|
SHOGUN v0.9.3
|
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: 00035 CBinaryFile(FILE* f, const char* name=NULL); 00036 00043 CBinaryFile(char* fname, char rw='r', const char* name=NULL); 00044 00046 virtual ~CBinaryFile(); 00047 00055 virtual void get_byte_vector(uint8_t*& vector, int32_t& len); 00056 virtual void get_char_vector(char*& vector, int32_t& len); 00057 virtual void get_int_vector(int32_t*& vector, int32_t& len); 00058 virtual void get_real_vector(float64_t*& vector, int32_t& len); 00059 virtual void get_shortreal_vector(float32_t*& vector, int32_t& len); 00060 virtual void get_short_vector(int16_t*& vector, int32_t& len); 00061 virtual void get_word_vector(uint16_t*& vector, int32_t& len); 00063 00072 virtual void get_byte_matrix( 00073 uint8_t*& matrix, int32_t& num_feat, int32_t& num_vec); 00074 virtual void get_char_matrix( 00075 char*& matrix, int32_t& num_feat, int32_t& num_vec); 00076 virtual void get_int_matrix( 00077 int32_t*& matrix, int32_t& num_feat, int32_t& num_vec); 00078 virtual void get_uint_matrix( 00079 uint32_t*& matrix, int32_t& num_feat, int32_t& num_vec); 00080 virtual void get_long_matrix( 00081 int64_t*& matrix, int32_t& num_feat, int32_t& num_vec); 00082 virtual void get_ulong_matrix( 00083 uint64_t*& matrix, int32_t& num_feat, int32_t& num_vec); 00084 virtual void get_shortreal_matrix( 00085 float32_t*& matrix, int32_t& num_feat, int32_t& num_vec); 00086 virtual void get_real_matrix( 00087 float64_t*& matrix, int32_t& num_feat, int32_t& num_vec); 00088 virtual void get_longreal_matrix( 00089 floatmax_t*& matrix, int32_t& num_feat, int32_t& num_vec); 00090 virtual void get_short_matrix( 00091 int16_t*& matrix, int32_t& num_feat, int32_t& num_vec); 00092 virtual void get_word_matrix( 00093 uint16_t*& matrix, int32_t& num_feat, int32_t& num_vec); 00095 00104 virtual void get_byte_ndarray( 00105 uint8_t*& array, int32_t*& dims, int32_t& num_dims); 00106 virtual void get_char_ndarray( 00107 char*& array, int32_t*& dims, int32_t& num_dims); 00108 virtual void get_int_ndarray( 00109 int32_t*& array, int32_t*& dims, int32_t& num_dims); 00110 virtual void get_shortreal_ndarray( 00111 float32_t*& array, int32_t*& dims, int32_t& num_dims); 00112 virtual void get_real_ndarray( 00113 float64_t*& array, int32_t*& dims, int32_t& num_dims); 00114 virtual void get_short_ndarray( 00115 int16_t*& array, int32_t*& dims, int32_t& num_dims); 00116 virtual void get_word_ndarray( 00117 uint16_t*& array, int32_t*& dims, int32_t& num_dims); 00119 00128 virtual void get_bool_sparsematrix( 00129 TSparse<bool>*& matrix, int32_t& num_feat, int32_t& num_vec); 00130 virtual void get_byte_sparsematrix( 00131 TSparse<uint8_t>*& matrix, int32_t& num_feat, int32_t& num_vec); 00132 virtual void get_char_sparsematrix( 00133 TSparse<char>*& matrix, int32_t& num_feat, int32_t& num_vec); 00134 virtual void get_int_sparsematrix( 00135 TSparse<int32_t>*& matrix, int32_t& num_feat, int32_t& num_vec); 00136 virtual void get_uint_sparsematrix( 00137 TSparse<uint32_t>*& matrix, int32_t& num_feat, int32_t& num_vec); 00138 virtual void get_long_sparsematrix( 00139 TSparse<int64_t>*& matrix, int32_t& num_feat, int32_t& num_vec); 00140 virtual void get_ulong_sparsematrix( 00141 TSparse<uint64_t>*& matrix, int32_t& num_feat, int32_t& num_vec); 00142 virtual void get_short_sparsematrix( 00143 TSparse<int16_t>*& matrix, int32_t& num_feat, int32_t& num_vec); 