umedia 0.3.0 → 0.4.1

package/lib/providers/x.js

@@ -1,69 +1,213 @@
 import { fetchJson, hostMatches } from "../util.js";
 import { Codes, UMediaError } from "../errors.js";
 
-/** X (Twitter) — syndication endpoint for public posts. Best-effort, honestly labeled. */
+/**
+ * X / Twitter — two proven public routes, no login:
+ *  1. syndication embed API with the computed token (the route embed widgets use)
+ *  2. GraphQL TweetDetail with a guest token (the route the logged-out web uses)
+ * Both are the exact techniques from yt-dlp's twitter extractor and cobalt's
+ * twitter service. Photos, videos (best bitrate mp4), gifs.
+ */
 export const name = "x";
 
+const BEARER = "AAAAAAAAAAAAAAAAAAAAANRILgAAAAAAnNwIzUejRCOuH5E6I8xnZz4puTs%3D1Zv7ttfk8LF81IUq16cHjhLTvJu4FA33AGWWjCpTnA";
+const GQL_FEATURES = JSON.stringify({
+  rweb_video_screen_enabled: false, payments_enabled: false, rweb_xchat_enabled: false,
+  profile_label_improvements_pcf_label_in_post_enabled: true, rweb_tipjar_consumption_enabled: true,
+  verified_phone_label_enabled: false, creator_subscriptions_tweet_preview_api_enabled: true,
+  responsive_web_graphql_timeline_navigation_enabled: true, responsive_web_graphql_skip_user_profile_image_extensions_enabled: false,
+  premium_content_api_read_enabled: false, communities_web_enable_tweet_community_results_fetch: true,
+  c9s_tweet_anatomy_moderator_badge_enabled: true, responsive_web_grok_analyze_button_fetch_trends_enabled: false,
+  responsive_web_grok_analyze_post_followups_enabled: true, responsive_web_jetfuel_frame: true,
+  responsive_web_grok_share_attachment_enabled: true, articles_preview_enabled: true,
+  responsive_web_edit_tweet_api_enabled: true, graphql_is_translatable_rweb_tweet_is_translatable_enabled: true,
+  view_counts_everywhere_api_enabled: true, longform_notetweets_consumption_enabled: true,
+  responsive_web_twitter_article_tweet_consumption_enabled: true, tweet_awards_web_tipping_enabled: false,
+  creator_subscriptions_quote_tweet_preview_enabled: false, freedom_of_speech_not_reach_fetch_enabled: true,
+  standardized_nudges_misinfo: true, tweet_with_visibility_results_prefer_gql_limited_actions_policy_enabled: true,
+  longform_notetweets_rich_text_read_enabled: true, longform_notetweets_inline_media_enabled: true,
+  responsive_web_grok_image_annotation_enabled: true, responsive_web_grok_imagine_annotation_enabled: true,
+  responsive_web_grok_community_note_auto_translation_is_enabled: false, responsive_web_enhance_cards_enabled: false,
+});
+
+const GQL_HEADERS = {
+  authorization: `Bearer ${BEARER}`,
+  "x-twitter-client-language": "en",
+  "x-twitter-active-user": "yes",
+  "accept-language": "en",
+  "user-agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/126.0 Safari/537.36",
+};
+
 export function canHandle(url) {
   try {
     return hostMatches(new URL(String(url)).hostname, ["twitter.com", "x.com"]);
   } catch { return false; }
 }
 
