path: root/src/plugins/NyaaSi.cpp

#include "../../plugins/NyaaSi.hpp"
#include "../../include/Program.hpp"
#include "../../include/Storage.hpp"
#include "../../include/Notification.hpp"
#include "../../include/StringUtils.hpp"
#include "../../include/NetUtils.hpp"
#include "../../include/Utils.hpp"
#include <quickmedia/HtmlSearch.h>

namespace QuickMedia {
    // Return end of td tag, or std::string::npos
    static size_t find_td_get_value(const std::string &str, size_t start_index, size_t end_index, std::string &result) {
        size_t td_begin = str.find("<td", start_index);
        if(td_begin == std::string::npos || td_begin >= end_index)
            return std::string::npos;

        size_t td_end = str.find("</td>", td_begin + 3);
        if(td_end == std::string::npos || td_end >= end_index)
            return std::string::npos;

        size_t value_begin = str.find('>', td_begin + 3);
        if(value_begin == std::string::npos || value_begin >= td_end)
            return std::string::npos;

        result = str.substr(value_begin + 1, td_end - (value_begin + 1));
        return td_end + 5;
    }

    static bool is_only_numbers(const char *str, size_t size) {
        for(size_t i = 0; i < size; ++i) {
            if(str[i] < '0' || str[i] > '9')
                return false;
        }
        return true;
    }

    static std::shared_ptr<BodyItem> create_front_page_item(const std::string &title, const std::string &category) {
        auto body_item = BodyItem::create(title);
        body_item->url = category;
        return body_item;
    }

    void get_nyaa_si_categories(BodyItems &result_items) {
        result_items.push_back(create_front_page_item("All categories", "0_0"));
        result_items.push_back(create_front_page_item("Anime", "1_0"));
        result_items.push_back(create_front_page_item("    Anime - Music video", "1_1"));
        result_items.push_back(create_front_page_item("    Anime - English translated", "1_2"));
        result_items.push_back(create_front_page_item("    Anime - Non-english translated", "1_3"));
        result_items.push_back(create_front_page_item("    Anime - Raw", "1_4"));
        result_items.push_back(create_front_page_item("Audio", "2_0"));
        result_items.push_back(create_front_page_item("    Audio - Lossless", "2_1"));
        result_items.push_back(create_front_page_item("    Anime - Lossy", "2_2"));
        result_items.push_back(create_front_page_item("Literature", "3_0"));
        result_items.push_back(create_front_page_item("    Literature - English translated", "3_1"));
        result_items.push_back(create_front_page_item("    Literature - Non-english translated", "3_1"));
        result_items.push_back(create_front_page_item("    Literature - Raw", "3_3"));
        result_items.push_back(create_front_page_item("Live Action", "4_0"));
        result_items.push_back(create_front_page_item("    Live Action - English translated", "4_1"));
        result_items.push_back(create_front_page_item("    Live Action - Non-english translated", "4_3"));
        result_items.push_back(create_front_page_item("    Live Action - Idol/Promotional video", "4_2"));
        result_items.push_back(create_front_page_item("    Live Action - Raw", "4_4"));
        result_items.push_back(create_front_page_item("Pictures", "5_0"));
        result_items.push_back(create_front_page_item("    Pictures - Graphics", "5_1"));
        result_items.push_back(create_front_page_item("    Pictures - Photos", "5_2"));
        result_items.push_back(create_front_page_item("Software", "6_0"));
        result_items.push_back(create_front_page_item("    Software - Applications", "6_1"));
        result_items.push_back(create_front_page_item("    Software - Games", "6_2"));
    }

    void get_sukebei_categories(BodyItems &result_items) {
        result_items.push_back(create_front_page_item("All categories", "0_0"));
        result_items.push_back(create_front_page_item("Art", "1_0"));
        result_items.push_back(create_front_page_item("    Anime", "1_1"));
        result_items.push_back(create_front_page_item("    Doujinshi", "1_2"));
        result_items.push_back(create_front_page_item("    Games", "1_3"));
        result_items.push_back(create_front_page_item("    Manga", "1_4"));
        result_items.push_back(create_front_page_item("    Pictures", "1_5"));
        result_items.push_back(create_front_page_item("Real Life", "2_0"));
        result_items.push_back(create_front_page_item("    Photobooks and Pictures", "2_1"));
        result_items.push_back(create_front_page_item("    Videos", "2_2"));
    }

    static time_t nyaa_si_time_to_unix_time(const char *time_str) {
        int year = 0;
        int month = 0;
        int day = 0;
        int hour = 0;
        int minute = 0;
        sscanf(time_str, "%d-%d-%d %d:%d", &year, &month, &day, &hour, &minute);
        if(year == 0) return 0;
        
        struct tm time;
        memset(&time, 0, sizeof(time));
        time.tm_year = year - 1900;
        time.tm_mon = month - 1;
        time.tm_mday = day;
        time.tm_hour = hour;
        time.tm_min = minute;
        time.tm_sec = 0;
        return timegm(&time);
    }

    static const std::array<const char*, 10> sort_type_names = {
        "Size desc 🡇",
        "Uploaded date desc 🡇",
        "Seeders desc 🡇",
        "Leechers desc 🡇",
        "Downloads desc 🡇",

