org.antlr.v4.parse.ANTLRParser.g Maven / Gradle / Ivy
/*
* Copyright (c) 2012 The ANTLR Project. All rights reserved.
* Use of this file is governed by the BSD-3-Clause license that
* can be found in the LICENSE.txt file in the project root.
*/
/** The definitive ANTLR v3 grammar to parse ANTLR v4 grammars.
* The grammar builds ASTs that are sniffed by subsequent stages.
*/
parser grammar ANTLRParser;
options {
// Target language is Java, which is the default but being specific
// here as this grammar is also meant as a good example grammar for
// for users.
language = Java;
// The output of this grammar is going to be an AST upon which
// we run a semantic checking phase, then the rest of the analysis
// including final code generation.
output = AST;
// The vocabulary (tokens and their int token types) we are using
// for the parser. This is generated by the lexer. The vocab will be extended
// to include the imaginary tokens below.
tokenVocab = ANTLRLexer;
ASTLabelType = GrammarAST;
}
// Imaginary Tokens
//
// Imaginary tokens do not exist as far as the lexer is concerned, and it cannot
// generate them. However we sometimes need additional 'tokens' to use as root
// nodes for the AST we are generating. The tokens section is where we
// specify any such tokens
tokens {
RULE;
PREC_RULE; // flip to this if we find that it's left-recursive
RULES;
RULEMODIFIERS;
RULEACTIONS;
BLOCK;
OPTIONAL;
CLOSURE;
POSITIVE_CLOSURE;
RANGE;
SET;
CHAR_RANGE;
EPSILON;
ALT;
ALTLIST;
ID;
ARG;
ARGLIST;
RET;
COMBINED;
INITACTION;
LABEL; // $x used in rewrite rules
TEMPLATE;
WILDCARD;
// A generic node indicating a list of something when we don't
// really need to distinguish what we have a list of as the AST
// will 'kinow' by context.
//
LIST;
ELEMENT_OPTIONS; // TOKEN
RESULT;
// lexer action stuff
LEXER_ALT_ACTION;
LEXER_ACTION_CALL; // ID(foo)
}
// Include the copyright in this source and also the generated source
//
@header {
/*
* Copyright (c) 2012 The ANTLR Project. All rights reserved.
* Use of this file is governed by the BSD-3-Clause license that
* can be found in the LICENSE.txt file in the project root.
*/
package org.antlr.v4.parse;
import org.antlr.v4.tool.*;
import org.antlr.v4.tool.ast.*;
import java.util.ArrayDeque;
import java.util.Deque;
}
@members {
Deque paraphrases = new ArrayDeque();
public void grammarError(ErrorType etype, org.antlr.runtime.Token token, Object... args) { }
}
// The main entry point for parsing a V3 grammar from top to toe. This is
// the method call from whence to obtain the AST for the parse.
//
grammarSpec
@after {
GrammarAST options = (GrammarAST)$tree.getFirstChildWithType(ANTLRParser.OPTIONS);
if ( options!=null ) {
Grammar.setNodeOptions($tree, options);
}
}
: // First we should see the type and name of the grammar file that
// we are about to parse.
//
grammarType id SEMI
// There now follows zero or more declaration sections that should
// be given to us before the rules are declared
//
// A number of things can be declared/stated before the grammar rules
// 'proper' are parsed. These include grammar imports (delegate), grammar
// options, imaginary token declarations, global scope declarations,
// and actions such as @header. In this rule we allow any number of
// these constructs in any order so that the grammar author is not
// constrained by some arbitrary order of declarations that nobody
// can remember. In the next phase of the parse, we verify that these
// constructs are valid, not repeated and so on.
sync ( prequelConstruct sync )*
// We should now see at least one ANTLR EBNF style rule
// declaration. If the rules are missing we will let the
// semantic verification phase tell the user about it.
//
rules
modeSpec*
// And we force ANTLR to process everything it finds in the input
// stream by specifying hte need to match End Of File before the
// parse is complete.
