Examples with ExprNodeConstantDesc org.apache.hadoop.hive.ql.plan.ExprNodeConstantDesc used on opensource projects

Example 96 with ExprNodeConstantDesc

use of org.apache.hadoop.hive.ql.plan.ExprNodeConstantDesc in project hive by apache.

the class TestOrcSplitElimination method testFooterExternalCacheImpl.

private void testFooterExternalCacheImpl(boolean isPpd) throws IOException {
    ObjectInspector inspector = createIO();
    writeFile(inspector, testFilePath);
    writeFile(inspector, testFilePath2);
    GenericUDF udf = new GenericUDFOPEqualOrLessThan();
    List<ExprNodeDesc> childExpr = Lists.newArrayList();
    createTestSarg(inspector, udf, childExpr);
    setupExternalCacheConfig(isPpd, testFilePath + "," + testFilePath2);
    // Get the base values w/o cache.
    conf.setBoolean(ConfVars.HIVE_ORC_MS_FOOTER_CACHE_ENABLED.varname, false);
    OrcInputFormatForTest.clearLocalCache();
    OrcInputFormat in0 = new OrcInputFormat();
    InputSplit[] originals = in0.getSplits(conf, -1);
    assertEquals(10, originals.length);
    HashSet<FsWithHash> originalHs = new HashSet<>();
    for (InputSplit original : originals) {
        originalHs.add(new FsWithHash((FileSplit) original));
    }
    // Populate the cache.
    conf.setBoolean(ConfVars.HIVE_ORC_MS_FOOTER_CACHE_ENABLED.varname, true);
    OrcInputFormatForTest in = new OrcInputFormatForTest();
    OrcInputFormatForTest.clearLocalCache();
    OrcInputFormatForTest.caches.resetCounts();
    OrcInputFormatForTest.caches.cache.clear();
    InputSplit[] splits = in.getSplits(conf, -1);
    // Puts, gets, hits, unused, unused.
    @SuppressWarnings("static-access") AtomicInteger[] counts = { in.caches.putCount, isPpd ? in.caches.getByExprCount : in.caches.getCount, isPpd ? in.caches.getHitByExprCount : in.caches.getHitCount, isPpd ? in.caches.getCount : in.caches.getByExprCount, isPpd ? in.caches.getHitCount : in.caches.getHitByExprCount };
    verifySplits(originalHs, splits);
    verifyCallCounts(counts, 2, 2, 0);
    assertEquals(2, OrcInputFormatForTest.caches.cache.size());
    // Verify we can get from cache.
    OrcInputFormatForTest.clearLocalCache();
    OrcInputFormatForTest.caches.resetCounts();
    splits = in.getSplits(conf, -1);
    verifySplits(originalHs, splits);
    verifyCallCounts(counts, 0, 2, 2);
    // Verify ORC SARG still works.
    OrcInputFormatForTest.clearLocalCache();
    OrcInputFormatForTest.caches.resetCounts();
    childExpr.set(1, new ExprNodeConstantDesc(5));
    conf.set("hive.io.filter.expr.serialized", SerializationUtilities.serializeExpression(new ExprNodeGenericFuncDesc(inspector, udf, childExpr)));
    splits = in.getSplits(conf, -1);
    InputSplit[] filtered = { originals[0], originals[4], originals[5], originals[9] };
    originalHs = new HashSet<>();
    for (InputSplit original : filtered) {
        originalHs.add(new FsWithHash((FileSplit) original));
    }
    verifySplits(originalHs, splits);
    verifyCallCounts(counts, 0, 2, 2);
    // Verify corrupted cache value gets replaced.
    OrcInputFormatForTest.clearLocalCache();
    OrcInputFormatForTest.caches.resetCounts();
    Map.Entry<Long, MockExternalCaches.MockItem> e = OrcInputFormatForTest.caches.cache.entrySet().iterator().next();
    Long key = e.getKey();
    byte[] someData = new byte[8];
    ByteBuffer toCorrupt = e.getValue().data;
    System.arraycopy(toCorrupt.array(), toCorrupt.arrayOffset(), someData, 0, someData.length);
    toCorrupt.putLong(0, 0L);
    splits = in.getSplits(conf, -1);
    verifySplits(originalHs, splits);
    if (!isPpd) {
        // Recovery is not implemented yet for PPD path.
        ByteBuffer restored = OrcInputFormatForTest.caches.cache.get(key).data;
        byte[] newData = new byte[someData.length];
        System.arraycopy(restored.array(), restored.arrayOffset(), newData, 0, newData.length);
        assertArrayEquals(someData, newData);
    }
}

