Parse errors in patterns and constructors now report the line number in the

space of the colm source file.

8 files changed, 69 insertions, 41 deletions

diff --git a/colm/fsmrun.cpp b/colm/fsmrun.cpp
index 8b0e205f..b027e028 100644
--- a/colm/fsmrun.cpp
+++ b/colm/fsmrun.cpp
@@ -45,9 +45,7 @@ void operator<<( ostream &out, exit_object & )
 FsmRun::FsmRun( Program *prg ) :
 	prg(prg),
 	tables(prg->rtd->fsmTables),
-	parser(0),
-	line(1),
-	position(0)
+	parser(0)
 {
 }
 
@@ -119,23 +117,27 @@ void FsmRun::streamPush( const char *data, long length )
 /* Keep the position up to date after consuming text. */
 void update_position( FsmRun *fsmRun, const char *data, long length )
 {
-	for ( int i = 0; i < length; i++ ) {
-		if ( data[i] == '\n' )
-			fsmRun->line += 1;
+	if ( !fsmRun->inputStream->handlesLine ) {
+		for ( int i = 0; i < length; i++ ) {
+			if ( data[i] == '\n' )
+				fsmRun->inputStream->line += 1;
+		}
 	}
 
-	fsmRun->position += length;
+	fsmRun->inputStream->position += length;
 }
 
 /* Keep the position up to date after sending back text. */
 void undo_position( FsmRun *fsmRun, const char *data, long length )
 {
-	for ( int i = 0; i < length; i++ ) {
-		if ( data[i] == '\n' )
-			fsmRun->line -= 1;
+	if ( !fsmRun->inputStream->handlesLine ) {
+		for ( int i = 0; i < length; i++ ) {
+			if ( data[i] == '\n' )
+				fsmRun->inputStream->line -= 1;
+		}
 	}
 
-	fsmRun->position -= length;
+	fsmRun->inputStream->position -= length;
 }
 
 /* Should only be sending back whole tokens/ignores, therefore the send back
@@ -665,7 +667,7 @@ void FsmRun::sendIgnore( long id )
 	parser->ignore( tree );
 
 	/* Prepare for more scanning. */
-	position += length;
+	inputStream->position += length;
 	region = parser->getNextRegion();
 	cs = tables->entryByRegion[region];
 
@@ -769,7 +771,7 @@ void FsmRun::attachInputStream( InputStream *in )
 	p = pe = runBuf->buf;
 	peof = 0;
 	eofSent = false;
-	position = 0;
+	inputStream->position = 0;
 }
 
 long PdaRun::run()
@@ -937,7 +939,7 @@ long FsmRun::run( PdaRun *destParser )
 			}
 
 			/* Machine failed before finding a token. */
-			cerr << "SCANNER ERROR" << endp;
+			cerr << "error:" << inputStream->line << ": scanner error" << endp;
 		}
 
 		space = runBuf->buf + FSM_BUFSIZE - pe;
diff --git a/colm/fsmrun.h b/colm/fsmrun.h
index af1edc10..7d96112c 100644
--- a/colm/fsmrun.h
+++ b/colm/fsmrun.h
@@ -121,8 +121,6 @@ struct FsmRun
 	bool eofSent;
 	RunBuf *runBuf;
 	bool gotoResume;
-	long line;
-	long position;
 	char *mark_enter[32];
 	char *mark_leave[32];
 };
diff --git a/colm/input.h b/colm/input.h
index 3796c554..41feef25 100644
--- a/colm/input.h
+++ b/colm/input.h
@@ -33,6 +33,11 @@ struct RunBuf;
 
 struct InputStream
 {
+	InputStream( bool handlesLine ) :
+		line(1),
+		position(0),
+		handlesLine(handlesLine) {}
+
 	virtual ~InputStream() {}
 
 	/* Basic functions. */
@@ -51,12 +56,20 @@ struct InputStream
 		{ assert( false ); return 0; }
 	virtual void pushBackNamed()
 		{ assert( false ); }
+	
+	long line;
+	long position;
+
+	/* This is set true for input streams that do their own line counting.
+	 * Causes FsmRun to ignore NLs. */
+	bool handlesLine;
 };
 
 struct InputStreamString : public InputStream
 {
-	InputStreamString( const String &data )
-		: data(data), offset(0), eof(false) {}
+	InputStreamString( const String &data ) : 
+		InputStream(false), 
+		data(data), offset(0), eof(false) {}
 
 	int getData( char *dest, int length );
 	int isEOF() { return eof; }
@@ -70,8 +83,9 @@ struct InputStreamString : public InputStream
 
