Examples with TupleReader org.apache.drill.exec.vector.accessor.TupleReader used on opensource projects

Example 6 with TupleReader

use of org.apache.drill.exec.vector.accessor.TupleReader in project drill by apache.

the class TestResultSetLoaderMapArray method testDoubleNestedArray.

/**
 * Test a doubly-nested array of maps.
 */
@Test
public void testDoubleNestedArray() {
    TupleMetadata schema = new SchemaBuilder().add("a", MinorType.INT).addMapArray("m1").add("b", MinorType.INT).addMapArray("m2").add("c", MinorType.INT).addArray("d", MinorType.VARCHAR).resumeMap().resumeSchema().buildSchema();
    ResultSetLoaderImpl.ResultSetOptions options = new ResultSetOptionBuilder().readerSchema(schema).build();
    ResultSetLoader rsLoader = new ResultSetLoaderImpl(fixture.allocator(), options);
    RowSetLoader rootWriter = rsLoader.writer();
    rsLoader.startBatch();
    ScalarWriter aWriter = rootWriter.scalar("a");
    ArrayWriter a1Writer = rootWriter.array("m1");
    TupleWriter m1Writer = a1Writer.tuple();
    ScalarWriter bWriter = m1Writer.scalar("b");
    ArrayWriter a2Writer = m1Writer.array("m2");
    TupleWriter m2Writer = a2Writer.tuple();
    ScalarWriter cWriter = m2Writer.scalar("c");
    ScalarWriter dWriter = m2Writer.array("d").scalar();
    for (int i = 0; i < 5; i++) {
        rootWriter.start();
        aWriter.setInt(i);
        for (int j = 0; j < 4; j++) {
            int a1Key = i + 10 + j;
            bWriter.setInt(a1Key);
            for (int k = 0; k < 3; k++) {
                int a2Key = a1Key * 10 + k;
                cWriter.setInt(a2Key);
                for (int l = 0; l < 2; l++) {
                    dWriter.setString("d-" + (a2Key * 10 + l));
                }
                a2Writer.save();
            }
            a1Writer.save();
        }
        rootWriter.save();
    }
    RowSet results = fixture.wrap(rsLoader.harvest());
    RowSetReader reader = results.reader();
    ScalarReader aReader = reader.scalar("a");
    ArrayReader a1Reader = reader.array("m1");
    TupleReader m1Reader = a1Reader.tuple();
    ScalarReader bReader = m1Reader.scalar("b");
    ArrayReader a2Reader = m1Reader.array("m2");
    TupleReader m2Reader = a2Reader.tuple();
    ScalarReader cReader = m2Reader.scalar("c");
    ArrayReader dArray = m2Reader.array("d");
    ScalarReader dReader = dArray.scalar();
    for (int i = 0; i < 5; i++) {
        assertTrue(reader.next());
        assertEquals(i, aReader.getInt());
        for (int j = 0; j < 4; j++) {
            assertTrue(a1Reader.next());
            int a1Key = i + 10 + j;
            assertEquals(a1Key, bReader.getInt());
            for (int k = 0; k < 3; k++) {
                assertTrue(a2Reader.next());
                int a2Key = a1Key * 10 + k;
                assertEquals(a2Key, cReader.getInt());
                for (int l = 0; l < 2; l++) {
                    assertTrue(dArray.next());
                    assertEquals("d-" + (a2Key * 10 + l), dReader.getString());
                }
            }
        }
    }
    rsLoader.close();
}

Example 7 with TupleReader

use of org.apache.drill.exec.vector.accessor.TupleReader in project drill by apache.

the class TestResultSetLoaderMaps method testOverwriteRow.