+export function tweetId(url) {
+  const m = String(url).match(/\/status(?:es)?\/(\d{8,25})/);
+  return m ? m[1] : null;
+}
+
+// the syndication embed token — computed from the tweet id (yt-dlp technique)
+const syndicationToken = (id) => ((Number(id) / 1e15) * Math.PI).toString(36).replace(/(0+|\.)/g, "");
+
+let _guestToken = null;
+async function guestToken() {
+  if (_guestToken) return _guestToken;
+  try {
+    const j = await fetchJson("https://api.x.com/1.1/guest/activate.json", {
+      method: "POST", headers: GQL_HEADERS,
+    }, 15_000);
+    if (j?.guest_token) return (_guestToken = j.guest_token);
+  } catch { /* syndication may still work */ }
+  return null;
+}
+
+function mediaFromEntities(entities, media) {
+  for (const m of entities || []) {
+    const isPhoto = m.type === "photo";
+    const isVideo = m.type === "video" || m.type === "animated_gif";
+    if (isPhoto && m.media_url_https) {
+      media.push({
+        type: "image", index: media.length, url: m.media_url_https, thumbnail: m.media_url_https,
+        mimeType: "image/jpeg", width: m.original_info?.width ?? null, height: m.original_info?.height ?? null,
+        duration: null, size: null, quality: "original",
+        hasAudio: false, hasVideo: false, source: name, variants: [],
+      });
+    } else if (isVideo && m.video_info?.variants?.length) {
+      const mp4s = m.video_info.variants
+        .filter((v) => v.content_type === "video/mp4" && v.url)
+        .map((v) => ({
+          height: null, width: null,
+          // cobalt strips the flaky `tag` param from twitter mp4 urls
+          url: (() => { try { const u = new URL(v.url); u.searchParams.delete("tag"); return u.toString(); } catch { return v.url; } })(),
+          hasAudio: true, hasVideo: true,
+          quality: v.bitrate ? `${Math.round(v.bitrate / 1000)}kbps` : "mp4",
+          mimeType: "video/mp4", bitrate: v.bitrate ?? null,
+        }))
+        .sort((a, b) => (b.bitrate || 0) - (a.bitrate || 0));
+      const best = mp4s[0]
+        || m.video_info.variants.filter((v) => v.url).map((v) => ({
+          height: null, width: null, url: v.url, hasAudio: true, hasVideo: true,
+          quality: null, mimeType: v.content_type || "application/x-mpegURL", bitrate: null,
+        }))[0];
+      if (best) {
+        media.push({
+          type: "video", index: media.length, url: best.url,
+          thumbnail: m.media_url_https || null,
+          mimeType: best.mimeType, width: m.original_info?.width ?? null, height: m.original_info?.height ?? null,
+          duration: m.video_info.duration_millis ? Math.round(m.video_info.duration_millis / 1000) : null,
+          size: null, quality: best.quality,
+          hasAudio: true, hasVideo: true, source: name, variants: mp4s,
+        });
+      }
+    }
+  }
+}
+
 export async function resolve(url, ctx = {}) {
   const attempt = { provider: name, status: "skipped" };
   ctx.attempts?.push(attempt);
-  const m = String(url).match(/status\/(\d+)/);
-  if (!m) throw new UMediaError(Codes.INVALID_URL, `No status id in ${url}`);
   const t0 = Date.now();
+
+  const id = tweetId(url);
+  if (!id) {
+    attempt.status = "failed";
+    throw new UMediaError(Codes.INVALID_URL, "Not a tweet URL (expected x.com/{user}/status/{id})", { attempts: [attempt] });
+  }
+
+  const media = [];
+  let title = null;
+  let author = null;
+
+  // 1) GraphQL TweetDetail with guest token (fullest data)
   try {
-    const tw = await fetchJson(`https://cdn.syndication.twimg.com/tweet-result?id=${m[1]}&lang=en`, {}, 12_000);
-    const media = [];
-    const photos = tw.photos || [];
-    photos.forEach((p, i) => media.push({
-      type: "image", index: i, url: p.url, thumbnail: p.url,
-      mimeType: "image/jpeg", width: p.width ?? null, height: p.height ?? null,
-      duration: null, size: null, quality: p.width && p.height ? `${p.width}x${p.height}` : null,
-      hasAudio: false, hasVideo: false, source: name, variants: [],
-    }));
-    const vids = tw.video?.variants || [];
-    const mp4s = vids.filter((v) => v.src?.includes(".mp4")).sort((a, b) => (b.bitrate || 0) - (a.bitrate || 0));
-    if (mp4s.length) {
-      const best = mp4s[0];
-      media.push({
-        type: "video", index: media.length, url: best.src, thumbnail: tw.video?.poster ?? null,
-        mimeType: "video/mp4", width: null, height: null,
-        duration: tw.video?.duration_millis ? tw.video.duration_millis / 1000 : null,
-        size: null, quality: best.bitrate ? `~${Math.round(best.bitrate / 1000)}kbps` : null,
-        hasAudio: true, hasVideo: true, source: name,
-        variants: mp4s.map((v) => ({
-          quality: v.bitrate ? `~${Math.round(v.bitrate / 1000)}kbps` : "unknown",
-          url: v.src, mimeType: "video/mp4", hasAudio: true, hasVideo: true,
-        })),
-      });
+    const token = await guestToken();
+    if (token) {
+      const u = new URL("https://api.x.com/graphql/4Siu98E55GquhG52zHdY5w/TweetDetail");
+      u.searchParams.set("variables", JSON.stringify({
