Search in sources :

Example 91 with Attribute

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

the class KeywordConjunctionTest method testWordInMultipleFieldsQuery.

/**
 * Verifies: data source has multiple attributes, and an entity can appear
 * in all the fields and multiple times.
 *
 * @throws Exception
 */
@Test
public void testWordInMultipleFieldsQuery() throws Exception {
    // Prepare the query
    String query = "lin clooney";
    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("lastName", 0, 11, "lin clooney", "lin clooney");
    Span span2 = new Span("description", 0, 3, "lin", "Lin", 0);
    Span span3 = new Span("description", 25, 28, "lin", "lin", 5);
    Span span4 = new Span("description", 4, 11, "clooney", "Clooney", 1);
    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] = new Attribute(RESULTS, AttributeType.LIST);
    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<>(list) };
    Tuple tuple1 = new Tuple(new Schema(schemaAttributes), fields1);
    List<Tuple> expectedResultList = new ArrayList<>();
    expectedResultList.add(tuple1);
    // Perform the query
    List<Tuple> resultList = KeywordTestHelper.getQueryResults(PEOPLE_TABLE, query, attributeNames, conjunction);
    // check the results
    boolean contains = TestUtils.equals(expectedResultList, resultList);
    Assert.assertTrue(contains);
}
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) 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 92 with Attribute

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

the class KeywordPhraseTest method testPhraseSearchForStringField.

/**
 * Verifies List<ITuple> returned by Phrase Matcher on multiple word query
 * on a String Field
 *
 * @throws Exception
 */
@Test
public void testPhraseSearchForStringField() throws Exception {
    // Prepare Query
    String query = "george lin lin";
    ArrayList<String> attributeNames = new ArrayList<>();
    attributeNames.add(TestConstants.FIRST_NAME);
    attributeNames.add(TestConstants.LAST_NAME);
    attributeNames.add(TestConstants.DESCRIPTION);
    // Prepare expected result list
    List<Span> list = new ArrayList<Span>();
    Span span1 = new Span("firstName", 0, 14, "george lin lin", "george lin lin");
    list.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] = new Attribute(RESULTS, AttributeType.LIST);
    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<>(list) };
    Tuple tuple1 = new Tuple(new Schema(schemaAttributes), fields1);
    List<Tuple> expectedResultList = new ArrayList<>();
    expectedResultList.add(tuple1);
    // Perform Query
    List<Tuple> resultList = KeywordTestHelper.getQueryResults(PEOPLE_TABLE, query, attributeNames, phrase);
    // Perform Check
    boolean contains = TestUtils.equals(expectedResultList, resultList);
    Assert.assertTrue(contains);
}
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) 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 93 with Attribute

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

the class KeywordPhraseTest method testWordInMultipleFieldsQueryWithStopWords4.

/**
 * Verifies: Query with Stop Words match corresponding phrases in the
 * document Used to cause exception if there is a special symbol with the
 * words, ex: big(
 *
 * @throws Exception
 *             with Medline data
 */
@Test
public void testWordInMultipleFieldsQueryWithStopWords4() throws Exception {
    // Prepare Query
    String query = "x-ray";
    ArrayList<String> attributeNames = new ArrayList<>();
    attributeNames.add(keywordTestConstants.ABSTRACT);
    // Prepare expected result list
    List<Span> list = new ArrayList<>();
    Span span1 = new Span(keywordTestConstants.ABSTRACT, 226, 231, "x-ray", "x-ray");
    list.add(span1);
    Attribute[] schemaAttributes = new Attribute[keywordTestConstants.ATTRIBUTES_MEDLINE.length + 1];
    for (int count = 0; count < schemaAttributes.length - 1; count++) {
        schemaAttributes[count] = keywordTestConstants.ATTRIBUTES_MEDLINE[count];
    }
    schemaAttributes[schemaAttributes.length - 1] = new Attribute(RESULTS, AttributeType.LIST);
    IField[] fields = { new IntegerField(17832788), new TextField(""), new TextField("Cosmic X-ray Sources."), new TextField("S Bowyer, E T Byram, T A Chubb, H Friedman"), new StringField("147-3656 Jan 22, 1965"), new TextField("Science (New York, N.Y.)"), new TextField(""), new TextField(""), new TextField("Eight new sources of cosmic x-rays were detected by two Aerobee surveys in 1964. One source, from Sagittarius, is close to the galactic center, and the other, " + "from Ophiuchus, may coincide with Kepler's 1604 supernova. All the x-ray sources are fairly close to the galactic plane."), new DoubleField(0.667832788), new ListField<>(list) };
    Tuple tuple1 = new Tuple(new Schema(schemaAttributes), fields);
    List<Tuple> expectedResultList = new ArrayList<>();
    expectedResultList.add(tuple1);
    List<Tuple> results = KeywordTestHelper.getQueryResults(MEDLINE_TABLE, query, attributeNames, phrase);
    // Perform Check
    boolean contains = TestUtils.equals(expectedResultList, results);
    Assert.assertTrue(contains);
}
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) 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) DoubleField(edu.uci.ics.texera.api.field.DoubleField) Tuple(edu.uci.ics.texera.api.tuple.Tuple) Test(org.junit.Test)

Example 94 with Attribute

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

the class KeywordPhraseTest method testWordInMultipleFieldsQueryWithStopWords5.

