Search in sources :

Example 86 with Schema

use of edu.uci.ics.texera.api.schema.Schema in project textdb by TextDB.

the class DictionaryMatcherTest method testMultipleWordQueryInTextFieldUsingScan1.

/**
 * Testcases for scan-based DictionaryMatcher, use getScanSourceResults method only.
 * @throws Exception
 */
@Test
public void testMultipleWordQueryInTextFieldUsingScan1() throws Exception {
    ArrayList<String> names = new ArrayList<String>(Arrays.asList("tall", "fair"));
    Dictionary dictionary = new Dictionary(names);
    // create a data tuple first
    List<Span> list = new ArrayList<Span>();
    List<Span> list1 = new ArrayList<Span>();
    Span span = new Span("description", 0, 4, "tall", "Tall");
    Span span1 = new Span("description", 5, 9, "fair", "Fair");
    list.add(span);
    list1.add(span);
    list1.add(span1);
    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>(list) };
    IField[] fields2 = { new StringField("christian john wayne"), new StringField("rock bale"), new IntegerField(42), new DoubleField(5.99), new DateField(new SimpleDateFormat("MM-dd-yyyy").parse("01-13-1974")), new TextField("Tall Fair"), new ListField<Span>(list1) };
    Tuple tuple1 = new Tuple(new Schema(schemaAttributes), fields1);
    Tuple tuple2 = new Tuple(new Schema(schemaAttributes), fields2);
    List<Tuple> expectedResults = new ArrayList<Tuple>();
    expectedResults.add(tuple1);
    expectedResults.add(tuple2);
    List<String> attributeNames = Arrays.asList(TestConstants.FIRST_NAME, TestConstants.LAST_NAME, TestConstants.DESCRIPTION);
    List<Tuple> returnedResults = DictionaryMatcherTestHelper.getQueryResults(PEOPLE_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 87 with Schema

use of edu.uci.ics.texera.api.schema.Schema 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 88 with Schema

use of edu.uci.ics.texera.api.schema.Schema in project textdb by TextDB.

the class DictionaryMatcherTest method testSingleWordQueryInStringFieldUsingScanChinese.

/**
 * Scenario: verifies GetNextTuple of DictionaryMatcher and single word
 * queries in String Field using SUBSTRING_SCANBASED
 * Test in Chinese.
 */
@Test
public void testSingleWordQueryInStringFieldUsingScanChinese() throws Exception {
    ArrayList<String> names = new ArrayList<String>(Arrays.asList("孔明"));
    Dictionary dictionary = new Dictionary(names);
    // create a data tuple first
    List<Span> list = new ArrayList<Span>();
    Span span = new Span("firstName", 0, 2, "孔明", "孔明");
    list.add(span);
    Schema resultSchema = new Schema.Builder().add(TestConstantsChinese.SCHEMA_PEOPLE).add(RESULTS_ATTRIBUTE).build();
    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(resultSchema, fields1);
    List<Tuple> expectedResults = new ArrayList<Tuple>();
    expectedResults.add(tuple1);
    List<String> attributeNames = Arrays.asList(TestConstants.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) 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 89 with Schema

use of edu.uci.ics.texera.api.schema.Schema in project textdb by TextDB.

the class DictionaryMatcherTest method testWordInMultipleFieldsQueryUsingPhrase.

/**
 * Scenario S-13:verifies: data source has multiple attributes, and an
 * entity can appear in all the fields and multiple times using PHRASE
 * OPERATOR.
 */
@Test
public void testWordInMultipleFieldsQueryUsingPhrase() throws Exception {
    ArrayList<String> names = new ArrayList<String>(Arrays.asList("lin clooney"));
    Dictionary dictionary = new Dictionary(names);
    // create a data tuple first
    List<Span> list = new ArrayList<Span>();
    Span span1 = new Span("lastName", 0, 11, "lin clooney", "lin clooney");
    Span span2 = new Span("description", 0, 11, "lin clooney", "Lin Clooney");
    Span span3 = new Span("description", 25, 36, "lin clooney", "lin clooney");
    list.add(span1);
    list.add(span2);
    list.add(span3);
    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("george lin lin"), new StringField("lin clooney"), new IntegerField(43), new DoubleField(6.06), new DateField(new SimpleDateFormat("MM-dd-yyyy").parse("01-13-1973")), new TextField("Lin Clooney is Short and lin clooney is Angry"), 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(TestConstants.FIRST_NAME, TestConstants.LAST_NAME, TestConstants.DESCRIPTION);
    List<Tuple> returnedResults = DictionaryMatcherTestHelper.getQueryResults(PEOPLE_TABLE, dictionary, attributeNames, KeywordMatchingType.PHRASE_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 90 with Schema

use of edu.uci.ics.texera.api.schema.Schema in project textdb by TextDB.

the class DictionaryMatcherTest method testWordInMultipleFieldsQueryUsingKeyword.

/**
 * Scenario S-12:verifies: data source has multiple attributes, and an
 * entity can appear in all the fields and multiple times using KEYWORD
 * OPERATOR.
 */
@Test
public void testWordInMultipleFieldsQueryUsingKeyword() throws Exception {
    ArrayList<String> names = new ArrayList<String>(Arrays.asList("lin clooney"));
    Dictionary dictionary = new Dictionary(names);
    // create a data tuple first
    List<Span> list = new ArrayList<Span>();
    Span span1 = new Span("lastName", 0, 11, "lin clooney", "lin clooney");
    Span span2 = new Span("description", 0, 3, "lin", "Lin", 0);
    Span span3 = new Span("description", 4, 11, "clooney", "Clooney", 1);
    Span span4 = new Span("description", 25, 28, "lin", "lin", 5);
    Span span5 = new Span("description", 29, 36, "clooney", "clooney", 6);
    list.add(span1);
    list.add(span2);
    list.add(span3);
    list.add(span4);
    list.add(span5);
    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("george lin lin"), new StringField("lin clooney"), new IntegerField(43), new DoubleField(6.06), new DateField(new SimpleDateFormat("MM-dd-yyyy").parse("01-13-1973")), new TextField("Lin Clooney is Short and lin clooney is Angry"), 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(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)

Aggregations

Schema (edu.uci.ics.texera.api.schema.Schema)134 Test (org.junit.Test)109 Tuple (edu.uci.ics.texera.api.tuple.Tuple)106 ArrayList (java.util.ArrayList)97 IField (edu.uci.ics.texera.api.field.IField)96 Span (edu.uci.ics.texera.api.span.Span)86 TextField (edu.uci.ics.texera.api.field.TextField)77 Attribute (edu.uci.ics.texera.api.schema.Attribute)76 StringField (edu.uci.ics.texera.api.field.StringField)72 IntegerField (edu.uci.ics.texera.api.field.IntegerField)71 DoubleField (edu.uci.ics.texera.api.field.DoubleField)60 DateField (edu.uci.ics.texera.api.field.DateField)57 SimpleDateFormat (java.text.SimpleDateFormat)54 Dictionary (edu.uci.ics.texera.dataflow.dictionarymatcher.Dictionary)29 ListField (edu.uci.ics.texera.api.field.ListField)21 IOperator (edu.uci.ics.texera.api.dataflow.IOperator)15 DataflowException (edu.uci.ics.texera.api.exception.DataflowException)14 AttributeType (edu.uci.ics.texera.api.schema.AttributeType)13 TexeraException (edu.uci.ics.texera.api.exception.TexeraException)9 JoinDistancePredicate (edu.uci.ics.texera.dataflow.join.JoinDistancePredicate)9