Method testJoinSchema

test/org/apache/pig/test/TestJoinBase.java:102–157

()

    @Test
    public void testJoinSchema() throws Exception {
        String[] input1 = {
                "1\t2",
                "2\t3",
                "3\t4"
        };
        String[] input2 = {
                "1\thello",
                "4\tbye",
        };

        String firstInput = createInputFile("a.txt", input1);
        String secondInput = createInputFile("b.txt", input2);
        Tuple expectedResult = (Tuple)Util.getPigConstant("(1,2,1,'hello',1,2,1,'hello')");

        // with schema
        String script = "a = load '"+ Util.encodeEscape(firstInput) +"' as (i:int, j:int); " +
                "b = load '"+ Util.encodeEscape(secondInput) +"' as (k:int, l:chararray); " +
                "c = join a by $0, b by $0;" +
                "d = foreach c generate i,j,k,l,a::i as ai,a::j as aj,b::k as bk,b::l as bl;";
        Util.registerMultiLineQuery(pigServer, script);
        Iterator<Tuple> it = pigServer.openIterator("d");
        assertTrue(it.hasNext());
        assertEquals(expectedResult, it.next());
        assertFalse(it.hasNext());

        // schema with duplicates
        script = "a = load '"+ Util.encodeEscape(firstInput) +"' as (i:int, j:int); " +
        "b = load '"+ Util.encodeEscape(secondInput) +"' as (i:int, l:chararray); " +
        "c = join a by $0, b by $0;" +
        "d = foreach c generate i,j,l,a::i as ai,a::j as aj,b::i as bi,b::l as bl;";
        boolean exceptionThrown = false;
        try{
            Util.registerMultiLineQuery(pigServer, script);
            pigServer.openIterator("d");
        }catch (Exception e) {
            PigException pe = LogUtils.getPigException(e);
            assertEquals(1025, pe.getErrorCode());
            exceptionThrown = true;
        }
        assertTrue(exceptionThrown);

        // schema with duplicates with resolution
        script = "a = load '"+ Util.encodeEscape(firstInput) +"' as (i:int, j:int); " +
        "b = load '"+ Util.encodeEscape(secondInput) +"' as (i:int, l:chararray); " +
        "c = join a by $0, b by $0;" +
        "d = foreach c generate a::i as ai1,j,b::i as bi1,l,a::i as ai2,a::j as aj2,b::i as bi3,b::l as bl3;";
        Util.registerMultiLineQuery(pigServer, script);
        it = pigServer.openIterator("d");
        assertTrue(it.hasNext());
        assertEquals(expectedResult, it.next());
        assertFalse(it.hasNext());
        deleteInputFile(firstInput);
        deleteInputFile(secondInput);
    }

    @Test