Examples with ColumnMapper org.apache.hadoop.hive.accumulo.columns.ColumnMapper used on opensource projects

Example 6 with ColumnMapper

use of org.apache.hadoop.hive.accumulo.columns.ColumnMapper in project hive by apache.

the class HiveAccumuloTableInputFormat method getSplits.

@Override
public InputSplit[] getSplits(JobConf jobConf, int numSplits) throws IOException {
    final AccumuloConnectionParameters accumuloParams = new AccumuloConnectionParameters(jobConf);
    final Instance instance = accumuloParams.getInstance();
    final ColumnMapper columnMapper;
    try {
        columnMapper = getColumnMapper(jobConf);
    } catch (TooManyAccumuloColumnsException e) {
        throw new IOException(e);
    }
    JobContext context = ShimLoader.getHadoopShims().newJobContext(Job.getInstance(jobConf));
    Path[] tablePaths = FileInputFormat.getInputPaths(context);
    try {
        UserGroupInformation ugi = UserGroupInformation.getCurrentUser();
        final Connector connector;
        // Need to get a Connector so we look up the user's authorizations if not otherwise specified
        if (accumuloParams.useSasl() && !ugi.hasKerberosCredentials()) {
            // In a YARN/Tez job, don't have the Kerberos credentials anymore, use the delegation token
            AuthenticationToken token = ConfiguratorBase.getAuthenticationToken(AccumuloInputFormat.class, jobConf);
            // Convert the stub from the configuration back into a normal Token
            // More reflection to support 1.6
            token = helper.unwrapAuthenticationToken(jobConf, token);
            connector = instance.getConnector(accumuloParams.getAccumuloUserName(), token);
        } else {
            // Still in the local JVM, use the username+password or Kerberos credentials
            connector = accumuloParams.getConnector(instance);
        }
        final List<ColumnMapping> columnMappings = columnMapper.getColumnMappings();
        final List<IteratorSetting> iterators = predicateHandler.getIterators(jobConf, columnMapper);
        final Collection<Range> ranges = predicateHandler.getRanges(jobConf, columnMapper);
        // We don't want that.
        if (null != ranges && ranges.isEmpty()) {
            return new InputSplit[0];
        }
        // Set the relevant information in the Configuration for the AccumuloInputFormat
        configure(jobConf, instance, connector, accumuloParams, columnMapper, iterators, ranges);
        int numColumns = columnMappings.size();
        List<Integer> readColIds = ColumnProjectionUtils.getReadColumnIDs(jobConf);
        // Sanity check
        if (numColumns < readColIds.size())
            throw new IOException("Number of column mappings (" + numColumns + ")" + " numbers less than the hive table columns. (" + readColIds.size() + ")");
        // get splits from Accumulo
        InputSplit[] splits = accumuloInputFormat.getSplits(jobConf, numSplits);
        HiveAccumuloSplit[] hiveSplits = new HiveAccumuloSplit[splits.length];
        for (int i = 0; i < splits.length; i++) {
            RangeInputSplit ris = (RangeInputSplit) splits[i];
            hiveSplits[i] = new HiveAccumuloSplit(ris, tablePaths[0]);
        }
        return hiveSplits;
    } catch (AccumuloException e) {
        log.error("Could not configure AccumuloInputFormat", e);
        throw new IOException(StringUtils.stringifyException(e));
    } catch (AccumuloSecurityException e) {
        log.error("Could not configure AccumuloInputFormat", e);
        throw new IOException(StringUtils.stringifyException(e));
    } catch (SerDeException e) {
        log.error("Could not configure AccumuloInputFormat", e);
        throw new IOException(StringUtils.stringifyException(e));
    }
}

Example 7 with ColumnMapper

use of org.apache.hadoop.hive.accumulo.columns.ColumnMapper in project hive by apache.

the class TestAccumuloPredicateHandler method testRowRangeGeneration.