//
EOF
// Having parsed everything in the file and accumulated the relevant
// subtrees, we can now rewrite everything into the main AST form
// that our tree walkers are expecting.
//
-> ^(grammarType // The grammar type is our root AST node
id // We need to identify the grammar of course
prequelConstruct* // The set of declarations we accumulated
rules // And of course, we need the set of rules we discovered
modeSpec*
)
;
grammarType
@after {
if ( $tg!=null ) throw new v3TreeGrammarException(tg);
if ( $t!=null ) ((GrammarRootAST)$tree).grammarType = $t.type;
else ((GrammarRootAST)$tree).grammarType=COMBINED;
}
: ( t=LEXER g=GRAMMAR -> GRAMMAR[$g, "LEXER_GRAMMAR", getTokenStream()]
| // A standalone parser specification
t=PARSER g=GRAMMAR -> GRAMMAR[$g, "PARSER_GRAMMAR", getTokenStream()]
// A combined lexer and parser specification
| g=GRAMMAR -> GRAMMAR[$g, "COMBINED_GRAMMAR", getTokenStream()]
| tg=TREE_GRAMMAR
)
;
// This is the list of all constructs that can be declared before
// the set of rules that compose the grammar, and is invoked 0..n
// times by the grammarPrequel rule.
prequelConstruct
: // A list of options that affect analysis and/or code generation
optionsSpec
| // A list of grammars to which this grammar will delegate certain
// parts of the parsing sequence - a set of imported grammars
delegateGrammars
| // The declaration of any token types we need that are not already
// specified by a preceeding grammar, such as when a parser declares
// imaginary tokens with which to construct the AST, or a rewriting
// tree parser adds further imaginary tokens to ones defined in a prior
// {tree} parser.
tokensSpec
| // A list of custom channels used by the grammar
channelsSpec
| // A declaration of language target implemented constructs. All such
// action sections start with '@' and are given to the language target's
// StringTemplate group. For instance @parser::header and @lexer::header
// are gathered here.
action
;
// A list of options that affect analysis and/or code generation
optionsSpec
: OPTIONS (option SEMI)* RBRACE -> ^(OPTIONS[$OPTIONS, "OPTIONS"] option*)
;
option
: id ASSIGN^ optionValue
;
// ------------
// Option Value
//
// The actual value of an option - Doh!
//
optionValue
: // If the option value is a single word that conforms to the
// lexical rules of token or rule names, then the user may skip quotes
// and so on. Many option values meet this description
qid
| STRING_LITERAL
| ACTION
| INT
;
// A list of grammars to which this grammar will delegate certain
// parts of the parsing sequence - a set of imported grammars
delegateGrammars
: IMPORT delegateGrammar (COMMA delegateGrammar)* SEMI -> ^(IMPORT delegateGrammar+)
;
// A possibly named grammar file that should be imported to this grammar
// and delgated to for the rules it specifies
delegateGrammar
: id ASSIGN^ id
| id
;
tokensSpec
: TOKENS_SPEC id (COMMA id)* COMMA? RBRACE -> ^(TOKENS_SPEC id+)
| TOKENS_SPEC RBRACE ->
| TOKENS_SPEC^ v3tokenSpec+ RBRACE!
{grammarError(ErrorType.V3_TOKENS_SYNTAX, $TOKENS_SPEC);}
;
v3tokenSpec
: id
( ASSIGN lit=STRING_LITERAL
{
grammarError(ErrorType.V3_ASSIGN_IN_TOKENS, $id.start,
$id.text, $lit.getText());
}
-> id // ignore assignment
| -> id
)
SEMI
;
channelsSpec
: CHANNELS^ (id (COMMA! id)* COMMA?)? RBRACE!
;
// A declaration of a language target specifc section,
// such as @header, @includes and so on. We do not verify these
// sections, they are just passed on to the language target.
