Search in sources :

Example 41 with IntegerField

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

the class TwitterSample method writeTwitterIndex.

public static void writeTwitterIndex() throws Exception {
    RelationManager relationManager = RelationManager.getRelationManager();
    relationManager.deleteTable(twitterClimateTable);
    relationManager.createTable(twitterClimateTable, "../index/twitter/", TwitterSchema.TWITTER_SCHEMA, LuceneAnalyzerConstants.standardAnalyzerString());
    DataWriter dataWriter = relationManager.getTableDataWriter(twitterClimateTable);
    dataWriter.open();
    int counter = 0;
    JsonNode jsonNode = new ObjectMapper().readTree(new File(twitterFilePath));
    for (JsonNode tweet : jsonNode) {
        try {
            String text = tweet.get("text").asText();
            Long id = tweet.get("id").asLong();
            String tweetLink = "https://twitter.com/statuses/" + id;
            JsonNode userNode = tweet.get("user");
            String userScreenName = userNode.get("screen_name").asText();
            String userLink = "https://twitter.com/" + userScreenName;
            String userName = userNode.get("name").asText();
            String userDescription = userNode.get("description").asText();
            Integer userFollowersCount = userNode.get("followers_count").asInt();
            Integer userFriendsCount = userNode.get("friends_count").asInt();
            JsonNode geoTagNode = tweet.get("geo_tag");
            String state = geoTagNode.get("stateName").asText();
            String county = geoTagNode.get("countyName").asText();
            String city = geoTagNode.get("cityName").asText();
            String createAt = tweet.get("create_at").asText();
            Tuple tuple = new Tuple(TwitterSchema.TWITTER_SCHEMA, new TextField(text), new StringField(tweetLink), new StringField(userLink), new TextField(userScreenName), new TextField(userName), new TextField(userDescription), new IntegerField(userFollowersCount), new IntegerField(userFriendsCount), new TextField(state), new TextField(county), new TextField(city), new StringField(createAt));
            dataWriter.insertTuple(tuple);
            counter++;
        } catch (RuntimeException e) {
            e.printStackTrace();
            continue;
        }
    }
    dataWriter.close();
    System.out.println("write twitter data finished");
    System.out.println(counter + " tweets written");
}
Also used : JsonNode(com.fasterxml.jackson.databind.JsonNode) IntegerField(edu.uci.ics.textdb.api.field.IntegerField) StringField(edu.uci.ics.textdb.api.field.StringField) TextField(edu.uci.ics.textdb.api.field.TextField) File(java.io.File) ObjectMapper(com.fasterxml.jackson.databind.ObjectMapper) Tuple(edu.uci.ics.textdb.api.tuple.Tuple) RelationManager(edu.uci.ics.textdb.storage.RelationManager) DataWriter(edu.uci.ics.textdb.storage.DataWriter)

Example 42 with IntegerField

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

the class ExcelSinkTest method writeSampleExcelFile.

/**
     * Create two tuples, write into a excel file. Need to manually delete the generated file.
     * @throws ParseException
     */
@Test
public void writeSampleExcelFile() 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 inputOperator = Mockito.mock(IOperator.class);
    Mockito.when(inputOperator.getOutputSchema()).thenReturn(new Schema(schemaAttributes)).thenReturn(null);
    Mockito.when(inputOperator.getNextTuple()).thenReturn(tuple1).thenReturn(tuple2).thenReturn(null);
    excelSink = new ExcelSink(new ExcelSinkPredicate());
    excelSink.setInputOperator(inputOperator);
    excelSink.open();
    excelSink.collectAllTuples();
    excelSink.close();
    Files.deleteIfExists(Paths.get(excelSink.getFilePath()));
}
Also used : Attribute(edu.uci.ics.textdb.api.schema.Attribute) IOperator(edu.uci.ics.textdb.api.dataflow.IOperator) Schema(edu.uci.ics.textdb.api.schema.Schema) ArrayList(java.util.ArrayList) IntegerField(edu.uci.ics.textdb.api.field.IntegerField) IField(edu.uci.ics.textdb.api.field.IField) Span(edu.uci.ics.textdb.api.span.Span) StringField(edu.uci.ics.textdb.api.field.StringField) TextField(edu.uci.ics.textdb.api.field.TextField) DateField(edu.uci.ics.textdb.api.field.DateField) SimpleDateFormat(java.text.SimpleDateFormat) DoubleField(edu.uci.ics.textdb.api.field.DoubleField) Tuple(edu.uci.ics.textdb.api.tuple.Tuple) Test(org.junit.Test)

Example 43 with IntegerField

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

the class DictionaryMatcherTest method testSingleWordQueryInTextFieldUsingScan.

/**
     * Scenario S-5:verifies GetNextTuple of DictionaryMatcher and single word
     * queries in Text Field using SCANOPERATOR
     */
@Test
public void testSingleWordQueryInTextFieldUsingScan() throws Exception {
    ArrayList<String> names = new ArrayList<String>(Arrays.asList("tall"));
    Dictionary dictionary = new Dictionary(names);
    // create a data tuple first
    List<Span> list = new ArrayList<Span>();
    Span span = new Span("description", 0, 4, "tall", "Tall");
    list.add(span);
    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>(list) };
    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.textdb.exp.dictionarymatcher.Dictionary) Attribute(edu.uci.ics.textdb.api.schema.Attribute) Schema(edu.uci.ics.textdb.api.schema.Schema) ArrayList(java.util.ArrayList) IntegerField(edu.uci.ics.textdb.api.field.IntegerField) IField(edu.uci.ics.textdb.api.field.IField) Span(edu.uci.ics.textdb.api.span.Span) StringField(edu.uci.ics.textdb.api.field.StringField) TextField(edu.uci.ics.textdb.api.field.TextField) DateField(edu.uci.ics.textdb.api.field.DateField) SimpleDateFormat(java.text.SimpleDateFormat) DoubleField(edu.uci.ics.textdb.api.field.DoubleField) Tuple(edu.uci.ics.textdb.api.tuple.Tuple) Test(org.junit.Test)

Example 44 with IntegerField

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

the class DictionaryMatcherTest method testWordInMultipleFieldsQueryUsingScan.

