refactor(rumble): un seul cœur de scraping, routeur fin, contrat 503 CF

- server/providers/rumble.mjs devient l'unique implémentation : fetch
  (cookie jar __cf_bm + curl_cffi + cooldown global 60 s après un échec
  total), parsing liste/vidéo avec surcouche JSON-LD partagée, vues via
  parseRumbleViews partout (« 1,2K » n'est plus lu en 12 sur /api/rumble).
- server/rumble.mjs = couche HTTP mince : limiter 20/min, cache 60 s,
  5 routes qui délèguent au cœur ; /search passe par le registre de
  providers (cache SQLite, negative cache et channelRef partagés avec
  /api/search).
- Contrat d'erreur : un blocage Cloudflare est un 503
  rumble_cloudflare_challenge, plus jamais un 404 (qui faisait retirer
  les vidéos du flux Shorts lors d'un challenge) ; page reçue sans
  identité = 404 rumble_video_not_found.
- Code mort supprimé : scrapeRumbleVideo d'index.mjs (191 l., 3e
  implémentation jamais appelée), rumbleLimiter 10/min jamais monté,
  import cheerio devenu inutile. Net -238 lignes.
- Tests : parseur générique (vues, durées, ids /shorts/), contrat 503
  sans réseau via le cooldown, durées absentes -> undefined.
- Docs : API_MCP_GUIDE §5.5 (routes, rate-limit réel, erreurs) + note
  README « Rumble peut être temporairement absent selon le réseau ».

server/index.mjs

@@ -12,7 +12,6 @@ import { execFile as execFileCb } from 'node:child_process';
 import { promisify } from 'node:util';
 import { fileURLToPath as serverFileURLToPath } from 'node:url';
 import ffmpegPath from 'ffmpeg-static';
-import * as cheerio from 'cheerio';
 import axios from 'axios';
 import rumbleRouter from './rumble.mjs';
 import { providerRegistry, getProviderAdapter, validateProviders, SUGGESTION_CONTRACT_VERSION } from './providers/registry.mjs';
@@ -1140,15 +1139,6 @@ function formatsCacheSet(key, data) {
   }
 }
 
-// Rate limiter for Rumble scraping to prevent being blocked
-const rumbleLimiter = rateLimit({
-  windowMs: 60 * 1000, // 1 min
-  max: 10, // Limit to 10 requests per minute per IP
-  standardHeaders: true,
-  legacyHeaders: false,
-  message: { error: 'Too many requests to Rumble API. Please try again later.' }
-});
-
 function makeAccessToken(userId, sessionId) {
   const payload = { sub: userId, sid: sessionId };
   return jwt.sign(payload, JWT_SECRET, { expiresIn: `${ACCESS_TTL_MIN}m` });
@@ -1300,198 +1290,6 @@ function providerUrlFrom(provider, videoId, { instance, slug, sourceUrl }) {
   }
 }
 
