Youtube: fallback to yt-dlp, add support for sponsorblock (as chapters)

4 files changed, 163 insertions, 44 deletions

diff --git a/src/Config.cpp b/src/Config.cpp
index 191beb4..cc1c476 100644
--- a/src/Config.cpp
+++ b/src/Config.cpp
@@ -280,6 +280,12 @@ namespace QuickMedia {
         if(youtube_json.isObject()) {
             get_json_value(youtube_json, "load_progress", config->youtube.load_progress);
             get_json_value(youtube_json, "invidious_instance", config->youtube.invidious_instance);
+
+            const Json::Value &sponsorblock_json = youtube_json["sponsorblock"];
+            if(sponsorblock_json.isObject()) {
+                get_json_value(sponsorblock_json, "enable", config->youtube.sponsorblock.enable);
+                get_json_value(sponsorblock_json, "min_votes", config->youtube.sponsorblock.min_votes);
+            }
         }
 
         bool has_known_matrix_homeservers_config = false;
diff --git a/src/QuickMedia.cpp b/src/QuickMedia.cpp
index b47ab13..c57540c 100644
--- a/src/QuickMedia.cpp
+++ b/src/QuickMedia.cpp
@@ -1089,18 +1089,58 @@ namespace QuickMedia {
             .related_media_thumbnail_handler({{"//div[data-role='video-relations']//img", "src", "xhcdn"}});
     }
 
-    void Program::check_youtube_dl_installed(const std::string &plugin_name) {
-        if(yt_dl_name)
-            return;
+    bool Program::youtube_dl_extract_url(const std::string &url, std::string &video_url, std::string &audio_url) {
+        const char *youtube_dl_name = get_youtube_dl_program_name();
+        if(!youtube_dl_name)
+            return false;
+
+        std::string ytdl_format;
+        if(no_video)
+            ytdl_format = "(bestaudio/best)";
+        else
+            ytdl_format = "(bestvideo[vcodec!*=av01][height<=?" + std::to_string(video_max_height) + "]+bestaudio/best)";
+
+        ytdl_format += "[protocol^=http]/" + ytdl_format + "[protocol^=m3u8]";
+
+        std::string result;
+        const char *args[] = { youtube_dl_name, "--no-warnings", "-f", ytdl_format.c_str(), "-g", "--", url.c_str(), nullptr };
+        if(exec_program(args, accumulate_string, &result) != 0)
+            return false;
+
+        string_split(result, '\n', [&](const char *str, size_t size) {
+            if(video_url.empty())
+                video_url.assign(str, size);
+            else if(audio_url.empty())
+                audio_url.assign(str, size);
+            return true;
+        }, false);
+
+        return true;
+    }
+
+    const char* Program::get_youtube_dl_program_name() {
+        if(yt_dl_name_checked)
+            return yt_dl_name;
 
         if(is_program_executable_by_name("yt-dlp")) {
             yt_dl_name = "yt-dlp";
         } else if(is_program_executable_by_name("youtube-dl")) {
             yt_dl_name = "youtube-dl";
         } else {
-            show_notification("QuickMedia", "yt-dlp or youtube-dl needs to be installed to play " + plugin_name + " videos", Urgency::CRITICAL);
-            exit(10);
+            yt_dl_name = nullptr;
         }
+
+        yt_dl_name_checked = true;
+        return yt_dl_name;
+    }
+
+    void Program::check_youtube_dl_installed(const std::string &plugin_name) {
+        get_youtube_dl_program_name();
+        if(yt_dl_name)
+            return;
+
+        show_notification("QuickMedia", "yt-dlp or youtube-dl needs to be installed to play " + plugin_name + " videos", Urgency::CRITICAL);
+        exit(10);
     }
 
     void Program::load_plugin_by_name(std::vector<Tab> &tabs, int &start_tab_index, FileManagerMimeType fm_mime_type, FileSelectionHandler file_selection_handler, std::string instance) {
diff --git a/src/VideoPlayer.cpp b/src/VideoPlayer.cpp
index 4642855..5369a07 100644
--- a/src/VideoPlayer.cpp
+++ b/src/VideoPlayer.cpp
@@ -47,7 +47,11 @@ namespace QuickMedia {
     static const int MAX_RETRIES_CONNECT = 1000;
     static const double READ_TIMEOUT_SEC = 3.0;
 
