arg)) positions.append(i) return i+1,args,positions current_arg.append(t) elif t.value == ',' and nesting == 1: args.append(self.tokenstrip(current_arg)) positions.append(i+1) current_arg = [] else: current_arg.append(t) i += 1 # Missing end argument self.error(self.source,tokenlist[-1].lineno,"Missing ')' in macro arguments") return 0, [],[] # ---------------------------------------------------------------------- # macro_prescan() # # Examine the macro value (token sequence) and identify patch points # This is used to speed up macro expansion later on---we'll know # right away where to apply patches to the value to form the expansion # ---------------------------------------------------------------------- def macro_prescan(self,macro): macro.patch = [] # Standard macro arguments macro.str_patch = [] # String conversion expansion macro.var_comma_patch = [] # Variadic macro comma patch i = 0 while i < len(macro.value): if macro.value[i].type == self.t_ID and macro.value[i].value in macro.arglist: argnum = macro.arglist.index(macro.value[i].value) # Conversion of argument to a string if i > 0 and macro.value[i-1].value == '#': macro.value[i] = copy.copy(macro.value[i]) macro.value[i].type = self.t_STRING del macro.value[i-1] macro.str_patch.append((argnum,i-1)) continue # Concatenation elif (i > 0 and macro.value[i-1].value == '##'): macro.patch.append(('c',argnum,i-1)) del macro.value[i-1] i -= 1 continue elif ((i+1) < len(macro.value) and macro.value[i+1].value == '##'): macro.patch.append(('c',argnum,i)) del macro.value[i + 1] continue # Standard expansion else: macro.patch.append(('e',argnum,i)) elif macro.value[i].value == '##': if macro.variadic and (i > 0) and (macro.value[i-1].value == ',') and \ ((i+1) < len(macro.value)) and (macro.value[i+1].type == self.t_ID) and \ (macro.value[i+1].value == macro.vararg): macro.var_comma_patch.append(i-1) i += 1 macro.patch.sort(key=lambda x: x[2],reverse=True) # ---------------------------------------------------------------------- # macro_expand_args() # # Given a Macro and list of arguments (each a token list), this method # returns an expanded version of a macro. The return value is a token sequence # representing the replacement macro tokens # ---------------------------------------------------------------------- def macro_expand_args(self,macro,args): # Make a copy of the macro token sequence rep = [copy.copy(_x) for _x in macro.value] # Make string expansion patches. These do not alter the length of the replacement sequence str_expansion = {} for argnum, i in macro.str_patch: if argnum not in str_expansion: str_expansion[argnum] = ('"%s"' % "".join([x.value for x in args[argnum]])).replace("\\","\\\\") rep[i] = copy.copy(rep[i]) rep[i].value = str_expansion[argnum] # Make the variadic macro comma patch. If the variadic macro argument is empty, we get rid comma_patch = False if macro.variadic and not args[-1]: for i in macro.var_comma_patch: rep[i] = None comma_patch = True # Make all other patches. The order of these matters. It is assumed that the patch list # has been sorted in reverse order of patch location since replacements will cause the # size of the replacement sequence to expand from the patch point. expanded = { } for ptype, argnum, i in macro.patch: # Concatenation. Argument is left unexpanded if ptype == 'c': rep[i:i+1] = args[argnum] # Normal expansion. Argument is macro expanded first elif ptype == 'e': if argnum not in expanded: expanded[argnum] = self.expand_macros(args[argnum]) rep[i:i+1] = expanded[argnum] # Get rid of removed comma if necessary if comma_patch: rep = [_i for _i in rep if _i] return rep # ---------------------------------------------------------------------- # expand_macros() # # Given a list of tokens, this function performs macro expansion. # The expanded argument is a dictionary that contains macros already # expanded. This is used to prevent infinite recursion. # ---------------------------------------------------------------------- def expand_macros(self,tokens,expanded=None): if expanded is None: expanded = {} i = 0 while i < len(tokens): t = tokens[i] if t.type == self.t_ID: if t.value in self.macros and t.value not in expanded: # Yes, we found a macro match expanded[t.value] = True m = self.macros[t.value] if not m.arglist: # A simple macro ex = self.expand_macros([copy.copy(_x) for _x in m.value],expanded) for e in ex: e.lineno = t.lineno tokens[i:i+1] = ex i += len(ex) else: # A macro with arguments j = i + 1 while j < len(tokens) and tokens[j].type in self.t_WS: j += 1 if j < len(tokens) and tokens[j].value == '(': tokcount,args,positions = self.collect_args(tokens[j:]) if not m.variadic and len(args) != len(m.arglist): self.error(self.source,t.lineno,"Macro %s requires %d arguments" % (t.value,len(m.arglist))) i = j + tokcount elif m.variadic and len(args) < len(m.arglist)-1: if len(m.arglist) > 2: self.error(self.source,t.lineno,"Macro %s must have at least %d arguments" % (t.value, len(m.arglist)-1)) else: self.error(self.source,t.lineno,"Macro %s must have at least %d argument" % (t.value, len(m.arglist)-1)) i = j + tokcount else: if m.variadic: if len(args) == len(m.arglist)-1: args.append([]) else: args[len(m.arglist)-1] = tokens[j+positions[len(m.arglist)-1]:j+tokcount-1] del args[len(m.arglist):] # Get macro replacement text rep = self.macro_expand_args(m,args) rep = self.expand_macros(rep,expanded) for r in rep: r.lineno = t.lineno tokens[i:j+tokcount] = rep i += len(rep) else: # This is not a macro. It is just a word which # equals to name of the macro. Hence, go to the # next token. i += 1 del expanded[t.value] continue elif t.value == '__LINE__': t.type = self.t_INTEGER t.value = self.t_INTEGER_TYPE(t.lineno) i += 1 return tokens # ---------------------------------------------------------------------- # evalexpr() # # Evaluate an expression token sequence for the purposes of evaluating # integral expressions. # ---------------------------------------------------------------------- def evalexpr(self,tokens): # tokens = tokenize(line) # Search for defined macros i = 0 while i < len(tokens): if tokens[i].type == self.t_ID and tokens[i].value == 'defined': j = i + 1 needparen = False result = "0L" while j < len(tokens): if tokens[j].type in self.t_WS: j += 1 continue elif tokens[j].type == self.t_ID: if tokens[j].value in self.macros: result = "1L" else: result = "0L" if not needparen: break elif tokens[j].value == '(': needparen = True elif tokens[j].value == ')': break else: self.error(self.source,tokens[i].lineno,"Malformed defined()") j += 1 tokens[i].type = self.t_INTEGER tokens[i].value = self.t_INTEGER_TYPE(result) del tokens[i+1:j+1] i += 1 tokens = self.expand_macros(tokens) for i,t in enumerate(tokens): if t.type == self.t_ID: tokens[i] = copy.copy(t) tokens[i].type = self.t_INTEGER tokens[i].value = self.t_INTEGER_TYPE("0L") elif t.type == self.t_INTEGER: tokens[i] = copy.copy(t) # Strip off any trailing suffixes tokens[i].value = str(tokens[i].value) while tokens[i].value[-1] not in "0123456789abcdefABCDEF": tokens[i].value = tokens[i].value[:-1] expr = "".join([str(x.value) for x in tokens]) expr = expr.replace("&&"," and ") expr = expr.replace("\|\|"," or ") expr = expr.replace("!"," not ") try: result = eval(expr) except Exception: self.error(self.source,tokens[0].lineno,"Couldn't evaluate expression") result = 0 return result # ---------------------------------------------------------------------- # parsegen() # # Parse an input string/ # ---------------------------------------------------------------------- def parsegen(self,input,source=None): # Replace trigraph sequences t = trigraph(input) lines = self.group_lines(t) if not source: source = "" self.define("__FILE__ \"%s\"" % source) self.source = source chunk = [] enable = True iftrigger = False ifstack = [] for x in lines: for i,tok in enumerate(x): if tok.type not in self.t_WS: break if tok.value == '#': # Preprocessor directive # insert necessary whitespace instead of eaten tokens for tok in x: if tok.type in self.t_WS and '\n' in tok.value: chunk.append(tok) dirtokens = self.tokenstrip(x[i+1:]) if dirtokens: name = dirtokens[0].value args = self.tokenstrip(dirtokens[1:]) else: name = "" args = [] if name == 'define': if enable: for tok in self.expand_macros(chunk): yield tok chunk = [] self.define(args) elif name == 'include': if enable: for tok in self.expand_macros(chunk): yield tok chunk = [] oldfile = self.macros['__FILE__'] for tok in self.include(args): yield tok self.macros['__FILE__'] = oldfile self.source = source elif name == 'undef': if enable: for tok in self.expand_macros(chunk): yield tok chunk = [] self.undef(args) elif name == 'ifdef': ifstack.append((enable,iftrigger)) if enable: if not args[0].value in self.macros: enable = False iftrigger = False else: iftrigger = True elif name == 'ifndef': ifstack.append((enable,iftrigger)) if enable: if args[0].value in self.macros: enable = False iftrigger = False else: iftrigger = True elif name == 'if': ifstack.append((enable,iftrigger)) if enable: result = self.evalexpr(args) if not result: enable = False iftrigger = False else: iftrigger = True elif name == 'elif': if ifstack: if ifstack[-1][0]: # We only pay attention if outer "if" allows this if enable: # If already true, we flip enable False enable = False elif not iftrigger: # If False, but not triggered yet, we'll check expression result = self.evalexpr(args) if result: enable = True iftrigger = True else: self.error(self.source,dirtokens[0].lineno,"Misplaced #elif") elif name == 'else': if ifstack: if ifstack[-1][0]: if enable: enable = False elif not iftrigger: enable = True iftrigger = True else: self.error(self.source,dirtokens[0].lineno,"Misplaced #else") elif name == 'endif': if ifstack: enable,iftrigger = ifstack.pop() else: self.error(self.source,dirtokens[0].lineno,"Misplaced #endif") else: # Unknown preprocessor directive pass else: # Normal text if enable: chunk.extend(x) for tok in self.expand_macros(chunk): yield tok chunk = [] # ---------------------------------------------------------------------- # include() # # Implementation of file-inclusion # ---------------------------------------------------------------------- def include(self,tokens): # Try to extract the filename and then process an include file if not tokens: return if tokens: if tokens[0].value != '<' and tokens[0].type != self.t_STRING: tokens = self.expand_macros(tokens) if tokens[0].value == '<': # Include <...> i = 1 while i < len(tokens): if tokens[i].value == '>': break i += 1 else: print("Malformed #include <...>") return filename = "".join([x.value for x in tokens[1:i]]) path = self.path + [""] + self.temp_path elif tokens[0].type == self.t_STRING: filename = tokens[0].value[1:-1] path = self.temp_path + [""] + self.path else: print("Malformed #include statement") return for p in path: iname = os.path.join(p,filename) try: data = open(iname,"r").read() dname = os.path.dirname(iname) if dname: self.temp_path.insert(0,dname) for tok in self.parsegen(data,filename): yield tok if dname: del self.temp_path[0] break except IOError: pass else: print("Couldn't find '%s'" % filename) # ---------------------------------------------------------------------- # define() # # Define a new macro # ---------------------------------------------------------------------- def define(self,tokens): if isinstance(tokens,STRING_TYPES): tokens = self.tokenize(tokens) linetok = tokens try: name = linetok[0] if len(linetok) > 1: mtype = linetok[1] else: mtype = None if not mtype: m = Macro(name.value,[]) self.macros[name.value] = m elif mtype.type in self.t_WS: # A normal macro m = Macro(name.value,self.tokenstrip(linetok[2:])) self.macros[name.value] = m elif mtype.value == '(': # A macro with arguments tokcount, args, positions = self.collect_args(linetok[1:]) variadic = False for a in args: if variadic: print("No more arguments may follow a variadic argument") break astr = "".join([str(_i.value) for _i in a]) if astr == "...": variadic = True a[0].type = self.t_ID a[0].value = '__VA_ARGS__' variadic = True del a[1:] continue elif astr[-3:] == "..." and a[0].type == self.t_ID: variadic = True del a[1:] # If, for some reason, "." is part of the identifier, strip off the name for the purposes # of macro expansion if a[0].value[-3:] == '...': a[0].value = a[0].value[:-3] continue if len(a) > 1 or a[0].type != self.t_ID: print("Invalid macro argument") break else: mvalue = self.tokenstrip(linetok[1+tokcount:]) i = 0 while i < len(mvalue): if i+1 < len(mvalue): if mvalue[i].type in self.t_WS and mvalue[i+1].value == '##': del mvalue[i] continue elif mvalue[i].value == '##' and mvalue[i+1].type in self.t_WS: del mvalue[i+1] i += 1 m = Macro(name.value,mvalue,[x[0].value for x in args],variadic) self.macro_prescan(m) self.macros[name.value] = m else: print("Bad macro definition") except LookupError: print("Bad macro definition") # ---------------------------------------------------------------------- # undef() # # Undefine a macro # ---------------------------------------------------------------------- def undef(self,tokens): id = tokens[0].value try: del self.macros[id] except LookupError: pass # ---------------------------------------------------------------------- # parse() # # Parse input text. # ---------------------------------------------------------------------- def parse(self,input,source=None,ignore={}): self.ignore = ignore self.parser = self.parsegen(input,source) # ---------------------------------------------------------------------- # token() # # Method to return individual tokens # ---------------------------------------------------------------------- def token(self): try: while True: tok = next(self.parser) if tok.type not in self.ignore: return tok except StopIteration: self.parser = None return None if __name__ == '__main__': import ply.lex as lex lexer = lex.lex() # Run a preprocessor import sys f = open(sys.argv[1]) input = f.read() p = Preprocessor(lexer) p.parse(input,sys.argv[1]) while True: tok = p.token() if not tok: break print(p.source, tok)	32.85 KB	2018-02-15 18:44
