SHOGUN: BinaryFile.h Source File

Go to the documentation of this file.
00001 /*
00002  * This program is free software; you can redistribute it and/or modify
00003  * it under the terms of the GNU General Public License as published by
00004  * the Free Software Foundation; either version 3 of the License, or
00005  * (at your option) any later version.
00006  *
00007  * Written (W) 2010 Soeren Sonnenburg
00008  * Copyright (C) 2010 Berlin Institute of Technology
00009  */
00010 #ifndef __BINARY_FILE_H__
00011 #define __BINARY_FILE_H__
00012 
00013 #include <shogun/lib/config.h>
00014 #include <shogun/lib/common.h>
00015 #include <shogun/base/SGObject.h>
00016 #include <shogun/lib/io.h>
00017 #include <shogun/lib/SimpleFile.h>
00018 
00019 namespace shogun
00020 {
00027 class CBinaryFile: public CFile
00028 {
00029 public:
00031     CBinaryFile(void);
00032 
00038     CBinaryFile(FILE* f, const char* name=NULL);
00039 
00046     CBinaryFile(char* fname, char rw='r', const char* name=NULL);
00047 
00049     virtual ~CBinaryFile();
00050 
00058     virtual void get_byte_vector(uint8_t*& vector, int32_t& len);
00059     virtual void get_char_vector(char*& vector, int32_t& len);
00060     virtual void get_int_vector(int32_t*& vector, int32_t& len);
00061     virtual void get_real_vector(float64_t*& vector, int32_t& len);
00062     virtual void get_shortreal_vector(float32_t*& vector, int32_t& len);
00063     virtual void get_short_vector(int16_t*& vector, int32_t& len);
00064     virtual void get_word_vector(uint16_t*& vector, int32_t& len);
00066 
00075     virtual void get_byte_matrix(
00076             uint8_t*& matrix, int32_t& num_feat, int32_t& num_vec);
00077     virtual void get_int8_matrix(
00078             int8_t*& matrix, int32_t& num_feat, int32_t& num_vec);
00079     virtual void get_char_matrix(
00080             char*& matrix, int32_t& num_feat, int32_t& num_vec);
00081     virtual void get_int_matrix(
00082             int32_t*& matrix, int32_t& num_feat, int32_t& num_vec);
00083     virtual void get_uint_matrix(
00084             uint32_t*& matrix, int32_t& num_feat, int32_t& num_vec);
00085     virtual void get_long_matrix(
00086             int64_t*& matrix, int32_t& num_feat, int32_t& num_vec);
00087     virtual void get_ulong_matrix(
00088             uint64_t*& matrix, int32_t& num_feat, int32_t& num_vec);
00089     virtual void get_shortreal_matrix(
00090             float32_t*& matrix, int32_t& num_feat, int32_t& num_vec);
00091     virtual void get_real_matrix(
00092             float64_t*& matrix, int32_t& num_feat, int32_t& num_vec);
00093     virtual void get_longreal_matrix(
00094             floatmax_t*& matrix, int32_t& num_feat, int32_t& num_vec);
00095     virtual void get_short_matrix(
00096             int16_t*& matrix, int32_t& num_feat, int32_t& num_vec);
00097     virtual void get_word_matrix(
00098             uint16_t*& matrix, int32_t& num_feat, int32_t& num_vec);
00100 
00109     virtual void get_byte_ndarray(
00110             uint8_t*& array, int32_t*& dims, int32_t& num_dims);
00111     virtual void get_char_ndarray(
00112             char*& array, int32_t*& dims, int32_t& num_dims);
00113     virtual void get_int_ndarray(
00114             int32_t*& array, int32_t*& dims, int32_t& num_dims);
00115     virtual void get_shortreal_ndarray(
00116             float32_t*& array, int32_t*& dims, int32_t& num_dims);
00117     virtual void get_real_ndarray(
00118             float64_t*& array, int32_t*& dims, int32_t& num_dims);
00119     virtual void get_short_ndarray(
00120             int16_t*& array, int32_t*& dims, int32_t& num_dims);
00121     virtual void get_word_ndarray(
00122             uint16_t*& array, int32_t*& dims, int32_t& num_dims);
00124 
00133     virtual void get_bool_sparsematrix(
00134             TSparse<bool>*& matrix, int32_t& num_feat, int32_t& num_vec);
00135     virtual void get_byte_sparsematrix(
00136             TSparse<uint8_t>*& matrix, int32_t& num_feat, int32_t& num_vec);
00137     virtual void get_int8_sparsematrix(
00138             TSparse<int8_t>*& matrix, int32_t& num_feat, int32_t& num_vec);
00139     virtual void get_char_sparsematrix(
00140             TSparse<char>*& matrix, int32_t& num_feat, int32_t& num_vec);
00141     virtual void get_int_sparsematrix(
00142             TSparse<int32_t>*& matrix, int32_t& num_feat, int32_t& num_vec);
