/** checks collection-level statistics on Terms */
public void assertTermsStatistics(Terms leftTerms, Terms rightTerms) throws Exception {
if (leftTerms.getDocCount() != -1 && rightTerms.getDocCount() != -1) {
assertEquals(leftTerms.getDocCount(), rightTerms.getDocCount());
}
if (leftTerms.getSumDocFreq() != -1 && rightTerms.getSumDocFreq() != -1) {
assertEquals(leftTerms.getSumDocFreq(), rightTerms.getSumDocFreq());
}
if (leftTerms.getSumTotalTermFreq() != -1 && rightTerms.getSumTotalTermFreq() != -1) {
assertEquals(leftTerms.getSumTotalTermFreq(), rightTerms.getSumTotalTermFreq());
}
if (leftTerms.size() != -1 && rightTerms.size() != -1) {
assertEquals(leftTerms.size(), rightTerms.size());
}
}
private void buildFieldStatistics(XContentBuilder builder, Terms curTerms) throws IOException {
long sumDocFreq = curTerms.getSumDocFreq();
int docCount = curTerms.getDocCount();
long sumTotalTermFrequencies = curTerms.getSumTotalTermFreq();
if (docCount > 0) {
assert ((sumDocFreq > 0)) : "docCount >= 0 but sumDocFreq ain't!";
assert ((sumTotalTermFrequencies > 0)) : "docCount >= 0 but sumTotalTermFrequencies ain't!";
builder.startObject(FieldStrings.FIELD_STATISTICS);
builder.field(FieldStrings.SUM_DOC_FREQ, sumDocFreq);
builder.field(FieldStrings.DOC_COUNT, docCount);
builder.field(FieldStrings.SUM_TTF, sumTotalTermFrequencies);
builder.endObject();
} else if (docCount == -1) { // this should only be -1 if the field
// statistics were not requested at all. In
// this case all 3 values should be -1
assert ((sumDocFreq == -1)) : "docCount was -1 but sumDocFreq ain't!";
assert ((sumTotalTermFrequencies == -1))
: "docCount was -1 but sumTotalTermFrequencies ain't!";
} else {
throw new ElasticsearchIllegalStateException(
"Something is wrong with the field statistics of the term vector request: Values are "
+ "\n"
+ FieldStrings.SUM_DOC_FREQ
+ " "
+ sumDocFreq
+ "\n"
+ FieldStrings.DOC_COUNT
+ " "
+ docCount
+ "\n"
+ FieldStrings.SUM_TTF
+ " "
+ sumTotalTermFrequencies);
}
}
@Override
public FieldStats stats(Terms terms, int maxDoc) throws IOException {
long minValue = NumericUtils.getMinInt(terms);
long maxValue = NumericUtils.getMaxInt(terms);
return new FieldStats.Long(
maxDoc,
terms.getDocCount(),
terms.getSumDocFreq(),
terms.getSumTotalTermFreq(),
minValue,
maxValue);
}
@Override
public FieldStats stats(Terms terms, int maxDoc) throws IOException {
float minValue = NumericUtils.sortableIntToFloat(NumericUtils.getMinInt(terms));
float maxValue = NumericUtils.sortableIntToFloat(NumericUtils.getMaxInt(terms));
return new FieldStats.Float(
maxDoc,
terms.getDocCount(),
terms.getSumDocFreq(),
terms.getSumTotalTermFreq(),
minValue,
maxValue);
}
@Override
public FieldStats.Long stats(IndexReader reader) throws IOException {
int maxDoc = reader.maxDoc();
Terms terms = org.apache.lucene.index.MultiFields.getTerms(reader, name());
if (terms == null) {
return null;
}
long minValue = LegacyNumericUtils.getMinInt(terms);
long maxValue = LegacyNumericUtils.getMaxInt(terms);
return new FieldStats.Long(
maxDoc,
terms.getDocCount(),
terms.getSumDocFreq(),
terms.getSumTotalTermFreq(),
isSearchable(),
isAggregatable(),
minValue,
maxValue);
}
public static TermsEnum filter(
TermsEnum toFilter, Terms terms, AtomicReader reader, Settings settings)
throws IOException {
int docCount = terms.getDocCount();
if (docCount == -1) {
docCount = reader.maxDoc();
}
final double minFrequency = settings.getAsDouble("min", 0d);
final double maxFrequency = settings.getAsDouble("max", docCount + 1d);