📄 ctokens.py # ---------------------------------------------------------------------- # ctokens.py # # Token specifications for symbols in ANSI C and C++. This file is # meant to be used as a library in other tokenizers. # ---------------------------------------------------------------------- # Reserved words tokens = [ # Literals (identifier, integer constant, float constant, string constant, char const) 'ID', 'TYPEID', 'INTEGER', 'FLOAT', 'STRING', 'CHARACTER', # Operators (+,-,,/,%,\|,&,~,^,<<,>>, \|\|, &&, !, <, <=, >, >=, ==, !=) 'PLUS', 'MINUS', 'TIMES', 'DIVIDE', 'MODULO', 'OR', 'AND', 'NOT', 'XOR', 'LSHIFT', 'RSHIFT', 'LOR', 'LAND', 'LNOT', 'LT', 'LE', 'GT', 'GE', 'EQ', 'NE', # Assignment (=, =, /=, %=, +=, -=, <<=, >>=, &=, ^=, \|=) 'EQUALS', 'TIMESEQUAL', 'DIVEQUAL', 'MODEQUAL', 'PLUSEQUAL', 'MINUSEQUAL', 'LSHIFTEQUAL','RSHIFTEQUAL', 'ANDEQUAL', 'XOREQUAL', 'OREQUAL', # Increment/decrement (++,--) 'INCREMENT', 'DECREMENT', # Structure dereference (->) 'ARROW', # Ternary operator (?) 'TERNARY', # Delimeters ( ) [ ] { } , . ; : 'LPAREN', 'RPAREN', 'LBRACKET', 'RBRACKET', 'LBRACE', 'RBRACE', 'COMMA', 'PERIOD', 'SEMI', 'COLON', # Ellipsis (...) 'ELLIPSIS', ] # Operators t_PLUS = r'\+' t_MINUS = r'-' t_TIMES = r'\' t_DIVIDE = r'/' t_MODULO = r'%' t_OR = r'\\|' t_AND = r'&' t_NOT = r'~' t_XOR = r'\^' t_LSHIFT = r'<<' t_RSHIFT = r'>>' t_LOR = r'\\|\\|' t_LAND = r'&&' t_LNOT = r'!' t_LT = r'<' t_GT = r'>' t_LE = r'<=' t_GE = r'>=' t_EQ = r'==' t_NE = r'!=' # Assignment operators t_EQUALS = r'=' t_TIMESEQUAL = r'\=' t_DIVEQUAL = r'/=' t_MODEQUAL = r'%=' t_PLUSEQUAL = r'\+=' t_MINUSEQUAL = r'-=' t_LSHIFTEQUAL = r'<<=' t_RSHIFTEQUAL = r'>>=' t_ANDEQUAL = r'&=' t_OREQUAL = r'\\|=' t_XOREQUAL = r'\^=' # Increment/decrement t_INCREMENT = r'\+\+' t_DECREMENT = r'--' # -> t_ARROW = r'->' # ? t_TERNARY = r'\?' # Delimeters t_LPAREN = r'\(' t_RPAREN = r'\)' t_LBRACKET = r'\[' t_RBRACKET = r'\]' t_LBRACE = r'\{' t_RBRACE = r'\}' t_COMMA = r',' t_PERIOD = r'\.' t_SEMI = r';' t_COLON = r':' t_ELLIPSIS = r'\.\.\.' # Identifiers t_ID = r'[A-Za-z_][A-Za-z0-9_]' # Integer literal t_INTEGER = r'\d+([uU]\|[lL]\|[uU][lL]\|[lL][uU])?' # Floating literal t_FLOAT = r'((\d+)(\.\d+)(e(\+\|-)?(\d+))? \| (\d+)e(\+\|-)?(\d+))([lL]\|[fF])?' # String literal t_STRING = r'\"([^\\\n]\|(\\.))?\"' # Character constant 'c' or L'c' t_CHARACTER = r'(L)?\'([^\\\n]\|(\\.))?\'' # Comment (C-Style) def t_COMMENT(t): r'/\(.\|\n)?\/' t.lexer.lineno += t.value.count('\n') return t # Comment (C++-Style) def t_CPPCOMMENT(t): r'//.*\n' t.lexer.lineno += 1 return t	3.08 KB	2018-02-15 18:44
📄 lex.py # ----------------------------------------------------------------------------- # ply: lex.py # # Copyright (C) 2001-2018 # 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. # ----------------------------------------------------------------------------- __version__ = '3.11' __tabversion__ = '3.10' import re import sys import types import copy import os import inspect # This tuple contains known string types try: # Python 2.6 StringTypes = (types.StringType, types.UnicodeType) except AttributeError: # Python 3.0 StringTypes = (str, bytes) # This regular expression is used to match valid token names _is_identifier = re.compile(r'^[a-zA-Z0-9_]+$') # Exception thrown when invalid token encountered and no default error # handler is defined. class LexError(Exception): def __init__(self, message, s): self.args = (message,) self.text = s # Token class. This class is used to represent the tokens produced. class LexToken(object): def __str__(self): return 'LexToken(%s,%r,%d,%d)' % (self.type, self.value, self.lineno, self.lexpos) def __repr__(self): return str(self) # This object is a stand-in for a logging object created by the # logging module. class PlyLogger(object): def __init__(self, f): self.f = f def critical(self, msg, args, kwargs): self.f.write((msg % args) + '\n') 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') info = critical debug = critical # 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 # ----------------------------------------------------------------------------- # === Lexing Engine === # # The following Lexer class implements the lexer runtime. There are only # a few public methods and attributes: # # input() - Store a new string in the lexer # token() - Get the next token # clone() - Clone the lexer # # lineno - Current line number # lexpos - Current position in the input string # ----------------------------------------------------------------------------- class Lexer: def __init__(self): self.lexre = None # Master regular expression. This is a list of # tuples (re, findex) where re is a compiled # regular expression and findex is a list # mapping regex group numbers to rules self.lexretext = None # Current regular expression strings self.lexstatere = {} # Dictionary mapping lexer states to master regexs self.lexstateretext = {} # Dictionary mapping lexer states to regex strings self.lexstaterenames = {} # Dictionary mapping lexer states to symbol names self.lexstate = 'INITIAL' # Current lexer state self.lexstatestack = [] # Stack of lexer states self.lexstateinfo = None # State information self.lexstateignore = {} # Dictionary of ignored characters for each state self.lexstateerrorf = {} # Dictionary of error functions for each state self.lexstateeoff = {} # Dictionary of eof functions for each state self.lexreflags = 0 # Optional re compile flags self.lexdata = None # Actual input data (as a string) self.lexpos = 0 # Current position in input text self.lexlen = 0 # Length of the input text self.lexerrorf = None # Error rule (if any) self.lexeoff = None # EOF rule (if any) self.lextokens = None # List of valid tokens self.lexignore = '' # Ignored characters self.lexliterals = '' # Literal characters that can be passed through self.lexmodule = None # Module self.lineno = 1 # Current line number self.lexoptimize = False # Optimized mode def clone(self, object=None): c = copy.copy(self) # If the object parameter has been supplied, it means we are attaching the # lexer to a new object. In this case, we have to rebind all methods in # the lexstatere and lexstateerrorf tables. if object: newtab = {} for key, ritem in self.lexstatere.items(): newre = [] for cre, findex in ritem: newfindex = [] for f in findex: if not f or not f[0]: newfindex.append(f) continue newfindex.append((getattr(object, f[0].__name__), f[1])) newre.append((cre, newfindex)) newtab[key] = newre c.lexstatere = newtab c.lexstateerrorf = {} for key, ef in self.lexstateerrorf.items(): c.lexstateerrorf[key] = getattr(object, ef.__name__) c.lexmodule = object return c # ------------------------------------------------------------ # writetab() - Write lexer information to a table file # ------------------------------------------------------------ def writetab(self, lextab, outputdir=''): if isinstance(lextab, types.ModuleType): raise IOError("Won't overwrite existing lextab module") basetabmodule = lextab.split('.')[-1] filename = os.path.join(outputdir, basetabmodule) + '.py' with open(filename, 'w') as tf: tf.write('# %s.py. This file automatically created by PLY (version %s). Don\'t edit!\n' % (basetabmodule, __version__)) tf.write('_tabversion = %s\n' % repr(__tabversion__)) tf.write('_lextokens = set(%s)\n' % repr(tuple(sorted(self.lextokens)))) tf.write('_lexreflags = %s\n' % repr(int(self.lexreflags))) tf.write('_lexliterals = %s\n' % repr(self.lexliterals)) tf.write('_lexstateinfo = %s\n' % repr(self.lexstateinfo)) # Rewrite the lexstatere table, replacing function objects with function names tabre = {} for statename, lre in self.lexstatere.items(): titem = [] for (pat, func), retext, renames in zip(lre, self.lexstateretext[statename], self.lexstaterenames[statename]): titem.append((retext, _funcs_to_names(func, renames))) tabre[statename] = titem tf.write('_lexstatere = %s\n' % repr(tabre)) tf.write('_lexstateignore = %s\n' % repr(self.lexstateignore)) taberr = {} for statename, ef in self.lexstateerrorf.items(): taberr[statename] = ef.__name__ if ef else None tf.write('_lexstateerrorf = %s\n' % repr(taberr)) tabeof = {} for statename, ef in self.lexstateeoff.items(): tabeof[statename] = ef.__name__ if ef else None tf.write('_lexstateeoff = %s\n' % repr(tabeof)) # ------------------------------------------------------------ # readtab() - Read lexer information from a tab file # ------------------------------------------------------------ def readtab(self, tabfile, fdict): if isinstance(tabfile, types.ModuleType): lextab = tabfile else: exec('import %s' % tabfile) lextab = sys.modules[tabfile] if getattr(lextab, '_tabversion', '0.0') != __tabversion__: raise ImportError('Inconsistent PLY version') self.lextokens = lextab._lextokens self.lexreflags = lextab._lexreflags self.lexliterals = lextab._lexliterals self.lextokens_all = self.lextokens \| set(self.lexliterals) self.lexstateinfo = lextab._lexstateinfo self.lexstateignore = lextab._lexstateignore self.lexstatere = {} self.lexstateretext = {} for statename, lre in lextab._lexstatere.items(): titem = [] txtitem = [] for pat, func_name in lre: titem.append((re.compile(pat, lextab._lexreflags), _names_to_funcs(func_name, fdict))) self.lexstatere[statename] = titem self.lexstateretext[statename] = txtitem self.lexstateerrorf = {} for statename, ef in lextab._lexstateerrorf.items(): self.lexstateerrorf[statename] = fdict[ef] self.lexstateeoff = {} for statename, ef in lextab._lexstateeoff.items(): self.lexstateeoff[statename] = fdict[ef] self.begin('INITIAL') # ------------------------------------------------------------ # input() - Push a new string into the lexer # ------------------------------------------------------------ def input(self, s): # Pull off the first character to see if s looks like a string c = s[:1] if not isinstance(c, StringTypes): raise ValueError('Expected a string') self.lexdata = s self.lexpos = 0 self.lexlen = len(s) # ------------------------------------------------------------ # begin() - Changes the lexing state # ------------------------------------------------------------ def begin(self, state): if state not in self.lexstatere: raise ValueError('Undefined state') self.lexre = self.lexstatere[state] self.lexretext = self.lexstateretext[state] self.lexignore = self.lexstateignore.get(state, '') self.lexerrorf = self.lexstateerrorf.get(state, None) self.lexeoff = self.lexstateeoff.get(state, None) self.lexstate = state # ------------------------------------------------------------ # push_state() - Changes the lexing state and saves old on stack # ------------------------------------------------------------ def push_state(self, state): self.lexstatestack.append(self.lexstate) self.begin(state) # ------------------------------------------------------------ # pop_state() - Restores the previous state # ------------------------------------------------------------ def pop_state(self): self.begin(self.lexstatestack.pop()) # ------------------------------------------------------------ # current_state() - Returns the current lexing state # ------------------------------------------------------------ def current_state(self): return self.lexstate # ------------------------------------------------------------ # skip() - Skip ahead n characters # ------------------------------------------------------------ def skip(self, n): self.lexpos += n # ------------------------------------------------------------ # opttoken() - Return the next token from the Lexer # # Note: This function has been carefully implemented to be as fast # as possible. Don't make changes unless you really know what # you are doing # ------------------------------------------------------------ def token(self): # Make local copies of frequently referenced attributes lexpos = self.lexpos lexlen = self.lexlen lexignore = self.lexignore lexdata = self.lexdata while lexpos < lexlen: # This code provides some short-circuit code for whitespace, tabs, and other ignored characters if lexdata[lexpos] in lexignore: lexpos += 1 continue # Look for a regular expression match for lexre, lexindexfunc in self.lexre: m = lexre.match(lexdata, lexpos) if not m: continue # Create a token for return tok = LexToken() tok.value = m.group() tok.lineno = self.lineno tok.lexpos = lexpos i = m.lastindex func, tok.type = lexindexfunc[i] if not func: # If no token type was set, it's an ignored token if tok.type: self.lexpos = m.end() return tok else: lexpos = m.end() break lexpos = m.end() # If token is processed by a function, call it tok.lexer = self # Set additional attributes useful in token rules self.lexmatch = m self.lexpos = lexpos newtok = func(tok) # Every function must return a token, if nothing, we just move to next token if not newtok: lexpos = self.lexpos # This is here in case user has updated lexpos. lexignore = self.lexignore # This is here in case there was a state change break # Verify type of the token. If not in the token map, raise an error if not self.lexoptimize: if newtok.type not in self.lextokens_all: raise LexError("%s:%d: Rule '%s' returned an unknown token type '%s'" % ( func.__code__.co_filename, func.__code__.co_firstlineno, func.__name__, newtok.type), lexdata[lexpos:]) return newtok else: # No match, see if in literals if lexdata[lexpos] in self.lexliterals: tok = LexToken() tok.value = lexdata[lexpos] tok.lineno = self.lineno tok.type = tok.value tok.lexpos = lexpos self.lexpos = lexpos + 1 return tok # No match. Call t_error() if defined. if self.lexerrorf: tok = LexToken() tok.value = self.lexdata[lexpos:] tok.lineno = self.lineno tok.type = 'error' tok.lexer = self tok.lexpos = lexpos self.lexpos = lexpos newtok = self.lexerrorf(tok) if lexpos == self.lexpos: # Error method didn't change text position at all. This is an error. raise LexError("Scanning error. Illegal character '%s'" % (lexdata[lexpos]), lexdata[lexpos:]) lexpos = self.lexpos if not newtok: continue return newtok self.lexpos = lexpos raise LexError("Illegal character '%s' at index %d" % (lexdata[lexpos], lexpos), lexdata[lexpos:]) if self.lexeoff: tok = LexToken() tok.type = 'eof' tok.value = '' tok.lineno = self.lineno tok.lexpos = lexpos tok.lexer = self self.lexpos = lexpos newtok = self.lexeoff(tok) return newtok self.lexpos = lexpos + 1 if self.lexdata is None: raise RuntimeError('No input string given with input()') return None # Iterator interface def __iter__(self): return self def next(self): t = self.token() if t is None: raise StopIteration return t __next__ = next # ----------------------------------------------------------------------------- # ==== Lex Builder === # # The functions and classes below are used to collect lexing information # and build a Lexer object from it. # ----------------------------------------------------------------------------- # ----------------------------------------------------------------------------- # _get_regex(func) # # Returns the regular expression assigned to a function either as a doc string # or as a .regex attribute attached by the @TOKEN decorator. # ----------------------------------------------------------------------------- def _get_regex(func): return getattr(func, 'regex', func.