Examples with RegexDimExtractionFn io.druid.query.extraction.RegexDimExtractionFn used on opensource projects

Example 1 with RegexDimExtractionFn

use of io.druid.query.extraction.RegexDimExtractionFn in project druid by druid-io.

the class ExtractionDimensionSpecTest method testSerdeWithType.

@Test
public void testSerdeWithType() throws Exception {
    final ObjectMapper objectMapper = new DefaultObjectMapper();
    final String oldJson = "{\n" + "    \"type\": \"extraction\",\n" + "    \"outputName\": \"first3Letters\",\n" + "    \"outputType\": \"LONG\",\n" + "    \"dimension\": \"myDim\"," + "    \"extractionFn\": {\n" + "        \"type\": \"regex\",\n" + "        \"expr\": \"(...).*\"\n" + "    }\n" + "}";
    final ExtractionDimensionSpec extractionDimensionSpec = (ExtractionDimensionSpec) objectMapper.readValue(oldJson, DimensionSpec.class);
    Assert.assertEquals("first3Letters", extractionDimensionSpec.getOutputName());
    Assert.assertEquals("myDim", extractionDimensionSpec.getDimension());
    Assert.assertNotNull(extractionDimensionSpec.getExtractionFn());
    Assert.assertEquals(ValueType.LONG, extractionDimensionSpec.getOutputType());
    Assert.assertTrue(extractionDimensionSpec.getExtractionFn() instanceof RegexDimExtractionFn);
    Assert.assertEquals(extractionDimensionSpec, objectMapper.readValue(objectMapper.writeValueAsBytes(extractionDimensionSpec), DimensionSpec.class));
}

Example 2 with RegexDimExtractionFn

use of io.druid.query.extraction.RegexDimExtractionFn in project druid by druid-io.

the class ExtractionDimensionSpecTest method testSerdeBackwardsCompatibility.

@Test
public void testSerdeBackwardsCompatibility() throws Exception {
    final ObjectMapper objectMapper = new DefaultObjectMapper();
    final String oldJson = "{\n" + "    \"type\": \"extraction\",\n" + "    \"outputName\": \"first3Letters\",\n" + "    \"dimension\": \"myDim\"," + "    \"dimExtractionFn\": {\n" + "        \"type\": \"regex\",\n" + "        \"expr\": \"(...).*\"\n" + "    }\n" + "}";
    final ExtractionDimensionSpec extractionDimensionSpec = (ExtractionDimensionSpec) objectMapper.readValue(oldJson, DimensionSpec.class);
    Assert.assertEquals("first3Letters", extractionDimensionSpec.getOutputName());
    Assert.assertEquals("myDim", extractionDimensionSpec.getDimension());
    Assert.assertNotNull(extractionDimensionSpec.getExtractionFn());
    Assert.assertTrue(extractionDimensionSpec.getExtractionFn() instanceof RegexDimExtractionFn);
    Assert.assertEquals(extractionDimensionSpec, objectMapper.readValue(objectMapper.writeValueAsBytes(extractionDimensionSpec), DimensionSpec.class));
    // new trumps old
    final String oldAndNewJson = "{\n" + "    \"type\": \"extraction\",\n" + "    \"outputName\": \"first3Letters\",\n" + "    \"dimension\": \"myDim\"," + "    \"extractionFn\": {\n" + "        \"type\": \"partial\",\n" + "        \"expr\": \"(...).*\"\n" + "    },\n" + "    \"dimExtractionFn\": {\n" + "        \"type\": \"regex\",\n" + "        \"expr\": \"(...).*\"\n" + "    }\n" + "}";
    Assert.assertTrue(objectMapper.readValue(oldAndNewJson, DimensionSpec.class).getExtractionFn() instanceof MatchingDimExtractionFn);
}

Example 3 with RegexDimExtractionFn

use of io.druid.query.extraction.RegexDimExtractionFn in project druid by druid-io.

the class GroupByQueryRunnerTest method testSubqueryWithExtractionFnInOuterQuery.