final double minSegmentSize = settings.getAsInt("min_segment_size", 0);
if (minSegmentSize < docCount) {
final int minFreq =
minFrequency >= 1.0 ? (int) minFrequency : (int) (docCount * minFrequency);
final int maxFreq =
maxFrequency >= 1.0 ? (int) maxFrequency : (int) (docCount * maxFrequency);
assert minFreq < maxFreq;
return new FrequencyFilter(toFilter, minFreq, maxFreq);
}
return toFilter;
}
@Override
public int getDocCount() throws IOException {
return delegateTerms.getDocCount();
}
private void checkAllInfo(
int numDocs,
String[] values,
int[] freq,
int[][] pos,
int[][] startOffset,
int[][] endOffset,
int i)
throws IOException {
TermVectorsRequestBuilder resp =
client()
.prepareTermVectors("test", "type1", Integer.toString(i))
.setPayloads(true)
.setOffsets(true)
.setPositions(true)
.setFieldStatistics(true)
.setTermStatistics(true)
.setSelectedFields();
assertThat(resp.request().fieldStatistics(), equalTo(true));
TermVectorsResponse response = resp.execute().actionGet();
assertThat("doc id: " + i + " doesn't exists but should", response.isExists(), equalTo(true));
Fields fields = response.getFields();
assertThat(fields.size(), equalTo(1));
Terms terms = fields.terms("field");
assertThat(terms.size(), equalTo(8l));
assertThat(terms.getSumTotalTermFreq(), Matchers.equalTo((long) (9 * numDocs)));
assertThat(terms.getDocCount(), Matchers.equalTo(numDocs));
assertThat(terms.getSumDocFreq(), equalTo((long) numDocs * values.length));
TermsEnum iterator = terms.iterator();
for (int j = 0; j < values.length; j++) {
String string = values[j];
BytesRef next = iterator.next();
assertThat(next, Matchers.notNullValue());
assertThat("expected " + string, string, equalTo(next.utf8ToString()));
assertThat(next, Matchers.notNullValue());
if (string.equals("the")) {
assertThat(
"expected ttf of " + string, numDocs * 2, equalTo((int) iterator.totalTermFreq()));
} else {
assertThat("expected ttf of " + string, numDocs, equalTo((int) iterator.totalTermFreq()));
}
PostingsEnum docsAndPositions = iterator.postings(null, null, PostingsEnum.ALL);
assertThat(docsAndPositions.nextDoc(), equalTo(0));
assertThat(freq[j], equalTo(docsAndPositions.freq()));
assertThat(iterator.docFreq(), equalTo(numDocs));
int[] termPos = pos[j];
int[] termStartOffset = startOffset[j];
int[] termEndOffset = endOffset[j];
assertThat(termPos.length, equalTo(freq[j]));
assertThat(termStartOffset.length, equalTo(freq[j]));
assertThat(termEndOffset.length, equalTo(freq[j]));
for (int k = 0; k < freq[j]; k++) {
int nextPosition = docsAndPositions.nextPosition();
assertThat("term: " + string, nextPosition, equalTo(termPos[k]));
assertThat("term: " + string, docsAndPositions.startOffset(), equalTo(termStartOffset[k]));
assertThat("term: " + string, docsAndPositions.endOffset(), equalTo(termEndOffset[k]));
assertThat("term: " + string, docsAndPositions.getPayload(), equalTo(new BytesRef("word")));
}
}
assertThat(iterator.next(), Matchers.nullValue());
XContentBuilder xBuilder = XContentFactory.jsonBuilder();
xBuilder.startObject();
response.toXContent(xBuilder, ToXContent.EMPTY_PARAMS);
xBuilder.endObject();
BytesStream bytesStream = xBuilder.bytesStream();
String utf8 = bytesStream.bytes().toUtf8().replaceFirst("\"took\":\\d+,", "");
;
String expectedString =
"{\"_index\":\"test\",\"_type\":\"type1\",\"_id\":\""
+ i