-    static std::string media_chapters_to_ffmetadata_chapters(const std::vector<MediaChapter> &chapters) {
+    static std::string media_chapters_to_ffmetadata_chapters(std::vector<MediaChapter> chapters) {
+        std::sort(chapters.begin(), chapters.end(), [](const MediaChapter &a, const MediaChapter &b) {
+            return a.start_seconds < b.start_seconds;
+        });
+
         std::string result = ";FFMETADATA1\n\n";
         for(size_t i = 0; i < chapters.size(); ++i) {
             const MediaChapter &chapter = chapters[i];
diff --git a/src/plugins/Youtube.cpp b/src/plugins/Youtube.cpp
index fef5fce..482843e 100644
--- a/src/plugins/Youtube.cpp
+++ b/src/plugins/Youtube.cpp
@@ -2857,6 +2857,12 @@ namespace QuickMedia {
     }
 
     std::string YoutubeVideoPage::get_video_url(int max_height, bool &has_embedded_audio, std::string &ext) {
+        if(use_youtube_dl_fallback) {
+            has_embedded_audio = youtube_dl_audio_fallback_url.empty();
+            ext = ".webm"; // TODO: Guess for now
+            return youtube_dl_video_fallback_url;
+        }
+
         if(!livestream_url.empty() && video_formats.empty() && audio_formats.empty()) {
             has_embedded_audio = true;
             return livestream_url;
@@ -2892,6 +2898,11 @@ namespace QuickMedia {
     }
 
     std::string YoutubeVideoPage::get_audio_url(std::string &ext) {
+        if(use_youtube_dl_fallback) {
+            ext = ".opus"; // TODO: Guess for now
+            return youtube_dl_audio_fallback_url;
+        }
+
         if(audio_formats.empty())
             return "";
 
@@ -3054,6 +3065,56 @@ namespace QuickMedia {
         video_details.duration = 0.0;
     }
 
+    static void sponsorblock_add_chapters(Page *page, const std::string &url, int min_votes, std::vector<MediaChapter> &chapters) {
+        std::string video_id;
+        if(!youtube_url_extract_id(url, video_id)) {
+            fprintf(stderr, "Failed to extract youtube id from %s\n", url.c_str());
+            return;
+        }
+
+        const std::string sponsorblock_url = "https://sponsor.ajay.app/api/skipSegments?videoID=" + video_id;
+        Json::Value json_root;
+        if(page->download_json(json_root, sponsorblock_url, {}) != DownloadResult::OK)
+            return;
+
+        if(!json_root.isArray())
+            return;
+
+        for(const Json::Value &item_json : json_root) {
+            if(!item_json.isObject())
+                continue;
+
+            const Json::Value &category_json = item_json["category"];
+            const Json::Value &action_type_json = item_json["actionType"];
+            const Json::Value &segment_json = item_json["segment"];
+            const Json::Value &votes_json = item_json["votes"];
+            if(!category_json.isString() || !action_type_json.isString() || !segment_json.isArray() || !votes_json.isInt())
+                continue;
+
+            if(strcmp(category_json.asCString(), "sponsor") != 0 || strcmp(action_type_json.asCString(), "skip") != 0)
+                continue;
+
+            if(segment_json.size() != 2)
+                continue;
+
+            if(!segment_json[0].isDouble() || !segment_json[1].isDouble())
+                continue;
+
+            if(votes_json.asInt() < min_votes)
+                continue;
+
+            MediaChapter ad_start;
+            ad_start.start_seconds = segment_json[0].asDouble();
+            ad_start.title = "Ad start";
+            chapters.push_back(std::move(ad_start));
+
+            MediaChapter ad_end;
+            ad_end.start_seconds = segment_json[1].asDouble();
+            ad_end.title = "Ad end";
+            chapters.push_back(std::move(ad_end));
+        }
+    }
+
     PluginResult YoutubeVideoPage::parse_video_response(const Json::Value &json_root, std::string &title, std::string &channel_url, std::vector<MediaChapter> &chapters, std::string &err_str) {
         livestream_url.clear();
         video_formats.clear();
@@ -3062,45 +3123,13 @@ namespace QuickMedia {
         title.clear();
         channel_url.clear();
         chapters.clear();
+        youtube_dl_video_fallback_url.clear();
+        youtube_dl_audio_fallback_url.clear();
         video_details_clear(video_details);
 
         if(!json_root.isObject())
             return PluginResult::ERR;
 
