SerializableHdf5Reader00.cpp

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 
00011 #include <shogun/lib/config.h>
00012 #ifdef HAVE_HDF5
00013 
00014 #include <shogun/io/SerializableHdf5Reader00.h>
00015 
00016 using namespace shogun;
00017 
00018 SerializableHdf5Reader00::SerializableHdf5Reader00(
00019     CSerializableHdf5File* file) { m_file = file; }
00020 
00021 SerializableHdf5Reader00::~SerializableHdf5Reader00() {}
00022 
00023 bool
00024 SerializableHdf5Reader00::read_scalar_wrapped(
00025     const TSGDataType* type, void* param)
00026 {
00027     CSerializableHdf5File::type_item_t* m
00028         = m_file->m_stack_type.back();
00029 
00030     switch (type->m_stype) {
00031     case ST_NONE:
00032         if (m->y != 0 || m->x != 0) return true;
00033         break;
00034     case ST_STRING:
00035         if (m->y == -1 || m->x == -1) break;
00036 
00037         if (m->sub_y != 0) return true;
00038 
00039         memcpy(param, m->vltype[m->x*m->dims[1] + m->y].p,
00040                m->vltype[m->x*m->dims[1] + m->y].len
00041                *type->sizeof_ptype());
00042 
00043         return true;
00044     case ST_SPARSE:
00045         if (m->sub_y != 0) return true;
00046         break;
00047     }
00048 
00049     hid_t mem_type_id;
00050     if ((mem_type_id = CSerializableHdf5File::new_stype2hdf5(
00051              type->m_stype, type->m_ptype)) < 0) return false;
00052 
00053     switch (type->m_stype) {
00054     case ST_NONE:
00055         if (H5Dread(m->dset, mem_type_id, H5S_ALL, H5S_ALL,
00056                     H5P_DEFAULT, param) < 0) return false;
00057         break;
00058     case ST_STRING:
00059         if (H5Dread(m->dset, mem_type_id, H5S_ALL, H5S_ALL,
00060                     H5P_DEFAULT, m->vltype) < 0) return false;
00061         break;
00062     case ST_SPARSE:
00063         if (H5Dread(m->dset, m->dtype, H5S_ALL, H5S_ALL,
00064                     H5P_DEFAULT, m->sparse_ptr) < 0) return false;
00065         break;
00066     }
00067 
00068     if (H5Tclose(mem_type_id) < 0) return false;
00069 
00070     return true;
00071 }
00072 
00073 bool
00074 SerializableHdf5Reader00::read_cont_begin_wrapped(
00075     const TSGDataType* type, index_t* len_read_y, index_t* len_read_x)
00076 {
00077     CSerializableHdf5File::type_item_t* m
00078         = m_file->m_stack_type.back();
00079 
00080     if (type->m_ptype != PT_SGOBJECT) {
00081         switch (type->m_ctype) {
00082         case CT_NDARRAY:
00083             SG_NOTIMPLEMENTED;
00084         case CT_SCALAR:
00085             SG_ERROR("read_cont_begin_wrapped(): Implementation error"
00086                      " during writing Hdf5File (0)!");
00087             return false;
00088         case CT_VECTOR: case CT_SGVECTOR: *len_read_y = m->dims[0]; break;
00089         case CT_MATRIX: case CT_SGMATRIX:
00090             *len_read_x = m->dims[0]; *len_read_y = m->dims[1];
00091             break;
00092         default: return false;
00093         }
00094 
00095         return true;
00096     }
00097 
00098     if (!m_file->attr_exists(STR_IS_CONT)) return false;
00099 
00100     string_t ctype_buf, buf;
00101     type->to_string(ctype_buf, STRING_LEN);
00102     if (!m_file->attr_read_string(STR_CTYPE_NAME, buf, STRING_LEN))
00103         return false;
00104     if (strcmp(ctype_buf, buf) != 0) return false;
00105 
