Update app.py

app.py

@@ -9,12 +9,16 @@ def save_binary_file(file_name, data):
     with open(file_name, "wb") as f:
         f.write(data)
 
-def generate_edit(prompt, image, api_key, model="gemini-2.0-flash-exp"):
+def generate_edit(prompt, pil_image, api_key, model="gemini-2.0-flash-exp"):
     # Initialize Gemini client with provided key or env fallback
     client = genai.Client(api_key=(api_key.strip() if api_key and api_key.strip() != "" else os.environ.get("GEMINI_API_KEY")))
 
-    # Upload image to Gemini and prepare content payload
-    files = [client.files.upload(file=image)]
+    # Save PIL Image to a temporary file path for upload
+    with tempfile.NamedTemporaryFile(suffix=".png", delete=False) as tmp_file:
+        image_path = tmp_file.name
+        pil_image.save(image_path)
+
+    files = [client.files.upload(file=image_path)]
     contents = [
         types.Content(
             role="user",
@@ -25,23 +29,20 @@ def generate_edit(prompt, image, api_key, model="gemini-2.0-flash-exp"):
         ),
     ]
 
-    # Configure generation with text and image response modalities
     generate_content_config = types.GenerateContentConfig(
         temperature=1,
         top_p=0.95,
         top_k=40,
         max_output_tokens=8192,
         response_modalities=["image", "text"],
-        response_mime_type="text/plain",  # Important: keep mime type text/plain to avoid errors
+        response_mime_type="text/plain",  # Important for streaming image inline data
     )
 
     text_response = ""
-    image_path = None
+    image_path_result = None
 
-    # Temporary file to write streamed image data
     with tempfile.NamedTemporaryFile(suffix=".png", delete=False) as tmp:
         temp_path = tmp.name
-        # Stream generation content, check each chunk for inline image data or text
         for chunk in client.models.generate_content_stream(
             model=model,
             contents=contents,
@@ -52,12 +53,13 @@ def generate_edit(prompt, image, api_key, model="gemini-2.0-flash-exp"):
             candidate = chunk.candidates[0].content.parts[0]
             if candidate.inline_data:
                 save_binary_file(temp_path, candidate.inline_data.data)
-                image_path = temp_path
-                break  # Stop streaming after getting image data
+                image_path_result = temp_path
+                break  # Stop on first image data chunk
             else:
                 text_response += chunk.text + "\n"
     del files
-    return image_path, text_response
+
+    return image_path_result, text_response
 
 def process_image_and_prompt(pil_image, prompt, api_key):
     try: