Examples with IntegerField edu.uci.ics.texera.api.field.IntegerField used on opensource projects

Example 26 with IntegerField

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

the class SimilarityJoinTest method test1.

/*
     * Tests the Similarity Join Predicate on two similar words:
     *   Donald J. Trump
     *   Donald Trump
     * Under the condition of similarity (NormalizedLevenshtein) > 0.8, these two words should match.
     *
     */
@Test
public void test1() throws TexeraException {
    JoinTestHelper.insertToTable(NEWS_TABLE_OUTER, JoinTestConstants.getNewsTuples().get(0));
    JoinTestHelper.insertToTable(NEWS_TABLE_INNER, JoinTestConstants.getNewsTuples().get(1));
    String trumpRegex = "[Dd]onald.{1,5}[Tt]rump";
    RegexMatcher regexMatcherInner = JoinTestHelper.getRegexMatcher(JoinTestHelper.NEWS_TABLE_INNER, trumpRegex, JoinTestConstants.NEWS_BODY);
    RegexMatcher regexMatcherOuter = JoinTestHelper.getRegexMatcher(JoinTestHelper.NEWS_TABLE_OUTER, trumpRegex, JoinTestConstants.NEWS_BODY);
    SimilarityJoinPredicate similarityJoinPredicate = new SimilarityJoinPredicate(JoinTestConstants.NEWS_BODY, 0.8);
    List<Tuple> results = JoinTestHelper.getJoinDistanceResults(regexMatcherInner, regexMatcherOuter, similarityJoinPredicate, Integer.MAX_VALUE, 0);
    Schema joinInputSchema = new Schema.Builder().add(JoinTestConstants.NEWS_SCHEMA).add(SchemaConstants.SPAN_LIST_ATTRIBUTE).build();
    Schema resultSchema = similarityJoinPredicate.generateOutputSchema(joinInputSchema, joinInputSchema);
    List<Span> resultSpanList = Arrays.asList(new Span("inner_" + JoinTestConstants.NEWS_BODY, 5, 20, trumpRegex, "Donald J. Trump", -1), new Span("outer_" + JoinTestConstants.NEWS_BODY, 18, 30, trumpRegex, "Donald Trump", -1));
    Tuple resultTuple = new Tuple(resultSchema, new IDField(UUID.randomUUID().toString()), new IntegerField(2), new TextField("Alternative Facts and the Costs of Trump-Branded Reality"), new TextField("When Donald J. Trump swore the presidential oath on Friday, he assumed " + "responsibility not only for the levers of government but also for one of " + "the United States’ most valuable assets, battered though it may be: its credibility. " + "The country’s sentimental reverence for truth and its jealously guarded press freedoms, " + "while never perfect, have been as important to its global standing as the strength of " + "its military and the reliability of its currency. It’s the bedrock of that " + "American exceptionalism we’ve heard so much about for so long."), new IntegerField(1), new TextField("UCI marchers protest as Trump begins his presidency"), new TextField("a few hours after Donald Trump was sworn in Friday as the nation’s 45th president, " + "a line of more than 100 UC Irvine faculty members and students took to the campus " + "in pouring rain to demonstrate their opposition to his policies on immigration and " + "other issues and urge other opponents to keep organizing during Trump’s presidency."), new ListField<>(resultSpanList));
    Assert.assertTrue(TestUtils.equals(Arrays.asList(resultTuple), results));
}

Example 27 with IntegerField

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

the class KeywordPhraseTest method testWordInMultipleFieldsQueryWithStopWords1.

/**
 * Verifies: Query with Stop Words match corresponding phrases in the
 * document
 *
 * @throws Exception
 */
@Test
public void testWordInMultipleFieldsQueryWithStopWords1() throws Exception {
    // Prepare Query
    String query = "lin and and angry";
    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 span1 = new Span("description", 25, 45, "lin and and angry", "lin clooney is Angry");
    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);
}

Example 28 with IntegerField

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

the class KeywordPhraseTest method testWordInMultipleFieldsQueryWithStopWords3.

/**
 * Verifies: Query with Stop Words match corresponding phrases with Medline
 * data
 *
 * @throws Exception
 *             with Medline data
 */
@Test
public void testWordInMultipleFieldsQueryWithStopWords3() throws Exception {
    // Prepare Query
    String query = "skin rash";
    ArrayList<String> attributeNames = new ArrayList<>();
    attributeNames.add(keywordTestConstants.ABSTRACT);
    // Prepare expected result list
    List<Span> list = new ArrayList<>();
    Span span1 = new Span(keywordTestConstants.ABSTRACT, 192, 201, "skin rash", "skin rash");
    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[] fields1 = { new IntegerField(14347980), new TextField(""), new TextField("CHRONIC MENINGOCOCCEMIA; EPIDEMIOLOGY, DIAGNOSIS AND TREATMENT."), new TextField("D S BLOOM"), new StringField("103 Aug, 1965"), new TextField("California medicine"), new TextField("DRUG THERAPY, MENINGOCOCCAL INFECTIONS, PENICILLIN G, SULFONAMIDES"), new TextField("Drug Therapy, Meningococcal Infections, Penicillin G, Sulfonamides"), new TextField("This report describes four cases of chronic meningococcemia with the characteristic manifestations of recurrent episodes of " + "fever, chills, night sweats, headache and anorexia, associated with skin rash and arthralgias. The diagnosis was established in all instances by blood culture. Administration " + "of sulfonamides in three cases and penicillin in the fourth resulted in prompt recovery. The recent finding of a strain of sulfonamide-resistant meningococci, however, indicates " + "that antibiotic-sensitivity tests should be carried out in all cases of meningococcal disease. While waiting for the results of such tests to be reported, the clinician should " + "initiate treatment with large doses of a sulfonamide and penicillin in combination."), new DoubleField(0.664347980), new ListField<>(list) };
    Tuple tuple1 = new Tuple(new Schema(schemaAttributes), fields1);
    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);
}

Example 29 with IntegerField

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

the class KeywordPhraseTest method testCombinedSpanInMultipleFieldsQuery.

/**
 * Verifies: getNextTuple should return Combined Span info for the phrase
 *
 * @throws Exception
 */
@Test
public void testCombinedSpanInMultipleFieldsQuery() throws Exception {
    // Prepare 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 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, 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] = 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);
}

Example 30 with IntegerField

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

the class KeywordPhraseTest method testWordInMultipleFieldsQueryWithStopWords2.

/**
 * Verifies: Query with Stop Words match corresponding phrases in the
 * document
 *
 * @throws Exception
 */
@Test
public void testWordInMultipleFieldsQueryWithStopWords2() throws Exception {
    // Prepare Query
    String query = "lin clooney and angry";
    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 span1 = new Span("description", 25, 45, "lin clooney and angry", "lin clooney is Angry");
    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);
}