/**
 * Version of the {#link TestResultSetLoaderProtocol#testOverwriteRow()} test
 * that uses nested columns.
 */
@Test
public void testOverwriteRow() {
    final TupleMetadata schema = new SchemaBuilder().add("a", MinorType.INT).addMap("m").add("b", MinorType.INT).add("c", MinorType.VARCHAR).resumeSchema().buildSchema();
    final ResultSetLoaderImpl.ResultSetOptions options = new ResultSetOptionBuilder().readerSchema(schema).rowCountLimit(ValueVector.MAX_ROW_COUNT).build();
    final ResultSetLoader rsLoader = new ResultSetLoaderImpl(fixture.allocator(), options);
    final RowSetLoader rootWriter = rsLoader.writer();
    // Can't use the shortcut to populate rows when doing overwrites.
    final ScalarWriter aWriter = rootWriter.scalar("a");
    final TupleWriter mWriter = rootWriter.tuple("m");
    final ScalarWriter bWriter = mWriter.scalar("b");
    final ScalarWriter cWriter = mWriter.scalar("c");
    // Write 100,000 rows, overwriting 99% of them. This will cause vector
    // overflow and data corruption if overwrite does not work; but will happily
    // produce the correct result if everything works as it should.
    final byte[] value = new byte[512];
    Arrays.fill(value, (byte) 'X');
    int count = 0;
    rsLoader.startBatch();
    while (count < 100_000) {
        rootWriter.start();
        count++;
        aWriter.setInt(count);
        bWriter.setInt(count * 10);
        cWriter.setBytes(value, value.length);
        if (count % 100 == 0) {
            rootWriter.save();
        }
    }
    // Verify using a reader.
    final RowSet result = fixture.wrap(rsLoader.harvest());
    assertEquals(count / 100, result.rowCount());
    final RowSetReader reader = result.reader();
    final TupleReader mReader = reader.tuple("m");
    int rowId = 1;
    while (reader.next()) {
        assertEquals(rowId * 100, reader.scalar("a").getInt());
        assertEquals(rowId * 1000, mReader.scalar("b").getInt());
        assertTrue(Arrays.equals(value, mReader.scalar("c").getBytes()));
        rowId++;
    }
    result.clear();
    rsLoader.close();
}

Example 8 with TupleReader

use of org.apache.drill.exec.vector.accessor.TupleReader in project drill by apache.

the class MetadataControllerBatch method getRowGroupMetadata.

private RowGroupMetadata getRowGroupMetadata(TupleReader reader, List<StatisticsHolder<?>> metadataStatistics, Map<SchemaPath, ColumnStatistics<?>> columnStatistics, int nestingLevel) {
    List<String> segmentColumns = popConfig.getContext().segmentColumns();
    String segmentKey = segmentColumns.size() > 0 ? reader.column(segmentColumns.iterator().next()).scalar().getString() : MetadataInfo.DEFAULT_SEGMENT_KEY;
    List<String> partitionValues = segmentColumns.stream().limit(nestingLevel - 2).map(columnName -> reader.column(columnName).scalar().getString()).collect(Collectors.toList());
    Path path = new Path(reader.column(MetastoreAnalyzeConstants.LOCATION_FIELD).scalar().getString());
    int rowGroupIndex = Integer.parseInt(reader.column(columnNamesOptions.rowGroupIndex()).scalar().getString());
    String metadataIdentifier = MetadataIdentifierUtils.getRowGroupMetadataIdentifier(partitionValues, path, rowGroupIndex);
    MetadataInfo metadataInfo = MetadataInfo.builder().type(MetadataType.ROW_GROUP).key(segmentKey).identifier(StringUtils.defaultIfEmpty(metadataIdentifier, null)).build();
    return RowGroupMetadata.builder().tableInfo(tableInfo).metadataInfo(metadataInfo).columnsStatistics(columnStatistics).metadataStatistics(metadataStatistics).hostAffinity(Collections.emptyMap()).rowGroupIndex(rowGroupIndex).path(path).lastModifiedTime(Long.parseLong(reader.column(columnNamesOptions.lastModifiedTime()).scalar().getString())).schema(TupleMetadata.of(reader.column(MetastoreAnalyzeConstants.SCHEMA_FIELD).scalar().getString())).build();
}