__doc__) # ----------------------------------------------------------------------------- # get_caller_module_dict() # # This function returns a dictionary containing all of the symbols defined within # a caller further down the call stack. This is used to get the environment # associated with the yacc() call if none was provided. # ----------------------------------------------------------------------------- def get_caller_module_dict(levels): f = sys._getframe(levels) ldict = f.f_globals.copy() if f.f_globals != f.f_locals: ldict.update(f.f_locals) return ldict # ----------------------------------------------------------------------------- # _funcs_to_names() # # Given a list of regular expression functions, this converts it to a list # suitable for output to a table file # ----------------------------------------------------------------------------- def _funcs_to_names(funclist, namelist): result = [] for f, name in zip(funclist, namelist): if f and f[0]: result.append((name, f[1])) else: result.append(f) return result # ----------------------------------------------------------------------------- # _names_to_funcs() # # Given a list of regular expression function names, this converts it back to # functions. # ----------------------------------------------------------------------------- def _names_to_funcs(namelist, fdict): result = [] for n in namelist: if n and n[0]: result.append((fdict[n[0]], n[1])) else: result.append(n) return result # ----------------------------------------------------------------------------- # _form_master_re() # # This function takes a list of all of the regex components and attempts to # form the master regular expression. Given limitations in the Python re # module, it may be necessary to break the master regex into separate expressions. # ----------------------------------------------------------------------------- def _form_master_re(relist, reflags, ldict, toknames): if not relist: return [] regex = '\|'.join(relist) try: lexre = re.compile(regex, reflags) # Build the index to function map for the matching engine lexindexfunc = [None] (max(lexre.groupindex.values()) + 1) lexindexnames = lexindexfunc[:] for f, i in lexre.groupindex.items(): handle = ldict.get(f, None) if type(handle) in (types.FunctionType, types.MethodType): lexindexfunc[i] = (handle, toknames[f]) lexindexnames[i] = f elif handle is not None: lexindexnames[i] = f if f.find('ignore_') > 0: lexindexfunc[i] = (None, None) else: lexindexfunc[i] = (None, toknames[f]) return [(lexre, lexindexfunc)], [regex], [lexindexnames] except Exception: m = int(len(relist)/2) if m == 0: m = 1 llist, lre, lnames = _form_master_re(relist[:m], reflags, ldict, toknames) rlist, rre, rnames = _form_master_re(relist[m:], reflags, ldict, toknames) return (llist+rlist), (lre+rre), (lnames+rnames) # ----------------------------------------------------------------------------- # def _statetoken(s,names) # # Given a declaration name s of the form "t_" and a dictionary whose keys are # state names, this function returns a tuple (states,tokenname) where states # is a tuple of state names and tokenname is the name of the token. For example, # calling this with s = "t_foo_bar_SPAM" might return (('foo','bar'),'SPAM') # ----------------------------------------------------------------------------- def _statetoken(s, names): parts = s.split('_') for i, part in enumerate(parts[1:], 1): if part not in names and part != 'ANY': break if i > 1: states = tuple(parts[1:i]) else: states = ('INITIAL',) if 'ANY' in states: states = tuple(names) tokenname = '_'.join(parts[i:]) return (states, tokenname) # ----------------------------------------------------------------------------- # LexerReflect() # # This class represents information needed to build a lexer as extracted from a # user's input file. # ----------------------------------------------------------------------------- class LexerReflect(object): def __init__(self, ldict, log=None, reflags=0): self.ldict = ldict self.error_func = None self.tokens = [] self.reflags = reflags self.stateinfo = {'INITIAL': 'inclusive'} self.modules = set() self.error = False self.log = PlyLogger(sys.stderr) if log is None else log # Get all of the basic information def get_all(self): self.get_tokens() self.get_literals() self.get_states() self.get_rules() # Validate all of the information def validate_all(self): self.validate_tokens() self.validate_literals() self.validate_rules() return self.error # Get the tokens map def get_tokens(self): tokens = self.ldict.get('tokens', None) if not tokens: self.log.error('No token list is defined') self.error = True return if not isinstance(tokens, (list, tuple)): self.log.error('tokens must be a list or tuple') self.error = True return if not tokens: self.log.error('tokens is empty') self.error = True return self.tokens = tokens # Validate the tokens def validate_tokens(self): terminals = {} for n in self.tokens: if not _is_identifier.match(n): self.log.error("Bad token name '%s'", n) self.error = True if n in terminals: self.log.warning("Token '%s' multiply defined", n) terminals[n] = 1 # Get the literals specifier def get_literals(self): self.literals = self.ldict.get('literals', '') if not self.literals: self.literals = '' # Validate literals def validate_literals(self): try: for c in self.literals: if not isinstance(c, StringTypes) or len(c) > 1: self.log.error('Invalid literal %s. Must be a single character', repr(c)) self.error = True except TypeError: self.log.error('Invalid literals specification. literals must be a sequence of characters') self.error = True def get_states(self): self.states = self.ldict.get('states', None) # Build statemap if self.states: if not isinstance(self.states, (tuple, list)): self.log.error('states must be defined as a tuple or list') self.error = True else: for s in self.states: if not isinstance(s, tuple) or len(s) != 2: self.log.error("Invalid state specifier %s. Must be a tuple (statename,'exclusive\|inclusive')", repr(s)) self.error = True continue name, statetype = s if not isinstance(name, StringTypes): self.log.error('State name %s must be a string', repr(name)) self.error = True continue if not (statetype == 'inclusive' or statetype == 'exclusive'): self.log.error("State type for state %s must be 'inclusive' or 'exclusive'", name) self.error = True continue if name in self.stateinfo: self.log.error("State '%s' already defined", name) self.error = True continue self.stateinfo[name] = statetype # Get all of the symbols with a t_ prefix and sort them into various # categories (functions, strings, error functions, and ignore characters) def get_rules(self): tsymbols = [f for f in self.ldict if f[:2] == 't_'] # Now build up a list of functions and a list of strings self.toknames = {} # Mapping of symbols to token names self.funcsym = {} # Symbols defined as functions self.strsym = {} # Symbols defined as strings self.ignore = {} # Ignore strings by state self.errorf = {} # Error functions by state self.eoff = {} # EOF functions by state for s in self.stateinfo: self.funcsym[s] = [] self.strsym[s] = [] if len(tsymbols) == 0: self.log.error('No rules of the form t_rulename are defined') self.error = True return for f in tsymbols: t = self.ldict[f] states, tokname = _statetoken(f, self.stateinfo) self.toknames[f] = tokname if hasattr(t, '__call__'): if tokname == 'error': for s in states: self.errorf[s] = t elif tokname == 'eof': for s in states: self.eoff[s] = t elif tokname == 'ignore': line = t.__code__.co_firstlineno file = t.__code__.co_filename self.log.error("%s:%d: Rule '%s' must be defined as a string", file, line, t.__name__) self.error = True else: for s in states: self.funcsym[s].append((f, t)) elif isinstance(t, StringTypes): if tokname == 'ignore': for s in states: self.ignore[s] = t if '\\' in t: self.log.warning("%s contains a literal backslash '\\'", f) elif tokname == 'error': self.log.error("Rule '%s' must be defined as a function", f) self.error = True else: for s in states: self.strsym[s].append((f, t)) else: self.log.error('%s not defined as a function or string', f) self.error = True # Sort the functions by line number for f in self.funcsym.values(): f.sort(key=lambda x: x[1].__code__.co_firstlineno) # Sort the strings by regular expression length for s in self.strsym.values(): s.sort(key=lambda x: len(x[1]), reverse=True) # Validate all of the t_rules collected def validate_rules(self): for state in self.stateinfo: # Validate all rules defined by functions for fname, f in self.funcsym[state]: line = f.__code__.co_firstlineno file = f.__code__.co_filename module = inspect.getmodule(f) self.modules.add(module) tokname = self.toknames[fname] if isinstance(f, types.MethodType): reqargs = 2 else: reqargs = 1 nargs = f.__code__.co_argcount if nargs > reqargs: self.log.error("%s:%d: Rule '%s' has too many arguments", file, line, f.__name__) self.error = True continue if nargs < reqargs: self.log.error("%s:%d: Rule '%s' requires an argument", file, line, f.__name__) self.error = True continue if not _get_regex(f): self.log.error("%s:%d: No regular expression defined for rule '%s'", file, line, f.__name__) self.error = True continue try: c = re.compile('(?P<%s>%s)' % (fname, _get_regex(f)), self.reflags) if c.match(''): self.log.error("%s:%d: Regular expression for rule '%s' matches empty string", file, line, f.__name__) self.error = True except re.error as e: self.log.error("%s:%d: Invalid regular expression for rule '%s'. %s", file, line, f.__name__, e) if '#' in _get_regex(f): self.log.error("%s:%d. Make sure '#' in rule '%s' is escaped with '\\#'", file, line, f.__name__) self.error = True # Validate all rules defined by strings for name, r in self.strsym[state]: tokname = self.toknames[name] if tokname == 'error': self.log.error("Rule '%s' must be defined as a function", name) self.error = True continue if tokname not in self.tokens and tokname.find('ignore_') < 0: self.log.error("Rule '%s' defined for an unspecified token %s", name, tokname) self.error = True continue try: c = re.compile('(?P<%s>%s)' % (name, r), self.reflags) if (c.match('')): self.log.error("Regular expression for rule '%s' matches empty string", name) self.error = True except re.error as e: self.log.error("Invalid regular expression for rule '%s'. %s", name, e) if '#' in r: self.log.error("Make sure '#' in rule '%s' is escaped with '\\#'", name) self.error = True if not self.funcsym[state] and not self.strsym[state]: self.log.error("No rules defined for state '%s'", state) self.error = True # Validate the error function efunc = self.errorf.get(state, None) if efunc: f = efunc line = f.__code__.co_firstlineno file = f.__code__.co_filename module = inspect.getmodule(f) self.modules.add(module) if isinstance(f, types.MethodType): reqargs = 2 else: reqargs = 1 nargs = f.__code__.co_argcount if nargs > reqargs: self.log.error("%s:%d: Rule '%s' has too many arguments", file, line, f.__name__) self.error = True if nargs < reqargs: self.log.error("%s:%d: Rule '%s' requires an argument", file, line, f.__name__) self.error = True for module in self.modules: self.validate_module(module) # ----------------------------------------------------------------------------- # validate_module() # # This checks to see if there are duplicated t_rulename() functions or strings # in the parser input file. This is done using a simple regular expression # match on each line in the source code of the given module. # ----------------------------------------------------------------------------- def validate_module(self, module): try: lines, linen = inspect.getsourcelines(module) except IOError: return fre = re.compile(r'\sdef\s+(t_[a-zA-Z_0-9])\(') sre = re.compile(r'\s(t_[a-zA-Z_0-9])\s*=') counthash = {} linen += 1 for line in lines: m = fre.match(line) if not m: m = sre.match(line) if m: name = m.group(1) prev = counthash.get(name) if not prev: counthash[name] = linen else: filename = inspect.getsourcefile(module) self.log.error('%s:%d: Rule %s redefined. Previously defined on line %d', filename, linen, name, prev) self.error = True linen += 1 # ----------------------------------------------------------------------------- # lex(module) # # Build all of the regular expression rules from definitions in the supplied module # ----------------------------------------------------------------------------- def lex(module=None, object=None, debug=False, optimize=False, lextab='lextab', reflags=int(re.VERBOSE), nowarn=False, outputdir=None, debuglog=None, errorlog=None): if lextab is None: lextab = 'lextab' global lexer ldict = None stateinfo = {'INITIAL': 'inclusive'} lexobj = Lexer() lexobj.lexoptimize = optimize global token, input if errorlog is None: errorlog = PlyLogger(sys.stderr) if debug: if debuglog is None: debuglog = PlyLogger(sys.stderr) # Get the module dictionary used for the lexer if object: module = object # Get the module dictionary used for the parser if module: _items = [(k, getattr(module, k)) for k in dir(module)] ldict = dict(_items) # If no __file__ attribute is available, try to obtain it from the __module__ instead if '__file__' not in ldict: ldict['__file__'] = sys.modules[ldict['__module__']].