localTube/Modules/Server/src/worker.cpp

//
// Created by s-har on 26.05.2021.
//
#include "worker.hpp"

#include <chrono>
#include <filesystem>
#include <functional>
#include <vector>

#include <boost/process.hpp>
#include <ctre.hpp>
#include <fmt/ostream.h>
#include <spdlog/spdlog.h>

using namespace std::chrono_literals;

void operator+=(std::vector<std::string>& vec, std::string const& s)
{
    vec.push_back(s);
}

namespace bp = boost::process;

std::string san_string_path(std::string name)
{
    std::string out;
    for (auto el : name)
    {
        if ((el >= 'a' && el <= 'z') || (el >= 'A' && el <= 'Z') || (el >= '0' && el <= '9') ||
            el == '_' || el == '.')
        {
            out += el;
        }
        else if(el == ' ')
        {
            out += '_';
        }
    }
    return out;
}

Worker::Worker(soci::session& sql) : thread(std::bind_front(&Worker::loop, this)), sql(sql)
{
    spdlog::info("Worker started.");
}

[[noreturn]] void Worker::loop()
{
    while (true)
    {
        Task task;
        if (!getTask(task))
        {
            std::this_thread::sleep_for(10s);
            continue;
        }

        spdlog::info("Worker Task: {}", task);

        spdlog::debug("Begin Download {}", task.id);
        task.status = FileStatus::DOWNLOADING;

        sql << "UPDATE Tasks SET status = :status WHERE id = :id", soci::use(task);

        if (download(task))
        {
            spdlog::debug("Download end {}", task.id);
            task.status = FileStatus::COMPLETE;
        } else
        {
            // Error
            task.status = FileStatus::DOWNLOAD_ERROR;
            spdlog::error("Download Error {}: ", task);
        }

        sql << "UPDATE Tasks SET status = :status WHERE id = :id", soci::use(task);
        std::this_thread::sleep_for(10s);
    }
}

bool Worker::getTask(Task& task)
{
    try
    {
        soci::rowset<Task> data = (sql.prepare << "SELECT * "
                                                  "FROM Tasks "
                                                  "WHERE status=:PENDING "
                                                  "ORDER BY timestamp LIMIT 1;",
                                   soci::use(static_cast<int>(FileStatus::PENDING), "PENDING"));

        for (auto const& el : data)
        {
            task = el;
            return true;
        }
    }
    catch (std::exception const& e)
    {
        spdlog::error("[Worker] GetTask Error: {}", e.what());
    }
    return false;
}

bool Worker::download(Task& task)
{
    bp::ipstream out;
    bp::ipstream err;

    /* ** args ** */
    std::vector<std::string> args;

    args += "-o";
    args += fmt::format("{}/%(title)s.%(ext)s", output_path);

    if (task.audio_only)
    {
        args += "--extract-audio";
    }

    if (task.format)
    {
        args += task.audio_only ? "--audio-format" : "-f";
        args += task.format.value();
    }

    args += "--restrict-filenames";
    args += task.url;

    /* ** Download ** */
    spdlog::debug("youtube-dl {}", fmt::join(args, " "));

#ifdef __linux__ 
    constexpr std::string_view youtube_dl = "youtube-dl";
#elif _WIN32
    constexpr std::string_view youtube_dl = "youtube-dl.exe";
#else

#endif

    bp::child c(std::string{youtube_dl}, bp::args(args), bp::std_out > out, bp::std_err > err);

    std::string destination_file;

    std::string line;
    while (c.running() || !out.eof() || !err.eof())
    {
        if (std::getline(out, line) && !line.empty())
        {
            spdlog::info("[youtube-dl] {}", line);

            if (auto [match, info, destination] =
                  ctre::match<R"(\[(.*)\] Destination: (.*))">(line);
                match)
            {
                destination_file = destination;
            }

            if(auto [match, info, destination] =
                ctre::match<R"regex(\[(.*)\] Merging formats into "(.*)")regex">(line);
              match)
            {
                destination_file = destination;
            }

        } else if (std::getline(err, line) && !line.empty())
        {
            if (auto [match, type, messages] = ctre::match<R"((.*): (.*))">(line); match)
            {
                if (type == "WARNING")
                    spdlog::warn("[youtube-dl] {}", messages);
            } else
            {
                spdlog::error("[youtube-dl] {}", line);
            }
        }
    }

    c.wait();
    auto exit_code = c.exit_code();

    if (exit_code != 0)
        return false;

    // Process Downloaded file
    namespace fs = std::filesystem;

    spdlog::debug("File: {}", destination_file);

    auto downloaded_path = std::filesystem::path(destination_file);
    spdlog::info(absolute(downloaded_path).string());

    auto san_filename = san_string_path(downloaded_path.filename().string());
    auto local_path = fs::path(output_path) / (task.hash + downloaded_path.extension().string());

    auto current_time = std::time(nullptr);

    try{
        fs::rename(fs::path(".") / downloaded_path, local_path );

        File file;
        file.id = task.id;
        file.filename = san_filename;
        file.local_path = local_path.string();
        file.hash = task.hash;
        file.format = downloaded_path.extension().string();
        file.timestamp = current_time;

        sql << "INSERT INTO Files (id, hash, timestamp, format, filename, local_path) "
               "values(:id, :hash, :timestamp, :format, :filename, :local_path) ", soci::use(file);
    }catch(std::exception const& e)
    {
        spdlog::error("File rename / insert error: {}", e.what());
        return false;
    }


    spdlog::debug("youtube-dl exit code {}", exit_code);
    return exit_code == 0;
}