use of org.apache.druid.java.util.common.parsers.JSONPathSpec in project druid by druid-io.
the class DecimalParquetReaderTest method testReadParquetDecimalFixedLen.
@Test
public void testReadParquetDecimalFixedLen() throws IOException {
final String file = "example/decimals/dec-in-fixed-len.parquet";
InputRowSchema schema = new InputRowSchema(new TimestampSpec("timestamp", "auto", DateTimes.of("2018-09-01T00:00:00.000Z")), new DimensionsSpec(DimensionsSpec.getDefaultSchemas(ImmutableList.of("fixed_len_dec"))), ColumnsFilter.all());
List<JSONPathFieldSpec> flattenExpr = ImmutableList.of(new JSONPathFieldSpec(JSONPathFieldType.ROOT, "fixed_len_dec", "fixed_len_dec"), new JSONPathFieldSpec(JSONPathFieldType.PATH, "metric1", "$.fixed_len_dec"));
JSONPathSpec flattenSpec = new JSONPathSpec(true, flattenExpr);
InputEntityReader reader = createReader(file, schema, flattenSpec);
List<InputRow> rows = readAllRows(reader);
Assert.assertEquals("2018-09-01T00:00:00.000Z", rows.get(1).getTimestamp().toString());
Assert.assertEquals("1.0", rows.get(1).getDimension("fixed_len_dec").get(0));
Assert.assertEquals(new BigDecimal("1.0"), rows.get(1).getMetric("metric1"));
reader = createReader(file, schema, flattenSpec);
List<InputRowListPlusRawValues> sampled = sampleAllRows(reader);
final String expectedJson = "{\n" + " \"fixed_len_dec\" : 1.0\n" + "}";
Assert.assertEquals(expectedJson, DEFAULT_JSON_WRITER.writeValueAsString(sampled.get(1).getRawValues()));
}
use of org.apache.druid.java.util.common.parsers.JSONPathSpec in project druid by druid-io.
the class DecimalParquetReaderTest method testReadParquetDecimali64.
@Test
public void testReadParquetDecimali64() throws IOException {
final String file = "example/decimals/dec-in-i64.parquet";
InputRowSchema schema = new InputRowSchema(new TimestampSpec("timestamp", "auto", DateTimes.of("2018-09-01T00:00:00.000Z")), new DimensionsSpec(DimensionsSpec.getDefaultSchemas(ImmutableList.of("i64_dec"))), ColumnsFilter.all());
List<JSONPathFieldSpec> flattenExpr = ImmutableList.of(new JSONPathFieldSpec(JSONPathFieldType.ROOT, "i32_dec", "i64_dec"), new JSONPathFieldSpec(JSONPathFieldType.PATH, "metric1", "$.i64_dec"));
JSONPathSpec flattenSpec = new JSONPathSpec(true, flattenExpr);
InputEntityReader reader = createReader(file, schema, flattenSpec);
List<InputRow> rows = readAllRows(reader);
Assert.assertEquals("2018-09-01T00:00:00.000Z", rows.get(1).getTimestamp().toString());
Assert.assertEquals("100", rows.get(1).getDimension("i64_dec").get(0));
Assert.assertEquals(new BigDecimal(100), rows.get(1).getMetric("metric1"));
reader = createReader(file, schema, flattenSpec);
List<InputRowListPlusRawValues> sampled = sampleAllRows(reader);
final String expectedJson = "{\n" + " \"i64_dec\" : 100\n" + "}";
Assert.assertEquals(expectedJson, DEFAULT_JSON_WRITER.writeValueAsString(sampled.get(1).getRawValues()));
}
use of org.apache.druid.java.util.common.parsers.JSONPathSpec in project druid by druid-io.
the class FlattenSpecParquetReaderTest method testFlat1FlattenSelectListItem.
@Test
public void testFlat1FlattenSelectListItem() throws IOException {
final String file = "example/flattening/test_flat_1.parquet";
InputRowSchema schema = new InputRowSchema(new TimestampSpec("timestamp", "auto", null), new DimensionsSpec(DimensionsSpec.getDefaultSchemas(ImmutableList.of("dim1", "dim2", "listExtracted"))), ColumnsFilter.all());
List<JSONPathFieldSpec> flattenExpr = ImmutableList.of(new JSONPathFieldSpec(JSONPathFieldType.ROOT, "timestamp", null), new JSONPathFieldSpec(JSONPathFieldType.ROOT, "dim1", null), new JSONPathFieldSpec(JSONPathFieldType.ROOT, "dim2", null), new JSONPathFieldSpec(JSONPathFieldType.PATH, "listExtracted", "$.listDim[1]"));
JSONPathSpec flattenSpec = new JSONPathSpec(false, flattenExpr);
InputEntityReader reader = createReader(file, schema, flattenSpec);
List<InputRow> rows = readAllRows(reader);
Assert.assertEquals(FlattenSpecParquetInputTest.TS1, rows.get(0).getTimestamp().toString());
Assert.assertEquals("d1v1", rows.get(0).getDimension("dim1").get(0));
Assert.assertEquals("d2v1", rows.get(0).getDimension("dim2").get(0));
Assert.assertEquals("listDim1v2", rows.get(0).getDimension("listExtracted").get(0));
Assert.assertEquals(1, rows.get(0).getMetric("metric1").longValue());
reader = createReader(file, schema, flattenSpec);
List<InputRowListPlusRawValues> sampled = sampleAllRows(reader);
Assert.assertEquals(FLAT_JSON, DEFAULT_JSON_WRITER.writeValueAsString(sampled.get(0).getRawValues()));
}
use of org.apache.druid.java.util.common.parsers.JSONPathSpec in project druid by druid-io.