+ "\",\"_version\":1,\"found\":true,\"term_vectors\":{\"field\":{\"field_statistics\":{\"sum_doc_freq\":120,\"doc_count\":15,\"sum_ttf\":135},\"terms\":{\"brown\":{\"doc_freq\":15,\"ttf\":15,\"term_freq\":1,\"tokens\":[{\"position\":2,\"start_offset\":10,\"end_offset\":15,\"payload\":\"d29yZA==\"}]},\"dog\":{\"doc_freq\":15,\"ttf\":15,\"term_freq\":1,\"tokens\":[{\"position\":8,\"start_offset\":40,\"end_offset\":43,\"payload\":\"d29yZA==\"}]},\"fox\":{\"doc_freq\":15,\"ttf\":15,\"term_freq\":1,\"tokens\":[{\"position\":3,\"start_offset\":16,\"end_offset\":19,\"payload\":\"d29yZA==\"}]},\"jumps\":{\"doc_freq\":15,\"ttf\":15,\"term_freq\":1,\"tokens\":[{\"position\":4,\"start_offset\":20,\"end_offset\":25,\"payload\":\"d29yZA==\"}]},\"lazy\":{\"doc_freq\":15,\"ttf\":15,\"term_freq\":1,\"tokens\":[{\"position\":7,\"start_offset\":35,\"end_offset\":39,\"payload\":\"d29yZA==\"}]},\"over\":{\"doc_freq\":15,\"ttf\":15,\"term_freq\":1,\"tokens\":[{\"position\":5,\"start_offset\":26,\"end_offset\":30,\"payload\":\"d29yZA==\"}]},\"quick\":{\"doc_freq\":15,\"ttf\":15,\"term_freq\":1,\"tokens\":[{\"position\":1,\"start_offset\":4,\"end_offset\":9,\"payload\":\"d29yZA==\"}]},\"the\":{\"doc_freq\":15,\"ttf\":30,\"term_freq\":2,\"tokens\":[{\"position\":0,\"start_offset\":0,\"end_offset\":3,\"payload\":\"d29yZA==\"},{\"position\":6,\"start_offset\":31,\"end_offset\":34,\"payload\":\"d29yZA==\"}]}}}}}";
assertThat(utf8, equalTo(expectedString));
}
private void duellReaders(CompositeReader other, LeafReader memIndexReader) throws IOException {
Fields memFields = memIndexReader.fields();
for (String field : MultiFields.getFields(other)) {
Terms memTerms = memFields.terms(field);
Terms iwTerms = memIndexReader.terms(field);
if (iwTerms == null) {
assertNull(memTerms);
} else {
NumericDocValues normValues = MultiDocValues.getNormValues(other, field);
NumericDocValues memNormValues = memIndexReader.getNormValues(field);
if (normValues != null) {
// mem idx always computes norms on the fly
assertNotNull(memNormValues);
assertEquals(normValues.get(0), memNormValues.get(0));
}
assertNotNull(memTerms);
assertEquals(iwTerms.getDocCount(), memTerms.getDocCount());
assertEquals(iwTerms.getSumDocFreq(), memTerms.getSumDocFreq());
assertEquals(iwTerms.getSumTotalTermFreq(), memTerms.getSumTotalTermFreq());
TermsEnum iwTermsIter = iwTerms.iterator();
TermsEnum memTermsIter = memTerms.iterator();
if (iwTerms.hasPositions()) {
final boolean offsets = iwTerms.hasOffsets() && memTerms.hasOffsets();
while (iwTermsIter.next() != null) {
assertNotNull(memTermsIter.next());
assertEquals(iwTermsIter.term(), memTermsIter.term());
PostingsEnum iwDocsAndPos = iwTermsIter.postings(null, PostingsEnum.ALL);
PostingsEnum memDocsAndPos = memTermsIter.postings(null, PostingsEnum.ALL);
while (iwDocsAndPos.nextDoc() != PostingsEnum.NO_MORE_DOCS) {
assertEquals(iwDocsAndPos.docID(), memDocsAndPos.nextDoc());
assertEquals(iwDocsAndPos.freq(), memDocsAndPos.freq());
for (int i = 0; i < iwDocsAndPos.freq(); i++) {
assertEquals(
"term: " + iwTermsIter.term().utf8ToString(),
iwDocsAndPos.nextPosition(),
memDocsAndPos.nextPosition());
if (offsets) {
assertEquals(iwDocsAndPos.startOffset(), memDocsAndPos.startOffset());
assertEquals(iwDocsAndPos.endOffset(), memDocsAndPos.endOffset());
}
if (iwTerms.hasPayloads()) {
assertEquals(iwDocsAndPos.getPayload(), memDocsAndPos.getPayload());
}
}
}
}
} else {
while (iwTermsIter.next() != null) {
assertEquals(iwTermsIter.term(), memTermsIter.term());
PostingsEnum iwDocsAndPos = iwTermsIter.postings(null);
PostingsEnum memDocsAndPos = memTermsIter.postings(null);
while (iwDocsAndPos.nextDoc() != PostingsEnum.NO_MORE_DOCS) {
assertEquals(iwDocsAndPos.docID(), memDocsAndPos.nextDoc());
assertEquals(iwDocsAndPos.freq(), memDocsAndPos.freq());
}
}
}
}
}
}