fold/tests/integration/test_scan_e2e.cpp

#include <clang/Tooling/CompilationDatabase.h>
#include <clang/Tooling/Tooling.h>

#include <catch2/catch_test_macros.hpp>
#include <filesystem>
#include <fstream>
#include <memory>
#include <nlohmann/json.hpp>
#include <sstream>
#include <string>
#include <string_view>
#include <tuple>
#include <vector>

#include "fold/detect/clone.hpp"
#include "fold/frontend/ast_visitor.hpp"
#include "fold/frontend/ndjson_keys.hpp"
#include "fold/report/finding.hpp"
#include "fold/report/sarif.hpp"

namespace {

// Materialize a compile_commands.json next to the fixtures at test time so the
// absolute paths inside the DB resolve correctly on the host machine. Returns
// the directory containing the generated DB.
std::string writeCompileDb(const std::string& fixtures_dir) {
    namespace fs = std::filesystem;
    fs::path dir = fs::path(fixtures_dir);

    nlohmann::json db = nlohmann::json::array();
    for (const auto& name : {"clones/exact.cpp", "clones/renamed.cpp", "clones/near_miss.cpp",
                             "clones/not_a_clone.cpp"}) {
        fs::path file = dir / name;
        db.push_back({
            {"directory", dir.string()},
            {"command", std::string("clang++ -std=c++20 -c ") + file.string()},
            {"file", file.string()},
        });
    }
    fs::path out = dir / "compile_commands.json";
    std::ofstream(out) << db.dump(2);
    return dir.string();
}

} // namespace

TEST_CASE("scan emits clusters for exact / renamed / near_miss but not for not_a_clone", "[e2e]") {
    // Catch2 wires command-line args through its singleton; for our test we
    // accept the fixtures path via env or fall back to the canonical location.
    const char* env = std::getenv("FOLD_FIXTURES_DIR");
    std::string fixtures = env ? env : "../fixtures";
    if (!std::filesystem::exists(fixtures + "/clones/exact.cpp")) {
        // Allow running from the meson build dir.
        fixtures = std::filesystem::current_path().string() + "/../fixtures";
    }
    REQUIRE(std::filesystem::exists(fixtures + "/clones/exact.cpp"));

    auto dir = writeCompileDb(fixtures);
    std::string err;
    auto db = clang::tooling::CompilationDatabase::autoDetectFromDirectory(dir, err);
    REQUIRE(db);
    auto files = db->getAllFiles();
    clang::tooling::ClangTool tool(*db, files);

    auto summaries = fold::collectFunctionSummaries(tool);
    REQUIRE(summaries.size() >= 8); // 4 fixture files * 2 funcs each

    fold::CloneDetectorOptions opts;
    opts.min_body_stmts = 1;
    auto findings = fold::detectClones(summaries, opts);

    // Expect three clusters: exact/exact, renamed/renamed, near_miss/near_miss.
    REQUIRE(findings.size() == 3);

    // SARIF should be valid JSON with a single run and three results.
    std::ostringstream sarif_buf;
    fold::writeSarif(sarif_buf, findings);
    auto sarif = nlohmann::json::parse(sarif_buf.str());
    REQUIRE(sarif["version"] == "2.1.0");
    REQUIRE(sarif["runs"].size() == 1);
    REQUIRE(sarif["runs"][0]["results"].size() == 3);
}

TEST_CASE("SARIF keeps stable rule metadata for structural findings", "[sarif]") {
    fold::Finding clone;
    clone.id = "clone-id";
    clone.kind = fold::FindingKind::Clone;
    clone.locations = {{"src/report/sarif.cpp", 1, 1, "clone"}};
    clone.message = "clone";
    clone.evidence.backend = "clang";

    fold::Finding complexity;
    complexity.id = "complexity-id";
    complexity.kind = fold::FindingKind::HighComplexity;
    complexity.locations = {{"src/report/sarif.cpp", 2, 1, "complexity"}};
    complexity.message = "complexity";

    std::ostringstream output;
    fold::writeSarif(output, {clone, complexity});
    const auto sarif = nlohmann::json::parse(output.str());
    const auto& driver = sarif["runs"][0]["tool"]["driver"];
    const auto& results = sarif["runs"][0]["results"];

