Update Mastodon.ts

Added video support.

src/engines/Mastodon.ts

@@ -7,26 +7,34 @@ import {
 } from "../ScrapeEngine.js";
 
 interface Attachment {
-  url: string;          // original URL, ?sensitive=true stripped
+  url: string;          // original media URL, ?sensitive=true stripped
-  previewUrl?: string;  // local cached preview (for video posters)
+  previewUrl?: string;  // poster / cached preview (for video posters)
   contentType: ContentType;
 }
 
 /**
  * Handles Mastodon-compatible instances, including forks such as baraag.net.
  *
- * On a Mastodon post page, media is rendered as:
+ * Mastodon renders media in three different shapes:
- *   <div class="media-gallery__item">
- *     <a class="media-gallery__item-thumbnail" href=".../original/....png">
- *       <img src=".../small/....png">
- *     </a>
- *   </div>
  *
- * The anchor href is the original (often federated) file. The <img> is a
+ *   1. Images:
- * scaled-down local cache. We read the anchor href, never the img src.
+ *        <a class="media-gallery__item-thumbnail" href=".../original/....png">
+ *          <img src=".../small/....png">
+ *        </a>
+ *      → read the anchor href.
  *
- * Multi-image posts repeat the same structure; we return one ScrapedPost
+ *   2. GIFVs:
- * per anchor, matching how Pixiv/Twitter engines behave.
+ *        <div class="media-gallery__gifv autoplay">
+ *          <video class="media-gallery__item-gifv-thumbnail"
+ *                 src=".../original/....mp4" autoplay loop></video>
+ *        </div>
+ *      → read the <video> src (already the original .mp4).
+ *
+ *   3. Videos:
+ *        <div role="menuitem" class="video-player">
+ *          <video src=".../original/....mp4" poster=".../small/....png"></video>
+ *        </div>
+ *      → read the <video> src and use the poster attribute for extraContentUrl.
  */
 export default class Mastodon extends ScrapeEngineBase {
   name = "mastodon";
@@ -35,8 +43,6 @@ export default class Mastodon extends ScrapeEngineBase {
     "Rating is assumed to be safe."
   ];
 
-  // Explicit hosts we claim. Other Mastodon instances are picked up by the
-  // URL-shape check in canImport().
   supportedHosts = ["baraag.net"];
 
   canImport(url: Location): boolean {
@@ -59,9 +65,7 @@ export default class Mastodon extends ScrapeEngineBase {
       post.contentUrl = att.url;
       post.contentType = att.contentType;
       post.referrer = document.location.origin;
-      // post.rating = "safe";
 
-      // Cached preview as a poster frame for videos.
       if (att.contentType === "video" && att.previewUrl) {
         post.extraContentUrl = att.previewUrl;
       }
@@ -78,20 +82,23 @@ export default class Mastodon extends ScrapeEngineBase {
 
   private findAttachments(document: Document): Attachment[] {
     const out: Attachment[] = [];
+    const seen = new Set<string>();
 
-    // Primary selector, confirmed against baraag's rendered HTML for both
+    const add = (att: Attachment) => {
-    // single- and multi-image posts.
+      if (seen.has(att.url)) return;
+      seen.add(att.url);
+      out.push(att);
+    };
+
+    // 1. Images: anchor href points at the original.
     let anchors = Array.from(
       document.querySelectorAll<HTMLAnchorElement>("a.media-gallery__item-thumbnail[href]")
     );
-
-    // Looser fallback for themes that drop the class.
     if (anchors.length === 0) {
       anchors = Array.from(
         document.querySelectorAll<HTMLAnchorElement>("a[href*='/media_attachments/']")
       );
     }
-
     for (const a of anchors) {
       const rawHref = a.getAttribute("href");
       if (!rawHref) continue;
@@ -99,14 +106,32 @@ export default class Mastodon extends ScrapeEngineBase {
       const img = a.querySelector("img");
       const preview = img?.currentSrc || img?.src;
 
-      const url = this.normalizeUrl(rawHref);
+      const url = this.normalizeUrl(rawHref) || this.largestFromSrcset(img);
-      const finalUrl = url || this.largestFromSrcset(img);
+      if (!url) continue;
-      if (!finalUrl) continue;
 
-      out.push({
+      add({
-        url: finalUrl,
+        url,
-        previewUrl: preview && preview !== finalUrl ? preview : undefined,
+        previewUrl: preview && preview !== url ? preview : undefined,
-        contentType: this.contentTypeFor(finalUrl),
+        contentType: this.contentTypeFor(url),
+      });
+    }
+
+    // 2 & 3. Videos and GIFVs: <video> elements carry the original .mp4 in src.
+    const videos = Array.from(document.querySelectorAll<HTMLVideoElement>("video[src]"));
+    for (const v of videos) {
+      const rawSrc = v.getAttribute("src");
+      if (!rawSrc) continue;
+
+      const url = this.normalizeUrl(rawSrc);
+      if (!url) continue;
+
+      // Use the poster as a thumbnail when available.
+      const poster = v.getAttribute("poster") || undefined;
+
+      add({
+        url,
+        previewUrl: poster && poster !== url ? poster : undefined,
+        contentType: "video",
       });
     }
 
@@ -114,9 +139,9 @@ export default class Mastodon extends ScrapeEngineBase {
   }
 
   /**
-   * Strip Mastodon's ?sensitive=true marker and report whether it was present.
+   * Strip Mastodon's ?sensitive=true marker. The marker is metadata, not
-   * The marker is metadata, not part of the media URL, and leaving it in can
+   * part of the media URL, and leaving it in can confuse booru deduplication.
-   * confuse booru deduplication and content-type sniffing.
+   * Returns an empty string if the URL is unusable.
    */
   private normalizeUrl(href: string): string {
     try {
@@ -128,7 +153,6 @@ export default class Mastodon extends ScrapeEngineBase {
     }
   }
 
-  /** Pick the highest-resolution candidate out of an <img srcset>. */
   private largestFromSrcset(img: HTMLImageElement | null): string | undefined {
     if (!img) return undefined;
     const srcset = img.getAttribute("srcset");