Examples with CsvInputFormat org.apache.druid.data.input.impl.CsvInputFormat used on opensource projects

Example 6 with CsvInputFormat

use of org.apache.druid.data.input.impl.CsvInputFormat in project druid by druid-io.

the class IndexTaskTest method testTransformSpec.

@Test
public void testTransformSpec() throws Exception {
    File tmpDir = temporaryFolder.newFolder();
    File tmpFile = File.createTempFile("druid", "index", tmpDir);
    try (BufferedWriter writer = Files.newWriter(tmpFile, StandardCharsets.UTF_8)) {
        writer.write("2014-01-01T00:00:10Z,a,an|array,1|2|3,1\n");
        writer.write("2014-01-01T01:00:20Z,b,another|array,3|4,1\n");
        writer.write("2014-01-01T02:00:30Z,c,and|another,0|1,1\n");
    }
    final DimensionsSpec dimensionsSpec = new DimensionsSpec(DimensionsSpec.getDefaultSchemas(Arrays.asList("ts", "dim", "dim_array", "dim_num_array", "dimt", "dimtarray1", "dimtarray2", "dimtnum_array")));
    final List<String> columns = Arrays.asList("ts", "dim", "dim_array", "dim_num_array", "val");
    final String listDelimiter = "|";
    final TransformSpec transformSpec = new TransformSpec(new SelectorDimFilter("dim", "b", null), ImmutableList.of(new ExpressionTransform("dimt", "concat(dim,dim)", ExprMacroTable.nil()), new ExpressionTransform("dimtarray1", "array(dim, dim)", ExprMacroTable.nil()), new ExpressionTransform("dimtarray2", "map(d -> concat(d, 'foo'), dim_array)", ExprMacroTable.nil()), new ExpressionTransform("dimtnum_array", "map(d -> d + 3, dim_num_array)", ExprMacroTable.nil())));
    final IndexTuningConfig tuningConfig = createTuningConfigWithMaxRowsPerSegment(2, false);
    final IndexIngestionSpec indexIngestionSpec;
    if (useInputFormatApi) {
        indexIngestionSpec = createIngestionSpec(jsonMapper, tmpDir, DEFAULT_TIMESTAMP_SPEC, dimensionsSpec, new CsvInputFormat(columns, listDelimiter, null, false, 0), transformSpec, null, tuningConfig, false, false);
    } else {
        indexIngestionSpec = createIngestionSpec(jsonMapper, tmpDir, new CSVParseSpec(DEFAULT_TIMESTAMP_SPEC, dimensionsSpec, listDelimiter, columns, false, 0), transformSpec, null, tuningConfig, false, false);
    }
    IndexTask indexTask = new IndexTask(null, null, indexIngestionSpec, null);
    Assert.assertEquals(indexTask.getId(), indexTask.getGroupId());
    final List<DataSegment> segments = runTask(indexTask).rhs;
    Assert.assertEquals(1, segments.size());
    DataSegment segment = segments.get(0);
    final File segmentFile = segmentCacheManager.getSegmentFiles(segment);
    final WindowedStorageAdapter adapter = new WindowedStorageAdapter(new QueryableIndexStorageAdapter(indexIO.loadIndex(segmentFile)), segment.getInterval());
    final Sequence<Cursor> cursorSequence = adapter.getAdapter().makeCursors(null, segment.getInterval(), VirtualColumns.EMPTY, Granularities.ALL, false, null);
    final List<Map<String, Object>> transforms = cursorSequence.map(cursor -> {
        final DimensionSelector selector1 = cursor.getColumnSelectorFactory().makeDimensionSelector(new DefaultDimensionSpec("dimt", "dimt"));
        final DimensionSelector selector2 = cursor.getColumnSelectorFactory().makeDimensionSelector(new DefaultDimensionSpec("dimtarray1", "dimtarray1"));
        final DimensionSelector selector3 = cursor.getColumnSelectorFactory().makeDimensionSelector(new DefaultDimensionSpec("dimtarray2", "dimtarray2"));
        final DimensionSelector selector4 = cursor.getColumnSelectorFactory().makeDimensionSelector(new DefaultDimensionSpec("dimtnum_array", "dimtnum_array"));
        Map<String, Object> row = new HashMap<>();
        row.put("dimt", selector1.defaultGetObject());
        row.put("dimtarray1", selector2.defaultGetObject());
        row.put("dimtarray2", selector3.defaultGetObject());
        row.put("dimtnum_array", selector4.defaultGetObject());
        cursor.advance();
        return row;
    }).toList();
    Assert.assertEquals(1, transforms.size());
    Assert.assertEquals("bb", transforms.get(0).get("dimt"));
    Assert.assertEquals(ImmutableList.of("b", "b"), transforms.get(0).get("dimtarray1"));
    Assert.assertEquals(ImmutableList.of("anotherfoo", "arrayfoo"), transforms.get(0).get("dimtarray2"));
    Assert.assertEquals(ImmutableList.of("6.0", "7.0"), transforms.get(0).get("dimtnum_array"));
    Assert.assertEquals(DATASOURCE, segments.get(0).getDataSource());
    Assert.assertEquals(Intervals.of("2014/P1D"), segments.get(0).getInterval());
    Assert.assertEquals(NumberedShardSpec.class, segments.get(0).getShardSpec().getClass());
    Assert.assertEquals(0, segments.get(0).getShardSpec().getPartitionNum());
}

