Examples with RegexMatcher edu.uci.ics.textdb.exp.regexmatcher.RegexMatcher used on opensource projects

Example 6 with RegexMatcher

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

the class SimilarityJoinTest method test4.

/*
     * Tests the Similarity Join Predicate on two similar words:
     *   Galaxy S8
     *   Galaxy Note 7
     * Under the condition of similarity (NormalizedLevenshtein) > 0.8, these two words should NOT match.
     *
     */
@Test
public void test4() throws TextDBException {
    JoinTestHelper.insertToTable(NEWS_TABLE_OUTER, JoinTestConstants.getNewsTuples().get(2));
    JoinTestHelper.insertToTable(NEWS_TABLE_INNER, JoinTestConstants.getNewsTuples().get(3));
    String phoneRegex = "[Gg]alaxy.{1,6}\\d";
    RegexMatcher regexMatcherInner = JoinTestHelper.getRegexMatcher(JoinTestHelper.NEWS_TABLE_INNER, phoneRegex, JoinTestConstants.NEWS_BODY);
    RegexMatcher regexMatcherOuter = JoinTestHelper.getRegexMatcher(JoinTestHelper.NEWS_TABLE_OUTER, phoneRegex, JoinTestConstants.NEWS_BODY);
    SimilarityJoinPredicate similarityJoinPredicate = new SimilarityJoinPredicate(JoinTestConstants.NEWS_BODY, 0.8);
    List<Tuple> results = JoinTestHelper.getJoinDistanceResults(regexMatcherInner, regexMatcherOuter, similarityJoinPredicate, Integer.MAX_VALUE, 0);
    Assert.assertTrue(results.isEmpty());
}

Example 7 with RegexMatcher

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

the class JoinTestHelper method getRegexMatcher.

public static RegexMatcher getRegexMatcher(String tableName, String query, String attrName) {
    try {
        ScanBasedSourceOperator scanBasedSourceOperator = new ScanBasedSourceOperator(new ScanSourcePredicate(tableName));
        RegexMatcher regexMatcher = new RegexMatcher(new RegexPredicate(query, Arrays.asList(attrName), SchemaConstants.SPAN_LIST));
        regexMatcher.setInputOperator(scanBasedSourceOperator);
        return regexMatcher;
    } catch (DataFlowException e) {
        e.printStackTrace();
        return null;
    }
}

Example 8 with RegexMatcher

use of edu.uci.ics.textdb.exp.regexmatcher.RegexMatcher 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);
}