Add datasets info for tests

src/tests/all_tables_config.cpp

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

src/tests/all_tables_config.h

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

src/tests/table_config.h

@@ -0,0 +1,41 @@
+#pragma once
+
+#include <filesystem>
+#include <string_view>
+#include <vector>
+
+#include "config/tabular_data/input_table_type.h"
+#include "parser/csv_parser/csv_parser.h"
+
+namespace tests {
+
+static auto const test_data_dir = std::filesystem::current_path() / "input_data";
+
+/// csv table configuration info to create an input table
+struct TableConfig {
+    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);
+    }
+};
+
+/// a struct consisting of a table config and the expected hash
+struct TableConfigHash {
+    TableConfig config;
+    size_t hash;
+};
+
+/// a struct consisting of a tables config and the expected hash
+struct TablesConfigHash {
+    std::vector<TableConfig> configs;
+    size_t hash;
+};
+
+}  // namespace tests

src/tests/test_ac_algorithm.cpp

@@ -7,7 +7,7 @@
 #include "algorithms/algebraic_constraints/bin_operation_enum.h"
 #include "algorithms/algo_factory.h"
 #include "config/names.h"
-#include "datasets.h"
+#include "table_config.h"
 #include "types.h"
 
 namespace {

src/tests/test_algo_interfaces.cpp

@@ -9,7 +9,7 @@
 #include "algorithms/fd/pyro/pyro.h"
 #include "config/error/type.h"
 #include "config/names.h"
-#include "datasets.h"
+#include "table_config.h"
 
 namespace tests {
 

src/tests/test_apriori.cpp

@@ -5,7 +5,7 @@
 #include "algorithms/algo_factory.h"
 #include "algorithms/association_rules/apriori.h"
 #include "config/names.h"
-#include "datasets.h"
+#include "table_config.h"
 
 namespace fs = std::filesystem;
 

src/tests/test_cfd_algos.cpp

@@ -9,7 +9,7 @@
 #include "algorithms/cfd/enums.h"
 #include "algorithms/cfd/fd_first_algorithm.h"
 #include "config/names.h"
-#include "datasets.h"
+#include "table_config.h"
 
 namespace tests {
 namespace fs = std::filesystem;

src/tests/test_cfd_relation_data.cpp

@@ -4,8 +4,8 @@
 #include <gtest/gtest.h>
 
 #include "algorithms/cfd/model/cfd_relation_data.h"
-#include "datasets.h"
 #include "parser/csv_parser/csv_parser.h"
+#include "table_config.h"
 
 namespace tests {
 

src/tests/test_data_stats.cpp

@@ -3,7 +3,7 @@
 
 #include "algorithms/algo_factory.h"
 #include "algorithms/statistics/data_stats.h"
-#include "datasets.h"
+#include "table_config.h"
 
 namespace tests {
 namespace mo = model;

src/tests/test_algorithm.cpp → src/tests/test_fd_algorithm.cpp

@@ -14,15 +14,17 @@
 #include "algorithms/fd/hyfd/hyfd.h"
 #include "algorithms/fd/pyro/pyro.h"
 #include "algorithms/fd/tane/tane.h"
-#include "datasets.h"
 #include "model/table/relational_schema.h"
-#include "testing_utils.h"
+#include "table_config.h"
+#include "test_fd_util.h"
 
 using std::string, std::vector;
 using ::testing::ContainerEq, ::testing::Eq;
 
 namespace fs = std::filesystem;
 
+namespace tests {
+
 /* This is a test suite for algorithm verification. It should be possible to run these tests for any
  * algorithm that:
  * 1. extends FDAlgorithm
@@ -76,20 +78,20 @@ std::set<std::pair<std::vector<unsigned int>, unsigned int>> FDsToSet(std::list<
 TYPED_TEST_SUITE_P(AlgorithmTest);
 
 TYPED_TEST_P(AlgorithmTest, ThrowsOnEmpty) {
-    auto algorithm = TestFixture::CreateAndConfToLoad(test_data_dir / "TestEmpty.csv", ',', true);
+    auto algorithm = TestFixture::CreateAndConfToLoad(kTestEmpty);
     ASSERT_THROW(algorithm->LoadData(), std::runtime_error);
 }
 
 TYPED_TEST_P(AlgorithmTest, ReturnsEmptyOnSingleNonKey) {
-    auto algorithm = TestFixture::CreateAlgorithmInstance("TestSingleColumn.csv", ',', true);
+    auto algorithm = TestFixture::CreateAlgorithmInstance(kTestSingleColumn);
     algorithm->Execute();
     ASSERT_TRUE(algorithm->FdList().empty());
 }
 
