testrewriteast.java

ANTLR(ANother Tool for Language Recognition)它是这样的一种工具

			"grammar T;\n" +
			"options {output=AST;}\n" +
			"tokens {BLOCK;}\n" +
			"a : lc='{' ID+ '}' -> ^(BLOCK[$lc,\"block\"] ID+) ;\n" +
			"type : 'int' | 'float' ;\n" +
			"ID : 'a'..'z'+ ;\n" +
			"INT : '0'..'9'+;\n" +
			"WS : (' '|'\\n') {channel=99;} ;\n";
		String found = execParser("t.g", grammar, "TParser", "TLexer",
				    "a", "{a b c}", debug);
		assertEquals("(block a b c)\n", found);
	}

	public void testMixedRewriteAndAutoAST() throws Exception {
		String grammar =
			"grammar T;\n" +
			"options {output=AST;}\n" +
			"tokens {BLOCK;}\n" +
			"a : b b^ ;\n" + // 2nd b matches only an INT; can make it root
			"b : ID INT -> INT ID\n" +
			"  | INT\n" +
			"  ;\n" +
			"ID : 'a'..'z'+ ;\n" +
			"INT : '0'..'9'+;\n" +
			"WS : (' '|'\\n') {channel=99;} ;\n";
		String found = execParser("t.g", grammar, "TParser", "TLexer",
				    "a", "a 1 2", debug);
		assertEquals("(2 1 a)\n", found);
	}

	public void testSubruleWithRewrite() throws Exception {
		String grammar =
			"grammar T;\n" +
			"options {output=AST;}\n" +
			"tokens {BLOCK;}\n" +
			"a : b b ;\n" +
			"b : (ID INT -> INT ID | INT INT -> INT+ )\n" +
			"  ;\n" +
			"ID : 'a'..'z'+ ;\n" +
			"INT : '0'..'9'+;\n" +
			"WS : (' '|'\\n') {channel=99;} ;\n";
		String found = execParser("t.g", grammar, "TParser", "TLexer",
				    "a", "a 1 2 3", debug);
		assertEquals("1 a 2 3\n", found);
	}

	public void testSubruleWithRewriteReferencingPreviousElement() throws Exception {
		String grammar =
			"grammar T;\n" +
			"options {output=AST;}\n" +
			"tokens {TYPE;}\n" +
			"a : b b ;\n" +
			"b : 'int'\n" +
			"    ( ID -> ^(TYPE 'int' ID)\n" +
			"    | ID '=' INT -> ^(TYPE 'int' ID INT)\n" +
			"    )\n" +
			"    ';'\n" +
			"  ;\n" +
			"ID : 'a'..'z'+ ;\n" +
			"INT : '0'..'9'+;\n" +
			"WS : (' '|'\\n') {channel=99;} ;\n";
		String found = execParser("t.g", grammar, "TParser", "TLexer",
				    "a", "int a; int b=3;", debug);
		assertEquals("(TYPE int a) (TYPE int b 3)\n", found);
	}

	public void testNestedRewriteShutsOffAutoAST() throws Exception {
		String grammar =
			"grammar T;\n" +
			"options {output=AST;}\n" +
			"tokens {BLOCK;}\n" +
			"a : b b ;\n" +
			"b : ID ( ID (last=ID -> $last)+ ) ';'\n" + // get last ID
			"  | INT\n" + // should still get auto AST construction
			"  ;\n" +
			"ID : 'a'..'z'+ ;\n" +
			"INT : '0'..'9'+;\n" +
			"WS : (' '|'\\n') {channel=99;} ;\n";
		String found = execParser("t.g", grammar, "TParser", "TLexer",
				    "a", "a b c d; 42", debug);
		assertEquals("d 42\n", found);
	}

	public void testRewriteActions() throws Exception {
		String grammar =
			"grammar T;\n" +
			"options {output=AST;}\n" +
			"a : atom -> ^({adaptor.create(INT,\"9\")} atom) ;\n" +
			"atom : INT ;\n" +
			"ID : 'a'..'z'+ ;\n" +
			"INT : '0'..'9'+;\n" +
			"WS : (' '|'\\n') {channel=99;} ;\n";
		String found = execParser("t.g", grammar, "TParser", "TLexer",
				    "a", "3", debug);
		assertEquals("(9 3)\n", found);
	}

