@Test
public void testHugeLabel() throws Exception {
Directory indexDir = newDirectory(), taxoDir = newDirectory();
IndexWriter indexWriter =
new IndexWriter(
indexDir, newIndexWriterConfig(TEST_VERSION_CURRENT, new MockAnalyzer(random())));
DirectoryTaxonomyWriter taxoWriter =
new DirectoryTaxonomyWriter(
taxoDir, OpenMode.CREATE, new Cl2oTaxonomyWriterCache(2, 1f, 1));
FacetFields facetFields = new FacetFields(taxoWriter);
// Add one huge label:
String bigs = null;
int ordinal = -1;
CategoryPath cp = null;
while (true) {
int len = CategoryPath.MAX_CATEGORY_PATH_LENGTH - 4; // for the dimension and separator
bigs = _TestUtil.randomSimpleString(random(), len, len);
cp = new CategoryPath("dim", bigs);
ordinal = taxoWriter.addCategory(cp);
Document doc = new Document();
facetFields.addFields(doc, Collections.singletonList(cp));
indexWriter.addDocument(doc);
break;
}
// Add tiny ones to cause a re-hash
for (int i = 0; i < 3; i++) {
String s = _TestUtil.randomSimpleString(random(), 1, 10);
taxoWriter.addCategory(new CategoryPath("dim", s));
Document doc = new Document();
facetFields.addFields(doc, Collections.singletonList(new CategoryPath("dim", s)));
indexWriter.addDocument(doc);
}
// when too large components were allowed to be added, this resulted in a new added category
assertEquals(ordinal, taxoWriter.addCategory(cp));
IOUtils.close(indexWriter, taxoWriter);
DirectoryReader indexReader = DirectoryReader.open(indexDir);
TaxonomyReader taxoReader = new DirectoryTaxonomyReader(taxoDir);
IndexSearcher searcher = new IndexSearcher(indexReader);
DrillDownQuery ddq = new DrillDownQuery(FacetIndexingParams.DEFAULT);
ddq.add(cp);
assertEquals(1, searcher.search(ddq, 10).totalHits);
IOUtils.close(indexReader, taxoReader);
IOUtils.close(indexDir, taxoDir);
}
public void testSortedSetDocValuesAccumulator() throws Exception {
assumeTrue("Test requires SortedSetDV support", defaultCodecSupportsSortedSet());
Directory dir = newDirectory();
RandomIndexWriter writer = new RandomIndexWriter(random(), dir);
// Use a custom delim char to make sure the impls
// respect it:
final char delim = ':';
FacetIndexingParams fip =
new FacetIndexingParams() {
@Override
public char getFacetDelimChar() {
return delim;
}
};
SortedSetDocValuesFacetFields dvFields = new SortedSetDocValuesFacetFields(fip);
Document doc = new Document();
// Mixup order we add these paths, to verify tie-break
// order is by label (unicode sort) and has nothing to
// do w/ order we added them:
List<CategoryPath> paths = new ArrayList<CategoryPath>();
paths.add(new CategoryPath("a", "foo"));
paths.add(new CategoryPath("a", "bar"));
paths.add(new CategoryPath("a", "zoo"));
Collections.shuffle(paths, random());
paths.add(new CategoryPath("b", "baz"));
paths.add(new CategoryPath("b" + FacetIndexingParams.DEFAULT_FACET_DELIM_CHAR, "bazfoo"));
dvFields.addFields(doc, paths);
writer.addDocument(doc);
if (random().nextBoolean()) {
writer.commit();
}
doc = new Document();
dvFields.addFields(doc, Collections.singletonList(new CategoryPath("a", "foo")));
writer.addDocument(doc);
// NRT open
IndexSearcher searcher = newSearcher(writer.getReader());
writer.close();
List<FacetRequest> requests = new ArrayList<FacetRequest>();
requests.add(new CountFacetRequest(new CategoryPath("a"), 10));
requests.add(new CountFacetRequest(new CategoryPath("b"), 10));
requests.add(
new CountFacetRequest(
new CategoryPath("b" + FacetIndexingParams.DEFAULT_FACET_DELIM_CHAR), 10));
final boolean doDimCount = random().nextBoolean();
CategoryListParams clp =
new CategoryListParams() {
@Override
public OrdinalPolicy getOrdinalPolicy(String dimension) {
return doDimCount ? OrdinalPolicy.NO_PARENTS : OrdinalPolicy.ALL_BUT_DIMENSION;
}
};
FacetSearchParams fsp = new FacetSearchParams(new FacetIndexingParams(clp), requests);
// Per-top-reader state:
SortedSetDocValuesReaderState state =
new SortedSetDocValuesReaderState(fip, searcher.getIndexReader());
// SortedSetDocValuesCollector c = new SortedSetDocValuesCollector(state);
// SortedSetDocValuesCollectorMergeBySeg c = new SortedSetDocValuesCollectorMergeBySeg(state);
FacetsCollector c = FacetsCollector.create(new SortedSetDocValuesAccumulator(fsp, state));
searcher.search(new MatchAllDocsQuery(), c);
// List<FacetResult> results = c.getFacetResults(requests);
List<FacetResult> results = c.getFacetResults();
assertEquals(3, results.size());
int dimCount = doDimCount ? 4 : 0;
assertEquals(
"a (" + dimCount + ")\n foo (2)\n bar (1)\n zoo (1)\n",
FacetTestUtils.toSimpleString(results.get(0)));
dimCount = doDimCount ? 1 : 0;
assertEquals(
"b (" + dimCount + ")\n baz (1)\n", FacetTestUtils.toSimpleString(results.get(1)));
dimCount = doDimCount ? 1 : 0;
assertEquals(
"b" + FacetIndexingParams.DEFAULT_FACET_DELIM_CHAR + " (" + dimCount + ")\n bazfoo (1)\n",
FacetTestUtils.toSimpleString(results.get(2)));
// DrillDown:
DrillDownQuery q = new DrillDownQuery(fip);
q.add(new CategoryPath("a", "foo"));
q.add(new CategoryPath("b", "baz"));
TopDocs hits = searcher.search(q, 1);
assertEquals(1, hits.totalHits);
q = new DrillDownQuery(fip);
q.add(new CategoryPath("a"));
hits = searcher.search(q, 1);
assertEquals(2, hits.totalHits);
searcher.getIndexReader().close();
dir.close();
}
