Capture YouTube titles from Discord's embed cache

CURRENT.md

@@ -40,6 +40,15 @@ mobile draft at 390px.
   - [x] C11 preview caching: Discord-hosted image/GIF previews (WebP 320px, hash-addressed, 5 MiB
     cap, one attempt) run after every walk in both modes. Live run `...-20260916090722`:
     `previewed: 2`; thumbs served over `/api/media/<scrape>/<item>/thumb` (200).
+  - [x] C58 Titles from Discord's embed cache. Gone YouTube videos have no oEmbed answer, but
+    Discord keeps the embed title (and, with the rebuilt core bundle, the author/channel). The
+    title lane now asks the source message's embed first and only then the provider oEmbed; the
+    core maps `authorName` from `embed.author.name`, extraction stores it as `title_author`, and
+    the viewer shows a prominent title/channel heading plus "Video unavailable (removed or
+    private)" when a YouTube row has neither. The 783 title errors were reset and the lane
+    backfilled them: the reported example (message 703788394337009754) now reads "Gun Test: Firing
+    Empty Shell's Handheld XM556 Microgun". The channel name populates after the Vesktop client
+    restarts on the new bundle (the client is externally started).
   - [x] C57 Page mode. The media grid has an Infinite/Page flippy toggle in the bar. Page mode
     renders one `pageSize` slice via `offset = (page-1)*pageSize`, shows Prev / editable page
     number / total / Next, persists the page in localStorage, and scrolls to the next page one at

media-scrape/src/engine.ts

@@ -100,6 +100,14 @@ export interface ScrapeEngineOptions {
     ) => Promise<string | null>;
     /** Delay before the single suspect retry. Default 50 s. */
     suspectRetryMs?: number;
+    /**
+     * Title and author from the source message's Discord embed. Preferred over the provider
+     * oEmbed: Discord keeps the embed data even after a video is deleted.
+     */
+    embedMeta?: (
+        scrapeId: string,
+        media: { id: number; url: string; playUrl: string | null }
+    ) => Promise<{ title: string; author: string | null } | null>;
     /** Read dimensions and duration from a downloaded file. Defaults to ffprobe. */
     probe?: (
         path: string
@@ -605,7 +613,15 @@ export class ScrapeEngine {
                     break;
                 }
 
-                const resolved = await resolveEmbedTitle(row.url, options);
+                // Discord's cached embed data survives a deleted video; ask it first.
+                const fromDiscord = await this.options.embedMeta?.(meta.id, {
+                    id: row.id,
+                    url: row.url,
+                    playUrl: row.play_url
+                });
+                const resolved =
+                    fromDiscord ??
+                    (await resolveEmbedTitle(row.url, options));
                 if (resolved) {
                     store.markTitle(row.id, resolved.title, resolved.author);
                     result.titled++;

media-scrape/src/extract.ts

@@ -34,6 +34,8 @@ export interface ExtractedMedia {
     previewUrl: string | null;
     /** Embed title, when Discord provides one (YouTube and friends). Null elsewhere. */
     title: string | null;
+    /** Embed author (the YouTube channel Discord shows). Null elsewhere. */
+    titleAuthor: string | null;
 }
 
 /**
@@ -95,7 +97,8 @@ function extractEmbed(embed: CoreEmbed): ExtractedMedia | null {
         width: embed.width ?? null,
         height: embed.height ?? null,
         durationSecs: null,
-        title: embed.title ?? null
+        title: embed.title ?? null,
+        titleAuthor: embed.authorName ?? null
     };
 
     if (embed.videoUrl) {
@@ -150,7 +153,8 @@ function extractSticker(sticker: CoreSticker): ExtractedMedia | null {
         durationSecs: null,
         downloaded: false,
         previewUrl: null,
-        title: null
+        title: null,
+        titleAuthor: null
     };
 }
 
@@ -178,7 +182,8 @@ function extractFrom(message: MediaMessage): ExtractedMedia[] {
                     : null,
             downloaded: false,
             previewUrl: null,
-            title: null
+            title: null,
+            titleAuthor: null
         });
     }
 

media-scrape/src/store.ts

@@ -404,7 +404,7 @@ export class ScrapeStore {
             "INSERT OR REPLACE INTO messages (id, author_id, author_name, timestamp, content) VALUES (?, ?, ?, ?, ?)"
         );
         const insertMedia = this.db.prepare(
-            "INSERT OR IGNORE INTO media (kind, source, author_id, author_name, posted_at, url, play_url, provider, content_type, size, width, height, aspect_ratio, duration_secs, downloaded, preview_url, title, title_key, created_at) VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)"
+            "INSERT OR IGNORE INTO media (kind, source, author_id, author_name, posted_at, url, play_url, provider, content_type, size, width, height, aspect_ratio, duration_secs, downloaded, preview_url, title, title_key, title_author, created_at) VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)"
         );
         const findMedia = this.db.prepare("SELECT id FROM media WHERE url = ?");
         const linkSource = this.db.prepare(
@@ -443,6 +443,7 @@ export class ScrapeStore {
                         item.previewUrl,
                         item.title,
                         item.title ? normalizeTitleKey(item.title) : null,
+                        item.titleAuthor,
                         createdAt
                     );
                     if (result.changes > 0) added[item.kind]++;

media-scrape/src/types.ts

@@ -35,6 +35,8 @@ export interface CoreEmbed {
     type: string | null;
     url: string | null;
     title: string | null;
+    /** Discord's embed author line (the YouTube channel name). */
+    authorName?: string | null;
     provider: string | null;
     videoUrl: string | null;
     imageUrl: string | null;