	public void testRewriteActions2() throws Exception {
		String grammar =
			"grammar T;\n" +
			"options {output=AST;}\n" +
			"a : atom -> {adaptor.create(INT,\"9\")} atom ;\n" +
			"atom : INT ;\n" +
			"ID : 'a'..'z'+ ;\n" +
			"INT : '0'..'9'+;\n" +
			"WS : (' '|'\\n') {channel=99;} ;\n";
		String found = execParser("t.g", grammar, "TParser", "TLexer",
				    "a", "3", debug);
		assertEquals("9 3\n", found);
	}

	public void testRefToOldValue() throws Exception {
		String grammar =
			"grammar T;\n" +
			"options {output=AST;}\n" +
			"tokens {BLOCK;}\n" +
			"a : (atom -> atom) (op='+' r=atom -> ^($op $a $r) )* ;\n" +
			"atom : INT ;\n" +
			"ID : 'a'..'z'+ ;\n" +
			"INT : '0'..'9'+;\n" +
			"WS : (' '|'\\n') {channel=99;} ;\n";
		String found = execParser("t.g", grammar, "TParser", "TLexer",
				    "a", "3+4+5", debug);
		assertEquals("(+ (+ 3 4) 5)\n", found);
	}

	public void testCopySemanticsForRules() throws Exception {
		String grammar =
			"grammar T;\n" +
			"options {output=AST;}\n" +
			"tokens {BLOCK;}\n" +
			"a : atom -> ^(atom atom) ;\n" + // NOT CYCLE! (dup atom)
			"atom : INT ;\n" +
			"ID : 'a'..'z'+ ;\n" +
			"INT : '0'..'9'+;\n" +
			"WS : (' '|'\\n') {channel=99;} ;\n";
		String found = execParser("t.g", grammar, "TParser", "TLexer",
				    "a", "3", debug);
		assertEquals("(3 3)\n", found);
	}

	public void testComplicatedMelange() throws Exception {
		String grammar =
			"grammar T;\n" +
			"options {output=AST;}\n" +
			"tokens {BLOCK;}\n" +
			"a : A A b=B B b=B c+=C C c+=C D {$D.text;} -> A+ B+ C+ D ;\n" +
			"type : 'int' | 'float' ;\n" +
			"A : 'a' ;\n" +
			"B : 'b' ;\n" +
			"C : 'c' ;\n" +
			"D : 'd' ;\n" +
			"WS : (' '|'\\n') {channel=99;} ;\n";
		String found = execParser("t.g", grammar, "TParser", "TLexer",
				    "a", "a a b b b c c c d", debug);
		assertEquals("a a b b b c c c d\n", found);
	}

	public void testRuleListLabel() throws Exception {
		String grammar =
			"grammar T;\n" +
			"options {output=AST;}\n" +
			"tokens {BLOCK;}\n" +
			"a : x+=b x+=b -> $x+;\n"+
			"b : ID ;\n"+
			"ID : 'a'..'z'+ ;\n" +
			"WS : (' '|'\\n') {channel=99;} ;\n";
		String found = execParser("t.g", grammar, "TParser", "TLexer",
				    "a", "a b", debug);
		assertEquals("a b\n", found);
	}

	public void testOptional() throws Exception {
		String grammar =
			"grammar T;\n" +
			"options {output=AST;}\n" +
			"tokens {BLOCK;}\n" +
			"a : x=b (y=b)? -> $x $y?;\n"+
			"b : ID ;\n"+
			"ID : 'a'..'z'+ ;\n" +
			"WS : (' '|'\\n') {channel=99;} ;\n";
		String found = execParser("t.g", grammar, "TParser", "TLexer",
				    "a", "a", debug);
		assertEquals("a\n", found);
	}

	public void testOptional2() throws Exception {
		String grammar =
			"grammar T;\n" +
			"options {output=AST;}\n" +
			"tokens {BLOCK;}\n" +
			"a : x=ID (y=b)? -> $x $y?;\n"+
			"b : ID ;\n"+
			"ID : 'a'..'z'+ ;\n" +
			"WS : (' '|'\\n') {channel=99;} ;\n";
		String found = execParser("t.g", grammar, "TParser", "TLexer",
				    "a", "a b", debug);
		assertEquals("a b\n", found);
	}

	public void testOptional3() throws Exception {
		String grammar =
			"grammar T;\n" +
			"options {output=AST;}\n" +
			"tokens {BLOCK;}\n" +
			"a : x=ID (y=b)? -> ($x $y)?;\n"+
			"b : ID ;\n"+
			"ID : 'a'..'z'+ ;\n" +
			"WS : (' '|'\\n') {channel=99;} ;\n";
		String found = execParser("t.g", grammar, "TParser", "TLexer",
				    "a", "a b", debug);
		assertEquals("a b\n", found);
	}