00106     switch (type->m_ctype) {
00107     case CT_NDARRAY:
00108         SG_NOTIMPLEMENTED;
00109     case CT_SCALAR:
00110         SG_ERROR("read_cont_begin_wrapped(): Implementation error"
00111                  " during writing Hdf5File (1)!");
00112         return false;
00113     case CT_MATRIX: case CT_SGMATRIX:
00114         if (!m_file->attr_read_scalar(TYPE_INDEX, STR_LENGTH_X,
00115                                       len_read_x))
00116             return false;
00117         /* break;  */
00118     case CT_VECTOR: case CT_SGVECTOR:
00119         if (!m_file->attr_read_scalar(TYPE_INDEX, STR_LENGTH_Y,
00120                                       len_read_y))
00121             return false;
00122         break;
00123     default: return false;
00124     }
00125 
00126     return true;
00127 }
00128 
00129 bool
00130 SerializableHdf5Reader00::read_cont_end_wrapped(
00131     const TSGDataType* type, index_t len_read_y, index_t len_read_x)
00132 {
00133     return true;
00134 }
00135 
00136 bool
00137 SerializableHdf5Reader00::read_string_begin_wrapped(
00138     const TSGDataType* type, index_t* length)
00139 {
00140     CSerializableHdf5File::type_item_t* m
00141         = m_file->m_stack_type.back();
00142 
00143     if (m->y == 0 && m->x == 0) {
00144         m->y = -1; m->x = -1;
00145         read_scalar_wrapped(type, NULL);
00146         m->y = 0; m->x = 0;
00147     }
00148 
00149     *length = m->vltype[m->x*m->dims[1] + m->y].len;
00150 
00151     return true;
00152 }
00153 
00154 bool
00155 SerializableHdf5Reader00::read_string_end_wrapped(
00156     const TSGDataType* type, index_t length)
00157 {
00158     return true;
00159 }
00160 
00161 bool
00162 SerializableHdf5Reader00::read_stringentry_begin_wrapped(
00163     const TSGDataType* type, index_t y)
00164 {
00165     CSerializableHdf5File::type_item_t* m
00166         = m_file->m_stack_type.back();
00167 
00168     m->sub_y = y;
00169 
00170     return true;
00171 }
00172 
00173 bool
00174 SerializableHdf5Reader00::read_stringentry_end_wrapped(
00175     const TSGDataType* type, index_t y)
00176 {
00177     return true;
00178 }
00179 
00180 bool
00181 SerializableHdf5Reader00::read_sparse_begin_wrapped(
00182     const TSGDataType* type, index_t* length)
00183 {
00184     CSerializableHdf5File::type_item_t* m_prev
00185         = m_file->m_stack_type.back();
00186 
00187     if(!m_file->dspace_select(type->m_ctype, m_prev->y, m_prev->x))
00188         return false;
00189 
00190     CSerializableHdf5File::type_item_t* m = new CSerializableHdf5File
00191         ::type_item_t(m_prev->name);
00192     m_file->m_stack_type.push_back(m);
00193 
00194     /* ************************************************************ */
00195 
00196     if (!m_file->group_open(m->name, STR_GROUP_PREFIX)) return false;
00197     if (!m_file->attr_exists(STR_IS_SPARSE)) return false;
00198 
00199     string_t name;
00200     CSerializableHdf5File::index2string(
00201         name, STRING_LEN, type->m_ctype, m_prev->y, m_prev->x);
00202     if ((m->dset = H5Dopen2(m_file->m_stack_h5stream.back(), name,
00203                             H5P_DEFAULT)) < 0)
00204         return false;
00205 
00206     if ((m->dtype = H5Dget_type(m->dset)) < 0) return false;
00207     if (!CSerializableHdf5File::isequal_stype2hdf5(
00208             type->m_stype, type->m_ptype, m->dtype)) return false;
