List of usage examples for org.apache.lucene.index PostingsEnum startOffset
public abstract int startOffset() throws IOException;
From source file:com.github.flaxsearch.api.PositionData.java
License:Apache License
public PositionData(PostingsEnum pe) throws IOException { this.position = pe.nextPosition(); this.startOffset = pe.startOffset(); this.endOffset = pe.endOffset(); this.payload = payloadToString(pe.getPayload()); }
From source file:nl.inl.blacklab.search.SearcherImpl.java
License:Apache License
@Override public void getCharacterOffsets(int doc, String fieldName, int[] startsOfWords, int[] endsOfWords, boolean fillInDefaultsIfNotFound) { if (startsOfWords.length == 0) return; // nothing to do try {/*from ww w . j ava 2 s .c o m*/ // Determine lowest and highest word position we'd like to know something about. // This saves a little bit of time for large result sets. int minP = -1, maxP = -1; int numStarts = startsOfWords.length; int numEnds = endsOfWords.length; for (int i = 0; i < numStarts; i++) { if (startsOfWords[i] < minP || minP == -1) minP = startsOfWords[i]; if (startsOfWords[i] > maxP) maxP = startsOfWords[i]; } for (int i = 0; i < numEnds; i++) { if (endsOfWords[i] < minP || minP == -1) minP = endsOfWords[i]; if (endsOfWords[i] > maxP) maxP = endsOfWords[i]; } if (minP < 0 || maxP < 0) throw new RuntimeException("Can't determine min and max positions"); String fieldPropName = ComplexFieldUtil.mainPropertyOffsetsField(indexStructure, fieldName); org.apache.lucene.index.Terms terms = reader.getTermVector(doc, fieldPropName); if (terms == null) throw new IllegalArgumentException( "Field " + fieldPropName + " in doc " + doc + " has no term vector"); if (!terms.hasPositions()) throw new IllegalArgumentException( "Field " + fieldPropName + " in doc " + doc + " has no character postion information"); //int lowestPos = -1, highestPos = -1; int lowestPosFirstChar = -1, highestPosLastChar = -1; int total = numStarts + numEnds; boolean[] done = new boolean[total]; // NOTE: array is automatically initialized to zeroes! int found = 0; // Iterate over terms TermsEnum termsEnum = terms.iterator(); while (termsEnum.next() != null) { PostingsEnum dpe = termsEnum.postings(null, null, PostingsEnum.POSITIONS); // Iterate over docs containing this term (NOTE: should be only one doc!) while (dpe.nextDoc() != DocIdSetIterator.NO_MORE_DOCS) { // Iterate over positions of this term in this doc int positionsRead = 0; int numberOfPositions = dpe.freq(); while (positionsRead < numberOfPositions) { int position = dpe.nextPosition(); if (position == -1) break; positionsRead++; // Keep track of the lowest and highest char pos, so // we can fill in the character positions we didn't find int startOffset = dpe.startOffset(); if (startOffset < lowestPosFirstChar || lowestPosFirstChar == -1) { lowestPosFirstChar = startOffset; } int endOffset = dpe.endOffset(); if (endOffset > highestPosLastChar) { highestPosLastChar = endOffset; } // We've calculated the min and max word positions in advance, so // we know we can skip this position if it's outside the range we're interested in. // (Saves a little time for large result sets) if (position < minP || position > maxP) { continue; } for (int m = 0; m < numStarts; m++) { if (!done[m] && position == startsOfWords[m]) { done[m] = true; startsOfWords[m] = startOffset; found++; } } for (int m = 0; m < numEnds; m++) { if (!done[numStarts + m] && position == endsOfWords[m]) { done[numStarts + m] = true; endsOfWords[m] = endOffset; found++; } } // NOTE: we might be tempted to break here if found == total, // but that would foul up our calculation of highestPosLastChar and // lowestPosFirstChar. } } } if (found < total) { if (!fillInDefaultsIfNotFound) throw new RuntimeException("Could not find all character offsets!"); if (lowestPosFirstChar < 0 || highestPosLastChar < 0) throw new RuntimeException("Could not find default char positions!"); for (int m = 0; m < numStarts; m++) { if (!done[m]) startsOfWords[m] = lowestPosFirstChar; } for (int m = 0; m < numEnds; m++) { if (!done[numStarts + m]) endsOfWords[m] = highestPosLastChar; } } } catch (IOException e) { throw ExUtil.wrapRuntimeException(e); } }
