Why is my tree grammar ambiguous? - antlr

Question

I'm a little confused. I have a grammar that works well and matches my language just as I want it to. Recently, I added a couple rules to the grammar and in the process of converting the new grammar rules to the tree grammar I am getting some strange errors. The first error I was getting was that the tree grammar was ambiguous.
The errors I received were:
[10:53:16] warning(200): ShiroDefinitionPass.g:129:17:
Decision can match input such as "'subjunctive node'" using multiple alternatives: 5, 6
As a result, alternative(s) 6 were disabled for that input
[10:53:16] warning(200): ShiroDefinitionPass.g:129:17:
Decision can match input such as "PORT_ASSIGNMENT" using multiple alternatives: 2, 6
and about 10 more similar errors.
I can't tell why the tree grammar is ambiguous. It was fine before I added the sNode, subjunctDeclNodeProd, subjunctDecl, and subjunctSelector rules.
My grammar is:
grammar Shiro;
options{
language = Java;
ASTLabelType=CommonTree;
output=AST;
}
tokens{
NEGATION;
STATE_DECL;
PORT_DECL;
PORT_INIT;
PORT_ASSIGNMENT;
PORT_TAG;
PATH;
PORT_INDEX;
EVAL_SELECT;
SUBJ_SELECT;
SUBJ_NODE_PROD;
ACTIVATION;
ACTIVATION_LIST;
PRODUCES;
}
shiro : statement+
;
statement
: nodestmt
| sNode
| graphDecl
| statestmt
| collection
| view
| NEWLINE!
;
view : 'view' IDENT mfName IDENT -> ^('view' IDENT mfName IDENT)
;
collection
: 'collection' IDENT orderingFunc path 'begin' NEWLINE
(collItem)+ NEWLINE?
'end'
-> ^('collection' IDENT orderingFunc path collItem+)
;
collItem: IDENT -> IDENT
;
orderingFunc
: IDENT -> IDENT
;
statestmt
: 'state' stateName 'begin' NEWLINE
stateHeader
'end' -> ^(STATE_DECL stateName stateHeader)
;
stateHeader
: (stateTimeStmt | stateCommentStmt | stateParentStmt | stateGraphStmt | activationPath | NEWLINE!)+
;
stateTimeStmt
: 'Time' time -> ^('Time' time)
;
stateCommentStmt
: 'Comment' comment -> ^('Comment' comment)
;
stateParentStmt
: 'Parent' stateParent -> ^('Parent' stateParent)
;
stateGraphStmt
: 'Graph' stateGraph -> ^('Graph' stateGraph)
;
stateName
: IDENT
;
time : STRING_LITERAL
;
comment : STRING_LITERAL
;
stateParent
: IDENT
;
stateGraph
: IDENT
;
activationPath
: l=activation ('.'^ (r=activation | activationList))*
;
activationList
: '<' activation (',' activation)* '>' -> ^(ACTIVATION_LIST activation+)
;
activation
: c=IDENT ('[' v=IDENT ']')? -> ^(ACTIVATION $c ($v)?)
;
graphDecl
: 'graph' IDENT 'begin' NEWLINE
graphLine+
'end'
-> ^('graph' IDENT graphLine+)
;
graphLine
: nodeProduction | portAssignment | NEWLINE!
;
nodeInternal
: (nodeProduction
| portAssignment
| portstmt
| nodestmt
| sNode
| NEWLINE!)+
;
nodestmt
: 'node'^ IDENT ('['! activeSelector ']'!)? 'begin'! NEWLINE!
nodeInternal
'end'!
;
sNode
: 'subjunctive node'^ IDENT '['! subjunctSelector ']'! 'begin'! NEWLINE!
(subjunctDeclNodeProd | subjunctDecl | NEWLINE!)+
'end'!
;
subjunctDeclNodeProd
: l=IDENT '->' r=IDENT 'begin' NEWLINE
nodeInternal
'end' -> ^(SUBJ_NODE_PROD $l $r nodeInternal )
;
subjunctDecl
: 'subjunct'^ IDENT ('['! activeSelector ']'!)? 'begin'! NEWLINE!
nodeInternal
'end'!
;
subjunctSelector
: IDENT -> ^(SUBJ_SELECT IDENT)
;
activeSelector
: IDENT -> ^(EVAL_SELECT IDENT)
;
nodeProduction
: path ('->'^ activationPath )+ NEWLINE!
;
portAssignment
: path '(' mfparams ')' NEWLINE -> ^(PORT_ASSIGNMENT path mfparams)
;
portDecl
: portType portName mfName -> ^(PORT_DECL ^(PORT_TAG portType) portName mfName)
;
portDeclInit
: portType portName mfCall -> ^(PORT_INIT ^(PORT_TAG portType) portName mfCall)
;
portstmt
: (portDecl | portDeclInit ) NEWLINE!
;
portName
: IDENT
;
portType: 'port'
| 'eval'
;
mfCall : mfName '(' mfparams ')' -> ^(mfName mfparams)
;
mfName : IDENT
;
mfparams: expression(',' expression)* -> expression+
;
// Path
path : (IDENT)('.' IDENT)*('[' pathIndex ']')? -> ^(PATH IDENT+ pathIndex?)
;
pathIndex
: portIndex -> ^(PORT_INDEX portIndex)
