#include "../src/data/csv_parser.h" #include "../src/data/libsvm_parser.h" #include "../src/data/libfm_parser.h" #include #include #include #include using namespace dmlc; using namespace dmlc::data; namespace parser_test { template class CSVParserTest : public CSVParser { public: explicit CSVParserTest(InputSplit *source, const std::map &args, int nthread) : CSVParser(source, args, nthread) {} void CallParseBlock(char *begin, char *end, RowBlockContainer *out) { CSVParser::ParseBlock(begin, end, out); } }; template class LibSVMParserTest : public LibSVMParser { public: explicit LibSVMParserTest(InputSplit *source, const std::map &args, int nthread) : LibSVMParser(source, args, nthread) {} void CallParseBlock(char *begin, char *end, RowBlockContainer *out) { LibSVMParser::ParseBlock(begin, end, out); } }; template class LibFMParserTest : public LibFMParser { public: explicit LibFMParserTest(InputSplit *source, const std::map &args, int nthread) : LibFMParser(source, args, nthread) {} void CallParseBlock(char *begin, char *end, RowBlockContainer *out) { LibFMParser::ParseBlock(begin, end, out); } }; } // namespace parser_test namespace { template static inline void CountDimensions(RowBlockContainer* rctr, size_t* out_num_row, size_t* out_num_col) { size_t num_row = rctr->label.size(); size_t num_col = 0; for (size_t i = rctr->offset[0]; i < rctr->offset[num_row]; ++i) { const IndexType index = rctr->index[i]; num_col = std::max(num_col, static_cast(index + 1)); } *out_num_row = num_row; *out_num_col = num_col; } } // namespace anonymous TEST(CSVParser, test_ignore_bom) { using namespace parser_test; InputSplit *source = nullptr; const std::map args; std::unique_ptr> parser( new CSVParserTest(source, args, 1)); std::string data = "\xEF\xBB\xBF\x31\n\xEF\xBB\x32\n"; char *out_data = (char *)data.c_str(); RowBlockContainer *rctr = new RowBlockContainer(); parser->CallParseBlock(out_data, out_data + data.size(), rctr); CHECK(rctr->value[0] == 1); CHECK(rctr->value[1] == 0); data = "\xEF\xBB\xBF\x31\n\xEF\xBB\xBF\x32\n"; out_data = (char *)data.c_str(); rctr = new RowBlockContainer(); parser->CallParseBlock(out_data, out_data + data.size(), rctr); CHECK(rctr->value[0] == 1); CHECK(rctr->value[1] == 2); } TEST(CSVParser, test_standard_case) { using namespace parser_test; InputSplit *source = nullptr; const std::map args; std::unique_ptr> parser( new CSVParserTest(source, args, 1)); RowBlockContainer *rctr = new RowBlockContainer(); std::string data = "0,1,2,3\n4,5,6,7\n8,9,10,11\n"; char *out_data = const_cast(data.c_str()); parser->CallParseBlock(out_data, out_data + data.size(), rctr); for (size_t i = 0; i < rctr->value.size(); i++) { CHECK(i == rctr->value[i]); } } TEST(CSVParser, test_int32_parse) { using namespace parser_test; InputSplit *source = nullptr; const std::map args; std::unique_ptr> parser( new CSVParserTest(source, args, 1)); RowBlockContainer *rctr = new RowBlockContainer(); std::string data = "20000000,20000001,20000002,20000003\n" "20000004,20000005,20000006,20000007\n" "20000008,20000009,20000010,20000011\n"; char *out_data = const_cast(data.c_str()); parser->CallParseBlock(out_data, out_data + data.size(), rctr); for (size_t i = 0; i < rctr->value.size(); i++) { CHECK((i+20000000) == rctr->value[i]); } } TEST(CSVParser, test_int64_parse) { using