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

Example 16 with DimensionSchema

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

the class SegmentAnalyzerTest method testMappedWorksHelper.

private void testMappedWorksHelper(EnumSet<SegmentMetadataQuery.AnalysisType> analyses) {
    final List<SegmentAnalysis> results = getSegmentAnalysises(new QueryableIndexSegment(TestIndex.getMMappedTestIndex(), SegmentId.dummy("test_1")), analyses);
    Assert.assertEquals(1, results.size());
    final SegmentAnalysis analysis = results.get(0);
    Assert.assertEquals(SegmentId.dummy("test_1").toString(), analysis.getId());
    final Map<String, ColumnAnalysis> columns = analysis.getColumns();
    Assert.assertEquals(TestIndex.COLUMNS.length + 3 - 1, columns.size());
    for (DimensionSchema schema : TestIndex.DIMENSION_SCHEMAS) {
        final String dimension = schema.getName();
        final ColumnAnalysis columnAnalysis = columns.get(dimension);
        if ("null_column".equals(dimension)) {
            Assert.assertNull(columnAnalysis);
        } else {
            final boolean isString = schema.getColumnType().is(ValueType.STRING);
            Assert.assertEquals(dimension, schema.getColumnType().toString(), columnAnalysis.getType());
            Assert.assertEquals(dimension, 0, columnAnalysis.getSize());
            if (isString) {
                if (analyses == null) {
                    Assert.assertTrue(dimension, columnAnalysis.getCardinality() > 0);
                } else {
                    Assert.assertEquals(dimension, 0, columnAnalysis.getCardinality().longValue());
                }
            } else {
                Assert.assertNull(dimension, columnAnalysis.getCardinality());
            }
        }
    }
    for (String metric : TestIndex.DOUBLE_METRICS) {
        final ColumnAnalysis columnAnalysis = columns.get(metric);
        Assert.assertEquals(metric, ValueType.DOUBLE.name(), columnAnalysis.getType());
        Assert.assertEquals(metric, 0, columnAnalysis.getSize());
        Assert.assertNull(metric, columnAnalysis.getCardinality());
    }
    for (String metric : TestIndex.FLOAT_METRICS) {
        final ColumnAnalysis columnAnalysis = columns.get(metric);
        Assert.assertEquals(metric, ValueType.FLOAT.name(), columnAnalysis.getType());
        Assert.assertEquals(metric, 0, columnAnalysis.getSize());
        Assert.assertNull(metric, columnAnalysis.getCardinality());
    }
}

Example 17 with DimensionSchema

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

the class NewestSegmentFirstIterator method needsCompaction.

