use of org.apache.druid.java.util.common.parsers.JSONPathFieldSpec in project druid by druid-io.
the class OrcReaderTest method testOrcFile11Format.
// This test is migrated from OrcHadoopInputRowParserTest
@Test
public void testOrcFile11Format() throws IOException {
final OrcInputFormat inputFormat = new OrcInputFormat(new JSONPathSpec(true, ImmutableList.of(new JSONPathFieldSpec(JSONPathFieldType.PATH, "struct_list_struct_int", "$.middle.list[1].int1"), new JSONPathFieldSpec(JSONPathFieldType.PATH, "struct_list_struct_intlist", "$.middle.list[*].int1"), new JSONPathFieldSpec(JSONPathFieldType.PATH, "struct_list_struct_middleListLength", "$.middle.list.length()"), new JSONPathFieldSpec(JSONPathFieldType.PATH, "list_struct_string", "$.list[0].string1"), new JSONPathFieldSpec(JSONPathFieldType.PATH, "map_struct_int", "$.map.chani.int1"))), null, new Configuration());
final InputEntityReader reader = createReader(new TimestampSpec("ts", "millis", null), new DimensionsSpec(null), inputFormat, "example/orc-file-11-format.orc");
try (CloseableIterator<InputRow> iterator = reader.read()) {
int actualRowCount = 0;
// Check the first row
Assert.assertTrue(iterator.hasNext());
InputRow row = iterator.next();
actualRowCount++;
Assert.assertEquals("false", Iterables.getOnlyElement(row.getDimension("boolean1")));
Assert.assertEquals("1", Iterables.getOnlyElement(row.getDimension("byte1")));
Assert.assertEquals("1024", Iterables.getOnlyElement(row.getDimension("short1")));
Assert.assertEquals("65536", Iterables.getOnlyElement(row.getDimension("int1")));
Assert.assertEquals("9223372036854775807", Iterables.getOnlyElement(row.getDimension("long1")));
Assert.assertEquals("1.0", Iterables.getOnlyElement(row.getDimension("float1")));
Assert.assertEquals("-15.0", Iterables.getOnlyElement(row.getDimension("double1")));
Assert.assertEquals("AAECAwQAAA==", Iterables.getOnlyElement(row.getDimension("bytes1")));
Assert.assertEquals("hi", Iterables.getOnlyElement(row.getDimension("string1")));
Assert.assertEquals("1.23456786547456E7", Iterables.getOnlyElement(row.getDimension("decimal1")));
Assert.assertEquals("2", Iterables.getOnlyElement(row.getDimension("struct_list_struct_int")));
Assert.assertEquals(ImmutableList.of("1", "2"), row.getDimension("struct_list_struct_intlist"));
Assert.assertEquals("good", Iterables.getOnlyElement(row.getDimension("list_struct_string")));
Assert.assertEquals("2", Iterables.getOnlyElement(row.getDimension("struct_list_struct_middleListLength")));
Assert.assertEquals(DateTimes.of("2000-03-12T15:00:00.0Z"), row.getTimestamp());
while (iterator.hasNext()) {
actualRowCount++;
row = iterator.next();
}
// Check the last row
Assert.assertEquals("true", Iterables.getOnlyElement(row.getDimension("boolean1")));
Assert.assertEquals("100", Iterables.getOnlyElement(row.getDimension("byte1")));
Assert.assertEquals("2048", Iterables.getOnlyElement(row.getDimension("short1")));
Assert.assertEquals("65536", Iterables.getOnlyElement(row.getDimension("int1")));
Assert.assertEquals("9223372036854775807", Iterables.getOnlyElement(row.getDimension("long1")));
Assert.assertEquals("2.0", Iterables.getOnlyElement(row.getDimension("float1")));
Assert.assertEquals("-5.0", Iterables.getOnlyElement(row.getDimension("double1")));
Assert.assertEquals("", Iterables.getOnlyElement(row.getDimension("bytes1")));
Assert.assertEquals("bye", Iterables.getOnlyElement(row.getDimension("string1")));
Assert.assertEquals("1.23456786547457E7", Iterables.getOnlyElement(row.getDimension("decimal1")));
Assert.assertEquals("2", Iterables.getOnlyElement(row.getDimension("struct_list_struct_int")));
Assert.assertEquals(ImmutableList.of("1", "2"), row.getDimension("struct_list_struct_intlist"));
Assert.assertEquals("cat", Iterables.getOnlyElement(row.getDimension("list_struct_string")));
Assert.assertEquals("5", Iterables.getOnlyElement(row.getDimension("map_struct_int")));
Assert.assertEquals(DateTimes.of("2000-03-12T15:00:01.000Z"), row.getTimestamp());
Assert.assertEquals(7500, actualRowCount);
}
}
use of org.apache.druid.java.util.common.parsers.JSONPathFieldSpec in project druid by druid-io.
the class OrcReaderTest method testTest2.