+        focalTweetId: id, with_rux_injections: false, rankingMode: "Relevance",
+        includePromotedContent: true, withCommunity: true,
+        withQuickPromoteEligibilityTweetFields: true, withBirdwatchNotes: true, withVoice: true,
+      }));
+      u.searchParams.set("features", GQL_FEATURES);
+      u.searchParams.set("fieldToggles", JSON.stringify({ withArticleRichContentState: true, withArticlePlainText: false }));
+      const j = await fetchJson(u.toString(), {
+        headers: { ...GQL_HEADERS, "content-type": "application/json", "x-guest-token": token, cookie: `guest_id=v1%3A${token}` },
+      }, 20_000);
+      const entries = j?.data?.threaded_conversation_with_injections_v2?.instructions
+        ?.find((i) => i.type === "TimelineAddEntries")?.entries || [];
+      const node = entries.find((e) => e.entryId === `tweet-${id}`)?.content?.itemContent?.tweet_results?.result;
+      let base = node?.legacy;
+      if (node?.__typename === "TweetWithVisibilityResults") base = node.tweet?.legacy;
+      const rep = base?.retweeted_status_result?.result?.legacy?.extended_entities
+        || base?.retweeted_status_result?.result?.tweet?.legacy?.extended_entities;
+      if (base?.full_text) title = base.full_text;
+      author = node?.core?.user_results?.result?.legacy?.screen_name
+        || node?.tweet?.core?.user_results?.result?.legacy?.screen_name || null;
+      mediaFromEntities(rep?.media || base?.extended_entities?.media, media);
     }
-    if (!media.length) throw new UMediaError(Codes.MEDIA_NOT_FOUND, "Post has no media (or it is not public)");
-    attempt.status = "success";
-    attempt.latencyMs = Date.now() - t0;
-    return {
-      result: {
-        sourceUrl: url, platform: "x", title: (tw.text || "").slice(0, 120) || null,
-        author: tw.user?.screen_name ?? null, thumbnail: media[0].thumbnail ?? null,
-        itemCount: media.length,
-        counts: {
-          images: media.filter((x) => x.type === "image").length,
-          videos: media.filter((x) => x.type === "video").length, audios: 0, other: 0,
-        },
-        truncated: false, media,
-      },
-      engine: { attempts: ctx.attempts || [attempt], finalProvider: name, totalLatencyMs: Date.now() - t0 },
-    };
   } catch (e) {
+    attempt.note = `graphql: ${String(e.message).slice(0, 60)}`;
+  }
+
+  // 2) syndication embed API with the computed token (embed-widget route)
+  if (!media.length) {
+    try {
+      const j = await fetchJson(
+        `https://cdn.syndication.twimg.com/tweet-result?id=${id}&token=${syndicationToken(id)}&lang=en`, {}, 15_000);
+      title = title || j?.text || null;
+      author = author || j?.user?.screen_name || null;
+      mediaFromEntities(j?.mediaDetails, media);
+      if (!media.length && j?.video?.variants) {
+        const mp4s = j.video.variants.filter((v) => v.type === "video/mp4").sort((a, b) => (b.bitrate || 0) - (a.bitrate || 0));
+        if (mp4s[0]) media.push({
+          type: "video", index: 0, url: mp4s[0].src, thumbnail: j.thumbnail?.image?.src || null,
+          mimeType: "video/mp4", width: null, height: null,
+          duration: j.video.duration_millis ? Math.round(j.video.duration_millis / 1000) : null,
+          size: null, quality: mp4s[0].bitrate ? `${Math.round(mp4s[0].bitrate / 1000)}kbps` : null,
+          hasAudio: true, hasVideo: true, source: name,
+          variants: mp4s.map((v) => ({ height: null, width: null, url: v.src, hasAudio: true, hasVideo: true, quality: v.bitrate ? `${Math.round(v.bitrate / 1000)}kbps` : null, mimeType: "video/mp4", bitrate: v.bitrate ?? null })),
+        });
+      }
+    } catch (e) {
+      attempt.note = `${attempt.note ? attempt.note + " · " : ""}syndication: ${String(e.message).slice(0, 60)}`;
+    }
+  }
+
+  if (!media.length) {
     attempt.status = "failed";
-    attempt.error = e.message;
-    if (e instanceof UMediaError) throw e;
-    throw new UMediaError(Codes.PROVIDER_UNAVAILABLE, `X unavailable: ${e.message}`, { attempts: [attempt] });
+    attempt.error = j_err(attempt.note);
+    throw new UMediaError(Codes.MEDIA_NOT_FOUND,
+      `X exposes no media for this tweet (${attempt.error}) — protected/removed tweets stay inaccessible; the yt-dlp tier adds cookie support.`, { attempts: [attempt] });
   }
+
+  attempt.status = "success";
+  attempt.latencyMs = Date.now() - t0;
+  return {
+    result: {
+      sourceUrl: url,
+      platform: name,
+      title: title ? title.slice(0, 200) : null,
+      author,
+      thumbnail: media[0].thumbnail,
+      itemCount: media.length,
+      counts: {
+        images: media.filter((m) => m.type === "image").length,
+        videos: media.filter((m) => m.type === "video").length,
+        audios: 0, other: 0,
+      },
+      truncated: false,
+      media,
+      streamUrlsAvailable: true,
+    },
+    engine: { attempts: ctx.attempts || [attempt], finalProvider: name, totalLatencyMs: Date.now() - t0 },
+  };
+}
+
+function j_err(note) {
+  return (note || "no media in public routes").replace(/^.*?(graphql|syndication)/, "$1");
 }
 