__file__ else: ldict = get_caller_module_dict(2) # Determine if the module is package of a package or not. # If so, fix the tabmodule setting so that tables load correctly pkg = ldict.get('__package__') if pkg and isinstance(lextab, str): if '.' not in lextab: lextab = pkg + '.' + lextab # Collect parser information from the dictionary linfo = LexerReflect(ldict, log=errorlog, reflags=reflags) linfo.get_all() if not optimize: if linfo.validate_all(): raise SyntaxError("Can't build lexer") if optimize and lextab: try: lexobj.readtab(lextab, ldict) token = lexobj.token input = lexobj.input lexer = lexobj return lexobj except ImportError: pass # Dump some basic debugging information if debug: debuglog.info('lex: tokens = %r', linfo.tokens) debuglog.info('lex: literals = %r', linfo.literals) debuglog.info('lex: states = %r', linfo.stateinfo) # Build a dictionary of valid token names lexobj.lextokens = set() for n in linfo.tokens: lexobj.lextokens.add(n) # Get literals specification if isinstance(linfo.literals, (list, tuple)): lexobj.lexliterals = type(linfo.literals[0])().join(linfo.literals) else: lexobj.lexliterals = linfo.literals lexobj.lextokens_all = lexobj.lextokens \| set(lexobj.lexliterals) # Get the stateinfo dictionary stateinfo = linfo.stateinfo regexs = {} # Build the master regular expressions for state in stateinfo: regex_list = [] # Add rules defined by functions first for fname, f in linfo.funcsym[state]: regex_list.append('(?P<%s>%s)' % (fname, _get_regex(f))) if debug: debuglog.info("lex: Adding rule %s -> '%s' (state '%s')", fname, _get_regex(f), state) # Now add all of the simple rules for name, r in linfo.strsym[state]: regex_list.append('(?P<%s>%s)' % (name, r)) if debug: debuglog.info("lex: Adding rule %s -> '%s' (state '%s')", name, r, state) regexs[state] = regex_list # Build the master regular expressions if debug: debuglog.info('lex: ==== MASTER REGEXS FOLLOW ====') for state in regexs: lexre, re_text, re_names = _form_master_re(regexs[state], reflags, ldict, linfo.toknames) lexobj.lexstatere[state] = lexre lexobj.lexstateretext[state] = re_text lexobj.lexstaterenames[state] = re_names if debug: for i, text in enumerate(re_text): debuglog.info("lex: state '%s' : regex[%d] = '%s'", state, i, text) # For inclusive states, we need to add the regular expressions from the INITIAL state for state, stype in stateinfo.items(): if state != 'INITIAL' and stype == 'inclusive': lexobj.lexstatere[state].extend(lexobj.lexstatere['INITIAL']) lexobj.lexstateretext[state].extend(lexobj.lexstateretext['INITIAL']) lexobj.lexstaterenames[state].extend(lexobj.lexstaterenames['INITIAL']) lexobj.lexstateinfo = stateinfo lexobj.lexre = lexobj.lexstatere['INITIAL'] lexobj.lexretext = lexobj.lexstateretext['INITIAL'] lexobj.lexreflags = reflags # Set up ignore variables lexobj.lexstateignore = linfo.ignore lexobj.lexignore = lexobj.lexstateignore.get('INITIAL', '') # Set up error functions lexobj.lexstateerrorf = linfo.errorf lexobj.lexerrorf = linfo.errorf.get('INITIAL', None) if not lexobj.lexerrorf: errorlog.warning('No t_error rule is defined') # Set up eof functions lexobj.lexstateeoff = linfo.eoff lexobj.lexeoff = linfo.eoff.get('INITIAL', None) # Check state information for ignore and error rules for s, stype in stateinfo.items(): if stype == 'exclusive': if s not in linfo.errorf: errorlog.warning("No error rule is defined for exclusive state '%s'", s) if s not in linfo.ignore and lexobj.lexignore: errorlog.warning("No ignore rule is defined for exclusive state '%s'", s) elif stype == 'inclusive': if s not in linfo.errorf: linfo.errorf[s] = linfo.errorf.get('INITIAL', None) if s not in linfo.ignore: linfo.ignore[s] = linfo.ignore.get('INITIAL', '') # Create global versions of the token() and input() functions token = lexobj.token input = lexobj.input lexer = lexobj # If in optimize mode, we write the lextab if lextab and optimize: if outputdir is None: # If no output directory is set, the location of the output files # is determined according to the following rules: # - If lextab specifies a package, files go into that package directory # - Otherwise, files go in the same directory as the specifying module if isinstance(lextab, types.ModuleType): srcfile = lextab.__file__ else: if '.' not in lextab: srcfile = ldict['__file__'] else: parts = lextab.split('.') pkgname = '.'.join(parts[:-1]) exec('import %s' % pkgname) srcfile = getattr(sys.modules[pkgname], '__file__', '') outputdir = os.path.dirname(srcfile) try: lexobj.writetab(lextab, outputdir) if lextab in sys.modules: del sys.modules[lextab] except IOError as e: errorlog.warning("Couldn't write lextab module %r. %s" % (lextab, e)) return lexobj # ----------------------------------------------------------------------------- # runmain() # # This runs the lexer as a main program # ----------------------------------------------------------------------------- def runmain(lexer=None, data=None): if not data: try: filename = sys.argv[1] f = open(filename) data = f.read() f.close() except IndexError: sys.stdout.write('Reading from standard input (type EOF to end):\n') data = sys.stdin.read() if lexer: _input = lexer.input else: _input = input _input(data) if lexer: _token = lexer.token else: _token = token while True: tok = _token() if not tok: break sys.stdout.write('(%s,%r,%d,%d)\n' % (tok.type, tok.value, tok.lineno, tok.lexpos)) # ----------------------------------------------------------------------------- # @TOKEN(regex) # # This decorator function can be used to set the regex expression on a function # when its docstring might need to be set in an alternative way # ----------------------------------------------------------------------------- def TOKEN(r): def set_regex(f): if hasattr(r, '__call__'): f.regex = _get_regex(r) else: f.regex = r return f return set_regex # Alternative spelling of the TOKEN decorator Token = TOKEN	41.9 KB	2018-02-15 18:44
📄 yacc.py # ----------------------------------------------------------------------------- # ply: yacc.py # # Copyright (C) 2001-2018 # 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 grammar 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! # ---------------------------------------------------------------------------- import re import types import sys import os.path import inspect import warnings __version__ = '3.11' __tabversion__ = '3.10' #----------------------------------------------------------------------------- # === User configurable parameters === # # Change these to modify the default behavior of yacc (if you wish) #----------------------------------------------------------------------------- yaccdebug = True # 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 = False # 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 # String type-checking compatibility if sys.version_info[0] < 3: string_types = basestring else: string_types = str MAXINT = sys.maxsize # 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)) # Panic mode error recovery support. This feature is being reworked--much of the # code here is to offer a deprecation/backwards compatible transition _errok = None _token = None _restart = None _warnmsg = '''PLY: Don't use global functions errok(), token(), and restart() in p_error(). Instead, invoke the methods on the associated parser instance: def p_error(p): ... # Use parser.errok(), parser.token(), parser.restart() ... parser = yacc.yacc() ''' def errok(): warnings.warn(_warnmsg) return _errok() def restart(): warnings.warn(_warnmsg) return _restart() def token(): warnings.warn(_warnmsg) return _token() # Utility function to call the p_error() function with some deprecation hacks def call_errorfunc(errorfunc, token, parser): global _errok, _token, _restart _errok = parser.errok _token = parser.token _restart = parser.restart r = errorfunc(token) try: del _errok, _token, _restart except NameError: pass return 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 isinstance(n, slice): return [s.value for s in self.slice[n]] elif 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 set_lexpos(self, n, lexpos): self.slice[n].lexpos = lexpos 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 self.set_defaulted_states() self.errorok = True def errok(self): self.errorok = True def restart(self): del self.statestack[:] del self.symstack[:] sym = YaccSymbol() sym.type = '$end' self.symstack.append(sym) self.statestack.append(0) # Defaulted state support. # This method identifies parser states where there is only one possible reduction action. # For such states, the parser can make a choose to make a rule reduction without consuming # the next look-ahead token. This delayed invocation of the tokenizer can be useful in # certain kinds of advanced parsing situations where the lexer and parser interact with # each other or change states (i.e., manipulation of scope, lexer states, etc.). # # See: http://www.gnu.org/software/bison/manual/html_node/Default-Reductions.html#Default-Reductions def set_defaulted_states(self): self.defaulted_states = {} for state, actions in self.action.items(): rules = list(actions.values()) if len(rules) == 1 and rules[0] < 0: self.defaulted_states[state] = rules[0] def disable_defaulted_states(self): self.defaulted_states = {} def parse(self, input=None, lexer=None, debug=False, tracking=False, 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. Optimized versions of this function # are automatically created by the ply/ygen.py script. This script cuts out # sections enclosed in markers such as this: # # #--! DEBUG # statements # #--! DEBUG # # !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! def parsedebug(self, input=None, lexer=None, debug=False, tracking=False, tokenfunc=None): #--! parsedebug-start 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.) defaulted_states = self.defaulted_states # Local reference to defaulted states 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: from . import 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 the parser() token method (sometimes used in error recovery) self.token = get_token # 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 = '$end' symstack.append(sym) state = 0 while True: # Get the next symbol on the input. If a lookahead symbol # is already set, we just use that. Otherwise, we'll pull # the next token off of the lookaheadstack or from the lexer #--! DEBUG debug.debug('') debug.debug('State : %s', state) #--! DEBUG if state not in defaulted_states: if not lookahead: if not lookaheadstack: lookahead = get_token() # Get the next token else: lookahead = lookaheadstack.pop() if not lookahead: lookahead = YaccSymbol() lookahead.type = '$end' # Check the action table ltype = lookahead.type t = actions[state].get(ltype) else: t = defaulted_states[state] #--! DEBUG debug.debug('Defaulted state %s: Reduce using %d', state, -t) #--! DEBUG #--! DEBUG debug.debug('Stack : %s', ('%s . %s' % (' '.join([xx.type for xx in symstack][1:]), str(lookahead))).lstrip()) #--! DEBUG 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:]])+']', goto[statestack[-1-plen]][pname]) else: debug.info('Action : Reduce rule [%s] with %s and goto state %d', p.str, [], goto[statestack[-1]][pname]) #--! 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:] self.state = state p.callable(pslice) del statestack[-plen:] #--! 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) # Save the current lookahead token symstack.extend(targ[1:-1]) # Put the production slice back on the stack statestack.pop() # Pop back one state (before the reduce) state = statestack[-1] sym.type = 'error' sym.value = 'error' lookahead = sym errorcount = error_count self.errorok = False 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 self.state = state 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) # Save the current lookahead token statestack.pop() # Pop back one state (before the reduce) state = statestack[-1] sym.type = 'error' sym.value = 'error' lookahead = sym errorcount = error_count self.errorok = False 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 is 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 = False errtoken = lookahead if errtoken.type == '$end': errtoken = None # End of file! if self.errorfunc: if errtoken and not hasattr(errtoken, 'lexer'): errtoken.lexer = lexer self.state = state tok = call_errorfunc(self.errorfunc, errtoken, self) if self.errorok: # User must have done some kind of panic # mode recovery on their own. The # returned token is the next lookahead lookahead = tok errtoken = None continue else: if errtoken: if hasattr(errtoken, 'lineno'): lineno = lookahead.lineno else: lineno = 0 if lineno: sys.stderr.write('yacc: Syntax error at line %d, token=%s\n' % (lineno, errtoken.type)) else: sys.stderr.write('yacc: Syntax error, token=%s' % errtoken.type) else: sys.stderr.write('yacc: Parse error in input. EOF\n') return else: errorcount = error_count # case 1: the statestack only has 1 entry on it. If we're in this state, the # entire parse has been rolled back and we're completely hosed. The token is # discarded and we just keep going. if len(statestack) <= 1 and lookahead.type != '$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 #--! TRACKING if tracking: sym.endlineno = getattr(lookahead, 'lineno', sym.lineno) sym.endlexpos = getattr(lookahead, 'lexpos', sym.lexpos) #--! TRACKING lookahead = None continue # Create the error symbol for the first time and make it the new lookahead symbol t = YaccSymbol() t.type = 'error' if hasattr(lookahead, 'lineno'): t.lineno = t.endlineno = lookahead.lineno if hasattr(lookahead, 'lexpos'): t.lexpos = t.endlexpos = lookahead.lexpos t.value = lookahead lookaheadstack.append(lookahead) lookahead = t else: sym = symstack.pop() #--! TRACKING if tracking: lookahead.lineno = sym.lineno lookahead.lexpos = sym.lexpos #--! TRACKING statestack.pop() state = statestack[-1] continue # Call an error function here raise RuntimeError('yacc: internal parser error!!!\n') #--! parsedebug-end # !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! # parseopt(). # # Optimized version of parse() method. DO NOT EDIT THIS CODE DIRECTLY! # This code is automatically generated by the ply/ygen.py script. Make # changes to the parsedebug() method instead. # !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! def parseopt(self, input=None, lexer=None, debug=False, tracking=False, tokenfunc=None): #--! parseopt-start 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.) defaulted_states = self.defaulted_states # Local reference to defaulted states 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: from . import 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 the parser() token method (sometimes used in error recovery) self.token = get_token # 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 = '$end' symstack.append(sym) state = 0 while True: # Get the next symbol on the input. If a lookahead symbol # is already set, we just use that. Otherwise, we'll pull # the next token off of the lookaheadstack or from the lexer if state not in defaulted_states: if not lookahead: if not lookaheadstack: lookahead = get_token() # Get the next token else: lookahead = lookaheadstack.pop() if not lookahead: lookahead = YaccSymbol() lookahead.type = '$end' # Check the action table ltype = lookahead.type t = actions[state].get(ltype) else: t = defaulted_states[state] 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:] self.state = state p.callable(pslice) del statestack[-plen:] 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) # Save the current lookahead token symstack.extend(targ[1:-1]) # Put the production slice back on the stack statestack.pop() # Pop back one state (before the reduce) state = statestack[-1] sym.type = 'error' sym.value = 'error' lookahead = sym errorcount = error_count self.errorok = False 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 self.state = state 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) # Save the current lookahead token statestack.pop() # Pop back one state (before the reduce) state = statestack[-1] sym.type = 'error' sym.value = 'error' lookahead = sym errorcount = error_count self.errorok = False continue # !