Spaces:
Running
Running
| from gradio import Interface, File | |
| import requests | |
| import io | |
| import base64 | |
| import json | |
| # Replace with your Google AI Studio API key | |
| API_KEY = "GOOGLE_API_KEY" | |
| project_id= "genai-test-396922" | |
| location= "us-central1" | |
| model_id= "gemini-2.0-flash-exp" | |
| # Endpoint for Gemini 2.0 Text Generation (adapt if needed for specific multimodal endpoints) | |
| ENDPOINT = "https://aiplatform.googleapis.com/v1/projects/{project_id}/locations/{location}/models/{model_id}:textGenerate" | |
| def encode_file(file_obj, file_type): | |
| """Encodes a file (image, video, or PDF) to base64.""" | |
| try: | |
| if file_type == "pdf": | |
| content = file_obj.read() | |
| else: | |
| content = file_obj.getvalue() # For images and videos | |
| encoded_content = base64.b64encode(content).decode("utf-8") | |
| return encoded_content | |
| except Exception as e: | |
| print(f"Error encoding file: {e}") | |
| return None | |
| def generate_response(text_prompt, files): | |
| """Handles multiple file inputs (images, videos, PDFs).""" | |
| url = ENDPOINT.format(project_id=project_id, location=location, model_id=model_id) | |
| headers = {"Authorization": f"Bearer {API_KEY}", "Content-Type": "application/json"} | |
| inputs = [{"text": text_prompt}] | |
| for file_data in files: | |
| if file_data is None: | |
| continue | |
| file_path = file_data.name | |
| file_type = file_path.split(".")[-1].lower() | |
| try: | |
| with open(file_path, "rb") as file_obj: | |
| encoded_content = encode_file(file_obj, file_type) | |
| if encoded_content: | |
| inputs.append({file_type: encoded_content}) # Use file type as key | |
| except Exception as e: | |
| return f"Error processing file {file_path}: {e}" | |
| request_body = {"inputs": inputs} | |
| try: | |
| response = requests.post(url, headers=headers, json=request_body) | |
| response.raise_for_status() # Raise an exception for bad status codes (4xx or 5xx) | |
| return response.json().get("generated_text", "No generated text found in response.") # Handle missing key | |
| except requests.exceptions.RequestException as e: | |
| if response is not None: | |
| try: | |
| error_message = response.json().get("error", {}).get("message", "Unknown error") | |
| except json.JSONDecodeError: | |
| error_message = response.text | |
| return f"API Error: {response.status_code} - {error_message}" | |
| else: | |
| return f"Request Exception: {e}" | |
| except Exception as e: | |
| return f"An unexpected error occurred: {e}" | |
| # Define Gradio interface with multiple file input | |
| interface = Interface( | |
| fn=generate_response, | |
| inputs=[ | |
| "text", | |
| [File(label="Upload Files (Images, Videos, PDFs)", file_types=[".png", ".jpg", ".jpeg", ".mp4", ".mov", ".pdf"], file_count="multiple")], # Multiple files | |
| ], | |
| outputs="text", | |
| title="Multimodal Prompt with Gemini 2.0 (Experimental)", | |
| description="Enter a text prompt and upload one or more files (images, videos, PDFs).", | |
| ) | |
| interface.launch(share=True) |