the class FlattenSpecParquetReaderTest method testNested1FlattenSelectListItem.
@Test
public void testNested1FlattenSelectListItem() throws IOException {
final String file = "example/flattening/test_nested_1.parquet";
InputRowSchema schema = new InputRowSchema(new TimestampSpec("timestamp", "auto", null), new DimensionsSpec(DimensionsSpec.getDefaultSchemas(ImmutableList.of())), ColumnsFilter.all());
List<JSONPathFieldSpec> flattenExpr = ImmutableList.of(new JSONPathFieldSpec(JSONPathFieldType.ROOT, "timestamp", null), new JSONPathFieldSpec(JSONPathFieldType.ROOT, "dim1", null), new JSONPathFieldSpec(JSONPathFieldType.PATH, "dim2", "$.nestedData.dim2"), new JSONPathFieldSpec(JSONPathFieldType.PATH, "dim3", "$.nestedData.dim3"), new JSONPathFieldSpec(JSONPathFieldType.PATH, "listextracted", "$.nestedData.listDim[1]"));
JSONPathSpec flattenSpec = new JSONPathSpec(true, flattenExpr);
InputEntityReader reader = createReader(file, schema, flattenSpec);
List<InputRow> rows = readAllRows(reader);
Assert.assertEquals(FlattenSpecParquetInputTest.TS1, rows.get(0).getTimestamp().toString());
Assert.assertEquals("d1v1", rows.get(0).getDimension("dim1").get(0));
Assert.assertEquals("d2v1", rows.get(0).getDimension("dim2").get(0));
Assert.assertEquals("1", rows.get(0).getDimension("dim3").get(0));
Assert.assertEquals("listDim1v2", rows.get(0).getDimension("listextracted").get(0));
Assert.assertEquals(1, rows.get(0).getMetric("metric1").longValue());
reader = createReader(file, schema, flattenSpec);
List<InputRowListPlusRawValues> sampled = sampleAllRows(reader);
Assert.assertEquals(NESTED_JSON, DEFAULT_JSON_WRITER.writeValueAsString(sampled.get(0).getRawValues()));
}
use of org.apache.druid.java.util.common.parsers.JSONPathSpec in project druid by druid-io.
the class FlattenSpecParquetReaderTest method testFlat1NoFlattenSpec.
@Test
public void testFlat1NoFlattenSpec() throws IOException {
final String file = "example/flattening/test_flat_1.parquet";
InputRowSchema schema = new InputRowSchema(new TimestampSpec("timestamp", "auto", null), new DimensionsSpec(DimensionsSpec.getDefaultSchemas(ImmutableList.of("dim1", "dim2", "dim3", "listDim"))), ColumnsFilter.all());
JSONPathSpec flattenSpec = new JSONPathSpec(false, ImmutableList.of());
InputEntityReader reader = createReader(file, schema, flattenSpec);
List<InputRow> rows = readAllRows(reader);
Assert.assertEquals(FlattenSpecParquetInputTest.TS1, rows.get(0).getTimestamp().toString());
Assert.assertEquals("d1v1", rows.get(0).getDimension("dim1").get(0));
Assert.assertEquals("d2v1", rows.get(0).getDimension("dim2").get(0));
Assert.assertEquals("1", rows.get(0).getDimension("dim3").get(0));
Assert.assertEquals("listDim1v1", rows.get(0).getDimension("listDim").get(0));
Assert.assertEquals("listDim1v2", rows.get(0).getDimension("listDim").get(1));
Assert.assertEquals(1, rows.get(0).getMetric("metric1").longValue());
reader = createReader(file, schema, flattenSpec);
List<InputRowListPlusRawValues> sampled = sampleAllRows(reader);
Assert.assertEquals(FLAT_JSON, DEFAULT_JSON_WRITER.writeValueAsString(sampled.get(0).getRawValues()));
}
Aggregations