        "Size asc 🡅",
        "Uploaded date asc 🡅",
        "Seeders asc 🡅",
        "Leechers asc 🡅",
        "Downloads asc 🡅"
    };

    static const std::array<const char*, 10> sort_params = {
        "s=size&o=desc",
        "s=id&o=desc",
        "s=seeders&o=desc",
        "s=leechers&o=desc",
        "s=downloads&o=desc",

        "s=size&o=asc",
        "s=id&o=asc",
        "s=seeders&o=asc",
        "s=leechers&o=asc",
        "s=downloads&o=asc",
    };

    static std::shared_ptr<BodyItem> create_sort_body_item(const std::string &title, NyaaSiSortType sort_type) {
        auto body_item = BodyItem::create(title);
        body_item->userdata = (void*)sort_type;
        return body_item;
    }

    static void sort_page_create_body_items(Body *body, NyaaSiSortType sort_type) {
        for(size_t i = 0; i < sort_type_names.size(); ++i) {
            std::string prefix = "  ";
            if((NyaaSiSortType)i == sort_type)
                prefix = "* ";
            body->append_item(create_sort_body_item(prefix + sort_type_names[i], (NyaaSiSortType)i));
        }
    }

    // TODO: Also show the number of comments for each torrent. TODO: Optimize?
    // TODO: Show each field as seperate columns instead of seperating by |
    static SearchResult search_page(const std::string &domain, const std::string &list_url, const std::string &text, int page, NyaaSiSortType sort_type, BodyItems &result_items) {
        std::string full_url = "https://" + domain + "/?c=" + list_url + "&f=0&p=" + std::to_string(page) + "&q=";
        full_url += url_param_encode(text);
        full_url += std::string("&") + sort_params[(size_t)sort_type];

        std::string website_data;
        if(download_to_string(full_url, website_data, {}, true) != DownloadResult::OK)
            return SearchResult::NET_ERR;

        const bool is_sukebei = (domain == "sukebei.nyaa.si");

        size_t tbody_begin = website_data.find("<tbody>");
        if(tbody_begin == std::string::npos)
            return SearchResult::OK;

        size_t tbody_end = website_data.find("</tbody>", tbody_begin + 7);
        if(tbody_end == std::string::npos)
            return SearchResult::ERR;

        size_t index = tbody_begin + 7;
        while(index < tbody_end) {
            size_t tr_begin = website_data.find("<tr", index);
            if(tr_begin == std::string::npos || tr_begin >= tbody_end)
                break;

            size_t tr_end = website_data.find("</tr>", tr_begin + 3);
            if(tr_end == std::string::npos || tr_end >= tbody_end)
                return SearchResult::ERR;

            index = tr_begin + 3;

            bool is_trusted = false;
            bool is_remake = false;
            size_t tr_class_begin = website_data.find("class=\"", index);
            if(tr_class_begin != std::string::npos && tr_class_begin < tr_end) {
                size_t tr_class_end = website_data.find('"', tr_class_begin + 7);
                size_t class_length = tr_class_end - (tr_class_begin + 7);
                if(strncmp(website_data.c_str() + tr_class_begin + 7, "success", class_length) == 0)
                    is_trusted = true;
                else if(strncmp(website_data.c_str() + tr_class_begin + 7, "danger", class_length) == 0)
                    is_remake = true;
                index = tr_class_end + 1;
            }
            
            size_t category_begin = website_data.find("/?c=", index);
            if(category_begin == std::string::npos || category_begin >= tr_end)
                return SearchResult::ERR;

            size_t category_end = website_data.find('"', category_begin + 4);
            if(category_end == std::string::npos || category_end >= tr_end)
                return SearchResult::ERR;

            index = category_end + 1;
            size_t view_begin = website_data.find("/view/", index);
            if(view_begin == std::string::npos || view_begin >= tr_end)
                return SearchResult::ERR;

            size_t view_end = website_data.find('"', view_begin + 6);
            if(view_end == std::string::npos || view_end >= tr_end)
                return SearchResult::ERR;

            std::string view_url = website_data.substr(view_begin, view_end - view_begin);

