Search in sources :

Example 41 with SimpleRecordSchema

use of org.apache.nifi.serialization.SimpleRecordSchema in project nifi by apache.

the class TestJsonTreeRowRecordReader method testPerformanceOnIndividualMessages.

@Test
@Ignore("Intended only for manual testing to determine performance before/after modifications")
public void testPerformanceOnIndividualMessages() throws IOException, MalformedRecordException {
    final RecordSchema schema = new SimpleRecordSchema(Collections.emptyList());
    final File file = new File("/devel/nifi/nifi-assembly/target/nifi-1.2.0-SNAPSHOT-bin/nifi-1.2.0-SNAPSHOT/1.prov.json");
    final byte[] data = Files.readAllBytes(file.toPath());
    final ComponentLog logger = Mockito.mock(ComponentLog.class);
    int recordCount = 0;
    final int iterations = 1_000_000;
    for (int j = 0; j < 5; j++) {
        final long start = System.nanoTime();
        for (int i = 0; i < iterations; i++) {
            try (final InputStream in = new ByteArrayInputStream(data);
                final JsonTreeRowRecordReader reader = new JsonTreeRowRecordReader(in, logger, schema, dateFormat, timeFormat, timestampFormat)) {
                while (reader.nextRecord() != null) {
                    recordCount++;
                }
            }
        }
        final long nanos = System.nanoTime() - start;
        final long millis = TimeUnit.NANOSECONDS.toMillis(nanos);
        System.out.println("Took " + millis + " millis to read " + recordCount + " records");
    }
}
Also used : SimpleRecordSchema(org.apache.nifi.serialization.SimpleRecordSchema) ByteArrayInputStream(java.io.ByteArrayInputStream) ByteArrayInputStream(java.io.ByteArrayInputStream) FileInputStream(java.io.FileInputStream) InputStream(java.io.InputStream) RecordSchema(org.apache.nifi.serialization.record.RecordSchema) SimpleRecordSchema(org.apache.nifi.serialization.SimpleRecordSchema) File(java.io.File) ComponentLog(org.apache.nifi.logging.ComponentLog) Ignore(org.junit.Ignore) Test(org.junit.Test)

Example 42 with SimpleRecordSchema

use of org.apache.nifi.serialization.SimpleRecordSchema in project nifi by apache.

the class TestJsonTreeRowRecordReader method testReadArrayDifferentSchemasWithOptionalElementOverridden.

