path: root/src/plugins/MediaGeneric.cpp

#include "../../plugins/MediaGeneric.hpp"
#include "../../include/StringUtils.hpp"
#include <quickmedia/HtmlSearch.h>

namespace QuickMedia {
    static bool string_view_contains(const QuickMediaStringView str, const char *sub) {
        return memmem(str.data, str.size, sub, strlen(sub));
    }

    using HtmlPathCallback = std::function<void(QuickMediaMatchNode*)>;
    static int quickmedia_html_find_nodes_xpath(QuickMediaHtmlSearch *self, const char *xpath, HtmlPathCallback callback) {
        return quickmedia_html_find_nodes_xpath(self, xpath, [](QuickMediaMatchNode *node, void *userdata) {
            HtmlPathCallback *callback = (HtmlPathCallback*)userdata;
            (*callback)(node);
            return 0;
        }, &callback);
    }

    static QuickMediaStringView html_attr_or_inner_text(QuickMediaMatchNode *node, const char *field_name) {
        if(strcmp(field_name, "text") == 0)
            return quickmedia_html_node_get_text(node);
        else
            return quickmedia_html_node_get_attribute_value(node, field_name);
    }

    static void body_items_prepend_website_url(BodyItems &body_items, const std::string &website_url) {
        for(auto &body_item : body_items) {
            if(string_starts_with(body_item->url, "//"))
                body_item->url = "https://" + body_item->url.substr(2);
            else if(string_starts_with(body_item->url, "/"))
                body_item->url = website_url + body_item->url.substr(1);

            if(string_starts_with(body_item->thumbnail_url, "//"))
                body_item->thumbnail_url = "https://" + body_item->thumbnail_url.substr(2);
            else if(string_starts_with(body_item->thumbnail_url, "/"))
                body_item->thumbnail_url = website_url + body_item->thumbnail_url.substr(1);
        }
    }

    static PluginResult fetch_page_results(
        const std::string &url, const std::string &website_url, const std::vector<MediaTextQuery> &text_queries,
        const std::vector<MediaThumbnailQuery> &thumbnail_queries, mgl::vec2i thumbnail_max_size,
        MediaRelatedCustomHandler *related_custom_handler,
        BodyItems &result_items, bool cloudflare_bypass, const std::vector<CommandArg> &extra_commands)
    {
        std::vector<CommandArg> args = extra_commands;
        if(!website_url.empty())
            args.push_back({ "--header", "referer: " + website_url });

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

        if(website_data.empty())
            return PluginResult::OK;

        if(related_custom_handler && *related_custom_handler) {
            std::vector<MediaRelatedItem> media_related_items = (*related_custom_handler)(website_data);
            for(MediaRelatedItem &media_related_item : media_related_items) {
                auto body_item = BodyItem::create(media_related_item.title);
                body_item->url = std::move(media_related_item.url);
                body_item->thumbnail_url = std::move(media_related_item.thumbnail_url);
                body_item->thumbnail_size = thumbnail_max_size;
                result_items.push_back(std::move(body_item));
            }
            body_items_prepend_website_url(result_items, website_url);
            return PluginResult::OK;
        }

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

        for(const MediaTextQuery &text_query : text_queries) {
            if(!text_query.html_query || !text_query.title_field) {
                assert(false);
                result = -1;
                goto cleanup;
            }

            result = quickmedia_html_find_nodes_xpath(&html_search, text_query.html_query, [&text_query, &result_items](QuickMediaMatchNode *node) {
                QuickMediaStringView title_value = html_attr_or_inner_text(node, text_query.title_field);
                QuickMediaStringView url_value = html_attr_or_inner_text(node, text_query.url_field);
                if(title_value.data && url_value.data && (!text_query.url_contains || string_view_contains(url_value, text_query.url_contains))) {
                    std::string field1_fixed(title_value.data, title_value.size);
                    html_unescape_sequences(field1_fixed);
                    auto item = BodyItem::create(std::move(field1_fixed));
                    item->url.assign(url_value.data, url_value.size);
                    result_items.push_back(std::move(item));
                }
            });
            if(result != 0)
                goto cleanup;
        }

        for(const MediaThumbnailQuery &thumbnail_query : thumbnail_queries) {
            assert(thumbnail_query.html_query && thumbnail_query.field_name);
            if(thumbnail_query.html_query && thumbnail_query.field_name) {
                size_t index = 0;
                result = quickmedia_html_find_nodes_xpath(&html_search, thumbnail_query.html_query, [&thumbnail_query, &result_items, &index, thumbnail_max_size](QuickMediaMatchNode *node) {
                    QuickMediaStringView field_value = html_attr_or_inner_text(node, thumbnail_query.field_name);
                    if(index < result_items.size() && field_value.data && (!thumbnail_query.field_contains || string_view_contains(field_value, thumbnail_query.field_contains))) {
                        result_items[index]->thumbnail_url.assign(field_value.data, field_value.size);
                        result_items[index]->thumbnail_size = thumbnail_max_size;
                        ++index;
                    }
                });
                if(result != 0)
                    goto cleanup;
            }
        }

        body_items_prepend_website_url(result_items, website_url);

        cleanup:
        quickmedia_html_search_deinit(&html_search);
        if(result == 0) {
            return PluginResult::OK;
        } else {
            result_items.clear();
            return PluginResult::ERR;
        }
    }