Example 97 with ExprNodeConstantDesc

use of org.apache.hadoop.hive.ql.plan.ExprNodeConstantDesc in project hive by apache.

the class TestParquetRowGroupFilter method testRowGroupFilterTakeEffect.

@Test
public void testRowGroupFilterTakeEffect() throws Exception {
    // define schema
    columnNames = "intCol";
    columnTypes = "int";
    StructObjectInspector inspector = getObjectInspector(columnNames, columnTypes);
    MessageType fileSchema = MessageTypeParser.parseMessageType("message hive_schema {\n" + "  optional int32 intCol;\n" + "}\n");
    conf.set(ColumnProjectionUtils.READ_COLUMN_NAMES_CONF_STR, "intCol");
    conf.set("columns", "intCol");
    conf.set("columns.types", "int");
    // create Parquet file with specific data
    Path testPath = writeDirect("RowGroupFilterTakeEffect", fileSchema, new DirectWriter() {

        @Override
        public void write(RecordConsumer consumer) {
            for (int i = 0; i < 100; i++) {
                consumer.startMessage();
                consumer.startField("int", 0);
                consumer.addInteger(i);
                consumer.endField("int", 0);
                consumer.endMessage();
            }
        }
    });
    // > 50
    GenericUDF udf = new GenericUDFOPGreaterThan();
    List<ExprNodeDesc> children = Lists.newArrayList();
    ExprNodeColumnDesc columnDesc = new ExprNodeColumnDesc(Integer.class, "intCol", "T", false);
    ExprNodeConstantDesc constantDesc = new ExprNodeConstantDesc(50);
    children.add(columnDesc);
    children.add(constantDesc);
    ExprNodeGenericFuncDesc genericFuncDesc = new ExprNodeGenericFuncDesc(inspector, udf, children);
    String searchArgumentStr = SerializationUtilities.serializeExpression(genericFuncDesc);
    conf.set(TableScanDesc.FILTER_EXPR_CONF_STR, searchArgumentStr);
    ParquetRecordReaderWrapper recordReader = (ParquetRecordReaderWrapper) new MapredParquetInputFormat().getRecordReader(new FileSplit(testPath, 0, fileLength(testPath), (String[]) null), conf, null);
    Assert.assertEquals("row group is not filtered correctly", 1, recordReader.getFiltedBlocks().size());
    // > 100
    constantDesc = new ExprNodeConstantDesc(100);
    children.set(1, constantDesc);
    genericFuncDesc = new ExprNodeGenericFuncDesc(inspector, udf, children);
    searchArgumentStr = SerializationUtilities.serializeExpression(genericFuncDesc);
    conf.set(TableScanDesc.FILTER_EXPR_CONF_STR, searchArgumentStr);
    recordReader = (ParquetRecordReaderWrapper) new MapredParquetInputFormat().getRecordReader(new FileSplit(testPath, 0, fileLength(testPath), (String[]) null), conf, null);
    Assert.assertEquals("row group is not filtered correctly", 0, recordReader.getFiltedBlocks().size());
}

Example 98 with ExprNodeConstantDesc

use of org.apache.hadoop.hive.ql.plan.ExprNodeConstantDesc in project hive by apache.

the class TestUtilities method testSerializeTimestamp.