private boolean needsCompaction(DataSourceCompactionConfig config, SegmentsToCompact candidates) {
    Preconditions.checkState(!candidates.isEmpty(), "Empty candidates");
    final ClientCompactionTaskQueryTuningConfig tuningConfig = ClientCompactionTaskQueryTuningConfig.from(config.getTuningConfig(), config.getMaxRowsPerSegment());
    final PartitionsSpec partitionsSpecFromConfig = findPartitionsSpecFromConfig(tuningConfig);
    final CompactionState lastCompactionState = candidates.segments.get(0).getLastCompactionState();
    if (lastCompactionState == null) {
        log.info("Candidate segment[%s] is not compacted yet. Needs compaction.", candidates.segments.get(0).getId());
        return true;
    }
    final boolean allCandidatesHaveSameLastCompactionState = candidates.segments.stream().allMatch(segment -> lastCompactionState.equals(segment.getLastCompactionState()));
    if (!allCandidatesHaveSameLastCompactionState) {
        log.info("[%s] Candidate segments were compacted with different partitions spec. Needs compaction.", candidates.segments.size());
        log.debugSegments(candidates.segments, "Candidate segments compacted with different partiton spec");
        return true;
    }
    final PartitionsSpec segmentPartitionsSpec = lastCompactionState.getPartitionsSpec();
    final IndexSpec segmentIndexSpec = objectMapper.convertValue(lastCompactionState.getIndexSpec(), IndexSpec.class);
    final IndexSpec configuredIndexSpec;
    if (tuningConfig.getIndexSpec() == null) {
        configuredIndexSpec = new IndexSpec();
    } else {
        configuredIndexSpec = tuningConfig.getIndexSpec();
    }
    if (!Objects.equals(partitionsSpecFromConfig, segmentPartitionsSpec)) {
        log.info("Configured partitionsSpec[%s] is differenet from " + "the partitionsSpec[%s] of segments. Needs compaction.", partitionsSpecFromConfig, segmentPartitionsSpec);
        return true;
    }
    // segmentIndexSpec cannot be null.
    if (!segmentIndexSpec.equals(configuredIndexSpec)) {
        log.info("Configured indexSpec[%s] is different from the one[%s] of segments. Needs compaction", configuredIndexSpec, segmentIndexSpec);
        return true;
    }
    if (config.getGranularitySpec() != null) {
        final ClientCompactionTaskGranularitySpec existingGranularitySpec = lastCompactionState.getGranularitySpec() != null ? objectMapper.convertValue(lastCompactionState.getGranularitySpec(), ClientCompactionTaskGranularitySpec.class) : null;
        // Checks for segmentGranularity
        if (config.getGranularitySpec().getSegmentGranularity() != null) {
            final Granularity existingSegmentGranularity = existingGranularitySpec != null ? existingGranularitySpec.getSegmentGranularity() : null;
            if (existingSegmentGranularity == null) {
                // Candidate segments were all compacted without segment granularity set.
                // We need to check if all segments have the same segment granularity as the configured segment granularity.
                boolean needsCompaction = candidates.segments.stream().anyMatch(segment -> !config.getGranularitySpec().getSegmentGranularity().isAligned(segment.getInterval()));
                if (needsCompaction) {
                    log.info("Segments were previously compacted but without segmentGranularity in auto compaction." + " Configured segmentGranularity[%s] is different from granularity implied by segment intervals. Needs compaction", config.getGranularitySpec().getSegmentGranularity());
                    return true;
                }
            } else if (!config.getGranularitySpec().getSegmentGranularity().equals(existingSegmentGranularity)) {
                log.info("Configured segmentGranularity[%s] is different from the segmentGranularity[%s] of segments. Needs compaction", config.getGranularitySpec().getSegmentGranularity(), existingSegmentGranularity);
                return true;
            }
        }
        // Checks for rollup
        if (config.getGranularitySpec().isRollup() != null) {
            final Boolean existingRollup = existingGranularitySpec != null ? existingGranularitySpec.isRollup() : null;
            if (existingRollup == null || !config.getGranularitySpec().isRollup().equals(existingRollup)) {
                log.info("Configured rollup[%s] is different from the rollup[%s] of segments. Needs compaction", config.getGranularitySpec().isRollup(), existingRollup);
                return true;
            }
        }
        // Checks for queryGranularity
        if (config.getGranularitySpec().getQueryGranularity() != null) {
            final Granularity existingQueryGranularity = existingGranularitySpec != null ? existingGranularitySpec.getQueryGranularity() : null;
            if (!config.getGranularitySpec().getQueryGranularity().equals(existingQueryGranularity)) {
                log.info("Configured queryGranularity[%s] is different from the queryGranularity[%s] of segments. Needs compaction", config.getGranularitySpec().getQueryGranularity(), existingQueryGranularity);
                return true;
            }
        }
    }
    if (config.getDimensionsSpec() != null) {
        final DimensionsSpec existingDimensionsSpec = lastCompactionState.getDimensionsSpec();
        // Checks for list of dimensions
        if (config.getDimensionsSpec().getDimensions() != null) {
            final List<DimensionSchema> existingDimensions = existingDimensionsSpec != null ? existingDimensionsSpec.getDimensions() : null;
            if (!config.getDimensionsSpec().getDimensions().equals(existingDimensions)) {
                log.info("Configured dimensionsSpec is different from the dimensionsSpec of segments. Needs compaction");
                return true;
            }
        }
    }
    if (config.getTransformSpec() != null) {
        final ClientCompactionTaskTransformSpec existingTransformSpec = lastCompactionState.getTransformSpec() != null ? objectMapper.convertValue(lastCompactionState.getTransformSpec(), ClientCompactionTaskTransformSpec.class) : null;
        // Checks for filters
        if (config.getTransformSpec().getFilter() != null) {
            final DimFilter existingFilters = existingTransformSpec != null ? existingTransformSpec.getFilter() : null;
            if (!config.getTransformSpec().getFilter().equals(existingFilters)) {
                log.info("Configured filter[%s] is different from the filter[%s] of segments. Needs compaction", config.getTransformSpec().getFilter(), existingFilters);
                return true;
            }
        }
    }
    if (ArrayUtils.isNotEmpty(config.getMetricsSpec())) {
        final AggregatorFactory[] existingMetricsSpec = lastCompactionState.getMetricsSpec() == null || lastCompactionState.getMetricsSpec().isEmpty() ? null : objectMapper.convertValue(lastCompactionState.getMetricsSpec(), AggregatorFactory[].class);
        if (existingMetricsSpec == null || !Arrays.deepEquals(config.getMetricsSpec(), existingMetricsSpec)) {
            log.info("Configured metricsSpec[%s] is different from the metricsSpec[%s] of segments. Needs compaction", Arrays.toString(config.getMetricsSpec()), Arrays.toString(existingMetricsSpec));
            return true;
        }
    }
    return false;
}