TermVectorsReader.h

Go to the documentation of this file.

// Copyright (c) 2009-2014 Alan Wright. All rights reserved.
// Distributable under the terms of either the Apache License (Version 2.0)
// or the GNU Lesser General Public License.
 
#ifndef TERMVECTORSREADER_H
#define TERMVECTORSREADER_H
 
#include "TermVectorMapper.h"
 
namespace Lucene {
 
class LPPAPI TermVectorsReader : public LuceneObject {
public:
    TermVectorsReader();
    TermVectorsReader(const DirectoryPtr& d, const String& segment, const FieldInfosPtr& fieldInfos);
    TermVectorsReader(const DirectoryPtr& d, const String& segment, const FieldInfosPtr& fieldInfos,
                      int32_t readBufferSize, int32_t docStoreOffset = -1, int32_t size = 0);
    virtual ~TermVectorsReader();
 
    LUCENE_CLASS(TermVectorsReader);
 
public:
    static const int32_t FORMAT_VERSION;

    static const int32_t FORMAT_VERSION2;

    static const int32_t FORMAT_UTF8_LENGTH_IN_BYTES;

    static const int32_t FORMAT_CURRENT;

    static const int32_t FORMAT_SIZE;
 
    static const uint8_t STORE_POSITIONS_WITH_TERMVECTOR;
    static const uint8_t STORE_OFFSET_WITH_TERMVECTOR;
 
protected:
    FieldInfosPtr fieldInfos;
 
    IndexInputPtr tvx;
    IndexInputPtr tvd;
    IndexInputPtr tvf;
    int32_t _size;
    int32_t numTotalDocs;

    int32_t docStoreOffset;
 
    int32_t format;
 
public:
    IndexInputPtr getTvdStream();

    IndexInputPtr getTvfStream();
 
    bool canReadRawDocs();

    void rawDocs(Collection<int32_t> tvdLengths, Collection<int32_t> tvfLengths, int32_t startDocID, int32_t numDocs);
 
    void close();

    int32_t size();
 
    void get(int32_t docNum, const String& field, const TermVectorMapperPtr& mapper);

    TermFreqVectorPtr get(int32_t docNum, const String& field);

    Collection<TermFreqVectorPtr> get(int32_t docNum);
 
    void get(int32_t docNumber, const TermVectorMapperPtr& mapper);
 
    virtual LuceneObjectPtr clone(const LuceneObjectPtr& other = LuceneObjectPtr());
 
protected:
    void ConstructReader(const DirectoryPtr& d, const String& segment, const FieldInfosPtr& fieldInfos, int32_t readBufferSize, int32_t docStoreOffset, int32_t size);
 
    void seekTvx(int32_t docNum);
 
    int32_t checkValidFormat(const IndexInputPtr& in);

    Collection<String> readFields(int32_t fieldCount);

    Collection<int64_t> readTvfPointers(int32_t fieldCount);
 
    Collection<TermFreqVectorPtr> readTermVectors(int32_t docNum, Collection<String> fields, Collection<int64_t> tvfPointers);
    void readTermVectors(Collection<String> fields, Collection<int64_t> tvfPointers, const TermVectorMapperPtr& mapper);

    void readTermVector(const String& field, int64_t tvfPointer, const TermVectorMapperPtr& mapper);
};

class ParallelArrayTermVectorMapper : public TermVectorMapper {
public:
    ParallelArrayTermVectorMapper();
    virtual ~ParallelArrayTermVectorMapper();
 
    LUCENE_CLASS(ParallelArrayTermVectorMapper);
 
protected:
    Collection<String> terms;
    Collection<int32_t> termFreqs;
    Collection< Collection<int32_t> > positions;
    Collection< Collection<TermVectorOffsetInfoPtr> > offsets;
    int32_t currentPosition;
    bool storingOffsets;
    bool storingPositions;
    String field;
 
public:
    virtual void setExpectations(const String& field, int32_t numTerms, bool storeOffsets, bool storePositions);

    virtual void map(const String& term, int32_t frequency, Collection<TermVectorOffsetInfoPtr> offsets, Collection<int32_t> positions);

    TermFreqVectorPtr materializeVector();
};
 
}
 
#endif