// This test is migrated from OrcHadoopInputRowParserTest
@Test
public void testTest2() throws IOException {
final InputFormat inputFormat = new OrcInputFormat(new JSONPathSpec(true, Collections.singletonList(new JSONPathFieldSpec(JSONPathFieldType.PATH, "col7-subcol7", "$.col7.subcol7"))), null, new Configuration());
final InputEntityReader reader = createReader(new TimestampSpec("timestamp", "auto", null), new DimensionsSpec(null), inputFormat, "example/test_2.orc");
try (CloseableIterator<InputRow> iterator = reader.read()) {
Assert.assertTrue(iterator.hasNext());
final InputRow row = iterator.next();
Assert.assertEquals(DateTimes.of("2016-01-01T00:00:00.000Z"), row.getTimestamp());
Assert.assertEquals("bar", Iterables.getOnlyElement(row.getDimension("col1")));
Assert.assertEquals(ImmutableList.of("dat1", "dat2", "dat3"), row.getDimension("col2"));
Assert.assertEquals("1.1", Iterables.getOnlyElement(row.getDimension("col3")));
Assert.assertEquals("2", Iterables.getOnlyElement(row.getDimension("col4")));
Assert.assertEquals("3.5", Iterables.getOnlyElement(row.getDimension("col5")));
Assert.assertTrue(row.getDimension("col6").isEmpty());
Assert.assertFalse(iterator.hasNext());
}
}
use of org.apache.druid.java.util.common.parsers.JSONPathFieldSpec in project druid by druid-io.
the class InputRowParserSerdeTest method testFlattenParse.
@Test
public void testFlattenParse() throws Exception {
List<JSONPathFieldSpec> fields = new ArrayList<>();
fields.add(JSONPathFieldSpec.createNestedField("foobar1", "$.foo.bar1"));
fields.add(JSONPathFieldSpec.createNestedField("foobar2", "$.foo.bar2"));
fields.add(JSONPathFieldSpec.createNestedField("baz0", "$.baz[0]"));
fields.add(JSONPathFieldSpec.createNestedField("baz1", "$.baz[1]"));
fields.add(JSONPathFieldSpec.createNestedField("baz2", "$.baz[2]"));
fields.add(JSONPathFieldSpec.createNestedField("hey0barx", "$.hey[0].barx"));
fields.add(JSONPathFieldSpec.createNestedField("metA", "$.met.a"));
fields.add(JSONPathFieldSpec.createNestedField("missing", "$.nonexistent.nested.field"));
fields.add(JSONPathFieldSpec.createRootField("timestamp"));
fields.add(JSONPathFieldSpec.createRootField("foo.bar1"));
JSONPathSpec flattenSpec = new JSONPathSpec(true, fields);
final StringInputRowParser parser = new StringInputRowParser(new JSONParseSpec(new TimestampSpec("timestamp", "iso", null), DimensionsSpec.EMPTY, flattenSpec, null, null), null);
final StringInputRowParser parser2 = jsonMapper.readValue(jsonMapper.writeValueAsBytes(parser), StringInputRowParser.class);
final InputRow parsed = parser2.parse("{\"blah\":[4,5,6], \"newmet\":5, \"foo\":{\"bar1\":\"aaa\", \"bar2\":\"bbb\"}, \"baz\":[1,2,3], \"timestamp\":\"2999\", \"foo.bar1\":\"Hello world!\", \"hey\":[{\"barx\":\"asdf\"}], \"met\":{\"a\":456}}");
Assert.assertEquals(ImmutableList.of("foobar1", "foobar2", "baz0", "baz1", "baz2", "hey0barx", "metA", "missing", "timestamp", "foo.bar1", "blah", "newmet", "baz"), parsed.getDimensions());
Assert.assertEquals(ImmutableList.of("aaa"), parsed.getDimension("foobar1"));
Assert.assertEquals(ImmutableList.of("bbb"), parsed.getDimension("foobar2"));
Assert.assertEquals(ImmutableList.of("1"), parsed.getDimension("baz0"));
Assert.assertEquals(ImmutableList.of("2"), parsed.getDimension("baz1"));
Assert.assertEquals(ImmutableList.of("3"), parsed.getDimension("baz2"));
Assert.assertEquals(ImmutableList.of("Hello world!"), parsed.getDimension("foo.bar1"));
