Search in sources :

Example 21 with LabelAndValue

use of org.apache.lucene.facet.LabelAndValue in project lucene-solr by apache.

the class TestTaxonomyFacetCounts method testManyFacetsInOneDocument.

// LUCENE-4583: make sure if we require > 32 KB for one
// document, we don't hit exc when using Facet42DocValuesFormat
public void testManyFacetsInOneDocument() throws Exception {
    assumeTrue("default Codec doesn't support huge BinaryDocValues", TestUtil.fieldSupportsHugeBinaryDocValues(FacetsConfig.DEFAULT_INDEX_FIELD_NAME));
    Directory dir = newDirectory();
    Directory taxoDir = newDirectory();
    IndexWriterConfig iwc = newIndexWriterConfig(new MockAnalyzer(random()));
    RandomIndexWriter writer = new RandomIndexWriter(random(), dir, iwc);
    DirectoryTaxonomyWriter taxoWriter = new DirectoryTaxonomyWriter(taxoDir, IndexWriterConfig.OpenMode.CREATE);
    FacetsConfig config = new FacetsConfig();
    config.setMultiValued("dim", true);
    int numLabels = TestUtil.nextInt(random(), 40000, 100000);
    Document doc = new Document();
    doc.add(newTextField("field", "text", Field.Store.NO));
    for (int i = 0; i < numLabels; i++) {
        doc.add(new FacetField("dim", "" + i));
    }
    writer.addDocument(config.build(taxoWriter, doc));
    // NRT open
    IndexSearcher searcher = newSearcher(writer.getReader());
    // NRT open
    TaxonomyReader taxoReader = new DirectoryTaxonomyReader(taxoWriter);
    Facets facets = getAllFacets(FacetsConfig.DEFAULT_INDEX_FIELD_NAME, searcher, taxoReader, config);
    FacetResult result = facets.getTopChildren(Integer.MAX_VALUE, "dim");
    assertEquals(numLabels, result.labelValues.length);
    Set<String> allLabels = new HashSet<>();
    for (LabelAndValue labelValue : result.labelValues) {
        allLabels.add(labelValue.label);
        assertEquals(1, labelValue.value.intValue());
    }
    assertEquals(numLabels, allLabels.size());
    writer.close();
    IOUtils.close(searcher.getIndexReader(), taxoWriter, taxoReader, dir, taxoDir);
}
Also used : IndexSearcher(org.apache.lucene.search.IndexSearcher) FacetsConfig(org.apache.lucene.facet.FacetsConfig) Facets(org.apache.lucene.facet.Facets) DirectoryTaxonomyReader(org.apache.lucene.facet.taxonomy.directory.DirectoryTaxonomyReader) FacetField(org.apache.lucene.facet.FacetField) Document(org.apache.lucene.document.Document) LabelAndValue(org.apache.lucene.facet.LabelAndValue) DirectoryTaxonomyWriter(org.apache.lucene.facet.taxonomy.directory.DirectoryTaxonomyWriter) MockAnalyzer(org.apache.lucene.analysis.MockAnalyzer) FacetResult(org.apache.lucene.facet.FacetResult) RandomIndexWriter(org.apache.lucene.index.RandomIndexWriter) Directory(org.apache.lucene.store.Directory) IndexWriterConfig(org.apache.lucene.index.IndexWriterConfig) DirectoryTaxonomyReader(org.apache.lucene.facet.taxonomy.directory.DirectoryTaxonomyReader) HashSet(java.util.HashSet)

Example 22 with LabelAndValue

use of org.apache.lucene.facet.LabelAndValue in project searchcode-server by boyter.

the class TimeCodeSearcher method getOwnerFacetResults.

/**
     * Returns the matching owner facets for a given query
     */
private List<CodeFacetOwner> getOwnerFacetResults(IndexSearcher searcher, IndexReader reader, Query query) {
    List<CodeFacetOwner> codeFacetRepo = new ArrayList<>();
    try {
        SortedSetDocValuesReaderState state = new DefaultSortedSetDocValuesReaderState(reader, Values.CODEOWNER);
        FacetsCollector fc = new FacetsCollector();
        FacetsCollector.search(searcher, query, 10, fc);
        Facets facets = new SortedSetDocValuesFacetCounts(state, fc);
        FacetResult result = facets.getTopChildren(200, Values.CODEOWNER);
        if (result != null) {
            int stepThru = result.childCount > 200 ? 200 : result.childCount;
            for (int i = 0; i < stepThru; i++) {
                LabelAndValue lv = result.labelValues[i];
                if (lv != null && lv.value != null) {
                    codeFacetRepo.add(new CodeFacetOwner(lv.label, lv.value.intValue()));
                }
            }
        }
    } catch (IOException ex) {
        LOGGER.warning(" caught a " + ex.getClass() + "\n with message: " + ex.getMessage());
    } catch (Exception ex) {
        LOGGER.warning(" caught a " + ex.getClass() + "\n with message: " + ex.getMessage());
    }
    return codeFacetRepo;
}
Also used : SortedSetDocValuesReaderState(org.apache.lucene.facet.sortedset.SortedSetDocValuesReaderState) DefaultSortedSetDocValuesReaderState(org.apache.lucene.facet.sortedset.DefaultSortedSetDocValuesReaderState) Facets(org.apache.lucene.facet.Facets) ArrayList(java.util.ArrayList) IOException(java.io.IOException) LabelAndValue(org.apache.lucene.facet.LabelAndValue) IOException(java.io.IOException) FacetsCollector(org.apache.lucene.facet.FacetsCollector) DefaultSortedSetDocValuesReaderState(org.apache.lucene.facet.sortedset.DefaultSortedSetDocValuesReaderState) SortedSetDocValuesFacetCounts(org.apache.lucene.facet.sortedset.SortedSetDocValuesFacetCounts) FacetResult(org.apache.lucene.facet.FacetResult)

