Update app.py

app.py

@@ -108,7 +108,7 @@ class HTML_TO_MARKDOWN_CONVERTER:
             return f"\n\n![{alt}]({full_src})\n\n"
         return inner_md
 
-async def perform_web_browse(query: str, browser_name: str, search_engine_name: str):
+async def perform_web_browse(action: str, query: str, browser_name: str, search_engine_name: str):
     browser_key = browser_name.lower()
     if "playwright" not in PLAYWRIGHT_STATE:
         PLAYWRIGHT_STATE["playwright"] = await async_playwright().start()
@@ -126,11 +126,12 @@ async def perform_web_browse(query: str, browser_name: str, search_engine_name:
     
     browser_instance = PLAYWRIGHT_STATE[browser_key]
 
-    is_direct_url = urllib.parse.urlparse(query).scheme in ['http', 'https'] and '.' in urllib.parse.urlparse(query).netloc
-    
-    if is_direct_url:
-        url = query
-    else:
+    if action == "Scrape URL":
+        if not query.startswith(('http://', 'https://')):
+            url = f"http://{query}"
+        else:
+            url = query
+    else: # action == "Search"
         url_template = SEARCH_ENGINES.get(search_engine_name)
         if not url_template:
             return {"status": "error", "query": query, "error_message": f"Invalid search engine: '{search_engine_name}'."}
@@ -159,7 +160,7 @@ async def perform_web_browse(query: str, browser_name: str, search_engine_name:
         markdown_text = converter.convert()
         status_code = response.status if response else 0
 
-        return {"status": "success", "query": query, "final_url": final_url, "page_title": title, "http_status": status_code, "proxy_used": proxy_server_used, "markdown_content": markdown_text}
+        return {"status": "success", "query": query, "action": action, "final_url": final_url, "page_title": title, "http_status": status_code, "proxy_used": proxy_server_used, "markdown_content": markdown_text}
     except Exception as e:
         error_message = str(e).splitlines()[0]
         if "Timeout" in error_message:
@@ -173,16 +174,17 @@ with gr.Blocks(title="Web Browse API", theme=gr.themes.Soft()) as demo:
     gr.Markdown("# Web Browse API")
     gr.Markdown(f"This interface exposes a stateless API endpoint (`/api/web_browse`) to fetch and parse web content. {REVOLVER.count()} proxies loaded.")
     
-    query_input = gr.Textbox(label="URL or Search Query", placeholder="e.g., https://gradio.app or 'how does gradio work'")
+    action_input = gr.Radio(label="Action", choices=["Search", "Scrape URL"], value="Search")
+    query_input = gr.Textbox(label="Query or URL", placeholder="e.g., 'best cat food' or 'www.wikipedia.org'")
     
     with gr.Row():
         browser_input = gr.Dropdown(label="Browser", choices=["firefox", "chromium", "webkit"], value="firefox", scale=1)
-        search_engine_input = gr.Dropdown(label="Search Engine (for non-URL queries)", choices=sorted(list(SEARCH_ENGINES.keys())), value="DuckDuckGo", scale=2)
+        search_engine_input = gr.Dropdown(label="Search Engine (if action is Search)", choices=sorted(list(SEARCH_ENGINES.keys())), value="DuckDuckGo", scale=2)
     
     submit_button = gr.Button("Browse", variant="primary")
     output_json = gr.JSON(label="API Result")
     
-    submit_button.click(fn=perform_web_browse, inputs=[query_input, browser_input, search_engine_input], outputs=output_json, api_name="web_browse")
+    submit_button.click(fn=perform_web_browse, inputs=[action_input, query_input, browser_input, search_engine_input], outputs=output_json, api_name="web_browse")
 
 if __name__ == "__main__":
     demo.launch()