/** Match stuff like @parser::members {int i;} */
action
: AT (actionScopeName COLONCOLON)? id ACTION -> ^(AT actionScopeName? id ACTION)
;
/** Sometimes the scope names will collide with keywords; allow them as
* ids for action scopes.
*/
actionScopeName
: id
| LEXER -> ID[$LEXER]
| PARSER -> ID[$PARSER]
;
modeSpec
: MODE id SEMI sync (lexerRule sync)* -> ^(MODE id lexerRule*)
;
rules
: sync (rule sync)*
// Rewrite with an enclosing node as this is good for counting
// the number of rules and an easy marker for the walker to detect
// that there are no rules.
->^(RULES rule*)
;
sync
@init {
BitSet followSet = computeErrorRecoverySet();
if ( input.LA(1)!=Token.EOF && !followSet.member(input.LA(1)) ) {
reportError(new NoViableAltException("",0,0,input));
beginResync();
consumeUntil(input, followSet);
endResync();
}
} :
;
rule: parserRule
| lexerRule
;
// The specification of an EBNF rule in ANTLR style, with all the
// rule level parameters, declarations, actions, rewrite specs and so
// on.
//
// Note that here we allow any number of rule declaration sections (such
// as scope, returns, etc) in any order and we let the upcoming semantic
// verification of the AST determine if things are repeated or if a
// particular functional element is not valid in the context of the
// grammar type, such as using returns in lexer rules and so on.
parserRule
@init { paraphrases.push("matching a rule"); }
@after {
paraphrases.pop();
GrammarAST options = (GrammarAST)$tree.getFirstChildWithType(ANTLRParser.OPTIONS);
if ( options!=null ) {
Grammar.setNodeOptions($tree, options);
}
}
: // Start with the rule name. Here we do not distinguish between
// parser or lexer rules, the semantic verification phase will
// reject any rules that make no sense, such as lexer rules in
// a pure parser or tree parser.
RULE_REF
// Immediately following the rulename, there may be a specification
// of input parameters for the rule. We do not do anything with the
// parameters here except gather them for future phases such as
// semantic verifcation, type assignment etc. We require that
// the input parameters are the next syntactically significant element
// following the rule id.
ARG_ACTION?
ruleReturns?
throwsSpec?
localsSpec?
// Now, before the rule specification itself, which is introduced
// with a COLON, we may have zero or more configuration sections.
// As usual we just accept anything that is syntactically valid for
// one form of the rule or another and let the semantic verification
// phase throw out anything that is invalid.
// At the rule level, a programmer may specify a number of sections, such
// as scope declarations, rule return elements, @ sections (which may be
// language target specific) and so on. We allow any number of these in any
// order here and as usual rely onthe semantic verification phase to reject
// anything invalid using its addinotal context information. Here we are
// context free and just accept anything that is a syntactically correct
// construct.
//
rulePrequels
COLON
// The rule is, at the top level, just a list of alts, with
// finer grained structure defined within the alts.
ruleBlock
SEMI
exceptionGroup
-> ^( RULE RULE_REF ARG_ACTION?
ruleReturns? throwsSpec? localsSpec? rulePrequels? ruleBlock exceptionGroup*
)
;
// Many language targets support exceptions and the rule will
// generally be able to throw the language target equivalent
// of a recognition exception. The grammar programmar can
// specify a list of exceptions to catch or a generic catch all
// and the target language code generation template is
// responsible for generating code that makes sense.
exceptionGroup
: exceptionHandler* finallyClause?
;
// Specifies a handler for a particular type of exception
// thrown by a rule
exceptionHandler
: CATCH ARG_ACTION ACTION -> ^(CATCH ARG_ACTION ACTION)
;
finallyClause
: FINALLY ACTION -> ^(FINALLY ACTION)
;
rulePrequels
@init { paraphrases.push("matching rule preamble"); }
@after { paraphrases.pop(); }
: sync (rulePrequel sync)* -> rulePrequel*
;
// An individual rule level configuration as referenced by the ruleActions
// rule above.
