Update app.py

app.py

@@ -4,6 +4,7 @@ import urllib.parse
 import asyncio
 from typing import Dict, Optional
 from itertools import cycle
+import json 
 
 # Install playwright if not present
 if os.getenv("PLAYWRIGHT_INSTALL_RUN", "false").lower() != "true":
@@ -42,18 +43,13 @@ class CredentialRevolver:
     def count(self) -> int:
         return len(self.proxies)
 
-# <<< REMOVED >>> The global PLAYWRIGHT_STATE is removed to prevent event loop conflicts.
-# PLAYWRIGHT_STATE: Dict = {}
 REVOLVER = CredentialRevolver(os.getenv("PROXY_LIST", ""))
 
 SEARCH_ENGINES = {
-    # ... (this dictionary is unchanged)
     "Google": "https://www.google.com/search?q={query}&hl=en", "DuckDuckGo": "https://duckduckgo.com/html/?q={query}", "Bing": "https://www.bing.com/search?q={query}", "Brave": "https://search.brave.com/search?q={query}", "Ecosia": "https://www.ecosia.org/search?q={query}", "Yahoo": "https://search.yahoo.com/search?p={query}", "Startpage": "https://www.startpage.com/sp/search?q={query}", "Qwant": "https://www.qwant.com/?q={query}", "Swisscows": "https://swisscows.com/web?query={query}", "You.com": "https://you.com/search?q={query}", "SearXNG": "https://searx.be/search?q={query}", "MetaGer": "https://metager.org/meta/meta.ger-en?eingabe={query}", "Yandex": "https://yandex.com/search/?text={query}", "Baidu": "https://www.baidu.com/s?wd={query}", "Perplexity": "https://www.perplexity.ai/search?q={query}",
 }
 
-# --- HTML to Markdown Conversion (unchanged) ---
 class HTML_TO_MARKDOWN_CONVERTER:
-    # ... (this class is unchanged)
     def __init__(self, soup: BeautifulSoup, base_url: str): self.soup = soup; self.base_url = base_url
     def _cleanup_html(self): selectors_to_remove = ['nav', 'footer', 'header', 'aside', 'form', 'script', 'style', 'svg', 'button', 'input', 'textarea', '[role="navigation"]', '[role="search"]', '[id*="comment"]', '[class*="comment-"]', '[id*="sidebar"]', '[class*="sidebar"]', '[id*="related"]', '[class*="related"]', '[id*="share"]', '[class*="share"]', '[id*="social"]', '[class*="social"]', '[id*="cookie"]', '[class*="cookie"]', '[aria-hidden="true"]']; [element.decompose() for selector in selectors_to_remove for element in self.soup.select(selector)]
     def convert(self): self._cleanup_html(); content_node = self.soup.find('main') or self.soup.find('article') or self.soup.find('body'); return re.sub(r'\n{3,}', '\n\n', self._process_node(content_node)).strip() if content_node else ""
@@ -83,9 +79,7 @@ class HTML_TO_MARKDOWN_CONVERTER:
 async def perform_web_browse(action: str, query: str, browser_name: str, search_engine: str):
     playwright = None
     browser = None
-    # <<< CHANGE >>> Use a try/finally block to ensure resources are always cleaned up.
     try:
-        # <<< CHANGE >>> Start Playwright inside the function.
         playwright = await async_playwright().start()
         
         browser_key = browser_name.lower()
@@ -136,12 +130,10 @@ async def perform_web_browse(action: str, query: str, browser_name: str, search_
             if 'page' in locals() and not page.is_closed(): await page.close()
             if 'context' in locals(): await context.close()
 
-    # <<< CHANGE >>> This block now handles errors during browser launch as well.
     except Exception as e:
         app.logger.error(f"A critical error occurred in perform_web_browse: {e}", exc_info=True)
         return {"status": "error", "query": query, "error_message": f"Failed to initialize browser resources: {str(e).splitlines()[0]}"}
     
-    # <<< CHANGE >>> Ensure browser and playwright are always closed down.
     finally:
         if browser:
             await browser.close()
@@ -149,15 +141,12 @@ async def perform_web_browse(action: str, query: str, browser_name: str, search_
             await playwright.stop()
 
 
-# --- API Endpoint Definitions (unchanged) ---
 @app.route('/', methods=['GET'])
 def index():
-    # ... (this function is unchanged)
-    return jsonify({ "status": "online", "message": "Welcome to the Web Browse API!", "api_endpoint": "/web_browse", "instructions": "Send a POST request to /web_browse with a JSON payload to use the service.", "payload_format": { "action": "string (required: 'Search' or 'Scrape URL')", "query": "string (required: a search term or a full URL)", "browser": "string (optional, default: 'firefox'; options: 'firefox', 'chromium', 'webkit')", "search_engine": "string (optional, default: 'DuckDuckGo'; see code for all options)" }, "example_curl": """curl -X POST YOUR_SPACE_URL/web_browse -H "Content-Type: application/json" -d '{"action": "Search", "query": "latest news on AI", "browser": "webkit"}'""" })
+    return json.dumps({ "status": "online", "message": "Welcome to the Web Browse API!", "api_endpoint": "/web_browse", "instructions": "Send a POST request to /web_browse with a JSON payload to use the service.", "payload_format": { "action": "string (required: 'Search' or 'Scrape URL')", "query": "string (required: a search term or a full URL)", "browser": "string (optional, default: 'firefox'; options: 'firefox', 'chromium', 'webkit')", "search_engine": "string (optional, default: 'DuckDuckGo'; see code for all options)" }, "example_curl": """curl -X POST YOUR_SPACE_URL/web_browse -H "Content-Type: application/json" -d '{"action": "Search", "query": "latest news on AI", "browser": "webkit"}'""" })
 
 @app.route('/web_browse', methods=['POST'])
 def web_browse():
-    # ... (this function is unchanged)
     if not request.is_json: return jsonify({"status": "error", "error_message": "Invalid input: payload must be JSON"}), 400
     data = request.get_json()
     action = data.get('action')
@@ -174,7 +163,6 @@ def web_browse():
         app.logger.error(f"An unexpected server error occurred: {e}", exc_info=True)
         return jsonify({"status": "error", "query": query, "error_message": f"An unexpected server error occurred: {str(e)}"}), 500
 
-# --- Main Application Runner (unchanged) ---
 if __name__ == "__main__":
     port = int(os.environ.get("PORT", 7860))
     print(f"Flask server starting on port {port}... {REVOLVER.count()} proxies loaded.")