namespace parser_test; InputSplit *source = nullptr; const std::map args; std::unique_ptr> parser( new CSVParserTest(source, args, 1)); RowBlockContainer *rctr = new RowBlockContainer(); std::string data = "2147483648,2147483649,2147483650,2147483651\n" "2147483652,2147483653,2147483654,2147483655\n" "2147483656,2147483657,2147483658,2147483659\n"; char *out_data = const_cast(data.c_str()); parser->CallParseBlock(out_data, out_data + data.size(), rctr); for (size_t i = 0; i < rctr->value.size(); i++) { CHECK((i+2147483648) == rctr->value[i]); } } TEST(CSVParser, test_different_newlines) { using namespace parser_test; InputSplit *source = nullptr; const std::map args; std::unique_ptr> parser( new CSVParserTest(source, args, 1)); RowBlockContainer *rctr = new RowBlockContainer(); std::string data = "0,1,2,3\r\n4,5,6,7\r\n8,9,10,11\r\n"; char *out_data = const_cast(data.c_str()); parser->CallParseBlock(out_data, out_data + data.size(), rctr); for (size_t i = 0; i < rctr->value.size(); i++) { CHECK(i == rctr->value[i]); } } TEST(CSVParser, test_noeol) { using namespace parser_test; InputSplit *source = nullptr; const std::map args; std::unique_ptr> parser( new CSVParserTest(source, args, 1)); RowBlockContainer *rctr = new RowBlockContainer(); std::string data = "0,1,2,3\r\n4,5,6,7\r\n8,9,10,11"; char *out_data = const_cast(data.c_str()); parser->CallParseBlock(out_data, out_data + data.size(), rctr); for (size_t i = 0; i < rctr->value.size(); i++) { CHECK(i == rctr->value[i]); } } TEST(CSVParser, test_delimiter) { using namespace parser_test; InputSplit *source = nullptr; const std::map args{ {"delimiter", " "} }; std::unique_ptr> parser( new CSVParserTest(source, args, 1)); RowBlockContainer *rctr = new RowBlockContainer(); std::string data = "0 1 2 3\n4 5 6 7\n8 9 10 11"; char *out_data = const_cast(data.c_str()); parser->CallParseBlock(out_data, out_data + data.size(), rctr); for (size_t i = 0; i < rctr->value.size(); i++) { CHECK(i == rctr->value[i]); } } TEST(CSVParser, test_weight_column) { using namespace parser_test; InputSplit *source = nullptr; const std::map args{ {"weight_column", "2"} }; std::unique_ptr> parser( new CSVParserTest(source, args, 1)); RowBlockContainer *rctr = new RowBlockContainer(); std::string data = "0,1,2,3\n4,5,6,7\n8,9,10,11"; char *out_data = const_cast(data.c_str()); parser->CallParseBlock(out_data, out_data + data.size(), rctr); CHECK_EQ(rctr->weight.size(), 3U); for (size_t i = 0; i < rctr->weight.size(); i++) { CHECK_EQ(rctr->weight[i], 2.0f + 4.0f * i); } const std::vector expected_values{0.0f, 1.0f, 3.0f, 4.0f, 5.0f, 7.0f, 8.0f, 9.0f, 11.0f}; CHECK_EQ(rctr->value.size(), expected_values.size()); for (size_t i = 0; i < rctr->value.size(); i++) { CHECK_EQ(rctr->value[i], expected_values[i]); } } TEST(CSVParser, test_weight_column_2) { using namespace parser_test; InputSplit *source = nullptr; const std::map args; std::unique_ptr> parser( new CSVParserTest(source, args, 1)); RowBlockContainer *rctr = new RowBlockContainer(); std::string data = "0,1,2,3\n4,5,6,7\n8,9,10,11"; char *out_data = const_cast(data.c_str()); parser->CallParseBlock(out_data, out_data + data.size(), rctr); CHECK(rctr->weight.empty()); CHECK_EQ(rctr->value.size(), 12U); for (size_t