public FreqProxTermsWriterPerField(TermsHashPerField termsHashPerField, FreqProxTermsWriterPerThread perThread, FieldInfo fieldInfo) { this.termsHashPerField = termsHashPerField; this.perThread = perThread; this.fieldInfo = fieldInfo; docState = termsHashPerField.docState; fieldState = termsHashPerField.fieldState; omitTermFreqAndPositions = fieldInfo.omitTermFreqAndPositions; }
// TODO: would be nice to factor out more of this, eg the // FreqProxFieldMergeState, and code to visit all Fields // under the same FieldInfo together, up into TermsHash*. // Other writers would presumably share alot of this... public override void Flush(System.Collections.IDictionary threadsAndFields, SegmentWriteState state) { // Gather all FieldData's that have postings, across all // ThreadStates System.Collections.ArrayList allFields = new System.Collections.ArrayList(); System.Collections.IEnumerator it = new System.Collections.Hashtable(threadsAndFields).GetEnumerator(); while (it.MoveNext()) { System.Collections.DictionaryEntry entry = (System.Collections.DictionaryEntry)it.Current; System.Collections.ICollection fields = (System.Collections.ICollection)entry.Value; System.Collections.IEnumerator fieldsIt = fields.GetEnumerator(); while (fieldsIt.MoveNext()) { FreqProxTermsWriterPerField perField = (FreqProxTermsWriterPerField)((System.Collections.DictionaryEntry)fieldsIt.Current).Key; if (perField.termsHashPerField.numPostings > 0) { allFields.Add(perField); } } } // Sort by field name allFields.Sort(); int numAllFields = allFields.Count; // TODO: allow Lucene user to customize this consumer: FormatPostingsFieldsConsumer consumer = new FormatPostingsFieldsWriter(state, fieldInfos); /* * Current writer chain: * FormatPostingsFieldsConsumer * -> IMPL: FormatPostingsFieldsWriter * -> FormatPostingsTermsConsumer * -> IMPL: FormatPostingsTermsWriter * -> FormatPostingsDocConsumer * -> IMPL: FormatPostingsDocWriter * -> FormatPostingsPositionsConsumer * -> IMPL: FormatPostingsPositionsWriter */ int start = 0; while (start < numAllFields) { FieldInfo fieldInfo = ((FreqProxTermsWriterPerField)allFields[start]).fieldInfo; System.String fieldName = fieldInfo.name; int end = start + 1; while (end < numAllFields && ((FreqProxTermsWriterPerField)allFields[end]).fieldInfo.name.Equals(fieldName)) { end++; } FreqProxTermsWriterPerField[] fields = new FreqProxTermsWriterPerField[end - start]; for (int i = start; i < end; i++) { fields[i - start] = (FreqProxTermsWriterPerField)allFields[i]; // Aggregate the storePayload as seen by the same // field across multiple threads fieldInfo.storePayloads |= fields[i - start].hasPayloads; } // If this field has postings then add them to the // segment AppendPostings(fields, consumer); for (int i = 0; i < fields.Length; i++) { TermsHashPerField perField = fields[i].termsHashPerField; int numPostings = perField.numPostings; perField.Reset(); perField.ShrinkHash(numPostings); fields[i].Reset(); } start = end; } it = new System.Collections.Hashtable(threadsAndFields).GetEnumerator(); while (it.MoveNext()) { System.Collections.DictionaryEntry entry = (System.Collections.DictionaryEntry)it.Current; FreqProxTermsWriterPerThread perThread = (FreqProxTermsWriterPerThread)entry.Key; perThread.termsHashPerThread.Reset(true); } consumer.Finish(); }
// TODO: would be nice to factor out more of this, eg the // FreqProxFieldMergeState, and code to visit all Fields // under the same FieldInfo together, up into TermsHash*. // Other writers would presumably share alot of this... public override void Flush(Support.Dictionary <TermsHashConsumerPerThread, IList <TermsHashConsumerPerField> > threadsAndFields, SegmentWriteState state) { // Gather all FieldData's that have postings, across all // ThreadStates List <FreqProxTermsWriterPerField> allFields = new List <FreqProxTermsWriterPerField>(); foreach (KeyValuePair <TermsHashConsumerPerThread, IList <TermsHashConsumerPerField> > entry in threadsAndFields) { IList <TermsHashConsumerPerField> fields = entry.Value; foreach (TermsHashConsumerPerField i in fields) { FreqProxTermsWriterPerField perField = (FreqProxTermsWriterPerField)i; if (perField.termsHashPerField.numPostings > 0) { allFields.Add(perField); } } } // Sort by field name allFields.Sort(); int numAllFields = allFields.Count; // TODO: allow Lucene user to customize this consumer: FormatPostingsFieldsConsumer consumer = new FormatPostingsFieldsWriter(state, fieldInfos); /* * Current writer chain: * FormatPostingsFieldsConsumer * -> IMPL: FormatPostingsFieldsWriter * -> FormatPostingsTermsConsumer * -> IMPL: FormatPostingsTermsWriter * -> FormatPostingsDocConsumer * -> IMPL: FormatPostingsDocWriter * -> FormatPostingsPositionsConsumer * -> IMPL: FormatPostingsPositionsWriter */ int start = 0; while (start < numAllFields) { FieldInfo fieldInfo = ((FreqProxTermsWriterPerField)allFields[start]).fieldInfo; System.String fieldName = fieldInfo.name; int end = start + 1; while (end < numAllFields && ((FreqProxTermsWriterPerField)allFields[end]).fieldInfo.name.Equals(fieldName)) { end++; } FreqProxTermsWriterPerField[] fields = new FreqProxTermsWriterPerField[end - start]; for (int i = start; i < end; i++) { fields[i - start] = (FreqProxTermsWriterPerField)allFields[i]; // Aggregate the storePayload as seen by the same // field across multiple threads fieldInfo.storePayloads |= fields[i - start].hasPayloads; } // If this field has postings then add them to the // segment AppendPostings(fields, consumer); for (int i = 0; i < fields.Length; i++) { TermsHashPerField perField = fields[i].termsHashPerField; int numPostings = perField.numPostings; perField.Reset(); perField.ShrinkHash(numPostings); fields[i].Reset(); } start = end; } foreach (KeyValuePair <TermsHashConsumerPerThread, IList <TermsHashConsumerPerField> > entry in threadsAndFields) { FreqProxTermsWriterPerThread perThread = (FreqProxTermsWriterPerThread)entry.Key; perThread.termsHashPerThread.Reset(true); } consumer.Finish(); }
// TODO: would be nice to factor out morme of this, eg the // FreqProxFieldMergeState, and code to visit all Fields // under the same FieldInfo together, up into TermsHash*. // Other writers would presumably share alot of this... internal override void flush(IDictionary <object, object> threadsAndFields, DocumentsWriter.FlushState state) { // Gather all FieldData's that have postings, across all // ThreadStates List <object> allFields = new List <object>(); IEnumerator <KeyValuePair <object, object> > it = threadsAndFields.GetEnumerator(); while (it.MoveNext()) { KeyValuePair <object, object> entry = (KeyValuePair <object, object>)it.Current; ICollection <object> fields = (ICollection <object>)entry.Value; IEnumerator <object> fieldsIt = fields.GetEnumerator(); while (fieldsIt.MoveNext()) { FreqProxTermsWriterPerField perField = (FreqProxTermsWriterPerField)fieldsIt.Current; if (perField.termsHashPerField.numPostings > 0) { allFields.Add(perField); } } } // Sort by field name allFields.Sort(); int numAllFields = allFields.Count; TermInfosWriter termsOut = new TermInfosWriter(state.directory, state.segmentName, fieldInfos, state.docWriter.writer.GetTermIndexInterval()); IndexOutput freqOut = state.directory.CreateOutput(state.SegmentFileName(IndexFileNames.FREQ_EXTENSION)); IndexOutput proxOut; if (fieldInfos.HasProx()) { proxOut = state.directory.CreateOutput(state.SegmentFileName(IndexFileNames.PROX_EXTENSION)); } else { proxOut = null; } DefaultSkipListWriter skipListWriter = new DefaultSkipListWriter(termsOut.skipInterval, termsOut.maxSkipLevels, state.numDocsInRAM, freqOut, proxOut); int start = 0; while (start < numAllFields) { FieldInfo fieldInfo = ((FreqProxTermsWriterPerField)allFields[start]).fieldInfo; string fieldName = fieldInfo.name; int end = start + 1; while (end < numAllFields && ((FreqProxTermsWriterPerField)allFields[end]).fieldInfo.name.Equals(fieldName)) { end++; } FreqProxTermsWriterPerField[] fields = new FreqProxTermsWriterPerField[end - start]; for (int i = start; i < end; i++) { fields[i - start] = (FreqProxTermsWriterPerField)allFields[i]; // Aggregate the storePayload as seen by the same // field across multiple threads fieldInfo.storePayloads |= fields[i - start].hasPayloads; } // If this field has postings then add them to the // segment AppendPostings(state, fields, termsOut, freqOut, proxOut, skipListWriter); for (int i = 0; i < fields.Length; i++) { TermsHashPerField perField = fields[i].termsHashPerField; int numPostings = perField.numPostings; perField.reset(); perField.shrinkHash(numPostings); fields[i].reset(); } start = end; } it = threadsAndFields.GetEnumerator(); while (it.MoveNext()) { KeyValuePair <object, object> entry = (KeyValuePair <object, object>)it.Current; FreqProxTermsWriterPerThread perThread = (FreqProxTermsWriterPerThread)entry.Key; perThread.termsHashPerThread.reset(true); } freqOut.Close(); if (proxOut != null) { state.flushedFiles[state.SegmentFileName(IndexFileNames.PROX_EXTENSION)] = state.SegmentFileName(IndexFileNames.PROX_EXTENSION); proxOut.Close(); } termsOut.Close(); // Record all files we have flushed state.flushedFiles[state.SegmentFileName(IndexFileNames.FIELD_INFOS_EXTENSION)] = state.SegmentFileName(IndexFileNames.FIELD_INFOS_EXTENSION); state.flushedFiles[state.SegmentFileName(IndexFileNames.FREQ_EXTENSION)] = state.SegmentFileName(IndexFileNames.FREQ_EXTENSION); state.flushedFiles[state.SegmentFileName(IndexFileNames.TERMS_EXTENSION)] = state.SegmentFileName(IndexFileNames.TERMS_EXTENSION); state.flushedFiles[state.SegmentFileName(IndexFileNames.TERMS_INDEX_EXTENSION)] = state.SegmentFileName(IndexFileNames.TERMS_INDEX_EXTENSION); }