From f0d3bd96b1036498f04714502b3dc924ee20fbbd Mon Sep 17 00:00:00 2001
From: thurston <thurston@052ea7fc-9027-0410-9066-f65837a77df0>
Date: Sun, 20 Apr 2008 23:16:19 +0000
Subject: [PATCH] Bug fix for eofTrans. When not using indicies we can't use a
 transitions's id to identify the transition to take. Instead add the
 transition to the end of the list and store it's position in a new var called
 pos. The pos var is then used as the index.

git-svn-id: http://svn.complang.org/ragel/trunk@437 052ea7fc-9027-0410-9066-f65837a77df0
---
 redfsm/redfsm.h                 |  3 ++-
 rlgen-cd/flatcodegen.cpp        | 11 ++++++++---
 rlgen-cd/fsmcodegen.h           |  4 ++--
 rlgen-cd/tabcodegen.cpp         | 34 +++++++++++++++++++++++++++++++---
 rlgen-csharp/flatcodegen.cpp    | 11 ++++++++---
 rlgen-csharp/fsmcodegen.h       |  4 ++--
 rlgen-csharp/tabcodegen.cpp     | 31 ++++++++++++++++++++++++++++---
 rlgen-java/javacodegen.cpp      | 30 +++++++++++++++++++++++++++---
 rlgen-java/javacodegen.h        |  4 ++--
 rlgen-ruby/ruby-codegen.h       |  4 ++--
 rlgen-ruby/ruby-flatcodegen.cpp | 11 ++++++++---
 rlgen-ruby/ruby-tabcodegen.cpp  | 26 +++++++++++++++++++++++---
 test/scan2.rl                   | 34 ++++++++++++++++++++++++++++++++++
 test/scan3.rl                   | 32 ++++++++++++++++++++++++++++++++
 14 files changed, 209 insertions(+), 30 deletions(-)
 create mode 100644 test/scan2.rl
 create mode 100644 test/scan3.rl

diff --git a/redfsm/redfsm.h b/redfsm/redfsm.h
index 8e53b9e..47906f7 100644
--- a/redfsm/redfsm.h
+++ b/redfsm/redfsm.h
@@ -217,11 +217,12 @@ struct RedTransAp
 	public AvlTreeEl<RedTransAp>
 {
 	RedTransAp( RedStateAp *targ, RedAction *action, int id )
-		: targ(targ), action(action), id(id), labelNeeded(true) { }
+		: targ(targ), action(action), id(id), pos(-1), labelNeeded(true) { }
 
 	RedStateAp *targ;
 	RedAction *action;
 	int id;
+	int pos;
 	bool partitionBoundary;
 	bool labelNeeded;
 };
diff --git a/rlgen-cd/flatcodegen.cpp b/rlgen-cd/flatcodegen.cpp
index 783cde3..7195bc8 100644
--- a/rlgen-cd/flatcodegen.cpp
+++ b/rlgen-cd/flatcodegen.cpp
@@ -236,8 +236,10 @@ std::ostream &FlatCodeGen::EOF_TRANS()
 		/* Write any eof action. */
 
 		long trans = 0;
-		if ( st->eofTrans != 0 )
-			trans = st->eofTrans->id+1;
+		if ( st->eofTrans != 0 ) {
+			assert( st->eofTrans->pos >= 0 );
+			trans = st->eofTrans->pos+1;
+		}
 		out << trans;
 
