brave-real-browser-mcp-server 2.25.0 → 2.26.0

package/dist/handlers/advanced-tools.js

@@ -2998,17 +2998,62 @@ export async function handleWebCrawler(page, args) {
             browserPoolOptions: {
                 maxOpenPagesPerBrowser: 1,
             },
-            // Pre-navigation hook for rate limiting and proxy
+            // Pre-navigation hook for rate limiting, popup blocking, and movie streaming optimizations
             preNavigationHooks: [
                 async (crawlingContext) => {
                     await enforceRateLimit();
+                    const pg = crawlingContext.page;
                     // Set custom user agent if provided
                     if (args.userAgent) {
-                        await crawlingContext.page.setUserAgent(args.userAgent);
+                        await pg.setUserAgent(args.userAgent);
                     }
                     // Set custom headers if provided
                     if (args.headers) {
-                        await crawlingContext.page.setExtraHTTPHeaders(args.headers);
+                        await pg.setExtraHTTPHeaders(args.headers);
+                    }
+                    // Block popups and overlay ads (default: true for movie streaming)
+                    if (args.blockPopups !== false) {
+                        await pg.evaluateOnNewDocument(() => {
+                            // Block window.open popups
+                            window.open = () => null;
+                            // Block alert, confirm, prompt
+                            window.alert = () => { };
+                            window.confirm = () => true;
+                            window.prompt = () => null;
+                            // Block popup via createElement
+                            const origCreate = document.createElement.bind(document);
+                            document.createElement = (tag) => {
+                                if (tag.toLowerCase() === 'a' && arguments[1]?.target === '_blank') {
+                                    return origCreate('span');
+                                }
+                                return origCreate(tag);
+                            };
+                        });
+                    }
+                    // Block overlay ads and floating elements
+                    if (args.blockOverlayAds !== false) {
+                        await pg.evaluateOnNewDocument(() => {
+                            // Remove overlay ads after DOM load
+                            const removeOverlays = () => {
+                                const selectors = [
+                                    '[class*="popup"]', '[class*="modal"]', '[class*="overlay"]',
+                                    '[id*="popup"]', '[id*="modal"]', '[id*="overlay"]',
+                                    '[class*="ad-"]', '[class*="-ad"]', '[class*="advert"]',
+                                    '[class*="banner"]', '[class*="sticky"]', '[class*="float"]',
+                                    'div[style*="position: fixed"]', 'div[style*="z-index: 9"]',
+                                ];
+                                selectors.forEach(sel => {
+                                    document.querySelectorAll(sel).forEach(el => {
+                                        const style = window.getComputedStyle(el);
+                                        if (style.position === 'fixed' || style.zIndex > '1000') {
+                                            el.style.display = 'none';
+                                        }
+                                    });
+                                });
+                            };
+                            document.addEventListener('DOMContentLoaded', removeOverlays);
+                            setInterval(removeOverlays, 2000);
+                        });
                     }
                 },
             ],
@@ -3051,6 +3096,85 @@ export async function handleWebCrawler(page, args) {
                             }
                         }
                     }
+                    // Extract video links (JWPlayer, DooPlayer, iframes, ajax sources)
+                    if (args.extractVideoLinks !== false) {
+                        result.videoLinks = await crawlerPage.evaluate(() => {
+                            const videoLinks = [];
+                            const videoPatterns = /\.(m3u8|mp4|mkv|webm|avi|mov|flv|wmv|ts)(\?|$)/i;
+                            // 1. JWPlayer detection
+                            if (window.jwplayer) {
+                                try {
+                                    const players = document.querySelectorAll('.jwplayer, [id*="jwplayer"]');
+                                    players.forEach((_, idx) => {
+                                        try {
+                                            const player = window.jwplayer(idx);
+                                            if (player && player.getPlaylistItem) {
+                                                const item = player.getPlaylistItem();
+                                                if (item?.file) {
+                                                    videoLinks.push({ url: item.file, type: item.file.includes('.m3u8') ? 'm3u8' : 'mp4', source: 'jwplayer' });
+                                                }
+                                                if (item?.sources) {
+                                                    item.sources.forEach((s) => {
+                                                        if (s.file)
+                                                            videoLinks.push({ url: s.file, type: s.type || 'mp4', source: 'jwplayer' });
+                                                    });
+                                                }
+                                            }
+                                        }
+                                        catch { }
+                                    });
+                                }
+                                catch { }
+                            }
+                            // 2. DooPlayer detection (common in movie sites)
+                            if (window.dooPlayer || document.querySelector('[id*="doo"]')) {
+                                try {
+                                    const dooConfig = window.dooPlayer?.config || window.player_config;
+                                    if (dooConfig?.source) {
+                                        videoLinks.push({ url: dooConfig.source, type: 'm3u8', source: 'dooplayer' });
+                                    }
+                                }
+                                catch { }
+                            }
+                            // 3. Iframe video sources
+                            document.querySelectorAll('iframe').forEach(iframe => {
+                                const src = iframe.src || iframe.getAttribute('data-src') || '';
+                                if (src && (src.includes('embed') || src.includes('player') || src.includes('stream'))) {
+                                    videoLinks.push({ url: src, type: 'iframe', source: 'iframe' });
+                                }
+                            });
+                            // 4. Video tags
+                            document.querySelectorAll('video source, video').forEach(el => {
+                                const src = el.getAttribute('src') || el.src;
+                                if (src && videoPatterns.test(src)) {
+                                    const ext = src.match(videoPatterns)?.[1] || 'mp4';
+                                    videoLinks.push({ url: src, type: ext, source: 'video-tag' });
+                                }
+                            });
+                            // 5. Hidden links in scripts (ajax pattern)
+                            document.querySelectorAll('script:not([src])').forEach(script => {
+                                const content = script.textContent || '';
+                                // m3u8/mp4 in script
+                                const matches = content.match(/https?:\/\/[^\s"'<>]+\.(m3u8|mp4|mkv)[^\s"'<>]*/gi);
+                                if (matches) {
+                                    matches.forEach(url => {
+                                        const ext = url.match(videoPatterns)?.[1] || 'mp4';
+                                        videoLinks.push({ url, type: ext, source: 'ajax-script' });
+                                    });
+                                }
+                            });
+                            // 6. Data attributes with video URLs
+                            document.querySelectorAll('[data-file], [data-source], [data-video], [data-stream]').forEach(el => {
+                                const url = el.getAttribute('data-file') || el.getAttribute('data-source') ||
+                                    el.getAttribute('data-video') || el.getAttribute('data-stream');
+                                if (url && (videoPatterns.test(url) || url.includes('m3u8'))) {
+                                    videoLinks.push({ url, type: url.includes('m3u8') ? 'm3u8' : 'mp4', source: 'data-attr' });
+                                }
+                            });
+                            // Deduplicate
+                            return [...new Map(videoLinks.map(v => [v.url, v])).values()];
+                        });
+                    }
                     // Follow links if enabled and depth allows
                     if (args.followLinks !== false && depth < maxDepth && results.length < maxPages) {
                         // Get all links

