Search in sources :

Example 6 with IOperator

use of edu.uci.ics.textdb.api.dataflow.IOperator in project textdb by TextDB.

the class LogicalPlanTest method testLogicalPlan3.

/*
     * Test a valid operator graph.
     * 
     *                  --> RegexMatcher -->
     *                  |                    >-- Join1
     * KeywordSource --< -> NlpEntityOperator -->          >-- Join2 --> TupleSink
     *                  |                           /
     *                  --> FuzzyTokenMatcher ----->
     * 
     */
@Test
public void testLogicalPlan3() throws Exception {
    LogicalPlan logicalPlan = getLogicalPlan3();
    Plan queryPlan = logicalPlan.buildQueryPlan();
    ISink tupleSink = queryPlan.getRoot();
    Assert.assertTrue(tupleSink instanceof TupleSink);
    IOperator join2 = ((TupleSink) tupleSink).getInputOperator();
    Assert.assertTrue(join2 instanceof Join);
    IOperator join2Input1 = ((Join) join2).getOuterInputOperator();
    Assert.assertTrue(join2Input1 instanceof Join);
    IOperator join2Input2 = ((Join) join2).getInnerInputOperator();
    Assert.assertTrue(join2Input2 instanceof FuzzyTokenMatcher);
    IOperator join1Input1 = ((Join) join2Input1).getInnerInputOperator();
    Assert.assertTrue(join1Input1 instanceof RegexMatcher);
    IOperator join1Input2 = ((Join) join2Input1).getOuterInputOperator();
    Assert.assertTrue(join1Input2 instanceof NlpEntityOperator);
    IOperator connectorOut1 = ((RegexMatcher) join1Input1).getInputOperator();
    Assert.assertTrue(connectorOut1 instanceof ConnectorOutputOperator);
    IOperator connectorOut2 = ((NlpEntityOperator) join1Input2).getInputOperator();
    Assert.assertTrue(connectorOut2 instanceof ConnectorOutputOperator);
    IOperator connectorOut3 = ((FuzzyTokenMatcher) join2Input2).getInputOperator();
    Assert.assertTrue(connectorOut3 instanceof ConnectorOutputOperator);
    HashSet<Integer> connectorIndices = new HashSet<>();
    connectorIndices.add(((ConnectorOutputOperator) connectorOut1).getOutputIndex());
    connectorIndices.add(((ConnectorOutputOperator) connectorOut2).getOutputIndex());
    connectorIndices.add(((ConnectorOutputOperator) connectorOut3).getOutputIndex());
    Assert.assertEquals(connectorIndices.size(), 3);
    OneToNBroadcastConnector connector1 = ((ConnectorOutputOperator) connectorOut1).getOwnerConnector();
    OneToNBroadcastConnector connector2 = ((ConnectorOutputOperator) connectorOut2).getOwnerConnector();
    OneToNBroadcastConnector connector3 = ((ConnectorOutputOperator) connectorOut3).getOwnerConnector();
    Assert.assertSame(connector1, connector2);
    Assert.assertSame(connector1, connector3);
    IOperator keywordSource = connector1.getInputOperator();
    Assert.assertTrue(keywordSource instanceof KeywordMatcherSourceOperator);
}
Also used : TupleSink(edu.uci.ics.textdb.exp.sink.tuple.TupleSink) IOperator(edu.uci.ics.textdb.api.dataflow.IOperator) Join(edu.uci.ics.textdb.exp.join.Join) Plan(edu.uci.ics.textdb.api.engine.Plan) FuzzyTokenMatcher(edu.uci.ics.textdb.exp.fuzzytokenmatcher.FuzzyTokenMatcher) KeywordMatcherSourceOperator(edu.uci.ics.textdb.exp.keywordmatcher.KeywordMatcherSourceOperator) ISink(edu.uci.ics.textdb.api.dataflow.ISink) ConnectorOutputOperator(edu.uci.ics.textdb.exp.connector.OneToNBroadcastConnector.ConnectorOutputOperator) NlpEntityOperator(edu.uci.ics.textdb.exp.nlp.entity.NlpEntityOperator) RegexMatcher(edu.uci.ics.textdb.exp.regexmatcher.RegexMatcher) OneToNBroadcastConnector(edu.uci.ics.textdb.exp.connector.OneToNBroadcastConnector) HashSet(java.util.HashSet) Test(org.junit.Test)

Example 7 with IOperator

use of edu.uci.ics.textdb.api.dataflow.IOperator in project textdb by TextDB.

the class LogicalPlanTest method testLogicalPlan1.