Example 23 with LabelAndValue

use of org.apache.lucene.facet.LabelAndValue in project searchcode-server by boyter.

the class TimeCodeSearcher method getRevisionFacetResults.

/**
     * Returns the matching revision facets for a given query
     */
private List<CodeFacetRevision> getRevisionFacetResults(IndexSearcher searcher, IndexReader reader, Query query) {
    List<CodeFacetRevision> revisionFacets = new ArrayList<>();
    try {
        SortedSetDocValuesReaderState state = new DefaultSortedSetDocValuesReaderState(reader, Values.REVISION);
        FacetsCollector fc = new FacetsCollector();
        FacetsCollector.search(searcher, query, 10, fc);
        Facets facets = new SortedSetDocValuesFacetCounts(state, fc);
        FacetResult result = facets.getTopChildren(200, Values.REVISION);
        if (result != null) {
            int stepThru = result.childCount > 200 ? 200 : result.childCount;
            for (int i = 0; i < stepThru; i++) {
                LabelAndValue lv = result.labelValues[i];
                if (lv != null && lv.value != null) {
                    revisionFacets.add(new CodeFacetRevision(lv.label, lv.value.intValue()));
                }
            }
        }
    } catch (IOException ex) {
        LOGGER.warning(" caught a " + ex.getClass() + "\n with message: " + ex.getMessage());
    } catch (Exception ex) {
        LOGGER.warning(" caught a " + ex.getClass() + "\n with message: " + ex.getMessage());
    }
    return revisionFacets;
}
Also used : SortedSetDocValuesReaderState(org.apache.lucene.facet.sortedset.SortedSetDocValuesReaderState) DefaultSortedSetDocValuesReaderState(org.apache.lucene.facet.sortedset.DefaultSortedSetDocValuesReaderState) Facets(org.apache.lucene.facet.Facets) ArrayList(java.util.ArrayList) IOException(java.io.IOException) LabelAndValue(org.apache.lucene.facet.LabelAndValue) IOException(java.io.IOException) FacetsCollector(org.apache.lucene.facet.FacetsCollector) DefaultSortedSetDocValuesReaderState(org.apache.lucene.facet.sortedset.DefaultSortedSetDocValuesReaderState) SortedSetDocValuesFacetCounts(org.apache.lucene.facet.sortedset.SortedSetDocValuesFacetCounts) FacetResult(org.apache.lucene.facet.FacetResult)

Example 24 with LabelAndValue

use of org.apache.lucene.facet.LabelAndValue in project searchcode-server by boyter.

the class TimeCodeSearcher method getYearFacetResults.

/**
     * Returns the matching yearmonth facets for a given query
     */
private List<CodeFacetYear> getYearFacetResults(IndexSearcher searcher, IndexReader reader, Query query) {
    List<CodeFacetYear> codeFacetYear = new ArrayList<>();
    try {
        SortedSetDocValuesReaderState state = new DefaultSortedSetDocValuesReaderState(reader, Values.DATEYEAR);
        FacetsCollector fc = new FacetsCollector();
        FacetsCollector.search(searcher, query, 10, fc);
        Facets facets = new SortedSetDocValuesFacetCounts(state, fc);
        FacetResult result = facets.getTopChildren(200, Values.DATEYEAR);
        if (result != null) {
            int stepThru = result.childCount > 200 ? 200 : result.childCount;
            for (int i = 0; i < stepThru; i++) {
                LabelAndValue lv = result.labelValues[i];
                if (lv != null && lv.value != null) {
                    codeFacetYear.add(new CodeFacetYear(lv.label, lv.value.intValue()));
                }
            }
        }
    } catch (IOException ex) {
        LOGGER.warning(" caught a " + ex.getClass() + "\n with message: " + ex.getMessage());
    } catch (Exception ex) {
        LOGGER.warning(" caught a " + ex.getClass() + "\n with message: " + ex.getMessage());
    }
    return codeFacetYear;
}
Also used : SortedSetDocValuesReaderState(org.apache.lucene.facet.sortedset.SortedSetDocValuesReaderState) DefaultSortedSetDocValuesReaderState(org.apache.lucene.facet.sortedset.DefaultSortedSetDocValuesReaderState) Facets(org.apache.lucene.facet.Facets) ArrayList(java.util.ArrayList) IOException(java.io.IOException) LabelAndValue(org.apache.lucene.facet.LabelAndValue) IOException(java.io.IOException) FacetsCollector(org.apache.lucene.facet.FacetsCollector) DefaultSortedSetDocValuesReaderState(org.apache.lucene.facet.sortedset.DefaultSortedSetDocValuesReaderState) SortedSetDocValuesFacetCounts(org.apache.lucene.facet.sortedset.SortedSetDocValuesFacetCounts) FacetResult(org.apache.lucene.facet.FacetResult)