From source file:org.elasticsearch.action.termvectors.AbstractTermVectorsTestCase.java
License:Apache License
protected void validateResponse(TermVectorsResponse esResponse, Fields luceneFields, TestConfig testConfig) throws IOException { assertThat(esResponse.getIndex(), equalTo(testConfig.doc.index)); TestDoc testDoc = testConfig.doc;//from w ww . j a v a 2s. c om HashSet<String> selectedFields = testConfig.selectedFields == null ? null : new HashSet<>(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(); TermsEnum luceneTermEnum = luceneTerms.iterator(); while (esTermEnum.next() != null) { assertNotNull(luceneTermEnum.next()); assertThat(esTermEnum.totalTermFreq(), equalTo(luceneTermEnum.totalTermFreq())); PostingsEnum esDocsPosEnum = esTermEnum.postings(null, PostingsEnum.POSITIONS); PostingsEnum luceneDocsPosEnum = luceneTermEnum.postings(null, PostingsEnum.POSITIONS); 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()); } }
From source file:org.elasticsearch.action.termvectors.AbstractTermVectorsTests.java
License:Apache License
protected void validateResponse(TermVectorsResponse esResponse, Fields luceneFields, TestConfig testConfig) throws IOException { assertThat(esResponse.getIndex(), equalTo(testConfig.doc.index)); TestDoc testDoc = testConfig.doc;/*from w w w .jav a 2 s . com*/ HashSet<String> selectedFields = testConfig.selectedFields == null ? null : new HashSet<>(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(); TermsEnum luceneTermEnum = luceneTerms.iterator(); while (esTermEnum.next() != null) { assertNotNull(luceneTermEnum.next()); assertThat(esTermEnum.totalTermFreq(), equalTo(luceneTermEnum.totalTermFreq())); PostingsEnum esDocsPosEnum = esTermEnum.postings(null, null, PostingsEnum.POSITIONS); PostingsEnum luceneDocsPosEnum = luceneTermEnum.postings(null, null, PostingsEnum.POSITIONS); 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()); } }
From source file:org.elasticsearch.action.termvectors.GetTermVectorsCheckDocFreqIT.java
License:Apache License
private void checkWithoutFieldStatistics(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).setTermStatistics(true) .setFieldStatistics(false).setSelectedFields(); 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) -1)); assertThat(terms.getDocCount(), Matchers.equalTo(-1)); assertThat(terms.getSumDocFreq(), equalTo((long) -1)); 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 {//from ww w. j ava 2s . c o m assertThat("expected ttf of " + string, numDocs, equalTo((int) iterator.totalTermFreq())); } PostingsEnum docsAndPositions = iterator.postings(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, null); xBuilder.endObject(); String utf8 = xBuilder.bytes().toUtf8().replaceFirst("\"took\":\\d+,", ""); ; String expectedString = "{\"_index\":\"test\",\"_type\":\"type1\",\"_id\":\"" + i + "\",\"_version\":1,\"found\":true,\"term_vectors\":{\"field\":{\"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)); }
From source file:org.elasticsearch.action.termvectors.GetTermVectorsCheckDocFreqIT.java
License:Apache License
private void checkWithoutTermStatistics(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).setTermStatistics(false) .setFieldStatistics(true).setSelectedFields(); assertThat(resp.request().termStatistics(), equalTo(false)); 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()); assertThat("expected ttf of " + string, -1, equalTo((int) iterator.totalTermFreq())); PostingsEnum docsAndPositions = iterator.postings(null, PostingsEnum.ALL); assertThat(docsAndPositions.nextDoc(), equalTo(0)); assertThat(freq[j], equalTo(docsAndPositions.freq())); assertThat(iterator.docFreq(), equalTo(-1)); 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"))); }//www .j av a2 s.c o m } assertThat(iterator.next(), Matchers.nullValue()); XContentBuilder xBuilder = XContentFactory.jsonBuilder(); xBuilder.startObject(); response.toXContent(xBuilder, null); xBuilder.endObject(); String utf8 = xBuilder.