 export async function search() {

package/lib/registry.js

@@ -10,13 +10,17 @@ import * as bluesky from "./providers/bluesky.js";
 import * as reddit from "./providers/reddit.js";
 import * as xprov from "./providers/x.js";
 import * as ig from "./providers/instagram.js";
-import * as ogembed from "./providers/ogembed.js";
+import * as facebook from "./providers/facebook.js";
+import * as threads from "./providers/threads.js";
+import * as snapchat from "./providers/snapchat.js";
+import * as rumble from "./providers/rumble.js";
+import * as tumblr from "./providers/tumblr.js";
 import * as ytdlp from "./providers/ytdlp.js";
 import * as itunes from "./providers/itunes.js";
 import { Codes, UMediaError } from "./errors.js";
 import { validateUrl } from "./util.js";
 
-const NATIVE = [yt, tiktok, pinterest, soundcloud, dailymotion, vimeo, streamable, twitch, bluesky, reddit, xprov, ig, ogembed];
+const NATIVE = [yt, tiktok, pinterest, soundcloud, dailymotion, vimeo, streamable, twitch, bluesky, reddit, xprov, ig, facebook, threads, snapchat, rumble, tumblr];
 
 export function detectPlatform(url) {
   for (const p of NATIVE) {
@@ -57,6 +61,7 @@ export async function resolve(url) {
   }
 
