Update gemini_vision.py

gemini_vision.py

@@ -15,26 +15,6 @@ genai.configure()
 # Initialize the Gemini Vision Pro model
 model = genai.GenerativeModel('gemini-1.5-flash')
 
-
-import time
-import google.api_core.exceptions  # Import this to handle specific exceptions
-
-
-def retry_with_backoff(func, retries=3, initial_wait=2):
-    for i in range(retries):
-        try:
-            return func()  # Call the function that uses the API
-        except google.api_core.exceptions.ResourceExhausted as e:
-            if i < retries - 1:
-                wait_time = initial_wait * (2 ** i)  # Exponential backoff
-                print(f"API quota exhausted. Retrying in {wait_time} seconds...")
-                time.sleep(wait_time)
-            else:
-                raise e  # Raise error after last retry
-        except Exception as e:
-            print(f"An error occurred: {e}")
-            break  # Break on non-recoverable errors
-
 def process_frame_with_gemini(frame, mode="image"):
     if mode == "image":
         # Convert OpenCV frame to PIL Image
@@ -44,8 +24,21 @@ def process_frame_with_gemini(frame, mode="image"):
         image = Image.new('RGB', (1, 1), color='white')
         audio_path = frame  # In this case, 'frame' is actually the audio file path
 
-    # Define the prompt
+    # Process the image or audio with Gemini Vision Pro
     if mode == "image":
-        prompt = ["Describe the content of this frame", image]
+        response = model.generate_content(["Describe the content of this frame", image])
     else:
+        response = model.generate_content(["Transcribe the audio file at this path", audio_path, image])
+    
+    return response.text
+
+def summarize_with_gemini(text, max_words):
+    text_model = genai.GenerativeModel('gemini-pro')
+    response = text_model.generate_content(f"Summarize the following text in about {max_words} words: {text}")
+    return response.text
+
+def extract_code_with_gemini(text):
+    text_model = genai.GenerativeModel('gemini-pro')
+    response = text_model.generate_content(f"Extract and format any code snippets from the following text: {text}")
+    return response.text