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\":{\"term_freq\":1,\"tokens\":[{\"position\":2,\"start_offset\":10,\"end_offset\":15,\"payload\":\"d29yZA==\"}]},\"dog\":{\"term_freq\":1,\"tokens\":[{\"position\":8,\"start_offset\":40,\"end_offset\":43,\"payload\":\"d29yZA==\"}]},\"fox\":{\"term_freq\":1,\"tokens\":[{\"position\":3,\"start_offset\":16,\"end_offset\":19,\"payload\":\"d29yZA==\"}]},\"jumps\":{\"term_freq\":1,\"tokens\":[{\"position\":4,\"start_offset\":20,\"end_offset\":25,\"payload\":\"d29yZA==\"}]},\"lazy\":{\"term_freq\":1,\"tokens\":[{\"position\":7,\"start_offset\":35,\"end_offset\":39,\"payload\":\"d29yZA==\"}]},\"over\":{\"term_freq\":1,\"tokens\":[{\"position\":5,\"start_offset\":26,\"end_offset\":30,\"payload\":\"d29yZA==\"}]},\"quick\":{\"term_freq\":1,\"tokens\":[{\"position\":1,\"start_offset\":4,\"end_offset\":9,\"payload\":\"d29yZA==\"}]},\"the\":{\"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)); }
From source file:org.elasticsearch.action.termvectors.GetTermVectorsCheckDocFreqIT.java
License:Apache License
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 {/*from www . j ava 2 s .c om*/ assertThat("expected ttf of " + string, numDocs, equalTo((int) iterator.totalTermFreq())); } PostingsEnum docsAndPositions = iterator.postings(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(); String utf8 = xBuilder.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)); }
From source file:org.elasticsearch.action.termvectors.GetTermVectorsCheckDocFreqTests.java
License:Apache License
private void checkWithoutFieldStatistics(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).setTermStatistics(true) .setFieldStatistics(false).setSelectedFields(); 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) -1)); assertThat(terms.getDocCount(), Matchers.equalTo(-1)); assertThat(terms.getSumDocFreq(), equalTo((long) -1)); 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 {/* w w w .j a va 2 s.com*/ 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, null); 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\":{\"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)); }
From source file:org.elasticsearch.action.termvectors.GetTermVectorsCheckDocFreqTests.java
License:Apache License
private void checkWithoutTermStatistics(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).setTermStatistics(false) .setFieldStatistics(true).setSelectedFields(); assertThat(resp.request().termStatistics(), equalTo(false)); 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()); assertThat("expected ttf of " + string, -1, 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(-1)); 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"))); }//from w w w .j a v a 2 s. c o m } assertThat(iterator.next(), Matchers.nullValue()); XContentBuilder xBuilder = XContentFactory.jsonBuilder(); xBuilder.startObject(); response.toXContent(xBuilder, null); 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\":{\"term_freq\":1,\"tokens\":[{\"position\":2,\"start_offset\":10,\"end_offset\":15,\"payload\":\"d29yZA==\"}]},\"dog\":{\"term_freq\":1,\"tokens\":[{\"position\":8,\"start_offset\":40,\"end_offset\":43,\"payload\":\"d29yZA==\"}]},\"fox\":{\"term_freq\":1,\"tokens\":[{\"position\":3,\"start_offset\":16,\"end_offset\":19,\"payload\":\"d29yZA==\"}]},\"jumps\":{\"term_freq\":1,\"tokens\":[{\"position\":4,\"start_offset\":20,\"end_offset\":25,\"payload\":\"d29yZA==\"}]},\"lazy\":{\"term_freq\":1,\"tokens\":[{\"position\":7,\"start_offset\":35,\"end_offset\":39,\"payload\":\"d29yZA==\"}]},\"over\":{\"term_freq\":1,\"tokens\":[{\"position\":5,\"start_offset\":26,\"end_offset\":30,\"payload\":\"d29yZA==\"}]},\"quick\":{\"term_freq\":1,\"tokens\":[{\"position\":1,\"start_offset\":4,\"end_offset\":9,\"payload\":\"d29yZA==\"}]},\"the\":{\"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)); }
From source file:org.elasticsearch.action.termvectors.GetTermVectorsCheckDocFreqTests.java
License:Apache License
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 {/* w w w .j a v a 2s . c om*/ 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)); }