            // Torrents with comments have two /view/, one for comments and one for the title
            if(!is_only_numbers(website_data.c_str() + view_begin + 6, view_end - (view_begin + 6))) {
                size_t view_begin2 = website_data.find("/view/", view_end + 1);
                if(view_begin2 == std::string::npos || view_begin2 >= tr_end)
                    return SearchResult::ERR;

                size_t view_end2 = website_data.find('"', view_begin2 + 6);
                if(view_end2 == std::string::npos || view_end2 >= tr_end)
                    return SearchResult::ERR;

                view_end = view_end2;
            }

            size_t title_begin = website_data.find('>', view_end + 1);
            if(title_begin == std::string::npos || title_begin >= tr_end)
                return SearchResult::ERR;
            size_t title_end = website_data.find("</a>", title_begin + 1);
            if(title_end == std::string::npos || title_end >= tr_end)
                return SearchResult::ERR;
            std::string title = website_data.substr(title_begin + 1, title_end - (title_begin + 1));
            html_unescape_sequences(title);
            title = strip(title);

            index = title_end + 4;
            size_t magnet_begin = website_data.find("magnet:?xt", index);
            if(magnet_begin == std::string::npos || magnet_begin >= tr_end)
                return SearchResult::ERR;

            size_t magnet_end = website_data.find('"', magnet_begin + 10);
            if(magnet_end == std::string::npos || magnet_end >= tr_end)
                return SearchResult::ERR;

            index = magnet_end + 1;
            std::string size;
            index = find_td_get_value(website_data, index, tr_end, size);
            if(index == std::string::npos)
                return SearchResult::ERR;

            std::string timestamp;
            index = find_td_get_value(website_data, index, tr_end, timestamp);
            if(index == std::string::npos)
                return SearchResult::ERR;

            std::string seeders;
            index = find_td_get_value(website_data, index, tr_end, seeders);
            if(index == std::string::npos)
                return SearchResult::ERR;

            std::string leechers;
            index = find_td_get_value(website_data, index, tr_end, leechers);
            if(index == std::string::npos)
                return SearchResult::ERR;

            std::string completed;
            index = find_td_get_value(website_data, index, tr_end, completed);
            if(index == std::string::npos)
                return SearchResult::ERR;

            index = tr_end + 5;

            std::string description = "Size: " + size + " | Published: " + unix_time_to_local_time_str(nyaa_si_time_to_unix_time(timestamp.c_str())) + " | Seeders: " + seeders + " | Leechers: " + leechers + " | Completed: " + completed;
            auto body_item = BodyItem::create(std::move(title));
            body_item->thumbnail_url = "https://" + domain + "/static/img/icons/" + (is_sukebei ? "sukebei" : "nyaa") + "/" + website_data.substr(category_begin + 4, category_end - (category_begin + 4)) + ".png";
            body_item->set_description(std::move(description));
            body_item->url = "https://" + domain + std::move(view_url);
            if(is_trusted)
                body_item->set_title_color(mgl::Color(43, 255, 47));
            else if(is_remake)
                body_item->set_title_color(mgl::Color(255, 45, 47));
            body_item->thumbnail_size = mgl::vec2i(80, 28);
            result_items.push_back(std::move(body_item));
        }

        return SearchResult::OK;
    }

    PluginResult NyaaSiCategoryPage::submit(const SubmitArgs &args, std::vector<Tab> &result_tabs) {
        std::string domain = is_sukebei ? "sukebei.nyaa.si" : "nyaa.si";

        BodyItems result_items;
        SearchResult search_result = search_page(domain, args.url, "", 1, NyaaSiSortType::UPLOAD_DATE_DESC, result_items);
        if(search_result != SearchResult::OK) return search_result_to_plugin_result(search_result);

        auto search_page = std::make_unique<NyaaSiSearchPage>(program, strip(args.title), args.url, std::move(domain));
        NyaaSiSearchPage *search_page_p = search_page.get();
        auto body = create_body();
        body->set_items(std::move(result_items));
        result_tabs.push_back(Tab{std::move(body), std::move(search_page), create_search_bar("Search...", 500)});

        auto sort_order_page_body = create_body();
        Body *sort_order_page_body_p = sort_order_page_body.get();
        sort_page_create_body_items(sort_order_page_body_p, NyaaSiSortType::UPLOAD_DATE_DESC);
        result_tabs.push_back(Tab{std::move(sort_order_page_body), std::make_unique<NyaaSiSortOrderPage>(program, sort_order_page_body_p, search_page_p), nullptr});
        return PluginResult::OK;
    }