 		if ( !st.last() ) {
@@ -394,8 +396,11 @@ std::ostream &FlatCodeGen::TRANS_TARGS()
 	out << '\t';
 	int totalStates = 0;
 	for ( int t = 0; t < redFsm->transSet.length(); t++ ) {
-		/* Write out the target state. */
+		/* Save the position. Needed for eofTargs. */
 		RedTransAp *trans = transPtrs[t];
+		trans->pos = t;
+
+		/* Write out the target state. */
 		out << trans->targ->id;
 		if ( t < redFsm->transSet.length()-1 ) {
 			out << ", ";
diff --git a/rlgen-cd/fsmcodegen.h b/rlgen-cd/fsmcodegen.h
index 7390448..9088370 100644
--- a/rlgen-cd/fsmcodegen.h
+++ b/rlgen-cd/fsmcodegen.h
@@ -116,8 +116,8 @@ protected:
 	string SL() { return "_" + DATA_PREFIX() + "single_lengths"; }
 	string RL() { return "_" + DATA_PREFIX() + "range_lengths"; }
 	string A() { return "_" + DATA_PREFIX() + "actions"; }
-	string TA() { return "_" + DATA_PREFIX() + "trans_actions_wi"; }
-	string TT() { return "_" + DATA_PREFIX() + "trans_targs_wi"; }
+	string TA() { return "_" + DATA_PREFIX() + "trans_actions"; }
+	string TT() { return "_" + DATA_PREFIX() + "trans_targs"; }
 	string TSA() { return "_" + DATA_PREFIX() + "to_state_actions"; }
 	string FSA() { return "_" + DATA_PREFIX() + "from_state_actions"; }
 	string EA() { return "_" + DATA_PREFIX() + "eof_actions"; }
diff --git a/rlgen-cd/tabcodegen.cpp b/rlgen-cd/tabcodegen.cpp
index 94d5c08..977ec63 100644
--- a/rlgen-cd/tabcodegen.cpp
+++ b/rlgen-cd/tabcodegen.cpp
@@ -334,8 +334,10 @@ std::ostream &TabCodeGen::EOF_TRANS()
 	for ( RedStateList::Iter st = redFsm->stateList; st.lte(); st++ ) {
 		/* Write any eof action. */
 		long trans = 0;
-		if ( st->eofTrans != 0 )
-			trans = st->eofTrans->id+1;
+		if ( st->eofTrans != 0 ) {
+			assert( st->eofTrans->pos >= 0 );
+			trans = st->eofTrans->pos+1;
+		}
 		out << trans;
 
 		if ( !st.last() ) {
@@ -489,6 +491,18 @@ std::ostream &TabCodeGen::TRANS_TARGS()
 		}
 	}
 
+	/* Add any eof transitions that have not yet been written out above. */
+	for ( RedStateList::Iter st = redFsm->stateList; st.lte(); st++ ) {
+		if ( st->eofTrans != 0 ) {
+			RedTransAp *trans = st->eofTrans;
+			trans->pos = totalTrans;
+			out << trans->targ->id << ", ";
+			if ( ++totalTrans % IALL == 0 )
+				out << "\n\t";
+		}
+	}
+
+
 	/* Output one last number so we don't have to figure out when the last
 	 * entry is and avoid writing a comma. */
 	out << 0 << "\n";
@@ -526,6 +540,17 @@ std::ostream &TabCodeGen::TRANS_ACTIONS()
 		}
 	}
 
+	/* Add any eof transitions that have not yet been written out above. */
+	for ( RedStateList::Iter st = redFsm->stateList; st.lte(); st++ ) {
+		if ( st->eofTrans != 0 ) {
+			RedTransAp *trans = st->eofTrans;
+			TRANS_ACTION( trans ) << ", ";
+			if ( ++totalTrans % IALL == 0 )
+				out << "\n\t";
+		}
+	}
+
+
 	/* Output one last number so we don't have to figure out when the last
 	 * entry is and avoid writing a comma. */
 	out << 0 << "\n";
@@ -543,8 +568,11 @@ std::ostream &TabCodeGen::TRANS_TARGS_WI()
 	out << '\t';
 	int totalStates = 0;
 	for ( int t = 0; t < redFsm->transSet.length(); t++ ) {
-		/* Write out the target state. */
+		/* Record the position, need this for eofTrans. */
 		RedTransAp *trans = transPtrs[t];
+		trans->pos = t;
+
+		/* Write out the target state. */
 		out << trans->targ->id;
 		if ( t < redFsm->transSet.length()-1 ) {
 			out << ", ";
diff --git a/rlgen-csharp/flatcodegen.cpp b/rlgen-csharp/flatcodegen.cpp
index f939c02..bf3ba24 100644
--- a/rlgen-csharp/flatcodegen.cpp
+++ b/rlgen-csharp/flatcodegen.cpp
@@ -236,8 +236,10 @@ std::ostream &CSharpFlatCodeGen::EOF_TRANS()
 		/* Write any eof action. */
 
