Search in sources :

Example 46 with StreamEvent

use of co.cask.cdap.api.flow.flowlet.StreamEvent in project cdap by caskdata.

the class DelimitedStringsRecordFormatTest method testCSV.

@Test
public void testCSV() throws Exception {
    FormatSpecification spec = new FormatSpecification(Formats.CSV, null, Collections.<String, String>emptyMap());
    RecordFormat<StreamEvent, StructuredRecord> format = RecordFormats.createInitializedFormat(spec);
    String body = "userX,actionY,itemZ";
    StructuredRecord output = format.read(new StreamEvent(ByteBuffer.wrap(Bytes.toBytes(body))));
    String[] actual = output.get("body");
    String[] expected = body.split(",");
    Assert.assertArrayEquals(expected, actual);
}
Also used : StreamEvent(co.cask.cdap.api.flow.flowlet.StreamEvent) FormatSpecification(co.cask.cdap.api.data.format.FormatSpecification) StructuredRecord(co.cask.cdap.api.data.format.StructuredRecord) Test(org.junit.Test)

Example 47 with StreamEvent

use of co.cask.cdap.api.flow.flowlet.StreamEvent in project cdap by caskdata.

the class DelimitedStringsRecordFormatTest method testTSV.

@Test
public void testTSV() throws Exception {
    FormatSpecification spec = new FormatSpecification(Formats.TSV, null, Collections.<String, String>emptyMap());
    RecordFormat<StreamEvent, StructuredRecord> format = RecordFormats.createInitializedFormat(spec);
    String body = "userX\tactionY\titemZ";
    StructuredRecord output = format.read(new StreamEvent(ByteBuffer.wrap(Bytes.toBytes(body))));
    String[] actual = output.get("body");
    String[] expected = body.split("\t");
    Assert.assertArrayEquals(expected, actual);
}
Also used : StreamEvent(co.cask.cdap.api.flow.flowlet.StreamEvent) FormatSpecification(co.cask.cdap.api.data.format.FormatSpecification) StructuredRecord(co.cask.cdap.api.data.format.StructuredRecord) Test(org.junit.Test)

Example 48 with StreamEvent

use of co.cask.cdap.api.flow.flowlet.StreamEvent in project cdap by caskdata.

the class DelimitedStringsRecordFormatTest method testFormatRecordWithSchema.

@Test
public void testFormatRecordWithSchema() throws UnsupportedTypeException, UnexpectedFormatException {
    Schema schema = Schema.recordOf("event", Schema.Field.of("f1", Schema.unionOf(Schema.of(Schema.Type.BOOLEAN), Schema.of(Schema.Type.NULL))), Schema.Field.of("f2", Schema.unionOf(Schema.of(Schema.Type.INT), Schema.of(Schema.Type.NULL))), Schema.Field.of("f3", Schema.unionOf(Schema.of(Schema.Type.FLOAT), Schema.of(Schema.Type.NULL))), Schema.Field.of("f4", Schema.unionOf(Schema.of(Schema.Type.DOUBLE), Schema.of(Schema.Type.NULL))), Schema.Field.of("f5", Schema.unionOf(Schema.of(Schema.Type.BYTES), Schema.of(Schema.Type.NULL))), Schema.Field.of("f6", Schema.unionOf(Schema.of(Schema.Type.STRING), Schema.of(Schema.Type.NULL))), Schema.Field.of("f7", Schema.arrayOf(Schema.of(Schema.Type.STRING))));
    DelimitedStringsRecordFormat format = new DelimitedStringsRecordFormat();
    FormatSpecification spec = new FormatSpecification(DelimitedStringsRecordFormat.class.getCanonicalName(), schema, ImmutableMap.of(DelimitedStringsRecordFormat.DELIMITER, ","));
    format.initialize(spec);
    boolean booleanVal = false;
    int intVal = Integer.MAX_VALUE;
    float floatVal = Float.MAX_VALUE;
    double doubleVal = Double.MAX_VALUE;
    byte[] bytesVal = new byte[] { 0, 1, 2 };
    String stringVal = "foo bar";
    String[] arrayVal = new String[] { "extra1", "extra2", "extra3" };
    String body = new StringBuilder().append(booleanVal).append(",").append(intVal).append(",").append(floatVal).append(",").append(doubleVal).append(",").append(Bytes.toStringBinary(bytesVal)).append(",").append(stringVal).append(",").append(arrayVal[0]).append(",").append(arrayVal[1]).append(",").append(arrayVal[2]).toString();
    StructuredRecord output = format.read(new StreamEvent(ByteBuffer.wrap(Bytes.toBytes(body))));
    Assert.assertEquals(booleanVal, output.get("f1"));
    Assert.assertEquals(intVal, (int) output.get("f2"));
    Assert.assertEquals(floatVal, output.get("f3"), 0.0001f);
    Assert.assertEquals(doubleVal, output.get("f4"), 0.0001d);
    Assert.assertArrayEquals(bytesVal, (byte[]) output.get("f5"));
    Assert.assertEquals(stringVal, output.get("f6"));
    Assert.assertArrayEquals(arrayVal, (String[]) output.get("f7"));
    // now try with null fields.
    output = format.read(new StreamEvent(ByteBuffer.wrap(Bytes.toBytes("true,,3.14159,,,hello world,extra1"))));
    Assert.assertTrue((Boolean) output.get("f1"));
    Assert.assertNull(output.get("f2"));
    Assert.assertEquals(3.14159f, output.get("f3"), 0.0001f);
    Assert.assertNull(output.get("f4"));
    Assert.assertNull(output.get("f5"));
    Assert.assertEquals("hello world", output.get("f6"));
    Assert.assertArrayEquals(new String[] { "extra1" }, (String[]) output.get("f7"));
}
Also used : Schema(co.cask.cdap.api.data.schema.Schema) StreamEvent(co.cask.cdap.api.flow.flowlet.StreamEvent) FormatSpecification(co.cask.cdap.api.data.format.FormatSpecification) StructuredRecord(co.cask.cdap.api.data.format.StructuredRecord) Test(org.junit.Test)

