/*
* Utility function to display a term vector.
*/
static void termVectorDisplay(Terms terms) throws IOException {
if ((terms == null) || (terms.size() == -1)) System.out.println(" The field is not stored.");
else {
/*
* The terms for this field are stored.
*/
System.out.println(" Vocabulary size: " + terms.size() + " terms");
TermsEnum ithTerm = terms.iterator(null);
/*
* Iterate over the terms in this document.
* Information about a term's occurrences (tf and
* positions) is accessed via the indexing API, which
* returns inverted lists that describe (only) the
* current document.
*/
while (ithTerm.next() != null) {
System.out.format(
" %10d %-20s %d ",
ithTerm.ord(), ithTerm.term().utf8ToString(), ithTerm.totalTermFreq());
DocsAndPositionsEnum currDoc = ithTerm.docsAndPositions(null, null);
currDoc.nextDoc();
for (int jthPosition = 0; jthPosition < ithTerm.totalTermFreq(); jthPosition++)
System.out.print(currDoc.nextPosition() + " ");
System.out.println();
}
;
}
;
}
/**
* We will implement this according to the Lucene specification the formula used: sum ( IDF(qi) *
* (df(qi,D) * (k+1)) / (df(qi,D) + k * (1-b + b*|D| / avgFL)) IDF and avgFL computation are
* described above.
*
* @param doc
* @param terms
* @param context
* @return
*/
@Override
public float extract(Document doc, Terms terms, RerankerContext context) {
Set<String> queryTokens = new HashSet<>(context.getQueryTokens());
TermsEnum termsEnum = null;
try {
termsEnum = terms.iterator();
} catch (IOException e) {
LOG.warn("Error computing BM25, unable to retrieve terms enum");
return 0.0f;
}
IndexReader reader = context.getIndexSearcher().getIndexReader();
long maxDocs = reader.numDocs();
long sumTotalTermFreq = getSumTermFrequency(reader, context.getField());
// Compute by iterating
long docSize = 0L;
// NOTE df cannot be retrieved just from the term vector,
// the term vector here is only a partial term vector that treats this as if we only have 1
// document in the index
Map<String, Integer> docFreqMap = null;
try {
docFreqMap = getDocFreqs(reader, context.getQueryTokens(), context.getField());
} catch (IOException e) {
LOG.warn("Unable to retrieve document frequencies.");
docFreqMap = new HashMap<>();
}
Map<String, Long> termFreqMap = new HashMap<>();
try {
while (termsEnum.next() != null) {
String termString = termsEnum.term().utf8ToString();
docSize += termsEnum.totalTermFreq();
if (queryTokens.contains(termString)) {
termFreqMap.put(termString, termsEnum.totalTermFreq());
}
}
} catch (IOException e) {
LOG.warn("Unable to retrieve termsEnum, treating as 0");
}
float score = 0.0f;
// Iterate over the query tokens
double avgFL = computeAvgFL(sumTotalTermFreq, maxDocs);
for (String token : queryTokens) {
long docFreq = docFreqMap.containsKey(token) ? docFreqMap.get(token) : 0;
double termFreq = termFreqMap.containsKey(token) ? termFreqMap.get(token) : 0;
double numerator = (this.k1 + 1) * termFreq;
double docLengthFactor = this.b * (docSize / avgFL);
double denominator = termFreq + (this.k1) * (1 - this.b + docLengthFactor);
score += computeIDF(docFreq, maxDocs) * numerator / denominator;
}
return score;
}
private void buildTermStatistics(XContentBuilder builder, TermsEnum termIter) throws IOException {
// write term statistics. At this point we do not naturally have a
// boolean that says if these values actually were requested.
