mason/src2/download.cpp

#include "download.h"

#include <regex>

#include "collection.h"

constexpr std::string_view downloads("downloads");
constexpr std::string_view headerAcceptJson("accept: application/json");

static const std::regex repo("(^https?):\\/\\/([\\w\\d\\.\\-\\_]+)\\/([\\w\\d\\-\\_]+)\\/([\\w\\d\\-\\_]+)");

unsigned int Download::instances(0);
AtomicMutex Download::amx{};

void Download::CurlDeleter::operator()(CURL* curl) const {
    curl_easy_cleanup(curl);
}

void Download::FileDeleter::operator()(FILE* file) const {
    fclose(file);
}


Download::Download(
    const std::string& url,
    const std::filesystem::path& destination,
    const std::shared_ptr<Collection>& collection,
    const std::shared_ptr<Logger>& logger
) :
    Loggable(logger),
    curl(),
    collection(collection),
    url(url),
    destination(destination)
{
    std::lock_guard lock(amx);
    if (instances == 0) {
        CURLcode res = curl_global_init(CURL_GLOBAL_ALL);
        if (res != CURLE_OK)
            throw CurlError(std::string("Error initializing curl global ") + curl_easy_strerror(res));
    }

    ++instances;

    createCurl();
}

Download::~Download() {
    std::lock_guard lock(amx);
    --instances;
    if (instances == 0)
        curl_global_cleanup();
}


void Download::proceed() {
    bool success = false;
    std::optional<RepoInfo> repo = testRepo();
    if (repo.has_value()) {
        std::optional<std::filesystem::path> path = downloadAsRepo(repo.value());
    }
}

std::optional<nlohmann::json> Download::repoInfoGiteaApi1(const RepoInfo& repo)  {
    std::string url = repo.origin() + "/api/v1/repos/" + repo.project();
    std::string data;
    CURLcode code = httpGet(url, data, {headerAcceptJson});
    if (code == CURLE_OK) {
        try {
            return nlohmann::json::parse(data);
        } catch (const nlohmann::json::exception& e) {
            warn(e.what());
        }
    }

    return std::nullopt;
}

std::optional<std::filesystem::path> Download::downloadAsRepo(const RepoInfo& repo) {
    info("Trying Gitea v1 API");
    std::optional<nlohmann::json> repoInfo = repoInfoGiteaApi1(repo);
    if (repoInfo.has_value()) {
        nlohmann::json::const_iterator itr = repoInfo.value().find("default_branch");
        if (itr != repoInfo.value().end() && itr->is_string()) {
            std::string branchName = *itr;
            info("Gitea v1 API seem to have worked");
            info("Default branch is " + branchName);

            std::optional<std::filesystem::path> path = downloadRepoGiteaApi1(repo, branchName);
        }
    }

    return std::nullopt;
}

std::optional<std::filesystem::path> Download::downloadRepoGiteaApi1(const RepoInfo& repo, const std::string& branch) {

    std::string fileName = branch + ".tar.gz";
    std::string url = repo.origin() + "/api/v1/repos/" + repo.project() + "/archive/" + fileName;
    std::filesystem::path path = destination/downloads/fileName;
    CURLcode code = httpDownload(url, path);
    if (code == CURLE_OK)
        return path;

    warn("Couldn't download file " + url + ": " + curl_easy_strerror(code));
    minor("Removing " + path.string());
    if (std::filesystem::exists(path))
        std::filesystem::remove_all(path);

    return std::nullopt;
}


CURLcode Download::httpGet(
    const std::string& url,
    std::string& result,
    const std::vector<std::string_view>& headers)
{
    setHeaders(headers);

    curl_easy_setopt(curl.get(), CURLOPT_URL, url.c_str());
    curl_easy_setopt(curl.get(), CURLOPT_WRITEFUNCTION, writeString);
    curl_easy_setopt(curl.get(), CURLOPT_WRITEDATA, &result);

    return curl_easy_perform(curl.get());
}

CURLcode Download::httpDownload(const std::string& url, const std::filesystem::path& path, const std::vector<std::string_view>& headers) {
    std::filesystem::create_directories(path.parent_path());
    if (std::filesystem::exists(path)) {
        minor("File " + path.string() + " already exists, will be overwritten");

        if (std::filesystem::is_directory(path))
            std::filesystem::remove_all(path);
    }

    std::unique_ptr<FILE, FileDeleter> filePtr(fopen(path.c_str(), "wb"));
    if (!filePtr)
        throw FileError("Could not open file " + path.string() + " to write");

    setHeaders(headers);
    curl_easy_setopt(curl.get(), CURLOPT_URL, url.c_str());
    curl_easy_setopt(curl.get(), CURLOPT_WRITEFUNCTION, writeFile);
    curl_easy_setopt(curl.get(), CURLOPT_WRITEDATA, filePtr.get());

    return curl_easy_perform(curl.get());

    //     return res;
    //     if (res != CURLE_OK) {
    //         Project::warn("Couldn't download file " + url + ": " + curl_easy_strerror(res));
    //     } else {
    //         uint32_t code;
    //         curl_easy_getinfo(curl, CURLINFO_RESPONSE_CODE, &code);
    //         if (code == 200)
    //             result = true;
    //         else
    //             Project::warn("Couldn't download file " + url + ": response code " + std::to_string(code));
    //     }
    // } else {
    //     Project::error(std::string("Couldn't open output file ") + destination.c_str());
    // }
}

void Download::setHeaders(const std::vector<std::string_view>& headers) {
    struct curl_slist* curlHeaders = nullptr;
    for (const std::string_view& header : headers)
        curlHeaders = curl_slist_append(curlHeaders, header.data());

    curl_easy_setopt(curl.get(), CURLOPT_HTTPHEADER, curlHeaders);
}

void Download::createCurl() {
    CURL* crl = curl_easy_init();
    if (!crl)
        throw CurlError("Error creating curl instalce");

	curl_easy_setopt(crl, CURLOPT_USERAGENT, "libcurl-agent/1.0" );
    curl_easy_setopt(crl, CURLOPT_VERBOSE, 1L);
    curl_easy_setopt(crl, CURLOPT_DEBUGFUNCTION, trace);
    curl_easy_setopt(crl, CURLOPT_DEBUGDATA, this);
    curl_easy_setopt(crl, CURLOPT_FOLLOWLOCATION, 1L);
    curl_easy_setopt(crl, CURLOPT_NOPROGRESS, 1L);
    curl = std::unique_ptr<CURL, CurlDeleter>(crl);
}


std::optional<Download::RepoInfo> Download::testRepo() const {
    std::smatch results;
    if (std::regex_search(url, results, repo)) {
        info(url + " appears to be a git repository");
        return RepoInfo {
            results[1],
            results[2],
            results[3],
            results[4]
        };
    }

    return std::nullopt;
}

Download::CurlError::CurlError(const std::string& message) :
    std::runtime_error(message) {}

Download::FileError::FileError(const std::string& message) :
    std::runtime_error(message) {}

std::string Download::RepoInfo::origin() const {
    return protocol + "://" + host;
}

std::string Download::RepoInfo::project() const {
    return owner + "/" + name;
}