   const ctx = { attempts: [] };
+  let lastTyped = null;
   for (const p of NATIVE) {
     if (!p.canHandle(url)) continue;
     try {
@@ -69,6 +74,7 @@ export async function resolve(url) {
         throw e;
       }
       // transient/provider failure → try the next layer
+      if (e instanceof UMediaError) lastTyped = e;
       continue;
     }
   }
@@ -82,6 +88,12 @@ export async function resolve(url) {
     }
   }
   const last = ctx.attempts[ctx.attempts.length - 1];
+  // surface the real provider error when every matching provider failed —
+  // never mask a typed outcome behind a generic message
+  if (lastTyped) {
+    lastTyped.attempts = ctx.attempts;
+    throw lastTyped;
+  }
   throw new UMediaError(
     detectPlatform(url) === "unknown" ? Codes.INVALID_URL : Codes.PROVIDER_UNAVAILABLE,
     `No provider could resolve this URL.${last?.error ? " Last error: " + last.error : ""}` +
@@ -115,10 +127,14 @@ export function capabilities() {
       { platform: "streamable", search: false, resolve: true, download: true, notes: "direct mp4 renditions" },
       { platform: "twitch", search: false, resolve: true, download: true, notes: "clips = direct mp4s; VODs/livestreams via yt-dlp tier" },
       { platform: "bluesky", search: false, resolve: true, download: true, notes: "public AppView API: full-size images + video playlists" },
-      { platform: "instagram", search: false, resolve: true, download: true, notes: "public embed best-effort" },
-      { platform: "reddit", search: false, resolve: true, download: true, notes: "posts + galleries" },
-      { platform: "x", search: false, resolve: true, download: true, notes: "public posts via syndication" },
-      { platform: "facebook · threads · snapchat · rumble · tumblr", search: false, resolve: "best-effort", download: "best-effort", notes: "official Open Graph media; often login-gated on datacenter IPs — yt-dlp tier reliable" },
+      { platform: "instagram", search: false, resolve: true, download: true, notes: "oEmbed→mobile info API: photos, videos, full carousels" },
+      { platform: "reddit", search: false, resolve: true, download: true, notes: "posts + galleries + video with audio sidecar" },
+      { platform: "x", search: false, resolve: true, download: true, notes: "syndication token + guest GraphQL — photos/videos/gifs" },
+      { platform: "facebook", search: false, resolve: true, download: true, notes: "reels/watch/videos/shares/fb.watch via the player JSON" },
+      { platform: "threads", search: false, resolve: true, download: true, notes: "post SSR state + og media" },
+      { platform: "snapchat", search: false, resolve: true, download: true, notes: "spotlight videos + public story snap lists" },
+      { platform: "rumble", search: false, resolve: true, download: true, notes: "embed API: direct mp4 renditions" },
+      { platform: "tumblr", search: false, resolve: true, download: true, notes: "mobile API: video/audio/photo posts + reblog trails" },
       { platform: "itunes", search: true, resolve: false, download: "previews", notes: "real 30s clips" },
       { platform: "ytdlp", search: false, resolve: Boolean(ytdlp.binary()), download: Boolean(ytdlp.binary()), notes: "optional tier: 1800+ sites — install yt-dlp" },
     ],
@@ -140,10 +156,14 @@ export function status() {
       { name: "streamable", state: "operational", via: "public video api" },
       { name: "twitch", state: "operational", via: "public GraphQL (clips)" },
       { name: "bluesky", state: "operational", via: "public AppView XRPC" },
-      { name: "instagram", state: "degraded", via: "public embed (best-effort)" },
+      { name: "instagram", state: "operational", via: "oEmbed + mobile info api" },
       { name: "reddit", state: "operational", via: "public json" },
-      { name: "x", state: "degraded", via: "syndication (best-effort)" },
-      { name: "ogembed", state: "degraded", via: "open graph meta (best-effort)" },
+      { name: "x", state: "operational", via: "syndication + guest graphql" },
+      { name: "facebook", state: "operational", via: "web player json" },
+      { name: "threads", state: "operational", via: "share page state" },
+      { name: "snapchat", state: "operational", via: "spotlight + story state" },
+      { name: "rumble", state: "operational", via: "embed json api" },
+      { name: "tumblr", state: "operational", via: "mobile api" },
       { name: "itunes", state: "operational", via: "search api" },
       { name: "ytdlp", state: ytdlp.binary() ? "operational" : "unavailable", via: "external binary" },
     ],

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "umedia",
-  "version": "0.3.0",
+  "version": "0.4.1",
   "description": "Standalone media engine — search, resolve and download media from YouTube, TikTok, Instagram, Reddit, X, iTunes (and 1800+ sites via the yt-dlp tier) entirely on your machine. No hosted API, no keys, no server bills. Honest quality labels, complete galleries, typed errors.",
   "type": "module",
   "main": "./index.js",