@Test
public void testSerializeTimestamp() {
    Timestamp ts = Timestamp.ofEpochMilli(1374554702000L, 123456);
    ExprNodeConstantDesc constant = new ExprNodeConstantDesc(ts);
    List<ExprNodeDesc> children = new ArrayList<ExprNodeDesc>(1);
    children.add(constant);
    ExprNodeGenericFuncDesc desc = new ExprNodeGenericFuncDesc(TypeInfoFactory.timestampTypeInfo, new GenericUDFFromUtcTimestamp(), children);
    assertEquals(desc.getExprString(), SerializationUtilities.deserializeExpression(SerializationUtilities.serializeExpression(desc)).getExprString());
}

Example 99 with ExprNodeConstantDesc

use of org.apache.hadoop.hive.ql.plan.ExprNodeConstantDesc in project hive by apache.

the class MapJoinTestConfig method makeInterceptSelectOperator.

private static Operator<SelectDesc> makeInterceptSelectOperator(MapJoinOperator mapJoinOperator, int bigTableKeySize, int bigTableRetainSize, String[] outputColumnNames, TypeInfo[] outputTypeInfos) {
    MapJoinDesc mapJoinDesc = (MapJoinDesc) mapJoinOperator.getConf();
    List<ExprNodeDesc> selectExprList = new ArrayList<ExprNodeDesc>();
    List<String> selectOutputColumnNameList = new ArrayList<String>();
    for (int i = 0; i < bigTableRetainSize; i++) {
        String selectOutputColumnName = HiveConf.getColumnInternalName(i);
        selectOutputColumnNameList.add(selectOutputColumnName);
        TypeInfo outputTypeInfo = outputTypeInfos[i];
        if (i < bigTableKeySize) {
            // Big Table key.
            ExprNodeColumnDesc keyColumnExpr = new ExprNodeColumnDesc(outputTypeInfo, outputColumnNames[i], "test", false);
            selectExprList.add(keyColumnExpr);
        } else {
            // For row-mode, substitute NULL constant for any non-key extra Big Table columns.
            ExprNodeConstantDesc nullExtraColumnExpr = new ExprNodeConstantDesc(outputTypeInfo, null);
            nullExtraColumnExpr.setFoldedFromCol(outputColumnNames[i]);
            selectExprList.add(nullExtraColumnExpr);
        }
    }
    SelectDesc selectDesc = new SelectDesc(selectExprList, selectOutputColumnNameList);
    Operator<SelectDesc> selectOperator = OperatorFactory.get(new CompilationOpContext(), selectDesc);
    return selectOperator;
}

Example 100 with ExprNodeConstantDesc

use of org.apache.hadoop.hive.ql.plan.ExprNodeConstantDesc in project hive by apache.

the class TestDecimalStringValidation method params.

@Parameterized.Parameters(name = "{index}: {0}")
public static Collection<FunctionCall> params() throws Exception {
    ExprNodeDesc[] characterExps = new ExprNodeDesc[] { new ExprNodeColumnDesc(TypeInfoFactory.varcharTypeInfo, "varchar_col", null, false), new ExprNodeColumnDesc(TypeInfoFactory.charTypeInfo, "char_col", null, false), new ExprNodeColumnDesc(TypeInfoFactory.stringTypeInfo, "string_col", null, false), new ExprNodeConstantDesc(TypeInfoFactory.varcharTypeInfo, "123.3"), new ExprNodeConstantDesc(TypeInfoFactory.charTypeInfo, "123.3"), new ExprNodeConstantDesc(TypeInfoFactory.stringTypeInfo, "123.3") };
    ExprNodeDesc[] numericExps = new ExprNodeDesc[] { new ExprNodeColumnDesc(TypeInfoFactory.decimalTypeInfo, "decimal_col", null, false), new ExprNodeConstantDesc(TypeInfoFactory.decimalTypeInfo, 123.3) };
    FunctionInfo[] functions = new FunctionInfo[] { FunctionRegistry.getFunctionInfo("="), FunctionRegistry.getFunctionInfo("<"), FunctionRegistry.getFunctionInfo(">"), FunctionRegistry.getFunctionInfo("<>"), FunctionRegistry.getFunctionInfo("<="), FunctionRegistry.getFunctionInfo(">="), FunctionRegistry.getFunctionInfo("<=>") };
    Collection<FunctionCall> input = new ArrayList<>();
    for (ExprNodeDesc chrExp : characterExps) {
        for (ExprNodeDesc numExp : numericExps) {
            for (FunctionInfo function : functions) {
                input.add(new FunctionCall(chrExp, numExp, function));
                input.add(new FunctionCall(numExp, chrExp, function));
            }
        }
    }
    return input;
}