    MediaGenericSearchPage::MediaGenericSearchPage(Program *program, const char *website_url, mgl::vec2i thumbnail_max_size, bool cloudflare_bypass, std::vector<CommandArg> extra_commands) :
        Page(program), website_url(website_url ? website_url : ""), thumbnail_max_size(thumbnail_max_size), cloudflare_bypass(cloudflare_bypass), extra_commands(std::move(extra_commands))
    {
        if(!this->website_url.empty()) {
            if(this->website_url.back() != '/')
                this->website_url.push_back('/');
        }
    }

    SearchResult MediaGenericSearchPage::search(const std::string &str, BodyItems &result_items) {
        return plugin_result_to_search_result(get_page(str, 0, result_items));
    }

    PluginResult MediaGenericSearchPage::get_page(const std::string &str, int page, BodyItems &result_items) {
        std::string url = search_query.search_template;
        string_replace_all(url, "%s", url_param_encode(str));
        string_replace_all(url, "%p", std::to_string(search_query.page_start + page));
        return fetch_page_results(url, website_url, text_queries, thumbnail_queries, thumbnail_max_size, nullptr, result_items, cloudflare_bypass, extra_commands);
    }

    PluginResult MediaGenericSearchPage::submit(const SubmitArgs &args, std::vector<Tab> &result_tabs) {
        result_tabs.push_back(Tab{nullptr, std::make_unique<MediaGenericVideoPage>(program, this, args.url), nullptr});
        return PluginResult::OK;
    }

    PluginResult MediaGenericSearchPage::get_related_media(const std::string &url, BodyItems &result_items) {
        return fetch_page_results(url, website_url, related_media_text_queries, related_media_thumbnail_queries, thumbnail_max_size, &related_custom_handler, result_items, cloudflare_bypass, extra_commands);
    }

    MediaGenericSearchPage& MediaGenericSearchPage::search_handler(const char *search_template, int page_start) {
        search_query.search_template = search_template;
        search_query.page_start = page_start;
        return *this;
    }

    MediaGenericSearchPage& MediaGenericSearchPage::text_handler(std::vector<MediaTextQuery> queries) {
        text_queries = std::move(queries);
        return *this;
    }

    MediaGenericSearchPage& MediaGenericSearchPage::thumbnail_handler(std::vector<MediaThumbnailQuery> queries) {
        thumbnail_queries = std::move(queries);
        return *this;
    }

    MediaGenericSearchPage& MediaGenericSearchPage::video_url_custom_handler(MediaVideoUrlCustomHandler handler) {
        video_custom_handler = std::move(handler);
        return *this;
    }

    MediaGenericSearchPage& MediaGenericSearchPage::related_media_text_handler(std::vector<MediaTextQuery> queries) {
        related_media_text_queries = std::move(queries);
        related_custom_handler = nullptr;
        return *this;
    }

    MediaGenericSearchPage& MediaGenericSearchPage::related_media_thumbnail_handler(std::vector<MediaThumbnailQuery> queries) {
        related_media_thumbnail_queries = std::move(queries);
        related_custom_handler = nullptr;
        return *this;
    }

    MediaGenericSearchPage& MediaGenericSearchPage::related_media_custom_handler(MediaRelatedCustomHandler handler) {
        related_custom_handler = std::move(handler);
        related_media_text_queries.clear();
        related_media_thumbnail_queries.clear();
        return *this;
    }

    PluginResult MediaGenericRelatedPage::submit(const SubmitArgs &args, std::vector<Tab> &result_tabs) {
        result_tabs.push_back(Tab{nullptr, std::make_unique<MediaGenericVideoPage>(program, search_page, args.url), nullptr});
        return PluginResult::OK;
    }

    BodyItems MediaGenericVideoPage::get_related_media(const std::string &url) {
        BodyItems result_items;
        search_page->get_related_media(url, result_items);
        return result_items;
    }

    PluginResult MediaGenericVideoPage::get_related_pages(const BodyItems &related_videos, const std::string&, std::vector<Tab> &result_tabs) {
        auto related_page_body = create_body(false, true);
        related_page_body->set_items(related_videos);
        result_tabs.push_back(Tab{std::move(related_page_body), std::make_unique<MediaGenericRelatedPage>(program, search_page), create_search_bar("Search...", SEARCH_DELAY_FILTER)});
        return PluginResult::OK;
    }

    std::string MediaGenericVideoPage::get_download_url(int max_height) {
        // TODO: Use max_height, if possible
        (void)max_height;
        return video_url;
    }

    std::string MediaGenericVideoPage::get_video_url(int max_height, bool &has_embedded_audio, std::string &ext) {
        // TODO: Use max_height, if possible
        (void)max_height;
        has_embedded_audio = true;
        ext = "m3u8";
        return video_url;
    }

    PluginResult MediaGenericVideoPage::load(std::string&, std::string&, std::vector<MediaChapter>&, std::string &err_msg) {
        video_url.clear();
        if(!search_page->video_custom_handler) {
            video_url = url;
            return PluginResult::OK;
        }

        std::vector<CommandArg> args = search_page->extra_commands;
        if(!url.empty())
            args.push_back({ "--header", "referer: " + url });

        std::string website_data;
        if(download_to_string(url, website_data, args, true, true, search_page->cloudflare_bypass) != DownloadResult::OK)
            return PluginResult::NET_ERR;

        video_url = search_page->video_custom_handler(website_data);
        if(video_url.empty()) {
            err_msg = "Failed to extract video url";
            return PluginResult::ERR;
        } else {
            return PluginResult::OK;
        }
    }
}