package/lib/providers/ogembed.js

@@ -1,111 +0,0 @@
-import { fetchText, hostMatches } from "../util.js";
-import { Codes, UMediaError } from "../errors.js";
-
-/**
- * og-embed tier — Facebook, Threads, Snapchat Spotlight, Rumble, Tumblr.
- * These platforms publish official media previews as Open Graph meta tags on
- * their public share pages; some networks gate them (typed honest errors there,
- * the yt-dlp tier is the reliable path). No scraping tricks — their own tags.
- */
-export const name = "ogembed";
-
-const HOSTS = ["facebook.com", "fb.watch", "threads.net", "threads.com", "snapchat.com", "rumble.com", "tumblr.com"];
-
-export function canHandle(url) {
-  try {
-    return hostMatches(new URL(String(url)).hostname, HOSTS);
-  } catch { return false; }
-}
-
-function meta(h, key) {
-  const pats = [
-    new RegExp(`<meta[^>]+(?:property|name)="${key}"[^>]+content="([^"]*)"`, "i"),
-    new RegExp(`<meta[^>]+content="([^"]*)"[^>]+(?:property|name)="${key}"`, "i"),
-  ];
-  for (const p of pats) {
-    const m = h.match(p);
-    if (m) return m[1].replace(/&amp;/g, "&").replace(/&#(\d+);/g, (_, n) => String.fromCharCode(n));
-  }
-  return null;
-}
-
-export async function resolve(url, ctx = {}) {
-  const attempt = { provider: name, status: "skipped" };
-  ctx.attempts?.push(attempt);
-  const t0 = Date.now();
-  const host = (() => { try { return new URL(String(url)).hostname.replace(/^www\./, ""); } catch { return "?"; } })();
-
-  let html;
-  try {
-    const r = await fetchText(String(url), {}, 18_000);
-    if (r.status >= 400) {
-      attempt.status = "failed";
-      attempt.error = `HTTP ${r.status}`;
-      throw new UMediaError(Codes.PROVIDER_UNAVAILABLE, `${host} returned HTTP ${r.status}`, { attempts: [attempt] });
-    }
-    html = r.text;
-  } catch (e) {
-    if (e instanceof UMediaError) throw e;
-    attempt.status = "failed";
-    attempt.error = e.message;
-    throw new UMediaError(Codes.PROVIDER_UNAVAILABLE, `${host} not retrievable. ${e.message}`, { attempts: [attempt] });
-  }
-
-  const video = meta(html, "og:video:secure_url") || meta(html, "og:video") || meta(html, "og:video:url");
-  const image = meta(html, "og:image:secure_url") || meta(html, "og:image");
-  const title = meta(html, "og:title");
-  const desc = meta(html, "og:description");
-  const author = meta(html, "og:article:author") || meta(html, "twitter:creator") || null;
-
-  const media = [];
-  if (video && /\.mp4|\.m3u8|video/i.test(video)) {
-    media.push({
-      type: "video", index: 0, url: video, thumbnail: image,
-      mimeType: /\.m3u8/i.test(video) ? "application/x-mpegURL" : "video/mp4",
-      width: parseInt(meta(html, "og:video:width") || "", 10) || null,
-      height: parseInt(meta(html, "og:video:height") || "", 10) || null,
-      duration: parseInt(meta(html, "video:duration") || meta(html, "og:video:duration") || "", 10) || null,
-      size: null, quality: null,
-      hasAudio: true, hasVideo: true, source: name, variants: [],
-    });
-  }
-  if (image && (!video || !/\.mp4/i.test(video))) {
-    media.push({
-      type: "image", index: media.length, url: image, thumbnail: image,
-      mimeType: "image/jpeg", width: null, height: null, duration: null, size: null,
-      quality: null, hasAudio: false, hasVideo: false, source: name, variants: [],
-    });
-  }
-
-  if (!media.length) {
-    attempt.status = "failed";
-    attempt.error = "no media in the public share page (login-gated or removed)";
-    throw new UMediaError(Codes.MEDIA_NOT_FOUND,
-      `${host} exposes no media on its public page (often login-gated from datacenter IPs) — the yt-dlp tier is the reliable path. ${attempt.error}`, { attempts: [attempt] });
-  }
-
-  attempt.status = "success";
-  attempt.latencyMs = Date.now() - t0;
-  return {
-    result: {
-      sourceUrl: url,
-      platform: host,
-      title: title || desc || null,
-      author,
-      thumbnail: image || media[0].thumbnail,
-      itemCount: media.length,
-      counts: {
-        images: media.filter((m) => m.type === "image").length,
-        videos: media.filter((m) => m.type === "video").length,
-        audios: 0, other: 0,
-      },
-      truncated: false,
-      media,
-    },
-    engine: { attempts: ctx.attempts || [attempt], finalProvider: name, totalLatencyMs: Date.now() - t0 },
-  };
-}
-
-export async function search() {
-  return [];
-}