//
rulePrequel
: optionsSpec
| ruleAction
;
// A rule can return elements that it constructs as it executes.
// The return values are specified in a 'returns' prequel element,
// which contains COMMA separated declarations, where the declaration
// is target language specific. Here we see the returns declaration
// as a single lexical action element, to be processed later.
//
ruleReturns
: RETURNS^ ARG_ACTION
;
// --------------
// Exception spec
//
// Some target languages, such as Java and C# support exceptions
// and they are specified as a prequel element for each rule that
// wishes to throw its own exception type. Note that the name of the
// exception is just a single word, so the header section of the grammar
// must specify the correct import statements (or language equivalent).
// Target languages that do not support exceptions just safely ignore
// them.
//
throwsSpec
: THROWS qid (COMMA qid)* -> ^(THROWS qid+)
;
// locals [Cat x, float g]
localsSpec : LOCALS^ ARG_ACTION ;
// @ Sections are generally target language specific things
// such as local variable declarations, code to run before the
// rule starts and so on. Fir instance most targets support the
// @init {} section where declarations and code can be placed
// to run before the rule is entered. The C target also has
// an @declarations {} section, where local variables are declared
// in order that the generated code is C89 copmliant.
//
/** Match stuff like @init {int i;} */
ruleAction
: AT id ACTION -> ^(AT id ACTION)
;
// A set of alts, rewritten as a BLOCK for generic processing
// in tree walkers. Used by the rule 'rule' so that the list of
// alts for a rule appears as a BLOCK containing the alts and
// can be processed by the generic BLOCK rule. Note that we
// use a separate rule so that the BLOCK node has start and stop
// boundaries set correctly by rule post processing of rewrites.
ruleBlock
@init {Token colon = input.LT(-1);}
: ruleAltList -> ^(BLOCK[colon,"BLOCK"] ruleAltList)
;
catch [ResyncToEndOfRuleBlock e] {
// just resyncing; ignore error
retval.tree = (GrammarAST)adaptor.errorNode(input, retval.start, input.LT(-1), null);
}
ruleAltList
: labeledAlt (OR labeledAlt)* -> labeledAlt+
;
labeledAlt
: alternative
( POUND! id! {((AltAST)$alternative.tree).altLabel=$id.tree;}
)?
;
lexerRule
@init { paraphrases.push("matching a lexer rule"); }
@after {
paraphrases.pop();
}
: FRAGMENT?
TOKEN_REF COLON lexerRuleBlock SEMI
-> ^( RULE TOKEN_REF
^(RULEMODIFIERS FRAGMENT)? lexerRuleBlock
)
;
lexerRuleBlock
@init {Token colon = input.LT(-1);}
: lexerAltList -> ^(BLOCK[colon,"BLOCK"] lexerAltList)
;
catch [ResyncToEndOfRuleBlock e] {
// just resyncing; ignore error
retval.tree = (GrammarAST)adaptor.errorNode(input, retval.start, input.LT(-1), null);
}
lexerAltList
: lexerAlt (OR lexerAlt)* -> lexerAlt+
;
lexerAlt
: lexerElements
