Search in sources :

Example 1 with DoubleField

use of edu.uci.ics.textdb.api.field.DoubleField 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();
    }
}
Also used : StringField(edu.uci.ics.textdb.api.field.StringField) TextField(edu.uci.ics.textdb.api.field.TextField) IntegerField(edu.uci.ics.textdb.api.field.IntegerField) DateField(edu.uci.ics.textdb.api.field.DateField) ParseException(java.text.ParseException) IField(edu.uci.ics.textdb.api.field.IField) SimpleDateFormat(java.text.SimpleDateFormat) DoubleField(edu.uci.ics.textdb.api.field.DoubleField)

Example 2 with DoubleField

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

the class TestConstantsChineseWordCount method getSamplePeopleTuples.

public static List<Tuple> getSamplePeopleTuples() {
    try {
        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("中新社北京4月26日电 (记者 刘育英)“中国制造2025”政策措施实施以来,“为稳定工业增长、加快制造业转型升级发" + "挥了重要作用”,效果初步显现。") };
        IField[] fields2 = { new StringField("tom hanks"), new StringField("cruise"), new IntegerField(45), new DoubleField(5.95), new DateField(new SimpleDateFormat("MM-dd-yyyy").parse("01-13-1971")), new TextField("  中国2015年发布了“中国制造2025”通知。中国工业和信息化部运行监测协调局副" + "局长黄利斌26日在国新办新闻发布会上表示,自“中国制造2025”实施以来,国家制造业创新中心建设、智能制造" + "、工业强基、绿色制造、高端装备创新等“五大工程”扎实推进;2016年度15个重大标志性项目中,7个完全落实" + ",4个基本落实,其余正在推进。") };
        IField[] fields3 = { new StringField("brad lie angelina"), new StringField("pitt"), new IntegerField(44), new DoubleField(6.10), new DateField(new SimpleDateFormat("MM-dd-yyyy").parse("01-12-1972")), new TextField("  2017年,工信部将重点推进六方面工作:加大“五大工程”实施力度," + "积极推进创新中心建设;扩大试点示范城市(群)覆盖面;实施新一轮重大技术改造升级工程;" + "推进" + "制造业与互联网融合发展;优化制造业发展环境。") };
        IField[] fields4 = { 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("  黄利斌说,今年继续开展“互联" + "网+”制造业试点示范,加快工业互联网基础设施改造升级。现在,47%的大企业" + "搭建了运营协同创新平台,两化融合(信息化和工业化融合)管理体系贯标企业运" + "营成本平均下降了8.8%,经营利润平均增长了6.9%。") };
        IField[] fields5 = { 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("  工信部今年还将" + "选取20-30个城市(群)继续开展“中国制造2025”试点示范创建," + "指导试点示范城市(群),在落实新发展理念等方面先行先试。") };
        IField[] fields6 = { 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("资料图:由驻日中" + "资太阳能企业开发、承建并运营维护的日本岛根县滨田市第二期12兆瓦光伏电站项目(滨田MS太阳能发电站),4月25日在当" + "地举行竣工典礼。该大型太阳能电站的九成设备来自“中国制造”。在日本并网发电的特高压太阳能电站中,这是中国产设备占" + "比最高的项目。中新社记者 王健 摄") };
        Tuple tuple1 = new Tuple(SCHEMA_PEOPLE, fields1);
        Tuple tuple2 = new Tuple(SCHEMA_PEOPLE, fields2);
        Tuple tuple3 = new Tuple(SCHEMA_PEOPLE, fields3);
        Tuple tuple4 = new Tuple(SCHEMA_PEOPLE, fields4);
        Tuple tuple5 = new Tuple(SCHEMA_PEOPLE, fields5);
        Tuple tuple6 = new Tuple(SCHEMA_PEOPLE, fields6);
        return Arrays.asList(tuple1, tuple2, tuple3, tuple4, tuple5, tuple6);
    //            return Arrays.asList(tuple1);
    } catch (ParseException e) {
        // exception should not happen because we know the data is correct
        e.printStackTrace();
        return Arrays.asList();
    }
}
Also used : StringField(edu.uci.ics.textdb.api.field.StringField) TextField(edu.uci.ics.textdb.api.field.TextField) IntegerField(edu.uci.ics.textdb.api.field.IntegerField) DateField(edu.uci.ics.textdb.api.field.DateField) ParseException(java.text.ParseException) IField(edu.uci.ics.textdb.api.field.IField) SimpleDateFormat(java.text.SimpleDateFormat) DoubleField(edu.uci.ics.textdb.api.field.DoubleField)

Example 3 with DoubleField

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

the class ExcelSinkTest method attributeTypeTest.

