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

Example 1 with Firehose

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

the class LocalFirehoseFactoryTest method testConnect.

@Test
public void testConnect() throws IOException {
    try (final Firehose firehose = factory.connect(new StringInputRowParser(new CSVParseSpec(new TimestampSpec("timestamp", "auto", null), new DimensionsSpec(DimensionsSpec.getDefaultSchemas(Arrays.asList("timestamp", "a"))), ",", Arrays.asList("timestamp", "a"), false, 0), StandardCharsets.UTF_8.name()), null)) {
        final List<Row> rows = new ArrayList<>();
        while (firehose.hasMore()) {
            rows.add(firehose.nextRow());
        }
        Assert.assertEquals(5, rows.size());
        rows.sort(Comparator.comparing(Row::getTimestamp));
        for (int i = 0; i < 5; i++) {
            final List<String> dimVals = rows.get(i).getDimension("a");
            Assert.assertEquals(1, dimVals.size());
            Assert.assertEquals(i + "th test file", dimVals.get(0));
        }
    }
}

Example 2 with Firehose

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

the class SqlFirehoseFactoryTest method testWithoutCacheAndFetch.

@Test
public void testWithoutCacheAndFetch() throws Exception {
    derbyConnector = derbyConnectorRule.getConnector();
    SqlTestUtils testUtils = new SqlTestUtils(derbyConnector);
    testUtils.createAndUpdateTable(TABLE_NAME_1, 10);
    final SqlFirehoseFactory factory = new SqlFirehoseFactory(SQLLIST1, 0L, 0L, 0L, 0L, true, testUtils.getDerbyFirehoseConnector(), mapper);
    final List<Row> rows = new ArrayList<>();
    final File firehoseTmpDir = createFirehoseTmpDir("testWithoutCacheAndFetch");
    try (Firehose firehose = factory.connect(parser, firehoseTmpDir)) {
        while (firehose.hasMore()) {
            rows.add(firehose.nextRow());
        }
    }
    assertResult(rows, SQLLIST1);
    assertNumRemainingCacheFiles(firehoseTmpDir, 0);
    testUtils.dropTable(TABLE_NAME_1);
}

Example 3 with Firehose

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

the class AppenderatorDriverRealtimeIndexTaskTest method testReportParseExceptionsOnBadMetric.

@Test(timeout = 60_000L)
public void testReportParseExceptionsOnBadMetric() throws Exception {
    expectPublishedSegments(0);
    final AppenderatorDriverRealtimeIndexTask task = makeRealtimeTask(null, true);
    final ListenableFuture<TaskStatus> statusFuture = runTask(task);
    // Wait for firehose to show up, it starts off null.
    while (task.getFirehose() == null) {
        Thread.sleep(50);
    }
    final TestFirehose firehose = (TestFirehose) task.getFirehose();
    firehose.addRows(ImmutableList.of(ImmutableMap.of("t", 2000000L, "dim1", "foo", "met1", "1"), ImmutableMap.of("t", 3000000L, "dim1", "foo", "met1", "foo"), ImmutableMap.of("t", now.minus(new Period("P1D")).getMillis(), "dim1", "foo", "met1", "foo"), ImmutableMap.of("t", 4000000L, "dim2", "bar", "met1", 2.0)));
    // Stop the firehose, this will drain out existing events.
    firehose.close();
    // Wait for the task to finish.
    TaskStatus status = statusFuture.get();
    Assert.assertTrue(status.getErrorMsg().contains("org.apache.druid.java.util.common.RE: Max parse exceptions[0] exceeded"));
    IngestionStatsAndErrorsTaskReportData reportData = getTaskReportData();
    List<LinkedHashMap> parseExceptionReports = (List<LinkedHashMap>) reportData.getUnparseableEvents().get(RowIngestionMeters.BUILD_SEGMENTS);
    List<String> expectedMessages = ImmutableList.of("Unable to parse value[foo] for field[met1]");
    List<String> actualMessages = parseExceptionReports.stream().map((r) -> {
        return ((List<String>) r.get("details")).get(0);
    }).collect(Collectors.toList());
    Assert.assertEquals(expectedMessages, actualMessages);
    List<String> expectedInputs = ImmutableList.of("{t=3000000, dim1=foo, met1=foo}");
    List<String> actualInputs = parseExceptionReports.stream().map((r) -> {
        return (String) r.get("input");
    }).collect(Collectors.toList());
    Assert.assertEquals(expectedInputs, actualInputs);
}

Example 4 with Firehose

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

the class AppenderatorDriverRealtimeIndexTaskTest method testRestoreCorruptData.