poc/public/media.css

@@ -885,3 +885,9 @@
     font: 400 11px/1.5 ui-monospace, Menlo, monospace;
     color: var(--color-neutral-500);
 }
+
+.vk-media-viewer-title {
+    font: 600 13px/1.5 ui-monospace, Menlo, monospace;
+    color: var(--color-text);
+    margin-bottom: 6px;
+}

poc/public/media.js

@@ -2669,6 +2669,18 @@
         }
 
         meta.replaceChildren();
+
+        // The clip title matters most for finding footage; show it big when we have one.
+        const heading =
+            item.title || (item.titleError && (item.provider ?? "").toLowerCase().includes("youtube"))
+                ? item.title
+                    ? `${item.title}${item.titleAuthor ? ` — ${item.titleAuthor}` : ""}`
+                    : "Video unavailable (removed or private)"
+                : "";
+        if (heading) {
+            meta.append(el("div", "vk-media-viewer-title", heading));
+        }
+
         const values = metaValues(item);
         for (const [key, label] of META_FIELDS) {
             if (!values[key]) continue;

poc/src/media.ts

@@ -136,6 +136,8 @@ export interface MediaItemResult {
     remotePreviewUrl: string | null;
     /** Why no preview exists, when a fetch failed (for example "http 404"). */
     previewError: string | null;
+    /** Why no title was resolved (provider oEmbed and Discord's embed both failed). */
+    titleError: string | null;
     /** Local animated GIF for hover playback, when the original was cached. */
     animUrl: string | null;
     localUrl: string | null;
@@ -217,6 +219,7 @@ export class MediaService {
             maxRetries: tunables.scrape.maxRetries,
             previewMaxAttempts: tunables.backfill.maxAttempts,
             suspectRetryMs: 50_000,
+            embedMeta: async (scrapeId, media) => this.embedMetaFor(scrapeId, media),
             refreshUrl: async (scrapeId, media) => {
                 const meta = this.engine.get(scrapeId);
                 if (!meta) return null;
@@ -683,6 +686,67 @@ export class MediaService {
         this.engine.stop(id);
     }
 
+    /**
+     * Title and channel from the source message's Discord embed. Discord keeps the embed data
+     * after a video is deleted, so this is the only title source for gone YouTube videos.
+     */
+    private async embedMetaFor(
+        scrapeId: string,
+        media: { id: number; url: string; playUrl: string | null }
+    ): Promise<{ title: string; author: string | null } | null> {
+        const meta = this.engine.get(scrapeId);
+        if (!meta) return null;
+
+        try {
+            const store = new ScrapeStore(scrapeDir(this.rootDir, scrapeId));
+            let messageId: string | null = null;
+            try {
+                messageId = store.sourceMessageId(media.id);
+            } finally {
+                store.close();
+            }
+            if (!messageId) return null;
+
+            const message = await this.fetchSourceMessage(meta, messageId);
+            if (!message) return null;
+
+            const path = (url: string) => url.split("?")[0];
+            const embeds: any[] = message.embeds ?? [];
+            const embed = embeds.find(
+                candidate =>
+                    (candidate?.url && path(candidate.url) === path(media.url)) ||
+                    (media.playUrl && candidate?.videoUrl && path(candidate.videoUrl) === path(media.playUrl)) ||
+                    (candidate?.video?.url && media.playUrl && path(candidate.video.url) === path(media.playUrl))
+            );
+
+            const title = embed?.title ?? null;
+            if (!title) return null;
+            const author = embed?.authorName ?? embed?.author?.name ?? null;
+            return { title, author };
+        } catch (error) {
+            this.deps.log?.(`embed meta failed for ${scrapeId}:${media.id}: ${String(error)}`);
+            return null;
+        }
+    }
+
+    /** The source message through slopbot REST (guild scrapes) or the puppet bridge. */
+    private async fetchSourceMessage(meta: ScrapeMeta, messageId: string): Promise<any | null> {
+        if (this.bot && meta.guildId && meta.backend === "slopbot") {
+            const channel = await this.bot.channels.fetch(meta.channelId);
+            if (channel?.isTextBased()) return await channel.messages.fetch(messageId);
+            return null;
+        }
+
+        const core = this.deps.getCore();
+        if (!core || !this.deps.isCoreConnected()) return null;
+
+        const result = (await core.fetchMessages(meta.channelId, 1, {
+            before: (BigInt(messageId) + 1n).toString()
+        })) as { messages?: Array<Record<string, any>> };
+
+        return result?.messages?.find(candidate => candidate?.id === messageId) ?? result?.messages?.[0] ?? null;
+    }
+
     /** Continue a paused scrape from its cursor with the same backend family. */
     async resumeScrape(scrapeId: string): Promise<ScrapeMeta | null> {
         const meta = this.engine.get(scrapeId);
@@ -1224,6 +1288,7 @@ export class MediaService {
                 playUrl: row.play_url,
                 remotePreviewUrl: row.preview_url,
                 previewError: row.preview_error,
+                titleError: row.title_error,
                 animUrl: row.anim_hash ? `/api/media/${meta.id}/${row.id}/anim` : null,
                 localUrl:
                     row.file_path && !meta.downloadDir

src/renderer/puppet/actions.ts

@@ -229,6 +229,7 @@ function mapEmbeds(message: any) {
             type: embed?.type ?? null,
             url: embed?.url ?? null,
             title: embed?.title ?? null,
+            authorName: embed?.author?.name ?? null,
             provider: embed?.provider?.name ?? null,
             videoUrl: embedUrl(embed?.video),
             imageUrl: embedUrl(embed?.image),