00001
00002
00003
00004
00005
00006
00007
00008
00009
00010
00011 #include <shogun/features/RealFileFeatures.h>
00012 #include <shogun/features/Features.h>
00013 #include <shogun/io/SGIO.h>
00014
00015 #include <stdio.h>
00016 #include <string.h>
00017
00018 using namespace shogun;
00019
00020 CRealFileFeatures::CRealFileFeatures()
00021 {
00022 SG_UNSTABLE("CRealFileFeatures::CRealFileFeatures()", "\n");
00023
00024 working_file=NULL;
00025 working_filename=strdup("");
00026 intlen=0;
00027 doublelen=0;
00028 endian=0;
00029 fourcc=0;
00030 preprocd=0;
00031 labels=NULL;
00032 status=false;
00033 }
00034
00035 CRealFileFeatures::CRealFileFeatures(int32_t size, char* fname)
00036 : CDenseFeatures<float64_t>(size)
00037 {
00038 working_file=fopen(fname, "r");
00039 working_filename=strdup(fname);
00040 ASSERT(working_file);
00041 intlen=0;
00042 doublelen=0;
00043 endian=0;
00044 fourcc=0;
00045 preprocd=0;
00046 labels=NULL;
00047 status=load_base_data();
00048 }
00049
00050 CRealFileFeatures::CRealFileFeatures(int32_t size, FILE* file)
00051 : CDenseFeatures<float64_t>(size), working_file(file), working_filename(NULL)
00052 {
00053 ASSERT(working_file);
00054 intlen=0;
00055 doublelen=0;
00056 endian=0;
00057 fourcc=0;
00058 preprocd=0;
00059 labels=NULL;
00060 status=load_base_data();
00061 }
00062
00063 CRealFileFeatures::~CRealFileFeatures()
00064 {
00065 SG_FREE(working_filename);
00066 SG_FREE(labels);
00067 }
00068
00069 CRealFileFeatures::CRealFileFeatures(const CRealFileFeatures & orig)
00070 : CDenseFeatures<float64_t>(orig), working_file(orig.working_file), status(orig.status)
00071 {
00072 if (orig.working_filename)
00073 working_filename=strdup(orig.working_filename);
00074 if (orig.labels && get_num_vectors())
00075 {
00076 labels=SG_MALLOC(int32_t, get_num_vectors());
00077 memcpy(labels, orig.labels, sizeof(int32_t)*get_num_vectors());
00078 }
00079 }
00080
00081 float64_t* CRealFileFeatures::compute_feature_vector(
00082 int32_t num, int32_t &len, float64_t* target)
00083 {
00084 ASSERT(num<num_vectors);
00085 len=num_features;
00086 float64_t* featurevector=target;
00087 if (!featurevector)
00088 featurevector=SG_MALLOC(float64_t, num_features);
00089 ASSERT(working_file);
00090 fseek(working_file, filepos+num_features*doublelen*num, SEEK_SET);
00091 ASSERT(fread(featurevector, doublelen, num_features, working_file)==(size_t) num_features);
00092 return featurevector;
00093 }
00094
00095 float64_t* CRealFileFeatures::load_feature_matrix()
00096 {
00097 ASSERT(working_file);
00098 fseek(working_file, filepos, SEEK_SET);
00099 free_feature_matrix();
00100
00101 SG_INFO( "allocating feature matrix of size %.2fM\n", sizeof(double)*num_features*num_vectors/1024.0/1024.0);
00102 free_feature_matrix();
00103 feature_matrix=SGMatrix<float64_t>(num_features,num_vectors);
00104
00105 SG_INFO( "loading... be patient.\n");
00106
00107 for (int32_t i=0; i<(int32_t) num_vectors; i++)
00108 {
00109 if (!(i % (num_vectors/10+1)))
00110 SG_PRINT( "%02d%%.", (int) (100.0*i/num_vectors));
00111 else if (!(i % (num_vectors/200+1)))
00112 SG_PRINT( ".");
00113
00114 ASSERT(fread(&feature_matrix.matrix[num_features*i], doublelen, num_features, working_file)==(size_t) num_features);
00115 }
00116 SG_DONE();
00117
00118 return feature_matrix.matrix;
00119 }
00120
00121 int32_t CRealFileFeatures::get_label(int32_t idx)
00122 {
00123 ASSERT(idx<num_vectors);
00124 if (labels)
00125 return labels[idx];
00126 return 0;
00127 }
00128
00129 bool CRealFileFeatures::load_base_data()
00130 {
00131 ASSERT(working_file);
00132 uint32_t num_vec=0;
00133 uint32_t num_feat=0;
00134
00135 ASSERT(fread(&intlen, sizeof(uint8_t), 1, working_file)==1);
00136 ASSERT(fread(&doublelen, sizeof(uint8_t), 1, working_file)==1);
00137 ASSERT(fread(&endian, (uint32_t) intlen, 1, working_file)== 1);
00138 ASSERT(fread(&fourcc, (uint32_t) intlen, 1, working_file)==1);
00139 ASSERT(fread(&num_vec, (uint32_t) intlen, 1, working_file)==1);
00140 ASSERT(fread(&num_feat, (uint32_t) intlen, 1, working_file)==1);
00141 ASSERT(fread(&preprocd, (uint32_t) intlen, 1, working_file)==1);
00142 SG_INFO( "detected: intsize=%d, doublesize=%d, num_vec=%d, num_feat=%d, preprocd=%d\n", intlen, doublelen, num_vec, num_feat, preprocd);
00143 filepos=ftell(working_file);
00144 set_num_vectors(num_vec);
00145 set_num_features(num_feat);
00146 fseek(working_file, filepos+num_features*num_vectors*doublelen, SEEK_SET);
00147 SG_FREE(labels);
00148 labels=SG_MALLOC(int, num_vec);
00149 ASSERT(fread(labels, intlen, num_vec, working_file) == num_vec);
00150 return true;
00151 }