@Test
public void testReadArrayDifferentSchemasWithOptionalElementOverridden() throws IOException, MalformedRecordException {
    final RecordSchema schema = new SimpleRecordSchema(getDefaultFields());
    try (final InputStream in = new FileInputStream(new File("src/test/resources/json/bank-account-array-optional-balance.json"));
        final JsonTreeRowRecordReader reader = new JsonTreeRowRecordReader(in, Mockito.mock(ComponentLog.class), schema, dateFormat, timeFormat, timestampFormat)) {
        final List<String> fieldNames = schema.getFieldNames();
        final List<String> expectedFieldNames = Arrays.asList(new String[] { "id", "name", "balance", "address", "city", "state", "zipCode", "country" });
        assertEquals(expectedFieldNames, fieldNames);
        final List<RecordFieldType> dataTypes = schema.getDataTypes().stream().map(dt -> dt.getFieldType()).collect(Collectors.toList());
        final List<RecordFieldType> expectedTypes = Arrays.asList(new RecordFieldType[] { RecordFieldType.INT, RecordFieldType.STRING, RecordFieldType.DOUBLE, RecordFieldType.STRING, RecordFieldType.STRING, RecordFieldType.STRING, RecordFieldType.STRING, RecordFieldType.STRING });
        assertEquals(expectedTypes, dataTypes);
        final Object[] firstRecordValues = reader.nextRecord().getValues();
        Assert.assertArrayEquals(new Object[] { 1, "John Doe", 4750.89, "123 My Street", "My City", "MS", "11111", "USA" }, firstRecordValues);
        final Object[] secondRecordValues = reader.nextRecord().getValues();
        Assert.assertArrayEquals(new Object[] { 2, "Jane Doe", null, "321 Your Street", "Your City", "NY", "33333", "USA" }, secondRecordValues);
        final Object[] thirdRecordValues = reader.nextRecord().getValues();
        Assert.assertArrayEquals(new Object[] { 3, "Jimmy Doe", null, "321 Your Street", "Your City", "NY", "33333", "USA" }, thirdRecordValues);
        assertNull(reader.nextRecord());
    }
}
Also used : SimpleRecordSchema(org.apache.nifi.serialization.SimpleRecordSchema) Arrays(java.util.Arrays) DataType(org.apache.nifi.serialization.record.DataType) HashMap(java.util.HashMap) ComponentLog(org.apache.nifi.logging.ComponentLog) ArrayList(java.util.ArrayList) ChoiceDataType(org.apache.nifi.serialization.record.type.ChoiceDataType) RecordSchema(org.apache.nifi.serialization.record.RecordSchema) ByteArrayInputStream(java.io.ByteArrayInputStream) Map(java.util.Map) Record(org.apache.nifi.serialization.record.Record) MalformedRecordException(org.apache.nifi.serialization.MalformedRecordException) Files(java.nio.file.Files) RecordField(org.apache.nifi.serialization.record.RecordField) Assert.assertTrue(org.junit.Assert.assertTrue) IOException(java.io.IOException) SimpleRecordSchema(org.apache.nifi.serialization.SimpleRecordSchema) Test(org.junit.Test) FileInputStream(java.io.FileInputStream) Collectors(java.util.stream.Collectors) File(java.io.File) TimeUnit(java.util.concurrent.TimeUnit) Mockito(org.mockito.Mockito) List(java.util.List) Assert.assertNull(org.junit.Assert.assertNull) Ignore(org.junit.Ignore) Assert(org.junit.Assert) Collections(java.util.Collections) RecordFieldType(org.apache.nifi.serialization.record.RecordFieldType) Assert.assertEquals(org.junit.Assert.assertEquals) InputStream(java.io.InputStream) ByteArrayInputStream(java.io.ByteArrayInputStream) FileInputStream(java.io.FileInputStream) InputStream(java.io.InputStream) ComponentLog(org.apache.nifi.logging.ComponentLog) FileInputStream(java.io.FileInputStream) RecordSchema(org.apache.nifi.serialization.record.RecordSchema) SimpleRecordSchema(org.apache.nifi.serialization.SimpleRecordSchema) File(java.io.File) RecordFieldType(org.apache.nifi.serialization.record.RecordFieldType) Test(org.junit.Test)

Example 43 with SimpleRecordSchema

use of org.apache.nifi.serialization.SimpleRecordSchema in project nifi by apache.

the class TestJsonTreeRowRecordReader method testReadArray.

