label+='foo' wasn't generating good code. It was generating token type as variable name. Now, I gen "s<ttype>" for implicit labels on string literals. Augmented unit test. Fixes antlr/antlr4#90

CHANGES.txt

@@ -11,6 +11,9 @@ December 1, 2012
   some tests in TestSets appeared to allow ~('a'|B) but it was randomly working.
   ('a'|B) works, though doesn't collapse to a set.
 
+* label+='foo' wasn't generating good code. It was generating token type as
+  variable name. Now, I gen "s<ttype>" for implicit labels on string literals.
+
 November 30, 2012
 
 * Maven updates (cleanup, unification, and specify Java 6 bootstrap classpath)

tool/src/org/antlr/v4/codegen/Target.java

@@ -370,10 +370,14 @@ public class Target {
 		return Utils.capitalize(r.name)+gen.templates.getInstanceOf("RuleContextNameSuffix").render();
 	}
 
-	// should be same for all refs to same token like $ID within single rule function
+	// should be same for all refs to same token like ctx.ID within single rule function
+	// for literals like 'while', we gen _s<ttype>
 	public String getImplicitTokenLabel(String tokenName) {
 		ST st = gen.templates.getInstanceOf("ImplicitTokenLabel");
 		int ttype = gen.g.getTokenType(tokenName);
+		if ( tokenName.startsWith("'") ) {
+			return "s"+ttype;
+		}
 		String text = getTokenTypeAsTargetLabel(gen.g, ttype);
 		st.add("tokenName", text);
 		return st.render();

tool/test/org/antlr/v4/test/TestParserExec.java

@@ -37,14 +37,14 @@ public class TestParserExec extends BaseTest {
 	@Test public void testLabels() throws Exception {
 		String grammar =
 			"grammar T;\n" +
-			"a : b1=b b2+=b*;\n" +
+			"a : b1=b b2+=b* b3+=';' ;\n" +
 			"b : id=ID val+=INT*;\n" +
 			"ID : 'a'..'z'+ ;\n" +
 			"INT : '0'..'9'+;\n" +
 			"WS : (' '|'\\n') -> skip ;\n";
 
 		String found = execParser("T.g4", grammar, "TParser", "TLexer", "a",
-								  "abc 34", false);
+								  "abc 34;", false);
 		assertEquals("", found);
 		assertEquals(null, stderrDuringParse);
 	}