@Test
public // writing 10000 tuples
void attributeTypeTest() throws Exception {
    ArrayList<String> attributeNames = new ArrayList<>();
    attributeNames.add(TestConstants.FIRST_NAME);
    attributeNames.add(TestConstants.LAST_NAME);
    attributeNames.add(TestConstants.AGE);
    attributeNames.add(TestConstants.HEIGHT);
    attributeNames.add(TestConstants.DATE_OF_BIRTH);
    attributeNames.add(TestConstants.DESCRIPTION);
    // Prepare Schema
    Attribute[] schemaAttributes = new Attribute[TestConstants.ATTRIBUTES_PEOPLE.length];
    for (int count = 0; count < schemaAttributes.length; count++) {
        schemaAttributes[count] = TestConstants.ATTRIBUTES_PEOPLE[count];
    }
    // Prepare 10000 tuples as a tupleList
    int testSize = 10000;
    Random rand = new Random();
    List<Tuple> tupleList = new ArrayList<Tuple>();
    for (int i = 0; i < testSize; i++) {
        IField[] fields = { new StringField(getRandomString()), new StringField(getRandomString()), new IntegerField(rand.nextInt()), new DoubleField(rand.nextDouble() * rand.nextInt()), new DateField(getRandomDate()), new TextField(getRandomString()) };
        tupleList.add(new Tuple(new Schema(schemaAttributes), fields));
    }
    assert (tupleList.size() == testSize);
    IOperator inputOperator = Mockito.mock(IOperator.class);
    Mockito.when(inputOperator.getOutputSchema()).thenReturn(new Schema(schemaAttributes));
    OngoingStubbing<Tuple> stubbing = Mockito.when(inputOperator.getNextTuple());
    for (Tuple t : tupleList) {
        stubbing = stubbing.thenReturn(t);
    }
    stubbing = stubbing.thenReturn(null);
    // excel writing test
    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) Random(java.util.Random) StringField(edu.uci.ics.textdb.api.field.StringField) TextField(edu.uci.ics.textdb.api.field.TextField) DateField(edu.uci.ics.textdb.api.field.DateField) Tuple(edu.uci.ics.textdb.api.tuple.Tuple) DoubleField(edu.uci.ics.textdb.api.field.DoubleField) Test(org.junit.Test)

Example 4 with DoubleField

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

the class KeywordSubstringTest method testSubstringWithStopwordQuery.

/**
     * Verifies: Verifies List<ITuple> returned multiple results by Substring
     * Matcher on query with spaces on both left side and right side.
     * 
     * @throws Exception
     */
@Test
public void testSubstringWithStopwordQuery() throws Exception {
    // Prepare Query
    String query = "is";
    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", 12, 14, "is", "is");
    Span span2 = new Span("description", 37, 39, "is", "is");
    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] = 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.getScanSourceResults(PEOPLE_TABLE, query, attributeNames, substring, Integer.MAX_VALUE, 0);
    // Perform Check
    boolean contains = TestUtils.equals(expectedResultList, resultList);
    Assert.assertTrue(contains);
}
Also used : 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 5 with DoubleField

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

the class KeywordSubstringTest method testCombinedSpanInMultipleFieldsQuery.

/**
     * Verifies: Verifies List<ITuple> returned multiple results by Substring
     * Matcher on query with spaces on both left side and right side.
     * 
     * @throws Exception
     */
@Test
public void testCombinedSpanInMultipleFieldsQuery() throws Exception {
    // Prepare Query
    String query = " 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 span1 = new Span("description", 24, 29, " 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.getScanSourceResults(PEOPLE_TABLE, query, attributeNames, substring, Integer.MAX_VALUE, 0);
    // Perform Check
    boolean contains = TestUtils.equals(expectedResultList, resultList);
    Assert.assertTrue(contains);
}
Also used : 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)

Aggregations

DoubleField (edu.uci.ics.textdb.api.field.DoubleField)69 IField (edu.uci.ics.textdb.api.field.IField)68 IntegerField (edu.uci.ics.textdb.api.field.IntegerField)68 StringField (edu.uci.ics.textdb.api.field.StringField)68 TextField (edu.uci.ics.textdb.api.field.TextField)68 DateField (edu.uci.ics.textdb.api.field.DateField)64 Tuple (edu.uci.ics.textdb.api.tuple.Tuple)64 SimpleDateFormat (java.text.SimpleDateFormat)63 ArrayList (java.util.ArrayList)63 Test (org.junit.Test)63 Attribute (edu.uci.ics.textdb.api.schema.Attribute)56 Schema (edu.uci.ics.textdb.api.schema.Schema)51 Span (edu.uci.ics.textdb.api.span.Span)49 Dictionary (edu.uci.ics.textdb.exp.dictionarymatcher.Dictionary)24 ParseException (java.text.ParseException)4 IOperator (edu.uci.ics.textdb.api.dataflow.IOperator)2 JsonNode (com.fasterxml.jackson.databind.JsonNode)1 ListField (edu.uci.ics.textdb.api.field.ListField)1 Random (java.util.Random)1