Search in sources :

Example 11 with RegexMatcher

use of edu.uci.ics.texera.dataflow.regexmatcher.RegexMatcher 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.texera.dataflow.sink.tuple.TupleSink) IOperator(edu.uci.ics.texera.api.dataflow.IOperator) Join(edu.uci.ics.texera.dataflow.join.Join) Plan(edu.uci.ics.texera.api.engine.Plan) FuzzyTokenMatcher(edu.uci.ics.texera.dataflow.fuzzytokenmatcher.FuzzyTokenMatcher) KeywordMatcherSourceOperator(edu.uci.ics.texera.dataflow.keywordmatcher.KeywordMatcherSourceOperator) ISink(edu.uci.ics.texera.api.dataflow.ISink) ConnectorOutputOperator(edu.uci.ics.texera.dataflow.connector.OneToNBroadcastConnector.ConnectorOutputOperator) NlpEntityOperator(edu.uci.ics.texera.dataflow.nlp.entity.NlpEntityOperator) RegexMatcher(edu.uci.ics.texera.dataflow.regexmatcher.RegexMatcher) OneToNBroadcastConnector(edu.uci.ics.texera.dataflow.connector.OneToNBroadcastConnector) HashSet(java.util.HashSet) Test(org.junit.Test)

Example 12 with RegexMatcher

use of edu.uci.ics.texera.dataflow.regexmatcher.RegexMatcher in project textdb by TextDB.

the class LogicalPlanTest method testGetOutputSchema2.

/*
     * Test getOutputSchema on a valid operator graph.
     *                  -> RegexMatcher -->
     * KeywordSource --<                     >-- Join --> TupleSink
     *                  -> NlpEntityOperator -->
     *
     */
@Test
public void testGetOutputSchema2() throws Exception {
    LogicalPlan logicalPlan = getLogicalPlan2();
    Plan queryPlan = logicalPlan.buildQueryPlan();
    ISink tupleSink = queryPlan.getRoot();
    IOperator join = ((TupleSink) tupleSink).getInputOperator();
    IOperator joinInput1 = ((Join) join).getInnerInputOperator();
    IOperator joinInput2 = ((Join) join).getOuterInputOperator();
    IOperator connectorOut1 = ((RegexMatcher) joinInput1).getInputOperator();
    IOperator connectorOut2 = ((NlpEntityOperator) joinInput2).getInputOperator();
    OneToNBroadcastConnector connector1 = ((ConnectorOutputOperator) connectorOut1).getOwnerConnector();
    OneToNBroadcastConnector connector2 = ((ConnectorOutputOperator) connectorOut2).getOwnerConnector();
    IOperator keywordSource = connector1.getInputOperator();
    join.open();
    Schema expectedJoinOutputSchema = join.getOutputSchema();
    Schema expectedSourceOutputSchema = keywordSource.getOutputSchema();
    Schema expectedMatcherOutputSchema = joinInput1.getOutputSchema();
    Schema expectedNlpEntityOutputSchema = joinInput2.getOutputSchema();
    join.close();
    Schema joinOutputSchema = logicalPlan.getOperatorOutputSchema(JOIN_DISTANCE_ID);
    Schema sourceOutputSchema = logicalPlan.getOperatorOutputSchema(KEYWORD_SOURCE_ID);
    Schema matcherOutputSchema = logicalPlan.getOperatorOutputSchema(REGEX_ID);
    Schema nlpEntityOutputSchema = logicalPlan.getOperatorOutputSchema(NLP_ENTITY_ID);
    Assert.assertEquals(expectedJoinOutputSchema, joinOutputSchema);
    Assert.assertEquals(expectedSourceOutputSchema, sourceOutputSchema);
    Assert.assertEquals(expectedMatcherOutputSchema, matcherOutputSchema);
    Assert.assertEquals(expectedNlpEntityOutputSchema, nlpEntityOutputSchema);
}
Also used : ISink(edu.uci.ics.texera.api.dataflow.ISink) ConnectorOutputOperator(edu.uci.ics.texera.dataflow.connector.OneToNBroadcastConnector.ConnectorOutputOperator) TupleSink(edu.uci.ics.texera.dataflow.sink.tuple.TupleSink) IOperator(edu.uci.ics.texera.api.dataflow.IOperator) NlpEntityOperator(edu.uci.ics.texera.dataflow.nlp.entity.NlpEntityOperator) Schema(edu.uci.ics.texera.api.schema.Schema) Join(edu.uci.ics.texera.dataflow.join.Join) RegexMatcher(edu.uci.ics.texera.dataflow.regexmatcher.RegexMatcher) Plan(edu.uci.ics.texera.api.engine.Plan) OneToNBroadcastConnector(edu.uci.ics.texera.dataflow.connector.OneToNBroadcastConnector) Test(org.junit.Test)

Aggregations

RegexMatcher (edu.uci.ics.texera.dataflow.regexmatcher.RegexMatcher)12 Test (org.junit.Test)11 IOperator (edu.uci.ics.texera.api.dataflow.IOperator)7 ISink (edu.uci.ics.texera.api.dataflow.ISink)7 Plan (edu.uci.ics.texera.api.engine.Plan)7 TupleSink (edu.uci.ics.texera.dataflow.sink.tuple.TupleSink)7 Schema (edu.uci.ics.texera.api.schema.Schema)6 Tuple (edu.uci.ics.texera.api.tuple.Tuple)4 SimilarityJoinPredicate (edu.uci.ics.texera.dataflow.join.SimilarityJoinPredicate)4 OneToNBroadcastConnector (edu.uci.ics.texera.dataflow.connector.OneToNBroadcastConnector)3 ConnectorOutputOperator (edu.uci.ics.texera.dataflow.connector.OneToNBroadcastConnector.ConnectorOutputOperator)3 Join (edu.uci.ics.texera.dataflow.join.Join)3 KeywordMatcherSourceOperator (edu.uci.ics.texera.dataflow.keywordmatcher.KeywordMatcherSourceOperator)3 NlpEntityOperator (edu.uci.ics.texera.dataflow.nlp.entity.NlpEntityOperator)3 IDField (edu.uci.ics.texera.api.field.IDField)2 IntegerField (edu.uci.ics.texera.api.field.IntegerField)2 TextField (edu.uci.ics.texera.api.field.TextField)2 Span (edu.uci.ics.texera.api.span.Span)2 RegexPredicate (edu.uci.ics.texera.dataflow.regexmatcher.RegexPredicate)2 HashSet (java.util.HashSet)2