From ae6fb457ca385540e0f9b1347ef9c3c84815b16d Mon Sep 17 00:00:00 2001 From: dec05eba Date: Tue, 3 Nov 2020 20:29:20 +0100 Subject: Youtube add channel page, fix search pagination (update to correct continuation token) --- src/plugins/Matrix.cpp | 8 +- src/plugins/Youtube.cpp | 425 ++++++++++++++++++++++++++++++++++++++++-------- 2 files changed, 361 insertions(+), 72 deletions(-) (limited to 'src') diff --git a/src/plugins/Matrix.cpp b/src/plugins/Matrix.cpp index 96ce789..198b2fd 100644 --- a/src/plugins/Matrix.cpp +++ b/src/plugins/Matrix.cpp @@ -818,6 +818,7 @@ namespace QuickMedia { return; assert(!this->delegate); + assert(!access_token.empty()); // Need to be logged in this->delegate = delegate; Path matrix_cache_dir = get_cache_dir().join("matrix"); @@ -846,7 +847,7 @@ namespace QuickMedia { } sync_is_cache = false; - // Filter with account data. TODO: Test if this is needed for encrypted chats + // Filter with account data // {"presence":{"limit":0,"types":[""]},"account_data":{"not_types":["im.vector.setting.breadcrumbs","m.push_rules","im.vector.setting.allowed_widgets","io.element.recent_emoji"]},"room":{"state":{"limit":1,"not_types":["m.room.related_groups","m.room.power_levels","m.room.join_rules","m.room.history_visibility"],"lazy_load_members":true},"timeline":{"limit":3,"lazy_load_members":true},"ephemeral":{"limit":0,"types":[""],"lazy_load_members":true},"account_data":{"limit":1,"types":["m.fully_read"],"lazy_load_members":true}}} // Filter without account data const char *filter = "{\"presence\":{\"limit\":0,\"types\":[\"\"]},\"account_data\":{\"limit\":0,\"types\":[\"\"]},\"room\":{\"state\":{\"not_types\":[\"m.room.related_groups\",\"m.room.power_levels\",\"m.room.join_rules\",\"m.room.history_visibility\"],\"lazy_load_members\":true},\"timeline\":{\"limit\":3,\"lazy_load_members\":true},\"ephemeral\":{\"limit\":0,\"types\":[\"\"],\"lazy_load_members\":true},\"account_data\":{\"limit\":1,\"types\":[\"m.fully_read\",\"m.tag\"],\"lazy_load_members\":true}}}"; @@ -861,11 +862,6 @@ namespace QuickMedia { PluginResult result; bool initial_sync = true; while(sync_running) { - std::vector additional_args = { - { "-H", "Authorization: Bearer " + access_token }, - { "-m", "35" } - }; - char url[1024]; if(next_batch.empty()) snprintf(url, sizeof(url), "%s/_matrix/client/r0/sync?filter=%s&timeout=0", homeserver.c_str(), filter_encoded.c_str()); diff --git a/src/plugins/Youtube.cpp b/src/plugins/Youtube.cpp index a157a8c..1ca25a3 100644 --- a/src/plugins/Youtube.cpp +++ b/src/plugins/Youtube.cpp @@ -1,80 +1,62 @@ #include "../../plugins/Youtube.hpp" #include "../../include/Storage.hpp" #include "../../include/NetUtils.hpp" +#include "../../include/Scale.hpp" #include -#include namespace QuickMedia { - static std::shared_ptr parse_common_video_item(const Json::Value &video_item_json, std::unordered_set &added_videos) { - const Json::Value &video_id_json = video_item_json["videoId"]; - if(!video_id_json.isString()) + // This is a common setup of text in the youtube json + static const char* yt_json_get_text(const Json::Value &json, const char *root_name) { + if(!json.isObject()) return nullptr; - std::string video_id_str = video_id_json.asString(); - if(added_videos.find(video_id_str) != added_videos.end()) + const Json::Value &text_json = json[root_name]; + if(!text_json.isObject()) return nullptr; - std::string thumbnail_url = "https://img.youtube.com/vi/" + video_id_str + "/hqdefault.jpg"; - - const char *date = nullptr; - const Json::Value &published_time_text_json = video_item_json["publishedTimeText"]; - if(published_time_text_json.isObject()) { - const Json::Value &text_json = published_time_text_json["simpleText"]; - if(text_json.isString()) - date = text_json.asCString(); - } - - const char *length = nullptr; - const Json::Value &length_text_json = video_item_json["lengthText"]; - if(length_text_json.isObject()) { - const Json::Value &text_json = length_text_json["simpleText"]; - if(text_json.isString()) - length = text_json.asCString(); - } - - const char *view_count_text = nullptr; - const Json::Value &view_count_text_json = video_item_json["viewCountText"]; - if(view_count_text_json.isObject()) { - const