Files
daggy/libdaggy/tests/unit_utilities.cpp
2022-01-28 10:23:21 -04:00

109 lines
3.6 KiB
C++

#include <algorithm>
#include <catch2/catch.hpp>
#include <chrono>
#include <filesystem>
#include <fstream>
#include <iomanip>
#include <iostream>
#include "daggy/Serialization.hpp"
#include "daggy/Utilities.hpp"
#include "daggy/executors/task/NoopTaskExecutor.hpp"
TEST_CASE("string_utilities", "[utilities_string]")
{
std::string test = "/this/is/{{A}}/test/{{A}}";
auto res = daggy::globalSub(test, "{{A}}", "hello");
REQUIRE(res == "/this/is/hello/test/hello");
}
TEST_CASE("generate_cart_values", "[cartesian_values]")
{
std::string testParams{
R"({"DATE": ["2021-05-06", "2021-05-07" ], "SOURCE": "name", "TYPE": ["a", "b", "c"]})"};
auto params = daggy::configFromJSON(testParams);
auto result = daggy::generateCartesianValues(params);
REQUIRE(result.size() == 6);
for (const auto& valset : result) {
REQUIRE(valset.size() == 3);
REQUIRE(valset.count("DATE") == 1);
REQUIRE(valset.count("SOURCE") == 1);
REQUIRE(valset.count("TYPE") == 1);
}
}
TEST_CASE("string_expansion", "[utilities][parameter_expansion]")
{
SECTION("Basic expansion")
{
std::string testParams{
R"({"DATE": ["2021-05-06", "2021-05-07" ], "SOURCE": "name", "TYPE": ["a", "b", "c"]})"};
auto params = daggy::configFromJSON(testParams);
std::vector<std::string> cmd{"/bin/echo", "{{DATE}}", "{{SOURCE}}",
"{{TYPE}}"};
auto allCommands = daggy::interpolateValues(cmd, params);
REQUIRE(allCommands.size() == 6);
}
SECTION("No expansion")
{
std::string testParams{
R"({"DATE": ["2021-05-06", "2021-05-07" ], "SOURCE": "name", "TYPE": ["a", "b", "c"]})"};
auto params = daggy::configFromJSON(testParams);
std::vector<std::string> cmd{"/bin/echo"};
auto allCommands = daggy::interpolateValues(cmd, params);
REQUIRE(allCommands.size() == 1);
}
SECTION("Skip over unused parameters")
{
std::string testParams{
R"({"DATE": ["2021-05-06", "2021-05-07" ], "SOURCE": "name", "TYPE": ["a", "b", "c"]})"};
auto params = daggy::configFromJSON(testParams);
std::vector<std::string> cmd{"/bin/echo", "{{DATE}}", "{{SOURCE}}"};
auto allCommands = daggy::interpolateValues(cmd, params);
// TYPE isn't used, so it's just |DATE| * |SOURCE|
REQUIRE(allCommands.size() == 2);
}
SECTION("Expand within a command part")
{
std::string testParams{
R"({"DATE": ["2021-05-06", "2021-05-07" ], "SOURCE": ["A", "B"], "TYPE": ["a", "b", "c"]})"};
auto params = daggy::configFromJSON(testParams);
std::vector<std::string> cmd{"/bin/touch", "{{DATE}}_{{SOURCE}}"};
auto result = daggy::interpolateValues(cmd, params);
// TYPE isn't used, so it's just |DATE| * |SOURCE|
REQUIRE(result.size() == 4);
}
}
TEST_CASE("expand_taskset", "[utilities][expand_taskset]")
{
daggy::executors::task::NoopTaskExecutor ex;
daggy::DAGSpec dagSpec;
std::string testParams{R"({"DATE": ["2021-05-06", "2021-05-07" ]})"};
dagSpec.taskConfig.variables = daggy::configFromJSON(testParams);
std::stringstream jsonTasks;
jsonTasks
<< R"({ "A": { "job": {"command": [ "/bin/cat", "/etc/passwd"]} )"
<< R"(, "children": ["C"], "isGenerator": true},)"
<< R"("C": { "job": {"command": [ "/bin/echo", "hello!"]} } })";
dagSpec.tasks = daggy::tasksFromJSON(jsonTasks.str());
REQUIRE(dagSpec.tasks.size() == 2);
REQUIRE(dagSpec.tasks["A"].children == std::unordered_set<std::string>{"C"});
dagSpec.tasks =
daggy::expandTaskSet(dagSpec.tasks, ex, dagSpec.taskConfig.variables);
REQUIRE(dagSpec.tasks.size() == 2);
REQUIRE(dagSpec.tasks["A_0"].children ==
std::unordered_set<std::string>{"C"});
}