 		long trans = 0;
-		if ( st->eofTrans != 0 )
-			trans = st->eofTrans->id+1;
+		if ( st->eofTrans != 0 ) {
+			assert( st->eofTrans->pos >= 0 );
+			trans = st->eofTrans->pos+1;
+		}
 		out << trans;
 
 		if ( !st.last() ) {
@@ -394,8 +396,11 @@ std::ostream &CSharpFlatCodeGen::TRANS_TARGS()
 	out << '\t';
 	int totalStates = 0;
 	for ( int t = 0; t < redFsm->transSet.length(); t++ ) {
-		/* Write out the target state. */
+		/* Record the position, need this for eofTrans. */
 		RedTransAp *trans = transPtrs[t];
+		trans->pos = t;
+
+		/* Write out the target state. */
 		out << trans->targ->id;
 		if ( t < redFsm->transSet.length()-1 ) {
 			out << ", ";
diff --git a/rlgen-csharp/fsmcodegen.h b/rlgen-csharp/fsmcodegen.h
index 439187e..d29d551 100644
--- a/rlgen-csharp/fsmcodegen.h
+++ b/rlgen-csharp/fsmcodegen.h
@@ -117,8 +117,8 @@ protected:
 	string SL() { return "_" + DATA_PREFIX() + "single_lengths"; }
 	string RL() { return "_" + DATA_PREFIX() + "range_lengths"; }
 	string A() { return "_" + DATA_PREFIX() + "actions"; }
-	string TA() { return "_" + DATA_PREFIX() + "trans_actions_wi"; }
-	string TT() { return "_" + DATA_PREFIX() + "trans_targs_wi"; }
+	string TA() { return "_" + DATA_PREFIX() + "trans_actions"; }
+	string TT() { return "_" + DATA_PREFIX() + "trans_targs"; }
 	string TSA() { return "_" + DATA_PREFIX() + "to_state_actions"; }
 	string FSA() { return "_" + DATA_PREFIX() + "from_state_actions"; }
 	string EA() { return "_" + DATA_PREFIX() + "eof_actions"; }
diff --git a/rlgen-csharp/tabcodegen.cpp b/rlgen-csharp/tabcodegen.cpp
index 0e65568..9233246 100644
--- a/rlgen-csharp/tabcodegen.cpp
+++ b/rlgen-csharp/tabcodegen.cpp
@@ -334,8 +334,10 @@ std::ostream &CSharpTabCodeGen::EOF_TRANS()
 	for ( RedStateList::Iter st = redFsm->stateList; st.lte(); st++ ) {
 		/* Write any eof action. */
 		long trans = 0;
-		if ( st->eofTrans != 0 )
-			trans = st->eofTrans->id+1;
+		if ( st->eofTrans != 0 ) {
+			assert( st->eofTrans->pos >= 0 );
+			trans = st->eofTrans->pos+1;
+		}
 		out << trans;
 
 		if ( !st.last() ) {
@@ -489,6 +491,17 @@ std::ostream &CSharpTabCodeGen::TRANS_TARGS()
 		}
 	}
 
+	for ( RedStateList::Iter st = redFsm->stateList; st.lte(); st++ ) {
+		if ( st->eofTrans != 0 ) {
+			RedTransAp *trans = st->eofTrans;
+			trans->pos = totalTrans;
+			out << trans->targ->id << ", ";
+			if ( ++totalTrans % IALL == 0 )
+				out << "\n\t";
+		}
+	}
+
+
 	/* Output one last number so we don't have to figure out when the last
 	 * entry is and avoid writing a comma. */
 	out << 0 << "\n";
@@ -526,6 +539,15 @@ std::ostream &CSharpTabCodeGen::TRANS_ACTIONS()
 		}
 	}
 