// However, we can assume that they were not if the statistic values are
// <= 0.
assert (((termIter.docFreq() > 0) && (termIter.totalTermFreq() > 0))
|| ((termIter.docFreq() == -1) && (termIter.totalTermFreq() == -1)));
int docFreq = termIter.docFreq();
if (docFreq > 0) {
builder.field(FieldStrings.DOC_FREQ, docFreq);
builder.field(FieldStrings.TTF, termIter.totalTermFreq());
}
}
/**
* Adds terms and frequencies found in vector into the Map termFreqMap
*
* @param field2termFreqMap a Map of terms and their frequencies per field
* @param vector List of terms and their frequencies for a doc/field
*/
private void addTermFrequencies(
Map<String, Map<String, Int>> field2termFreqMap, Terms vector, String fieldName)
throws IOException {
Map<String, Int> termFreqMap = field2termFreqMap.get(fieldName);
if (termFreqMap == null) {
termFreqMap = new HashMap<>();
field2termFreqMap.put(fieldName, termFreqMap);
}
final TermsEnum termsEnum = vector.iterator();
final CharsRefBuilder spare = new CharsRefBuilder();
BytesRef text;
while ((text = termsEnum.next()) != null) {
spare.copyUTF8Bytes(text);
final String term = spare.toString();
if (isNoiseWord(term)) {
continue;
}
final int freq = (int) termsEnum.totalTermFreq();
// increment frequency
Int cnt = termFreqMap.get(term);
if (cnt == null) {
cnt = new Int();
termFreqMap.put(term, cnt);
cnt.x = freq;
} else {
cnt.x += freq;
}
}
}
/*
* listTermDictionary displays the term dictionary for a field.
*/
static void listTermDictionary(IndexReader reader, String fieldName) throws IOException {
System.out.println("\nTerm Dictionary: field " + fieldName);
/*
Grant says:
MultiFields.getTerms(IndexReader, fieldName)
*/
Terms terms = MultiFields.getTerms(reader, fieldName);
if ((terms == null) || (terms.size() == -1))
System.out.println(" The term dictionary is empty.");
else {
System.out.println(" Vocabulary size: " + terms.size() + " terms");
TermsEnum ithTerm = terms.iterator(null);
/*
* Iterate over the terms in this document.
* Information about a term's occurrences (tf and
* positions) is accessed via the indexing API, which
* returns inverted lists that describe (only) the
* current document.
*/
while (ithTerm.next() != null) {
System.out.format(
" %-30s %d %d\n",
ithTerm.term().utf8ToString(), ithTerm.docFreq(), ithTerm.totalTermFreq());
}
;
}
;
}
protected void compareTermVectors(Terms terms, Terms memTerms, String field_name)
throws IOException {
TermsEnum termEnum = terms.iterator();
TermsEnum memTermEnum = memTerms.iterator();
while (termEnum.next() != null) {
assertNotNull(memTermEnum.next());
assertThat(termEnum.totalTermFreq(), equalTo(memTermEnum.totalTermFreq()));
PostingsEnum docsPosEnum = termEnum.postings(null, PostingsEnum.POSITIONS);
PostingsEnum memDocsPosEnum = memTermEnum.postings(null, PostingsEnum.POSITIONS);
String currentTerm = termEnum.term().utf8ToString();
assertThat(
"Token mismatch for field: " + field_name,
currentTerm,
equalTo(memTermEnum.term().utf8ToString()));
docsPosEnum.nextDoc();
memDocsPosEnum.nextDoc();
int freq = docsPosEnum.freq();
assertThat(freq, equalTo(memDocsPosEnum.freq()));
for (int i = 0; i < freq; i++) {
String failDesc = " (field:" + field_name + " term:" + currentTerm + ")";
int memPos = memDocsPosEnum.nextPosition();
int pos = docsPosEnum.nextPosition();
assertThat("Position test failed" + failDesc, memPos, equalTo(pos));
assertThat(
"Start offset test failed" + failDesc,
memDocsPosEnum.startOffset(),
equalTo(docsPosEnum.startOffset()));