    NyaaSiSearchPage::NyaaSiSearchPage(Program *program, std::string category_name, std::string category_id, std::string domain) :
        Page(program), category_name(std::move(category_name)), category_id(std::move(category_id)), domain(std::move(domain))
    {
        set_sort_type(NyaaSiSortType::UPLOAD_DATE_DESC);
    }

    SearchResult NyaaSiSearchPage::search(const std::string &str, BodyItems &result_items) {
        return search_page(domain, category_id, str, 1, sort_type, result_items);
    }

    PluginResult NyaaSiSearchPage::get_page(const std::string &str, int page, BodyItems &result_items) {
        return search_result_to_plugin_result(search_page(domain, category_id, str, 1 + page, sort_type, result_items));
    }

    struct ResultItemExtra {
        BodyItems *result_items;
        const std::string *domain;
    };

    PluginResult NyaaSiSearchPage::submit(const SubmitArgs &args, std::vector<Tab> &result_tabs) {
        size_t comments_start_index;
        std::string title;
        BodyItems result_items;

        auto torrent_item = BodyItem::create("💾 Download magnet");
        std::string magnet_url;
        std::string description;

        ResultItemExtra result_item_extra;
        result_item_extra.result_items = &result_items;
        result_item_extra.domain = &domain;

        std::string website_data;
        if(download_to_string(args.url, website_data, {}, true) != DownloadResult::OK)
            return PluginResult::NET_ERR;

        QuickMediaHtmlSearch html_search;
        int result = quickmedia_html_search_init(&html_search, website_data.c_str(), website_data.size());
        if(result != 0)
            goto cleanup;

        result = quickmedia_html_find_nodes_xpath(&html_search, "//h3[class='panel-title']",
            [](QuickMediaMatchNode *node, void *userdata) {
                std::string *title = (std::string*)userdata;
                QuickMediaStringView text = quickmedia_html_node_get_text(node);
                if(title->empty() && text.data) {
                    title->assign(text.data, text.size);
                }
                return 0;
            }, &title);

        if(result != 0)
            goto cleanup;

        if(title.empty()) {
            fprintf(stderr, "Error: nyaa.si: failed to get title\n");
            result = -1;
            goto cleanup;
        }

        result = quickmedia_html_find_nodes_xpath(&html_search, "//div[class='panel-body']//div[class='row']//a",
            [](QuickMediaMatchNode *node, void *userdata) {
                ResultItemExtra *item_data = (ResultItemExtra*)userdata;
                QuickMediaStringView href = quickmedia_html_node_get_attribute_value(node, "href");
                QuickMediaStringView text = quickmedia_html_node_get_text(node);
                if(item_data->result_items->empty() && href.data && text.data && href.size >= 6 && memcmp(href.data, "/user/", 6) == 0) {
                    auto body_item = BodyItem::create("");
                    body_item->set_description("Submitter: " + std::string(text.data, text.size));
                    body_item->url = "https://" + *item_data->domain + "/" + std::string(href.data, href.size);
                    item_data->result_items->push_back(std::move(body_item));
                }
                return 0;
            }, &result_item_extra);

        if(result != 0)
            goto cleanup;

        if(result_items.empty()) {
            auto body_item = BodyItem::create("");
            body_item->set_description("Submitter: Anonymous");
            result_items.push_back(std::move(body_item));
        }

        result_items.front()->set_title(title);

        result = quickmedia_html_find_nodes_xpath(&html_search, "//div[id='torrent-description']",
            [](QuickMediaMatchNode *node, void *userdata) {
                std::string *description = (std::string*)userdata;
                QuickMediaStringView text = quickmedia_html_node_get_text(node);
                if(description->empty() && text.data) {
                    std::string desc(text.data, text.size);
                    html_unescape_sequences(desc);
                    *description = std::move(desc);
                }
                return 0;
            }, &description);

        if(result != 0)
            goto cleanup;

        if(!description.empty())
            result_items.front()->set_description(result_items.front()->get_description() + "\nDescription:\n" + description);

        result = quickmedia_html_find_nodes_xpath(&html_search, "//div[class='container']//a",
            [](QuickMediaMatchNode *node, void *userdata) {
                std::string *magnet_url = (std::string*)userdata;
                QuickMediaStringView href = quickmedia_html_node_get_attribute_value(node, "href");
                if(magnet_url->empty() && href.data && href.size >= 8 && memcmp(href.data, "magnet:?", 8) == 0) {
                    magnet_url->assign(href.data, href.size);
                }
                return 0;
            }, &magnet_url);