	public void testOptional4() throws Exception {
		String grammar =
			"grammar T;\n" +
			"options {output=AST;}\n" +
			"tokens {BLOCK;}\n" +
			"a : x+=ID (y=b)? -> ($x $y)?;\n"+
			"b : ID ;\n"+
			"ID : 'a'..'z'+ ;\n" +
			"WS : (' '|'\\n') {channel=99;} ;\n";
		String found = execParser("t.g", grammar, "TParser", "TLexer",
				    "a", "a b", debug);
		assertEquals("a b\n", found);
	}

	public void testArbitraryExprType() throws Exception {
		String grammar =
			"grammar T;\n" +
			"options {output=AST;}\n" +
			"tokens {BLOCK;}\n" +
			"a : x+=b x+=b -> {new Object()};\n"+
			"b : ID ;\n"+
			"ID : 'a'..'z'+ ;\n" +
			"WS : (' '|'\\n') {channel=99;} ;\n";
		String found = execParser("t.g", grammar, "TParser", "TLexer",
				    "a", "a b", debug);
		//assertEquals("[not sure what this should be!]\n", found);
//ATTENTION: I changed this one's behavior from the above.  Is it right?
		assertEquals("nil\n", found);
	}

	// E R R O R S

	public void testUnknownRule() throws Exception {
		ErrorQueue equeue = new ErrorQueue();
		ErrorManager.setErrorListener(equeue);

		String grammar =
			"grammar T;\n" +
			"options {output=AST;}\n" +
			"a : INT -> ugh ;\n" +
			"ID : 'a'..'z'+ ;\n" +
			"INT : '0'..'9'+;\n" +
			"WS : (' '|'\\n') {channel=99;} ;\n";

		Grammar g = new Grammar(grammar);
		Tool antlr = newTool();
		antlr.setOutputDirectory(null); // write to /dev/null
		CodeGenerator generator = new CodeGenerator(antlr, g, "Java");
		g.setCodeGenerator(generator);
		generator.genRecognizer();

		int expectedMsgID = ErrorManager.MSG_UNDEFINED_RULE_REF;
		Object expectedArg = "ugh";
		Object expectedArg2 = null;
		GrammarSemanticsMessage expectedMessage =
			new GrammarSemanticsMessage(expectedMsgID, g, null, expectedArg, expectedArg2);

		checkError(equeue, expectedMessage);
	}

	public void testKnownRuleButNotInLHS() throws Exception {
		ErrorQueue equeue = new ErrorQueue();
		ErrorManager.setErrorListener(equeue);

		String grammar =
			"grammar T;\n" +
			"options {output=AST;}\n" +
			"a : INT -> b ;\n" +
			"b : 'b' ;\n" +
			"ID : 'a'..'z'+ ;\n" +
			"INT : '0'..'9'+;\n" +
			"WS : (' '|'\\n') {channel=99;} ;\n";

		Grammar g = new Grammar(grammar);
		Tool antlr = newTool();
		antlr.setOutputDirectory(null); // write to /dev/null
		CodeGenerator generator = new CodeGenerator(antlr, g, "Java");
		g.setCodeGenerator(generator);
		generator.genRecognizer();

		int expectedMsgID = ErrorManager.MSG_REWRITE_ELEMENT_NOT_PRESENT_ON_LHS;
		Object expectedArg = "b";
		Object expectedArg2 = null;
		GrammarSemanticsMessage expectedMessage =
			new GrammarSemanticsMessage(expectedMsgID, g, null, expectedArg, expectedArg2);

		checkError(equeue, expectedMessage);
	}

	public void testUnknownToken() throws Exception {
		ErrorQueue equeue = new ErrorQueue();
		ErrorManager.setErrorListener(equeue);

		String grammar =
			"grammar T;\n" +
			"options {output=AST;}\n" +
			"a : INT -> ICK ;\n" +
			"ID : 'a'..'z'+ ;\n" +
			"INT : '0'..'9'+;\n" +
			"WS : (' '|'\\n') {channel=99;} ;\n";

		Grammar g = new Grammar(grammar);
		Tool antlr = newTool();
		antlr.setOutputDirectory(null); // write to /dev/null
		CodeGenerator generator = new CodeGenerator(antlr, g, "Java");
		g.setCodeGenerator(generator);
		generator.genRecognizer();

		int expectedMsgID = ErrorManager.MSG_UNDEFINED_TOKEN_REF_IN_REWRITE;
		Object expectedArg = "ICK";
		Object expectedArg2 = null;
		GrammarSemanticsMessage expectedMessage =
			new GrammarSemanticsMessage(expectedMsgID, g, null, expectedArg, expectedArg2);

		checkError(equeue, expectedMessage);
	}

	public void testUnknownLabel() throws Exception {
		ErrorQueue equeue = new ErrorQueue();
		ErrorManager.setErrorListener(equeue);