00209 
00210     if ((m->dspace = H5Dget_space(m->dset)) < 0) return false;
00211     if (H5Sget_simple_extent_ndims(m->dspace) != 1) return false;
00212 
00213 
00214     if ((m->rank = H5Sget_simple_extent_dims(m->dspace, m->dims, NULL)
00215             ) < 0) return false;
00216 
00217     if (H5Sget_simple_extent_type(m->dspace) != H5S_NULL
00218         && m->rank != 1) return false;
00219 
00220     *length = m->dims[0];
00221 
00222     /* ************************************************************ */
00223 
00224     char* buf = SG_MALLOC(char, CSerializableHdf5File::sizeof_sparsetype());
00225 
00226     hid_t mem_type_id;
00227     if ((mem_type_id = CSerializableHdf5File::new_sparsetype()) < 0)
00228         return false;
00229 
00230     hid_t mem_space_id;
00231     if ((mem_space_id = H5Screate_simple(0, NULL, NULL)) < 0)
00232         return false;
00233 
00234     if (H5Dread(m_prev->dset, mem_type_id, mem_space_id,
00235                 m_prev->dspace, H5P_DEFAULT, buf) < 0) return false;
00236 
00237     if (H5Sclose(mem_space_id) < 0) return false;
00238     if (H5Tclose(mem_type_id) < 0) return false;
00239 
00240     delete buf;
00241 
00242     return true;
00243 }
00244 
00245 bool
00246 SerializableHdf5Reader00::read_sparse_end_wrapped(
00247     const TSGDataType* type, index_t length)
00248 {
00249     if (!m_file->group_close()) return false;
00250 
00251     delete m_file->m_stack_type.back();
00252     m_file->m_stack_type.pop_back();
00253 
00254     return true;
00255 }
00256 
00257 bool
00258 SerializableHdf5Reader00::read_sparseentry_begin_wrapped(
00259     const TSGDataType* type, SGSparseVectorEntry<char>* first_entry,
00260     index_t* feat_index, index_t y)
00261 {
00262     CSerializableHdf5File::type_item_t* m
00263         = m_file->m_stack_type.back();
00264 
00265     m->sparse_ptr = first_entry;
00266     m->sub_y = y;
00267 
00268     return true;
00269 }
00270 
00271 bool
00272 SerializableHdf5Reader00::read_sparseentry_end_wrapped(
00273     const TSGDataType* type, SGSparseVectorEntry<char>* first_entry,
00274     index_t* feat_index, index_t y)
00275 {
00276     return true;
00277 }
00278 
00279 bool
00280 SerializableHdf5Reader00::read_item_begin_wrapped(
00281     const TSGDataType* type, index_t y, index_t x)
00282 {
00283     CSerializableHdf5File::type_item_t* m
00284         = m_file->m_stack_type.back();
00285     m->y = y; m->x = x;
00286 
00287     if (type->m_ptype != PT_SGOBJECT) return true;
00288 
00289     string_t name;
00290     if (!CSerializableHdf5File::index2string(
00291             name, STRING_LEN, type->m_ctype, y, x)) return false;
00292     if (!m_file->group_open(name, "")) return false;
00293 
00294     return true;
00295 }
00296 
00297 bool
00298 SerializableHdf5Reader00::read_item_end_wrapped(
00299     const TSGDataType* type, index_t y, index_t x)
00300 {
00301     if (type->m_ptype == PT_SGOBJECT)
00302         if (!m_file->group_close()) return false;
00303 
00304     return true;
00305 }
00306 
00307 bool
00308 SerializableHdf5Reader00::read_sgserializable_begin_wrapped(
00309     const TSGDataType* type, char* sgserializable_name,
00310     EPrimitiveType* generic)
00311 {
00312     if (!m_file->attr_exists(STR_IS_SGSERIALIZABLE)) return false;
00313 
00314     if (m_file->attr_exists(STR_IS_NULL)) {