Example 7 with CsvInputFormat

use of org.apache.druid.data.input.impl.CsvInputFormat in project druid by druid-io.

the class CsvInputSourceSamplerTest method testCSVColumnAllNull.

@Test
public void testCSVColumnAllNull() {
    final TimestampSpec timestampSpec = new TimestampSpec(null, null, DateTimes.of("1970"));
    final DimensionsSpec dimensionsSpec = new DimensionsSpec(null);
    final DataSchema dataSchema = new DataSchema("sampler", timestampSpec, dimensionsSpec, null, null, null);
    final List<String> strCsvRows = ImmutableList.of("FirstName,LastName,Number,Gender", "J,G,,Male", "Kobe,Bryant,,Male", "Lisa, Krystal,,Female", "Michael,Jackson,,Male");
    final InputSource inputSource = new InlineInputSource(String.join("\n", strCsvRows));
    final InputFormat inputFormat = new CsvInputFormat(null, null, null, true, 0);
    final InputSourceSampler inputSourceSampler = new InputSourceSampler();
    final SamplerResponse response = inputSourceSampler.sample(inputSource, inputFormat, dataSchema, null);
    Assert.assertEquals(4, response.getNumRowsRead());
    Assert.assertEquals(4, response.getNumRowsIndexed());
    Assert.assertEquals(4, response.getData().size());
    List<SamplerResponseRow> data = response.getData();
    Assert.assertEquals(new SamplerResponseRow(new SamplerTestUtils.MapAllowingNullValuesBuilder<String, Object>().put("Number", null).put("FirstName", "J").put("LastName", "G").put("Gender", "Male").build(), new SamplerTestUtils.MapAllowingNullValuesBuilder<String, Object>().put("__time", 0L).put("Number", null).put("FirstName", "J").put("LastName", "G").put("Gender", "Male").build(), null, null), data.get(0));
    Assert.assertEquals(new SamplerResponseRow(new SamplerTestUtils.MapAllowingNullValuesBuilder<String, Object>().put("Number", null).put("FirstName", "Kobe").put("LastName", "Bryant").put("Gender", "Male").build(), new SamplerTestUtils.MapAllowingNullValuesBuilder<String, Object>().put("Number", null).put("__time", 0L).put("FirstName", "Kobe").put("LastName", "Bryant").put("Gender", "Male").build(), null, null), data.get(1));
    Assert.assertEquals(new SamplerResponseRow(new SamplerTestUtils.MapAllowingNullValuesBuilder<String, Object>().put("Number", null).put("FirstName", "Lisa").put("LastName", " Krystal").put("Gender", "Female").build(), new SamplerTestUtils.MapAllowingNullValuesBuilder<String, Object>().put("Number", null).put("__time", 0L).put("FirstName", "Lisa").put("LastName", " Krystal").put("Gender", "Female").build(), null, null), data.get(2));
    Assert.assertEquals(new SamplerResponseRow(new SamplerTestUtils.MapAllowingNullValuesBuilder<String, Object>().put("Number", null).put("FirstName", "Michael").put("LastName", "Jackson").put("Gender", "Male").build(), new SamplerTestUtils.MapAllowingNullValuesBuilder<String, Object>().put("__time", 0L).put("Number", null).put("FirstName", "Michael").put("LastName", "Jackson").put("Gender", "Male").build(), null, null), data.get(3));
}

Example 8 with CsvInputFormat

use of org.apache.druid.data.input.impl.CsvInputFormat in project druid by druid-io.

the class RecordSupplierInputSourceTest method testRead.