Example 9 with TupleReader

use of org.apache.drill.exec.vector.accessor.TupleReader in project drill by apache.

the class MetadataControllerBatch method getSegmentMetadata.

private SegmentMetadata getSegmentMetadata(TupleReader reader, List<StatisticsHolder<?>> metadataStatistics, Map<SchemaPath, ColumnStatistics<?>> columnStatistics, int nestingLevel) {
    List<String> segmentColumns = popConfig.getContext().segmentColumns();
    String segmentKey = segmentColumns.size() > 0 ? reader.column(segmentColumns.iterator().next()).scalar().getString() : MetadataInfo.DEFAULT_SEGMENT_KEY;
    // and therefore all values should be used when forming metadata identifier
    if (popConfig.getContext().multiValueSegments()) {
        nestingLevel = segmentColumns.size();
    }
    List<String> allPartitionValues = segmentColumns.stream().limit(nestingLevel).map(columnName -> reader.column(columnName).scalar().getString()).collect(Collectors.toList());
    String metadataIdentifier = MetadataIdentifierUtils.getMetadataIdentifierKey(allPartitionValues);
    MetadataInfo metadataInfo = MetadataInfo.builder().type(MetadataType.SEGMENT).key(segmentKey).identifier(StringUtils.defaultIfEmpty(metadataIdentifier, null)).build();
    int segmentLevel = nestingLevel - 1;
    // for the case of multi-value segments, there is no nesting,
    // so all partition column values should be used
    List<String> partitionValues = popConfig.getContext().multiValueSegments() ? allPartitionValues : Collections.singletonList(allPartitionValues.get(segmentLevel));
    return SegmentMetadata.builder().tableInfo(tableInfo).metadataInfo(metadataInfo).columnsStatistics(columnStatistics).metadataStatistics(metadataStatistics).path(new Path(reader.column(MetastoreAnalyzeConstants.LOCATION_FIELD).scalar().getString())).locations(getIncomingLocations(reader)).column(segmentColumns.size() > 0 ? SchemaPath.getSimplePath(segmentColumns.get(segmentLevel)) : null).partitionValues(partitionValues).lastModifiedTime(Long.parseLong(reader.column(columnNamesOptions.lastModifiedTime()).scalar().getString())).schema(TupleMetadata.of(reader.column(MetastoreAnalyzeConstants.SCHEMA_FIELD).scalar().getString())).build();
}

Example 10 with TupleReader

use of org.apache.drill.exec.vector.accessor.TupleReader in project drill by apache.

the class MetadataControllerBatch method getPartitionMetadata.

private PartitionMetadata getPartitionMetadata(TupleReader reader, List<StatisticsHolder<?>> metadataStatistics, Map<SchemaPath, ColumnStatistics<?>> columnStatistics, int nestingLevel) {
    List<String> segmentColumns = popConfig.getContext().segmentColumns();
    String segmentKey = segmentColumns.size() > 0 ? reader.column(segmentColumns.iterator().next()).scalar().getString() : MetadataInfo.DEFAULT_SEGMENT_KEY;
    List<String> partitionValues = segmentColumns.stream().limit(nestingLevel).map(columnName -> reader.column(columnName).scalar().getString()).collect(Collectors.toList());
    String metadataIdentifier = MetadataIdentifierUtils.getMetadataIdentifierKey(partitionValues);
    MetadataInfo metadataInfo = MetadataInfo.builder().type(MetadataType.PARTITION).key(segmentKey).identifier(StringUtils.defaultIfEmpty(metadataIdentifier, null)).build();
    return PartitionMetadata.builder().tableInfo(tableInfo).metadataInfo(metadataInfo).columnsStatistics(columnStatistics).metadataStatistics(metadataStatistics).locations(getIncomingLocations(reader)).lastModifiedTime(Long.parseLong(reader.column(columnNamesOptions.lastModifiedTime()).scalar().getString())).schema(TupleMetadata.of(reader.column(MetastoreAnalyzeConstants.SCHEMA_FIELD).scalar().getString())).build();
}