01-ragel-6.7.GT2.patch

GT2 patch as found in ragel-6-gt2 branch.

More info: http://www.complang.org/pipermail/ragel-users/2012-May/002831.html

--- ragel/cdgoto.cpp
+++ ragel/cdgoto.cpp
@@ -34,6 +34,11 @@
 	return out;
 }
 
+int GotoCodeGen::TRANS_NR( RedTransAp *trans )
+{
+	return trans->id;
+}
+
 std::ostream &GotoCodeGen::TO_STATE_ACTION_SWITCH()
 {
 	/* Walk the list of functions, printing the cases. */
@@ -109,6 +114,86 @@
 }
 
 
+void GotoCodeGen::emitTableSwitch( RedStateAp *state )
+{
+	/* Load up the singles. */
+	int numSingles = state->outSingle.length();
+	RedTransEl *sdata = state->outSingle.data;
+
+	/* Load up the ranges.  */
+	int numRanges = state->outRange.length();
+	RedTransEl *rdata = state->outRange.data;
+
+	int minId = INT_MAX;
+	int maxId = INT_MIN;
+
+	for ( Key i = keyOps->minKey; i <= keyOps->maxKey; i.increment() ) {
+		RedTransAp *trans = state->defTrans;
+		for ( int j = 0; j < numRanges; j++ )
+			if ( ( rdata[j].lowKey.getVal() <= i ) && ( i <= rdata[j].highKey.getVal() ) )
+				trans = rdata[j].value;
+		for ( int j = 0; j < numSingles; j++ )
+			if ( i == sdata[j].lowKey.getVal() )
+				trans = sdata[j].value;
+		if ( minId > TRANS_NR(trans) )
+			minId = TRANS_NR(trans);
+		if ( maxId < TRANS_NR(trans) )
+			maxId = TRANS_NR(trans);
+	}
+
+	if ( (maxId - minId) <= UCHAR_MAX )
+		out << "\t{\n" <<
+			"\t\tstatic const unsigned char jump_table[] = ";
+	else if ( (maxId - minId) <= USHRT_MAX )
+		out << "\t{\n" <<
+			"\t\tstatic const unsigned short jump_table[] = ";
+	else
+		out << "\t{\n" <<
+			"\t\tstatic const int jump_table[] = ";
+
+	char delimiter = '{';
+	for ( Key i = keyOps->minKey; i <= keyOps->maxKey; i.increment() ) {
+		RedTransAp *trans = state->defTrans;
+		for ( int j = 0; j < numRanges; j++ )
+			if ( ( rdata[j].lowKey.getVal() <= i ) && ( i <= rdata[j].highKey.getVal() ) )
+				trans = rdata[j].value;
+		for ( int j = 0; j < numSingles; j++ )
+			if ( i == sdata[j].lowKey.getVal() )
+				trans = sdata[j].value;
+		if ( minId >= 0 && maxId <= UCHAR_MAX )
+			out << delimiter << " " << TRANS_NR(trans);
+		else if ( (maxId - minId) <= UCHAR_MAX )
+			out << delimiter << " " << (TRANS_NR(trans) - minId);
+		else if ( minId >= 0 && maxId <= USHRT_MAX )
+			out << delimiter << " " << TRANS_NR(trans);
+		else if ( (maxId - minId) <= USHRT_MAX )
+			out << delimiter << " " << (TRANS_NR(trans) - minId);
+		else
+			out << delimiter << " " << TRANS_NR(trans);
+		delimiter = ',';
+	}
+	out << " };\n";
+
+	if (keyOps->minKey != 0 )
+		out << "\t\t" << vCS() << " = jump_table[" << GET_WIDE_KEY(state) << "-" <<
+			WIDE_KEY(state, keyOps->minKey) << "]";
+	else
+		out << "\t\t" << vCS() << " = jump_table[" << GET_WIDE_KEY(state) << "]";
+	if ( minId >= 0 && maxId <= UCHAR_MAX )
+		out << ";\n";
+	else if ( (maxId - minId) <= UCHAR_MAX )
+		out << " + " << minId << ";\n";
+	else if ( minId >= 0 && maxId <= USHRT_MAX )
+		out << ";\n";
+	else if ( (maxId - minId) <= USHRT_MAX )
+		out << " + " << minId << ";\n";
+	else
+		out << ";\n";
+
+	out << "\t\tgoto _again;\n"
+	       "\t}\n";
+}
+
 void GotoCodeGen::emitSingleSwitch( RedStateAp *state )
 {
 	/* Load up the singles. */
@@ -361,16 +446,24 @@
 				emitCondBSearch( st, 1, 0, st->stateCondVect.length() - 1 );
 			}
 
-			/* Try singles. */
-			if ( st->outSingle.length() > 0 )
-				emitSingleSwitch( st );
-
-			/* Default case is to binary search for the ranges, if that fails then */
-			if ( st->outRange.length() > 0 )
-				emitRangeBSearch( st, 1, 0, st->outRange.length() - 1 );
+			if ( (st->outSingle.length() + st->outRange.length() * 2) > maxTransitions) {
+
+				emitTableSwitch( st );
+
+			} else {
 
