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

Example 11 with MapWork

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

the class TestUtilities method runTestGetInputSummary.

private ContentSummary runTestGetInputSummary(JobConf jobConf, Properties properties, int numOfPartitions, int bytesPerFile, Class<? extends InputFormat> inputFormatClass) throws IOException {
    // creates scratch directories needed by the Context object
    SessionState.start(new HiveConf());
    MapWork mapWork = new MapWork();
    Context context = new Context(jobConf);
    LinkedHashMap<Path, PartitionDesc> pathToPartitionInfo = new LinkedHashMap<>();
    LinkedHashMap<Path, ArrayList<String>> pathToAliasTable = new LinkedHashMap<>();
    TableScanOperator scanOp = new TableScanOperator();
    PartitionDesc partitionDesc = new PartitionDesc(new TableDesc(inputFormatClass, null, properties), null);
    String testTableName = "testTable";
    Path testTablePath = new Path(testTableName);
    Path[] testPartitionsPaths = new Path[numOfPartitions];
    for (int i = 0; i < numOfPartitions; i++) {
        String testPartitionName = "p=" + 1;
        testPartitionsPaths[i] = new Path(testTablePath, "p=" + i);
        pathToPartitionInfo.put(testPartitionsPaths[i], partitionDesc);
        pathToAliasTable.put(testPartitionsPaths[i], Lists.newArrayList(testPartitionName));
        mapWork.getAliasToWork().put(testPartitionName, scanOp);
    }
    mapWork.setPathToAliases(pathToAliasTable);
    mapWork.setPathToPartitionInfo(pathToPartitionInfo);
    FileSystem fs = FileSystem.getLocal(jobConf);
    try {
        fs.mkdirs(testTablePath);
        byte[] data = new byte[bytesPerFile];
        for (int i = 0; i < numOfPartitions; i++) {
            fs.mkdirs(testPartitionsPaths[i]);
            FSDataOutputStream out = fs.create(new Path(testPartitionsPaths[i], "test1.txt"));
            out.write(data);
            out.close();
        }
        return Utilities.getInputSummary(context, mapWork, null);
    } finally {
        if (fs.exists(testTablePath)) {
            fs.delete(testTablePath, true);
        }
    }
}

Example 12 with MapWork

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

the class TestUtilities method testGetInputPathsWithEmptyTables.

/**
   * Check that calling {@link Utilities#getInputPaths(JobConf, MapWork, Path, Context, boolean)}
   * can process two different empty tables without throwing any exceptions.
   */
@Test
public void testGetInputPathsWithEmptyTables() throws Exception {
    String alias1Name = "alias1";
    String alias2Name = "alias2";
    MapWork mapWork1 = new MapWork();
    MapWork mapWork2 = new MapWork();
    JobConf jobConf = new JobConf();
    Path nonExistentPath1 = new Path(UUID.randomUUID().toString());
    Path nonExistentPath2 = new Path(UUID.randomUUID().toString());
    PartitionDesc mockPartitionDesc = mock(PartitionDesc.class);
    TableDesc mockTableDesc = mock(TableDesc.class);
    when(mockTableDesc.isNonNative()).thenReturn(false);
    when(mockTableDesc.getProperties()).thenReturn(new Properties());
    when(mockPartitionDesc.getProperties()).thenReturn(new Properties());
    when(mockPartitionDesc.getTableDesc()).thenReturn(mockTableDesc);
    doReturn(HiveSequenceFileOutputFormat.class).when(mockPartitionDesc).getOutputFileFormatClass();
    mapWork1.setPathToAliases(new LinkedHashMap<>(ImmutableMap.of(nonExistentPath1, Lists.newArrayList(alias1Name))));
    mapWork1.setAliasToWork(new LinkedHashMap<String, Operator<? extends OperatorDesc>>(ImmutableMap.of(alias1Name, (Operator<?>) mock(Operator.class))));
    mapWork1.setPathToPartitionInfo(new LinkedHashMap<>(ImmutableMap.of(nonExistentPath1, mockPartitionDesc)));
    mapWork2.setPathToAliases(new LinkedHashMap<>(ImmutableMap.of(nonExistentPath2, Lists.newArrayList(alias2Name))));
    mapWork2.setAliasToWork(new LinkedHashMap<String, Operator<? extends OperatorDesc>>(ImmutableMap.of(alias2Name, (Operator<?>) mock(Operator.class))));
    mapWork2.setPathToPartitionInfo(new LinkedHashMap<>(ImmutableMap.of(nonExistentPath2, mockPartitionDesc)));
    List<Path> inputPaths = new ArrayList<>();
    try {
        Path scratchDir = new Path(HiveConf.getVar(jobConf, HiveConf.ConfVars.LOCALSCRATCHDIR));
        inputPaths.addAll(Utilities.getInputPaths(jobConf, mapWork1, scratchDir, mock(Context.class), false));
        inputPaths.addAll(Utilities.getInputPaths(jobConf, mapWork2, scratchDir, mock(Context.class), false));
        assertEquals(inputPaths.size(), 2);
    } finally {
        File file;
        for (Path path : inputPaths) {
            file = new File(path.toString());
            if (file.exists()) {
                file.delete();
            }
        }
    }
}

Example 13 with MapWork

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

the class TestDynamicPartitionPruner method testSingleSourceMultipleFiltersOrdering2.

