Examples with CompiledAutomaton org.apache.lucene.util.automaton.CompiledAutomaton used on opensource projects

Example 11 with CompiledAutomaton

use of org.apache.lucene.util.automaton.CompiledAutomaton in project lucene-solr by apache.

the class TestBlockPostingsFormat3 method assertTerms.

// following code is almost an exact dup of code from TestDuelingCodecs: sorry!
public void assertTerms(Terms leftTerms, Terms rightTerms, boolean deep) throws Exception {
    if (leftTerms == null || rightTerms == null) {
        assertNull(leftTerms);
        assertNull(rightTerms);
        return;
    }
    assertTermsStatistics(leftTerms, rightTerms);
    // NOTE: we don't assert hasOffsets/hasPositions/hasPayloads because they are allowed to be different
    boolean bothHavePositions = leftTerms.hasPositions() && rightTerms.hasPositions();
    TermsEnum leftTermsEnum = leftTerms.iterator();
    TermsEnum rightTermsEnum = rightTerms.iterator();
    assertTermsEnum(leftTermsEnum, rightTermsEnum, true, bothHavePositions);
    assertTermsSeeking(leftTerms, rightTerms);
    if (deep) {
        int numIntersections = atLeast(3);
        for (int i = 0; i < numIntersections; i++) {
            String re = AutomatonTestUtil.randomRegexp(random());
            CompiledAutomaton automaton = new CompiledAutomaton(new RegExp(re, RegExp.NONE).toAutomaton());
            if (automaton.type == CompiledAutomaton.AUTOMATON_TYPE.NORMAL) {
                // TODO: test start term too
                TermsEnum leftIntersection = leftTerms.intersect(automaton, null);
                TermsEnum rightIntersection = rightTerms.intersect(automaton, null);
                assertTermsEnum(leftIntersection, rightIntersection, rarely(), bothHavePositions);
            }
        }
    }
}

Example 12 with CompiledAutomaton

use of org.apache.lucene.util.automaton.CompiledAutomaton in project lucene-solr by apache.

the class BaseDocValuesFormatTestCase method testSortedTermsEnum.

public void testSortedTermsEnum() throws IOException {
    Directory directory = newDirectory();
    Analyzer analyzer = new MockAnalyzer(random());
    IndexWriterConfig iwconfig = newIndexWriterConfig(analyzer);
    iwconfig.setMergePolicy(newLogMergePolicy());
    RandomIndexWriter iwriter = new RandomIndexWriter(random(), directory, iwconfig);
    Document doc = new Document();
    doc.add(new SortedDocValuesField("field", new BytesRef("hello")));
    iwriter.addDocument(doc);
    doc = new Document();
    doc.add(new SortedDocValuesField("field", new BytesRef("world")));
    iwriter.addDocument(doc);
    doc = new Document();
    doc.add(new SortedDocValuesField("field", new BytesRef("beer")));
    iwriter.addDocument(doc);
    iwriter.forceMerge(1);
    DirectoryReader ireader = iwriter.getReader();
    iwriter.close();
    SortedDocValues dv = getOnlyLeafReader(ireader).getSortedDocValues("field");
    assertEquals(3, dv.getValueCount());
    TermsEnum termsEnum = dv.termsEnum();
    // next()
    assertEquals("beer", termsEnum.next().utf8ToString());
    assertEquals(0, termsEnum.ord());
    assertEquals("hello", termsEnum.next().utf8ToString());
    assertEquals(1, termsEnum.ord());
    assertEquals("world", termsEnum.next().utf8ToString());
    assertEquals(2, termsEnum.ord());
    // seekCeil()
    assertEquals(SeekStatus.NOT_FOUND, termsEnum.seekCeil(new BytesRef("ha!")));
    assertEquals("hello", termsEnum.term().utf8ToString());
    assertEquals(1, termsEnum.ord());
    assertEquals(SeekStatus.FOUND, termsEnum.seekCeil(new BytesRef("beer")));
    assertEquals("beer", termsEnum.term().utf8ToString());
    assertEquals(0, termsEnum.ord());
    assertEquals(SeekStatus.END, termsEnum.seekCeil(new BytesRef("zzz")));
    assertEquals(SeekStatus.NOT_FOUND, termsEnum.seekCeil(new BytesRef("aba")));
    assertEquals(0, termsEnum.ord());
    // seekExact()
    assertTrue(termsEnum.seekExact(new BytesRef("beer")));
    assertEquals("beer", termsEnum.term().utf8ToString());
    assertEquals(0, termsEnum.ord());
    assertTrue(termsEnum.seekExact(new BytesRef("hello")));
    assertEquals(Codec.getDefault().toString(), "hello", termsEnum.term().utf8ToString());
    assertEquals(1, termsEnum.ord());
    assertTrue(termsEnum.seekExact(new BytesRef("world")));
    assertEquals("world", termsEnum.term().utf8ToString());
    assertEquals(2, termsEnum.ord());
    assertFalse(termsEnum.seekExact(new BytesRef("bogus")));
    // seek(ord)
    termsEnum.seekExact(0);
    assertEquals("beer", termsEnum.term().utf8ToString());
    assertEquals(0, termsEnum.ord());
    termsEnum.seekExact(1);
    assertEquals("hello", termsEnum.term().utf8ToString());
    assertEquals(1, termsEnum.ord());
    termsEnum.seekExact(2);
    assertEquals("world", termsEnum.term().utf8ToString());
    assertEquals(2, termsEnum.ord());
    // NORMAL automaton
    termsEnum = dv.intersect(new CompiledAutomaton(new RegExp(".*l.*").toAutomaton()));
    assertEquals("hello", termsEnum.next().utf8ToString());
    assertEquals(1, termsEnum.ord());
    assertEquals("world", termsEnum.next().utf8ToString());
    assertEquals(2, termsEnum.ord());
    assertNull(termsEnum.next());
    // SINGLE automaton
    termsEnum = dv.intersect(new CompiledAutomaton(new RegExp("hello").toAutomaton()));
    assertEquals("hello", termsEnum.next().utf8ToString());
    assertEquals(1, termsEnum.ord());
    assertNull(termsEnum.next());
    ireader.close();
    directory.close();
}