( lexerCommands -> ^(LEXER_ALT_ACTION lexerElements lexerCommands)
| -> lexerElements
)
;
lexerElements
: lexerElement+ -> ^(ALT lexerElement+)
| -> ^(ALT EPSILON) // empty alt
;
lexerElement
@init {
paraphrases.push("looking for lexer rule element");
int m = input.mark();
}
@after { paraphrases.pop(); }
: labeledLexerElement
( ebnfSuffix -> ^( ebnfSuffix ^(BLOCK[$labeledLexerElement.start,"BLOCK"] ^(ALT labeledLexerElement) ) )
| -> labeledLexerElement
)
| lexerAtom
( ebnfSuffix -> ^( ebnfSuffix ^(BLOCK[$lexerAtom.start,"BLOCK"] ^(ALT lexerAtom) ) )
| -> lexerAtom
)
| lexerBlock
( ebnfSuffix -> ^(ebnfSuffix lexerBlock)
| -> lexerBlock
)
| actionElement // actions only allowed at end of outer alt actually,
// but preds can be anywhere
;
catch [RecognitionException re] {
retval.tree = (GrammarAST)adaptor.errorNode(input, retval.start, input.LT(-1), re);
int ttype = input.get(input.range()).getType(); // seems to be next token
// look for anything that really belongs at the start of the rule minus the initial ID
if ( ttype==COLON || ttype==RETURNS || ttype==CATCH || ttype==FINALLY || ttype==AT || ttype==EOF ) {
RecognitionException missingSemi =
new v4ParserException("unterminated rule (missing ';') detected at '"+
input.LT(1).getText()+" "+input.LT(2).getText()+"'", input);
reportError(missingSemi);
if ( ttype==EOF ) {
input.seek(input.index()+1);
}
else if ( ttype==CATCH || ttype==FINALLY ) {
input.seek(input.range()); // ignore what's before rule trailer stuff
}
else if ( ttype==RETURNS || ttype==AT ) { // scan back looking for ID of rule header
int p = input.index();
Token t = input.get(p);
while ( t.getType()!=RULE_REF && t.getType()!=TOKEN_REF ) {
p--;
t = input.get(p);
}
input.seek(p);
}
throw new ResyncToEndOfRuleBlock(); // make sure it goes back to rule block level to recover
}
reportError(re);
recover(input,re);
}
labeledLexerElement
: id (ass=ASSIGN|ass=PLUS_ASSIGN)
( lexerAtom -> ^($ass id lexerAtom)
| lexerBlock -> ^($ass id lexerBlock)
)
;
lexerBlock
@after {
GrammarAST options = (GrammarAST)$tree.getFirstChildWithType(ANTLRParser.OPTIONS);
if ( options!=null ) {
Grammar.setNodeOptions($tree, options);
}
}
: LPAREN
( optionsSpec COLON )?
lexerAltList
RPAREN
-> ^(BLOCK[$LPAREN,"BLOCK"] optionsSpec? lexerAltList )
;
// channel=HIDDEN, skip, more, mode(INSIDE), push(INSIDE), pop
lexerCommands
: RARROW lexerCommand (COMMA lexerCommand)* -> lexerCommand+
;
lexerCommand
: lexerCommandName LPAREN lexerCommandExpr RPAREN -> ^(LEXER_ACTION_CALL lexerCommandName lexerCommandExpr)
| lexerCommandName
;
lexerCommandExpr
: id
| INT
;
lexerCommandName
: id
| MODE ->ID[$MODE]
;
altList
: alternative (OR alternative)* -> alternative+
;
// An individual alt with an optional alt option like
alternative
@init { paraphrases.push("matching alternative"); }
@after {
paraphrases.pop();
Grammar.setNodeOptions($tree, $o.tree);
}
: o=elementOptions?