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! if t == 0: n = symstack[-1] result = getattr(n, 'value', None) return result if t is 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 = False errtoken = lookahead if errtoken.type == '$end': errtoken = None # End of file! if self.errorfunc: if errtoken and not hasattr(errtoken, 'lexer'): errtoken.lexer = lexer self.state = state tok = call_errorfunc(self.errorfunc, errtoken, self) if self.errorok: # User must have done some kind of panic # mode recovery on their own. The # returned token is the next lookahead lookahead = tok errtoken = None continue else: if errtoken: if hasattr(errtoken, 'lineno'): lineno = lookahead.lineno else: lineno = 0 if lineno: sys.stderr.write('yacc: Syntax error at line %d, token=%s\n' % (lineno, errtoken.type)) else: sys.stderr.write('yacc: Syntax error, token=%s' % errtoken.type) else: sys.stderr.write('yacc: Parse error in input. EOF\n') return else: errorcount = error_count # case 1: the statestack only has 1 entry on it. If we're in this state, the # entire parse has been rolled back and we're completely hosed. The token is # discarded and we just keep going. if len(statestack) <= 1 and lookahead.type != '$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 #--! TRACKING if tracking: sym.endlineno = getattr(lookahead, 'lineno', sym.lineno) sym.endlexpos = getattr(lookahead, 'lexpos', sym.lexpos) #--! TRACKING lookahead = None continue # Create the error symbol for the first time and make it the new lookahead symbol t = YaccSymbol() t.type = 'error' if hasattr(lookahead, 'lineno'): t.lineno = t.endlineno = lookahead.lineno if hasattr(lookahead, 'lexpos'): t.lexpos = t.endlexpos = lookahead.lexpos t.value = lookahead lookaheadstack.append(lookahead) lookahead = t else: sym = symstack.pop() #--! TRACKING if tracking: lookahead.lineno = sym.lineno lookahead.lexpos = sym.lexpos #--! TRACKING statestack.pop() state = statestack[-1] continue # Call an error function here raise RuntimeError('yacc: internal parser error!!!\n') #--! parseopt-end # !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! # parseopt_notrack(). # # Optimized version of parseopt() with line number tracking removed. # DO NOT EDIT THIS CODE DIRECTLY. This code is automatically generated # by the ply/ygen.py script. Make changes to the parsedebug() method instead. # !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! def parseopt_notrack(self, input=None, lexer=None, debug=False, tracking=False, tokenfunc=None): #--! parseopt-notrack-start 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.) defaulted_states = self.defaulted_states # Local reference to defaulted states 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: from . import 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 the parser() token method (sometimes used in error recovery) self.token = get_token # 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 = '$end' symstack.append(sym) state = 0 while True: # Get the next symbol on the input. If a lookahead symbol # is already set, we just use that. Otherwise, we'll pull # the next token off of the lookaheadstack or from the lexer if state not in defaulted_states: if not lookahead: if not lookaheadstack: lookahead = get_token() # Get the next token else: lookahead = lookaheadstack.pop() if not lookahead: lookahead = YaccSymbol() lookahead.type = '$end' # Check the action table ltype = lookahead.type t = actions[state].get(ltype) else: t = defaulted_states[state] 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:] self.state = state p.callable(pslice) del statestack[-plen:] 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) # Save the current lookahead token symstack.extend(targ[1:-1]) # Put the production slice back on the stack statestack.pop() # Pop back one state (before the reduce) state = statestack[-1] sym.type = 'error' sym.value = 'error' lookahead = sym errorcount = error_count self.errorok = False 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 self.state = state 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) # Save the current lookahead token statestack.pop() # Pop back one state (before the reduce) state = statestack[-1] sym.type = 'error' sym.value = 'error' lookahead = sym errorcount = error_count self.errorok = False continue # !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! if t == 0: n = symstack[-1] result = getattr(n, 'value', None) return result if t is 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 = False errtoken = lookahead if errtoken.type == '$end': errtoken = None # End of file! if self.errorfunc: if errtoken and not hasattr(errtoken, 'lexer'): errtoken.lexer = lexer self.state = state tok = call_errorfunc(self.errorfunc, errtoken, self) if self.errorok: # User must have done some kind of panic # mode recovery on their own. The # returned token is the next lookahead lookahead = tok errtoken = None continue else: if errtoken: if hasattr(errtoken, 'lineno'): lineno = lookahead.lineno else: lineno = 0 if lineno: sys.stderr.write('yacc: Syntax error at line %d, token=%s\n' % (lineno, errtoken.type)) else: sys.stderr.write('yacc: Syntax error, token=%s' % errtoken.type) else: sys.stderr.write('yacc: Parse error in input. EOF\n') return else: errorcount = error_count # case 1: the statestack only has 1 entry on it. If we're in this state, the # entire parse has been rolled back and we're completely hosed. The token is # discarded and we just keep going. if len(statestack) <= 1 and lookahead.type != '$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 == '$end': # Whoa. We're really hosed here. Bail out return if lookahead.type != 'error': sym = symstack[-1] if sym.type == 'error': # Hmmm. Error is on top of stack, we'll just nuke input # symbol and continue lookahead = None continue # Create the error symbol for the first time and make it the new lookahead symbol t = YaccSymbol() t.type = 'error' if hasattr(lookahead, 'lineno'): t.lineno = t.endlineno = lookahead.lineno if hasattr(lookahead, 'lexpos'): t.lexpos = t.endlexpos = lookahead.lexpos t.value = lookahead lookaheadstack.append(lookahead) lookahead = t else: sym = symstack.pop() statestack.pop() state = statestack[-1] continue # Call an error function here raise RuntimeError('yacc: internal parser error!!!\n') #--! parseopt-notrack-end # ----------------------------------------------------------------------------- # === Grammar Representation === # # The following functions, classes, and variables are used to represent and # manipulate the rules that make up a grammar. # ----------------------------------------------------------------------------- # regex matching identifiers _is_identifier = re.compile(r'^[a-zA-Z0-9_-]+$') # ----------------------------------------------------------------------------- # 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 -> <empty>' % 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. try: p.lr_after = self.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 -> <empty>' % 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 = set() # 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 %r' % 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 %r. Already defined as a token' % (file, line, prodname)) if prodname == 'error': raise GrammarError('%s:%d: Illegal rule name %r. error is a reserved word' % (file, line, prodname)) if not _is_identifier.match(prodname): raise GrammarError('%s:%d: Illegal rule name %r' % (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 %r may only be a single character' % (file, line, s, prodname)) if c not 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 %r in rule %r' % (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) if not prodprec: raise GrammarError('%s:%d: Nothing known about the precedence of %r' % (file, line, precname)) else: self.UsedPrecedence.add(precname) 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 prodname not 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 t not 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] # ----------------------------------------------------------------------------- # 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 s in reachable: return reachable.add(s) for p in self.Prodnames.get(s, []): for r in p.prod: mark_reachable_from(r) reachable = set() mark_reachable_from(self.Productions[0].prod[0]) return [s for s in self.Nonterminals if s not in reachable] # ----------------------------------------------------------------------------- # 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] = True terminates['$end'] = True # Nonterminals: # Initialize to false: for n in self.Nonterminals: terminates[n] = False # Then propagate termination until no change: while True: some_change = False 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 = False break else: # didn't break from the loop, # so every symbol s terminates # so production p terminates. p_terminates = True if p_terminates: # symbol n terminates! if not terminates[n]: terminates[n] = True some_change = True # 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 s not in self.Prodnames and s not 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 s not in self.Prodnames and s not 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 all symbols. # ----------------------------------------------------------------------------- def unused_terminals(self): unused_tok = [] for s, v in self.Terminals.items(): if s != 'error' and not v: unused_tok.append(s) return unused_tok # ------------------------------------------------------------------------------ # unused_rules() # # Find all grammar rules that were defined, but not used (maybe not reachable) # Returns a list of productions. # ------------------------------------------------------------------------------ def unused_rules(self): unused_prod = [] for s, v in self.Nonterminals.items(): if not v: p = self.Prodnames[s][0] unused_prod.append(p) return unused_prod # ----------------------------------------------------------------------------- # unused_precedence() # # Returns a list of tuples (term,precedence) corresponding to precedence # rules that were never used by the grammar. term is the name of the terminal # on which precedence was applied and precedence is a string such as 'left' or # 'right' corresponding to the type of precedence. # ----------------------------------------------------------------------------- def unused_precedence(self): unused = [] for termname in self.Precedence: if not (termname in self.Terminals or termname in self.UsedPrecedence): unused.append((termname, self.Precedence[termname][0])) return unused # ------------------------------------------------------------------------- # _first() # # Compute the value of FIRST1(beta) where beta is a tuple of symbols. # # During execution of compute_first1, the result may be incomplete. # Afterward (e.g., when called from compute_follow()), it will be complete. # ------------------------------------------------------------------------- def _first(self, beta): # We are computing First(x1,x2,x3,...,xn) result = [] for x in beta: x_produces_empty = False # Add all the non-<empty> symbols of First[x] to the result. for f in self.First[x]: if f == '<empty>': x_produces_empty = True else: if f not in result: result.append(f) if x_produces_empty: # We have to consider the next x in beta, # i.e. stay in the loop. pass else: # We don't have to consider any further symbols in beta. break else: # There was no 'break' from the loop, # so x_produces_empty was true for all x in beta, # so beta produces empty as well. result.append('<empty>') return result # ------------------------------------------------------------------------- # compute_first() # # Compute the value of FIRST1(X) for all symbols # ------------------------------------------------------------------------- def compute_first(self): if self.First: return self.First # Terminals: for t in self.Terminals: self.First[t] = [t] self.First['$end'] = ['$end'] # Nonterminals: # Initialize to the empty set: for n in self.Nonterminals: self.First[n] = [] # Then propagate symbols until no change: while True: some_change = False for n in self.Nonterminals: for p in self.Prodnames[n]: for f in self._first(p.prod): if f not in self.First[n]: self.First[n].append(f) some_change = True if not some_change: break return self.First # --------------------------------------------------------------------- # compute_follow() # # Computes all of the follow sets for every non-terminal symbol. The # follow set is the set of all symbols that might follow a given # non-terminal. See the Dragon book, 2nd Ed. p. 189. # --------------------------------------------------------------------- def compute_follow(self, start=None): # If already computed, return the result if self.Follow: return self.Follow # If first sets not computed yet, do that first. if not self.First: self.compute_first() # Add '$end' to the follow list of the start symbol for k in self.Nonterminals: self.Follow[k] = [] if not start: start = self.Productions[1].name self.Follow[start] = ['$end'] while True: didadd = False for p in self.Productions[1:]: # Here is the production set for i, B in enumerate(p.prod): if B in self.Nonterminals: # Okay. We got a non-terminal in a production fst = self._first(p.prod[i+1:]) hasempty = False for f in fst: if f != '<empty>' and f not in self.Follow[B]: self.Follow[B].append(f) didadd = True if f == '<empty>': hasempty = True if hasempty or i == (len(p.prod)-1): # Add elements of follow(a) to follow(b) for f in self.Follow[p.name]: if f not in self.Follow[B]: self.Follow[B].append(f) didadd = True if not didadd: break return self.Follow # ----------------------------------------------------------------------------- # build_lritems() # # This function walks the list of productions and builds a complete set of the # LR items. The LR items are stored in two ways: First, they are uniquely # numbered and placed in the list _lritems. Second, a linked list of LR items # is built for each production. For example: # # E -> E PLUS E # # Creates the list # # [E -> . E PLUS E, E -> E . PLUS E, E -> E PLUS . E, E -> E PLUS E . ] # ----------------------------------------------------------------------------- def build_lritems(self): for p in self.Productions: lastlri = p i = 0 lr_items = [] while True: if i > len(p): lri = None else: lri = LRItem(p, i) # Precompute the list of productions immediately following try: lri.lr_after = self.Prodnames[lri.prod[i+1]] except (IndexError, KeyError): lri.lr_after = [] try: lri.lr_before = lri.prod[i-1] except IndexError: lri.lr_before = None lastlri.lr_next = lri if not lri: break lr_items.append(lri) lastlri = lri i += 1 p.lr_items = lr_items # ----------------------------------------------------------------------------- # == Class LRTable == # # This basic class represents a basic table of LR parsing information. # Methods for generating the tables are not defined here. They are defined # in the derived class LRGeneratedTable. # ----------------------------------------------------------------------------- class VersionError(YaccError): pass class LRTable(object): def __init__(self): self.lr_action = None self.lr_goto = None self.lr_productions = None self.lr_method = None def read_table(self, module): if isinstance(module, types.ModuleType): parsetab = module else: exec('import %s' % module) parsetab = sys.modules[module] if parsetab._tabversion != __tabversion__: raise VersionError('yacc table file version is out of date') self.lr_action = parsetab._lr_action self.lr_goto = parsetab._lr_goto self.lr_productions = [] for p in parsetab._lr_productions: self.lr_productions.append(MiniProduction(p)) self.lr_method = parsetab._lr_method return parsetab._lr_signature def read_pickle(self, filename): try: import cPickle as pickle except ImportError: import pickle if not os.path.exists(filename): raise ImportError in_f = open(filename, 'rb') tabversion = pickle.load(in_f) if tabversion != __tabversion__: raise VersionError('yacc table file version is out of date') self.lr_method = pickle.load(in_f) signature = pickle.load(in_f) self.lr_action = pickle.load(in_f) self.lr_goto = pickle.load(in_f) productions = pickle.load(in_f) self.lr_productions = [] for p in productions: self.lr_productions.append(MiniProduction(p)) in_f.close() return signature # Bind all production function names to callable objects in pdict def bind_callables(self, pdict): for p in self.lr_productions: p.bind(pdict) # ----------------------------------------------------------------------------- # === LR Generator === # # The following classes and functions are used to generate LR parsing tables on # a grammar. # ----------------------------------------------------------------------------- # ----------------------------------------------------------------------------- # digraph() # traverse() # # The following two functions are used to compute set valued functions # of the form: # # F(x) = F'(x) U U{F(y) \| x R y} # # This is used to compute the values of Read() sets as well as FOLLOW sets # in LALR(1) generation. # # Inputs: X - An input set # R - A relation # FP - Set-valued function # ------------------------------------------------------------------------------ def digraph(X, R, FP): N = {} for x in X: N[x] = 0 stack = [] F = {} for x in X: if N[x] == 0: traverse(x, N, stack, F, X, R, FP) return F def traverse(x, N, stack, F, X, R, FP): stack.append(x) d = len(stack) N[x] = d F[x] = FP(x) # F(X) <- F'(x) rel = R(x) # Get y's related to x for y in rel: if N[y] == 0: traverse(y, N, stack, F, X, R, FP) N[x] = min(N[x], N[y]) for a in F.get(y, []): if a not in F[x]: F[x].append(a) if N[x] == d: N[stack[-1]] = MAXINT F[stack[-1]] = F[x] element = stack.pop() while element != x: N[stack[-1]] = MAXINT F[stack[-1]] = F[x] element = stack.pop() class LALRError(YaccError): pass # ----------------------------------------------------------------------------- # == LRGeneratedTable == # # This class implements the LR table generation algorithm. There are no # public methods except for write() # ----------------------------------------------------------------------------- class LRGeneratedTable(LRTable): def __init__(self, grammar, method='LALR', log=None): if method not in ['SLR', 'LALR']: raise LALRError('Unsupported method %s' % method) self.grammar = grammar self.lr_method = method # Set up the logger if not log: log = NullLogger() self.log = log # Internal attributes self.lr_action = {} # Action table self.lr_goto = {} # Goto table self.lr_productions = grammar.Productions # Copy of grammar Production array self.lr_goto_cache = {} # Cache of computed gotos self.lr0_cidhash = {} # Cache of closures self._add_count = 0 # Internal counter used to detect cycles # Diagonistic information filled in by the table generator self.sr_conflict = 0 self.rr_conflict = 0 self.conflicts = [] # List of conflicts self.sr_conflicts = [] self.rr_conflicts = [] # Build the tables self.grammar.build_lritems() self.grammar.compute_first() self.grammar.compute_follow() self.lr_parse_table() # Compute the LR(0) closure operation on I, where I is a set of LR(0) items. def lr0_closure(self, I): self._add_count += 1 # Add everything in I to J J = I[:] didadd = True while didadd: didadd = False for j in J: for x in j.lr_after: if getattr(x, 'lr0_added', 0) == self._add_count: continue # Add B --> .G to J J.append(x.lr_next) x.lr0_added = self._add_count didadd = True return J # Compute the LR(0) goto function goto(I,X) where I is a set # of LR(0) items and X is a grammar symbol. This function is written # in a way that guarantees uniqueness of the generated goto sets # (i.e. the same goto set will never be returned as two different Python # objects). With uniqueness, we can later do fast set comparisons using # id(obj) instead of element-wise comparison. def lr0_goto(self, I, x): # First we look for a previously cached entry g = self.lr_goto_cache.get((id(I), x)) if g: return g # Now we generate the goto set in a way that guarantees uniqueness # of the result s = self.lr_goto_cache.get(x) if not s: s = {} self.lr_goto_cache[x] = s gs = [] for p in I: n = p.lr_next if n and n.lr_before == x: s1 = s.get(id(n)) if not s1: s1 = {} s[id(n)] = s1 gs.append(n) s = s1 g = s.get('$end') if not g: if gs: g = self.lr0_closure(gs) s['$end'] = g else: s['$end'] = gs self.lr_goto_cache[(id(I), x)] = g return g # Compute the LR(0) sets of item function def lr0_items(self): C = [self.lr0_closure([self.grammar.Productions[0].lr_next])] i = 0 for I in C: self.lr0_cidhash[id(I)] = i i += 1 # Loop over the items in C and each grammar symbols i = 0 while i < len(C): I = C[i] i += 1 # Collect all of the symbols that could possibly be in the goto(I,X) sets asyms = {} for ii in I: for s in ii.usyms: asyms[s] = None for x in asyms: g = self.lr0_goto(I, x) if not g or id(g) in self.lr0_cidhash: continue self.lr0_cidhash[id(g)] = len(C) C.append(g) return C # ----------------------------------------------------------------------------- # ==== LALR(1) Parsing ==== # # LALR(1) parsing is almost exactly the same as SLR except that instead of # relying upon Follow() sets when performing reductions, a more selective # lookahead set that incorporates the state of the LR(0) machine is utilized. # Thus, we mainly just have to focus on calculating the lookahead sets. # # The method used here is due to DeRemer and Pennelo (1982). # # DeRemer, F. L., and T. J. Pennelo: "Efficient Computation of LALR(1) # Lookahead Sets", ACM Transactions on Programming Languages and Systems, # Vol. 4, No. 4, Oct. 1982, pp. 615-649 # # Further details can also be found in: # # J. Tremblay and P. Sorenson, "The Theory and Practice of Compiler Writing", # McGraw-Hill Book Company, (1985). # # ----------------------------------------------------------------------------- # ----------------------------------------------------------------------------- # compute_nullable_nonterminals() # # Creates a dictionary containing all of the non-terminals that might produce # an empty production. # ----------------------------------------------------------------------------- def compute_nullable_nonterminals(self): nullable = set() num_nullable = 0 while True: for p in self.grammar.Productions[1:]: if p.len == 0: nullable.add(p.name) continue for t in p.prod: if t not in nullable: break else: nullable.add(p.name) if len(nullable) == num_nullable: break num_nullable = len(nullable) return nullable # ----------------------------------------------------------------------------- # find_nonterminal_trans(C) # # Given a set of LR(0) items, this functions finds all of the non-terminal # transitions. These are transitions in which a dot appears immediately before # a non-terminal. Returns a list of tuples of the form (state,N) where state # is the state number and N is the nonterminal symbol. # # The input C is the set of LR(0) items. # ----------------------------------------------------------------------------- def find_nonterminal_transitions(self, C): trans = [] for stateno, state in enumerate(C): for p in state: if p.lr_index < p.len - 1: t = (stateno, p.prod[p.lr_index+1]) if t[1] in self.grammar.Nonterminals: if t not in trans: trans.append(t) return trans # ----------------------------------------------------------------------------- # dr_relation() # # Computes the DR(p,A) relationships for non-terminal transitions. The input # is a tuple (state,N) where state is a number and N is a nonterminal symbol. # # Returns a list of terminals. # ----------------------------------------------------------------------------- def dr_relation(self, C, trans, nullable): state, N = trans terms = [] g = self.lr0_goto(C[state], N) for p in g: if p.lr_index < p.len - 1: a = p.prod[p.lr_index+1] if a in self.grammar.Terminals: if a not in terms: terms.append(a) # This extra bit is to handle the start state if state == 0 and N == self.grammar.Productions[0].prod[0]: terms.append('$end') return terms # ----------------------------------------------------------------------------- # reads_relation() # # Computes the READS() relation (p,A) READS (t,C). # ----------------------------------------------------------------------------- def reads_relation(self, C, trans, empty): # Look for empty transitions rel = [] state, N = trans g = self.lr0_goto(C[state], N) j = self.lr0_cidhash.get(id(g), -1) for p in g: if p.lr_index < p.len - 1: a = p.prod[p.lr_index + 1] if a in empty: rel.append((j, a)) return rel # ----------------------------------------------------------------------------- # compute_lookback_includes() # # Determines the lookback and includes relations # # LOOKBACK: # # This relation is determined by running the LR(0) state machine forward. # For example, starting with a production "N : . A B C", we run it forward # to obtain "N : A B C ." We then build a relationship between this final # state and the starting state. These relationships are stored in a dictionary # lookdict. # # INCLUDES: # # Computes the INCLUDE() relation (p,A) INCLUDES (p',B). # # This relation is used to determine non-terminal transitions that occur # inside of other non-terminal transition states. (p,A) INCLUDES (p', B) # if the following holds: # # B -> LAT, where T -> epsilon and p' -L-> p # # L is essentially a prefix (which may be empty), T is a suffix that must be # able to derive an empty string. State p' must lead to state p with the string L. # # ----------------------------------------------------------------------------- def compute_lookback_includes(self, C, trans, nullable): lookdict = {} # Dictionary of lookback relations includedict = {} # Dictionary of include relations # Make a dictionary of non-terminal transitions dtrans = {} for t in trans: dtrans[t] = 1 # Loop over all transitions and compute lookbacks and includes for state, N in trans: lookb = [] includes = [] for p in C[state]: if p.name != N: continue # Okay, we have a name match. We now follow the production all the way # through the state machine until we get the . on the right hand side lr_index = p.lr_index j = state while lr_index < p.len - 1: lr_index = lr_index + 1 t = p.prod[lr_index] # Check to see if this symbol and state are a non-terminal transition if (j, t) in dtrans: # Yes. Okay, there is some chance that this is an includes relation # the only way to know for certain is whether the rest of the # production derives empty li = lr_index + 1 while li < p.len: if p.prod[li] in self.grammar.Terminals: break # No forget it if p.prod[li] not in nullable: break li = li + 1 else: # Appears to be a relation between (j,t) and (state,N) includes.append((j, t)) g = self.lr0_goto(C[j], t) # Go to next set j = self.lr0_cidhash.get(id(g), -1) # Go to next state # When we get here, j is the final state, now we have to locate the production for r in C[j]: if r.name != p.name: continue if r.len != p.len: continue i = 0 # This look is comparing a production ". A B C" with "A B C ." while i < r.lr_index: if r.prod[i] != p.prod[i+1]: break i = i + 1 else: lookb.append((j, r)) for i in includes: if i not in includedict: includedict[i] = [] includedict[i].append((state, N)) lookdict[(state, N)] = lookb return lookdict, includedict # ----------------------------------------------------------------------------- # compute_read_sets() # # Given a set of LR(0) items, this function computes the read sets. # # Inputs: C = Set of LR(0) items # ntrans = Set of nonterminal transitions # nullable = Set of empty transitions # # Returns a set containing the read sets # ----------------------------------------------------------------------------- def compute_read_sets(self, C, ntrans, nullable): FP = lambda x: self.dr_relation(C, x, nullable) R = lambda x: self.reads_relation(C, x, nullable) F = digraph(ntrans, R, FP) return F # ----------------------------------------------------------------------------- # compute_follow_sets() # # Given a set of LR(0) items, a set of non-terminal transitions, a readset, # and an include set, this function computes the follow sets # # Follow(p,A) = Read(p,A) U U {Follow(p',B) \| (p,A) INCLUDES (p',B)} # # Inputs: # ntrans = Set of nonterminal transitions # readsets = Readset (previously computed) # inclsets = Include sets (previously computed) # # Returns a set containing the follow sets # ----------------------------------------------------------------------------- def compute_follow_sets(self, ntrans, readsets, inclsets): FP = lambda x: readsets[x] R = lambda x: inclsets.get(x, []) F = digraph(ntrans, R, FP) return F # ----------------------------------------------------------------------------- # add_lookaheads() # # Attaches the lookahead symbols to grammar rules. # # Inputs: lookbacks - Set of lookback relations # followset - Computed follow set # # This function directly attaches the lookaheads to productions contained # in the lookbacks set # ----------------------------------------------------------------------------- def add_lookaheads(self, lookbacks, followset): for trans, lb in lookbacks.items(): # Loop over productions in lookback for state, p in lb: if state not in p.lookaheads: p.lookaheads[state] = [] f = followset.get(trans, []) for a in f: if a not in p.lookaheads[state]: p.lookaheads[state].append(a) # ----------------------------------------------------------------------------- # add_lalr_lookaheads() # # This function does all of the work of adding lookahead information for use # with LALR parsing # ----------------------------------------------------------------------------- def