-async function scrapeRumbleVideo(videoId) {
-  const url = `https://rumble.com/${videoId}`;
-  
-  try {
-    const response = await axios.get(url, {
-      headers: {
-        'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.36',
-        'Accept': 'text/html,application/xhtml+xml,application/xml;q=0.9,image/webp,*/*;q=0.8',
-        'Accept-Language': 'en-US,en;q=0.5',
-        'Accept-Encoding': 'gzip, deflate, br',
-        'DNT': '1',
-        'Connection': 'keep-alive',
-        'Upgrade-Insecure-Requests': '1',
-        'Cache-Control': 'no-cache'
-      },
-      timeout: 15000,
-      maxRedirects: 5,
-      validateStatus: function (status) {
-        return status >= 200 && status < 400; // Accept redirects
-      }
-    });
-    
-    const $ = cheerio.load(response.data);
-    const html = response.data;
-    
-    // Extract basic video information
-    let title = '';
-    let thumbnail = '';
-    let uploaderName = '';
-    let uploaderAvatar = '';
-    let views = 0;
-    let duration = 0;
-    let uploadedDate = '';
-    let description = '';
-    
-    // Try multiple selectors for title (Rumble's HTML structure can vary)
-    title = $('h1.video-title, .video-title h1, [data-video-title]').first().text().trim() ||
-             $('meta[property="og:title"]').attr('content') ||
-             $('title').text().trim() ||
-             $('h1').first().text().trim() || '';
-    
-    // Clean up title (remove site name if present)
-    title = title.replace(/\s*\|\s*Rumble$/i, '').trim();
-    
-    // Extract thumbnail with fallbacks
-    thumbnail = $('meta[property="og:image"], meta[name="twitter:image"]').attr('content') ||
-                $('meta[property="og:image:secure_url"]').attr('content') ||
-                $('.video-thumbnail img, .thumbnail img').attr('src') || '';
-    
-    // Make thumbnail URL absolute if relative
-    if (thumbnail && !thumbnail.startsWith('http')) {
-      thumbnail = thumbnail.startsWith('//') ? 'https:' + thumbnail : 'https://rumble.com' + thumbnail;
-    }
-    
-    // Extract uploader information with multiple selectors
-    uploaderName = $('.media-by--a, .channel-name, .uploader-name').first().text().trim() || 
-                   $('meta[property="article:author"]').attr('content') ||
-                   $('.author-name, .channel-link').first().text().trim() || '';
-    
-    uploaderAvatar = $('.channel-avatar img, .uploader-avatar img').attr('src') ||
-                     $('.author-avatar img').attr('src') || '';
-    
-    // Make uploader avatar URL absolute
-    if (uploaderAvatar && !uploaderAvatar.startsWith('http')) {
-      uploaderAvatar = uploaderAvatar.startsWith('//') ? 'https:' + uploaderAvatar : 'https://rumble.com' + uploaderAvatar;
-    }
-    
-    // Extract views with better parsing
-    const viewsText = $('.rumbles-views, .video-views, .views-count, .video-info .views').first().text().trim();
-    if (viewsText) {
-      const viewsMatch = viewsText.match(/([\d,]+(?:\.\d+)?)\s*(K|M|B)?/i);
-      if (viewsMatch) {
-        let num = parseFloat(viewsMatch[1].replace(/,/g, ''));
-        const multiplier = viewsMatch[2]?.toUpperCase();
-        if (multiplier === 'K') num *= 1000;
-        else if (multiplier === 'M') num *= 1000000;
-        else if (multiplier === 'B') num *= 1000000000;
-        views = Math.floor(num);
-      } else {
-        // Try direct number parsing
-        const directMatch = viewsText.match(/(\d+(?:,\d+)*)/);
-        if (directMatch) {
-          views = parseInt(directMatch[1].replace(/,/g, ''));
-        }
-      }
-    }
-    
-    // Extract duration with improved parsing
-    const durationText = $('meta[property="video:duration"]').attr('content') ||
-                        $('video').attr('duration') ||
-                        $('.video-duration, .duration, .video-time').first().text().trim() ||
-                        $('.time-duration').text().trim();
-    
-    if (durationText) {
-      if (!isNaN(durationText)) {
-        duration = parseInt(durationText);
-      } else {
-        // Parse various duration formats
-        const timeMatch = durationText.match(/(\d+):(\d+)(?::(\d+))?/);
-        if (timeMatch) {
-          const hours = parseInt(timeMatch[3] || '0');
-          const minutes = parseInt(timeMatch[1]);
-          const seconds = parseInt(timeMatch[2]);
-          duration = hours * 3600 + minutes * 60 + seconds;
-        } else {
-          // Try HH:MM:SS format or MM:SS
-          const parts = durationText.split(':').map(p => parseInt(p.trim()) || 0);
-          if (parts.length === 3) {
-            duration = parts[0] * 3600 + parts[1] * 60 + parts[2];
-          } else if (parts.length === 2) {
-            duration = parts[0] * 60 + parts[1];
-          }
-        }
-      }
-    }
-    
-    // Extract upload date
-    uploadedDate = $('meta[property="article:published_time"]').attr('content') ||
-                   $('.upload-date, .published-date, .video-date').first().text().trim() || '';
-    
-    // Try to parse relative dates
-    if (!uploadedDate || uploadedDate.includes('ago')) {
-      const relativeDate = $('.upload-date, .published-date').first().text().trim();
-      if (relativeDate && relativeDate.includes('ago')) {
-        // Convert relative date to ISO string (simple conversion)
-        const now = new Date();
-        if (relativeDate.includes('hour')) {
-          const hours = parseInt(relativeDate.match(/(\d+)/)?.[1] || '1');
-          now.setHours(now.getHours() - hours);
-          uploadedDate = now.toISOString();
-        } else if (relativeDate.includes('day')) {
-          const days = parseInt(relativeDate.match(/(\d+)/)?.[1] || '1');
-          now.setDate(now.getDate() - days);
-          uploadedDate = now.toISOString();
-        }
-      }
-    }
-    
-    // Extract description
-    description = $('meta[property="og:description"]').attr('content') ||
-                  $('.video-description, .description, .video-summary').first().text().trim() || '';
-    
-    // Extract video ID from various sources
-    let extractedVideoId = videoId;
-    const videoIdMatch = html.match(/"video_id"\s*:\s*"([^"]+)"/) || 
-                        html.match(/video[_-]id["\s:]+([^"\s]+)/) ||
-                        html.match(/embed\/([^/?]+)/);
-    if (videoIdMatch && videoIdMatch[1]) {
-      extractedVideoId = videoIdMatch[1];
-    }
-    
-    // Validate extracted data
-    const isValidVideo = title || thumbnail || uploaderName;
-    
-    return {
-      videoId: extractedVideoId,
-      title: title || 'Untitled Video',
-      thumbnail,
-      uploaderName: uploaderName || 'Unknown Uploader',
-      uploaderAvatar: uploaderAvatar || thumbnail,
-      views: Math.max(0, views),
-      duration: Math.max(0, duration),
-      uploadedDate: uploadedDate || new Date().toISOString(),
-      description,
-      url,
-      type: 'video',
-      scraped: true,
-      confidence: isValidVideo ? 'high' : 'low'
-    };
-  } catch (error) {
-    console.error('Erreur scraping Rumble:', error.message);
-    
-    // Return minimal data for fallback with error info
-    return {
-      videoId,
-      title: 'Video unavailable',
-      thumbnail: '',
-      uploaderName: 'Unknown',
-      uploaderAvatar: '',
-      views: 0,
-      duration: 0,
-      uploadedDate: '',
-      description: '',
-      url,
-      type: 'video',
-      error: error.message,
-      scraped: false,
-      confidence: 'none'
-    };
-  }
-}
-
 function guessContentTypeByExt(ext) {
   const e = String(ext || '').toLowerCase();
   if (e === 'mp4' || e === 'm4v') return 'video/mp4';