public void testCopyFieldsAndFieldBoostsAndDocBoosts() throws Exception {
SolrCore core = h.getCore();
IndexSchema schema = core.getLatestSchema();
SolrInputDocument doc = new SolrInputDocument();
final float DOC_BOOST = 3.0F;
doc.setDocumentBoost(DOC_BOOST);
doc.addField("id", "42");
SolrInputField inTitle = new SolrInputField("title");
inTitle.addValue("titleA", 2.0F);
inTitle.addValue("titleB", 7.0F);
final float TITLE_BOOST = 2.0F * 7.0F;
assertEquals(TITLE_BOOST, inTitle.getBoost(), 0.0F);
doc.put(inTitle.getName(), inTitle);
SolrInputField inFoo = new SolrInputField("foo_t");
inFoo.addValue("summer time", 1.0F);
inFoo.addValue("in the city", 5.0F);
inFoo.addValue("living is easy", 11.0F);
final float FOO_BOOST = 1.0F * 5.0F * 11.0F;
assertEquals(FOO_BOOST, inFoo.getBoost(), 0.0F);
doc.put(inFoo.getName(), inFoo);
Document out = DocumentBuilder.toDocument(doc, schema);
IndexableField[] outTitle = out.getFields(inTitle.getName());
assertEquals("wrong number of title values", 2, outTitle.length);
IndexableField[] outNoNorms = out.getFields("title_stringNoNorms");
assertEquals("wrong number of nonorms values", 2, outNoNorms.length);
IndexableField[] outFoo = out.getFields(inFoo.getName());
assertEquals("wrong number of foo values", 3, outFoo.length);
IndexableField[] outText = out.getFields("text");
assertEquals("wrong number of text values", 5, outText.length);
// since Lucene no longer has native document boosts, we should find
// the doc boost multiplied into the boost on the first field value
// of each field. All other field values should be 1.0f
// (lucene will multiply all of the field value boosts later)
assertEquals(TITLE_BOOST * DOC_BOOST, outTitle[0].boost(), 0.0F);
assertEquals(1.0F, outTitle[1].boost(), 0.0F);
//
assertEquals(FOO_BOOST * DOC_BOOST, outFoo[0].boost(), 0.0F);
assertEquals(1.0F, outFoo[1].boost(), 0.0F);
assertEquals(1.0F, outFoo[2].boost(), 0.0F);
//
assertEquals(TITLE_BOOST * DOC_BOOST, outText[0].boost(), 0.0F);
assertEquals(1.0F, outText[1].boost(), 0.0F);
assertEquals(FOO_BOOST, outText[2].boost(), 0.0F);
assertEquals(1.0F, outText[3].boost(), 0.0F);
assertEquals(1.0F, outText[4].boost(), 0.0F);
// copyField dest with no norms should not have recieved any boost
assertEquals(1.0F, outNoNorms[0].boost(), 0.0F);
assertEquals(1.0F, outNoNorms[1].boost(), 0.0F);
// now index that SolrInputDocument to check the computed norms
assertU(adoc(doc));
assertU(commit());
SolrQueryRequest req = req("q", "id:42");
try {
// very hack-ish
SolrQueryResponse rsp = new SolrQueryResponse();
core.execute(core.getRequestHandler(req.getParams().get(CommonParams.QT)), req, rsp);
DocList dl = ((ResultContext) rsp.getValues().get("response")).docs;
assertTrue("can't find the doc we just added", 1 == dl.size());
int docid = dl.iterator().nextDoc();
SolrIndexSearcher searcher = req.getSearcher();
AtomicReader reader =
SlowCompositeReaderWrapper.wrap(searcher.getTopReaderContext().reader());
assertTrue(
"similarity doesn't extend DefaultSimilarity, "
+ "config or defaults have changed since test was written",
searcher.getSimilarity() instanceof DefaultSimilarity);
DefaultSimilarity sim = (DefaultSimilarity) searcher.getSimilarity();
NumericDocValues titleNorms = reader.getNormValues("title");
NumericDocValues fooNorms = reader.getNormValues("foo_t");
NumericDocValues textNorms = reader.getNormValues("text");
assertEquals(expectedNorm(sim, 2, TITLE_BOOST * DOC_BOOST), titleNorms.get(docid));
assertEquals(expectedNorm(sim, 8 - 3, FOO_BOOST * DOC_BOOST), fooNorms.get(docid));
assertEquals(
expectedNorm(sim, 2 + 8 - 3, TITLE_BOOST * FOO_BOOST * DOC_BOOST), textNorms.get(docid));
} finally {
req.close();
}
}
public static NamedList<Integer> getCounts(
SolrIndexSearcher searcher,
DocSet docs,
String fieldName,
int offset,
int limit,
int mincount,
boolean missing,
String sort,
String prefix)
throws IOException {
SchemaField schemaField = searcher.getSchema().getField(fieldName);
FieldType ft = schemaField.getType();
NamedList<Integer> res = new NamedList<Integer>();
final SortedSetDocValues si; // for term lookups only
OrdinalMap ordinalMap = null; // for mapping per-segment ords to global ones
if (schemaField.multiValued()) {
si = searcher.getAtomicReader().getSortedSetDocValues(fieldName);
if (si instanceof MultiSortedSetDocValues) {
ordinalMap = ((MultiSortedSetDocValues) si).mapping;
}
} else {
SortedDocValues single = searcher.getAtomicReader().getSortedDocValues(fieldName);
si = single == null ? null : new SingletonSortedSetDocValues(single);
if (single instanceof MultiSortedDocValues) {
ordinalMap = ((MultiSortedDocValues) single).mapping;
}
}
if (si == null) {
return finalize(res, searcher, schemaField, docs, -1, missing);
}
if (si.getValueCount() >= Integer.MAX_VALUE) {
throw new UnsupportedOperationException(
"Currently this faceting method is limited to " + Integer.MAX_VALUE + " unique terms");
}
final BytesRef br = new BytesRef();
final BytesRef prefixRef;
if (prefix == null) {
prefixRef = null;
} else if (prefix.length() == 0) {
prefix = null;
prefixRef = null;
} else {
prefixRef = new BytesRef(prefix);
}
int startTermIndex, endTermIndex;
if (prefix != null) {
startTermIndex = (int) si.lookupTerm(prefixRef);
if (startTermIndex < 0) startTermIndex = -startTermIndex - 1;
prefixRef.append(UnicodeUtil.BIG_TERM);
endTermIndex = (int) si.lookupTerm(prefixRef);
assert endTermIndex < 0;
endTermIndex = -endTermIndex - 1;
} else {
startTermIndex = -1;
endTermIndex = (int) si.getValueCount();
}
final int nTerms = endTermIndex - startTermIndex;
int missingCount = -1;
final CharsRef charsRef = new CharsRef(10);
if (nTerms > 0 && docs.size() >= mincount) {
// count collection array only needs to be as big as the number of terms we are
// going to collect counts for.