        if(result != 0)
            goto cleanup;

        if(magnet_url.empty()) {
            fprintf(stderr, "Error: %s: failed to get magnet link\n", domain.c_str());
            result = -1;
            goto cleanup;
        }

        html_unescape_sequences(magnet_url);
        torrent_item->url = std::move(magnet_url);
        result_items.push_back(std::move(torrent_item));
        comments_start_index = result_items.size();

        result = quickmedia_html_find_nodes_xpath(&html_search, "//div[id='comments']//a",
            [](QuickMediaMatchNode *node, void *userdata) {
                auto *item_data = (BodyItems*)userdata;
                QuickMediaStringView href = quickmedia_html_node_get_attribute_value(node, "href");
                QuickMediaStringView text = quickmedia_html_node_get_text(node);
                if(href.data && text.data && href.size >= 6 && memcmp(href.data, "/user/", 6) == 0) {
                    auto body_item = BodyItem::create(std::string(text.data, text.size));
                    //body_item->url = "https://nyaa.si/" + std::string(href);
                    item_data->push_back(std::move(body_item));
                }
                return 0;
            }, &result_items);

        if(result != 0 || result_items.size() == comments_start_index)
            goto cleanup;

        BodyItemContext body_item_image_context;
        body_item_image_context.body_items = &result_items;
        body_item_image_context.index = comments_start_index;

        result = quickmedia_html_find_nodes_xpath(&html_search, "//div[id='comments']//img[class='avatar']",
            [](QuickMediaMatchNode *node, void *userdata) {
                auto *item_data = (BodyItemContext*)userdata;
                QuickMediaStringView src = quickmedia_html_node_get_attribute_value(node, "src");
                if(src.data && item_data->index < item_data->body_items->size()) {
                    (*item_data->body_items)[item_data->index]->thumbnail_url.assign(src.data, src.size);
                    (*item_data->body_items)[item_data->index]->thumbnail_size = mgl::vec2i(120, 120);
                    item_data->index++;
                }
                return 0;
            }, &body_item_image_context);

        if(result != 0)
            goto cleanup;

        body_item_image_context.index = comments_start_index;

        result = quickmedia_html_find_nodes_xpath(&html_search, "//div[id='comments']//div[class='comment-content']",
            [](QuickMediaMatchNode *node, void *userdata) {
                auto *item_data = (BodyItemContext*)userdata;
                QuickMediaStringView text = quickmedia_html_node_get_text(node);
                if(text.data && item_data->index < item_data->body_items->size()) {
                    std::string desc(text.data, text.size);
                    html_unescape_sequences(desc);
                    (*item_data->body_items)[item_data->index]->set_description(std::move(desc));
                    item_data->index++;
                }
                return 0;
            }, &body_item_image_context);

        cleanup:
        quickmedia_html_search_deinit(&html_search);
        if(result != 0)
            return PluginResult::ERR;

        auto body = create_body();
        body->set_items(std::move(result_items));
        result_tabs.push_back(Tab{std::move(body), std::make_unique<NyaaSiTorrentPage>(program), nullptr});
        return PluginResult::OK;
    }

    void NyaaSiSearchPage::set_sort_type(NyaaSiSortType sort_type) {
        this->sort_type = sort_type;
        title = category_name + " | " + sort_type_names[(size_t)sort_type];
        title.erase(title.end() - 5, title.end()); // Erase emoji character and space at the end. TODO: Remove this when tabs support emojis.
    }

    PluginResult NyaaSiSortOrderPage::submit(const SubmitArgs &args, std::vector<Tab>&) {
        const NyaaSiSortType sort_type = (NyaaSiSortType)(size_t)args.userdata;
        body->clear_items();
        sort_page_create_body_items(body, sort_type);
        search_page->set_sort_type(sort_type);
        search_page->needs_refresh = true;
        return PluginResult::OK;
    }

    PluginResult NyaaSiTorrentPage::submit(const SubmitArgs &submit_args, std::vector<Tab> &result_tabs) {
        (void)result_tabs;
        if(strncmp(submit_args.url.c_str(), "magnet:?", 8) == 0) {
            if(!is_program_executable_by_name("xdg-open")) {
                show_notification("QuickMedia", "xdg-utils which provides xdg-open needs to be installed to download torrents", Urgency::CRITICAL);
                return PluginResult::ERR;
            }
            const char *args[] = { "xdg-open", submit_args.url.c_str(), nullptr };
            exec_program_async(args, nullptr);
        }
        return PluginResult::OK;
    }
}