		String grammar =
			"grammar T;\n" +
			"options {output=AST;}\n" +
			"a : INT -> $foo ;\n" +
			"ID : 'a'..'z'+ ;\n" +
			"INT : '0'..'9'+;\n" +
			"WS : (' '|'\\n') {channel=99;} ;\n";

		Grammar g = new Grammar(grammar);
		Tool antlr = newTool();
		antlr.setOutputDirectory(null); // write to /dev/null
		CodeGenerator generator = new CodeGenerator(antlr, g, "Java");
		g.setCodeGenerator(generator);
		generator.genRecognizer();

		int expectedMsgID = ErrorManager.MSG_UNDEFINED_LABEL_REF_IN_REWRITE;
		Object expectedArg = "foo";
		Object expectedArg2 = null;
		GrammarSemanticsMessage expectedMessage =
			new GrammarSemanticsMessage(expectedMsgID, g, null, expectedArg, expectedArg2);

		checkError(equeue, expectedMessage);
	}

	public void testUnknownCharLiteralToken() throws Exception {
		ErrorQueue equeue = new ErrorQueue();
		ErrorManager.setErrorListener(equeue);

		String grammar =
			"grammar T;\n" +
			"options {output=AST;}\n" +
			"a : INT -> 'a' ;\n" +
			"ID : 'a'..'z'+ ;\n" +
			"INT : '0'..'9'+;\n" +
			"WS : (' '|'\\n') {channel=99;} ;\n";

		Grammar g = new Grammar(grammar);
		Tool antlr = newTool();
		antlr.setOutputDirectory(null); // write to /dev/null
		CodeGenerator generator = new CodeGenerator(antlr, g, "Java");
		g.setCodeGenerator(generator);
		generator.genRecognizer();

		int expectedMsgID = ErrorManager.MSG_UNDEFINED_TOKEN_REF_IN_REWRITE;
		Object expectedArg = "'a'";
		Object expectedArg2 = null;
		GrammarSemanticsMessage expectedMessage =
			new GrammarSemanticsMessage(expectedMsgID, g, null, expectedArg, expectedArg2);

		checkError(equeue, expectedMessage);
	}

	public void testUnknownStringLiteralToken() throws Exception {
		ErrorQueue equeue = new ErrorQueue();
		ErrorManager.setErrorListener(equeue);

		String grammar =
			"grammar T;\n" +
			"options {output=AST;}\n" +
			"a : INT -> 'foo' ;\n" +
			"ID : 'a'..'z'+ ;\n" +
			"INT : '0'..'9'+;\n" +
			"WS : (' '|'\\n') {channel=99;} ;\n";

		Grammar g = new Grammar(grammar);
		Tool antlr = newTool();
		antlr.setOutputDirectory(null); // write to /dev/null
		CodeGenerator generator = new CodeGenerator(antlr, g, "Java");
		g.setCodeGenerator(generator);
		generator.genRecognizer();

		int expectedMsgID = ErrorManager.MSG_UNDEFINED_TOKEN_REF_IN_REWRITE;
		Object expectedArg = "'foo'";
		Object expectedArg2 = null;
		GrammarSemanticsMessage expectedMessage =
			new GrammarSemanticsMessage(expectedMsgID, g, null, expectedArg, expectedArg2);

		checkError(equeue, expectedMessage);
	}

	// S U P P O R T

	protected void checkError(ErrorQueue equeue,
							  GrammarSemanticsMessage expectedMessage)
		throws Exception
	{
		//System.out.println("errors="+equeue);
		Message foundMsg = null;
		for (int i = 0; i < equeue.errors.size(); i++) {
			Message m = (Message)equeue.errors.get(i);
			if (m.msgID==expectedMessage.msgID ) {
				foundMsg = m;
			}
		}
		assertTrue("no error; "+expectedMessage.msgID+" expected", equeue.errors.size()>0);
		assertTrue("too many errors; "+equeue.errors, equeue.errors.size()<=1);
		assertNotNull("couldn't find expected error: "+expectedMessage.msgID, foundMsg);
		assertTrue("error is not a GrammarSemanticsMessage",
				   foundMsg instanceof GrammarSemanticsMessage);
		assertEquals(expectedMessage.arg, foundMsg.arg);
		assertEquals(expectedMessage.arg2, foundMsg.arg2);
		ErrorManager.resetErrorState(); // wack errors for next test
	}

}