add_lalr_lookaheads(self, C): # Determine all of the nullable nonterminals nullable = self.compute_nullable_nonterminals() # Find all non-terminal transitions trans = self.find_nonterminal_transitions(C) # Compute read sets readsets = self.compute_read_sets(C, trans, nullable) # Compute lookback/includes relations lookd, included = self.compute_lookback_includes(C, trans, nullable) # Compute LALR FOLLOW sets followsets = self.compute_follow_sets(trans, readsets, included) # Add all of the lookaheads self.add_lookaheads(lookd, followsets) # ----------------------------------------------------------------------------- # lr_parse_table() # # This function constructs the parse tables for SLR or LALR # ----------------------------------------------------------------------------- def lr_parse_table(self): Productions = self.grammar.Productions Precedence = self.grammar.Precedence goto = self.lr_goto # Goto array action = self.lr_action # Action array log = self.log # Logger for output actionp = {} # Action production array (temporary) log.info('Parsing method: %s', self.lr_method) # Step 1: Construct C = { I0, I1, ... IN}, collection of LR(0) items # This determines the number of states C = self.lr0_items() if self.lr_method == 'LALR': self.add_lalr_lookaheads(C) # Build the parser table, state by state st = 0 for I in C: # Loop over each production in I actlist = [] # List of actions st_action = {} st_actionp = {} st_goto = {} log.info('') log.info('state %d', st) log.info('') for p in I: log.info(' (%d) %s', p.number, p) log.info('') for p in I: if p.len == p.lr_index + 1: if p.name == "S'": # Start symbol. Accept! st_action['$end'] = 0 st_actionp['$end'] = p else: # We are at the end of a production. Reduce! if self.lr_method == 'LALR': laheads = p.lookaheads[st] else: laheads = self.grammar.Follow[p.name] for a in laheads: actlist.append((a, p, 'reduce using rule %d (%s)' % (p.number, p))) r = st_action.get(a) if r is not None: # Whoa. Have a shift/reduce or reduce/reduce conflict if r > 0: # Need to decide on shift or reduce here # By default we favor shifting. Need to add # some precedence rules here. # Shift precedence comes from the token sprec, slevel = Precedence.get(a, ('right', 0)) # Reduce precedence comes from rule being reduced (p) rprec, rlevel = Productions[p.number].prec if (slevel < rlevel) or ((slevel == rlevel) and (rprec == 'left')): # We really need to reduce here. st_action[a] = -p.number st_actionp[a] = p if not slevel and not rlevel: log.info(' ! shift/reduce conflict for %s resolved as reduce', a) self.sr_conflicts.append((st, a, 'reduce')) Productions[p.number].reduced += 1 elif (slevel == rlevel) and (rprec == 'nonassoc'): st_action[a] = None else: # Hmmm. Guess we'll keep the shift if not rlevel: log.info(' ! shift/reduce conflict for %s resolved as shift', a) self.sr_conflicts.append((st, a, 'shift')) elif r < 0: # Reduce/reduce conflict. In this case, we favor the rule # that was defined first in the grammar file oldp = Productions[-r] pp = Productions[p.number] if oldp.line > pp.line: st_action[a] = -p.number st_actionp[a] = p chosenp, rejectp = pp, oldp Productions[p.number].reduced += 1 Productions[oldp.number].reduced -= 1 else: chosenp, rejectp = oldp, pp self.rr_conflicts.append((st, chosenp, rejectp)) log.info(' ! reduce/reduce conflict for %s resolved using rule %d (%s)', a, st_actionp[a].number, st_actionp[a]) else: raise LALRError('Unknown conflict in state %d' % st) else: st_action[a] = -p.number st_actionp[a] = p Productions[p.number].reduced += 1 else: i = p.lr_index a = p.prod[i+1] # Get symbol right after the "." if a in self.grammar.Terminals: g = self.lr0_goto(I, a) j = self.lr0_cidhash.get(id(g), -1) if j >= 0: # We are in a shift state actlist.append((a, p, 'shift and go to state %d' % j)) r = st_action.get(a) if r is not None: # Whoa have a shift/reduce or shift/shift conflict if r > 0: if r != j: raise LALRError('Shift/shift conflict in state %d' % st) elif r < 0: # Do a precedence check. # - if precedence of reduce rule is higher, we reduce. # - if precedence of reduce is same and left assoc, we reduce. # - otherwise we shift # Shift precedence comes from the token sprec, slevel = Precedence.get(a, ('right', 0)) # Reduce precedence comes from the rule that could have been reduced rprec, rlevel = Productions[st_actionp[a].number].prec if (slevel > rlevel) or ((slevel == rlevel) and (rprec == 'right')): # We decide to shift here... highest precedence to shift Productions[st_actionp[a].number].reduced -= 1 st_action[a] = j st_actionp[a] = p if not rlevel: log.info(' ! shift/reduce conflict for %s resolved as shift', a) self.sr_conflicts.append((st, a, 'shift')) elif (slevel == rlevel) and (rprec == 'nonassoc'): st_action[a] = None else: # Hmmm. Guess we'll keep the reduce if not slevel and not rlevel: log.info(' ! shift/reduce conflict for %s resolved as reduce', a) self.sr_conflicts.append((st, a, 'reduce')) else: raise LALRError('Unknown conflict in state %d' % st) else: st_action[a] = j st_actionp[a] = p # Print the actions associated with each terminal _actprint = {} for a, p, m in actlist: if a in st_action: if p is st_actionp[a]: log.info(' %-15s %s', a, m) _actprint[(a, m)] = 1 log.info('') # Print the actions that were not used. (debugging) not_used = 0 for a, p, m in actlist: if a in st_action: if p is not st_actionp[a]: if not (a, m) in _actprint: log.debug(' ! %-15s [ %s ]', a, m) not_used = 1 _actprint[(a, m)] = 1 if not_used: log.debug('') # Construct the goto table for this state nkeys = {} for ii in I: for s in ii.usyms: if s in self.grammar.Nonterminals: nkeys[s] = None for n in nkeys: g = self.lr0_goto(I, n) j = self.lr0_cidhash.get(id(g), -1) if j >= 0: st_goto[n] = j log.info(' %-30s shift and go to state %d', n, j) action[st] = st_action actionp[st] = st_actionp goto[st] = st_goto st += 1 # ----------------------------------------------------------------------------- # write() # # This function writes the LR parsing tables to a file # ----------------------------------------------------------------------------- def write_table(self, tabmodule, outputdir='', signature=''): if isinstance(tabmodule, types.ModuleType): raise IOError("Won't overwrite existing tabmodule") basemodulename = tabmodule.split('.')[-1] filename = os.path.join(outputdir, basemodulename) + '.py' try: f = open(filename, 'w') f.write(''' # %s # This file is automatically generated. Do not edit. # pylint: disable=W,C,R _tabversion = %r _lr_method = %r _lr_signature = %r ''' % (os.path.basename(filename), __tabversion__, self.lr_method, signature)) # Change smaller to 0 to go back to original tables smaller = 1 # Factor out names to try and make smaller if smaller: items = {} for s, nd in self.lr_action.items(): for name, v in nd.items(): i = items.get(name) if not i: i = ([], []) items[name] = i i[0].append(s) i[1].append(v) f.write('\n_lr_action_items = {') for k, v in items.items(): f.write('%r:([' % k) for i in v[0]: f.write('%r,' % i) f.write('],[') for i in v[1]: f.write('%r,' % i) f.write(']),') f.write('}\n') f.write(''' _lr_action = {} for _k, _v in _lr_action_items.items(): for _x,_y in zip(_v[0],_v[1]): if not _x in _lr_action: _lr_action[_x] = {} _lr_action[_x][_k] = _y del _lr_action_items ''') else: f.write('\n_lr_action = { ') for k, v in self.lr_action.items(): f.write('(%r,%r):%r,' % (k[0], k[1], v)) f.write('}\n') if smaller: # Factor out names to try and make smaller items = {} for s, nd in self.lr_goto.items(): for name, v in nd.items(): i = items.get(name) if not i: i = ([], []) items[name] = i i[0].append(s) i[1].append(v) f.write('\n_lr_goto_items = {') for k, v in items.items(): f.write('%r:([' % k) for i in v[0]: f.write('%r,' % i) f.write('],[') for i in v[1]: f.write('%r,' % i) f.write(']),') f.write('}\n') f.write(''' _lr_goto = {} for _k, _v in _lr_goto_items.items(): for _x, _y in zip(_v[0], _v[1]): if not _x in _lr_goto: _lr_goto[_x] = {} _lr_goto[_x][_k] = _y del _lr_goto_items ''') else: f.write('\n_lr_goto = { ') for k, v in self.lr_goto.items(): f.write('(%r,%r):%r,' % (k[0], k[1], v)) f.write('}\n') # Write production table f.write('_lr_productions = [\n') for p in self.lr_productions: if p.func: f.write(' (%r,%r,%d,%r,%r,%d),\n' % (p.str, p.name, p.len, p.func, os.path.basename(p.file), p.line)) else: f.write(' (%r,%r,%d,None,None,None),\n' % (str(p), p.name, p.len)) f.write(']\n') f.close() except IOError as e: raise # ----------------------------------------------------------------------------- # pickle_table() # # This function pickles the LR parsing tables to a supplied file object # ----------------------------------------------------------------------------- def pickle_table(self, filename, signature=''): try: import cPickle as pickle except ImportError: import pickle with open(filename, 'wb') as outf: pickle.dump(__tabversion__, outf, pickle_protocol) pickle.dump(self.lr_method, outf, pickle_protocol) pickle.dump(signature, outf, pickle_protocol) pickle.dump(self.lr_action, outf, pickle_protocol) pickle.dump(self.lr_goto, outf, pickle_protocol) outp = [] for p in self.lr_productions: if p.func: outp.append((p.str, p.name, p.len, p.func, os.path.basename(p.file), p.line)) else: outp.append((str(p), p.name, p.len, None, None, None)) pickle.dump(outp, outf, pickle_protocol) # ----------------------------------------------------------------------------- # === INTROSPECTION === # # The following functions and classes are used to implement the PLY # introspection features followed by the yacc() function itself. # ----------------------------------------------------------------------------- # ----------------------------------------------------------------------------- # get_caller_module_dict() # # This function returns a dictionary containing all of the symbols defined within # a caller further down the call stack. This is used to get the environment # associated with the yacc() call if none was provided. # ----------------------------------------------------------------------------- def get_caller_module_dict(levels): f = sys._getframe(levels) ldict = f.f_globals.copy() if f.f_globals != f.f_locals: ldict.update(f.f_locals) return ldict # ----------------------------------------------------------------------------- # parse_grammar() # # This takes a raw grammar rule string and parses it into production data # ----------------------------------------------------------------------------- def parse_grammar(doc, file, line): grammar = [] # Split the doc string into lines pstrings = doc.splitlines() lastp = None dline = line for ps in pstrings: dline += 1 p = ps.split() if not p: continue try: if p[0] == '\|': # This is a continuation of a previous rule if not lastp: raise SyntaxError("%s:%d: Misplaced '\|'" % (file, dline)) prodname = lastp syms = p[1:] else: prodname = p[0] lastp = prodname syms = p[2:] assign = p[1] if assign != ':' and assign != '::=': raise SyntaxError("%s:%d: Syntax error. Expected ':'" % (file, dline)) grammar.append((file, dline, prodname, syms)) except SyntaxError: raise except Exception: raise SyntaxError('%s:%d: Syntax error in rule %r' % (file, dline, ps.strip())) return grammar # ----------------------------------------------------------------------------- # ParserReflect() # # This class represents information extracted for building a parser including # start symbol, error function, tokens, precedence list, action functions, # etc. # ----------------------------------------------------------------------------- class ParserReflect(object): def __init__(self, pdict, log=None): self.pdict = pdict self.start = None self.error_func = None self.tokens = None self.modules = set() self.grammar = [] self.error = False if log is None: self.log = PlyLogger(sys.stderr) else: self.log = log # Get all of the basic information def get_all(self): self.get_start() self.get_error_func() self.get_tokens() self.get_precedence() self.get_pfunctions() # Validate all of the information def validate_all(self): self.validate_start() self.validate_error_func() self.validate_tokens() self.validate_precedence() self.validate_pfunctions() self.validate_modules() return self.error # Compute a signature over the grammar def signature(self): parts = [] try: if self.start: parts.append(self.start) if self.prec: parts.append(''.join([''.join(p) for p in self.prec])) if self.tokens: parts.append(' '.join(self.tokens)) for f in self.pfuncs: if f[3]: parts.append(f[3]) except (TypeError, ValueError): pass return ''.join(parts) # ----------------------------------------------------------------------------- # validate_modules() # # This method checks to see if there are duplicated p_rulename() functions # in the parser module file. Without this function, it is really easy for # users to make mistakes by cutting and pasting code fragments (and it's a real # bugger to try and figure out why the resulting parser doesn't work). Therefore, # we just do a little regular expression pattern matching of def statements # to try and detect duplicates. # ----------------------------------------------------------------------------- def validate_modules(self): # Match def p_funcname( fre = re.compile(r'\sdef\s+(p_[a-zA-Z_0-9]*)\(') for module in self.modules: try: lines, linen = inspect.getsourcelines(module) except IOError: continue counthash = {} for linen, line in enumerate(lines): linen += 1 m = fre.match(line) if m: name = m.group(1) prev = counthash.get(name) if not prev: counthash[name] = linen else: filename = inspect.getsourcefile(module) self.log.warning('%s:%d: Function %s redefined. Previously defined on line %d', filename, linen, name, prev) # Get the start symbol def get_start(self): self.start = self.pdict.get('start') # Validate the start symbol def validate_start(self): if self.start is not None: if not isinstance(self.start, string_types): self.log.error("'start' must be a string") # Look for error handler def get_error_func(self): self.error_func = self.pdict.get('p_error') # Validate the error function def validate_error_func(self): if self.error_func: if isinstance(self.error_func, types.FunctionType): ismethod = 0 elif isinstance(self.error_func, types.MethodType): ismethod = 1 else: self.log.error("'p_error' defined, but is not a function or method") self.error = True return eline = self.error_func.__code__.co_firstlineno efile = self.error_func.__code__.co_filename module = inspect.getmodule(self.error_func) self.modules.add(module) argcount = self.error_func.