-			/* Write the default transition. */
-			TRANS_GOTO( st->defTrans, 1 ) << "\n";
+				/* Try singles. */
+				if ( st->outSingle.length() > 0 )
+					emitSingleSwitch( st );
+
+				/* Default case is to binary search for the ranges, if that fails then */
+				if ( st->outRange.length() > 0 )
+					emitRangeBSearch( st, 1, 0, st->outRange.length() - 1 );
+
+				/* Write the default transition. */
+				TRANS_GOTO( st->defTrans, 1 ) << "\n";
+
+			}
 		}
 	}
 	return out;
--- ragel/cdgoto.h
+++ ragel/cdgoto.h
@@ -74,7 +74,9 @@
 	void STATE_CONDS( RedStateAp *state, bool genDefault ); 
 
 	virtual std::ostream &TRANS_GOTO( RedTransAp *trans, int level );
+	virtual int TRANS_NR( RedTransAp *trans );
 
+	void emitTableSwitch( RedStateAp *state );
 	void emitSingleSwitch( RedStateAp *state );
 	void emitRangeBSearch( RedStateAp *state, int level, int low, int high );
 
--- ragel/cdipgoto.cpp	2011-02-11 07:14:43.000000000 +0300
+++ ragel/cdipgoto.cpp	2012-05-13 15:20:11.433272943 +0400
@@ -27,9 +27,20 @@
 #include "gendata.h"
 #include "bstmap.h"
 
+bool IpGotoCodeGen::useTransInAgainLabel()
+{
+	for ( RedStateList::Iter st = redFsm->stateList; st.lte(); st++ )
+		if ( st == redFsm->errState )
+			continue;
+		else if ( (st->outSingle.length() + st->outRange.length() * 2) > maxTransitions)
+			return true;
+	return false;
+}
+
 bool IpGotoCodeGen::useAgainLabel()
 {
-	return redFsm->anyRegActionRets() || 
+	return useTransInAgainLabel() ||
+			redFsm->anyRegActionRets() || 
 			redFsm->anyRegActionByValControl() || 
 			redFsm->anyRegNextStmt();
 }
@@ -243,6 +254,18 @@
 	return out;
 }
 
+int IpGotoCodeGen::TRANS_NR( RedTransAp *trans )
+{
+	if ( trans->action != 0 ) {
+		/* Go to the transition which will go to the state. */
+		return trans->id + redFsm->stateList.length();
+	}
+	else {
+		/* Go directly to the target state. */
+		return trans->targ->id;
+	}
+}
+
 std::ostream &IpGotoCodeGen::EXIT_STATES()
 {
 	for ( RedStateList::Iter st = redFsm->stateList; st.lte(); st++ ) {
@@ -258,6 +281,13 @@
 std::ostream &IpGotoCodeGen::AGAIN_CASES()
 {
 	for ( RedStateList::Iter st = redFsm->stateList; st.lte(); st++ ) {
+		if ( useTransInAgainLabel() )
+			for ( int it = 0; it < st->numInTrans; it++ ) {
+				RedTransAp *trans = st->inTrans[it];
+				if ( trans->action != 0 && trans->labelNeeded )
+					out <<
+						"\t\tcase " << (trans->id + redFsm->stateList.length()) << ": goto tr" << trans->id << ";\n";
+			}
 		out << 
 			"		case " << st->id << ": goto st" << st->id << ";\n";
 	}
@@ -409,7 +439,6 @@
 			out << 
 				"	" << P() << " += 1;\n";
 		}
-
 		out << "_resume:\n";
 	}
 
--- ragel/cdipgoto.h
+++ ragel/cdipgoto.h
@@ -40,6 +40,7 @@
 
 	std::ostream &EXIT_STATES();
 	std::ostream &TRANS_GOTO( RedTransAp *trans, int level );
+	int TRANS_NR( RedTransAp *trans );
 	std::ostream &FINISH_CASES();
 	std::ostream &AGAIN_CASES();
 
@@ -58,6 +59,7 @@
 	virtual void writeExec();
 
 protected:
+	bool useTransInAgainLabel();
 	bool useAgainLabel();
 
 	/* Called from GotoCodeGen::STATE_GOTOS just before writing the gotos for
--- ragel/main.cpp
+++ ragel/main.cpp
@@ -83,6 +83,8 @@
 /* Target language and output style. */
 CodeStyle codeStyle = GenTables;
 
+long maxTransitions = LONG_MAX;
+
 int numSplitPartitions = 0;
 bool noLineDirectives = false;
 
@@ -139,6 +140,7 @@
 "   -G1                  Faster goto-driven FSM\n"
 "code style: (C/D)\n"
 "   -G2                  Really fast goto-driven FSM\n"
+"   -GT2                 Really fast goto-driven FSM with table-driven fallback\n"
 "   -P<N>                N-Way Split really fast goto-driven FSM\n"
 	;	
 
@@ -407,7 +409,10 @@
 					codeStyle = GenFGoto;
 				else if ( pc.paramArg[0] == '2' )
 					codeStyle = GenIpGoto;
-				else {
+				else if ( pc.paramArg[0] == 'T' && pc.paramArg[1] == '2' ) {
+					codeStyle = GenIpGoto;
+					maxTransitions = 32;
+				} else {
 					error() << "-G" << pc.paramArg[0] << 
 							" is an invalid argument" << endl;
 					exit(1);
--- ragel/ragel.h
+++ ragel/ragel.h
@@ -109,6 +109,8 @@
 extern int numSplitPartitions;
 extern bool noLineDirectives;
 
+extern long maxTransitions;
+
 std::ostream &error();
 
 /* Target language and output style. */