00315         *sgserializable_name = '\0'; return true;
00316     }
00317 
00318     if (!m_file->attr_read_string(
00319             STR_INSTANCE_NAME, sgserializable_name, STRING_LEN))
00320         return false;
00321 
00322     if (m_file->attr_exists(STR_GENERIC_NAME)) {
00323         string_t buf;
00324         if (!m_file->attr_read_string(
00325                 STR_GENERIC_NAME, buf, STRING_LEN)) return false;
00326         if (!TSGDataType::string_to_ptype(generic, buf))
00327             return false;
00328     }
00329 
00330     return true;
00331 }
00332 
00333 bool
00334 SerializableHdf5Reader00::read_sgserializable_end_wrapped(
00335     const TSGDataType* type, const char* sgserializable_name,
00336     EPrimitiveType generic)
00337 {
00338     return true;
00339 }
00340 
00341 bool
00342 SerializableHdf5Reader00::read_type_begin_wrapped(
00343     const TSGDataType* type, const char* name, const char* prefix)
00344 {
00345     CSerializableHdf5File::type_item_t* m = new CSerializableHdf5File
00346         ::type_item_t(name);
00347     m_file->m_stack_type.push_back(m);
00348 
00349     if (type->m_ptype == PT_SGOBJECT) {
00350         if (!m_file->group_open(name, "")) return false;
00351         return true;
00352     }
00353 
00354     if ((m->dset = H5Dopen2(m_file->m_stack_h5stream.back(), name,
00355                             H5P_DEFAULT)) < 0)
00356         return false;
00357 
00358     if ((m->dtype = H5Dget_type(m->dset)) < 0) return false;
00359     if (!CSerializableHdf5File::isequal_stype2hdf5(
00360             type->m_stype, type->m_ptype, m->dtype)) return false;
00361 
00362     if ((m->dspace = H5Dget_space(m->dset)) < 0) return false;
00363 
00364     if (H5Sget_simple_extent_ndims(m->dspace) > 2) return false;
00365     if ((m->rank = H5Sget_simple_extent_dims(m->dspace, m->dims, NULL)
00366             ) < 0) return false;
00367 
00368     switch (type->m_ctype) {
00369     case CT_NDARRAY:
00370         SG_NOTIMPLEMENTED;
00371     case CT_SCALAR:
00372         if (m->rank != 0) return false;
00373         if (type->m_stype == ST_STRING) m->vltype = SG_MALLOC(hvl_t, 1);
00374         break;
00375     case CT_VECTOR: case CT_SGVECTOR:
00376         if (H5Sget_simple_extent_type(m->dspace) != H5S_NULL
00377             && m->rank != 1) return false;
00378         if (type->m_stype == ST_STRING)
00379             m->vltype = SG_MALLOC(hvl_t, m->dims[0]);
00380         break;
00381     case CT_MATRIX: case CT_SGMATRIX:
00382         if (H5Sget_simple_extent_type(m->dspace) != H5S_NULL
00383             && m->rank != 2) return false;
00384         if (type->m_stype == ST_STRING)
00385             m->vltype = SG_MALLOC(hvl_t, m->dims[0] *m->dims[1]);
00386         break;
00387     default: return false;
00388     }
00389 
00390     return true;
00391 }
00392 
00393 bool
00394 SerializableHdf5Reader00::read_type_end_wrapped(
00395     const TSGDataType* type, const char* name, const char* prefix)
00396 {
00397     if (type->m_ptype == PT_SGOBJECT)
00398         if (!m_file->group_close()) return false;
00399 
00400     delete m_file->m_stack_type.back();
00401     m_file->m_stack_type.pop_back();
00402     return true;
00403 }
00404 
00405 #endif /* HAVE_HDF5  */
 All Classes Namespaces Files Functions Variables Typedefs Enumerations Enumerator Friends Defines

SHOGUN Machine Learning Toolbox - Documentation