assertThat(
"End offset test failed" + failDesc,
memDocsPosEnum.endOffset(),
equalTo(docsPosEnum.endOffset()));
assertThat(
"Missing payload test failed" + failDesc,
docsPosEnum.getPayload(),
equalTo(docsPosEnum.getPayload()));
}
}
assertNull("Still some tokens not processed", memTermEnum.next());
}
/* Copied from lucene 4.2.x core */
private static long totalTermFreq(IndexReader r, String field, BytesRef text) throws IOException {
final Terms terms = MultiFields.getTerms(r, field);
if (terms != null) {
final TermsEnum termsEnum = terms.iterator(null);
if (termsEnum.seekExact(text, true)) {
return termsEnum.totalTermFreq();
}
}
return 0;
}
public void collectTermContext(
IndexReader reader,
List<LeafReaderContext> leaves,
TermContext[] contextArray,
Term[] queryTerms)
throws IOException {
TermsEnum termsEnum = null;
for (LeafReaderContext context : leaves) {
final Fields fields = context.reader().fields();
for (int i = 0; i < queryTerms.length; i++) {
Term term = queryTerms[i];
TermContext termContext = contextArray[i];
final Terms terms = fields.terms(term.field());
if (terms == null) {
// field does not exist
continue;
}
termsEnum = terms.iterator();
assert termsEnum != null;
if (termsEnum == TermsEnum.EMPTY) continue;
if (termsEnum.seekExact(term.bytes())) {
if (termContext == null) {
contextArray[i] =
new TermContext(
reader.getContext(),
termsEnum.termState(),
context.ord,
termsEnum.docFreq(),
termsEnum.totalTermFreq());
} else {
termContext.register(
termsEnum.termState(), context.ord, termsEnum.docFreq(), termsEnum.totalTermFreq());
}
}
}
}
}
@Override
public long totalTermFreq() throws IOException {
return termsEnum.totalTermFreq();
}
@Test
public void luceneNgramMetaCollectorTest() throws Exception {
File tmpDir = folder.newFolder();
CollectionReaderDescription reader =
CollectionReaderFactory.createReaderDescription(
TextReader.class,
TextReader.PARAM_SOURCE_LOCATION,
"src/test/resources/data/",
TextReader.PARAM_LANGUAGE,
"en",
TextReader.PARAM_PATTERNS,
"text*.txt");
AnalysisEngineDescription segmenter =
AnalysisEngineFactory.createEngineDescription(BreakIteratorSegmenter.class);
AnalysisEngineDescription metaCollector =
AnalysisEngineFactory.createEngineDescription(
LuceneNGramMetaCollector.class, LuceneNGramDFE.PARAM_LUCENE_DIR, tmpDir);
for (JCas jcas : new JCasIterable(reader, segmenter, metaCollector)) {
// System.out.println(jcas.getDocumentText().length());
}
int i = 0;
IndexReader index;
try {
index = DirectoryReader.open(FSDirectory.open(tmpDir));
Fields fields = MultiFields.getFields(index);
if (fields != null) {
Terms terms = fields.terms(LuceneNGramDFE.LUCENE_NGRAM_FIELD);
if (terms != null) {
TermsEnum termsEnum = terms.iterator(null);
// Bits liveDocs = MultiFields.getLiveDocs(index);
// DocsEnum docs = termsEnum.docs(liveDocs, null);
// int docId;
// while((docId = docs.nextDoc()) != DocsEnum.NO_MORE_DOCS) {
// index.g
// }
BytesRef text = null;
while ((text = termsEnum.next()) != null) {
// System.out.println(text.utf8ToString() + " - " +
// termsEnum.totalTermFreq());
// System.out.println(termsEnum.docFreq());
if (text.utf8ToString().equals("this")) {
assertEquals(2, termsEnum.docFreq());
assertEquals(3, termsEnum.totalTermFreq());
}
i++;
}
}
}
} catch (Exception e) {
throw new ResourceInitializationException(e);
}
assertEquals(35, i);
}
private void mapOneVector(
NamedList<Object> docNL,
FieldOptions fieldOptions,
IndexReader reader,
int docID,
TermsEnum termsEnum,