/*
     * Test a valid operator graph.
     * 
     * KeywordSource --> RegexMatcher --> TupleSink
     * 
     */
@Test
public void testLogicalPlan1() throws Exception {
    LogicalPlan logicalPlan = getLogicalPlan1();
    Plan queryPlan = logicalPlan.buildQueryPlan();
    ISink tupleSink = queryPlan.getRoot();
    Assert.assertTrue(tupleSink instanceof TupleSink);
    IOperator regexMatcher = ((TupleSink) tupleSink).getInputOperator();
    Assert.assertTrue(regexMatcher instanceof RegexMatcher);
    IOperator keywordSource = ((RegexMatcher) regexMatcher).getInputOperator();
    Assert.assertTrue(keywordSource instanceof KeywordMatcherSourceOperator);
}
Also used : ISink(edu.uci.ics.textdb.api.dataflow.ISink) TupleSink(edu.uci.ics.textdb.exp.sink.tuple.TupleSink) IOperator(edu.uci.ics.textdb.api.dataflow.IOperator) RegexMatcher(edu.uci.ics.textdb.exp.regexmatcher.RegexMatcher) Plan(edu.uci.ics.textdb.api.engine.Plan) KeywordMatcherSourceOperator(edu.uci.ics.textdb.exp.keywordmatcher.KeywordMatcherSourceOperator) Test(org.junit.Test)

Example 8 with IOperator

use of edu.uci.ics.textdb.api.dataflow.IOperator in project textdb by TextDB.

the class LogicalPlanTest method testLogicalPlan2.

/*
     * Test a valid operator graph.
     *                  -> RegexMatcher -->
     * KeywordSource --<                     >-- Join --> TupleSink
     *                  -> NlpEntityOperator -->
     * 
     */
@Test
public void testLogicalPlan2() throws Exception {
    LogicalPlan logicalPlan = getLogicalPlan2();
    Plan queryPlan = logicalPlan.buildQueryPlan();
    ISink tupleSink = queryPlan.getRoot();
    Assert.assertTrue(tupleSink instanceof TupleSink);
    IOperator join = ((TupleSink) tupleSink).getInputOperator();
    Assert.assertTrue(join instanceof Join);
    IOperator joinInput1 = ((Join) join).getInnerInputOperator();
    Assert.assertTrue(joinInput1 instanceof RegexMatcher);
    IOperator joinInput2 = ((Join) join).getOuterInputOperator();
    Assert.assertTrue(joinInput2 instanceof NlpEntityOperator);
    IOperator connectorOut1 = ((RegexMatcher) joinInput1).getInputOperator();
    Assert.assertTrue(connectorOut1 instanceof ConnectorOutputOperator);
    IOperator connectorOut2 = ((NlpEntityOperator) joinInput2).getInputOperator();
    Assert.assertTrue(connectorOut2 instanceof ConnectorOutputOperator);
    HashSet<Integer> connectorIndices = new HashSet<>();
    connectorIndices.add(((ConnectorOutputOperator) connectorOut1).getOutputIndex());
    connectorIndices.add(((ConnectorOutputOperator) connectorOut2).getOutputIndex());
    Assert.assertEquals(connectorIndices.size(), 2);
    OneToNBroadcastConnector connector1 = ((ConnectorOutputOperator) connectorOut1).getOwnerConnector();
    OneToNBroadcastConnector connector2 = ((ConnectorOutputOperator) connectorOut2).getOwnerConnector();
    Assert.assertSame(connector1, connector2);
    IOperator keywordSource = connector1.getInputOperator();
    Assert.assertTrue(keywordSource instanceof KeywordMatcherSourceOperator);
}
Also used : TupleSink(edu.uci.ics.textdb.exp.sink.tuple.TupleSink) IOperator(edu.uci.ics.textdb.api.dataflow.IOperator) Join(edu.uci.ics.textdb.exp.join.Join) Plan(edu.uci.ics.textdb.api.engine.Plan) KeywordMatcherSourceOperator(edu.uci.ics.textdb.exp.keywordmatcher.KeywordMatcherSourceOperator) ISink(edu.uci.ics.textdb.api.dataflow.ISink) ConnectorOutputOperator(edu.uci.ics.textdb.exp.connector.OneToNBroadcastConnector.ConnectorOutputOperator) NlpEntityOperator(edu.uci.ics.textdb.exp.nlp.entity.NlpEntityOperator) RegexMatcher(edu.uci.ics.textdb.exp.regexmatcher.RegexMatcher) OneToNBroadcastConnector(edu.uci.ics.textdb.exp.connector.OneToNBroadcastConnector) HashSet(java.util.HashSet) Test(org.junit.Test)