 TYPED_TEST_P(AlgorithmTest, WorksOnLongDataset) {
     std::set<std::pair<std::vector<unsigned int>, unsigned int>> true_fd_collection{{{2}, 1}};
 
-    auto algorithm = TestFixture::CreateAlgorithmInstance("TestLong.csv", ',', true);
+    auto algorithm = TestFixture::CreateAlgorithmInstance(kTestLong);
     algorithm->Execute();
     ASSERT_TRUE(CheckFdListEquality(true_fd_collection, algorithm->FdList()));
 }
@@ -98,51 +100,25 @@ TYPED_TEST_P(AlgorithmTest, WorksOnWideDataset) {
     std::set<std::pair<std::vector<unsigned int>, unsigned int>> true_fd_collection{
             {{0}, 2}, {{0}, 4}, {{2}, 0}, {{2}, 4}, {{4}, 0}, {{4}, 2}, {{}, 1}, {{}, 3}};
 
-    auto algorithm = TestFixture::CreateAlgorithmInstance("TestWide.csv", ',', true);
+    auto algorithm = TestFixture::CreateAlgorithmInstance(kTestWide);
     algorithm->Execute();
     ASSERT_TRUE(CheckFdListEquality(true_fd_collection, algorithm->FdList()));
 }
 
 TYPED_TEST_P(AlgorithmTest, LightDatasetsConsistentHash) {
-    try {
-        for (auto const& dataset : LightDatasets::datasets_) {
-            auto algorithm = TestFixture::CreateAlgorithmInstance(dataset.name, dataset.separator,
-                                                                  dataset.has_header);
-            algorithm->Execute();
-            std::cout << dataset.name << std::endl;
-            EXPECT_EQ(algorithm->Fletcher16(), dataset.hash)
-                    << "FD collection hash changed for " << dataset.name;
-        }
-    } catch (std::runtime_error& e) {
-        std::cout << "Exception raised in test: " << e.what() << std::endl;
-        FAIL();
-    }
-    SUCCEED();
+    TestFixture::PerformConsistentHashTestOn(TestFixture::light_datasets_);
 }
 
 TYPED_TEST_P(AlgorithmTest, HeavyDatasetsConsistentHash) {
-    try {
-        for (auto const& dataset : HeavyDatasets::datasets_) {
-            auto algorithm = TestFixture::CreateAlgorithmInstance(dataset.name, dataset.separator,
-                                                                  dataset.has_header);
-            algorithm->Execute();
-            EXPECT_EQ(algorithm->Fletcher16(), dataset.hash)
-                    << "The new algorithm and Pyro yield different results at " << dataset.name;
-        }
-    } catch (std::runtime_error& e) {
-        std::cout << "Exception raised in test: " << e.what() << std::endl;
-        FAIL();
-    }
-    SUCCEED();
+    TestFixture::PerformConsistentHashTestOn(TestFixture::heavy_datasets_);
 }
 
 TYPED_TEST_P(AlgorithmTest, ConsistentRepeatedExecution) {
-    auto const path = test_data_dir / "WDC_astronomical.csv";
-    auto algorithm = TestFixture::CreateAlgorithmInstance(path, ',', true);
+    auto algorithm = TestFixture::CreateAlgorithmInstance(kWDC_astronomical);
     algorithm->Execute();
     auto first_res = FDsToSet(algorithm->FdList());
     for (int i = 0; i < 3; ++i) {
-        algos::ConfigureFromMap(*algorithm, TestFixture::GetParamMap(path, ',', true));
+        algos::ConfigureFromMap(*algorithm, TestFixture::GetParamMap(kWDC_astronomical));
         algorithm->Execute();
         ASSERT_TRUE(CheckFdListEquality(first_res, algorithm->FdList()));
     }
@@ -155,3 +131,5 @@ REGISTER_TYPED_TEST_SUITE_P(AlgorithmTest, ThrowsOnEmpty, ReturnsEmptyOnSingleNo
 using Algorithms = ::testing::Types<algos::Tane, algos::Pyro, algos::FastFDs, algos::DFD,
                                     algos::Depminer, algos::FDep, algos::FUN, algos::hyfd::HyFD>;
 INSTANTIATE_TYPED_TEST_SUITE_P(AlgorithmTest, AlgorithmTest, Algorithms);
+
+}  // namespace tests