 struct InputStreamFile : public InputStream
 {
-	InputStreamFile( FILE *file )
-		: file(file), queue(0) {}
+	InputStreamFile( FILE *file ) :
+		InputStream(false), 
+		file(file), queue(0) {}
 
 	int getData( char *dest, int length );
 	int isEOF();
@@ -85,8 +99,9 @@ struct InputStreamFile : public InputStream
 
 struct InputStreamFD : public InputStream
 {
-	InputStreamFD( long fd )
-		: fd(fd), eof(false), queue(0) {}
+	InputStreamFD( long fd ) :
+		InputStream(false), 
+		fd(fd), eof(false), queue(0) {}
 
 	int isEOF();
 	int needFlush();
diff --git a/colm/lmparse.kl b/colm/lmparse.kl
index ef34d20f..dc81e9d9 100644
--- a/colm/lmparse.kl
+++ b/colm/lmparse.kl
@@ -399,7 +399,7 @@ litpat_el_list: ;
 
 litpat_el: TK_LitPat
 	final {
-		PatternItem *patternItem = new PatternItem( $1->data,
+		PatternItem *patternItem = new PatternItem( $1->loc, $1->data,
 			PatternItem::InputText );
 		patternItemList->append( patternItem );
 	};
@@ -444,7 +444,7 @@ pattern_el_type_or_lit: region_qual TK_Word opt_repeat
 	final {
 		PdaFactor *factor = new PdaFactor( $2->loc, false, $1->nspaceQual, 
 				$2->data, 0, $3->opt, $3->repeat );
-		$$->patternItem = new PatternItem( factor, PatternItem::FactorType );
+		$$->patternItem = new PatternItem( $2->loc, factor, PatternItem::FactorType );
 		patternItemList->append( $$->patternItem );
 	};
 
@@ -453,7 +453,7 @@ pattern_el_type_or_lit: region_qual TK_Literal opt_repeat
 		PdaLiteral *literal = new PdaLiteral( $2->loc, *$2 );
 		PdaFactor *factor = new PdaFactor( $2->loc, false, $1->nspaceQual, 
 				literal, 0, $3->opt, $3->repeat );
-		$$->patternItem = new PatternItem( factor, PatternItem::FactorType );
+		$$->patternItem = new PatternItem( $2->loc, factor, PatternItem::FactorType );
 		patternItemList->append( $$->patternItem );
 	};
 
@@ -488,7 +488,7 @@ litrepl_el_list: ;
 
 litrepl_el: TK_LitPat
 	final {
-		ReplItem *replItem = new ReplItem( ReplItem::InputText, $1->data );
+		ReplItem *replItem = new ReplItem( $1->loc, ReplItem::InputText, $1->data );
 		replItemList->append( replItem );
 	};
 
@@ -499,7 +499,7 @@ repl_el_list: ;
 
 repl_el: var_ref
 	final {
-		ReplItem *replItem = new ReplItem( ReplItem::VarRefType, $1->varRef );
+		ReplItem *replItem = new ReplItem( $1->varRef->loc, ReplItem::VarRefType, $1->varRef );
 		replItemList->append( replItem );
 	};
 
@@ -508,7 +508,7 @@ repl_el: region_qual TK_Literal
 		PdaLiteral *literal = new PdaLiteral( $2->loc, *$2 );
 		PdaFactor *factor = new PdaFactor( $2->loc, false, $1->nspaceQual, 
 				literal, 0, false, false );
-		ReplItem *replItem = new ReplItem( ReplItem::FactorType, factor );
+		ReplItem *replItem = new ReplItem( $2->loc, ReplItem::FactorType, factor );
 		replItemList->append( replItem );
 	};
 
diff --git a/colm/lmscan.rl b/colm/lmscan.rl
index d772a887..83283c35 100644
--- a/colm/lmscan.rl
+++ b/colm/lmscan.rl
@@ -165,6 +165,9 @@ void Scanner::token( int type )
 		loc.line = line;
 		loc.col = column;
 
+		if ( tokdata != 0 && tokdata[toklen-1] == '\n' )
+			loc.line -= 1;
+
 		parser->token( loc, type, tokdata, toklen );
 	}
 
@@ -442,8 +445,6 @@ void Scanner::endSection( )
 
 		# Whitespace other than newline.
 		[ \t\r]+ => { updateCol(); };
-
-		# If we are in a single line machine then newline may end the spec.
 		NL => { updateCol(); };
 
 		# Consume eof.
diff --git a/colm/parsedata.cpp b/colm/parsedata.cpp
index fd38e64e..2effebcf 100644
--- a/colm/parsedata.cpp
+++ b/colm/parsedata.cpp
@@ -1386,6 +1386,7 @@ void ParseData::initEmptyScanners()
 