    // fold lives only on git.trevon.dev; SARIF consumers follow this link.
    REQUIRE(driver["informationUri"] == "https://git.trevon.dev/trevon/fold");
    REQUIRE(results[0]["ruleId"] == "fold.clone");
    REQUIRE(results[0]["properties"]["backend"] == "clang");
    REQUIRE(results[1]["ruleId"] == "fold.high-complexity");

    const auto clone_rule =
        std::find_if(driver["rules"].begin(), driver["rules"].end(),
                     [](const auto& rule) { return rule["id"] == "fold.clone"; });
    REQUIRE(clone_rule != driver["rules"].end());
    REQUIRE((*clone_rule)["shortDescription"]["text"] == "Duplicate function (syntactic)");
}

TEST_CASE("filtered SARIF writer preserves an existing report-visible set", "[sarif]") {
    fold::Finding finding;
    finding.id = "already-filtered";
    finding.kind = fold::FindingKind::ProductionReachability;
    finding.message = "already filtered";

    std::ostringstream output;
    fold::writeFilteredSarif(output, {finding});
    const auto sarif = nlohmann::json::parse(output.str());
    REQUIRE(sarif["runs"][0]["results"].size() == 1);
    REQUIRE(sarif["runs"][0]["results"][0]["properties"]["id"] == "already-filtered");
}

TEST_CASE("SARIF expands canonical reachability witnesses into compatibility fields", "[sarif]") {
    fold::Finding finding;
    finding.id = "reachability-witness";
    finding.kind = fold::FindingKind::ProductionReachability;
    finding.message = "reachable under another configuration";
    finding.evidence.quality = fold::QualityEvidence{};
    finding.evidence.reachability = fold::ReachabilityEvidence{.witness = {"main", "feature"}};

    std::ostringstream output;
    fold::writeFilteredSarif(output, {finding});
    const auto properties =
        nlohmann::json::parse(output.str())["runs"][0]["results"][0]["properties"];
    REQUIRE(properties["pathTrace"] == nlohmann::json::array({"main", "feature"}));
    REQUIRE(properties["qualityEvidence"]["member_ids"] ==
            nlohmann::json::array({"main", "feature"}));
    REQUIRE(properties["reachabilityEvidence"]["witness"] ==
            nlohmann::json::array({"main", "feature"}));
}

TEST_CASE("SARIF declares every emitted foreign and security rule", "[sarif]") {
    std::vector<fold::Finding> findings;
    for (const auto [kind, id, line] :
         {std::tuple{fold::FindingKind::FormatString, "format", 10},
          std::tuple{fold::FindingKind::SwiftUnsafeCast, "swift", 11},
          std::tuple{fold::FindingKind::UnknownForeign, "foreign", 12}}) {
        fold::Finding finding;
        finding.id = id;
        finding.kind = kind;
        finding.locations = {{"src/report/sarif.cpp", line, 1, id}};
        finding.message = id;
        findings.push_back(finding);
    }

    std::ostringstream output;
    fold::writeSarif(output, findings);
    const auto sarif = nlohmann::json::parse(output.str());
    const auto& driver_rules = sarif["runs"][0]["tool"]["driver"]["rules"];
    for (const auto& result : sarif["runs"][0]["results"]) {
        const auto rule =
            std::find_if(driver_rules.begin(), driver_rules.end(), [&](const auto& candidate) {
                return candidate["id"] == result["ruleId"];
            });
        REQUIRE(rule != driver_rules.end());
    }
}

TEST_CASE("frontend NDJSON finding envelope key remains stable", "[frontend][wire-format]") {
    REQUIRE(std::string_view(fold::frontend::ndjson_keys::kKindFinding) == "finding");
}