@Test(timeout = 60_000L)
public void testRestoreCorruptData() throws Exception {
    final AppenderatorDriverRealtimeIndexTask task1 = makeRealtimeTask(null);
    // First run:
    {
        expectPublishedSegments(0);
        final ListenableFuture<TaskStatus> statusFuture = runTask(task1);
        // Wait for firehose to show up, it starts off null.
        while (task1.getFirehose() == null) {
            Thread.sleep(50);
        }
        final TestFirehose firehose = (TestFirehose) task1.getFirehose();
        firehose.addRows(ImmutableList.of(ImmutableMap.of("t", now.getMillis(), "dim1", "foo")));
        // Trigger graceful shutdown.
        task1.stopGracefully(taskToolboxFactory.build(task1).getConfig());
        // Wait for the task to finish. The status doesn't really matter, but we'll check it anyway.
        final TaskStatus taskStatus = statusFuture.get();
        Assert.assertEquals(TaskState.SUCCESS, taskStatus.getStatusCode());
        // Nothing should be published.
        Assert.assertTrue(publishedSegments.isEmpty());
    }
    Optional<File> optional = FileUtils.listFiles(baseDir, null, true).stream().filter(f -> f.getName().equals("00000.smoosh")).findFirst();
    Assert.assertTrue("Could not find smoosh file", optional.isPresent());
    // Corrupt the data:
    final File smooshFile = optional.get();
    Files.write(smooshFile.toPath(), StringUtils.toUtf8("oops!"));
    // Second run:
    {
        expectPublishedSegments(0);
        final AppenderatorDriverRealtimeIndexTask task2 = makeRealtimeTask(task1.getId());
        final ListenableFuture<TaskStatus> statusFuture = runTask(task2);
        // Wait for the task to finish.
        TaskStatus status = statusFuture.get();
        Map<String, Object> expectedMetrics = ImmutableMap.of(RowIngestionMeters.BUILD_SEGMENTS, ImmutableMap.of(RowIngestionMeters.PROCESSED_WITH_ERROR, 0, RowIngestionMeters.PROCESSED, 0, RowIngestionMeters.UNPARSEABLE, 0, RowIngestionMeters.THROWN_AWAY, 0));
        IngestionStatsAndErrorsTaskReportData reportData = getTaskReportData();
        Assert.assertEquals(expectedMetrics, reportData.getRowStats());
        Pattern errorPattern = Pattern.compile("(?s)java\\.lang\\.IllegalArgumentException.*\n" + "\tat (java\\.base/)?java\\.nio\\.Buffer\\..*");
        Assert.assertTrue(errorPattern.matcher(status.getErrorMsg()).matches());
    }
}

Example 5 with Firehose

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

the class IngestSegmentFirehoseFactoryTest method testTransformSpec.

@Test
public void testTransformSpec() throws IOException {
    Assert.assertEquals(MAX_SHARD_NUMBER.longValue(), SEGMENT_SET.size());
    Integer rowcount = 0;
    final TransformSpec transformSpec = new TransformSpec(new SelectorDimFilter(ColumnHolder.TIME_COLUMN_NAME, "1", null), ImmutableList.of(new ExpressionTransform(METRIC_FLOAT_NAME, METRIC_FLOAT_NAME + " * 10", ExprMacroTable.nil())));
    int skipped = 0;
    try (final Firehose firehose = factory.connect(transformSpec.decorate(rowParser), TMP_DIR)) {
        while (firehose.hasMore()) {
            InputRow row = firehose.nextRow();
            if (row == null) {
                skipped++;
                continue;
            }
            Assert.assertArrayEquals(new String[] { DIM_NAME }, row.getDimensions().toArray());
            Assert.assertArrayEquals(new String[] { DIM_VALUE }, row.getDimension(DIM_NAME).toArray());
            Assert.assertEquals(METRIC_LONG_VALUE.longValue(), row.getMetric(METRIC_LONG_NAME).longValue());
            Assert.assertEquals(METRIC_FLOAT_VALUE * 10, row.getMetric(METRIC_FLOAT_NAME).floatValue(), METRIC_FLOAT_VALUE * 0.0001);
            ++rowcount;
        }
    }
    Assert.assertEquals(90, skipped);
    Assert.assertEquals((int) MAX_ROWS, (int) rowcount);
}