Merge pull request #1489 from gsomix/feature/libsvm

Probably fix for issues #1401 #1402

src/shogun/io/CSVFile.cpp

@@ -180,6 +180,8 @@ void CCSVFile::fname(sg_type*& matrix, int32_t& num_feat, int32_t& num_vec) \
 	skip_lines(m_num_to_skip); \
 	num_lines=get_stats(num_tokens); \
 	\
+	SG_SET_LOCALE_C; \
+	\
 	matrix=SG_MALLOC(sg_type, num_lines*num_tokens); \
 	skip_lines(m_num_to_skip); \
 	while (m_line_reader->has_next()) \
@@ -200,6 +202,8 @@ void CCSVFile::fname(sg_type*& matrix, int32_t& num_feat, int32_t& num_vec) \
 		current_line_idx++; \
 	} \
 	\
+	SG_RESET_LOCALE; \
+	\
 	if (!is_data_transposed) \
 	{ \
 		num_feat=num_tokens; \
@@ -229,6 +233,8 @@ GET_MATRIX(get_matrix, read_ulong, uint64_t)
 #define SET_VECTOR(fname, format, sg_type) \
 void CCSVFile::fname(const sg_type* vector, int32_t len) \
 { \
+	SG_SET_LOCALE_C; \
+	\
 	if (!is_data_transposed) \
 	{ \
 		for (int32_t i=0; i<len; i++) \
@@ -240,6 +246,8 @@ void CCSVFile::fname(const sg_type* vector, int32_t len) \
 			fprintf(file, "%" format "%c", vector[i], m_delimiter); \
 		fprintf(file, "\n"); \
 	} \
+	\
+	SG_RESET_LOCALE; \
 }
 
 SET_VECTOR(set_vector, SCNi8, int8_t)
@@ -259,6 +267,8 @@ SET_VECTOR(set_vector, SCNu16, uint16_t)
 #define SET_MATRIX(fname, format, sg_type) \
 void CCSVFile::fname(const sg_type* matrix, int32_t num_feat, int32_t num_vec) \
 { \
+	SG_SET_LOCALE_C; \
+	\
 	if (!is_data_transposed) \
 	{ \
 		for (int32_t i=0; i<num_vec; i++) \
@@ -277,6 +287,8 @@ void CCSVFile::fname(const sg_type* matrix, int32_t num_feat, int32_t num_vec) \
 			fprintf(file, "\n"); \
 		} \
 	} \
+	\
+	SG_RESET_LOCALE; \
 }
 
 SET_MATRIX(set_matrix, SCNi8, int8_t)

src/shogun/io/LibSVMFile.cpp

@@ -115,6 +115,8 @@ void CLibSVMFile::get_sparse_matrix(SGSparseVector<sg_type>*& matrix, int32_t& n
 	if (load_labels) \
 		labels=SG_MALLOC(float64_t, num_vec); \
 	\
+	SG_SET_LOCALE_C; \
+	\
 	while (m_line_reader->has_next()) \
 	{ \
 		num_entries=0; \
@@ -158,6 +160,8 @@ void CLibSVMFile::get_sparse_matrix(SGSparseVector<sg_type>*& matrix, int32_t& n
 		SG_PROGRESS(current_line_ind, 0, num_vec, 1, "LOADING:\t") \
 	} \
 	\
+	SG_RESET_LOCALE; \
+	\
 	SG_INFO("file successfully read\n") \
 }
 
@@ -203,6 +207,8 @@ void CLibSVMFile::set_sparse_matrix( \
 			const SGSparseVector<sg_type>* matrix, int32_t num_feat, int32_t num_vec, \
 			const float64_t* labels) \
 { \
+	SG_SET_LOCALE_C; \
+	\
 	for (int32_t i=0; i<num_vec; i++) \
 	{ \
 		if (labels!=NULL) \
@@ -217,6 +223,8 @@ void CLibSVMFile::set_sparse_matrix( \
 		} \
 		fprintf(file, "\n"); \
 	} \
+	\
+	SG_RESET_LOCALE; \
 }
 
