All Downloads are FREE. Search and download functionalities are using the official Maven repository.

Python3.src.antlr4.atn.ATN.py Maven / Gradle / Ivy

There is a newer version: 4.13.2
Show newest version
# [The "BSD license"]
#  Copyright (c) 2013 Terence Parr
#  Copyright (c) 2013 Sam Harwell
#  Copyright (c) 2014 Eric Vergnaud
#  All rights reserved.
#
#  Redistribution and use in source and binary forms, with or without
#  modification, are permitted provided that the following conditions
#  are met:
#
#  1. Redistributions of source code must retain the above copyright
#     notice, this list of conditions and the following disclaimer.
#  2. 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.
#  3. The name of the author may not be used to endorse or promote products
#     derived from this software without specific prior written permission.
#
#  THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``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 AUTHOR 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.
#/
from antlr4.IntervalSet import IntervalSet

from antlr4.RuleContext import RuleContext

from antlr4.Token import Token
from antlr4.atn.ATNType import ATNType
from antlr4.atn.ATNState import ATNState, DecisionState


class ATN(object):

    INVALID_ALT_NUMBER = 0

    # Used for runtime deserialization of ATNs from strings#/
    def __init__(self, grammarType:ATNType , maxTokenType:int ):
        # The type of the ATN.
        self.grammarType = grammarType
        # The maximum value for any symbol recognized by a transition in the ATN.
        self.maxTokenType = maxTokenType
        self.states = []
        # Each subrule/rule is a decision point and we must track them so we
        #  can go back later and build DFA predictors for them.  This includes
        #  all the rules, subrules, optional blocks, ()+, ()* etc...
        self.decisionToState = []
        # Maps from rule index to starting state number.
        self.ruleToStartState = []
        # Maps from rule index to stop state number.
        self.ruleToStopState = None
        self.modeNameToStartState = dict()
        # For lexer ATNs, this maps the rule index to the resulting token type.
        # For parser ATNs, this maps the rule index to the generated bypass token
        # type if the
        # {@link ATNDeserializationOptions#isGenerateRuleBypassTransitions}
        # deserialization option was specified; otherwise, this is {@code null}.
        self.ruleToTokenType = None
        # For lexer ATNs, this is an array of {@link LexerAction} objects which may
        # be referenced by action transitions in the ATN.
        self.lexerActions = None
        self.modeToStartState = []

    # Compute the set of valid tokens that can occur starting in state {@code s}.
    #  If {@code ctx} is null, the set of tokens will not include what can follow
    #  the rule surrounding {@code s}. In other words, the set will be
    #  restricted to tokens reachable staying within {@code s}'s rule.
    def nextTokensInContext(self, s:ATNState, ctx:RuleContext):
        from antlr4.LL1Analyzer import LL1Analyzer
        anal = LL1Analyzer(self)
        return anal.LOOK(s, ctx=ctx)

    # Compute the set of valid tokens that can occur starting in {@code s} and
    # staying in same rule. {@link Token#EPSILON} is in set if we reach end of
    # rule.
    def nextTokensNoContext(self, s:ATNState):
        if s.nextTokenWithinRule is not None:
            return s.nextTokenWithinRule
        s.nextTokenWithinRule = self.nextTokensInContext(s, None)
        s.nextTokenWithinRule.readonly = True
        return s.nextTokenWithinRule

    def nextTokens(self, s:ATNState, ctx:RuleContext = None):
        if ctx==None:
            return self.nextTokensNoContext(s)
        else:
            return self.nextTokensInContext(s, ctx)

    def addState(self, state:ATNState):
        if state is not None:
            state.atn = self
            state.stateNumber = len(self.states)
        self.states.append(state)

    def removeState(self, state:ATNState):
        self.states[state.stateNumber] = None # just free mem, don't shift states in list

    def defineDecisionState(self, s:DecisionState):
        self.decisionToState.append(s)
        s.decision = len(self.decisionToState)-1
        return s.decision

    def getDecisionState(self, decision:int):
        if len(self.decisionToState)==0:
            return None
        else:
            return self.decisionToState[decision]

    # Computes the set of input symbols which could follow ATN state number
    # {@code stateNumber} in the specified full {@code context}. This method
    # considers the complete parser context, but does not evaluate semantic
    # predicates (i.e. all predicates encountered during the calculation are
    # assumed true). If a path in the ATN exists from the starting state to the
    # {@link RuleStopState} of the outermost context without matching any
    # symbols, {@link Token#EOF} is added to the returned set.
    #
    # 

If {@code context} is {@code null}, it is treated as # {@link ParserRuleContext#EMPTY}.

# # @param stateNumber the ATN state number # @param context the full parse context # @return The set of potentially valid input symbols which could follow the # specified state in the specified context. # @throws IllegalArgumentException if the ATN does not contain a state with # number {@code stateNumber} #/ def getExpectedTokens(self, stateNumber:int, ctx:RuleContext ): if stateNumber < 0 or stateNumber >= len(self.states): raise Exception("Invalid state number.") s = self.states[stateNumber] following = self.nextTokens(s) if Token.EPSILON not in following: return following expected = IntervalSet() expected.addSet(following) expected.removeOne(Token.EPSILON) while (ctx != None and ctx.invokingState >= 0 and Token.EPSILON in following): invokingState = self.states[ctx.invokingState] rt = invokingState.transitions[0] following = self.nextTokens(rt.followState) expected.addSet(following) expected.removeOne(Token.EPSILON) ctx = ctx.parentCtx if Token.EPSILON in following: expected.addOne(Token.EOF) return expected




© 2015 - 2025 Weber Informatics LLC | Privacy Policy