package/dist/tool-definitions.js

@@ -623,11 +623,11 @@ export const TOOLS = [
         },
     },
     // ============================================================
-    // WEB CRAWLER TOOL (Crawlee-inspired)
+    // WEB CRAWLER TOOL (Movie Streaming Optimized)
     // ============================================================
     {
         name: 'web_crawler',
-        description: 'Advanced web crawler with Crawlee-like features: URL queue (breadth/depth-first), proxy rotation, session management, auto-retry, rate limiting, concurrency control, and data extraction. Supports both browser and HTTP modes.',
+        description: 'Advanced web crawler optimized for movie downloading and streaming websites. Features: URL queue (breadth/depth-first), proxy rotation, auto-retry, rate limiting, JavaScript popup blocking, overlay ads blocking, and video link extraction. Uses brave-real-puppeteer-core with 50+ stealth features.',
         inputSchema: {
             type: 'object',
             additionalProperties: false,
@@ -635,7 +635,7 @@ export const TOOLS = [
                 startUrls: {
                     type: 'array',
                     items: { type: 'string' },
-                    description: 'Initial URLs to start crawling from'
+                    description: 'Initial URLs to start crawling from (movie/streaming pages)'
                 },
                 maxDepth: {
                     type: 'number',
@@ -680,9 +680,25 @@ export const TOOLS = [
                     description: 'Follow discovered links',
                     default: true
                 },
+                // Movie streaming specific options
+                blockPopups: {
+                    type: 'boolean',
+                    description: 'Block JavaScript popup ads and window.open calls',
+                    default: true
+                },
+                blockOverlayAds: {
+                    type: 'boolean',
+                    description: 'Block overlay ads, modal popups, and floating elements',
+                    default: true
+                },
+                extractVideoLinks: {
+                    type: 'boolean',
+                    description: 'Auto-extract m3u8, mp4, mkv video links from pages',
+                    default: true
+                },
                 downloadMedia: {
                     type: 'boolean',
-                    description: 'Download images/videos/files',
+                    description: 'Download video/audio files',
                     default: false
                 },
                 savePath: {
@@ -715,11 +731,6 @@ export const TOOLS = [
                     description: 'Crawl mode (browser = Puppeteer, http = fast HTTP)',
                     default: 'browser'
                 },
-                respectRobotsTxt: {
-                    type: 'boolean',
-                    description: 'Respect robots.txt rules',
-                    default: true
-                },
                 userAgent: {
                     type: 'string',
                     description: 'Custom User-Agent string'

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "brave-real-browser-mcp-server",
-  "version": "2.25.0",
+  "version": "2.26.0",
   "description": "🦁 MCP server for Brave Real Browser - NPM Workspaces Monorepo with anti-detection features, SSE streaming, and LSP compatibility",
   "type": "module",
   "main": "dist/index.js",
@@ -50,7 +50,7 @@
   "dependencies": {
     "@modelcontextprotocol/sdk": "latest",
     "@types/turndown": "latest",
-    "brave-real-browser": "^2.6.0",
+    "brave-real-browser": "^2.7.0",
     "crawlee": "^3.15.3",
     "puppeteer-core": "^24.35.0",
     "turndown": "latest",