00144 virtual void get_word_sparsematrix( 00145 TSparse<uint16_t>*& matrix, int32_t& num_feat, int32_t& num_vec); 00146 virtual void get_shortreal_sparsematrix( 00147 TSparse<float32_t>*& matrix, int32_t& num_feat, int32_t& num_vec); 00148 virtual void get_real_sparsematrix( 00149 TSparse<float64_t>*& matrix, int32_t& num_feat, int32_t& num_vec); 00150 virtual void get_longreal_sparsematrix( 00151 TSparse<floatmax_t>*& matrix, int32_t& num_feat, int32_t& num_vec); 00153 00154 00163 virtual void get_byte_string_list( 00164 T_STRING<uint8_t>*& strings, int32_t& num_str, 00165 int32_t& max_string_len); 00166 virtual void get_char_string_list( 00167 T_STRING<char>*& strings, int32_t& num_str, 00168 int32_t& max_string_len); 00169 virtual void get_int_string_list( 00170 T_STRING<int32_t>*& strings, int32_t& num_str, 00171 int32_t& max_string_len); 00172 virtual void get_uint_string_list( 00173 T_STRING<uint32_t>*& strings, int32_t& num_str, 00174 int32_t& max_string_len); 00175 virtual void get_short_string_list( 00176 T_STRING<int16_t>*& strings, int32_t& num_str, 00177 int32_t& max_string_len); 00178 virtual void get_word_string_list( 00179 T_STRING<uint16_t>*& strings, int32_t& num_str, 00180 int32_t& max_string_len); 00181 virtual void get_long_string_list( 00182 T_STRING<int64_t>*& strings, int32_t& num_str, 00183 int32_t& max_string_len); 00184 virtual void get_ulong_string_list( 00185 T_STRING<uint64_t>*& strings, int32_t& num_str, 00186 int32_t& max_string_len); 00187 virtual void get_shortreal_string_list( 00188 T_STRING<float32_t>*& strings, int32_t& num_str, 00189 int32_t& max_string_len); 00190 virtual void get_real_string_list( 00191 T_STRING<float64_t>*& strings, int32_t& num_str, 00192 int32_t& max_string_len); 00193 virtual void get_longreal_string_list( 00194 T_STRING<floatmax_t>*& strings, int32_t& num_str, 00195 int32_t& max_string_len); 00197 00205 virtual void set_byte_vector(const uint8_t* vector, int32_t len); 00206 virtual void set_char_vector(const char* vector, int32_t len); 00207 virtual void set_int_vector(const int32_t* vector, int32_t len); 00208 virtual void set_shortreal_vector( const float32_t* vector, int32_t len); 00209 virtual void set_real_vector(const float64_t* vector, int32_t len); 00210 virtual void set_short_vector(const int16_t* vector, int32_t len); 00211 virtual void set_word_vector(const uint16_t* vector, int32_t len); 00213 00214 00222 virtual void set_byte_matrix( 00223 const uint8_t* matrix, int32_t num_feat, int32_t num_vec); 00224 virtual void set_char_matrix( 00225 const char* matrix, int32_t num_feat, int32_t num_vec); 00226 virtual void set_int_matrix( 00227 const int32_t* matrix, int32_t num_feat, int32_t num_vec); 00228 virtual void set_uint_matrix( 00229 const uint32_t* matrix, int32_t num_feat, int32_t num_vec); 00230 virtual void set_long_matrix( 00231 const int64_t* matrix, int32_t num_feat, int32_t num_vec); 00232 virtual void set_ulong_matrix( 00233 const uint64_t* matrix, int32_t num_feat, int32_t num_vec); 00234 virtual void set_shortreal_matrix( 00235 const float32_t* matrix, int32_t num_feat, int32_t num_vec); 00236 virtual void set_real_matrix( 00237 const float64_t* matrix, int32_t num_feat, int32_t num_vec); 00238 virtual void set_longreal_matrix( 00239 const floatmax_t* matrix, int32_t num_feat, int32_t num_vec); 00240 virtual void set_short_matrix( 