Search in sources :

Example 51 with TextField

use of edu.uci.ics.texera.api.field.TextField in project textdb by TextDB.

the class SpanTupleTest method testGetters.

@Test
public void testGetters() throws ParseException {
    // create data tuple first
    Attribute[] attributes = new Attribute[TestConstants.ATTRIBUTES_PEOPLE.length + 1];
    for (int count = 0; count < attributes.length - 1; count++) {
        attributes[count] = TestConstants.ATTRIBUTES_PEOPLE[count];
    }
    attributes[attributes.length - 1] = SchemaConstants.SPAN_LIST_ATTRIBUTE;
    List<IField> fields = new ArrayList<IField>(Arrays.asList(new IField[] { new StringField("bruce"), new StringField("lee"), new IntegerField(46), new DoubleField(5.50), new DateField(new SimpleDateFormat("MM-dd-yyyy").parse("01-14-1970")), new TextField("bruce was born in new york city and was grown up in los angeles") }));
    IField spanField = createSpanListField();
    fields.add(spanField);
    spanTuple = new Tuple(new Schema(attributes), fields.toArray(new IField[fields.size()]));
    IField spanFieldRetrieved = spanTuple.getField(SchemaConstants.SPAN_LIST);
    Assert.assertTrue(spanFieldRetrieved instanceof ListField);
    Assert.assertSame(spanField, spanFieldRetrieved);
}
Also used : Attribute(edu.uci.ics.texera.api.schema.Attribute) Schema(edu.uci.ics.texera.api.schema.Schema) ArrayList(java.util.ArrayList) IntegerField(edu.uci.ics.texera.api.field.IntegerField) ListField(edu.uci.ics.texera.api.field.ListField) IField(edu.uci.ics.texera.api.field.IField) StringField(edu.uci.ics.texera.api.field.StringField) TextField(edu.uci.ics.texera.api.field.TextField) DateField(edu.uci.ics.texera.api.field.DateField) SimpleDateFormat(java.text.SimpleDateFormat) DoubleField(edu.uci.ics.texera.api.field.DoubleField) Tuple(edu.uci.ics.texera.api.tuple.Tuple) Test(org.junit.Test)

Example 52 with TextField

use of edu.uci.ics.texera.api.field.TextField in project textdb by TextDB.

the class ExcelSinkTest method attributeTypeTest.

@Test
public // writing 10000 tuples
void attributeTypeTest() throws Exception {
    ArrayList<String> attributeNames = new ArrayList<>();
    attributeNames.add(TestConstants.FIRST_NAME);
    attributeNames.add(TestConstants.LAST_NAME);
    attributeNames.add(TestConstants.AGE);
    attributeNames.add(TestConstants.HEIGHT);
    attributeNames.add(TestConstants.DATE_OF_BIRTH);
    attributeNames.add(TestConstants.DESCRIPTION);
    // Prepare Schema
    Attribute[] schemaAttributes = new Attribute[TestConstants.ATTRIBUTES_PEOPLE.length];
    for (int count = 0; count < schemaAttributes.length; count++) {
        schemaAttributes[count] = TestConstants.ATTRIBUTES_PEOPLE[count];
    }
    // Prepare 10000 tuples as a tupleList
    int testSize = 10000;
    Random rand = new Random();
    List<Tuple> tupleList = new ArrayList<Tuple>();
    for (int i = 0; i < testSize; i++) {
        IField[] fields = { new StringField(getRandomString()), new StringField(getRandomString()), new IntegerField(rand.nextInt()), new DoubleField(rand.nextDouble() * rand.nextInt()), new DateField(getRandomDate()), new TextField(getRandomString()) };
        tupleList.add(new Tuple(new Schema(schemaAttributes), fields));
    }
    assert (tupleList.size() == testSize);
    IOperator inputOperator = Mockito.mock(IOperator.class);
    Mockito.when(inputOperator.getOutputSchema()).thenReturn(new Schema(schemaAttributes));
    OngoingStubbing<Tuple> stubbing = Mockito.when(inputOperator.getNextTuple());
    for (Tuple t : tupleList) {
        stubbing = stubbing.thenReturn(t);
    }
    stubbing = stubbing.thenReturn(null);
    // excel writing test
    excelSink = new ExcelSink(new ExcelSinkPredicate());
    excelSink.setInputOperator(inputOperator);
    excelSink.open();
    excelSink.collectAllTuples();
    excelSink.close();
    Files.deleteIfExists(excelSink.getFilePath());
}
Also used : Attribute(edu.uci.ics.texera.api.schema.Attribute) IOperator(edu.uci.ics.texera.api.dataflow.IOperator) Schema(edu.uci.ics.texera.api.schema.Schema) ArrayList(java.util.ArrayList) IntegerField(edu.uci.ics.texera.api.field.IntegerField) IField(edu.uci.ics.texera.api.field.IField) Random(java.util.Random) StringField(edu.uci.ics.texera.api.field.StringField) TextField(edu.uci.ics.texera.api.field.TextField) DateField(edu.uci.ics.texera.api.field.DateField) Tuple(edu.uci.ics.texera.api.tuple.Tuple) DoubleField(edu.uci.ics.texera.api.field.DoubleField) Test(org.junit.Test)