-        const Json::Value &playability_status_json = json_root["playabilityStatus"];
-        if(playability_status_json.isObject()) {
-            const Json::Value &status_json = playability_status_json["status"];
-            if(status_json.isString() && (strcmp(status_json.asCString(), "UNPLAYABLE") == 0 || strcmp(status_json.asCString(), "LOGIN_REQUIRED") == 0)) {
-                const Json::Value &reason_json = playability_status_json["reason"];
-                if(reason_json.isString())
-                    err_str = reason_json.asString();
-                fprintf(stderr, "Unable to play video, status: %s, reason: %s\n", status_json.asCString(), reason_json.isString() ? reason_json.asCString() : "Unknown");
-                return PluginResult::ERR;
-            }
-        }
-
-        const Json::Value *streaming_data_json = &json_root["streamingData"];
-        if(!streaming_data_json->isObject())
-            return PluginResult::ERR;
-
-        // TODO: Verify if this always works (what about copyrighted live streams?), also what about choosing video quality for live stream? Maybe use mpv --hls-bitrate option?
-        const Json::Value &hls_manifest_url_json = (*streaming_data_json)["hlsManifestUrl"];
-        if(hls_manifest_url_json.isString())
-            livestream_url = hls_manifest_url_json.asString();
-        
-        /*
-        const Json::Value &dash_manifest_url_json = (*streaming_data_json)["dashManifestUrl"];
-        if(livestream_url.empty() && dash_manifest_url_json.isString()) {
-            // TODO: mpv cant properly play dash videos. Video goes back and replays.
-            // So for now return here (get_video_info only hash dash stream and no hls stream) which will fallback to the player youtube endpoint which has hls stream.
-            return PluginResult::ERR;
-        }
-        */
-
-        parse_formats(*streaming_data_json);
-        if(video_formats.empty() && audio_formats.empty() && livestream_url.empty())
-            return PluginResult::ERR;
-
         const Json::Value &video_details_json = json_root["videoDetails"];
         if(video_details_json.isObject()) {
             const Json::Value &channel_id_json = video_details_json["channelId"];
@@ -3166,6 +3195,48 @@ namespace QuickMedia {
             }
         }
 
+        const Json::Value &playability_status_json = json_root["playabilityStatus"];
+        if(playability_status_json.isObject()) {
+            const Json::Value &status_json = playability_status_json["status"];
+            if(status_json.isString() && (strcmp(status_json.asCString(), "UNPLAYABLE") == 0 || strcmp(status_json.asCString(), "LOGIN_REQUIRED") == 0)) {
+                fprintf(stderr, "Failed to load youtube video, trying with yt-dlp instead\n");
+                if(program->youtube_dl_extract_url(url, youtube_dl_video_fallback_url, youtube_dl_audio_fallback_url)) {
+                    if(get_config().youtube.sponsorblock.enable)
+                        sponsorblock_add_chapters(this, url, get_config().youtube.sponsorblock.min_votes, chapters);
+                    use_youtube_dl_fallback = true;
+                    return PluginResult::OK;
+                } else {
+                    const Json::Value &reason_json = playability_status_json["reason"];
+                    if(reason_json.isString())
+                        err_str = reason_json.asString();
+                    fprintf(stderr, "Unable to play video, status: %s, reason: %s\n", status_json.asCString(), reason_json.isString() ? reason_json.asCString() : "Unknown");
+                    return PluginResult::ERR;
+                }
+            }
+        }
+
+        const Json::Value *streaming_data_json = &json_root["streamingData"];
+        if(!streaming_data_json->isObject())
+            return PluginResult::ERR;
+
+        // TODO: Verify if this always works (what about copyrighted live streams?), also what about choosing video quality for live stream? Maybe use mpv --hls-bitrate option?
+        const Json::Value &hls_manifest_url_json = (*streaming_data_json)["hlsManifestUrl"];
+        if(hls_manifest_url_json.isString())
+            livestream_url = hls_manifest_url_json.asString();
+        
+        /*
+        const Json::Value &dash_manifest_url_json = (*streaming_data_json)["dashManifestUrl"];
+        if(livestream_url.empty() && dash_manifest_url_json.isString()) {
+            // TODO: mpv cant properly play dash videos. Video goes back and replays.
+            // So for now return here (get_video_info only hash dash stream and no hls stream) which will fallback to the player youtube endpoint which has hls stream.
+            return PluginResult::ERR;
+        }
+        */
+
+        parse_formats(*streaming_data_json);
+        if(video_formats.empty() && audio_formats.empty() && livestream_url.empty())
+            return PluginResult::ERR;
+
         std::sort(video_formats.begin(), video_formats.end(), [](const YoutubeVideoFormat &format1, const YoutubeVideoFormat &format2) {
             return format1.base.bitrate > format2.base.bitrate;
         });
@@ -3174,6 +3245,8 @@ namespace QuickMedia {
             return format1.base.bitrate > format2.base.bitrate;
         });
 
+        if(get_config().youtube.sponsorblock.enable)
+            sponsorblock_add_chapters(this, url, get_config().youtube.sponsorblock.min_votes, chapters);
         return PluginResult::OK;
     }
 
@@ -3184,10 +3257,6 @@ namespace QuickMedia {
             return PluginResult::ERR;
         }
 
-        // The first one works for copyrighted videos and regular videos but only if they can be embedded.
-        // The second one works for age restricted videos and regular videos but only if they can be embedded. It doesn't work for copyrighted videos.
-        // The third one works for all non-copyrighted, non-age restricted videos, embeddable or not.
-
         const int num_request_types = 1;
         std::string request_data[num_request_types] = {
 R"END(