public static IDictionary<int, Interval> GetStateToGrammarRegionMap(GrammarRootAST ast, IntervalSet grammarTokenTypes)
{
IDictionary<int, Interval> stateToGrammarRegionMap = new Dictionary<int, Interval>();
if (ast == null)
return stateToGrammarRegionMap;
IList<GrammarAST> nodes = ast.GetNodesWithType(grammarTokenTypes);
foreach (GrammarAST n in nodes)
{
if (n.atnState != null)
{
Interval tokenRegion = Interval.Of(n.TokenStartIndex, n.TokenStopIndex);
Antlr.Runtime.Tree.ITree ruleNode = null;
// RULEs, BLOCKs of transformed recursive rules point to original token interval
switch (n.Type)
{
case ANTLRParser.RULE:
ruleNode = n;
break;
case ANTLRParser.BLOCK:
case ANTLRParser.CLOSURE:
ruleNode = n.GetAncestor(ANTLRParser.RULE);
break;
}
if (ruleNode is RuleAST)
{
string ruleName = ((RuleAST)ruleNode).GetRuleName();
Rule r = ast.g.GetRule(ruleName);
if (r is LeftRecursiveRule)
{
RuleAST originalAST = ((LeftRecursiveRule)r).GetOriginalAST();
tokenRegion = Interval.Of(originalAST.TokenStartIndex, originalAST.TokenStopIndex);
}
}
stateToGrammarRegionMap[n.atnState.stateNumber] = tokenRegion;
}
}
return stateToGrammarRegionMap;
}
/** Return list of (TOKEN_NAME node, 'literal' node) pairs */
public static IList<System.Tuple<GrammarAST, GrammarAST>> GetStringLiteralAliasesFromLexerRules(GrammarRootAST ast)
{
string[] patterns = {
"(RULE %name:TOKEN_REF (BLOCK (ALT %lit:STRING_LITERAL)))",
"(RULE %name:TOKEN_REF (BLOCK (ALT %lit:STRING_LITERAL ACTION)))",
"(RULE %name:TOKEN_REF (BLOCK (ALT %lit:STRING_LITERAL SEMPRED)))",
"(RULE %name:TOKEN_REF (BLOCK (LEXER_ALT_ACTION (ALT %lit:STRING_LITERAL) .)))",
"(RULE %name:TOKEN_REF (BLOCK (LEXER_ALT_ACTION (ALT %lit:STRING_LITERAL) . .)))",
"(RULE %name:TOKEN_REF (BLOCK (LEXER_ALT_ACTION (ALT %lit:STRING_LITERAL) (LEXER_ACTION_CALL . .))))",
"(RULE %name:TOKEN_REF (BLOCK (LEXER_ALT_ACTION (ALT %lit:STRING_LITERAL) . (LEXER_ACTION_CALL . .))))",
"(RULE %name:TOKEN_REF (BLOCK (LEXER_ALT_ACTION (ALT %lit:STRING_LITERAL) (LEXER_ACTION_CALL . .) .)))",
// TODO: allow doc comment in there
};
GrammarASTAdaptor adaptor = new GrammarASTAdaptor(ast.Token.InputStream);
Antlr.Runtime.Tree.TreeWizard wiz = new Antlr.Runtime.Tree.TreeWizard(adaptor, ANTLRParser.tokenNames);
IList<System.Tuple<GrammarAST, GrammarAST>> lexerRuleToStringLiteral =
new List<System.Tuple<GrammarAST, GrammarAST>>();
IList<GrammarAST> ruleNodes = ast.GetNodesWithType(ANTLRParser.RULE);
if (ruleNodes == null || ruleNodes.Count == 0)
return null;
foreach (GrammarAST r in ruleNodes)
{
//tool.log("grammar", r.toStringTree());
// System.out.println("chk: "+r.toStringTree());
Antlr.Runtime.Tree.ITree name = r.GetChild(0);
if (name.Type == ANTLRParser.TOKEN_REF)
{
// check rule against patterns
bool isLitRule;
foreach (string pattern in patterns)
{
isLitRule =
DefAlias(r, pattern, wiz, lexerRuleToStringLiteral);
if (isLitRule)
break;
}
// if ( !isLitRule ) System.out.println("no pattern matched");
}
}
return lexerRuleToStringLiteral;
}