Example 53 with TextField

use of edu.uci.ics.texera.api.field.TextField in project textdb by TextDB.

the class MysqlSinkTest method test2TupleInsertion.

/**
 * Create two tuples, insert into mysql
 * @throws ParseException
 */
public void test2TupleInsertion() throws Exception {
    ArrayList<String> attributeNames = new ArrayList<>();
    attributeNames.add(TestConstants.FIRST_NAME);
    attributeNames.add(TestConstants.LAST_NAME);
    attributeNames.add(TestConstants.DESCRIPTION);
    // Prepare the expected result list
    List<Span> list = new ArrayList<>();
    Span span1 = new Span("firstName", 0, 5, "bruce", "bruce");
    Span span2 = new Span("lastnName", 0, 5, "jacki", "jacki");
    list.add(span1);
    list.add(span2);
    Attribute[] schemaAttributes = new Attribute[TestConstants.ATTRIBUTES_PEOPLE.length + 1];
    for (int count = 0; count < schemaAttributes.length - 1; count++) {
        schemaAttributes[count] = TestConstants.ATTRIBUTES_PEOPLE[count];
    }
    schemaAttributes[schemaAttributes.length - 1] = SchemaConstants.SPAN_LIST_ATTRIBUTE;
    IField[] fields1 = { new StringField("bruce"), new StringField("john Lee"), new IntegerField(46), new DoubleField(5.50), new DateField(new SimpleDateFormat("MM-dd-yyyy").parse("01-14-1970")), new TextField("Tall Angry"), new ListField<>(list) };
    IField[] fields2 = { new StringField("test"), new StringField("jackie chan"), new IntegerField(0), new DoubleField(6.0), new DateField(new SimpleDateFormat("MM-dd-yyyy").parse("09-18-1994")), new TextField("Angry Bird"), new ListField<>(list) };
    Tuple tuple1 = new Tuple(new Schema(schemaAttributes), fields1);
    Tuple tuple2 = new Tuple(new Schema(schemaAttributes), fields2);
    IOperator localInputOperator = Mockito.mock(IOperator.class);
    Mockito.when(localInputOperator.getOutputSchema()).thenReturn(new Schema(schemaAttributes)).thenReturn(null);
    Mockito.when(localInputOperator.getNextTuple()).thenReturn(tuple1).thenReturn(tuple2).thenReturn(null);
    mysqlSink.setInputOperator(localInputOperator);
    mysqlSink.open();
    mysqlSink.processTuples();
    mysqlSink.close();
}
Also used : Attribute(edu.uci.ics.texera.api.schema.Attribute) IOperator(edu.uci.ics.texera.api.dataflow.IOperator) Schema(edu.uci.ics.texera.api.schema.Schema) ArrayList(java.util.ArrayList) IntegerField(edu.uci.ics.texera.api.field.IntegerField) IField(edu.uci.ics.texera.api.field.IField) Span(edu.uci.ics.texera.api.span.Span) StringField(edu.uci.ics.texera.api.field.StringField) TextField(edu.uci.ics.texera.api.field.TextField) DateField(edu.uci.ics.texera.api.field.DateField) SimpleDateFormat(java.text.SimpleDateFormat) DoubleField(edu.uci.ics.texera.api.field.DoubleField) Tuple(edu.uci.ics.texera.api.tuple.Tuple)

Example 54 with TextField

use of edu.uci.ics.texera.api.field.TextField in project textdb by TextDB.

the class FileSourceOperatorTest method test4.