final int[] counts = new int[nTerms];
Filter filter = docs.getTopFilter();
List<AtomicReaderContext> leaves = searcher.getTopReaderContext().leaves();
for (int subIndex = 0; subIndex < leaves.size(); subIndex++) {
AtomicReaderContext leaf = leaves.get(subIndex);
DocIdSet dis =
filter.getDocIdSet(leaf, null); // solr docsets already exclude any deleted docs
DocIdSetIterator disi = null;
if (dis != null) {
disi = dis.iterator();
}
if (disi != null) {
if (schemaField.multiValued()) {
SortedSetDocValues sub = leaf.reader().getSortedSetDocValues(fieldName);
if (sub == null) {
sub = SortedSetDocValues.EMPTY;
}
if (sub instanceof SingletonSortedSetDocValues) {
// some codecs may optimize SORTED_SET storage for single-valued fields
final SortedDocValues values =
((SingletonSortedSetDocValues) sub).getSortedDocValues();
accumSingle(counts, startTermIndex, values, disi, subIndex, ordinalMap);
} else {
accumMulti(counts, startTermIndex, sub, disi, subIndex, ordinalMap);
}
} else {
SortedDocValues sub = leaf.reader().getSortedDocValues(fieldName);
if (sub == null) {
sub = SortedDocValues.EMPTY;
}
accumSingle(counts, startTermIndex, sub, disi, subIndex, ordinalMap);
}
}
}
if (startTermIndex == -1) {
missingCount = counts[0];
}
// IDEA: we could also maintain a count of "other"... everything that fell outside
// of the top 'N'
int off = offset;
int lim = limit >= 0 ? limit : Integer.MAX_VALUE;
if (sort.equals(FacetParams.FACET_SORT_COUNT)
|| sort.equals(FacetParams.FACET_SORT_COUNT_LEGACY)) {
int maxsize = limit > 0 ? offset + limit : Integer.MAX_VALUE - 1;
maxsize = Math.min(maxsize, nTerms);
LongPriorityQueue queue =
new LongPriorityQueue(Math.min(maxsize, 1000), maxsize, Long.MIN_VALUE);
int min = mincount - 1; // the smallest value in the top 'N' values
for (int i = (startTermIndex == -1) ? 1 : 0; i < nTerms; i++) {
int c = counts[i];
if (c > min) {
// NOTE: we use c>min rather than c>=min as an optimization because we are going in
// index order, so we already know that the keys are ordered. This can be very
// important if a lot of the counts are repeated (like zero counts would be).
// smaller term numbers sort higher, so subtract the term number instead
long pair = (((long) c) << 32) + (Integer.MAX_VALUE - i);
boolean displaced = queue.insert(pair);
if (displaced) min = (int) (queue.top() >>> 32);
}
}
// if we are deep paging, we don't have to order the highest "offset" counts.
int collectCount = Math.max(0, queue.size() - off);
assert collectCount <= lim;
// the start and end indexes of our list "sorted" (starting with the highest value)
int sortedIdxStart = queue.size() - (collectCount - 1);
int sortedIdxEnd = queue.size() + 1;
final long[] sorted = queue.sort(collectCount);
for (int i = sortedIdxStart; i < sortedIdxEnd; i++) {
long pair = sorted[i];
int c = (int) (pair >>> 32);
int tnum = Integer.MAX_VALUE - (int) pair;
si.lookupOrd(startTermIndex + tnum, br);
ft.indexedToReadable(br, charsRef);
res.add(charsRef.toString(), c);
}
} else {
// add results in index order
int i = (startTermIndex == -1) ? 1 : 0;
if (mincount <= 0) {
// if mincount<=0, then we won't discard any terms and we know exactly
// where to start.
i += off;
off = 0;
}
for (; i < nTerms; i++) {
int c = counts[i];
if (c < mincount || --off >= 0) continue;
if (--lim < 0) break;
si.lookupOrd(startTermIndex + i, br);
ft.indexedToReadable(br, charsRef);
res.add(charsRef.toString(), c);
}
}
}
return finalize(res, searcher, schemaField, docs, missingCount, missing);
}
