Since there is no longer a continually run scanner loop that pushes tokens, we

can initialize FsmRun::cs and FsmRun::tokstart at the beginning of teach run of
the scanner in which we attempt to fetch a token.

2 files changed, 4 insertions, 45 deletions

diff --git a/colm/fsmrun.cpp b/colm/fsmrun.cpp
index 811899f1..f5968585 100644
--- a/colm/fsmrun.cpp
+++ b/colm/fsmrun.cpp
@@ -192,9 +192,6 @@ void FsmRun::sendBackText( const char *data, long length )
 	assert( memcmp( data, p, length ) == 0 );
 		
 	undo_position( this, data, length );
-
-	/* We are adjusting p so this must be reset. */
-	tokstart = 0;
 }
 
 void FsmRun::queueBack( PdaRun *parser, Kid *input )
@@ -397,10 +394,6 @@ void send_queued_tokens( FsmRun *fsmRun, PdaRun *parser )
 			
 			parser->ignore( send->tree );
 			fsmRun->prg->kidPool.free( send );
-
-			/* Set the current state from the next region. */
-			fsmRun->region = parser->getNextRegion();
-			fsmRun->cs = fsmRun->tables->entryByRegion[fsmRun->region];
 		}
 		else {
 			#ifdef COLM_LOG_PARSE
@@ -583,10 +576,6 @@ void send_handle_error( FsmRun *fsmRun, PdaRun *parser, Kid *input )
 		parser->parse_error(id, input->tree) << "parse error" << endp;
 	}
 	else {
-		/* Set the current state from the next region. */
-		fsmRun->region = parser->getNextRegion();
-		fsmRun->cs = fsmRun->tables->entryByRegion[fsmRun->region];
-
 		if ( parser->isParserStopFinished() ) {
 			#ifdef COLM_LOG_PARSE
 			if ( colm_log_parse ) {
@@ -636,11 +625,8 @@ void exec_gen( FsmRun *fsmRun, PdaRun *parser, long id )
 	 * data is pulled from the stream. */
 
 	fsmRun->p = fsmRun->tokstart;
-	fsmRun->tokstart = 0;
 
 	generation_action( fsmRun, parser, id, tokdata, false, 0 );
-
-	memset( fsmRun->mark, 0, sizeof(fsmRun->mark) );
 }
 
 void send_ignore( FsmRun *fsmRun, PdaRun *parser, long id )
@@ -659,7 +645,6 @@ void send_ignore( FsmRun *fsmRun, PdaRun *parser, long id )
 	int length = fsmRun->p - fsmRun->tokstart;
 	Head *ignoreStr = string_alloc_const( fsmRun->prg, fsmRun->tokstart, length );
 	update_position( fsmRun, fsmRun->tokstart, length );
-	fsmRun->tokstart = 0;
 	
 	Tree *tree = fsmRun->prg->treePool.allocate();
 	tree->refs = 1;
@@ -668,12 +653,6 @@ void send_ignore( FsmRun *fsmRun, PdaRun *parser, long id )
 
 	/* Send it to the parser. */
 	parser->ignore( tree );
-
-	/* Prepare for more scanning. */
-	fsmRun->region = parser->getNextRegion();
-	fsmRun->cs = fsmRun->tables->entryByRegion[fsmRun->region];
-
-	memset( fsmRun->mark, 0, sizeof(fsmRun->mark) );
 }
 
 void send_token( FsmRun *fsmRun, PdaRun *parser, long id )
@@ -693,14 +672,8 @@ void send_token( FsmRun *fsmRun, PdaRun *parser, long id )
 	Head *tokdata = string_alloc_const( fsmRun->prg, fsmRun->tokstart, length );
 	update_position( fsmRun, fsmRun->tokstart, length );
 