+	for ( RedStateList::Iter st = redFsm->stateList; st.lte(); st++ ) {
+		if ( st->eofTrans != 0 ) {
+			RedTransAp *trans = st->eofTrans;
+			TRANS_ACTION( trans ) << ", ";
+			if ( ++totalTrans % IALL == 0 )
+				out << "\n\t";
+		}
+	}
+
 	/* Output one last number so we don't have to figure out when the last
 	 * entry is and avoid writing a comma. */
 	out << 0 << "\n";
@@ -543,8 +565,11 @@ std::ostream &CSharpTabCodeGen::TRANS_TARGS_WI()
 	out << '\t';
 	int totalStates = 0;
 	for ( int t = 0; t < redFsm->transSet.length(); t++ ) {
-		/* Write out the target state. */
+		/* Record the position, need this for eofTrans. */
 		RedTransAp *trans = transPtrs[t];
+		trans->pos = t;
+
+		/* Write out the target state. */
 		out << trans->targ->id;
 		if ( t < redFsm->transSet.length()-1 ) {
 			out << ", ";
diff --git a/rlgen-java/javacodegen.cpp b/rlgen-java/javacodegen.cpp
index e70a7b7..fdce2b3 100644
--- a/rlgen-java/javacodegen.cpp
+++ b/rlgen-java/javacodegen.cpp
@@ -611,8 +611,10 @@ std::ostream &JavaTabCodeGen::EOF_TRANS()
 	for ( RedStateList::Iter st = redFsm->stateList; st.lte(); st++ ) {
 		/* Write any eof action. */
 		long trans = 0;
-		if ( st->eofTrans != 0 )
-			trans = st->eofTrans->id+1;
+		if ( st->eofTrans != 0 ) {
+			assert( st->eofTrans->pos >= 0 );
+			trans = st->eofTrans->pos+1;
+		}
 
 		/* Write any eof action. */
 		ARRAY_ITEM( INT(trans), st.last() );
@@ -705,23 +707,35 @@ std::ostream &JavaTabCodeGen::INDICIES()
 
 std::ostream &JavaTabCodeGen::TRANS_TARGS()
 {
+	int totalTrans = 0;
 	for ( RedStateList::Iter st = redFsm->stateList; st.lte(); st++ ) {
 		/* Walk the singles. */
 		for ( RedTransList::Iter stel = st->outSingle; stel.lte(); stel++ ) {
 			RedTransAp *trans = stel->value;
 			ARRAY_ITEM( KEY( trans->targ->id ), false );
+			totalTrans++;
 		}
 
 		/* Walk the ranges. */
 		for ( RedTransList::Iter rtel = st->outRange; rtel.lte(); rtel++ ) {
 			RedTransAp *trans = rtel->value;
 			ARRAY_ITEM( KEY( trans->targ->id ), false );
+			totalTrans++;
 		}
 
 		/* The state's default target state. */
 		if ( st->defTrans != 0 ) {
 			RedTransAp *trans = st->defTrans;
 			ARRAY_ITEM( KEY( trans->targ->id ), false );
+			totalTrans++;
+		}
+	}
+
+	for ( RedStateList::Iter st = redFsm->stateList; st.lte(); st++ ) {
+		if ( st->eofTrans != 0 ) {
+			RedTransAp *trans = st->eofTrans;
+			trans->pos = totalTrans++;
+			ARRAY_ITEM( KEY( trans->targ->id ), false );
 		}
 	}
 
@@ -754,6 +768,13 @@ std::ostream &JavaTabCodeGen::TRANS_ACTIONS()
 		}
 	}
 
+	for ( RedStateList::Iter st = redFsm->stateList; st.lte(); st++ ) {
+		if ( st->eofTrans != 0 ) {
+			RedTransAp *trans = st->eofTrans;
+			ARRAY_ITEM( INT(TRANS_ACTION( trans )), false );
+		}
+	}
+
 	/* Output one last number so we don't have to figure out when the last
 	 * entry is and avoid writing a comma. */
 	ARRAY_ITEM( INT(0), true );
@@ -769,8 +790,11 @@ std::ostream &JavaTabCodeGen::TRANS_TARGS_WI()
 