@Test
public void testReadArray() throws IOException, MalformedRecordException {
    final RecordSchema schema = new SimpleRecordSchema(getDefaultFields());
    try (final InputStream in = new FileInputStream(new File("src/test/resources/json/bank-account-array.json"));
        final JsonTreeRowRecordReader reader = new JsonTreeRowRecordReader(in, Mockito.mock(ComponentLog.class), schema, dateFormat, timeFormat, timestampFormat)) {
        final List<String> fieldNames = schema.getFieldNames();
        final List<String> expectedFieldNames = Arrays.asList(new String[] { "id", "name", "balance", "address", "city", "state", "zipCode", "country" });
        assertEquals(expectedFieldNames, fieldNames);
        final List<RecordFieldType> dataTypes = schema.getDataTypes().stream().map(dt -> dt.getFieldType()).collect(Collectors.toList());
        final List<RecordFieldType> expectedTypes = Arrays.asList(new RecordFieldType[] { RecordFieldType.INT, RecordFieldType.STRING, RecordFieldType.DOUBLE, RecordFieldType.STRING, RecordFieldType.STRING, RecordFieldType.STRING, RecordFieldType.STRING, RecordFieldType.STRING });
        assertEquals(expectedTypes, dataTypes);
        final Object[] firstRecordValues = reader.nextRecord().getValues();
        Assert.assertArrayEquals(new Object[] { 1, "John Doe", 4750.89, "123 My Street", "My City", "MS", "11111", "USA" }, firstRecordValues);
        final Object[] secondRecordValues = reader.nextRecord().getValues();
        Assert.assertArrayEquals(new Object[] { 2, "Jane Doe", 4820.09, "321 Your Street", "Your City", "NY", "33333", "USA" }, secondRecordValues);
        assertNull(reader.nextRecord());
    }
}
Also used : SimpleRecordSchema(org.apache.nifi.serialization.SimpleRecordSchema) Arrays(java.util.Arrays) DataType(org.apache.nifi.serialization.record.DataType) HashMap(java.util.HashMap) ComponentLog(org.apache.nifi.logging.ComponentLog) ArrayList(java.util.ArrayList) ChoiceDataType(org.apache.nifi.serialization.record.type.ChoiceDataType) RecordSchema(org.apache.nifi.serialization.record.RecordSchema) ByteArrayInputStream(java.io.ByteArrayInputStream) Map(java.util.Map) Record(org.apache.nifi.serialization.record.Record) MalformedRecordException(org.apache.nifi.serialization.MalformedRecordException) Files(java.nio.file.Files) RecordField(org.apache.nifi.serialization.record.RecordField) Assert.assertTrue(org.junit.Assert.assertTrue) IOException(java.io.IOException) SimpleRecordSchema(org.apache.nifi.serialization.SimpleRecordSchema) Test(org.junit.Test) FileInputStream(java.io.FileInputStream) Collectors(java.util.stream.Collectors) File(java.io.File) TimeUnit(java.util.concurrent.TimeUnit) Mockito(org.mockito.Mockito) List(java.util.List) Assert.assertNull(org.junit.Assert.assertNull) Ignore(org.junit.Ignore) Assert(org.junit.Assert) Collections(java.util.Collections) RecordFieldType(org.apache.nifi.serialization.record.RecordFieldType) Assert.assertEquals(org.junit.Assert.assertEquals) InputStream(java.io.InputStream) ByteArrayInputStream(java.io.ByteArrayInputStream) FileInputStream(java.io.FileInputStream) InputStream(java.io.InputStream) ComponentLog(org.apache.nifi.logging.ComponentLog) FileInputStream(java.io.FileInputStream) RecordSchema(org.apache.nifi.serialization.record.RecordSchema) SimpleRecordSchema(org.apache.nifi.serialization.SimpleRecordSchema) File(java.io.File) RecordFieldType(org.apache.nifi.serialization.record.RecordFieldType) Test(org.junit.Test)

Example 44 with SimpleRecordSchema

use of org.apache.nifi.serialization.SimpleRecordSchema in project nifi by apache.

the class TestJsonTreeRowRecordReader method testElementWithNestedArray.