Json::Value &text_json = view_count_text_json["simpleText"]; - if(text_json.isString()) - view_count_text = text_json.asCString(); - } - - const char *owner_text = nullptr; - const Json::Value &owner_text_json = video_item_json["shortBylineText"]; - if(owner_text_json.isObject()) { - const Json::Value &runs_json = owner_text_json["runs"]; + const Json::Value &simple_text_json = text_json["simpleText"]; + if(simple_text_json.isString()) { + return simple_text_json.asCString(); + } else { + const Json::Value &runs_json = text_json["runs"]; if(runs_json.isArray() && !runs_json.empty()) { const Json::Value &first_runs_json = runs_json[0]; if(first_runs_json.isObject()) { const Json::Value &text_json = first_runs_json["text"]; if(text_json.isString()) - owner_text = text_json.asCString(); + return text_json.asCString(); } } } - const char *title = nullptr; - const Json::Value &title_json = video_item_json["title"]; - if(title_json.isObject()) { - const Json::Value &simple_text_json = title_json["simpleText"]; - if(simple_text_json.isString()) { - title = simple_text_json.asCString(); - } else { - const Json::Value &runs_json = title_json["runs"]; - if(runs_json.isArray() && !runs_json.empty()) { - const Json::Value &first_runs_json = runs_json[0]; - if(first_runs_json.isObject()) { - const Json::Value &text_json = first_runs_json["text"]; - if(text_json.isString()) - title = text_json.asCString(); - } - } - } - } + return nullptr; + } + + static std::shared_ptr parse_common_video_item(const Json::Value &video_item_json, std::unordered_set &added_videos) { + const Json::Value &video_id_json = video_item_json["videoId"]; + if(!video_id_json.isString()) + return nullptr; + + std::string video_id_str = video_id_json.asString(); + if(added_videos.find(video_id_str) != added_videos.end()) + return nullptr; + const char *title = yt_json_get_text(video_item_json, "title"); if(!title) return nullptr; + + const char *date = yt_json_get_text(video_item_json, "publishedTimeText"); + const char *view_count_text = yt_json_get_text(video_item_json, "viewCountText"); + const char *owner_text = yt_json_get_text(video_item_json, "shortBylineText"); + const char *length = yt_json_get_text(video_item_json, "lengthText"); + if(!length) { + const Json::Value &thumbnail_overlays_json = video_item_json["thumbnailOverlays"]; + if(thumbnail_overlays_json.isArray() && !thumbnail_overlays_json.empty()) { + const Json::Value &thumbnail_overlay_json = thumbnail_overlays_json[0]; + if(thumbnail_overlay_json.isObject()) + length = yt_json_get_text(thumbnail_overlay_json["thumbnailOverlayTimeStatusRenderer"], "text"); + } + } auto body_item = BodyItem::create(title); std::string desc; @@ -97,7 +79,7 @@ namespace QuickMedia { } body_item->set_description(std::move(desc)); body_item->url = "https://www.youtube.com/watch?v=" + video_id_str; - body_item->thumbnail_url = std::move(thumbnail_url); + body_item->thumbnail_url = "https://img.youtube.com/vi/" + video_id_str + "/hqdefault.jpg"; body_item->thumbnail_size = sf::Vector2i(175, 131); added_videos.insert(video_id_str); return body_item; @@ -114,6 +96,96 @@ namespace QuickMedia { return parse_common_video_item(video_renderer_json, added_videos); } + struct Thumbnail { + const char *url; + int width; + int height; + }; + + static std::optional yt_json_get_largest_thumbnail(const Json::Value &thumbnail_json) { + if(!thumbnail_json.isObject()) + return std::nullopt; + + const Json::Value &thumbnails_json = thumbnail_json["thumbnails"]; + if(!thumbnails_json.isArray()) + return std::nullopt; + + std::vector thumbnails; + for(const Json::Value &thumbnail_data_json : thumbnails_json) { + if(!thumbnail_data_json.isObject()) + continue; + + const Json::Value &url_json = thumbnail_data_json["url"]; + if(!url_json.isString()) + continue; + + const Json::Value &width_json = thumbnail_data_json["width"]; + if(!width_json.isInt()) + continue; + + const Json::Value &height_json = thumbnail_data_json["height"]; + if(!height_json.isInt()) + continue; + + thumbnails.push_back({ url_json.asCString(), width_json.asInt(), height_json.asInt() }); + } + + return *std::max_element(thumbnails.begin(), thumbnails.end(), [](const Thumbnail &thumbnail1, const Thumbnail &thumbnail2) { + int size1 = thumbnail1.width * thumbnail1.height; + int size2 = thumbnail2.width * thumbnail2.height; + return size1 < size2; + }); + } + + static std::shared_ptr parse_channel_renderer(const Json::Value &channel_renderer_json) { + if(!channel_renderer_json.isObject()) + return nullptr; + + const Json::Value &channel_id_json = channel_renderer_json["channelId"]; + if(!channel_id_json.isString()) + return nullptr; + + const char *title = yt_json_get_text(channel_renderer_json, "title"); + if(!title) + return nullptr; + + const char *description = yt_json_get_text(channel_renderer_json, "descriptionSnippet"); + const char *video_count = yt_json_get_text(channel_renderer_json, "videoCountText"); + const char *subscribers = yt_json_get_text(channel_renderer_json, "subscriberCountText"); + + const Json::Value &thumbnail_json = channel_renderer_json["thumbnail"]; + std::optional thumbnail = yt_json_get_largest_thumbnail(thumbnail_json); + + auto body_item = BodyItem::create(title); + std::string desc; + if(subscribers) + desc += subscribers; + if(video_count) { + if(!desc.empty()) + desc += " • "; + desc += video_count; + if(strcmp(video_count, "1") == 0) + desc += " video"; + else + desc += " videos"; + } + if(description) { + if(!desc.empty()) + desc += '\n'; + desc += description; + } + body_item->set_description(std::move(desc)); + body_item->url = "https://www.youtube.com/channel/" + channel_id_json.asString(); + if(thumbnail) { + body_item->thumbnail_url = std::string("https:") + thumbnail->url; + body_item->thumbnail_mask_type = ThumbnailMaskType::CIRCLE; + body_item->thumbnail_size.x = thumbnail->width; + body_item->thumbnail_size.y = thumbnail->height; + body_item->thumbnail_size = clamp_to_size(body_item->thumbnail_size, sf::Vector2i(136, 136)); + } + return body_item; + } + // Returns empty string if continuation token can't be found static std::string item_section_renderer_get_continuation_token(const Json::Value &item_section_renderer_json) { const Json::Value &continuation_item_renderer_json = item_section_renderer_json["continuationItemRenderer"]; @@ -135,6 +207,27 @@ namespace QuickMedia { return token_json.asString(); } + static std::string grid_renderer_get_continuation_token(const Json::Value &grid_renderer_json) { + const Json::Value &continuations_json = grid_renderer_json["continuations"]; + if(!continuations_json.isArray()) + return ""; + + for(const Json::Value &continuation_json : continuations_json) { + if(!continuation_json.isObject()) + continue; + + const Json::Value &next_continuation_data_json = continuation_json["nextContinuationData"]; + if(!next_continuation_data_json.isObject()) + continue; + + const Json::Value &continuation_item_json = next_continuation_data_json["continuation"]; + if(continuation_item_json.isString()) + return continuation_item_json.asString(); + } + + return ""; + } + static void parse_item_section_renderer(const Json::Value &item_section_renderer_json, std::unordered_set &added_videos, BodyItems &result_items) { const Json::Value &item_contents_json = item_section_renderer_json["contents"]; if(!item_contents_json.isArray()) @@ -168,6 +261,10 @@ namespace QuickMedia { if(body_item) result_items.push_back(std::move(body_item)); } + } else if(key.isString() && strcmp(key.asCString(), "channelRenderer") == 0) { + std::shared_ptr body_item = parse_channel_renderer(*it); + if(body_item) + result_items.push_back(std::move(body_item)); } else { std::shared_ptr body_item = parse_content_video_renderer(content_item_json, added_videos); if(body_item) @@ -193,9 +290,106 @@ namespace QuickMedia { return parse_common_video_item(compact_video_renderer_json, added_videos); } + static BodyItems parse_channel_videos(const Json::Value &json_root, std::string &continuation_token, std::unordered_set &added_videos) { + BodyItems body_items; + if(!json_root.isArray()) + return body_items; + + std::string new_continuation_token; + for(const Json::Value &json_item : json_root) { + if(!json_item.isObject()) + continue; + + const Json::Value &response_json = json_item["response"]; + if(!response_json.isObject()) + continue; + + const Json::Value &contents_json = response_json["contents"]; + if(!contents_json.isObject()) + continue; + + const Json::Value &tcbrr_json = contents_json["twoColumnBrowseResultsRenderer"]; + if(!tcbrr_json.isObject()) + continue; + + const Json::Value &tabs_json = tcbrr_json["tabs"]; + if(!tabs_json.isArray()) + continue; + + for(const Json::Value &tab_json : tabs_json) { + if(!tab_json.isObject()) + continue; + + const Json::Value &tab_renderer_json = tab_json["tabRenderer"]; + if(!tab_renderer_json.isObject()) + continue; + + const Json::Value &content_json = tab_renderer_json["content"]; + if(!content_json.isObject()) + continue; + + const Json::Value §ion_list_renderer = content_json["sectionListRenderer"]; + if(!section_list_renderer.isObject()) + continue; + + const Json::Value &contents2_json = section_list_renderer["contents"]; + if(!contents2_json.isArray()) + continue; + + for(const Json::Value &content_item_json : contents2_json) { + if(!content_item_json.isObject()) + continue; + + const Json::Value &item_section_renderer_json = content_item_json["itemSectionRenderer"]; + if(!item_section_renderer_json.isObject()) + continue; + + const Json::Value &item_contents_json = item_section_renderer_json["contents"]; + if(!item_contents_json.isArray()) + continue; + + for(const Json::Value &content_json : item_contents_json) { + if(!content_json.isObject()) + continue; + + const Json::Value &grid_renderer_json = content_json["gridRenderer"]; + if(!grid_renderer_json.isObject()) + continue; + + if(new_continuation_token.empty()) + new_continuation_token = grid_renderer_get_continuation_token(grid_renderer_json); + + const Json::Value &items_json = grid_renderer_json["items"]; + if(!items_json.isArray()) + continue; + + for(const Json::Value &item_json : items_json) { + if(!item_json.isObject()) + continue; + + const Json::Value &grid_video_renderer = item_json["gridVideoRenderer"]; + if(!grid_video_renderer.isObject()) + continue; + + auto body_item = parse_common_video_item(grid_video_renderer, added_videos); + if(body_item) + body_items.push_back(std::move(body_item)); + } + } + } + } + } + + if(!new_continuation_token.empty()) + continuation_token = std::move(new_continuation_token); + + return body_items; + } + SearchResult YoutubeSearchPage::search(const std::string &str, BodyItems &result_items) { continuation_token.clear(); current_page = 0; + added_videos.clear(); search_url = "https://youtube.com/results?search_query="; search_url += url_param_encode(str); @@ -217,8 +411,6 @@ namespace QuickMedia { if(!json_root.isArray()) return SearchResult::ERR; - std::unordered_set added_videos; /* The input contains duplicates, filter them out! */ - for(const Json::Value &json_item : json_root) { if(!json_item.isObject()) continue; @@ -265,8 +457,7 @@ namespace QuickMedia { return SearchResult::OK; } - PluginResult YoutubeSearchPage::get_page(const std::string &str, int page, BodyItems &result_items) { - (void)str; + PluginResult YoutubeSearchPage::get_page(const std::string&, int page, BodyItems &result_items) { while(current_page < page) { PluginResult plugin_result = search_get_continuation(search_url, continuation_token, result_items); if(plugin_result != PluginResult::OK) return plugin_result; @@ -276,9 +467,32 @@ namespace QuickMedia { } PluginResult YoutubeSearchPage::submit(const std::string &title, const std::string &url, std::vector &result_tabs) { - (void)title; - (void)url; - result_tabs.push_back(Tab{nullptr, std::make_unique(program), nullptr}); + if(strncmp(url.c_str(), "https://www.youtube.com/channel/", 32) == 0) { + std::vector additional_args = { + { "-H", "x-spf-referer: " + url }, + { "-H", "x-youtube-client-name: 1" }, + { "-H", "x-spf-previous: " + url }, + { "-H", "x-youtube-client-version: 2.20200626.03.00" }, + { "-H", "referer: " + url } + }; + + //std::vector cookies = get_cookies(); + //additional_args.insert(additional_args.end(), cookies.begin(), cookies.end()); + + Json::Value json_root; + DownloadResult result = download_json(json_root, url + "/videos?pbj=1", std::move(additional_args), true); + if(result != DownloadResult::OK) return download_result_to_plugin_result(result); + + auto channel_body = create_body(); + std::unordered_set added_videos; + std::string continuation_token; + channel_body->items = parse_channel_videos(json_root, continuation_token, added_videos); + auto channel_page = std::make_unique(program, url, std::move(continuation_token), title); + channel_page->added_videos = std::move(added_videos); + result_tabs.push_back(Tab{std::move(channel_body), std::move(channel_page), create_search_bar("Search...", SEARCH_DELAY_FILTER)}); + } else { + result_tabs.push_back(Tab{nullptr, std::make_unique(program), nullptr}); + } return PluginResult::OK; } @@ -303,9 +517,7 @@ namespace QuickMedia { if(!json_root.isArray()) return PluginResult::ERR; - std::unordered_set added_videos; std::string new_continuation_token; - for(const Json::Value &json_item : json_root) { if(!json_item.isObject()) continue; @@ -348,12 +560,93 @@ namespace QuickMedia { } } - if(new_continuation_token.empty()) + if(!new_continuation_token.empty()) + continuation_token = std::move(new_continuation_token); + + return PluginResult::OK; + } + + PluginResult YoutubeChannelPage::get_page(const std::string&, int page, BodyItems &result_items) { + while(current_page < page) { + PluginResult plugin_result = search_get_continuation(url, continuation_token, result_items); + if(plugin_result != PluginResult::OK) return plugin_result; + ++current_page; + } + return PluginResult::OK; + } + + PluginResult YoutubeChannelPage::search_get_continuation(const std::string &url, const std::string ¤t_continuation_token, BodyItems &result_items) { + std::string next_url = "https://www.youtube.com/browse_ajax?ctoken=" + current_continuation_token + "&continuation=" + current_continuation_token; + + std::vector additional_args = { + { "-H", "x-spf-referer: " + url }, + { "-H", "x-youtube-client-name: 1" }, + { "-H", "x-spf-previous: " + url }, + { "-H", "x-youtube-client-version: 2.20200626.03.00" }, + { "-H", "referer: " + url } + }; + + //std::vector cookies = get_cookies(); + //additional_args.insert(additional_args.end(), cookies.begin(), cookies.end()); + + Json::Value json_root; + DownloadResult result = download_json(json_root, next_url, std::move(additional_args), true); + if(result != DownloadResult::OK) return download_result_to_plugin_result(result); + + if(!json_root.isArray()) + return PluginResult::ERR; + + std::string new_continuation_token; + for(const Json::Value &json_item : json_root) { + if(!json_item.isObject()) + continue; + + const Json::Value &response_json = json_item["response"]; + if(!response_json.isObject()) + continue; + + const Json::Value &continuation_contents_json = response_json["continuationContents"]; + if(!continuation_contents_json.isObject()) + continue; + + const Json::Value &grid_continuation_json = continuation_contents_json["gridContinuation"]; + if(!grid_continuation_json.isObject()) + continue; + + if(new_continuation_token.empty()) { + // grid_continuation_json is compatible with grid_renderer + new_continuation_token = grid_renderer_get_continuation_token(grid_continuation_json); + } + + const Json::Value &items_json = grid_continuation_json["items"]; + if(!items_json.isArray()) + continue; + + for(const Json::Value &item_json : items_json) { + if(!item_json.isObject()) + continue; + + const Json::Value &grid_video_renderer = item_json["gridVideoRenderer"]; + if(!grid_video_renderer.isObject()) + continue; + + auto body_item = parse_common_video_item(grid_video_renderer, added_videos); + if(body_item) + result_items.push_back(std::move(body_item)); + } + } + + if(!new_continuation_token.empty()) continuation_token = std::move(new_continuation_token); return PluginResult::OK; } + PluginResult YoutubeChannelPage::submit(const std::string&, const std::string&, std::vector &result_tabs) { + result_tabs.push_back(Tab{nullptr, std::make_unique(program), nullptr}); + return PluginResult::OK; + } + // TODO: Make this faster by using string search instead of parsing html. // TODO: If the result is a play BodyItems YoutubeVideoPage::get_related_media(const std::string &url) { -- cgit v1.2.3