 	/* Keep a count of the num of items in the array written. */
 	for ( int t = 0; t < redFsm->transSet.length(); t++ ) {
-		/* Write out the target state. */
+		/* Save the position. Needed for eofTargs. */
 		RedTransAp *trans = transPtrs[t];
+		trans->pos = t;
+
+		/* Write out the target state. */
 		ARRAY_ITEM( INT(trans->targ->id), ( t >= redFsm->transSet.length()-1 ) );
 	}
 	delete[] transPtrs;
diff --git a/rlgen-java/javacodegen.h b/rlgen-java/javacodegen.h
index 9ee24c6..2626153 100644
--- a/rlgen-java/javacodegen.h
+++ b/rlgen-java/javacodegen.h
@@ -150,8 +150,8 @@ public:
 	string SL() { return "_" + DATA_PREFIX() + "single_lengths"; }
 	string RL() { return "_" + DATA_PREFIX() + "range_lengths"; }
 	string A() { return "_" + DATA_PREFIX() + "actions"; }
-	string TA() { return "_" + DATA_PREFIX() + "trans_actions_wi"; }
-	string TT() { return "_" + DATA_PREFIX() + "trans_targs_wi"; }
+	string TA() { return "_" + DATA_PREFIX() + "trans_actions"; }
+	string TT() { return "_" + DATA_PREFIX() + "trans_targs"; }
 	string TSA() { return "_" + DATA_PREFIX() + "to_state_actions"; }
 	string FSA() { return "_" + DATA_PREFIX() + "from_state_actions"; }
 	string EA() { return "_" + DATA_PREFIX() + "eof_actions"; }
diff --git a/rlgen-ruby/ruby-codegen.h b/rlgen-ruby/ruby-codegen.h
index af56318..21034ff 100644
--- a/rlgen-ruby/ruby-codegen.h
+++ b/rlgen-ruby/ruby-codegen.h
@@ -84,8 +84,8 @@ protected:
 	string SL() { return "_" + DATA_PREFIX() + "single_lengths"; }
 	string RL() { return "_" + DATA_PREFIX() + "range_lengths"; }
 	string A() { return "_" + DATA_PREFIX() + "actions"; }
-	string TA() { return "_" + DATA_PREFIX() + "trans_actions_wi"; }
-	string TT() { return "_" + DATA_PREFIX() + "trans_targs_wi"; }
+	string TA() { return "_" + DATA_PREFIX() + "trans_actions"; }
+	string TT() { return "_" + DATA_PREFIX() + "trans_targs"; }
 	string TSA() { return "_" + DATA_PREFIX() + "to_state_actions"; }
 	string FSA() { return "_" + DATA_PREFIX() + "from_state_actions"; }
 	string EA() { return "_" + DATA_PREFIX() + "eof_actions"; }
diff --git a/rlgen-ruby/ruby-flatcodegen.cpp b/rlgen-ruby/ruby-flatcodegen.cpp
index 71dbc9d..a8e7469 100644
--- a/rlgen-ruby/ruby-flatcodegen.cpp
+++ b/rlgen-ruby/ruby-flatcodegen.cpp
@@ -215,8 +215,10 @@ std::ostream &RubyFlatCodeGen::EOF_TRANS()
 	for ( RedStateList::Iter st = redFsm->stateList; st.lte(); st++ ) {
 		/* Write any eof action. */
 		long trans = 0;
-		if ( st->eofTrans != 0 )
-			trans = st->eofTrans->id+1;
+		if ( st->eofTrans != 0 ) {
+			assert( st->eofTrans->pos >= 0 );
+			trans = st->eofTrans->pos+1;
+		}
 
 		/* Write any eof action. */
 		ARRAY_ITEM( INT(trans), ++totalStateNum, st.last() );
@@ -237,8 +239,11 @@ std::ostream &RubyFlatCodeGen::TRANS_TARGS()
 
 	int totalStates = 0;
 	for ( int t = 0; t < redFsm->transSet.length(); t++ ) {
-		/* Write out the target state. */
+		/* Save the position. Needed for eofTargs. */
 		RedTransAp *trans = transPtrs[t];
+		trans->pos = t;
+
+		/* Write out the target state. */
 		ARRAY_ITEM( INT( trans->targ->id ), ++totalStates, t >= redFsm->transSet.length()-1  );
 	}
 	END_ARRAY_LINE();
diff --git a/rlgen-ruby/ruby-tabcodegen.cpp b/rlgen-ruby/ruby-tabcodegen.cpp
index 6ad01c6..8b97697 100644
--- a/rlgen-ruby/ruby-tabcodegen.cpp
+++ b/rlgen-ruby/ruby-tabcodegen.cpp
@@ -673,8 +673,10 @@ std::ostream &RubyTabCodeGen::EOF_TRANS()
 	for ( RedStateList::Iter st = redFsm->stateList; st.lte(); st++ ) {
 		/* Write any eof action. */
 		long trans = 0;
-		if ( st->eofTrans != 0 )
-			trans = st->eofTrans->id+1;
+		if ( st->eofTrans != 0 ) {
+			assert( st->eofTrans->pos >= 0 );
+			trans = st->eofTrans->pos+1;
+		}
 