@Test
public void testElementWithNestedArray() throws IOException, MalformedRecordException {
    final DataType accountRecordType = RecordFieldType.RECORD.getRecordDataType(getAccountSchema());
    final DataType accountsType = RecordFieldType.ARRAY.getArrayDataType(accountRecordType);
    final List<RecordField> fields = getDefaultFields();
    fields.add(new RecordField("accounts", accountsType));
    fields.remove(new RecordField("balance", RecordFieldType.DOUBLE.getDataType()));
    final RecordSchema schema = new SimpleRecordSchema(fields);
    try (final InputStream in = new FileInputStream(new File("src/test/resources/json/single-element-nested-array.json"));
        final JsonTreeRowRecordReader reader = new JsonTreeRowRecordReader(in, Mockito.mock(ComponentLog.class), schema, dateFormat, timeFormat, timestampFormat)) {
        final List<String> fieldNames = schema.getFieldNames();
        final List<String> expectedFieldNames = Arrays.asList(new String[] { "id", "name", "address", "city", "state", "zipCode", "country", "accounts" });
        assertEquals(expectedFieldNames, fieldNames);
        final List<RecordFieldType> dataTypes = schema.getDataTypes().stream().map(dt -> dt.getFieldType()).collect(Collectors.toList());
        final List<RecordFieldType> expectedTypes = Arrays.asList(new RecordFieldType[] { RecordFieldType.INT, RecordFieldType.STRING, RecordFieldType.STRING, RecordFieldType.STRING, RecordFieldType.STRING, RecordFieldType.STRING, RecordFieldType.STRING, RecordFieldType.ARRAY });
        assertEquals(expectedTypes, dataTypes);
        final Object[] firstRecordValues = reader.nextRecord().getValues();
        final Object[] nonArrayValues = Arrays.copyOfRange(firstRecordValues, 0, firstRecordValues.length - 1);
        Assert.assertArrayEquals(new Object[] { 1, "John Doe", "123 My Street", "My City", "MS", "11111", "USA" }, nonArrayValues);
        final Object lastRecord = firstRecordValues[firstRecordValues.length - 1];
        assertTrue(Object[].class.isAssignableFrom(lastRecord.getClass()));
        assertNull(reader.nextRecord());
    }
}
Also used : SimpleRecordSchema(org.apache.nifi.serialization.SimpleRecordSchema) Arrays(java.util.Arrays) DataType(org.apache.nifi.serialization.record.DataType) HashMap(java.util.HashMap) ComponentLog(org.apache.nifi.logging.ComponentLog) ArrayList(java.util.ArrayList) ChoiceDataType(org.apache.nifi.serialization.record.type.ChoiceDataType) RecordSchema(org.apache.nifi.serialization.record.RecordSchema) ByteArrayInputStream(java.io.ByteArrayInputStream) Map(java.util.Map) Record(org.apache.nifi.serialization.record.Record) MalformedRecordException(org.apache.nifi.serialization.MalformedRecordException) Files(java.nio.file.Files) RecordField(org.apache.nifi.serialization.record.RecordField) Assert.assertTrue(org.junit.Assert.assertTrue) IOException(java.io.IOException) SimpleRecordSchema(org.apache.nifi.serialization.SimpleRecordSchema) Test(org.junit.Test) FileInputStream(java.io.FileInputStream) Collectors(java.util.stream.Collectors) File(java.io.File) TimeUnit(java.util.concurrent.TimeUnit) Mockito(org.mockito.Mockito) List(java.util.List) Assert.assertNull(org.junit.Assert.assertNull) Ignore(org.junit.Ignore) Assert(org.junit.Assert) Collections(java.util.Collections) RecordFieldType(org.apache.nifi.serialization.record.RecordFieldType) Assert.assertEquals(org.junit.Assert.assertEquals) InputStream(java.io.InputStream) RecordField(org.apache.nifi.serialization.record.RecordField) ByteArrayInputStream(java.io.ByteArrayInputStream) FileInputStream(java.io.FileInputStream) InputStream(java.io.InputStream) ComponentLog(org.apache.nifi.logging.ComponentLog) FileInputStream(java.io.FileInputStream) DataType(org.apache.nifi.serialization.record.DataType) ChoiceDataType(org.apache.nifi.serialization.record.type.ChoiceDataType) RecordSchema(org.apache.nifi.serialization.record.RecordSchema) SimpleRecordSchema(org.apache.nifi.serialization.SimpleRecordSchema) File(java.io.File) RecordFieldType(org.apache.nifi.serialization.record.RecordFieldType) Test(org.junit.Test)

Example 45 with SimpleRecordSchema

use of org.apache.nifi.serialization.SimpleRecordSchema in project nifi by apache.

the class TestJsonTreeRowRecordReader method testReadArrayDifferentSchemas.