( e+=element+ -> ^(ALT elementOptions? $e+)
| -> ^(ALT elementOptions? EPSILON) // empty alt
)
;
element
@init {
paraphrases.push("looking for rule element");
int m = input.mark();
}
@after { paraphrases.pop(); }
: labeledElement
( ebnfSuffix -> ^( ebnfSuffix ^(BLOCK[$labeledElement.start,"BLOCK"] ^(ALT labeledElement ) ))
| -> labeledElement
)
| atom
( ebnfSuffix -> ^( ebnfSuffix ^(BLOCK[$atom.start,"BLOCK"] ^(ALT atom) ) )
| -> atom
)
| ebnf
| actionElement
;
catch [RecognitionException re] {
retval.tree = (GrammarAST)adaptor.errorNode(input, retval.start, input.LT(-1), re);
int ttype = input.get(input.range()).getType();
// look for anything that really belongs at the start of the rule minus the initial ID
if ( ttype==COLON || ttype==RETURNS || ttype==CATCH || ttype==FINALLY || ttype==AT ) {
RecognitionException missingSemi =
new v4ParserException("unterminated rule (missing ';') detected at '"+
input.LT(1).getText()+" "+input.LT(2).getText()+"'", input);
reportError(missingSemi);
if ( ttype==CATCH || ttype==FINALLY ) {
input.seek(input.range()); // ignore what's before rule trailer stuff
}
if ( ttype==RETURNS || ttype==AT ) { // scan back looking for ID of rule header
int p = input.index();
Token t = input.get(p);
while ( t.getType()!=RULE_REF && t.getType()!=TOKEN_REF ) {
p--;
t = input.get(p);
}
input.seek(p);
}
throw new ResyncToEndOfRuleBlock(); // make sure it goes back to rule block level to recover
}
reportError(re);
recover(input,re);
}
actionElement
@after {
GrammarAST options = (GrammarAST)$tree.getFirstChildWithType(ANTLRParser.ELEMENT_OPTIONS);
if ( options!=null ) {
Grammar.setNodeOptions($tree, options);
}
}
: ACTION
| ACTION elementOptions -> ^(ACTION elementOptions)
| SEMPRED
| SEMPRED elementOptions -> ^(SEMPRED elementOptions)
;
labeledElement
: id (ass=ASSIGN|ass=PLUS_ASSIGN)
( atom -> ^($ass id atom)
| block -> ^($ass id block)
)
;
// A block of grammar structure optionally followed by standard EBNF
// notation, or ANTLR specific notation. I.E. ? + ^ and so on
ebnf
: block
// And now we see if we have any of the optional suffixs and rewrite
// the AST for this rule accordingly
( blockSuffix -> ^(blockSuffix block)
| -> block
)
;
// The standard EBNF suffixes with additional components that make
// sense only to ANTLR, in the context of a grammar block.
blockSuffix
: ebnfSuffix // Standard EBNF
;
ebnfSuffix
: QUESTION nongreedy=QUESTION? -> OPTIONAL[$start, $nongreedy]
| STAR nongreedy=QUESTION? -> CLOSURE[$start, $nongreedy]
| PLUS nongreedy=QUESTION? -> POSITIVE_CLOSURE[$start, $nongreedy]
;
lexerAtom
: range
| terminal
| RULE_REF
| notSet
| wildcard
| LEXER_CHAR_SET
;
atom
: // Qualified reference delegate.rule. This must be
// lexically contiguous (no spaces either side of the DOT)
// otherwise it is two references with a wildcard in between
// and not a qualified reference.
/*
{
input.LT(1).getCharPositionInLine()+input.LT(1).getText().length()==
input.LT(2).getCharPositionInLine() &&
input.LT(2).getCharPositionInLine()+1==input.LT(3).getCharPositionInLine()
}?
id DOT ruleref -> ^(DOT id ruleref)
|
*/
range // Range x..y - only valid in lexers
| terminal
| ruleref
| notSet
| wildcard
;
catch [RecognitionException re] { throw re; } // pass upwards to element
wildcard
@after {
GrammarAST options = (GrammarAST)$tree.getFirstChildWithType(ANTLRParser.ELEMENT_OPTIONS);
if ( options!=null ) {
Grammar.setNodeOptions($tree, options);
}
}
: // Wildcard '.' means any character in a lexer, any
// token in parser and any node or subtree in a tree parser
// Because the terminal rule is allowed to be the node
// specification for the start of a tree rule, we must
// later check that wildcard was not used for that.
DOT elementOptions?
-> ^(WILDCARD[$DOT] elementOptions?)