 		/* Write any eof action. */
 		ARRAY_ITEM( INT(trans), ++totalStateNum, st.last() );
@@ -801,6 +803,14 @@ std::ostream &RubyTabCodeGen::TRANS_TARGS()
 		}
 	}
 
+	for ( RedStateList::Iter st = redFsm->stateList; st.lte(); st++ ) {
+		if ( st->eofTrans != 0 ) {
+			RedTransAp *trans = st->eofTrans;
+			trans->pos = totalTrans;
+			ARRAY_ITEM( KEY( trans->targ->id ), ++totalTrans, false );
+		}
+	}
+
 	/* Output one last number so we don't have to figure out when the last
 	 * entry is and avoid writing a comma. */
 	ARRAY_ITEM( INT(0), ++totalTrans, true );
@@ -833,6 +843,13 @@ std::ostream &RubyTabCodeGen::TRANS_ACTIONS()
 		}
 	}
 
+	for ( RedStateList::Iter st = redFsm->stateList; st.lte(); st++ ) {
+		if ( st->eofTrans != 0 ) {
+			RedTransAp *trans = st->eofTrans;
+			ARRAY_ITEM( INT(TRANS_ACTION( trans )), ++totalTrans, false );
+		}
+	}
+
 	/* Output one last number so we don't have to figure out when the last
 	 * entry is and avoid writing a comma. */
 	ARRAY_ITEM( INT(0), ++totalTrans, true );
@@ -851,8 +868,11 @@ std::ostream &RubyTabCodeGen::TRANS_TARGS_WI()
 	START_ARRAY_LINE();
 	int totalStates = 0;
 	for ( int t = 0; t < redFsm->transSet.length(); t++ ) {
-		/* Write out the target state. */
+		/* Save the position. Needed for eofTargs. */
 		RedTransAp *trans = transPtrs[t];
+		trans->pos = t;
+
+		/* Write out the target state. */
 		ARRAY_ITEM( INT(trans->targ->id), ++totalStates, ( t >= redFsm->transSet.length()-1 ) );
 	}
 	END_ARRAY_LINE();
diff --git a/test/scan2.rl b/test/scan2.rl
new file mode 100644
index 0000000..a1ae959
--- /dev/null
+++ b/test/scan2.rl
@@ -0,0 +1,34 @@
+/*
+ * @LANG: indep
+ */
+ptr ts;
+ptr te;
+int act;
+int token;
+%%
+%%{
+	machine scanner;
+
+	# Warning: changing the patterns or the input string will affect the
+	# coverage of the scanner action types.
+	main := |*
+        'a' => {
+			prints "pat1\n";
+		};
+
+        [ab]+ . 'c' => {
+			prints "pat2\n";
+		};
+
+        any => {
+			prints "any\n";
+		};
+	*|;
+}%%
+/* _____INPUT_____
+"a"
+_____INPUT_____ */
+/* _____OUTPUT_____
+pat1
+ACCEPT
+_____OUTPUT_____ */
diff --git a/test/scan3.rl b/test/scan3.rl
new file mode 100644
index 0000000..ca1a136
--- /dev/null
+++ b/test/scan3.rl
@@ -0,0 +1,32 @@
+/*
+ * @LANG: indep
+ */
+ptr ts;
+ptr te;
+int act;
+int token;
+%%
+%%{
+	machine scanner;
+
+	# Warning: changing the patterns or the input string will affect the
+	# coverage of the scanner action types.
+	main := |*
+		'a' => {
+			prints "pat1\n";
+		};
+		'b' => {
+			prints "pat2\n";
+		};
+        [ab] any*  => {
+			prints "pat3\n";
+		};
+	*|;
+}%%
+/* _____INPUT_____
+"ab89"
+_____INPUT_____ */
+/* _____OUTPUT_____
+pat3
+ACCEPT
+_____OUTPUT_____ */
-- 
2.7.4