i = 0; i < rctr->value.size(); i++) { CHECK(i == rctr->value[i]); } } TEST(LibSVMParser, test_qid) { using namespace parser_test; InputSplit *source = nullptr; const std::map args; std::unique_ptr> parser( new LibSVMParserTest(source, args, 1)); RowBlockContainer* rctr = new RowBlockContainer(); std::string data = R"qid(3 qid:1 1:1 2:1 3:0 4:0.2 5:0 2 qid:1 1:0 2:0 3:1 4:0.1 5:1 1 qid:1 1:0 2:1 3:0 4:0.4 5:0 1 qid:1 1:0 2:0 3:1 4:0.3 5:0 1 qid:2 1:0 2:0 3:1 4:0.2 5:0 2 qid:2 1:1 2:0 3:1 4:0.4 5:0 1 qid:2 1:0 2:0 3:1 4:0.1 5:0 1 qid:2 1:0 2:0 3:1 4:0.2 5:0 2 qid:3 1:0 2:0 3:1 4:0.1 5:1 3 qid:3 1:1 2:1 3:0 4:0.3 5:0 4 qid:3 1:1 2:0 3:0 4:0.4 5:1 1 qid:3 1:0 2:1 3:1 4:0.5 5:0)qid"; char* out_data = const_cast(data.c_str()); parser->CallParseBlock(out_data, out_data + data.size(), rctr); const std::vector expected_offset{ 0, 5, 10, 15, 20, 25, 30, 35, 40, 45, 50, 55, 60 }; const std::vector expected_label{ 3, 2, 1, 1, 1, 2, 1, 1, 2, 3, 4, 1 }; const std::vector expected_qid{ 1, 1, 1, 1, 2, 2, 2, 2, 3, 3, 3, 3 }; const std::vector expected_index{ 1, 2, 3, 4, 5, 1, 2, 3, 4, 5, 1, 2, 3, 4, 5, 1, 2, 3, 4, 5, 1, 2, 3, 4, 5, 1, 2, 3, 4, 5, 1, 2, 3, 4, 5, 1, 2, 3, 4, 5, 1, 2, 3, 4, 5, 1, 2, 3, 4, 5, 1, 2, 3, 4, 5, 1, 2, 3, 4, 5 }; const std::vector expected_value{ 1.0f, 1.0f, 0.0f, 0.2f, 0.0f, 0.0f, 0.0f, 1.0f, 0.1f, 1.0f, 0.0f, 1.0f, 0.0f, 0.4f, 0.0f, 0.0f, 0.0f, 1.0f, 0.3f, 0.0f, 0.0f, 0.0f, 1.0f, 0.2f, 0.0f, 1.0f, 0.0f, 1.0f, 0.4f, 0.0f, 0.0f, 0.0f, 1.0f, 0.1f, 0.0f, 0.0f, 0.0f, 1.0f, 0.2f, 0.0f, 0.0f, 0.0f, 1.0f, 0.1f, 1.0f, 1.0f, 1.0f, 0.0f, 0.3f, 0.0f, 1.0f, 0.0f, 0.0f, 0.4f, 1.0f, 0.0f, 1.0f, 1.0f, 0.5f, 0.0f }; CHECK(rctr->offset == expected_offset); CHECK(rctr->label == expected_label); CHECK(rctr->qid == expected_qid); CHECK(rctr->index == expected_index); CHECK(rctr->value == expected_value); } TEST(LibSVMParser, test_excess_decimal_digits) { using namespace parser_test; InputSplit *source = nullptr; const std::map args; std::unique_ptr> parser( new LibSVMParserTest(source, args, 1)); RowBlockContainer* rctr = new RowBlockContainer(); std::string data = "0 1:17.065995780200002000000 4:17.0659957802 " "6:0.00017065995780200002 8:0.000170659957802\n"; char* out_data = const_cast(data.c_str()); parser->CallParseBlock(out_data, out_data + data.size(), rctr); size_t num_row, num_col; CountDimensions(rctr, &num_row, &num_col); CHECK_EQ(num_row, 1U); CHECK_EQ(num_col, 9U); const std::vector expected_index{1, 4, 6, 8}; CHECK(rctr->index == expected_index); // perform element-wise comparsion CHECK_EQ(rctr->value[0], rctr->value[1]); CHECK_EQ(rctr->value[2], rctr->value[3]); } TEST(LibSVMParser, test_indexing_mode_0_based) { using namespace parser_test; InputSplit *source = nullptr; const std::map args; std::unique_ptr> parser( new LibSVMParserTest(source, args, 1)); RowBlockContainer* rctr = new RowBlockContainer(); std::string data = "1 1:1 2:-1\n0 1:-1 2:1\n1 1:-1 2:-1\n0 1:1 2:1\n"; char* out_data = const_cast(data.c_str()); parser->CallParseBlock(out_data, out_data + data.size(), rctr); size_t num_row, num_col; CountDimensions(rctr, &num_row, &num_col); CHECK_EQ(num_row, 4U); CHECK_EQ(num_col, 3U); const std::vector expected_index{1, 2, 1, 2, 1, 2, 1, 2}; const std::vector expected_value{1, -1, -1, 1, -1, -1, 1, 1}; CHECK(rctr->index == expected_index); // perform element-wise comparsion CHECK(rctr->value == expected_value); } TEST(LibSVMParser, test_indexing_mode_1_based) { using namespace parser_test; InputSplit *source = nullptr; const std::map args{{"indexing_mode", "1"}}; std::unique_ptr> parser( new LibSVMParserTest(source, args, 1)); RowBlockContainer* rctr = new RowBlockContainer(); std::string data = "1 1:1 2:-1\n0 1:-1 2:1\n1 1:-1 2:-1\n0 1:1 2:1\n"; char* out_data = const_cast(data.c_str()); parser->CallParseBlock(out_data, out_data + data.size(), rctr); size_t num_row, num_col; CountDimensions(rctr, &num_row, &num_col); CHECK_EQ(num_row, 4U); CHECK_EQ(num_col, 2U); const std::vector expected_index{0, 1, 0, 1, 0, 1, 0, 1}; // with indexing_mode=1, parser will subtract 1 from each feature index const std::vector expected_value{1, -1, -1, 1, -1, -1, 1, 1}; CHECK(rctr->index == expected_index); // perform element-wise comparsion CHECK(rctr->value == expected_value); } TEST(LibSVMParser, test_indexing_mode_auto_detect) { using namespace parser_test; InputSplit *source = nullptr; const std::map args{{"indexing_mode", "-1"}}; std::unique_ptr> parser( new LibSVMParserTest(source, args, 1)); RowBlockContainer* rctr = new RowBlockContainer(); std::string data = "1 1:1 2:-1\n0 1:-1 2:1\n1 1:-1 2:-1\n0 1:1 2:1\n"; char* out_data = const_cast(data.c_str()); parser->CallParseBlock(out_data, out_data + data.size(), rctr); size_t num_row, num_col; CountDimensions(rctr, &num_row, &num_col); CHECK_EQ(num_row, 4U); CHECK_EQ(num_col, 2U); const std::vector expected_index{0, 1, 0, 1, 0, 1, 0, 1}; // expect to detect 1-based indexing, since the least feature id is 1 const std::vector expected_value{1, -1, -1, 1, -1, -1, 1, 1}; CHECK(rctr->index == expected_index); // perform element-wise comparsion CHECK(rctr->value == expected_value); } TEST(LibSVMParser, test_indexing_mode_auto_detect_2) { using namespace parser_test; InputSplit *source = nullptr; const std::map args{{"indexing_mode", "-1"}}; std::unique_ptr> parser( new LibSVMParserTest(source, args, 1)); RowBlockContainer* rctr = new RowBlockContainer(); std::string data = "1 1:1 2:-1\n0 0:-2 1:-1 2:1\n1 1:-1 2:-1\n0 1:1 2:1\n"; char* out_data = const_cast(data.c_str()); parser->CallParseBlock(out_data, out_data + data.size(), rctr); size_t num_row, num_col; CountDimensions(rctr, &num_row, &num_col); CHECK_EQ(num_row, 4U); CHECK_EQ(num_col, 3U); const std::vector expected_index{1, 2, 0, 1, 2, 1, 2, 1, 2}; // expect to detect 0-based indexing, since the least feature id is 0 const std::vector expected_value{1, -1, -2, -1, 1, -1, -1, 1, 1}; CHECK(rctr->index == expected_index); // perform element-wise comparsion CHECK(rctr->value == expected_value); } TEST(LibFMParser, test_indexing_mode_0_based) { using namespace parser_test; InputSplit *source = nullptr; const std::map args; std::unique_ptr> parser( new LibFMParserTest(source, args, 1)); RowBlockContainer* rctr = new RowBlockContainer(); std::string data = "1 1:1:1 1:2:-1\n0 1:1:-1 2:2:1\n1 2:1:-1 1:2:-1\n0 2:1:1 2:2:1\n"; char* out_data = const_cast(data.c_str()); parser->CallParseBlock(out_data, out_data + data.size(), rctr); size_t num_row, num_col; CountDimensions(rctr, &num_row, &num_col); CHECK_EQ(num_row, 4U); CHECK_EQ(num_col, 3U); const