 SET_LABELED_SPARSE_MATRIX(SCNi32, bool)

src/shogun/io/Parser.cpp

@@ -87,28 +87,22 @@ bool CParser::read_bool()
 	SGVector<char> token=read_cstring();
 
 	if (token.vlen>0)
-		return (bool) strtol(token.vector, NULL, 10);
+		return (bool) strtod(token.vector, NULL);
 	else
 		return (bool) 0L;
 }
 
 #define READ_INT_METHOD(fname, convf, sg_type) \
-sg_type CParser::fname(int32_t base) \
+sg_type CParser::fname() \
 { \
 	SGVector<char> token=read_cstring(); \
 	\
 	if (token.vlen>0) \
-		return (sg_type) convf(token.vector, NULL, base); \
+		return (sg_type) convf(token.vector, NULL, 10); \
 	else \
 		return (sg_type) 0L; \
 }
 
-READ_INT_METHOD(read_char, strtol, char)
-READ_INT_METHOD(read_byte, strtoul, uint8_t)
-READ_INT_METHOD(read_short, strtol, int16_t)
-READ_INT_METHOD(read_word, strtoul, uint16_t)
-READ_INT_METHOD(read_int, strtol, int32_t)
-READ_INT_METHOD(read_uint, strtoul, uint32_t)
 READ_INT_METHOD(read_long, strtoll, int64_t)
 READ_INT_METHOD(read_ulong, strtoull, uint64_t)
 #undef READ_INT_METHOD
@@ -124,6 +118,13 @@ sg_type CParser::fname() \
 		return (sg_type) 0L; \
 }
 
+READ_REAL_METHOD(read_char, strtod, char)
+READ_REAL_METHOD(read_byte, strtod, uint8_t)
+READ_REAL_METHOD(read_short, strtod, int16_t)
+READ_REAL_METHOD(read_word, strtod, uint16_t)
+READ_REAL_METHOD(read_int, strtod, int32_t)
+READ_REAL_METHOD(read_uint, strtod, uint32_t)
+
 READ_REAL_METHOD(read_short_real, strtod, float32_t)
 READ_REAL_METHOD(read_real, strtod, float64_t)
 #ifdef HAVE_STRTOLD

src/shogun/io/Parser.h

@@ -50,14 +50,14 @@ class CParser : public CSGObject
 	/** read one of the several base data types. */
 	//@{
 	virtual bool read_bool();
-	virtual char read_char(int32_t base=10);
-	virtual uint8_t read_byte(int32_t base=10);
-	virtual int16_t read_short(int32_t base=10);
-	virtual uint16_t read_word(int32_t base=10);
-	virtual int32_t read_int(int32_t base=10);
-	virtual uint32_t read_uint(int32_t base=10);
-	virtual int64_t read_long(int32_t base=10);
-	virtual uint64_t read_ulong(int32_t base=10);
+	virtual char read_char();
+	virtual uint8_t read_byte();
+	virtual int16_t read_short();
+	virtual uint16_t read_word();
+	virtual int32_t read_int();
+	virtual uint32_t read_uint();
+	virtual int64_t read_long();
+	virtual uint64_t read_ulong();
 	virtual float32_t read_short_real();
 	virtual float64_t read_real();
 	virtual floatmax_t read_long_real();

tests/unit/io/Parser_unittest.cc

@@ -55,6 +55,8 @@ TEST(ParserTest, tokenization_reals)
 
 	CParser* reader=new CParser(cv, tokenizer);
 
+	SG_SET_LOCALE_C;
+
 	float64_t tmp=0;
 	int32_t num_tokens=0;
 	while (reader->has_next())
@@ -65,6 +67,8 @@ TEST(ParserTest, tokenization_reals)
 	}
 	EXPECT_EQ(num_tokens, ntokens);
 
+	SG_RESET_LOCALE;
+
 	SG_UNREF(reader);
 	SG_UNREF(tokenizer);
 }