/**
 * Verifies: Query with Stop Words match corresponding phrases in the
 * document Used to cause exception sometimes if there is a space between
 * words
 *
 * @throws Exception
 *             with Medline data
 */
@Test
public void testWordInMultipleFieldsQueryWithStopWords5() throws Exception {
    // Prepare Query
    String query = "gain weight";
    ArrayList<String> attributeNames = new ArrayList<>();
    attributeNames.add(keywordTestConstants.ABSTRACT);
    // Prepare expected result list
    List<Span> list = new ArrayList<>();
    Span span1 = new Span(keywordTestConstants.ABSTRACT, 26, 37, "gain weight", "gain weight");
    list.add(span1);
    Attribute[] schemaAttributes = new Attribute[keywordTestConstants.ATTRIBUTES_MEDLINE.length + 1];
    for (int count = 0; count < schemaAttributes.length - 1; count++) {
        schemaAttributes[count] = keywordTestConstants.ATTRIBUTES_MEDLINE[count];
    }
    schemaAttributes[schemaAttributes.length - 1] = new Attribute(RESULTS, AttributeType.LIST);
    IField[] fields = { new IntegerField(4566015), new TextField(""), new TextField("Significance of milk pH in newborn infants."), new TextField("V C Harrison, G Peat"), new StringField("4-5839 Dec 2, 1972"), new TextField("British medical journal"), new TextField(""), new TextField("Infant Nutritional Physiological Phenomena, Infant, Newborn, Milk"), new TextField("Bottle-fed infants do not gain weight as rapidly as breast-fed babies during the first week of life. This " + "weight lag can be corrected by the addition of a small amount of alkali (sodium bicarbonate or trometamol) to " + "the feeds. The alkali corrects the acidity of cow's milk which now assumes some of the properties of human breast " + "milk. It has a bacteriostatic effect on specific Escherichia coli in vitro, and in infants it produces a stool with" + " a preponderance of lactobacilli over E. coli organisms. When alkali is removed from the milk there is a decrease in" + " the weight of an infant and the stools contain excessive numbers of E. coli bacteria.A pH-corrected milk appears to" + " be more physiological than unaltered cow's milk and may provide some protection against gastroenteritis in early " + "life. Its bacteriostatic effect on specific E. coli may be of practical significance in feed preparations where " + "terminal sterilization and refrigeration are not available. The study was conducted during the week after birth, and " + "no conclusions are derived for older infants. The long-term effects of trometamol are unknown. No recommendation can " + "be given for the addition of sodium bicarbonate to milks containing a higher content of sodium."), new DoubleField(0.667832788), new ListField<>(list) };
    Tuple tuple1 = new Tuple(new Schema(schemaAttributes), fields);
    List<Tuple> expectedResultList = new ArrayList<>();
    expectedResultList.add(tuple1);
    List<Tuple> results = KeywordTestHelper.getQueryResults(MEDLINE_TABLE, query, attributeNames, phrase);
    // Perform Check
    boolean contains = TestUtils.equals(expectedResultList, results);
    Assert.assertTrue(contains);
}
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) 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) DoubleField(edu.uci.ics.texera.api.field.DoubleField) Tuple(edu.uci.ics.texera.api.tuple.Tuple) Test(org.junit.Test)

Example 95 with Attribute

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

the class JsonSerializationTest method testAttribute.

@Test
public void testAttribute() {
    Attribute attribute = new Attribute("attrName", AttributeType.TEXT);
    TestUtils.testJsonSerialization(attribute);
}
Also used : Attribute(edu.uci.ics.texera.api.schema.Attribute) Test(org.junit.Test)

Aggregations

Attribute (edu.uci.ics.texera.api.schema.Attribute)98 Test (org.junit.Test)81 Tuple (edu.uci.ics.texera.api.tuple.Tuple)78 ArrayList (java.util.ArrayList)76 Schema (edu.uci.ics.texera.api.schema.Schema)75 IField (edu.uci.ics.texera.api.field.IField)60 StringField (edu.uci.ics.texera.api.field.StringField)56 TextField (edu.uci.ics.texera.api.field.TextField)56 IntegerField (edu.uci.ics.texera.api.field.IntegerField)54 DoubleField (edu.uci.ics.texera.api.field.DoubleField)53 Span (edu.uci.ics.texera.api.span.Span)51 DateField (edu.uci.ics.texera.api.field.DateField)50 SimpleDateFormat (java.text.SimpleDateFormat)47 Dictionary (edu.uci.ics.texera.dataflow.dictionarymatcher.Dictionary)28 AttributeType (edu.uci.ics.texera.api.schema.AttributeType)9 IOperator (edu.uci.ics.texera.api.dataflow.IOperator)8 DataflowException (edu.uci.ics.texera.api.exception.DataflowException)6 TexeraException (edu.uci.ics.texera.api.exception.TexeraException)6 List (java.util.List)6 Collectors (java.util.stream.Collectors)5