Is C++ code generation in ANTLR 3.2 ready?

Question

Hi, I was trying hard to make ANTLR 3.2 generate parser/lexer in C++. It was fruitless. Things went well with Java & C though.

I was using this tutorial to get started: http://www.ibm.com/developerworks/aix/library/au-c%5Fplusplus%5Fantlr/index.html

When I checked the *.stg files, I found that:

CPP has only

./tool/src/main/resources/org/antlr/codegen/templates/CPP/CPP.stg

C has so many files:

./tool/src/main/resources/org/antlr/codegen/templates/C/AST.stg
./tool/src/main/resources/org/antlr/codegen/templates/C/ASTDbg.stg
./tool/src/main/resources/org/antlr/codegen/templates/C/ASTParser.stg
./tool/src/main/resources/org/antlr/codegen/templates/C/ASTTreeParser.stg
./tool/src/main/resources/org/antlr/codegen/templates/C/C.stg
./tool/src/main/resources/org/antlr/codegen/templates/C/Dbg.stg

And so other languages.

My C.g file:

grammar C;

options { language='CPP'; }

/** Match things like "call foo;" */
r : 'call' ID ';' {System.out.println("invoke "+$ID.text);} ;
ID: ('a'..'z'|'A'..'Z'|'_')('0'..'9'|'a'..'z'|'A'..'Z'|'_')* ;
WS: (' ' |'\n' |'\r' )+ {$channel=HIDDEN;} ; // ignore whitespace

Errors:

error(10):  internal error: group Cpp does not satisfy interface ANTLRCore: missing templates [lexerRuleRefAndListLabel, parameterSetAttributeRef, scopeSetAttributeRef, returnSetAttributeRef, lexerRulePropertyRef_text, lexerRulePropertyRef_type, lexerRulePropertyRef_line, lexerRulePropertyRef_pos, lexerRulePropertyRef_index, lexerRulePropertyRef_channel, lexerRulePropertyRef_start, lexerRulePropertyRef_stop, ruleSetPropertyRef_tree, ruleSetPropertyRef_st] 

error(10):  internal error: group Cpp does not satisfy interface ANTLRCore: mismatched arguments on these templates [outputFile(LEXER, PARSER, TREE_PARSER, actionScope, actions, docComment, recognizer, name, tokens, tokenNames, rules, cyclicDFAs, bitsets, buildTemplate, buildAST, rewriteMode, profile, backtracking, synpreds, memoize, numRules, fileName, ANTLRVersion, generatedTimestamp, trace, scopes, superClass, literals), optional headerFile(LEXER, PARSER, TREE_PARSER, actionScope, actions, docComment, recognizer, name, tokens, tokenNames, rules, cyclicDFAs, bitsets, buildTemplate, buildAST, rewriteMode, profile, backtracking, synpreds, memoize, numRules, fileName, ANTLRVersion, generatedTimestamp, trace, scopes, superClass, literals), lexer(grammar, name, tokens, scopes, rules, numRules, labelType, filterMode, superClass), rule(ruleName, ruleDescriptor, block, emptyRule, description, exceptions, finally, memoize), alt(elements, altNum, description, autoAST, outerAlt, treeLevel, rew), tokenRef(token, label, elementIndex, hetero), tokenRefAndListLabel(token, label, elementIndex, hetero), listLabel(label, elem), charRangeRef(a, b, label), ruleRef(rule, label, elementIndex, args, scope), ruleRefAndListLabel(rule, label, elementIndex, args, scope), lexerRuleRef(rule, label, args, elementIndex, scope), lexerMatchEOF(label, elementIndex), tree(root, actionsAfterRoot, children, nullableChildList, enclosingTreeLevel, treeLevel)] 

error(10):  internal error: C.g : java.lang.IllegalArgumentException: Can't find template actionGate.st; group hierarchy is [Cpp]

... and so on.

Please kindly advise. Thank you! I'm using Leopard 10.5.8 with

CLASSPATH=:/Users/vietlq/projects/antlr-3.2.jar:/Users/vietlq/projects/stringtemplate-3.2.1/lib/stringtemplate-3.2.1.jar:/Users/vietlq/projects/stringtemplate-3.2.1/lib/antlr-2.7.7.jar

Answer 1

It sounds like you've answered your own question: ANTLR's C++ lexer/parser generators are not yet functional.

For what it's worth, it's still possible to use ANTLR for parsing from C++, via the C target. I use ANTLR to generate a C language lexer and parser, which I then compile and link to my C++ code.

I have one C++ file that translates an ANTLR parse tree to my target abstract syntax tree classes, and the rest of my code doesn't care where the AST comes from. It works pretty well in practice! It would be easy to replace ANTLR with a different parser generator, and I find that the separation leads to cleaner ANTLR grammars.