SKILLS/image-generation-minimax/mcp_server/minimax_image_mcp.py

199 lines
6.7 KiB
Python

#!/usr/bin/env python3
"""
MiniMax Image Generation MCP Server
Provides image generation via MiniMax API
"""
import os
import base64
import requests
from urllib.parse import urlparse
from mcp.server.fastmcp import FastMCP
# Initialize MCP server
mcp = FastMCP("minimax-image-generator")
def read_local_image(file_path: str) -> str | None:
"""
Read local image file and return as base64 encoded string.
Returns the base64 string or None if failed.
"""
try:
with open(file_path, "rb") as f:
image_data = f.read()
return base64.b64encode(image_data).decode("utf-8")
except Exception as e:
return None
def build_subject_reference(subject_ref: str, subject_type: str) -> dict | None:
"""
Build subject_reference object from URL or local file path.
Args:
subject_ref: URL or local file path
subject_type: Type of subject (character, product, logo, etc.)
Returns:
dict with subject_reference structure or None if failed
"""
# Check if it's a URL or local file
if subject_ref.startswith(("http://", "https://")):
# It's a URL
return {
"type": subject_type,
"image_file": subject_ref
}
else:
# It's a local file - convert to base64
base64_data = read_local_image(subject_ref)
if base64_data:
return {
"type": subject_type,
"image_file": f"data:image/jpeg;base64,{base64_data}"
}
return None
@mcp.tool()
def generate_image(
prompt: str,
model: str = "image-01",
aspect_ratio: str = "1:1",
n: int = 1,
prompt_optimizer: bool = False,
subject_reference: str | None = None,
subject_type: str = "character",
seed: int | None = None,
output_path: str | None = None
) -> str:
"""
Generate images using MiniMax image generation API.
Args:
prompt: The image generation prompt describing what to generate
model: Image generation model (default: image-01)
aspect_ratio: Image aspect ratio (default: 1:1, options: 1:1, 16:9, 9:16, 4:3, 3:4)
n: Number of images to generate (default: 1, max: 3)
prompt_optimizer: Enable prompt optimizer (default: false)
subject_reference: Reference image URL or local file path for image-to-image generation
subject_type: Subject reference type (default: character, options: character, product, logo, video_subject, other)
seed: Random seed for reproducible generation
output_path: Optional local path to save the generated image
Returns:
Image URLs and generation status
"""
api_key = os.environ.get("MINIMAX_API_KEY", "")
api_base = "https://api.minimaxi.com"
if not api_key:
return "Error: MINIMAX_API_KEY environment variable is not set"
# Build request payload
payload = {
"model": model,
"prompt": prompt,
"aspect_ratio": aspect_ratio,
"response_format": "url",
"n": n,
"prompt_optimizer": prompt_optimizer
}
# Add seed if provided
if seed is not None:
payload["seed"] = seed
# Add subject_reference for image-to-image generation
if subject_reference:
subject_ref = build_subject_reference(subject_reference, subject_type)
if subject_ref:
payload["subject_reference"] = [subject_ref]
# Make API request
url = f"{api_base}/v1/image_generation"
headers = {
"Authorization": f"Bearer {api_key}",
"Content-Type": "application/json"
}
try:
response = requests.post(url, headers=headers, json=payload, timeout=60)
response.raise_for_status()
result = response.json()
# Check for API errors
if result.get("base_resp", {}).get("status_code") != 0:
error_msg = result.get("base_resp", {}).get("status_msg", "Unknown error")
return f"API Error: {error_msg}"
# Extract image URLs
image_urls = result.get("data", {}).get("image_urls", [])
metadata = result.get("metadata", {})
request_id = result.get("id", "N/A")
success_count = metadata.get("success_count", len(image_urls))
# Save image if output_path provided
saved_paths = []
if output_path and image_urls:
try:
import time
timestamp = int(time.time())
for i, img_url in enumerate(image_urls, 1):
img_response = requests.get(img_url, timeout=30)
img_response.raise_for_status()
# Determine file extension
parsed = urlparse(img_url)
ext = os.path.splitext(parsed.path)[1] if "." in parsed.path else ".jpeg"
if not ext or len(ext) > 5:
ext = ".jpeg"
# Handle multiple images
if len(image_urls) > 1:
base_path = output_path.rsplit('.', 1)[0] if '.' in output_path else output_path
file_ext = output_path.rsplit('.', 1)[1] if '.' in output_path else ext
img_path = f"{base_path}_{i}_{timestamp}.{file_ext}"
else:
if not output_path.endswith(ext):
img_path = output_path + ext
else:
img_path = output_path
# Ensure directory exists
os.makedirs(os.path.dirname(img_path) if os.path.dirname(img_path) else ".", exist_ok=True)
with open(img_path, "wb") as f:
f.write(img_response.content)
saved_paths.append(os.path.abspath(img_path))
except Exception as e:
return f"Failed to save image: {str(e)}"
# Build response
response_text = f"Successfully generated {success_count} image(s)\n\n"
response_text += f"Request ID: {request_id}\n\n"
if saved_paths:
response_text += "Saved to:\n"
for path in saved_paths:
response_text += f" - {path}\n"
response_text += "\n"
response_text += "Image URLs:\n"
for i, img_url in enumerate(image_urls, 1):
response_text += f" {i}. {img_url}\n"
return response_text
except requests.exceptions.RequestException as e:
return f"Request Error: {str(e)}"
except Exception as e:
return f"Unexpected Error: {str(e)}"
if __name__ == "__main__":
mcp.run()