Example 9 with IOperator

use of edu.uci.ics.textdb.api.dataflow.IOperator in project textdb by TextDB.

the class LogicalPlan method findSinkOperator.

/*
     * Finds the sink operator in the operator graph.
     * 
     * This function assumes that the graph is valid and there is only one sink in the graph.
     */
private ISink findSinkOperator(HashMap<String, IOperator> operatorObjectMap) throws PlanGenException {
    IOperator sinkOperator = adjacencyList.keySet().stream().filter(operator -> operatorPredicateMap.get(operator).getClass().toString().toLowerCase().contains("sink")).map(operator -> operatorObjectMap.get(operator)).findFirst().orElse(null);
    PlanGenUtils.planGenAssert(sinkOperator != null, "Error: sink operator doesn't exist.");
    PlanGenUtils.planGenAssert(sinkOperator instanceof ISink, "Error: sink operator's type doesn't match.");
    return (ISink) sinkOperator;
}
Also used : JsonProperty(com.fasterxml.jackson.annotation.JsonProperty) OneToNBroadcastConnector(edu.uci.ics.textdb.exp.connector.OneToNBroadcastConnector) HashMap(java.util.HashMap) PlanGenException(edu.uci.ics.textdb.api.exception.PlanGenException) InvocationTargetException(java.lang.reflect.InvocationTargetException) ArrayList(java.util.ArrayList) PredicateBase(edu.uci.ics.textdb.exp.common.PredicateBase) HashSet(java.util.HashSet) LinkedHashMap(java.util.LinkedHashMap) Join(edu.uci.ics.textdb.exp.join.Join) List(java.util.List) ISink(edu.uci.ics.textdb.api.dataflow.ISink) Plan(edu.uci.ics.textdb.api.engine.Plan) JsonCreator(com.fasterxml.jackson.annotation.JsonCreator) PropertyNameConstants(edu.uci.ics.textdb.exp.common.PropertyNameConstants) LinkedHashSet(java.util.LinkedHashSet) IOperator(edu.uci.ics.textdb.api.dataflow.IOperator) ISink(edu.uci.ics.textdb.api.dataflow.ISink) IOperator(edu.uci.ics.textdb.api.dataflow.IOperator)

Example 10 with IOperator

use of edu.uci.ics.textdb.api.dataflow.IOperator 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)

Aggregations

IOperator (edu.uci.ics.textdb.api.dataflow.IOperator)11 Test (org.junit.Test)7 ISink (edu.uci.ics.textdb.api.dataflow.ISink)5 Plan (edu.uci.ics.textdb.api.engine.Plan)5 ArrayList (java.util.ArrayList)5 Tuple (edu.uci.ics.textdb.api.tuple.Tuple)4 OneToNBroadcastConnector (edu.uci.ics.textdb.exp.connector.OneToNBroadcastConnector)4 IField (edu.uci.ics.textdb.api.field.IField)3 TextField (edu.uci.ics.textdb.api.field.TextField)3 Schema (edu.uci.ics.textdb.api.schema.Schema)3 Join (edu.uci.ics.textdb.exp.join.Join)3 KeywordMatcherSourceOperator (edu.uci.ics.textdb.exp.keywordmatcher.KeywordMatcherSourceOperator)3 RegexMatcher (edu.uci.ics.textdb.exp.regexmatcher.RegexMatcher)3 TupleSink (edu.uci.ics.textdb.exp.sink.tuple.TupleSink)3 HashSet (java.util.HashSet)3 DateField (edu.uci.ics.textdb.api.field.DateField)2 DoubleField (edu.uci.ics.textdb.api.field.DoubleField)2 IntegerField (edu.uci.ics.textdb.api.field.IntegerField)2 StringField (edu.uci.ics.textdb.api.field.StringField)2 Attribute (edu.uci.ics.textdb.api.schema.Attribute)2