use of edu.uci.ics.texera.api.field.IntegerField in project textdb by TextDB.
the class TwitterConverter method generateFieldsFromJson.
private List<IField> generateFieldsFromJson(String rawJsonData) {
try {
JsonNode tweet = new ObjectMapper().readTree(rawJsonData);
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();
ZonedDateTime zonedCreateAt = ZonedDateTime.parse(createAt, DateTimeFormatter.ISO_INSTANT.withZone(ZoneId.systemDefault()));
return Arrays.asList(new StringField(id.toString()), 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 DateTimeField(zonedCreateAt.toLocalDateTime()));
} catch (Exception e) {
return Arrays.asList();
}
}
use of edu.uci.ics.texera.api.field.IntegerField in project textdb by TextDB.
the class NltkSentimentOperator method popupOneTuple.
private Tuple popupOneTuple() {
Tuple outputTuple = tupleBuffer.get(0);
tupleBuffer.remove(0);
if (tupleBuffer.isEmpty()) {
tupleBuffer = null;
}
List<IField> outputFields = new ArrayList<>();
outputFields.addAll(outputTuple.getFields());
Integer className = idClassMap.get(outputTuple.getField(SchemaConstants._ID).getValue().toString());
outputFields.add(new IntegerField(className));
return new Tuple(outputSchema, outputFields);
}
use of edu.uci.ics.texera.api.field.IntegerField in project textdb by TextDB.
the class SpanTupleTest method testGetters.
@Test
public void testGetters() throws ParseException {
// create data tuple first
Attribute[] attributes = new Attribute[TestConstants.ATTRIBUTES_PEOPLE.length + 1];
for (int count = 0; count < attributes.length - 1; count++) {
attributes[count] = TestConstants.ATTRIBUTES_PEOPLE[count];
}
attributes[attributes.length - 1] = SchemaConstants.SPAN_LIST_ATTRIBUTE;
List<IField> fields = new ArrayList<IField>(Arrays.asList(new IField[] { new StringField("bruce"), new StringField("lee"), new IntegerField(46), new DoubleField(5.50), new DateField(new SimpleDateFormat("MM-dd-yyyy").parse("01-14-1970")), new TextField("bruce was born in new york city and was grown up in los angeles") }));
IField spanField = createSpanListField();
fields.add(spanField);
spanTuple = new Tuple(new Schema(attributes), fields.toArray(new IField[fields.size()]));
IField spanFieldRetrieved = spanTuple.getField(SchemaConstants.SPAN_LIST);
Assert.assertTrue(spanFieldRetrieved instanceof ListField);
Assert.assertSame(spanField, spanFieldRetrieved);
}
use of edu.uci.ics.texera.api.field.IntegerField in project textdb by TextDB.
the class TestConstantsChinese method getSamplePeopleTuples.
public static List<Tuple> getSamplePeopleTuples() {
try {
IField[] fields1 = { new StringField("无忌"), new StringField("长孙"), new IntegerField(46), new DoubleField(5.50), new DateField(new SimpleDateFormat("MM-dd-yyyy").parse("01-14-1970")), new TextField("北京大学电气工程学院") };
IField[] fields2 = { 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") };
IField[] fields3 = { new StringField("宋江"), new StringField("建筑"), new IntegerField(42), new DoubleField(5.99), new DateField(new SimpleDateFormat("MM-dd-yyyy").parse("01-13-1974")), new TextField("伟大的建筑是历史的坐标,具有传承的价值。") };
Tuple tuple1 = new Tuple(SCHEMA_PEOPLE, fields1);
Tuple tuple2 = new Tuple(SCHEMA_PEOPLE, fields2);
Tuple tuple3 = new Tuple(SCHEMA_PEOPLE, fields3);
return Arrays.asList(tuple1, tuple2, tuple3);
} catch (ParseException e) {
// exception should not happen because we know the data is correct
e.printStackTrace();
return Arrays.asList();
}
}
use of edu.uci.ics.texera.api.field.IntegerField in project textdb by TextDB.
the class TestConstantsRegexSplit method constructSamplePeopleTuples.
public static List<Tuple> constructSamplePeopleTuples() {
IField[] fields1 = { new StringField("bruce"), new StringField("john Lee"), new IntegerField(46), new DoubleField(5.50), new DateTimeField(LocalDateTime.parse("1970-01-01T11:11:11")), new TextField("banana") };
IField[] fields2 = { new StringField("tom hanks"), new StringField("cruise"), new IntegerField(45), new DoubleField(5.95), new DateTimeField(LocalDateTime.parse("1980-01-02T13:14:15")), new TextField("mississippi") };
Tuple tuple1 = new Tuple(SCHEMA_PEOPLE, fields1);
Tuple tuple2 = new Tuple(SCHEMA_PEOPLE, fields2);
return Arrays.asList(tuple1, tuple2);
}
Aggregations