@Test
public void testSubqueryWithExtractionFnInOuterQuery() {
    //https://github.com/druid-io/druid/issues/2556
    GroupByQuery subquery = GroupByQuery.builder().setDataSource(QueryRunnerTestHelper.dataSource).setQuerySegmentSpec(QueryRunnerTestHelper.firstToThird).setDimensions(Lists.<DimensionSpec>newArrayList(new DefaultDimensionSpec("quality", "alias"))).setDimFilter(new JavaScriptDimFilter("quality", "function(dim){ return true; }", null, JavaScriptConfig.getEnabledInstance())).setAggregatorSpecs(Arrays.asList(QueryRunnerTestHelper.rowsCount, new LongSumAggregatorFactory("idx", "index"), new LongSumAggregatorFactory("indexMaxPlusTen", "indexMaxPlusTen"))).setGranularity(QueryRunnerTestHelper.dayGran).build();
    GroupByQuery query = GroupByQuery.builder().setDataSource(subquery).setQuerySegmentSpec(new MultipleIntervalSegmentSpec(ImmutableList.of(new Interval("2011-04-01T00:00:00.000Z/2011-04-03T00:00:00.000Z")))).setDimensions(Lists.<DimensionSpec>newArrayList(new ExtractionDimensionSpec("alias", "alias", new RegexDimExtractionFn("(a).*", true, "a")))).setAggregatorSpecs(Arrays.<AggregatorFactory>asList(new LongSumAggregatorFactory("rows", "rows"), new LongSumAggregatorFactory("idx", "idx"))).setGranularity(QueryRunnerTestHelper.dayGran).build();
    List<Row> expectedResults = Arrays.asList(GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-01", "alias", "a", "rows", 13L, "idx", 6619L), GroupByQueryRunnerTestHelper.createExpectedRow("2011-04-02", "alias", "a", "rows", 13L, "idx", 5827L));
    // Subqueries are handled by the ToolChest
    Iterable<Row> results = GroupByQueryRunnerTestHelper.runQuery(factory, runner, query);
    TestHelper.assertExpectedObjects(expectedResults, results, "");
}

Example 4 with RegexDimExtractionFn

use of io.druid.query.extraction.RegexDimExtractionFn in project druid by druid-io.

the class SearchQueryDimFilterTest method testGetCacheKey.

@Test
public void testGetCacheKey() {
    SearchQueryDimFilter searchQueryDimFilter = new SearchQueryDimFilter("dim", new SearchQuerySpec() {

        @Override
        public boolean accept(String dimVal) {
            return false;
        }

        @Override
        public byte[] getCacheKey() {
            return StringUtils.toUtf8("value");
        }
    }, null);
    SearchQueryDimFilter searchQueryDimFilter2 = new SearchQueryDimFilter("di", new SearchQuerySpec() {

        @Override
        public boolean accept(String dimVal) {
            return false;
        }

        @Override
        public byte[] getCacheKey() {
            return StringUtils.toUtf8("mvalue");
        }
    }, null);
    Assert.assertFalse(Arrays.equals(searchQueryDimFilter.getCacheKey(), searchQueryDimFilter2.getCacheKey()));
    RegexDimExtractionFn regexFn = new RegexDimExtractionFn(".*", false, null);
    SearchQueryDimFilter searchQueryDimFilter3 = new SearchQueryDimFilter("dim", new SearchQuerySpec() {

        @Override
        public boolean accept(String dimVal) {
            return false;
        }

        @Override
        public byte[] getCacheKey() {
            return StringUtils.toUtf8("value");
        }
    }, regexFn);
    Assert.assertFalse(Arrays.equals(searchQueryDimFilter.getCacheKey(), searchQueryDimFilter3.getCacheKey()));
}

Example 5 with RegexDimExtractionFn

use of io.druid.query.extraction.RegexDimExtractionFn in project druid by druid-io.

the class SearchQueryDimFilterTest method testHashcode.

@Test
public void testHashcode() {
    SearchQueryDimFilter searchQueryDimFilter = new SearchQueryDimFilter("dim", new SearchQuerySpec() {

        @Override
        public boolean accept(String dimVal) {
            return false;
        }

        @Override
        public byte[] getCacheKey() {
            return StringUtils.toUtf8("value");
        }
    }, null);
    SearchQueryDimFilter searchQueryDimFilter2 = new SearchQueryDimFilter("di", new SearchQuerySpec() {

        @Override
        public boolean accept(String dimVal) {
            return false;
        }

        @Override
        public byte[] getCacheKey() {
            return StringUtils.toUtf8("mvalue");
        }
    }, null);
    Assert.assertNotEquals(searchQueryDimFilter.hashCode(), searchQueryDimFilter2.hashCode());
    RegexDimExtractionFn regexFn = new RegexDimExtractionFn(".*", false, null);
    SearchQueryDimFilter searchQueryDimFilter3 = new SearchQueryDimFilter("dim", new SearchQuerySpec() {

        @Override
        public boolean accept(String dimVal) {
            return false;
        }

        @Override
        public byte[] getCacheKey() {
            return StringUtils.toUtf8("value");
        }
    }, regexFn);
    Assert.assertNotEquals(searchQueryDimFilter.hashCode(), searchQueryDimFilter3.hashCode());
}