From 93b1123581c5aab25d5bdcf68fa8e91406170acf Mon Sep 17 00:00:00 2001
From: Michael Polyntsov <Arno9148@gmail.com>
Date: Sun, 13 Mar 2022 02:03:55 +0300
Subject: [PATCH] Implement tests for ColumnTypedData

---
 tests/TestTypedColumnData.cpp   | 103 ++++++++++++++++++++++++++++++++
 tests/inputData/SimpleTypes.csv |  10 ++++
 2 files changed, 113 insertions(+)
 create mode 100644 tests/TestTypedColumnData.cpp
 create mode 100644 tests/inputData/SimpleTypes.csv

diff --git a/tests/TestTypedColumnData.cpp b/tests/TestTypedColumnData.cpp
new file mode 100644
index 0000000000..2a81f3d5f8
--- /dev/null
+++ b/tests/TestTypedColumnData.cpp
@@ -0,0 +1,103 @@
+#include <filesystem>
+#include <memory>
+
+#include <gmock/gmock.h>
+#include <gtest/gtest.h>
+
+#include "ColumnLayoutTypedRelationData.h"
+#include "CSVParser.h"
+
+namespace tests {
+
+namespace fs = std::filesystem;
+namespace mo = model;
+
+using mo::TypeId;
+
+struct TypeParsingParams {
+    std::vector<mo::TypeId> expected;
+    std::string_view dataset;
+    char const sep;
+    bool const has_header;
+    TypeParsingParams(std::vector<mo::TypeId> expected,
+                      std::string_view dataset,
+                      char const sep = ',',
+                      bool const has_header = true) noexcept
+        : expected(std::move(expected)), dataset(dataset),
+          sep(sep), has_header(has_header) {}
+};
+
+class TestTypeParsing : public ::testing::TestWithParam<TypeParsingParams> {};
+
+static inline std::vector<mo::TypedColumnData> CreateColumnData(std::string_view data, char sep,
+                                                                bool has_header) {
+    auto const path = fs::current_path() / "inputData" / data;
+    CSVParser input_generator(path, sep, has_header);
+    std::unique_ptr<model::ColumnLayoutTypedRelationData> relation_data =
+        model::ColumnLayoutTypedRelationData::CreateFrom(input_generator, true, -1, -1);
+    std::vector<mo::TypedColumnData> col_data = std::move(relation_data->GetColumnData());
+    return col_data;
+}
+
+TEST_P(TestTypeParsing, DefaultTest) {
+    TypeParsingParams const& p = GetParam();
+    std::vector<mo::TypedColumnData> column_data = CreateColumnData(p.dataset, p.sep, p.has_header);
+
+    ASSERT_EQ(column_data.size(), p.expected.size());
+
+    size_t i = 0;
+    for (mo::TypedColumnData const& col_data : column_data) {
+        EXPECT_EQ(col_data.GetTypeId(), p.expected[i]) << "Column index: " << i;
+        ++i;
+    }
+}
+
+INSTANTIATE_TEST_SUITE_P(
+    TypeSystem, TestTypeParsing,
+    ::testing::Values(
+        TypeParsingParams({TypeId::kString, TypeId::kMixed, TypeId::kMixed},
+                          "WDC_appearances.csv"),
+        TypeParsingParams({TypeId::kString, TypeId::kString, TypeId::kString},
+                          "WDC_age.csv"),
+        TypeParsingParams({TypeId::kString, TypeId::kMixed, TypeId::kMixed,
+                           TypeId::kMixed},
+                          "WDC_kepler.csv"),
+        TypeParsingParams({TypeId::kString, TypeId::kString, TypeId::kMixed,
+                           TypeId::kMixed, TypeId::kMixed, TypeId::kString,
+                           TypeId::kString, TypeId::kString},
+                          "WDC_satellites.csv"),
+        TypeParsingParams({TypeId::kString, TypeId::kString, TypeId::kInt,
+                           TypeId::kInt, TypeId::kInt, TypeId::kInt,
+                           TypeId::kInt, TypeId::kUndefined, TypeId::kUndefined,
+                           TypeId::kUndefined, TypeId::kUndefined, TypeId::kInt,
+                           TypeId::kInt, TypeId::kInt, TypeId::kInt,
+                           TypeId::kInt, TypeId::kUndefined, TypeId::kUndefined},
+                          "CIPublicHighway700.csv"),
+        TypeParsingParams({TypeId::kInt, TypeId::kInt, TypeId::kMixed,
+                           TypeId::kInt, TypeId::kInt, TypeId::kInt,
+                           TypeId::kInt},
+                          "neighbors10k.csv"),
+        TypeParsingParams({TypeId::kDouble, TypeId::kDouble, TypeId::kDouble,
+                           TypeId::kDouble, TypeId::kString},
+                          "iris.csv", ',', false),
+        TypeParsingParams({TypeId::kUndefined,TypeId::kUndefined,TypeId::kUndefined,
+                           TypeId::kInt, TypeId::kString, TypeId::kDouble,
+                           TypeId::kBigInt, TypeId::kMixed, TypeId::kBigInt,
+                           TypeId::kMixed, TypeId::kInt},
+                          "SimpleTypes.csv")));
+
+TEST(TypeSystem, SumColumnDoubles) {
+    std::vector<mo::TypedColumnData> col_data = CreateColumnData("iris.csv", ',', false);
+    ASSERT_EQ(col_data.size(), 5);
+    mo::TypedColumnData const& col = col_data.front();
+    ASSERT_EQ(col.GetTypeId(), static_cast<TypeId>(TypeId::kDouble));
+    mo::INumericType const& type = static_cast<mo::INumericType const&>(col.GetType());
+    std::unique_ptr<std::byte[]> sum(type.Allocate());
+    for (std::byte const* value : col.GetData()) {
+        type.Add(sum.get(), value, sum.get());
+    }
+    mo::Double expected = 876.5;
+    EXPECT_DOUBLE_EQ(type.GetValue<mo::Double>(sum.get()), expected);
+}
+
+}  // namespace tests
diff --git a/tests/inputData/SimpleTypes.csv b/tests/inputData/SimpleTypes.csv
new file mode 100644
index 0000000000..9971846276
--- /dev/null
+++ b/tests/inputData/SimpleTypes.csv
@@ -0,0 +1,10 @@
+Undefined1,Undefined2,Undefined3,Int,String,Double,BigInt,IntAndDouble,BigIntWithInt,JustMixed,Int1
+NULL,NULL,,12341,asdb,3.142,33333333333333333333,111111,1,123,3123
+NULL,NULL,,12312341,justs0mestring,3.,33333333333173333333,3.123141,2,123.,2
+,NULL,,0,"some string within quotes",-3.142,82374045691542307589,3.,3,+202,,
+NULL,NULL,,-1,"some string\" with quote",-2312.,100000000000000000000,1111,4,0,NULL
+,NULL,,-10101,something,8321937193.0,NULL,7,170317031703170317031703,NULL,NULL
+,NULL,,1111111111111111111,"something with \n",1.1,98765432112345678909876,8,6,suddenly,NULL
+NULL,NULL,,17238917293719,le le le le le,2.0,170317031703170317031703,9,7,bigint,3
+NULL,NULL,,+123123828282,31a231234,+3.,1337733113377331133773311337,10,8,smallint,+3
+,NULL,,1337,sssssnake,+3.89,30000000000000000004,11,9,,-11