__code__.co_argcount - ismethod if argcount != 1: self.log.error('%s:%d: p_error() requires 1 argument', efile, eline) self.error = True # Get the tokens map def get_tokens(self): tokens = self.pdict.get('tokens') if not tokens: self.log.error('No token list is defined') self.error = True return if not isinstance(tokens, (list, tuple)): self.log.error('tokens must be a list or tuple') self.error = True return if not tokens: self.log.error('tokens is empty') self.error = True return self.tokens = sorted(tokens) # Validate the tokens def validate_tokens(self): # Validate the tokens. if 'error' in self.tokens: self.log.error("Illegal token name 'error'. Is a reserved word") self.error = True return terminals = set() for n in self.tokens: if n in terminals: self.log.warning('Token %r multiply defined', n) terminals.add(n) # Get the precedence map (if any) def get_precedence(self): self.prec = self.pdict.get('precedence') # Validate and parse the precedence map def validate_precedence(self): preclist = [] if self.prec: if not isinstance(self.prec, (list, tuple)): self.log.error('precedence must be a list or tuple') self.error = True return for level, p in enumerate(self.prec): if not isinstance(p, (list, tuple)): self.log.error('Bad precedence table') self.error = True return if len(p) < 2: self.log.error('Malformed precedence entry %s. Must be (assoc, term, ..., term)', p) self.error = True return assoc = p[0] if not isinstance(assoc, string_types): self.log.error('precedence associativity must be a string') self.error = True return for term in p[1:]: if not isinstance(term, string_types): self.log.error('precedence items must be strings') self.error = True return preclist.append((term, assoc, level+1)) self.preclist = preclist # Get all p_functions from the grammar def get_pfunctions(self): p_functions = [] for name, item in self.pdict.items(): if not name.startswith('p_') or name == 'p_error': continue if isinstance(item, (types.FunctionType, types.MethodType)): line = getattr(item, 'co_firstlineno', item.__code__.co_firstlineno) module = inspect.getmodule(item) p_functions.append((line, module, name, item.__doc__)) # Sort all of the actions by line number; make sure to stringify # modules to make them sortable, since `line` may not uniquely sort all # p functions p_functions.sort(key=lambda p_function: ( p_function[0], str(p_function[1]), p_function[2], p_function[3])) self.pfuncs = p_functions # Validate all of the p_functions def validate_pfunctions(self): grammar = [] # Check for non-empty symbols if len(self.pfuncs) == 0: self.log.error('no rules of the form p_rulename are defined') self.error = True return for line, module, name, doc in self.pfuncs: file = inspect.getsourcefile(module) func = self.pdict[name] if isinstance(func, types.MethodType): reqargs = 2 else: reqargs = 1 if func.__code__.co_argcount > reqargs: self.log.error('%s:%d: Rule %r has too many arguments', file, line, func.__name__) self.error = True elif func.__code__.co_argcount < reqargs: self.log.error('%s:%d: Rule %r requires an argument', file, line, func.__name__) self.error = True elif not func.__doc__: self.log.warning('%s:%d: No documentation string specified in function %r (ignored)', file, line, func.__name__) else: try: parsed_g = parse_grammar(doc, file, line) for g in parsed_g: grammar.append((name, g)) except SyntaxError as e: self.log.error(str(e)) self.error = True # Looks like a valid grammar rule # Mark the file in which defined. self.modules.add(module) # Secondary validation step that looks for p_ definitions that are not functions # or functions that look like they might be grammar rules. for n, v in self.pdict.items(): if n.startswith('p_') and isinstance(v, (types.FunctionType, types.MethodType)): continue if n.startswith('t_'): continue if n.startswith('p_') and n != 'p_error': self.log.warning('%r not defined as a function', n) if ((isinstance(v, types.FunctionType) and v.__code__.co_argcount == 1) or (isinstance(v, types.MethodType) and v.__func__.__code__.co_argcount == 2)): if v.__doc__: try: doc = v.__doc__.split(' ') if doc[1] == ':': self.log.warning('%s:%d: Possible grammar rule %r defined without p_ prefix', v.__code__.co_filename, v.__code__.co_firstlineno, n) except IndexError: pass self.grammar = grammar # ----------------------------------------------------------------------------- # yacc(module) # # Build a parser # ----------------------------------------------------------------------------- def yacc(method='LALR', debug=yaccdebug, module=None, tabmodule=tab_module, start=None, check_recursion=True, optimize=False, write_tables=True, debugfile=debug_file, outputdir=None, debuglog=None, errorlog=None, picklefile=None): if tabmodule is None: tabmodule = tab_module # Reference to the parsing method of the last built parser global parse # If pickling is enabled, table files are not created if picklefile: write_tables = 0 if errorlog is None: errorlog = PlyLogger(sys.stderr) # Get the module dictionary used for the parser if module: _items = [(k, getattr(module, k)) for k in dir(module)] pdict = dict(_items) # If no __file__ or __package__ attributes are available, try to obtain them # from the __module__ instead if '__file__' not in pdict: pdict['__file__'] = sys.modules[pdict['__module__']].__file__ if '__package__' not in pdict and '__module__' in pdict: if hasattr(sys.modules[pdict['__module__']], '__package__'): pdict['__package__'] = sys.modules[pdict['__module__']].__package__ else: pdict = get_caller_module_dict(2) if outputdir is None: # If no output directory is set, the location of the output files # is determined according to the following rules: # - If tabmodule specifies a package, files go into that package directory # - Otherwise, files go in the same directory as the specifying module if isinstance(tabmodule, types.ModuleType): srcfile = tabmodule.__file__ else: if '.' not in tabmodule: srcfile = pdict['__file__'] else: parts = tabmodule.split('.') pkgname = '.'.join(parts[:-1]) exec('import %s' % pkgname) srcfile = getattr(sys.modules[pkgname], '__file__', '') outputdir = os.path.dirname(srcfile) # Determine if the module is package of a package or not. # If so, fix the tabmodule setting so that tables load correctly pkg = pdict.get('__package__') if pkg and isinstance(tabmodule, str): if '.' not in tabmodule: tabmodule = pkg + '.' + tabmodule # Set start symbol if it's specified directly using an argument if start is not None: pdict['start'] = start # Collect parser information from the dictionary pinfo = ParserReflect(pdict, log=errorlog) pinfo.get_all() if pinfo.error: raise YaccError('Unable to build parser') # Check signature against table files (if any) signature = pinfo.signature() # Read the tables try: lr = LRTable() if picklefile: read_signature = lr.read_pickle(picklefile) else: read_signature = lr.read_table(tabmodule) if optimize or (read_signature == signature): try: lr.bind_callables(pinfo.pdict) parser = LRParser(lr, pinfo.error_func) parse = parser.parse return parser except Exception as e: errorlog.warning('There was a problem loading the table file: %r', e) except VersionError as e: errorlog.warning(str(e)) except ImportError: pass if debuglog is None: if debug: try: debuglog = PlyLogger(open(os.path.join(outputdir, debugfile), 'w')) except IOError as e: errorlog.warning("Couldn't open %r. %s" % (debugfile, e)) debuglog = NullLogger() else: debuglog = NullLogger() debuglog.info('Created by PLY version %s (http://www.dabeaz.com/ply)', __version__) errors = False # Validate the parser information if pinfo.validate_all(): raise YaccError('Unable to build parser') if not pinfo.error_func: errorlog.warning('no p_error() function is defined') # Create a grammar object grammar = Grammar(pinfo.tokens) # Set precedence level for terminals for term, assoc, level in pinfo.preclist: try: grammar.set_precedence(term, assoc, level) except GrammarError as e: errorlog.warning('%s', e) # Add productions to the grammar for funcname, gram in pinfo.grammar: file, line, prodname, syms = gram try: grammar.add_production(prodname, syms, funcname, file, line) except GrammarError as e: errorlog.error('%s', e) errors = True # Set the grammar start symbols try: if start is None: grammar.set_start(pinfo.start) else: grammar.set_start(start) except GrammarError as e: errorlog.error(str(e)) errors = True if errors: raise YaccError('Unable to build parser') # Verify the grammar structure undefined_symbols = grammar.undefined_symbols() for sym, prod in undefined_symbols: errorlog.error('%s:%d: Symbol %r used, but not defined as a token or a rule', prod.file, prod.line, sym) errors = True unused_terminals = grammar.unused_terminals() if unused_terminals: debuglog.info('') debuglog.info('Unused terminals:') debuglog.info('') for term in unused_terminals: errorlog.warning('Token %r defined, but not used', term) debuglog.info(' %s', term) # Print out all productions to the debug log if debug: debuglog.info('') debuglog.info('Grammar') debuglog.info('') for n, p in enumerate(grammar.Productions): debuglog.info('Rule %-5d %s', n, p) # Find unused non-terminals unused_rules = grammar.unused_rules() for prod in unused_rules: errorlog.warning('%s:%d: Rule %r defined, but not used', prod.file, prod.line, prod.name) if len(unused_terminals) == 1: errorlog.warning('There is 1 unused token') if len(unused_terminals) > 1: errorlog.warning('There are %d unused tokens', len(unused_terminals)) if len(unused_rules) == 1: errorlog.warning('There is 1 unused rule') if len(unused_rules) > 1: errorlog.warning('There are %d unused rules', len(unused_rules)) if debug: debuglog.info('') debuglog.info('Terminals, with rules where they appear') debuglog.info('') terms = list(grammar.Terminals) terms.sort() for term in terms: debuglog.info('%-20s : %s', term, ' '.join([str(s) for s in grammar.Terminals[term]])) debuglog.info('') debuglog.info('Nonterminals, with rules where they appear') debuglog.info('') nonterms = list(grammar.Nonterminals) nonterms.sort() for nonterm in nonterms: debuglog.info('%-20s : %s', nonterm, ' '.join([str(s) for s in grammar.Nonterminals[nonterm]])) debuglog.info('') if check_recursion: unreachable = grammar.find_unreachable() for u in unreachable: errorlog.warning('Symbol %r is unreachable', u) infinite = grammar.infinite_cycles() for inf in infinite: errorlog.error('Infinite recursion detected for symbol %r', inf) errors = True unused_prec = grammar.unused_precedence() for term, assoc in unused_prec: errorlog.error('Precedence rule %r defined for unknown symbol %r', assoc, term) errors = True if errors: raise YaccError('Unable to build parser') # Run the LRGeneratedTable on the grammar if debug: errorlog.debug('Generating %s tables', method) lr = LRGeneratedTable(grammar, method, debuglog) if debug: num_sr = len(lr.sr_conflicts) # Report shift/reduce and reduce/reduce conflicts if num_sr == 1: errorlog.warning('1 shift/reduce conflict') elif num_sr > 1: errorlog.warning('%d shift/reduce conflicts', num_sr) num_rr = len(lr.rr_conflicts) if num_rr == 1: errorlog.warning('1 reduce/reduce conflict') elif num_rr > 1: errorlog.warning('%d reduce/reduce conflicts', num_rr) # Write out conflicts to the output file if debug and (lr.sr_conflicts or lr.rr_conflicts): debuglog.warning('') debuglog.warning('Conflicts:') debuglog.warning('') for state, tok, resolution in lr.sr_conflicts: debuglog.warning('shift/reduce conflict for %s in state %d resolved as %s', tok, state, resolution) already_reported = set() for state, rule, rejected in lr.rr_conflicts: if (state, id(rule), id(rejected)) in already_reported: continue debuglog.warning('reduce/reduce conflict in state %d resolved using rule (%s)', state, rule) debuglog.warning('rejected rule (%s) in state %d', rejected, state) errorlog.warning('reduce/reduce conflict in state %d resolved using rule (%s)', state, rule) errorlog.warning('rejected rule (%s) in state %d', rejected, state) already_reported.add((state, id(rule), id(rejected))) warned_never = [] for state, rule, rejected in lr.rr_conflicts: if not rejected.reduced and (rejected not in warned_never): debuglog.warning('Rule (%s) is never reduced', rejected) errorlog.warning('Rule (%s) is never reduced', rejected) warned_never.append(rejected) # Write the table file if requested if write_tables: try: lr.write_table(tabmodule, outputdir, signature) if tabmodule in sys.modules: del sys.modules[tabmodule] except IOError as e: errorlog.warning("Couldn't create %r. %s" % (tabmodule, e)) # Write a pickled version of the tables if picklefile: try: lr.pickle_table(picklefile, signature) except IOError as e: errorlog.warning("Couldn't create %r. %s" % (picklefile, e)) # Build the parser lr.bind_callables(pinfo.pdict) parser = LRParser(lr, pinfo.error_func) parse = parser.parse return parser	134.51 KB	2018-02-15 18:44
📄 ygen.py # ply: ygen.py # # This is a support program that auto-generates different versions of the YACC parsing # function with different features removed for the purposes of performance. # # Users should edit the method LRParser.parsedebug() in yacc.py. The source code # for that method is then used to create the other methods. See the comments in # yacc.py for further details. import os.path import shutil def get_source_range(lines, tag): srclines = enumerate(lines) start_tag = '#--! %s-start' % tag end_tag = '#--! %s-end' % tag for start_index, line in srclines: if line.strip().startswith(start_tag): break for end_index, line in srclines: if line.strip().endswith(end_tag): break return (start_index + 1, end_index) def filter_section(lines, tag): filtered_lines = [] include = True tag_text = '#--! %s' % tag for line in lines: if line.strip().startswith(tag_text): include = not include elif include: filtered_lines.append(line) return filtered_lines def main(): dirname = os.path.dirname(__file__) shutil.copy2(os.path.join(dirname, 'yacc.py'), os.path.join(dirname, 'yacc.py.bak')) with open(os.path.join(dirname, 'yacc.py'), 'r') as f: lines = f.readlines() parse_start, parse_end = get_source_range(lines, 'parsedebug') parseopt_start, parseopt_end = get_source_range(lines, 'parseopt') parseopt_notrack_start, parseopt_notrack_end = get_source_range(lines, 'parseopt-notrack') # Get the original source orig_lines = lines[parse_start:parse_end] # Filter the DEBUG sections out parseopt_lines = filter_section(orig_lines, 'DEBUG') # Filter the TRACKING sections out parseopt_notrack_lines = filter_section(parseopt_lines, 'TRACKING') # Replace the parser source sections with updated versions lines[parseopt_notrack_start:parseopt_notrack_end] = parseopt_notrack_lines lines[parseopt_start:parseopt_end] = parseopt_lines lines = [line.rstrip()+'\n' for line in lines] with open(os.path.join(dirname, 'yacc.py'), 'w') as f: f.writelines(lines) print('Updated yacc.py') if __name__ == '__main__': main()	2.19 KB	2018-02-15 18:44

Terminal