@Test
public void testRowRangeGeneration() throws SerDeException {
    List<String> columnNames = Arrays.asList("key", "column");
    List<TypeInfo> columnTypes = Arrays.<TypeInfo>asList(TypeInfoFactory.stringTypeInfo, TypeInfoFactory.stringTypeInfo);
    conf.set(serdeConstants.LIST_COLUMNS, Joiner.on(',').join(columnNames));
    conf.set(serdeConstants.LIST_COLUMN_TYPES, "string,string");
    String columnMappingStr = ":rowID,cf:f1";
    conf.set(AccumuloSerDeParameters.COLUMN_MAPPINGS, columnMappingStr);
    columnMapper = new ColumnMapper(columnMappingStr, ColumnEncoding.STRING.getName(), columnNames, columnTypes);
    // 100 < key
    ExprNodeDesc column = new ExprNodeColumnDesc(TypeInfoFactory.intTypeInfo, "key", null, false);
    ExprNodeDesc constant = new ExprNodeConstantDesc(TypeInfoFactory.intTypeInfo, 100);
    List<ExprNodeDesc> children = Lists.newArrayList();
    children.add(constant);
    children.add(column);
    ExprNodeGenericFuncDesc node = new ExprNodeGenericFuncDesc(TypeInfoFactory.stringTypeInfo, new GenericUDFOPLessThan(), children);
    assertNotNull(node);
    String filterExpr = SerializationUtilities.serializeExpression(node);
    conf.set(TableScanDesc.FILTER_EXPR_CONF_STR, filterExpr);
    // Should make (100, +inf)
    List<Range> ranges = handler.getRanges(conf, columnMapper);
    Assert.assertEquals(1, ranges.size());
    Assert.assertEquals(new Range(new Text("100"), false, null, false), ranges.get(0));
}

Example 8 with ColumnMapper

use of org.apache.hadoop.hive.accumulo.columns.ColumnMapper in project hive by apache.

the class TestAccumuloPredicateHandler method testIgnoreIteratorPushdown.

@Test
public void testIgnoreIteratorPushdown() throws TooManyAccumuloColumnsException {
    // Override what's placed in the Configuration by setup()
    conf = new JobConf();
    List<String> columnNames = Arrays.asList("field1", "field2", "rid");
    List<TypeInfo> columnTypes = Arrays.<TypeInfo>asList(TypeInfoFactory.stringTypeInfo, TypeInfoFactory.intTypeInfo, TypeInfoFactory.stringTypeInfo);
    conf.set(serdeConstants.LIST_COLUMNS, Joiner.on(',').join(columnNames));
    conf.set(serdeConstants.LIST_COLUMN_TYPES, "string,int,string");
    String columnMappingStr = "cf:f1,cf:f2,:rowID";
    conf.set(AccumuloSerDeParameters.COLUMN_MAPPINGS, columnMappingStr);
    columnMapper = new ColumnMapper(columnMappingStr, ColumnEncoding.STRING.getName(), columnNames, columnTypes);
    ExprNodeDesc column = new ExprNodeColumnDesc(TypeInfoFactory.stringTypeInfo, "field1", null, false);
    ExprNodeDesc constant = new ExprNodeConstantDesc(TypeInfoFactory.stringTypeInfo, "aaa");
    List<ExprNodeDesc> children = Lists.newArrayList();
    children.add(column);
    children.add(constant);
    ExprNodeDesc node = new ExprNodeGenericFuncDesc(TypeInfoFactory.stringTypeInfo, new GenericUDFOPEqualOrLessThan(), children);
    assertNotNull(node);
    ExprNodeDesc column2 = new ExprNodeColumnDesc(TypeInfoFactory.intTypeInfo, "field2", null, false);
    ExprNodeDesc constant2 = new ExprNodeConstantDesc(TypeInfoFactory.intTypeInfo, 5);
    List<ExprNodeDesc> children2 = Lists.newArrayList();
    children2.add(column2);
    children2.add(constant2);
    ExprNodeDesc node2 = new ExprNodeGenericFuncDesc(TypeInfoFactory.stringTypeInfo, new GenericUDFOPGreaterThan(), children2);
    assertNotNull(node2);
    List<ExprNodeDesc> bothFilters = Lists.newArrayList();
    bothFilters.add(node);
    bothFilters.add(node2);
    ExprNodeGenericFuncDesc both = new ExprNodeGenericFuncDesc(TypeInfoFactory.stringTypeInfo, new GenericUDFOPAnd(), bothFilters);
    String filterExpr = SerializationUtilities.serializeExpression(both);
    conf.set(TableScanDesc.FILTER_EXPR_CONF_STR, filterExpr);
    conf.setBoolean(AccumuloSerDeParameters.ITERATOR_PUSHDOWN_KEY, false);
    try {
        List<IteratorSetting> iterators = handler.getIterators(conf, columnMapper);
        assertEquals(iterators.size(), 0);
    } catch (Exception e) {
        fail(StringUtils.stringifyException(e));
    }
}

