[antlr-interest] funny things with the generated C target

[Davyd Madeley]

Hi all,

I'm just getting started with using ANTLR. I wrote a grammar and knocked
together a quick tree walker in Python and everything looked good. Now
I'm looking to use the C target to build upon this parser.

I changed the language to C, generated and compiled the grammar. My
first problem was that I was unable to have a lexer rule named LT (?),
so I renamed it. Now it's segfaulting inside the generated code...

(gdb) r ../antlr-parse-1.def 
Starting program: /home/davyd/src/uniseis2/c/test ../antlr-parse-1.def

Program received signal SIGSEGV, Segmentation fault.
0x00fd6e42 in fillBuffer () from /home/davyd/install/lib/libantlr3c.so
(gdb) bt
#0  0x00fd6e42 in fillBuffer ()
from /home/davyd/install/lib/libantlr3c.so
#1  0x00fd721a in tokLT () from /home/davyd/install/lib/libantlr3c.so
#2  0x0804ac04 in phaseBlock (ctx=0x8fc19e8) at UniseisPhaseParser.c:527
#3  0x080489a0 in main (argc=1457883477, argv=0xb0ec8153)
    at SemanticChecker.c:56

I've attached my file, SemanticChecker.c, which currently does nothing
more than load and parse the file. I've also attached the grammar.

What might I be doing wrong?

Regards,
--davyd

-- 
Davyd Madeley        Software Engineer
Fugro Seismic Imaging, Perth Australia
-------------- next part --------------
A non-text attachment was scrubbed...
Name: SemanticChecker.c
Type: text/x-csrc
Size: 1296 bytes
Desc: not available
Url : http://www.antlr.org/pipermail/antlr-interest/attachments/20080717/79a3f310/attachment-0001.bin 
-------------- next part --------------
grammar UniseisPhase;

options
{
	output=AST;
	language=C;
}

tokens
{
	PHASE;
	CARD;
	PARAMETER;
	MANUAL;
}

// the definition for a phase block
//     phase NAME { properties }
// phases can accept
//    - properties
//    - manual entries
//    - card definitions
phaseBlock
	: 'phase' UNISEIS_NAME LINE_TERMINATOR? '{' LINE_TERMINATOR* (phaseDefinition LINE_TERMINATOR*)* '}'
	-> ^(PHASE UNISEIS_NAME phaseDefinition*)
	;

phaseDefinition
	: propertyAssignment
	| cardBlock
	| MANUAL_XML
	;

// the definition for a card block
//     card NAME { properties }
// cards can accept:
//    - properties
//    - manual entries
//    - parameter definitions
cardBlock
	: 'card' UNISEIS_NAME LINE_TERMINATOR? '{' LINE_TERMINATOR* (cardDefinition LINE_TERMINATOR*)* '}'
	-> ^(CARD UNISEIS_NAME cardDefinition*)
	;

cardDefinition
	: propertyAssignment
	| parameterBlock
	| MANUAL_XML
	;

// the definition for a parameter block:
//     parameter NAME { properties }
// parameters can accept:
//    - properties
//    - manual entries
parameterBlock
	: 'parameter' UNISEIS_NAME LINE_TERMINATOR? '{' LINE_TERMINATOR* (parameterDefinition LINE_TERMINATOR*)* '}'
	-> ^(PARAMETER UNISEIS_NAME parameterDefinition*)
	;

parameterDefinition
	: propertyAssignment
	| MANUAL_XML
	;

// rule for a property assignment
// properties are assigned using the syntax:
//    property: value
propertyAssignment
	: PROPERTY ':' values LINE_TERMINATOR
	-> ^(PROPERTY values)
	;

// these are the different types of values a property can take
fragment
values
	: STRING
	| DATE
	| BOOLEAN
	;

// These are the lexer rules
// Note that ORDER MATTERS, that is to say, the top-most rule will be matched first
// specific rules should go before generic rules
STRING
	: '"' ~('"')* '"'
	;

MANUAL_XML
	: '<manual>' .* '</manual>'
	;

BOOLEAN
	: 'true'
	| 'false'
	;

DATE
	: DIGIT DIGIT DIGIT DIGIT '-' DIGIT DIGIT '-' DIGIT DIGIT
	;

UNISEIS_NAME
	: ('A'..'Z')('A'..'Z'|DIGIT)+
	;

PROPERTY
	: ('a'..'z')('a'..'z'|'-')*
	;

fragment
DIGIT
	: '0'..'9'
	;

WS // Tab, vertical tab, form feed, space, non-breaking space and any other unicode "space separator".
	: ('\t' | '\v' | '\f' | ' ' | '\u00A0') {$channel=HIDDEN;}
	;

COMMENT
	: '/*' (options {greedy=false;} : . )* '*/' {$channel=HIDDEN;}
	;

LINE_COMMENT
	: '//' ~('\n'|'\r')* LINE_TERMINATOR {$channel=HIDDEN;}
	;

LINE_TERMINATOR
	: '\n'		// Line feed
	| '\r'		// Carriage return
	| '\u2028'	// Line separator
	| '\u2029'	// Paragraph separator
	;