Example 25 with LabelAndValue

use of org.apache.lucene.facet.LabelAndValue in project searchcode-server by boyter.

the class TimeCodeSearcher method getLanguageFacetResults.

/**
     * Returns the matching language facets for a given query
     */
private List<CodeFacetLanguage> getLanguageFacetResults(IndexSearcher searcher, IndexReader reader, Query query) {
    List<CodeFacetLanguage> codeFacetLanguages = new ArrayList<>();
    try {
        SortedSetDocValuesReaderState state = new DefaultSortedSetDocValuesReaderState(reader, Values.LANGUAGENAME);
        FacetsCollector fc = new FacetsCollector();
        FacetsCollector.search(searcher, query, 10, fc);
        Facets facets = new SortedSetDocValuesFacetCounts(state, fc);
        FacetResult result = facets.getTopChildren(200, Values.LANGUAGENAME);
        if (result != null) {
            int stepThru = result.childCount > 200 ? 200 : result.childCount;
            for (int i = 0; i < stepThru; i++) {
                LabelAndValue lv = result.labelValues[i];
                if (lv != null && lv.value != null) {
                    codeFacetLanguages.add(new CodeFacetLanguage(lv.label, lv.value.intValue()));
                }
            }
        }
    } catch (IOException ex) {
        LOGGER.warning(" caught a " + ex.getClass() + "\n with message: " + ex.getMessage());
    } catch (Exception ex) {
        LOGGER.warning(" caught a " + ex.getClass() + "\n with message: " + ex.getMessage());
    }
    return codeFacetLanguages;
}
Also used : SortedSetDocValuesReaderState(org.apache.lucene.facet.sortedset.SortedSetDocValuesReaderState) DefaultSortedSetDocValuesReaderState(org.apache.lucene.facet.sortedset.DefaultSortedSetDocValuesReaderState) Facets(org.apache.lucene.facet.Facets) ArrayList(java.util.ArrayList) IOException(java.io.IOException) LabelAndValue(org.apache.lucene.facet.LabelAndValue) IOException(java.io.IOException) FacetsCollector(org.apache.lucene.facet.FacetsCollector) DefaultSortedSetDocValuesReaderState(org.apache.lucene.facet.sortedset.DefaultSortedSetDocValuesReaderState) SortedSetDocValuesFacetCounts(org.apache.lucene.facet.sortedset.SortedSetDocValuesFacetCounts) FacetResult(org.apache.lucene.facet.FacetResult)

Aggregations

LabelAndValue (org.apache.lucene.facet.LabelAndValue)29 FacetResult (org.apache.lucene.facet.FacetResult)28 Facets (org.apache.lucene.facet.Facets)23 FacetsCollector (org.apache.lucene.facet.FacetsCollector)22 IndexSearcher (org.apache.lucene.search.IndexSearcher)12 IOException (java.io.IOException)11 ArrayList (java.util.ArrayList)11 DefaultSortedSetDocValuesReaderState (org.apache.lucene.facet.sortedset.DefaultSortedSetDocValuesReaderState)11 SortedSetDocValuesFacetCounts (org.apache.lucene.facet.sortedset.SortedSetDocValuesFacetCounts)11 SortedSetDocValuesReaderState (org.apache.lucene.facet.sortedset.SortedSetDocValuesReaderState)11 DirectoryTaxonomyReader (org.apache.lucene.facet.taxonomy.directory.DirectoryTaxonomyReader)9 Document (org.apache.lucene.document.Document)7 FacetsConfig (org.apache.lucene.facet.FacetsConfig)7 RandomIndexWriter (org.apache.lucene.index.RandomIndexWriter)7 MatchAllDocsQuery (org.apache.lucene.search.MatchAllDocsQuery)7 Directory (org.apache.lucene.store.Directory)7 DirectoryReader (org.apache.lucene.index.DirectoryReader)6 TermQuery (org.apache.lucene.search.TermQuery)5 HashMap (java.util.HashMap)4 Map (java.util.Map)4