/**
     * Scenario S-11:verifies: data source has multiple attributes, and an
     * entity can appear in all the fields and multiple times using SUBSTRING_SCANBASE
     * OPERATOR.
     */
@Test
public void testWordInMultipleFieldsQueryUsingScan() 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.SUBSTRING_SCANBASED);
    boolean contains = TestUtils.equals(expectedResults, returnedResults);
    Assert.assertTrue(contains);
}
Also used : Dictionary(edu.uci.ics.textdb.exp.dictionarymatcher.Dictionary) Attribute(edu.uci.ics.textdb.api.schema.Attribute) Schema(edu.uci.ics.textdb.api.schema.Schema) ArrayList(java.util.ArrayList) IntegerField(edu.uci.ics.textdb.api.field.IntegerField) IField(edu.uci.ics.textdb.api.field.IField) Span(edu.uci.ics.textdb.api.span.Span) StringField(edu.uci.ics.textdb.api.field.StringField) TextField(edu.uci.ics.textdb.api.field.TextField) DateField(edu.uci.ics.textdb.api.field.DateField) SimpleDateFormat(java.text.SimpleDateFormat) DoubleField(edu.uci.ics.textdb.api.field.DoubleField) Tuple(edu.uci.ics.textdb.api.tuple.Tuple) Test(org.junit.Test)

Example 45 with IntegerField

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

the class ComparableMatcherTest method testIntegerMatching1.

/**
     * Verifies the behavior of ComparableMatcher<Integer> with matching type EQUAL_TO
     *
     * @throws Exception
     */
@Test
public void testIntegerMatching1() throws Exception {
    // Prepare the query
    int threshold = 42;
    Attribute attribute = TestConstants.AGE_ATTR;
    String attributeName = attribute.getAttributeName();
    ComparisonType matchingType = ComparisonType.EQUAL_TO;
    // Perform the query
    List<Tuple> returnedResults = getIntegerQueryResults(attributeName, matchingType, threshold);
    IField[] fields1 = { 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") };
    IField[] fields2 = { new StringField("Mary brown"), new StringField("Lake Forest"), new IntegerField(42), new DoubleField(5.99), new DateField(new SimpleDateFormat("MM-dd-yyyy").parse("01-13-1974")), new TextField("Short angry") };
    List<Tuple> expectedResults = new ArrayList<>();
    expectedResults.add(new Tuple(TestConstants.SCHEMA_PEOPLE, fields1));
    expectedResults.add(new Tuple(TestConstants.SCHEMA_PEOPLE, fields2));
    // check the results
    Assert.assertEquals(2, returnedResults.size());
    Assert.assertTrue(TestUtils.equals(expectedResults, returnedResults));
}
Also used : Attribute(edu.uci.ics.textdb.api.schema.Attribute) ArrayList(java.util.ArrayList) IntegerField(edu.uci.ics.textdb.api.field.IntegerField) IField(edu.uci.ics.textdb.api.field.IField) StringField(edu.uci.ics.textdb.api.field.StringField) TextField(edu.uci.ics.textdb.api.field.TextField) DateField(edu.uci.ics.textdb.api.field.DateField) SimpleDateFormat(java.text.SimpleDateFormat) Tuple(edu.uci.ics.textdb.api.tuple.Tuple) DoubleField(edu.uci.ics.textdb.api.field.DoubleField) Test(org.junit.Test)

Aggregations

IntegerField (edu.uci.ics.textdb.api.field.IntegerField)85 StringField (edu.uci.ics.textdb.api.field.StringField)81 IField (edu.uci.ics.textdb.api.field.IField)80 TextField (edu.uci.ics.textdb.api.field.TextField)80 Tuple (edu.uci.ics.textdb.api.tuple.Tuple)80 ArrayList (java.util.ArrayList)76 Test (org.junit.Test)74 DoubleField (edu.uci.ics.textdb.api.field.DoubleField)68 DateField (edu.uci.ics.textdb.api.field.DateField)64 SimpleDateFormat (java.text.SimpleDateFormat)63 Schema (edu.uci.ics.textdb.api.schema.Schema)62 Span (edu.uci.ics.textdb.api.span.Span)60 Attribute (edu.uci.ics.textdb.api.schema.Attribute)57 Dictionary (edu.uci.ics.textdb.exp.dictionarymatcher.Dictionary)24 JoinDistancePredicate (edu.uci.ics.textdb.exp.join.JoinDistancePredicate)9 KeywordMatcherSourceOperator (edu.uci.ics.textdb.exp.keywordmatcher.KeywordMatcherSourceOperator)9 ParseException (java.text.ParseException)4 JsonNode (com.fasterxml.jackson.databind.JsonNode)2 IOperator (edu.uci.ics.textdb.api.dataflow.IOperator)2 IDField (edu.uci.ics.textdb.api.field.IDField)2