Update app.py

app.py

@@ -199,6 +199,46 @@ async def fetch_codepen_project(codepen_url):
     except Exception as e:
         return f"Error: {e}", "", ""
 
+# Web Data Extractor
+async def extract_web_data(url):
+    """Extracts additional web data like description, image preview, colors, fonts, similar code, videos, and files."""
+    try:
+        response = await asyncio.to_thread(requests.get, url, timeout=5)
+        response.raise_for_status()
+        soup = BeautifulSoup(response.text, "html.parser")
+
+        # Extract site description
+        description = soup.find("meta", attrs={"name": "description"})["content"] if soup.find("meta", attrs={"name": "description"}) else "No description available."
+
+        # Extract image preview (first image on the page)
+        image_preview = soup.find("img")["src"] if soup.find("img") else "No image preview available."
+
+        # Extract colors (from CSS or inline styles)
+        colors = []
+        for style in soup.find_all("style"):
+            colors.extend([color for color in style.text.split() if color.startswith("#")])
+        colors = list(set(colors))[:5]  # Limit to 5 unique colors
+
+        # Extract fonts (from CSS or Google Fonts)
+        fonts = []
+        for link in soup.find_all("link", attrs={"href": True}):
+            if "fonts.googleapis.com" in link["href"]:
+                fonts.append(link["href"])
+        fonts = list(set(fonts))[:5]  # Limit to 5 unique fonts
+
+        # Extract similar code (shorter version of the HTML)
+        similar_code = str(soup)[:1000]  # Limit to first 1000 characters
+
+        # Extract videos (embedded iframes)
+        videos = [iframe["src"] for iframe in soup.find_all("iframe") if "src" in iframe.attrs]
+
+        # Extract files (links to downloadable files)
+        files = [a["href"] for a in soup.find_all("a", attrs={"href": True}) if a["href"].endswith((".pdf", ".zip", ".doc", ".docx", ".xls", ".xlsx"))]
+
+        return description, image_preview, colors, fonts, similar_code, videos, files
+    except Exception as e:
+        return f"Error: {e}", "", [], [], "", [], []
+
 # Create the Gradio interface
 with gr.Blocks() as demo:
     gr.HTML(copy_button_html)  # Add the "Copy Code" script
@@ -326,5 +366,38 @@ with gr.Blocks() as demo:
                 outputs=[html_output, css_output, js_output]
             )
 
+        # Tab 5: Web Data Extractor
+        with gr.Tab("Web Data Extractor"):
+            gr.Markdown("## Web Data Extractor")
+            gr.Markdown("Enter a URL to extract additional web data like description, image preview, colors, fonts, similar code, videos, and files.")
+            
+            with gr.Row():
+                web_data_url_input = gr.Textbox(label="Enter URL", placeholder="https://example.com")
+            
+            with gr.Row():
+                description_output = gr.Textbox(label="Site Description", interactive=False)
+                image_preview_output = gr.Image(label="Image Preview", interactive=False)
+            
+            with gr.Row():
+                colors_output = gr.Textbox(label="Colors", interactive=False)
+                fonts_output = gr.Textbox(label="Fonts", interactive=False)
+            
+            with gr.Row():
+                similar_code_output = gr.Textbox(label="Similar Code", interactive=True, elem_id="similar-code-output")
+            
+            with gr.Row():
+                videos_output = gr.Textbox(label="Videos", interactive=False)
+                files_output = gr.Textbox(label="Files", interactive=False)
+            
+            with gr.Row():
+                gr.HTML("<button onclick='copyCode(\"similar-code-output\")'>Copy Code</button>")  # Add the "Copy Code" button
+            
+            submit_web_data_button = gr.Button("Extract Web Data")
+            submit_web_data_button.click(
+                fn=extract_web_data,
+                inputs=[web_data_url_input],
+                outputs=[description_output, image_preview_output, colors_output, fonts_output, similar_code_output, videos_output, files_output]
+            )
+
 # Launch the interface
 demo.launch()