Search in sources :

Example 96 with IntegerField

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

the class DictionaryMatcherTest method testSingleWordQueryInStringFieldUsingKeyword.

/**
 * Scenario: verifies GetNextTuple of DictionaryMatcher and multiple word
 * queries in String Field using KEYWORDOPERATOR
 */
@Test
public void testSingleWordQueryInStringFieldUsingKeyword() throws Exception {
    ArrayList<String> names = new ArrayList<String>(Arrays.asList("john Lee", "bruce"));
    Dictionary dictionary = new Dictionary(names);
    // create a data tuple first
    List<Span> list1 = new ArrayList<Span>();
    Span span1 = new Span("lastName", 0, 8, "john Lee", "john Lee");
    Span span2 = new Span("firstName", 0, 5, "bruce", "bruce");
    list1.add(span1);
    list1.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] = RESULTS_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<Span>(list1) };
    Tuple tuple1 = new Tuple(new Schema(schemaAttributes), fields1);
    List<Tuple> expectedResults = new ArrayList<Tuple>();
    expectedResults.add(tuple1);
    List<String> attributeNames = Arrays.asList(TestConstants.FIRST_NAME, TestConstants.LAST_NAME, TestConstants.DESCRIPTION);
    List<Tuple> returnedResults = DictionaryMatcherTestHelper.getQueryResults(PEOPLE_TABLE, dictionary, attributeNames, KeywordMatchingType.CONJUNCTION_INDEXBASED);
    boolean contains = TestUtils.equals(expectedResults, returnedResults);
    Assert.assertTrue(contains);
}
Also used : Dictionary(edu.uci.ics.texera.dataflow.dictionarymatcher.Dictionary) 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) 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) Test(org.junit.Test)

Example 97 with IntegerField

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

the class DictionaryMatcherTest method testMultipleWordsQueryUsingScanChinese.

/**
 * Scenario: verifies ITuple returned by DictionaryMatcher and multiple
 * word queries using SCAN OPERATOR in Chinese
 */
@Test
public void testMultipleWordsQueryUsingScanChinese() throws Exception {
    ArrayList<String> names = new ArrayList<String>(Arrays.asList("洛克贝尔"));
    Dictionary dictionary = new Dictionary(names);
    // create a data tuple first
    List<Span> list = new ArrayList<Span>();
    Span span = new Span("lastName", 0, 4, "洛克贝尔", "洛克贝尔");
    list.add(span);
    Attribute[] schemaAttributes = new Attribute[TestConstantsChinese.ATTRIBUTES_PEOPLE.length + 1];
    for (int count = 0; count < schemaAttributes.length - 1; count++) {
        schemaAttributes[count] = TestConstantsChinese.ATTRIBUTES_PEOPLE[count];
    }
    schemaAttributes[schemaAttributes.length - 1] = RESULTS_ATTRIBUTE;
    IField[] fields1 = { new StringField("孔明"), new StringField("洛克贝尔"), new IntegerField(42), new DoubleField(5.99), new DateField(new SimpleDateFormat("MM-dd-yyyy").parse("01-13-1974")), new TextField("北京大学计算机学院"), new ListField<Span>(list) };
    Tuple tuple1 = new Tuple(new Schema(schemaAttributes), fields1);
    List<Tuple> expectedResults = new ArrayList<Tuple>();
    expectedResults.add(tuple1);
    List<String> attributeNames = Arrays.asList(TestConstantsChinese.FIRST_NAME, TestConstantsChinese.LAST_NAME, TestConstantsChinese.DESCRIPTION);
    List<Tuple> returnedResults = DictionaryMatcherTestHelper.getQueryResults(CHINESE_TABLE, dictionary, attributeNames, KeywordMatchingType.SUBSTRING_SCANBASED);
    boolean contains = TestUtils.equals(expectedResults, returnedResults);
    Assert.assertTrue(contains);
}
Also used : Dictionary(edu.uci.ics.texera.dataflow.dictionarymatcher.Dictionary) 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) 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) Test(org.junit.Test)

Example 98 with IntegerField

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

the class PieChartSink method buildOtherDataField.

private IField buildOtherDataField(double value) {
    Attribute dataColumn = inputOperator.getOutputSchema().getAttribute(predicate.getDataColumn());
    AttributeType dataColumnType = dataColumn.getType();
    if (dataColumnType.equals(AttributeType.INTEGER)) {
        return new IntegerField((int) value);
    }
    return new DoubleField(value);
}
Also used : Attribute(edu.uci.ics.texera.api.schema.Attribute) AttributeType(edu.uci.ics.texera.api.schema.AttributeType) IntegerField(edu.uci.ics.texera.api.field.IntegerField) DoubleField(edu.uci.ics.texera.api.field.DoubleField)

Aggregations

IntegerField (edu.uci.ics.texera.api.field.IntegerField)98 Tuple (edu.uci.ics.texera.api.tuple.Tuple)90 IField (edu.uci.ics.texera.api.field.IField)88 StringField (edu.uci.ics.texera.api.field.StringField)87 TextField (edu.uci.ics.texera.api.field.TextField)79 ArrayList (java.util.ArrayList)75 Schema (edu.uci.ics.texera.api.schema.Schema)72 Test (org.junit.Test)70 Span (edu.uci.ics.texera.api.span.Span)66 DoubleField (edu.uci.ics.texera.api.field.DoubleField)65 DateField (edu.uci.ics.texera.api.field.DateField)59 SimpleDateFormat (java.text.SimpleDateFormat)57 Attribute (edu.uci.ics.texera.api.schema.Attribute)56 Dictionary (edu.uci.ics.texera.dataflow.dictionarymatcher.Dictionary)29 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 ObjectMapper (com.fasterxml.jackson.databind.ObjectMapper)4 IOperator (edu.uci.ics.texera.api.dataflow.IOperator)4 DateTimeField (edu.uci.ics.texera.api.field.DateTimeField)3