@Test
public void testReadArrayDifferentSchemas() throws IOException, MalformedRecordException {
    final RecordSchema schema = new SimpleRecordSchema(getDefaultFields());
    try (final InputStream in = new FileInputStream(new File("src/test/resources/json/bank-account-array-different-schemas.json"));
        final JsonTreeRowRecordReader reader = new JsonTreeRowRecordReader(in, Mockito.mock(ComponentLog.class), schema, dateFormat, timeFormat, timestampFormat)) {
        final List<String> fieldNames = schema.getFieldNames();
        final List<String> expectedFieldNames = Arrays.asList(new String[] { "id", "name", "balance", "address", "city", "state", "zipCode", "country" });
        assertEquals(expectedFieldNames, fieldNames);
        final List<RecordFieldType> dataTypes = schema.getDataTypes().stream().map(dt -> dt.getFieldType()).collect(Collectors.toList());
        final List<RecordFieldType> expectedTypes = Arrays.asList(new RecordFieldType[] { RecordFieldType.INT, RecordFieldType.STRING, RecordFieldType.DOUBLE, RecordFieldType.STRING, RecordFieldType.STRING, RecordFieldType.STRING, RecordFieldType.STRING, RecordFieldType.STRING });
        assertEquals(expectedTypes, dataTypes);
        final Object[] firstRecordValues = reader.nextRecord().getValues();
        Assert.assertArrayEquals(new Object[] { 1, "John Doe", 4750.89, "123 My Street", "My City", "MS", "11111", "USA" }, firstRecordValues);
        final Object[] secondRecordValues = reader.nextRecord().getValues();
        Assert.assertArrayEquals(new Object[] { 2, "Jane Doe", 4820.09, "321 Your Street", "Your City", "NY", "33333", null }, secondRecordValues);
        final Object[] thirdRecordValues = reader.nextRecord().getValues();
        Assert.assertArrayEquals(new Object[] { 3, "Jake Doe", 4751.89, "124 My Street", "My City", "MS", "11111", "USA" }, thirdRecordValues);
        assertNull(reader.nextRecord());
    }
}
Also used : SimpleRecordSchema(org.apache.nifi.serialization.SimpleRecordSchema) Arrays(java.util.Arrays) DataType(org.apache.nifi.serialization.record.DataType) HashMap(java.util.HashMap) ComponentLog(org.apache.nifi.logging.ComponentLog) ArrayList(java.util.ArrayList) ChoiceDataType(org.apache.nifi.serialization.record.type.ChoiceDataType) RecordSchema(org.apache.nifi.serialization.record.RecordSchema) ByteArrayInputStream(java.io.ByteArrayInputStream) Map(java.util.Map) Record(org.apache.nifi.serialization.record.Record) MalformedRecordException(org.apache.nifi.serialization.MalformedRecordException) Files(java.nio.file.Files) RecordField(org.apache.nifi.serialization.record.RecordField) Assert.assertTrue(org.junit.Assert.assertTrue) IOException(java.io.IOException) SimpleRecordSchema(org.apache.nifi.serialization.SimpleRecordSchema) Test(org.junit.Test) FileInputStream(java.io.FileInputStream) Collectors(java.util.stream.Collectors) File(java.io.File) TimeUnit(java.util.concurrent.TimeUnit) Mockito(org.mockito.Mockito) List(java.util.List) Assert.assertNull(org.junit.Assert.assertNull) Ignore(org.junit.Ignore) Assert(org.junit.Assert) Collections(java.util.Collections) RecordFieldType(org.apache.nifi.serialization.record.RecordFieldType) Assert.assertEquals(org.junit.Assert.assertEquals) InputStream(java.io.InputStream) ByteArrayInputStream(java.io.ByteArrayInputStream) FileInputStream(java.io.FileInputStream) InputStream(java.io.InputStream) ComponentLog(org.apache.nifi.logging.ComponentLog) FileInputStream(java.io.FileInputStream) RecordSchema(org.apache.nifi.serialization.record.RecordSchema) SimpleRecordSchema(org.apache.nifi.serialization.SimpleRecordSchema) File(java.io.File) RecordFieldType(org.apache.nifi.serialization.record.RecordFieldType) Test(org.junit.Test)

Aggregations

SimpleRecordSchema (org.apache.nifi.serialization.SimpleRecordSchema)171 RecordSchema (org.apache.nifi.serialization.record.RecordSchema)152 Test (org.junit.Test)146 RecordField (org.apache.nifi.serialization.record.RecordField)127 Record (org.apache.nifi.serialization.record.Record)120 ArrayList (java.util.ArrayList)114 MapRecord (org.apache.nifi.serialization.record.MapRecord)86 HashMap (java.util.HashMap)82 InputStream (java.io.InputStream)58 FileInputStream (java.io.FileInputStream)53 ByteArrayInputStream (java.io.ByteArrayInputStream)52 ComponentLog (org.apache.nifi.logging.ComponentLog)50 LinkedHashMap (java.util.LinkedHashMap)33 File (java.io.File)30 DataType (org.apache.nifi.serialization.record.DataType)26 ByteArrayOutputStream (java.io.ByteArrayOutputStream)21 MalformedRecordException (org.apache.nifi.serialization.MalformedRecordException)19 RecordFieldType (org.apache.nifi.serialization.record.RecordFieldType)19 List (java.util.List)18 IOException (java.io.IOException)17