 InputStreamPattern::InputStreamPattern( Pattern *pattern )
 : 
+	InputStream(true),
 	pattern(pattern),
 	patItem(pattern->list->head),
 	offset(0),
@@ -1408,6 +1409,7 @@ KlangEl *InputStreamPattern::getLangEl( long &bindId, char *&data, long &length
 	bindId = patItem->bindId;
 	data = 0;
 	length = 0;
+	line = patItem->loc.line;
 
 	patItem = patItem->next;
 	offset = 0;
@@ -1418,6 +1420,9 @@ KlangEl *InputStreamPattern::getLangEl( long &bindId, char *&data, long &length
 
 int InputStreamPattern::getData( char *dest, int length )
 { 
+	if ( offset == 0 )
+		line = patItem->loc.line;
+
 	assert ( patItem->type == PatternItem::InputText );
 	int available = patItem->data.length() - offset;
 
@@ -1491,6 +1496,7 @@ void InputStreamPattern::pushBackNamed()
 
 InputStreamRepl::InputStreamRepl( Replacement *replacement )
 : 
+	InputStream(true),
 	replacement(replacement),
 	replItem(replacement->list->head),
 	offset(0),
@@ -1517,6 +1523,7 @@ KlangEl *InputStreamRepl::getLangEl( long &bindId, char *&data, long &length )
 
 	data = 0;
 	length = 0;
+	line = replItem->loc.line;
 
 	if ( replItem->type == ReplItem::FactorType ) {
 		if ( replItem->factor->literal != 0 ) {
@@ -1538,6 +1545,9 @@ KlangEl *InputStreamRepl::getLangEl( long &bindId, char *&data, long &length )
 
 int InputStreamRepl::getData( char *dest, int length )
 { 
+	if ( offset == 0 )
+		line = replItem->loc.line;
+
 	assert ( replItem->type == ReplItem::InputText );
 	int available = replItem->data.length() - offset;
 
diff --git a/colm/parsetree.h b/colm/parsetree.h
index 96449c43..a9e4f6c1 100644
--- a/colm/parsetree.h
+++ b/colm/parsetree.h
@@ -907,14 +907,15 @@ struct PatternItem
 		InputText
 	};
 
-	PatternItem( const String &data, Type type ) : 
-			factor(0), data(data), type(type), region(0), 
+	PatternItem( const InputLoc &loc, const String &data, Type type ) : 
+			loc(loc), factor(0), data(data), type(type), region(0), 
 			varRef(0), bindId(0) {}
 
-	PatternItem( PdaFactor *factor, Type type ) : 
-			factor(factor), type(type), region(0), 
+	PatternItem( const InputLoc &loc, PdaFactor *factor, Type type ) : 
+			loc(loc), factor(factor), type(type), region(0), 
 			varRef(0), bindId(0) {}
 
+	InputLoc loc;
 	PdaFactor *factor;
 	String data;
 	Type type;
@@ -935,15 +936,16 @@ struct ReplItem
 		FactorType
 	};
 
-	ReplItem( Type type, const String &data ) : 
-		type(type), data(data), varRef(0), bindId(0) {}
+	ReplItem( const InputLoc &loc, Type type, const String &data ) : 
+		loc(loc), type(type), data(data), varRef(0), bindId(0) {}
 
-	ReplItem( Type type, LangVarRef *varRef ) : 
-		type(type), varRef(varRef), bindId(0) {}
+	ReplItem( const InputLoc &loc, Type type, LangVarRef *varRef ) : 
+		loc(loc), type(type), varRef(varRef), bindId(0) {}
 
-	ReplItem( Type type, PdaFactor *factor ) : 
-		type(type), factor(factor), bindId(0) {}
+	ReplItem( const InputLoc &loc, Type type, PdaFactor *factor ) : 
+		loc(loc), type(type), factor(factor), bindId(0) {}
 
+	InputLoc loc;
 	Type type;
 	String data;
 	LangVarRef *varRef;
diff --git a/colm/pdarun.cpp b/colm/pdarun.cpp
index d3a3fd87..07bd8d09 100644
--- a/colm/pdarun.cpp
+++ b/colm/pdarun.cpp
@@ -701,7 +701,7 @@ _out:
 
 ostream &PdaRun::parse_error( int tokId, Tree *tree )
 {
-	cerr << "error:" << fsmRun->line << ": at token ";
+	cerr << "error:" << fsmRun->inputStream->line << ": at token ";
 	if ( tokId < 128 )
 		cerr << "\"" << tables->rtd->lelInfo[tokId].name << "\"";
 	else