00241 const int16_t* matrix, int32_t num_feat, int32_t num_vec); 00242 virtual void set_word_matrix( 00243 const uint16_t* matrix, int32_t num_feat, int32_t num_vec); 00245 00253 virtual void set_bool_sparsematrix( 00254 const TSparse<bool>* matrix, int32_t num_feat, int32_t num_vec); 00255 virtual void set_byte_sparsematrix( 00256 const TSparse<uint8_t>* matrix, int32_t num_feat, int32_t num_vec); 00257 virtual void set_char_sparsematrix( 00258 const TSparse<char>* matrix, int32_t num_feat, int32_t num_vec); 00259 virtual void set_int_sparsematrix( 00260 const TSparse<int32_t>* matrix, int32_t num_feat, int32_t num_vec); 00261 virtual void set_uint_sparsematrix( 00262 const TSparse<uint32_t>* matrix, int32_t num_feat, int32_t num_vec); 00263 virtual void set_long_sparsematrix( 00264 const TSparse<int64_t>* matrix, int32_t num_feat, int32_t num_vec); 00265 virtual void set_ulong_sparsematrix( 00266 const TSparse<uint64_t>* matrix, int32_t num_feat, int32_t num_vec); 00267 virtual void set_short_sparsematrix( 00268 const TSparse<int16_t>* matrix, int32_t num_feat, int32_t num_vec); 00269 virtual void set_word_sparsematrix( 00270 const TSparse<uint16_t>* matrix, int32_t num_feat, int32_t num_vec); 00271 virtual void set_shortreal_sparsematrix( 00272 const TSparse<float32_t>* matrix, int32_t num_feat, int32_t num_vec); 00273 virtual void set_real_sparsematrix( 00274 const TSparse<float64_t>* matrix, int32_t num_feat, int32_t num_vec); 00275 virtual void set_longreal_sparsematrix( 00276 const TSparse<floatmax_t>* matrix, int32_t num_feat, int32_t num_vec); 00278 00279 00288 virtual void set_byte_string_list( 00289 const T_STRING<uint8_t>* strings, int32_t num_str); 00290 virtual void set_char_string_list( 00291 const T_STRING<char>* strings, int32_t num_str); 00292 virtual void set_int_string_list( 00293 const T_STRING<int32_t>* strings, int32_t num_str); 00294 virtual void set_uint_string_list( 00295 const T_STRING<uint32_t>* strings, int32_t num_str); 00296 virtual void set_short_string_list( 00297 const T_STRING<int16_t>* strings, int32_t num_str); 00298 virtual void set_word_string_list( 00299 const T_STRING<uint16_t>* strings, int32_t num_str); 00300 virtual void set_long_string_list( 00301 const T_STRING<int64_t>* strings, int32_t num_str); 00302 virtual void set_ulong_string_list( 00303 const T_STRING<uint64_t>* strings, int32_t num_str); 00304 virtual void set_shortreal_string_list( 00305 const T_STRING<float32_t>* strings, int32_t num_str); 00306 virtual void set_real_string_list( 00307 const T_STRING<float64_t>* strings, int32_t num_str); 00308 virtual void set_longreal_string_list( 00309 const T_STRING<floatmax_t>* strings, int32_t num_str); 00311 00313 inline virtual const char* get_name() const { return "BinaryFile"; } 00314 00315 protected: 00320 SGDataType read_header(); 00321 00326 void write_header(SGDataType datatype); 00327 00333 int32_t parse_first_header(SGDataType &type); 00334 00340 int32_t parse_next_header(SGDataType &type); 00341 00342 private: 00349 template <class DT> DT* load_data(DT* target, int64_t& num) 00350 { 00351 CSimpleFile<DT> f(filename, file); 00352 return f.load(target, num); 00353 } 00354 00361 template <class DT> bool save_data(DT* src, int64_t num) 00362 { 00363 CSimpleFile<DT> f(filename, file); 00364 return f.save(src, num); 00365 } 00366 }; 00367 } 00368 #endif //__BINARY_FILE_H__