String field)
throws IOException {
NamedList<Object> fieldNL = new NamedList<Object>();
docNL.add(field, fieldNL);
BytesRef text;
DocsAndPositionsEnum dpEnum = null;
while ((text = termsEnum.next()) != null) {
String term = text.utf8ToString();
NamedList<Object> termInfo = new NamedList<Object>();
fieldNL.add(term, termInfo);
final int freq = (int) termsEnum.totalTermFreq();
if (fieldOptions.termFreq == true) {
termInfo.add("tf", freq);
}
dpEnum = termsEnum.docsAndPositions(null, dpEnum);
boolean useOffsets = false;
boolean usePositions = false;
if (dpEnum != null) {
dpEnum.nextDoc();
usePositions = fieldOptions.positions;
useOffsets = fieldOptions.offsets;
}
NamedList<Integer> positionsNL = null;
NamedList<Number> theOffsets = null;
if (usePositions || useOffsets) {
for (int i = 0; i < freq; i++) {
final int pos = dpEnum.nextPosition();
if (usePositions && pos >= 0) {
if (positionsNL == null) {
positionsNL = new NamedList<Integer>();
termInfo.add("positions", positionsNL);
}
positionsNL.add("position", pos);
}
if (useOffsets && theOffsets == null) {
if (dpEnum.startOffset() == -1) {
useOffsets = false;
} else {
theOffsets = new NamedList<Number>();
termInfo.add("offsets", theOffsets);
}
}
if (theOffsets != null) {
theOffsets.add("start", dpEnum.startOffset());
theOffsets.add("end", dpEnum.endOffset());
}
}
}
int df = 0;
if (fieldOptions.docFreq || fieldOptions.tfIdf) {
df = reader.docFreq(new Term(field, text));
}
if (fieldOptions.docFreq) {
termInfo.add("df", df);
}
// TODO: this is not TF/IDF by anyone's definition!
if (fieldOptions.tfIdf) {
double tfIdfVal = ((double) freq) / df;
termInfo.add("tf-idf", tfIdfVal);
}
}
}
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));
}
public SparseInstances readIndex(String indexPath, String destFile, int threshold)
throws Exception {
if (indexPath == null || destFile == null) {
System.out.println("error: indexPath or destFile is null\n");
return null;
}
DirectoryReader reader = DirectoryReader.open(FSDirectory.open(Paths.get(indexPath)));
Terms terms = SlowCompositeReaderWrapper.wrap(reader).terms(reviewKey);
int capacity = (int) terms.size();
HashMap<String, Integer> wordDict = new HashMap<>(capacity);
capacity = capacity > 65535 ? 65535 : capacity;
SparseInstances instData = new SparseInstances(capacity, reader.numDocs());
TermsEnum termsEnum = terms.iterator();
int index = 0;
BytesRef term = null;
String strTerm = null;
while ((term = termsEnum.next()) != null) {
strTerm = term.toString();
if (termsEnum.totalTermFreq() < threshold) {
continue;
}
if (strTerm.isEmpty()) {
continue;
}
if (wordDict.get(strTerm) != null) {
continue;
}
instData.addAttribute(strTerm);
index++;
}
int numAtt = instData.numAttributes();
int numInst = instData.numInstances();
Integer attIndex = null;
String id = null;
int termIndex = 0;
for (int docIndex = 0; docIndex < numInst; docIndex++) {
id = reader.document(docIndex).getField(idKey).stringValue();
Terms docTerms = reader.getTermVector(docIndex, reviewKey);
if (docTerms == null) {
continue;
}
int[] indices = new int[(int) docTerms.size()];
double[] attValues = new double[(int) docTerms.size()];
termsEnum = docTerms.iterator();
termIndex = 0;
while ((term = termsEnum.next()) != null) {
strTerm = term.toString();
attIndex = wordDict.get(strTerm);
if (attIndex == null) {
continue;
}
indices[termIndex] = attIndex.intValue();
attValues[termIndex] = termsEnum.totalTermFreq();
}
ESparseInstance instance = new ESparseInstance(id, 1.0, attValues, indices, numAtt);
