Experimental parser: add backtracking

tools/lexer_generator/rule_parser.py

 # Copyright 2013 the V8 project authors. 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
@@ skipping 17 matching lines @@
 import ply.lex as lex
 import ply.yacc as yacc
 from term import Term
 from regex_parser import RegexParser, ParserBuilder
 from automaton import Action
 from nfa_builder import NfaBuilder
 from dfa import Dfa
 from dfa_optimizer import DfaOptimizer
 from dfa_minimizer import DfaMinimizer
 from transition_key import TransitionKey, KeyEncoding
+from backtracking_generator import BacktrackingGenerator
 
 class RuleLexer:
 
   tokens = (
     'DEFAULT_ACTION',
     'EPSILON',
     'EOS',
     'CATCH_ALL',
 
     'IDENTIFIER',
@@ skipping 299 matching lines @@
 
   def default_automata(self):
     return self.__automata['default']
 
   def default_action(self):
     return self.__default_action
 
   class Automata(object):
     'a container for the resulting automata, which are lazily built'
 
-    def __init__(self, encoding, character_classes, rule_map, name):
-      self.__encoding = encoding
+    def __init__(self, rule_processor, character_classes, rule_map, name):
+      self.__rule_processor = rule_processor
       self.__character_classes = character_classes
       self.__rule_map = rule_map
       self.__name = name
       self.__nfa = None
       self.__dfa = None
       self.__minimial_dfa = None
 
+    def encoding(self):
+      return self.__rule_processor.encoding()
+
     def name(self):
       return self.__name
 
     def rule_term(self):
       return self.__rule_map[self.__name]
 
     def nfa(self):
       if not self.__nfa:
         self.__nfa = NfaBuilder.nfa(
-          self.__encoding, self.__character_classes,
+          self.encoding(), self.__character_classes,
           self.__rule_map, self.__name)
       return self.__nfa
 
     def dfa(self):
       if not self.__dfa:
         (start, dfa_nodes) = self.nfa().compute_dfa()
-        self.__dfa = Dfa(self.nfa().encoding(), start, dfa_nodes)
+        dfa = Dfa(self.encoding(), start, dfa_nodes)
+        if self.name() == 'default':
+          dfa = BacktrackingGenerator.generate(
+            dfa, self.__rule_processor.default_action())
+        self.__dfa = dfa
       return self.__dfa
 
     def optimize_dfa(self, log = False):
       assert not self.__dfa
       assert not self.__minimial_dfa
       self.__dfa = DfaOptimizer.optimize(self.minimal_dfa(), log)
       self.__minimial_dfa = None
 
     def minimal_dfa(self):
       if not self.__minimial_dfa:
         self.__minimial_dfa = DfaMinimizer(self.dfa()).minimize()
       return self.__minimial_dfa
 
   def __process_parser_state(self):
     parser_state = self.__parser_state
     assert 'default' in parser_state.rules, "default lexer state required"
     # Check that we don't transition to a nonexistent state.
     assert parser_state.transitions <= set(parser_state.rules.keys())
     rule_map = {}
     # process all subgraphs
     for tree_name, v in parser_state.rules.items():
       assert v['trees'], "lexer state %s is empty" % tree_name
       rule_map[tree_name] = NfaBuilder.or_terms(v['trees'])
     # build the automata
     for name, tree in rule_map.items():
       self.__automata[name] = RuleProcessor.Automata(
-        parser_state.encoding, parser_state.character_classes, rule_map, name)
+        self, parser_state.character_classes, rule_map, name)
     # process default_action
     default_action = parser_state.rules['default']['default_action']
     self.__default_action = Action(default_action, 0)