00143     virtual void get_uint_sparsematrix(
00144             TSparse<uint32_t>*& matrix, int32_t& num_feat, int32_t& num_vec);
00145     virtual void get_long_sparsematrix(
00146             TSparse<int64_t>*& matrix, int32_t& num_feat, int32_t& num_vec);
00147     virtual void get_ulong_sparsematrix(
00148             TSparse<uint64_t>*& matrix, int32_t& num_feat, int32_t& num_vec);
00149     virtual void get_short_sparsematrix(
00150             TSparse<int16_t>*& matrix, int32_t& num_feat, int32_t& num_vec);
00151     virtual void get_word_sparsematrix(
00152             TSparse<uint16_t>*& matrix, int32_t& num_feat, int32_t& num_vec);
00153     virtual void get_shortreal_sparsematrix(
00154             TSparse<float32_t>*& matrix, int32_t& num_feat, int32_t& num_vec);
00155     virtual void get_real_sparsematrix(
00156             TSparse<float64_t>*& matrix, int32_t& num_feat, int32_t& num_vec);
00157     virtual void get_longreal_sparsematrix(
00158             TSparse<floatmax_t>*& matrix, int32_t& num_feat, int32_t& num_vec);
00160 
00161 
00170     virtual void get_byte_string_list(
00171             TString<uint8_t>*& strings, int32_t& num_str,
00172             int32_t& max_string_len);
00173     virtual void get_int8_string_list(
00174             TString<int8_t>*& strings, int32_t& num_str,
00175             int32_t& max_string_len);
00176     virtual void get_char_string_list(
00177             TString<char>*& strings, int32_t& num_str,
00178             int32_t& max_string_len);
00179     virtual void get_int_string_list(
00180             TString<int32_t>*& strings, int32_t& num_str,
00181             int32_t& max_string_len);
00182     virtual void get_uint_string_list(
00183             TString<uint32_t>*& strings, int32_t& num_str,
00184             int32_t& max_string_len);
00185     virtual void get_short_string_list(
00186             TString<int16_t>*& strings, int32_t& num_str,
00187             int32_t& max_string_len);
00188     virtual void get_word_string_list(
00189             TString<uint16_t>*& strings, int32_t& num_str,
00190             int32_t& max_string_len);
00191     virtual void get_long_string_list(
00192             TString<int64_t>*& strings, int32_t& num_str,
00193             int32_t& max_string_len);
00194     virtual void get_ulong_string_list(
00195             TString<uint64_t>*& strings, int32_t& num_str,
00196             int32_t& max_string_len);
00197     virtual void get_shortreal_string_list(
00198             TString<float32_t>*& strings, int32_t& num_str,
00199             int32_t& max_string_len);
00200     virtual void get_real_string_list(
00201             TString<float64_t>*& strings, int32_t& num_str,
00202             int32_t& max_string_len);
00203     virtual void get_longreal_string_list(
00204             TString<floatmax_t>*& strings, int32_t& num_str,
00205             int32_t& max_string_len);
00207 
00215     virtual void set_byte_vector(const uint8_t* vector, int32_t len);
00216     virtual void set_char_vector(const char* vector, int32_t len);
00217     virtual void set_int_vector(const int32_t* vector, int32_t len);
00218     virtual void set_shortreal_vector( const float32_t* vector, int32_t len);
00219     virtual void set_real_vector(const float64_t* vector, int32_t len);
00220     virtual void set_short_vector(const int16_t* vector, int32_t len);
00221     virtual void set_word_vector(const uint16_t* vector, int32_t len);
00223 
00224 
00232     virtual void set_byte_matrix(
00233             const uint8_t* matrix, int32_t num_feat, int32_t num_vec);
00234     virtual void set_int8_matrix(
00235             const int8_t* matrix, int32_t num_feat, int32_t num_vec);
00236     virtual void set_char_matrix(
00237             const char* matrix, int32_t num_feat, int32_t num_vec);
00238     virtual void set_int_matrix(
00239             const int32_t* matrix, int32_t num_feat, int32_t num_vec);
00240     virtual void set_uint_matrix(
00241             const uint32_t* matrix, int32_t num_feat, int32_t num_vec);
00242     virtual void set_long_matrix(
00243             const int64_t* matrix, int32_t num_feat, int32_t num_vec);
00244     virtual void set_ulong_matrix(
00245             const uint64_t* matrix, int32_t num_feat, int32_t num_vec);
00246     virtual void set_shortreal_matrix(
00247             const float32_t* matrix, int32_t num_feat, int32_t num_vec);