;
portIndex
: ( NUMBER |STRING_LITERAL )
;
// Expressions
term : path
| '(' expression ')' -> expression
| NUMBER
| STRING_LITERAL
;
unary : ('+'^ | '-'^)* term
;
mult : unary (('*'^ | '/'^ | '%'^) unary)*
;
add
: mult (( '+'^ | '-'^ ) mult)*
;
expression
: add (( '|'^ ) add)*
;
// LEXEMES
STRING_LITERAL
: '"' .* '"'
;
NUMBER : DIGIT+ ('.'DIGIT+)?
;
IDENT : (LCLETTER | UCLETTER | DIGIT)(LCLETTER | UCLETTER | DIGIT|'_')*
;
COMMENT
: '//' ~('\n'|'\r')* {$channel=HIDDEN;}
| '/*' ( options {greedy=false;} : . )* '*/' NEWLINE?{$channel=HIDDEN;}
;
WS
: (' ' | '\t' | '\f')+ {$channel = HIDDEN;}
;
NEWLINE : '\r'? '\n'
;
fragment
LCLETTER
: 'a'..'z'
;
fragment
UCLETTER: 'A'..'Z'
;
fragment
DIGIT : '0'..'9'
;
My tree grammar for the section looks like:
tree grammar ShiroDefinitionPass;
options{
tokenVocab=Shiro;
ASTLabelType=CommonTree;
}
shiro
: statement+
;
statement
: nodestmt
| sNode
| graphDecl
| statestmt
| collection
| view
;
view : ^('view' IDENT mfName IDENT)
;
collection
: ^('collection' IDENT orderingFunc path collItem+)
;
collItem: IDENT
;
orderingFunc
: IDENT
;
statestmt
: ^(STATE_DECL stateHeader)
;
stateHeader
: (stateTimeStmt | stateCommentStmt | stateParentStmt| stateGraphStmt | activation )+
;
stateTimeStmt
: ^('Time' time)
;
stateCommentStmt
: ^('Comment' comment)
;
stateParentStmt
: ^('Parent' stateParent)
;
stateGraphStmt
: ^('Graph' stateGraph)
;
stateName
: IDENT
;
time : STRING_LITERAL
;
comment : STRING_LITERAL
;
stateParent
: IDENT
;
stateGraph
: IDENT
;
activationPath
: l=activation ('.' (r=activation | activationList))*
;
activationList
: ^(ACTIVATION_LIST activation+)
;
activation
: ^(ACTIVATION IDENT IDENT?)
;
// Graph Declarations
graphDecl
: ^('graph' IDENT graphLine+)
;
graphLine
: nodeProduction
| portAssignmen
;
// End Graph declaration
nodeInternal
: (nodeProduction
|portAssignment
|portstmt
|nodestmt
|sNode )+
;
nodestmt
: ^('node' IDENT activeSelector? nodeInternal)
;
sNode
: ^('subjunctive node' IDENT subjunctSelector (subjunctDeclNodeProd | subjunctDecl)*)
;
subjunctDeclNodeProd
: ^(SUBJ_NODE_PROD IDENT IDENT nodeInternal+ )
;
subjunctDecl
: ^('subjunct' IDENT activeSelector? nodeInternal )
;
subjunctSelector
: ^(SUBJ_SELECT IDENT)
;
activeSelector returns
: ^(EVAL_SELECT IDENT)
;
nodeProduction
: ^('->' nodeProduction)
| path
;
portAssignment
: ^(PORT_ASSIGNMENT path)
;
// Port Statement
portDecl
: ^(PORT_DECL ^(PORT_TAG portType) portName mfName)
;
portDeclInit
: ^(PORT_INIT ^(PORT_TAG portType) portName mfCall)
;
portstmt
: (portDecl | portDeclInit)
;
portName
: IDENT
;
portType returns
: 'port' | 'eval'
;
mfCall
: ^(mfName mfparams)
;
mfName
: IDENT
;
mfparams
: (exps=expression)+
;
// Path
path
: ^(PATH (id=IDENT)+ (pathIndex)? )
;
pathIndex
: ^(PORT_INDEX portIndex)
;
portIndex
: ( NUMBER
|STRING_LITERAL
)
;
// Expressions
expression
: ^('+' op1=expression op2=expression)
| ^('-' op1=expression op2=expression)
| ^('*' op1=expression op2=expression)
| ^('/' op1=expression op2=expression)
| ^('%' op1=expression op2=expression)
| ^('|' op1=expression op2=expression)
| NUMBER
| path
;

Answer 1

In your tree grammar, here is the declaration of rule nodeInternal:
nodeInternal
: (nodeProduction
|portAssignment
|portstmt
|nodestmt
|sNode)+
;
And here is your declaration of rule subjunctDeclNodeProd:
subjunctDeclNodeProd
: ^(SUBJ_NODE_PROD IDENT IDENT nodeInternal+ )
;
When subjunctDeclNodeProd is being processed, ANTLR doesn't know how to process an input such as this, with two PATH children:
^(SUBJ_NODE_PROD IDENT IDENT ^(PATH IDENT) ^(PATH IDENT))
Should it follow rule nodeInternal once and process nodeProduction, nodeProduction or should it follow nodeInternal twice and process nodeProduction each time?
Consider rewriting subjunctDeclNodeProd without the +:
subjunctDeclNodeProd
: ^(SUBJ_NODE_PROD IDENT IDENT nodeInternal)
;
I think that will take care of the problem.