mknn_dataset_loader.hpp

/*
 * Copyright (C) 2012-2015, Juan Manuel Barrios <juanmanuel@barrios.cl>
 * All rights reserved.
 *
 * This file is part of MetricKnn.
 * MetricKnn is made available under the terms of the BSD 2-Clause License.
 */

#ifndef MKNN_DATASET_LOADER_HPP_
#define MKNN_DATASET_LOADER_HPP_

#include "../metricknn_cpp.hpp"

namespace mknn {

class DatasetCustom {
public:

    virtual long long getNumObjects()= 0;
    virtual void *getObject(long long pos)= 0;

    virtual void pushObject(void *object)= 0;

    virtual ~DatasetCustom() = 0;

};

class DatasetConcatenate: public Dataset {
public:
    long long getNumSubDatasets();
    Dataset getSubDataset(long long num_subdataset);
    void getDatasetObject(long long posObject, long long *out_numSubdataset,
            long long *out_posObjectInSubdataset);

};

class DatasetMultiObject: public Dataset {
public:
    long long getNumSubDatasets();
    Dataset getSubDataset(long long num_subdataset);

};

class DatasetLoader {
public:

    static Dataset Custom(DatasetCustom *custom_dataset,
            bool delete_custom_dataset_on_dataset_release, Domain domain);

    static Dataset PointerArray(void **object_array, long long num_objects,
            Domain domain);

    static Dataset PointerCompactVectors(void *vectors_header,
            long long num_vectors, long long vector_num_dimensions,
            const std::string vector_dimension_datatype);

    static Dataset ParseVectorFile(std::string filename,
            const std::string datatype);

    static Dataset ParseStringsFile(std::string filename);

    static DatasetConcatenate Concatenate(
            const std::vector<Dataset> &subdatasets);

    static Dataset SubsetSegment(Dataset &superdataset, long long position_start,
            long long length);

    static Dataset SubsetPositions(Dataset &superdataset, long long *positions,
            long long num_positions);
    static Dataset UniformRandomVectors(long long num_objects, long long dimension,
            double dimension_minValueIncluded,
            double dimension_maxValueNotIncluded, const std::string datatype);

    static DatasetMultiObject MultiObject(
            const std::vector<Dataset> &subdatasets);

    static Dataset Empty(Domain domain);

};

}
#endif