-	/* By default the match is consumed and this is what we need. Just
-	 * need to reset tokstart. */
-	fsmRun->tokstart = 0;
-
 	Kid *input = make_token( fsmRun, parser, id, tokdata, false, 0 );
 	send_handle_error( fsmRun, parser, input );
-
-	memset( fsmRun->mark, 0, sizeof(fsmRun->mark) );
 }
 
 /* Load up a token, starting from tokstart if it is set. If not set then
@@ -774,10 +747,6 @@ void send_eof( FsmRun *fsmRun, PdaRun *parser )
 		parser->parse_error( input->tree->id, input->tree ) << 
 				"parse error" << endp;
 	}
-
-	fsmRun->tokstart = 0;
-	fsmRun->region = parser->getNextRegion();
-	fsmRun->cs = fsmRun->tables->entryByRegion[fsmRun->region];
 }
 
 
@@ -839,8 +808,6 @@ void scanner_error( FsmRun *fsmRun, PdaRun *parser )
 		parser->sendBackIgnore();
 
 		parser->nextRegionInd += 1;
-		fsmRun->region = parser->getNextRegion();
-		fsmRun->cs = fsmRun->tables->entryByRegion[fsmRun->region];
 		#ifdef COLM_LOG_PARSE
 		if ( colm_log_parse ) {
 			cerr << "new token region: " << 
@@ -866,8 +833,6 @@ void scanner_error( FsmRun *fsmRun, PdaRun *parser )
 			cerr << "PARSE ERROR" << endp;
 		}
 		else {
-			fsmRun->region = parser->getNextRegion();
-			fsmRun->cs = fsmRun->tables->entryByRegion[fsmRun->region];
 			#ifdef COLM_LOG_PARSE
 			if ( colm_log_parse ) {
 				cerr << "new token region: " << 
@@ -887,7 +852,7 @@ void parse( FsmRun *fsmRun, PdaRun *parser )
 	parser->init();
 
 	while ( true ) {
-		int tokenId = fsmRun->scan( parser );
+		int tokenId = fsmRun->scanToken( parser );
 
 		/* Check for EOF. */
 		if ( tokenId == SCAN_EOF ) {
@@ -932,10 +897,8 @@ void parse( FsmRun *fsmRun, PdaRun *parser )
 	}
 }
 
-long FsmRun::scan( PdaRun *parser )
+long FsmRun::scanToken( PdaRun *parser )
 {
-	long space;
-
 	/* Init the scanner vars. */
 	act = 0;
 	tokstart = 0;
@@ -949,9 +912,6 @@ long FsmRun::scan( PdaRun *parser )
 	/* Clear the mark array. */
 	memset( mark, 0, sizeof(mark) );
 
-	/* Start with the EOF test. The pattern and replacement input sources can
-	 * be EOF from the start. */
-
 	while ( true ) {
 		execute();
 
@@ -1002,7 +962,7 @@ long FsmRun::scan( PdaRun *parser )
 
 		/* There may be space left in the current buffer. If not then we need
 		 * to make some. */
-		space = runBuf->buf + FSM_BUFSIZE - pe;
+		long space = runBuf->buf + FSM_BUFSIZE - pe;
 		if ( space == 0 ) {
 			/* Create a new run buf. */
 			RunBuf *newBuf = new RunBuf;
@@ -1058,7 +1018,6 @@ long FsmRun::scan( PdaRun *parser )
 		pe = p + len;
 		if ( inputStream->needFlush() )
 			peof = pe;
-
 	}
 
 	/* Should not be reached. */
diff --git a/colm/fsmrun.h b/colm/fsmrun.h
index 920508f3..8fed546c 100644
--- a/colm/fsmrun.h
+++ b/colm/fsmrun.h
@@ -98,7 +98,7 @@ struct FsmRun
 	void sendBackText( const char *data, long length );
 	void execAction( GenAction *action );
 
-	long scan( PdaRun *parser );
+	long scanToken( PdaRun *parser );
 	void attachInputStream( InputStream *in );
 	void streamPush( const char *data, long length );
 	void undoStreamPush( long length );