;
// --------------------
// Inverted element set
//
// A set of characters (in a lexer) or terminal tokens, if a parser,
// that are then used to create the inverse set of them.
notSet
: NOT setElement -> ^(NOT[$NOT] ^(SET[$setElement.start,"SET"] setElement))
| NOT blockSet -> ^(NOT[$NOT] blockSet)
;
blockSet
@init {
Token t;
boolean ebnf = false;
}
: LPAREN setElement (OR setElement)* RPAREN
-> ^(SET[$LPAREN,"SET"] setElement+ )
;
setElement
: TOKEN_REF^ elementOptions?
| STRING_LITERAL^ elementOptions?
| range
| LEXER_CHAR_SET
;
// -------------
// Grammar Block
//
// Anywhere where an element is valid, the grammar may start a new block
// of alts by surrounding that block with ( ). A new block may also have a set
// of options, which apply only to that block.
//
block
@after {
GrammarAST options = (GrammarAST)$tree.getFirstChildWithType(ANTLRParser.OPTIONS);
if ( options!=null ) {
Grammar.setNodeOptions($tree, options);
}
}
: LPAREN
( optionsSpec? ra+=ruleAction* COLON )?
altList
RPAREN
-> ^(BLOCK[$LPAREN,"BLOCK"] optionsSpec? $ra* altList )
;
// ----------------
// Parser rule ref
//
// Reference to a parser rule with optional arguments and optional
// directive to become the root node or ignore the tree produced
//
ruleref
@after {
GrammarAST options = (GrammarAST)$tree.getFirstChildWithType(ANTLRParser.ELEMENT_OPTIONS);
if ( options!=null ) {
Grammar.setNodeOptions($tree, options);
}
}
: RULE_REF ARG_ACTION? elementOptions? -> ^(RULE_REF ARG_ACTION? elementOptions?)
;
catch [RecognitionException re] { throw re; } // pass upwards to element
// ---------------
// Character Range
//
// Specifies a range of characters. Valid for lexer rules only, but
// we do not check that here, the tree walkers shoudl do that.
// Note also that the parser also allows through more than just
// character literals so that we can produce a much nicer semantic
// error about any abuse of the .. operator.
//
range
: STRING_LITERAL RANGE^ STRING_LITERAL
;
terminal
@after {
GrammarAST options = (GrammarAST)$tree.getFirstChildWithType(ANTLRParser.ELEMENT_OPTIONS);
if ( options!=null ) {
Grammar.setNodeOptions($tree, options);
}
}
: TOKEN_REF elementOptions? -> ^(TOKEN_REF elementOptions?)
| STRING_LITERAL elementOptions? -> ^(STRING_LITERAL elementOptions?)
;
// Terminals may be adorned with certain options when
// reference in the grammar: TOK<,,,>
elementOptions
: LT (elementOption (COMMA elementOption)*)? GT
-> ^(ELEMENT_OPTIONS[$LT,"ELEMENT_OPTIONS"] elementOption*)
;
// When used with elements we can specify what the tree node type can
// be and also assign settings of various options (which we do not check here)
elementOption
: // This format indicates the default element option
qid
| id ASSIGN^ optionValue
;
// The name of the grammar, and indeed some other grammar elements may
// come through to the parser looking like a rule reference or a token
// reference, hence this rule is used to pick up whichever it is and rewrite
// it as a generic ID token.
id
@init { paraphrases.push("looking for an identifier"); }
@after { paraphrases.pop(); }
: RULE_REF ->ID[$RULE_REF]
| TOKEN_REF ->ID[$TOKEN_REF]
;
qid
@init { paraphrases.push("looking for a qualified identifier"); }
@after { paraphrases.pop(); }
: id (DOT id)* -> ID[$qid.start, $text]
;
alternativeEntry : alternative EOF ; // allow gunit to call alternative and see EOF afterwards
elementEntry : element EOF ;
ruleEntry : rule EOF ;
blockEntry : block EOF ;
© 2015 - 2025 Weber Informatics LLC | Privacy Policy