@Test
public void testRead() throws IOException {
    final RandomCsvSupplier supplier = new RandomCsvSupplier();
    final InputSource inputSource = new RecordSupplierInputSource<>("topic", supplier, false);
    final List<String> colNames = IntStream.range(0, NUM_COLS).mapToObj(i -> StringUtils.format("col_%d", i)).collect(Collectors.toList());
    final InputFormat inputFormat = new CsvInputFormat(colNames, null, null, false, 0);
    final InputSourceReader reader = inputSource.reader(new InputRowSchema(new TimestampSpec("col_0", "auto", null), new DimensionsSpec(DimensionsSpec.getDefaultSchemas(colNames.subList(1, colNames.size()))), ColumnsFilter.all()), inputFormat, temporaryFolder.newFolder());
    int read = 0;
    try (CloseableIterator<InputRow> iterator = reader.read()) {
        for (; read < NUM_ROWS && iterator.hasNext(); read++) {
            final InputRow inputRow = iterator.next();
            Assert.assertEquals(DateTimes.of(TIMESTAMP_STRING), inputRow.getTimestamp());
            Assert.assertEquals(NUM_COLS - 1, inputRow.getDimensions().size());
        }
    }
    Assert.assertEquals(NUM_ROWS, read);
    Assert.assertTrue(supplier.isClosed());
}

Example 9 with CsvInputFormat

use of org.apache.druid.data.input.impl.CsvInputFormat in project druid by druid-io.

the class OssInputSourceTest method testReader.

@Test
public void testReader() throws IOException {
    EasyMock.reset(OSSCLIENT);
    expectListObjects(PREFIXES.get(0), ImmutableList.of(EXPECTED_URIS.get(0)), CONTENT);
    expectListObjects(EXPECTED_URIS.get(1), ImmutableList.of(EXPECTED_URIS.get(1)), CONTENT);
    expectGetObject(EXPECTED_URIS.get(0));
    expectGetObject(EXPECTED_URIS.get(1));
    EasyMock.replay(OSSCLIENT);
    OssInputSource inputSource = new OssInputSource(OSSCLIENT, INPUT_DATA_CONFIG, null, ImmutableList.of(PREFIXES.get(0), EXPECTED_URIS.get(1)), null, null);
    InputRowSchema someSchema = new InputRowSchema(new TimestampSpec("time", "auto", null), new DimensionsSpec(DimensionsSpec.getDefaultSchemas(ImmutableList.of("dim1", "dim2"))), ColumnsFilter.all());
    InputSourceReader reader = inputSource.reader(someSchema, new CsvInputFormat(ImmutableList.of("time", "dim1", "dim2"), "|", false, null, 0), temporaryFolder.newFolder());
    CloseableIterator<InputRow> iterator = reader.read();
    while (iterator.hasNext()) {
        InputRow nextRow = iterator.next();
        Assert.assertEquals(NOW, nextRow.getTimestamp());
        Assert.assertEquals("hello", nextRow.getDimension("dim1").get(0));
        Assert.assertEquals("world", nextRow.getDimension("dim2").get(0));
    }
    EasyMock.verify(OSSCLIENT);
}

Example 10 with CsvInputFormat

use of org.apache.druid.data.input.impl.CsvInputFormat in project druid by druid-io.

the class GoogleCloudStorageInputSourceTest method testCompressedReader.

@Test
public void testCompressedReader() throws IOException {
    EasyMock.reset(STORAGE);
    EasyMock.reset(INPUT_DATA_CONFIG);
    addExpectedPrefixObjects(PREFIXES.get(0), ImmutableList.of(EXPECTED_COMPRESSED_URIS.get(0)));
    addExpectedGetCompressedObjectMock(EXPECTED_COMPRESSED_URIS.get(0));
    addExpectedPrefixObjects(PREFIXES.get(1), ImmutableList.of(EXPECTED_COMPRESSED_URIS.get(1)));
    addExpectedGetCompressedObjectMock(EXPECTED_COMPRESSED_URIS.get(1));
    EasyMock.expect(INPUT_DATA_CONFIG.getMaxListingLength()).andReturn(MAX_LISTING_LENGTH);
    EasyMock.replay(STORAGE);
    EasyMock.replay(INPUT_DATA_CONFIG);
    GoogleCloudStorageInputSource inputSource = new GoogleCloudStorageInputSource(STORAGE, INPUT_DATA_CONFIG, null, PREFIXES, null);
    InputRowSchema someSchema = new InputRowSchema(new TimestampSpec("time", "auto", null), new DimensionsSpec(DimensionsSpec.getDefaultSchemas(ImmutableList.of("dim1", "dim2"))), ColumnsFilter.all());
    InputSourceReader reader = inputSource.reader(someSchema, new CsvInputFormat(ImmutableList.of("time", "dim1", "dim2"), "|", false, null, 0), null);
    CloseableIterator<InputRow> iterator = reader.read();
    while (iterator.hasNext()) {
        InputRow nextRow = iterator.next();
        Assert.assertEquals(NOW, nextRow.getTimestamp());
        Assert.assertEquals("hello", nextRow.getDimension("dim1").get(0));
        Assert.assertEquals("world", nextRow.getDimension("dim2").get(0));
    }
}