instData.addInstance(instance);
}
return null;
}
/** checks term-level statistics */
public void assertTermStats(TermsEnum leftTermsEnum, TermsEnum rightTermsEnum) throws Exception {
assertEquals(leftTermsEnum.docFreq(), rightTermsEnum.docFreq());
if (leftTermsEnum.totalTermFreq() != -1 && rightTermsEnum.totalTermFreq() != -1) {
assertEquals(leftTermsEnum.totalTermFreq(), rightTermsEnum.totalTermFreq());
}
}
protected void validateResponse(
TermVectorResponse esResponse, Fields luceneFields, TestConfig testConfig)
throws IOException {
TestDoc testDoc = testConfig.doc;
HashSet<String> selectedFields =
testConfig.selectedFields == null
? null
: new HashSet<String>(Arrays.asList(testConfig.selectedFields));
Fields esTermVectorFields = esResponse.getFields();
for (TestFieldSetting field : testDoc.fieldSettings) {
Terms esTerms = esTermVectorFields.terms(field.name);
if (selectedFields != null && !selectedFields.contains(field.name)) {
assertNull(esTerms);
continue;
}
assertNotNull(esTerms);
Terms luceneTerms = luceneFields.terms(field.name);
TermsEnum esTermEnum = esTerms.iterator(null);
TermsEnum luceneTermEnum = luceneTerms.iterator(null);
while (esTermEnum.next() != null) {
assertNotNull(luceneTermEnum.next());
assertThat(esTermEnum.totalTermFreq(), equalTo(luceneTermEnum.totalTermFreq()));
DocsAndPositionsEnum esDocsPosEnum = esTermEnum.docsAndPositions(null, null, 0);
DocsAndPositionsEnum luceneDocsPosEnum = luceneTermEnum.docsAndPositions(null, null, 0);
if (luceneDocsPosEnum == null) {
// test we expect that...
assertFalse(field.storedOffset);
assertFalse(field.storedPayloads);
assertFalse(field.storedPositions);
continue;
}
String currentTerm = esTermEnum.term().utf8ToString();
assertThat(
"Token mismatch for field: " + field.name,
currentTerm,
equalTo(luceneTermEnum.term().utf8ToString()));
esDocsPosEnum.nextDoc();
luceneDocsPosEnum.nextDoc();
int freq = esDocsPosEnum.freq();
assertThat(freq, equalTo(luceneDocsPosEnum.freq()));
for (int i = 0; i < freq; i++) {
String failDesc = " (field:" + field.name + " term:" + currentTerm + ")";
int lucenePos = luceneDocsPosEnum.nextPosition();
int esPos = esDocsPosEnum.nextPosition();
if (field.storedPositions && testConfig.requestPositions) {
assertThat("Position test failed" + failDesc, lucenePos, equalTo(esPos));
} else {
assertThat("Missing position test failed" + failDesc, esPos, equalTo(-1));
}
if (field.storedOffset && testConfig.requestOffsets) {
assertThat(
"Offset test failed" + failDesc,
luceneDocsPosEnum.startOffset(),
equalTo(esDocsPosEnum.startOffset()));
assertThat(
"Offset test failed" + failDesc,
luceneDocsPosEnum.endOffset(),
equalTo(esDocsPosEnum.endOffset()));
} else {
assertThat(
"Missing offset test failed" + failDesc, esDocsPosEnum.startOffset(), equalTo(-1));
assertThat(
"Missing offset test failed" + failDesc, esDocsPosEnum.endOffset(), equalTo(-1));
}
if (field.storedPayloads && testConfig.requestPayloads) {
assertThat(
"Payload test failed" + failDesc,
luceneDocsPosEnum.getPayload(),
equalTo(esDocsPosEnum.getPayload()));
} else {
assertThat(
"Missing payload test failed" + failDesc,
esDocsPosEnum.getPayload(),
equalTo(null));
}
}
}
assertNull("Es returned terms are done but lucene isn't", luceneTermEnum.next());
}
}
/**
* Safe (but, slowish) default method to write every vector field in the document. This default
* implementation requires that the vectors implement both Fields.size and Terms.size.