/*
     * Test FileSourceOperator with a Directory with recursive = true and maxDepth = 2.
     * 
     * The files under the recursive sub-directories with recursive depth 2 will be read.
     *     
     * expected results: test1.txt, test2.txt and test4.txt will be included
     */
@Test
public void test4() throws Exception {
    String attrName = "content";
    Schema schema = new Schema(new Attribute(attrName, AttributeType.TEXT));
    FileSourcePredicate predicate = new FileSourcePredicate(tempFolderPath.toString(), attrName, true, 2);
    FileSourceOperator fileSource = new FileSourceOperator(predicate);
    Tuple tuple;
    ArrayList<Tuple> exactResults = new ArrayList<>();
    fileSource.open();
    while ((tuple = fileSource.getNextTuple()) != null) {
        exactResults.add(tuple);
    }
    fileSource.close();
    List<Tuple> expectedResults = Arrays.asList(new Tuple(schema, new TextField(tempFile1String)), new Tuple(schema, new TextField(tempFile2String)), new Tuple(schema, new TextField(tempFile4String)));
    Assert.assertTrue(TestUtils.equals(expectedResults, exactResults));
}
Also used : Attribute(edu.uci.ics.texera.api.schema.Attribute) Schema(edu.uci.ics.texera.api.schema.Schema) ArrayList(java.util.ArrayList) TextField(edu.uci.ics.texera.api.field.TextField) Tuple(edu.uci.ics.texera.api.tuple.Tuple) Test(org.junit.Test)

Example 55 with TextField

use of edu.uci.ics.texera.api.field.TextField in project textdb by TextDB.

the class FileSourceOperatorTest method test1.

/*
     * Test FileSourceOperator with a single "txt" file.
     * Optional parameters are all set to default.
     */
@Test
public void test1() throws Exception {
    String attrName = "content";
    Schema schema = new Schema(new Attribute(attrName, AttributeType.TEXT));
    FileSourcePredicate predicate = new FileSourcePredicate(tempFile1Path.toString(), attrName);
    FileSourceOperator fileSource = new FileSourceOperator(predicate);
    Tuple tuple;
    ArrayList<Tuple> exactResults = new ArrayList<>();
    fileSource.open();
    while ((tuple = fileSource.getNextTuple()) != null) {
        exactResults.add(tuple);
    }
    fileSource.close();
    List<Tuple> expectedResults = Arrays.asList(new Tuple(schema, new TextField(tempFile1String)));
    Assert.assertTrue(TestUtils.equals(expectedResults, exactResults));
}
Also used : Attribute(edu.uci.ics.texera.api.schema.Attribute) Schema(edu.uci.ics.texera.api.schema.Schema) ArrayList(java.util.ArrayList) TextField(edu.uci.ics.texera.api.field.TextField) Tuple(edu.uci.ics.texera.api.tuple.Tuple) Test(org.junit.Test)

Aggregations

TextField (edu.uci.ics.texera.api.field.TextField)115 IField (edu.uci.ics.texera.api.field.IField)99 Tuple (edu.uci.ics.texera.api.tuple.Tuple)89 ArrayList (java.util.ArrayList)84 IntegerField (edu.uci.ics.texera.api.field.IntegerField)78 StringField (edu.uci.ics.texera.api.field.StringField)78 Span (edu.uci.ics.texera.api.span.Span)78 Schema (edu.uci.ics.texera.api.schema.Schema)77 Test (org.junit.Test)76 DoubleField (edu.uci.ics.texera.api.field.DoubleField)63 DateField (edu.uci.ics.texera.api.field.DateField)58 Attribute (edu.uci.ics.texera.api.schema.Attribute)56 SimpleDateFormat (java.text.SimpleDateFormat)56 Dictionary (edu.uci.ics.texera.dataflow.dictionarymatcher.Dictionary)29 ListField (edu.uci.ics.texera.api.field.ListField)11 JoinDistancePredicate (edu.uci.ics.texera.dataflow.join.JoinDistancePredicate)9 KeywordMatcherSourceOperator (edu.uci.ics.texera.dataflow.keywordmatcher.KeywordMatcherSourceOperator)9 JsonNode (com.fasterxml.jackson.databind.JsonNode)5 IOperator (edu.uci.ics.texera.api.dataflow.IOperator)5 ObjectMapper (com.fasterxml.jackson.databind.ObjectMapper)4