std::vector expected_field{1, 1, 1, 2, 2, 1, 2, 2}; const std::vector expected_index{1, 2, 1, 2, 1, 2, 1, 2}; const std::vector expected_value{1, -1, -1, 1, -1, -1, 1, 1}; CHECK(rctr->field == expected_field); CHECK(rctr->index == expected_index); CHECK(rctr->value == expected_value); // perform element-wise comparsion } TEST(LibFMParser, test_indexing_mode_1_based) { using namespace parser_test; InputSplit *source = nullptr; const std::map args{{"indexing_mode", "1"}}; std::unique_ptr> parser( new LibFMParserTest(source, args, 1)); RowBlockContainer* rctr = new RowBlockContainer(); std::string data = "1 1:1:1 1:2:-1\n0 1:1:-1 2:2:1\n1 2:1:-1 1:2:-1\n0 2:1:1 2:2:1\n"; char* out_data = const_cast(data.c_str()); parser->CallParseBlock(out_data, out_data + data.size(), rctr); size_t num_row, num_col; CountDimensions(rctr, &num_row, &num_col); CHECK_EQ(num_row, 4U); CHECK_EQ(num_col, 2U); const std::vector expected_field{0, 0, 0, 1, 1, 0, 1, 1}; const std::vector expected_index{0, 1, 0, 1, 0, 1, 0, 1}; // with indexing_mode=1, parser will subtract 1 from field/feature indices const std::vector expected_value{1, -1, -1, 1, -1, -1, 1, 1}; CHECK(rctr->field == expected_field); CHECK(rctr->index == expected_index); CHECK(rctr->value == expected_value); // perform element-wise comparsion } TEST(LibFMParser, test_indexing_mode_auto_detect) { using namespace parser_test; InputSplit *source = nullptr; const std::map args{{"indexing_mode", "-1"}}; std::unique_ptr> parser( new LibFMParserTest(source, args, 1)); RowBlockContainer* rctr = new RowBlockContainer(); std::string data = "1 1:1:1 1:2:-1\n0 1:1:-1 2:2:1\n1 2:1:-1 1:2:-1\n0 2:1:1 2:2:1\n"; char* out_data = const_cast(data.c_str()); parser->CallParseBlock(out_data, out_data + data.size(), rctr); size_t num_row, num_col; CountDimensions(rctr, &num_row, &num_col); CHECK_EQ(num_row, 4U); CHECK_EQ(num_col, 2U); const std::vector expected_field{0, 0, 0, 1, 1, 0, 1, 1}; const std::vector expected_index{0, 1, 0, 1, 0, 1, 0, 1}; // expect to detect 1-based indexing, since all field/feature id's exceed 0 const std::vector expected_value{1, -1, -1, 1, -1, -1, 1, 1}; CHECK(rctr->field == expected_field); CHECK(rctr->index == expected_index); CHECK(rctr->value == expected_value); // perform element-wise comparsion } TEST(LibFMParser, test_indexing_mode_auto_detect_2) { using namespace parser_test; InputSplit *source = nullptr; const std::map args{{"indexing_mode", "-1"}}; std::unique_ptr> parser( new LibFMParserTest(source, args, 1)); RowBlockContainer* rctr = new RowBlockContainer(); std::string data = "1 1:1:1 1:2:-1\n0 0:0:-2 1:1:-1 2:2:1\n1 2:1:-1 1:2:-1\n0 2:1:1 2:2:1\n"; char* out_data = const_cast(data.c_str()); parser->CallParseBlock(out_data, out_data + data.size(), rctr); size_t num_row, num_col; CountDimensions(rctr, &num_row, &num_col); CHECK_EQ(num_row, 4U); CHECK_EQ(num_col, 3U); const std::vector expected_field{1, 1, 0, 1, 2, 2, 1, 2, 2}; const std::vector expected_index{1, 2, 0, 1, 2, 1, 2, 1, 2}; // expect to detect 0-based indexing, since second row has feature id 0 const std::vector expected_value{1, -1, -2, -1, 1, -1, -1, 1, 1}; CHECK(rctr->field == expected_field); CHECK(rctr->index == expected_index); CHECK(rctr->value == expected_value); // perform element-wise comparsion }