*/
protected final void addAllDocVectors(Fields vectors, FieldInfos fieldInfos) throws IOException {
if (vectors == null) {
startDocument(0);
return;
}
final int numFields = vectors.size();
if (numFields == -1) {
throw new IllegalStateException("vectors.size() must be implemented (it returned -1)");
}
startDocument(numFields);
final FieldsEnum fieldsEnum = vectors.iterator();
String fieldName;
String lastFieldName = null;
while ((fieldName = fieldsEnum.next()) != null) {
final FieldInfo fieldInfo = fieldInfos.fieldInfo(fieldName);
assert lastFieldName == null || fieldName.compareTo(lastFieldName) > 0
: "lastFieldName=" + lastFieldName + " fieldName=" + fieldName;
lastFieldName = fieldName;
final Terms terms = fieldsEnum.terms();
if (terms == null) {
// FieldsEnum shouldn't lie...
continue;
}
final int numTerms = (int) terms.size();
if (numTerms == -1) {
throw new IllegalStateException("terms.size() must be implemented (it returned -1)");
}
final TermsEnum termsEnum = terms.iterator(null);
DocsAndPositionsEnum docsAndPositionsEnum = null;
boolean startedField = false;
// NOTE: this is tricky, because TermVectors allow
// indexing offsets but NOT positions. So we must
// lazily init the field by checking whether first
// position we see is -1 or not.
int termCount = 0;
while (termsEnum.next() != null) {
termCount++;
final int freq = (int) termsEnum.totalTermFreq();
if (startedField) {
startTerm(termsEnum.term(), freq);
}
// TODO: we need a "query" API where we can ask (via
// flex API) what this term was indexed with...
// Both positions & offsets:
docsAndPositionsEnum = termsEnum.docsAndPositions(null, null, true);
final boolean hasOffsets;
boolean hasPositions = false;
if (docsAndPositionsEnum == null) {
// Fallback: no offsets
docsAndPositionsEnum = termsEnum.docsAndPositions(null, null, false);
hasOffsets = false;
} else {
hasOffsets = true;
}
if (docsAndPositionsEnum != null) {
final int docID = docsAndPositionsEnum.nextDoc();
assert docID != DocIdSetIterator.NO_MORE_DOCS;
assert docsAndPositionsEnum.freq() == freq;
for (int posUpto = 0; posUpto < freq; posUpto++) {
final int pos = docsAndPositionsEnum.nextPosition();
if (!startedField) {
assert numTerms > 0;
hasPositions = pos != -1;
startField(fieldInfo, numTerms, hasPositions, hasOffsets);
startTerm(termsEnum.term(), freq);
startedField = true;
}
final int startOffset;
final int endOffset;
if (hasOffsets) {
startOffset = docsAndPositionsEnum.startOffset();
endOffset = docsAndPositionsEnum.endOffset();
assert startOffset != -1;
assert endOffset != -1;
} else {
startOffset = -1;
endOffset = -1;
}
assert !hasPositions || pos >= 0;
addPosition(pos, startOffset, endOffset);
}
} else {
if (!startedField) {
assert numTerms > 0;
startField(fieldInfo, numTerms, hasPositions, hasOffsets);
startTerm(termsEnum.term(), freq);
startedField = true;
}
}
}
assert termCount == numTerms;
}
}