Update app.py

app.py

@@ -141,7 +141,11 @@ def process_urls(url_input, bulk_toggle, action_radio, max_urls, crawl_depth):
                     screenshot_image = take_screenshot(url)  # Capture screenshot
                     if screenshot_image is not None:
                         logging.info(f"Successfully captured screenshot of: {url}")
-                        embeddings[url] = screenshot_image
+                        # Preprocess the image and get CLIP embedding
+                        inputs = processor(images=screenshot_image, return_tensors="pt")
+                        with clip_model_lock:
+                            embedding = model.get_image_embeddings(inputs['pixel_values'])
+                        embeddings[url] = embedding.squeeze()  # Store the embedding tensor
                 except Exception as e:
                     logging.error(f"Error capturing screenshot for {url}: {str(e)}")
         
@@ -154,6 +158,17 @@ def process_urls(url_input, bulk_toggle, action_radio, max_urls, crawl_depth):
         logging.error(f"Unexpected error: {str(e)}")
         return json.dumps([], indent=2), None
 
+# Function to save the CLIP embedding to a JSON file
+def save_embedding(embedding, url):
+    date_dir = datetime.datetime.now().strftime("%Y%m%d")
+    embedding_dir = os.path.join('embeddings', date_dir)
+    if not os.path.exists(embedding_dir):
+        os.makedirs(embedding_dir)
+    filename = os.path.join(embedding_dir, f'embedding_{datetime.datetime.now().strftime("%Y%m%d%H%M%S")}_{uuid.uuid4().hex}.json')
+    with open(filename, 'w') as f:
+        json.dump({url: embedding.tolist()}, f)
+    return filename
+
 # Global variables for monitoring
 stop_event = threading.Event()
 monitor_thread = None