00248     virtual void set_real_matrix(
00249             const float64_t* matrix, int32_t num_feat, int32_t num_vec);
00250     virtual void set_longreal_matrix(
00251             const floatmax_t* matrix, int32_t num_feat, int32_t num_vec);
00252     virtual void set_short_matrix(
00253             const int16_t* matrix, int32_t num_feat, int32_t num_vec);
00254     virtual void set_word_matrix(
00255             const uint16_t* matrix, int32_t num_feat, int32_t num_vec);
00257 
00265     virtual void set_bool_sparsematrix(
00266             const TSparse<bool>* matrix, int32_t num_feat, int32_t num_vec);
00267     virtual void set_byte_sparsematrix(
00268             const TSparse<uint8_t>* matrix, int32_t num_feat, int32_t num_vec);
00269     virtual void set_int8_sparsematrix(
00270             const TSparse<int8_t>* matrix, int32_t num_feat, int32_t num_vec);
00271     virtual void set_char_sparsematrix(
00272             const TSparse<char>* matrix, int32_t num_feat, int32_t num_vec);
00273     virtual void set_int_sparsematrix(
00274             const TSparse<int32_t>* matrix, int32_t num_feat, int32_t num_vec);
00275     virtual void set_uint_sparsematrix(
00276             const TSparse<uint32_t>* matrix, int32_t num_feat, int32_t num_vec);
00277     virtual void set_long_sparsematrix(
00278             const TSparse<int64_t>* matrix, int32_t num_feat, int32_t num_vec);
00279     virtual void set_ulong_sparsematrix(
00280             const TSparse<uint64_t>* matrix, int32_t num_feat, int32_t num_vec);
00281     virtual void set_short_sparsematrix(
00282             const TSparse<int16_t>* matrix, int32_t num_feat, int32_t num_vec);
00283     virtual void set_word_sparsematrix(
00284             const TSparse<uint16_t>* matrix, int32_t num_feat, int32_t num_vec); 
00285     virtual void set_shortreal_sparsematrix(
00286             const TSparse<float32_t>* matrix, int32_t num_feat, int32_t num_vec);
00287     virtual void set_real_sparsematrix(
00288             const TSparse<float64_t>* matrix, int32_t num_feat, int32_t num_vec);
00289     virtual void set_longreal_sparsematrix(
00290             const TSparse<floatmax_t>* matrix, int32_t num_feat, int32_t num_vec);
00292 
00293 
00302     virtual void set_byte_string_list(
00303             const TString<uint8_t>* strings, int32_t num_str);
00304     virtual void set_int8_string_list(
00305             const TString<int8_t>* strings, int32_t num_str);
00306     virtual void set_char_string_list(
00307             const TString<char>* strings, int32_t num_str);
00308     virtual void set_int_string_list(
00309             const TString<int32_t>* strings, int32_t num_str);
00310     virtual void set_uint_string_list(
00311             const TString<uint32_t>* strings, int32_t num_str);
00312     virtual void set_short_string_list(
00313             const TString<int16_t>* strings, int32_t num_str);
00314     virtual void set_word_string_list(
00315             const TString<uint16_t>* strings, int32_t num_str);
00316     virtual void set_long_string_list(
00317             const TString<int64_t>* strings, int32_t num_str);
00318     virtual void set_ulong_string_list(
00319             const TString<uint64_t>* strings, int32_t num_str);
00320     virtual void set_shortreal_string_list(
00321             const TString<float32_t>* strings, int32_t num_str);
00322     virtual void set_real_string_list(
00323             const TString<float64_t>* strings, int32_t num_str);
00324     virtual void set_longreal_string_list(
00325             const TString<floatmax_t>* strings, int32_t num_str);
00327 
00329     inline virtual const char* get_name() const { return "BinaryFile"; }
00330 
00331 protected:
00336     void read_header(TSGDataType* dest);
00337 
00342     void write_header(const TSGDataType* datatype);
00343 
00349     int32_t parse_first_header(TSGDataType& type);
00350     
00356     int32_t parse_next_header(TSGDataType& type);
00357 
00358 private:
00365     template <class DT> DT* load_data(DT* target, int64_t& num)
00366     {
00367         CSimpleFile<DT> f(filename, file);
00368         return f.load(target, num);
00369     }
00370 
00377     template <class DT> bool save_data(DT* src, int64_t num)
00378     {
00379         CSimpleFile<DT> f(filename, file);
00380         return f.save(src, num);
00381     }
00382 };
00383 }
00384 #endif //__BINARY_FILE_H__