Example 49 with StreamEvent

use of co.cask.cdap.api.flow.flowlet.StreamEvent in project cdap by caskdata.

the class CombinedLogRecordFormatTest method testCLFLog.

@Test
public void testCLFLog() throws UnsupportedTypeException, UnexpectedFormatException {
    CombinedLogRecordFormat format = new CombinedLogRecordFormat();
    FormatSpecification spec = new FormatSpecification(CombinedLogRecordFormat.class.getCanonicalName(), null, ImmutableMap.of());
    format.initialize(spec);
    String data = "10.10.10.10 - - [01/Feb/2015:06:47:10 +0000] \"GET /browse/COOP-DBT-JOB1-238/artifact HTTP/1.1\"" + " 301 256 \"-\" \"Mozilla/5.0 (compatible; AhrefsBot/5.0; +http://ahrefs.com/robot/)\"";
    StructuredRecord output = format.read(new StreamEvent(ByteBuffer.wrap(Bytes.toBytes(data))));
    Assert.assertEquals("10.10.10.10", output.get("remote_host"));
    Assert.assertNull(output.get("remote_login"));
    Assert.assertNull(output.get("auth_user"));
    Assert.assertEquals("01/Feb/2015:06:47:10 +0000", output.get("request_time"));
    Assert.assertEquals("GET /browse/COOP-DBT-JOB1-238/artifact HTTP/1.1", output.get("request"));
    Assert.assertEquals(301, (int) output.get("status"));
    Assert.assertEquals(256, (int) output.get("content_length"));
    Assert.assertNull(output.get("referrer"));
    Assert.assertEquals("Mozilla/5.0 (compatible; AhrefsBot/5.0; +http://ahrefs.com/robot/)", output.get("user_agent"));
}
Also used : StreamEvent(co.cask.cdap.api.flow.flowlet.StreamEvent) FormatSpecification(co.cask.cdap.api.data.format.FormatSpecification) StructuredRecord(co.cask.cdap.api.data.format.StructuredRecord) Test(org.junit.Test)

Example 50 with StreamEvent

use of co.cask.cdap.api.flow.flowlet.StreamEvent in project cdap by caskdata.

the class CombinedLogRecordFormatTest method testInvalid.

@Test(expected = UnexpectedFormatException.class)
public void testInvalid() throws UnsupportedTypeException, UnexpectedFormatException {
    CombinedLogRecordFormat format = new CombinedLogRecordFormat();
    FormatSpecification spec = new FormatSpecification(CombinedLogRecordFormat.class.getCanonicalName(), null, ImmutableMap.of());
    format.initialize(spec);
    String data = "10.10.10.10[01/Feb/2015:06:47:10 +0000";
    StructuredRecord output = format.read(new StreamEvent(ByteBuffer.wrap(Bytes.toBytes(data))));
}
Also used : StreamEvent(co.cask.cdap.api.flow.flowlet.StreamEvent) FormatSpecification(co.cask.cdap.api.data.format.FormatSpecification) StructuredRecord(co.cask.cdap.api.data.format.StructuredRecord) Test(org.junit.Test)

Aggregations

StreamEvent (co.cask.cdap.api.flow.flowlet.StreamEvent)84 Test (org.junit.Test)65 Location (org.apache.twill.filesystem.Location)27 StreamId (co.cask.cdap.proto.id.StreamId)24 StructuredRecord (co.cask.cdap.api.data.format.StructuredRecord)19 FormatSpecification (co.cask.cdap.api.data.format.FormatSpecification)17 Schema (co.cask.cdap.api.data.schema.Schema)10 IOException (java.io.IOException)9 StreamConfig (co.cask.cdap.data2.transaction.stream.StreamConfig)8 ByteBuffer (java.nio.ByteBuffer)8 ConsumerConfig (co.cask.cdap.data2.queue.ConsumerConfig)7 StreamAdmin (co.cask.cdap.data2.transaction.stream.StreamAdmin)6 TransactionContext (org.apache.tephra.TransactionContext)6 BinaryDecoder (co.cask.cdap.common.io.BinaryDecoder)5 TypeToken (com.google.common.reflect.TypeToken)5 StreamEventCodec (co.cask.cdap.common.stream.StreamEventCodec)4 IdentityStreamEventDecoder (co.cask.cdap.data.stream.decoder.IdentityStreamEventDecoder)4 File (java.io.File)4 SchemaHash (co.cask.cdap.api.data.schema.SchemaHash)3 QueueName (co.cask.cdap.common.queue.QueueName)3