@Test(timeout = 5000)
public void testSingleSourceMultipleFiltersOrdering2() throws InterruptedException, SerDeException {
    InputInitializerContext mockInitContext = mock(InputInitializerContext.class);
    doReturn(2).when(mockInitContext).getVertexNumTasks("v1");
    MapWork mapWork = createMockMapWork(new TestSource("v1", 2));
    DynamicPartitionPruner pruner = new DynamicPartitionPrunerForEventTesting(mockInitContext, mapWork);
    PruneRunnable pruneRunnable = new PruneRunnable(pruner);
    Thread t = new Thread(pruneRunnable);
    t.start();
    try {
        pruneRunnable.start();
        InputInitializerEvent event = InputInitializerEvent.create("FakeTarget", "TargetInput", ByteBuffer.allocate(0));
        event.setSourceVertexName("v1");
        pruner.processVertex("v1");
        pruner.addEvent(event);
        pruner.addEvent(event);
        pruner.addEvent(event);
        pruner.addEvent(event);
        pruneRunnable.awaitEnd();
        assertFalse(pruneRunnable.inError.get());
    } finally {
        t.interrupt();
        t.join();
    }
}

Example 14 with MapWork

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

the class TestDynamicPartitionPruner method testMultipleSourcesOrdering2.

@Test(timeout = 5000)
public void testMultipleSourcesOrdering2() throws InterruptedException, SerDeException {
    InputInitializerContext mockInitContext = mock(InputInitializerContext.class);
    doReturn(2).when(mockInitContext).getVertexNumTasks("v1");
    doReturn(3).when(mockInitContext).getVertexNumTasks("v2");
    MapWork mapWork = createMockMapWork(new TestSource("v1", 2), new TestSource("v2", 1));
    DynamicPartitionPruner pruner = new DynamicPartitionPrunerForEventTesting(mockInitContext, mapWork);
    PruneRunnable pruneRunnable = new PruneRunnable(pruner);
    Thread t = new Thread(pruneRunnable);
    t.start();
    try {
        pruneRunnable.start();
        InputInitializerEvent eventV1 = InputInitializerEvent.create("FakeTarget", "TargetInput", ByteBuffer.allocate(0));
        eventV1.setSourceVertexName("v1");
        InputInitializerEvent eventV2 = InputInitializerEvent.create("FakeTarget", "TargetInput", ByteBuffer.allocate(0));
        eventV2.setSourceVertexName("v2");
        // 2 X 2 events for V1. 3 X 1 events for V2
        pruner.processVertex("v1");
        pruner.processVertex("v2");
        pruner.addEvent(eventV1);
        pruner.addEvent(eventV1);
        pruner.addEvent(eventV1);
        pruner.addEvent(eventV1);
        pruner.addEvent(eventV2);
        pruner.addEvent(eventV2);
        pruner.addEvent(eventV2);
        pruneRunnable.awaitEnd();
        assertFalse(pruneRunnable.inError.get());
    } finally {
        t.interrupt();
        t.join();
    }
}

Example 15 with MapWork

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

the class TestDynamicPartitionPruner method createMockMapWork.

private MapWork createMockMapWork(TestSource... testSources) {
    MapWork mapWork = mock(MapWork.class);
    Map<String, List<TableDesc>> tableMap = new HashMap<>();
    Map<String, List<String>> columnMap = new HashMap<>();
    Map<String, List<String>> typeMap = new HashMap<>();
    Map<String, List<ExprNodeDesc>> exprMap = new HashMap<>();
    int count = 0;
    for (TestSource testSource : testSources) {
        for (int i = 0; i < testSource.numExpressions; i++) {
            List<TableDesc> tableDescList = tableMap.get(testSource.vertexName);
            if (tableDescList == null) {
                tableDescList = new LinkedList<>();
                tableMap.put(testSource.vertexName, tableDescList);
            }
            tableDescList.add(mock(TableDesc.class));
            List<String> columnList = columnMap.get(testSource.vertexName);
            if (columnList == null) {
                columnList = new LinkedList<>();
                columnMap.put(testSource.vertexName, columnList);
            }
            columnList.add(testSource.vertexName + "c_" + count + "_" + i);
            List<String> typeList = typeMap.get(testSource.vertexName);
            if (typeList == null) {
                typeList = new LinkedList<>();
                typeMap.put(testSource.vertexName, typeList);
            }
            typeList.add("string");
            List<ExprNodeDesc> exprNodeDescList = exprMap.get(testSource.vertexName);
            if (exprNodeDescList == null) {
                exprNodeDescList = new LinkedList<>();
                exprMap.put(testSource.vertexName, exprNodeDescList);
            }
            exprNodeDescList.add(mock(ExprNodeDesc.class));
        }
        count++;
    }
    doReturn(tableMap).when(mapWork).getEventSourceTableDescMap();
    doReturn(columnMap).when(mapWork).getEventSourceColumnNameMap();
    doReturn(exprMap).when(mapWork).getEventSourcePartKeyExprMap();
    doReturn(typeMap).when(mapWork).getEventSourceColumnTypeMap();
    return mapWork;
}