Assert.assertEquals(ImmutableList.of("asdf"), parsed.getDimension("hey0barx"));
Assert.assertEquals(ImmutableList.of("456"), parsed.getDimension("metA"));
Assert.assertEquals(ImmutableList.of("5"), parsed.getDimension("newmet"));
Assert.assertEquals(ImmutableList.of(), parsed.getDimension("missing"));
Assert.assertEquals(DateTimes.of("2999").getMillis(), parsed.getTimestampFromEpoch());
String testSpec = "{\"enabled\": true,\"useFieldDiscovery\": true, \"fields\": [\"parseThisRootField\"]}";
final JSONPathSpec parsedSpec = jsonMapper.readValue(testSpec, JSONPathSpec.class);
List<JSONPathFieldSpec> fieldSpecs = parsedSpec.getFields();
Assert.assertEquals(JSONPathFieldType.ROOT, fieldSpecs.get(0).getType());
Assert.assertEquals("parseThisRootField", fieldSpecs.get(0).getName());
Assert.assertEquals("parseThisRootField", fieldSpecs.get(0).getExpr());
}
use of org.apache.druid.java.util.common.parsers.JSONPathFieldSpec in project druid by druid-io.
the class JSONParseSpecTest method testParseRowWithConditional.
@Test
public void testParseRowWithConditional() {
final JSONParseSpec parseSpec = new JSONParseSpec(new TimestampSpec("timestamp", "iso", null), new DimensionsSpec(DimensionsSpec.getDefaultSchemas(ImmutableList.of("foo"))), new JSONPathSpec(true, ImmutableList.of(new JSONPathFieldSpec(JSONPathFieldType.PATH, "foo", "$.[?(@.maybe_object)].maybe_object.foo.test"), new JSONPathFieldSpec(JSONPathFieldType.PATH, "baz", "$.maybe_object_2.foo.test"), new JSONPathFieldSpec(JSONPathFieldType.PATH, "bar", "$.[?(@.something_else)].something_else.foo"))), null, false);
final Map<String, Object> expected = new HashMap<>();
expected.put("foo", new ArrayList());
expected.put("baz", null);
expected.put("bar", Collections.singletonList("test"));
final Parser<String, Object> parser = parseSpec.makeParser();
final Map<String, Object> parsedRow = parser.parseToMap("{\"something_else\": {\"foo\": \"test\"}}");
Assert.assertNotNull(parsedRow);
Assert.assertEquals(expected, parsedRow);
}
use of org.apache.druid.java.util.common.parsers.JSONPathFieldSpec in project druid by druid-io.
the class JsonLineReaderTest method testFalseKeepNullColumns.
@Test
public void testFalseKeepNullColumns() throws IOException {
final JsonInputFormat format = new JsonInputFormat(new JSONPathSpec(true, ImmutableList.of(new JSONPathFieldSpec(JSONPathFieldType.PATH, "path_omg", "$.o.mg"))), null, false);
final ByteEntity source = new ByteEntity(StringUtils.toUtf8("{\"timestamp\":\"2019-01-01\",\"bar\":null,\"foo\":\"x\",\"o\":{\"mg\":\"a\"}}"));
final InputEntityReader reader = format.createReader(new InputRowSchema(new TimestampSpec("timestamp", "iso", null), new DimensionsSpec(DimensionsSpec.getDefaultSchemas(Collections.emptyList())), ColumnsFilter.all()), source, null);
final int numExpectedIterations = 1;
try (CloseableIterator<InputRow> iterator = reader.read()) {
int numActualIterations = 0;
while (iterator.hasNext()) {
final InputRow row = iterator.next();
Assert.assertEquals(Arrays.asList("path_omg", "timestamp", "foo"), row.getDimensions());
Assert.assertTrue(row.getDimension("bar").isEmpty());
Assert.assertEquals("x", Iterables.getOnlyElement(row.getDimension("foo")));
Assert.assertEquals("a", Iterables.getOnlyElement(row.getDimension("path_omg")));
numActualIterations++;
}
Assert.assertEquals(numExpectedIterations, numActualIterations);
}
}
Aggregations