Example 9 with ColumnMapper

use of org.apache.hadoop.hive.accumulo.columns.ColumnMapper in project hive by apache.

the class TestAccumuloPredicateHandler method testBinaryRangeGeneration.

@Test
public void testBinaryRangeGeneration() throws Exception {
    List<String> columnNames = Arrays.asList("key", "column");
    List<TypeInfo> columnTypes = Arrays.<TypeInfo>asList(TypeInfoFactory.intTypeInfo, TypeInfoFactory.stringTypeInfo);
    conf.set(serdeConstants.LIST_COLUMNS, Joiner.on(',').join(columnNames));
    conf.set(serdeConstants.LIST_COLUMN_TYPES, "int,string");
    String columnMappingStr = ":rowID#b,cf:f1";
    conf.set(AccumuloSerDeParameters.COLUMN_MAPPINGS, columnMappingStr);
    columnMapper = new ColumnMapper(columnMappingStr, ColumnEncoding.STRING.getName(), columnNames, columnTypes);
    int intValue = 100;
    // Make binary integer value in the bytearray
    ByteArrayOutputStream baos = new ByteArrayOutputStream();
    JavaIntObjectInspector intOI = (JavaIntObjectInspector) PrimitiveObjectInspectorFactory.getPrimitiveJavaObjectInspector(TypeInfoFactory.getPrimitiveTypeInfo(serdeConstants.INT_TYPE_NAME));
    LazyUtils.writePrimitive(baos, intValue, intOI);
    // 100 < key
    ExprNodeDesc column = new ExprNodeColumnDesc(TypeInfoFactory.intTypeInfo, "key", null, false);
    ExprNodeDesc constant = new ExprNodeConstantDesc(TypeInfoFactory.intTypeInfo, intValue);
    List<ExprNodeDesc> children = Lists.newArrayList();
    children.add(constant);
    children.add(column);
    ExprNodeGenericFuncDesc node = new ExprNodeGenericFuncDesc(TypeInfoFactory.stringTypeInfo, new GenericUDFOPLessThan(), children);
    assertNotNull(node);
    String filterExpr = SerializationUtilities.serializeExpression(node);
    conf.set(TableScanDesc.FILTER_EXPR_CONF_STR, filterExpr);
    // Should make (100, +inf)
    List<Range> ranges = handler.getRanges(conf, columnMapper);
    Assert.assertEquals(1, ranges.size());
    Assert.assertEquals(new Range(new Text(baos.toByteArray()), false, null, false), ranges.get(0));
}

Example 10 with ColumnMapper

use of org.apache.hadoop.hive.accumulo.columns.ColumnMapper in project hive by apache.

the class TestAccumuloPredicateHandler method setup.

@Before
public void setup() throws TooManyAccumuloColumnsException {
    FunctionRegistry.getFunctionNames();
    conf = new JobConf();
    List<String> columnNames = Arrays.asList("field1", "rid");
    List<TypeInfo> columnTypes = Arrays.<TypeInfo>asList(TypeInfoFactory.stringTypeInfo, TypeInfoFactory.stringTypeInfo);
    conf.set(serdeConstants.LIST_COLUMNS, Joiner.on(',').join(columnNames));
    conf.set(serdeConstants.LIST_COLUMN_TYPES, "string,string");
    String columnMappingStr = "cf:f1,:rowID";
    conf.set(AccumuloSerDeParameters.COLUMN_MAPPINGS, columnMappingStr);
    columnMapper = new ColumnMapper(columnMappingStr, ColumnEncoding.STRING.getName(), columnNames, columnTypes);
}