Add a file with datasets info for fd and ucc tests

src/tests/all_datasets_info.cpp

@@ -0,0 +1,36 @@
+#include "all_datasets_info.h"
+
+namespace tests {
+extern DatasetInfo const kWDC_astronomical{"WDC_astronomical.csv", ',', true};
+extern DatasetInfo const kWDC_symbols{"WDC_symbols.csv", ',', true};
+extern DatasetInfo const kWDC_science{"WDC_science.csv", ',', true};
+extern DatasetInfo const kWDC_satellites{"WDC_satellites.csv", ',', true};
+extern DatasetInfo const kWDC_appearances{"WDC_appearances.csv", ',', true};
+extern DatasetInfo const kWDC_astrology{"WDC_astrology.csv", ',', true};
+extern DatasetInfo const kWDC_game{"WDC_game.csv", ',', true};
+extern DatasetInfo const kWDC_kepler{"WDC_kepler.csv", ',', true};
+extern DatasetInfo const kWDC_planetz{"WDC_planetz.csv", ',', true};
+extern DatasetInfo const kWDC_age{"WDC_age.csv", ',', true};
+extern DatasetInfo const kTestWide{"TestWide.csv", ',', true};
+extern DatasetInfo const kabalone{"abalone.csv", ',', false};
+extern DatasetInfo const kiris{"iris.csv", ',', false};
+extern DatasetInfo const kadult{"adult.csv", ';', false};
+extern DatasetInfo const kbreast_cancer{"breast_cancer.csv", ',', true};
+extern DatasetInfo const kCIPublicHighway10k{"CIPublicHighway10k.csv", ',', true};
+extern DatasetInfo const kneighbors10k{"neighbors10k.csv", ',', true};
+extern DatasetInfo const kneighbors50k{"neighbors50k.csv", ',', true};
+extern DatasetInfo const kneighbors100k{"neighbors100k.csv", ',', true};
+extern DatasetInfo const kCIPublicHighway700{"CIPublicHighway700.csv", ',', true};
+extern DatasetInfo const kEpicVitals{"EpicVitals.csv", '|', true};
+extern DatasetInfo const kEpicMeds{"EpicMeds.csv", '|', true};
+extern DatasetInfo const kiowa1kk{"iowa1kk.csv", ',', true};
+extern DatasetInfo const kfd_reduced_30{"fd-reduced-30.csv", ',', true};
+extern DatasetInfo const kflight_1k{"flight_1k.csv", ';', true};
+extern DatasetInfo const kplista_1k{"plista_1k.csv", ';', false};
+extern DatasetInfo const kletter{"letter.csv", ',', false};
+extern DatasetInfo const kCIPublicHighway{"CIPublicHighway.csv", ',', true};
+extern DatasetInfo const kLegacyPayors{"LegacyPayors.csv", '|', true};
+extern DatasetInfo const kTestEmpty{"TestEmpty.csv", ',', true};
+extern DatasetInfo const kTestSingleColumn{"TestSingleColumn.csv", ',', true};
+extern DatasetInfo const kTestLong{"TestLong.csv", ',', true};
+}  // namespace tests

src/tests/all_datasets_info.h

@@ -0,0 +1,38 @@
+#pragma once
+
+#include "datasets.h"
+
+namespace tests {
+extern DatasetInfo const kWDC_astronomical;
+extern DatasetInfo const kWDC_symbols;
+extern DatasetInfo const kWDC_science;
+extern DatasetInfo const kWDC_satellites;
+extern DatasetInfo const kWDC_appearances;
+extern DatasetInfo const kWDC_astrology;
+extern DatasetInfo const kWDC_game;
+extern DatasetInfo const kWDC_kepler;
+extern DatasetInfo const kWDC_planetz;
+extern DatasetInfo const kWDC_age;
+extern DatasetInfo const kTestWide;
+extern DatasetInfo const kabalone;
+extern DatasetInfo const kiris;
+extern DatasetInfo const kadult;
+extern DatasetInfo const kbreast_cancer;
+extern DatasetInfo const kCIPublicHighway10k;
+extern DatasetInfo const kneighbors10k;
+extern DatasetInfo const kneighbors50k;
+extern DatasetInfo const kneighbors100k;
+extern DatasetInfo const kCIPublicHighway700;
+extern DatasetInfo const kEpicVitals;
+extern DatasetInfo const kEpicMeds;
+extern DatasetInfo const kiowa1kk;
+extern DatasetInfo const kfd_reduced_30;
+extern DatasetInfo const kflight_1k;
+extern DatasetInfo const kplista_1k;
+extern DatasetInfo const kletter;
+extern DatasetInfo const kCIPublicHighway;
+extern DatasetInfo const kLegacyPayors;
+extern DatasetInfo const kTestEmpty;
+extern DatasetInfo const kTestSingleColumn;
+extern DatasetInfo const kTestLong;
+}  // namespace tests

src/tests/datasets.h

@@ -4,7 +4,10 @@
 #include <string>
 #include <vector>
 
-static const auto test_data_dir = std::filesystem::current_path() / "input_data";
+#include "config/tabular_data/input_table_type.h"
+#include "parser/csv_parser/csv_parser.h"
+
+static auto const test_data_dir = std::filesystem::current_path() / "input_data";
 
 struct Dataset {
     std::string name;
@@ -13,6 +16,20 @@ struct Dataset {
     bool has_header;
 };
 
+struct DatasetInfo {
+    std::string_view name;
+    char separator;
+    bool has_header;
+
+    std::filesystem::path GetPath() const {
+        return test_data_dir / name;
+    }
+
+    config::InputTable MakeInputTable() const {
+        return std::make_shared<CSVParser>(GetPath(), separator, has_header);
+    }
+};
+
 class LightDatasets {
 public:
     static inline const std::array<Dataset, 11> datasets_ = {