File size: 15,632 Bytes
f201243 dd2695a f201243 b8b7791 f201243 2bfe32d f201243 dd2695a f201243 dd2695a f201243 dd2695a f201243 dd2695a f201243 8158a5c f201243 8158a5c f201243 dd2695a f201243 dd2695a f201243 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 |
"""
Image generation service supporting both Replicate and OpenAI APIs.
Supports multiple image generation models with automatic fallback.
"""
import os
import sys
import time
import random
import base64
from typing import Optional, Tuple, Dict, Any
# Add parent directory to path for imports
sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
import replicate
import httpx
from openai import AsyncOpenAI
import asyncio
from config import settings
# Model registry for PsyAdGenesis
MODEL_REGISTRY: Dict[str, Dict[str, Any]] = {
"nano-banana": {
"id": "google/nano-banana",
"param_name": "aspect_ratio",
"uses_dimensions": False,
},
"nano-banana-pro": {
"id": "google/nano-banana-pro",
"param_name": "aspect_ratio",
"uses_dimensions": False,
},
"imagen-4": {
"id": "google/imagen-4",
"param_name": "aspect_ratio",
"uses_dimensions": False,
},
"imagen-4-ultra": {
"id": "google/imagen-4-ultra",
"param_name": "aspect_ratio",
"uses_dimensions": False,
},
"z-image-turbo": {
"id": "prunaai/z-image-turbo",
"param_name": "height",
"uses_dimensions": True,
},
"seedream-3": {
"id": "bytedance/seedream-3",
"param_name": "aspect_ratio",
"uses_dimensions": False,
},
"recraft-v3": {
"id": "recraft-ai/recraft-v3",
"param_name": "aspect_ratio",
"uses_dimensions": False,
},
"photon": {
"id": "luma/photon",
"param_name": "aspect_ratio",
"uses_dimensions": False,
},
"ideogram-v3": {
"id": "ideogram-ai/ideogram-v3-quality",
"param_name": "aspect_ratio",
"uses_dimensions": False,
},
"seedream-4.5": {
"id": "bytedance/seedream-4.5",
"param_name": "aspect_ratio",
"uses_dimensions": False,
},
"flux-2-max": {
"id": "black-forest-labs/flux-2-max",
"param_name": "aspect_ratio",
"uses_dimensions": False,
},
"qwen-image": {
"id": "prunaai/z-image-turboqwen/qwen-image",
"param_name": "aspect_ratio",
"uses_dimensions": False,
},
"p-image": {
"id": "prunaai/p-image",
"param_name": "aspect_ratio",
"uses_dimensions": False,
},
"ideogram-v3-turbo": {
"id": "ideogram-ai/ideogram-v3-turbo",
"param_name": "aspect_ratio",
"uses_dimensions": False,
},
}
# Default model fallback chain (same as original project)
DEFAULT_FALLBACK_MODELS = ["nano-banana", "imagen-4", "z-image-turbo"]
RETRY_ATTEMPTS = 3
REQUEST_TIMEOUT = 60
def convert_dimensions_to_aspect_ratio(width: int, height: int) -> str:
"""Convert width/height to aspect ratio string."""
if width == height:
return "1:1"
elif width > height:
ratio = width / height
if abs(ratio - 16/9) < 0.1:
return "16:9"
elif abs(ratio - 4/3) < 0.1:
return "4:3"
elif abs(ratio - 3/2) < 0.1:
return "3:2"
else:
return "16:9"
else:
ratio = height / width
if abs(ratio - 16/9) < 0.1:
return "9:16"
elif abs(ratio - 4/3) < 0.1:
return "3:4"
elif abs(ratio - 3/2) < 0.1:
return "2:3"
else:
return "9:16"
class ImageService:
"""Image generation service supporting Replicate and OpenAI APIs."""
def __init__(self):
"""Initialize image generation clients."""
self.api_token = settings.replicate_api_token
if not self.api_token:
raise ValueError("REPLICATE_API_TOKEN not configured")
self.client = replicate.Client(api_token=self.api_token)
self.default_model = settings.image_model
# Initialize OpenAI client for gpt-image-1.5 support
self.openai_client = None
if hasattr(settings, 'openai_api_key') and settings.openai_api_key:
self.openai_client = AsyncOpenAI(api_key=settings.openai_api_key)
def _fetch_image(self, url: str) -> Optional[bytes]:
"""Fetch image from URL with retry logic."""
for attempt in range(RETRY_ATTEMPTS):
try:
response = httpx.get(
url,
timeout=REQUEST_TIMEOUT,
headers={
"Cache-Control": "no-cache",
"User-Agent": "AdGeneratorLite/1.0",
},
)
response.raise_for_status()
return response.content
except Exception as e:
if attempt == RETRY_ATTEMPTS - 1:
print(f"Failed to fetch image from {url}: {e}")
return None
time.sleep(1) # Sync fetch so sync sleep is fine here or use asyncio.sleep
return None
async def load_image(
self,
image_id: Optional[str] = None,
image_url: Optional[str] = None,
image_bytes: Optional[bytes] = None,
filepath: Optional[str] = None,
) -> Optional[bytes]:
"""
Load image from various sources (database ID, URL, bytes, or filepath).
Args:
image_id: Database ID of ad creative (will fetch from database)
image_url: Direct URL to image
image_bytes: Raw image bytes
filepath: Local file path
Returns:
Image bytes or None if failed
"""
# Priority: bytes > filepath > URL > database ID
if image_bytes:
return image_bytes
if filepath:
try:
with open(filepath, "rb") as f:
return f.read()
except Exception as e:
print(f"Failed to load image from filepath {filepath}: {e}")
return None
if image_url:
return self._fetch_image(image_url)
if image_id:
# Try to fetch from database
try:
from services.database import db_service
ad = await db_service.get_ad_creative(image_id)
if ad:
# Try image_url first
if ad.get("image_url"):
return self._fetch_image(ad["image_url"])
# Try local file
if ad.get("image_filename"):
filepath = os.path.join(settings.output_dir, ad["image_filename"])
if os.path.exists(filepath):
with open(filepath, "rb") as f:
return f.read()
except Exception as e:
print(f"Failed to load image from database ID {image_id}: {e}")
return None
return None
def _extract_image_from_output(self, output) -> Tuple[Optional[bytes], Optional[str]]:
"""
Extract image bytes and URL from Replicate output.
Returns:
Tuple of (image_bytes, image_url)
"""
try:
# Handle file-like object
if hasattr(output, 'read'):
url = getattr(output, 'url', None)
return output.read(), url
# Handle URL attribute
if hasattr(output, 'url'):
url = output.url
return self._fetch_image(url), url
# Handle list of outputs
if isinstance(output, list) and len(output) > 0:
first = output[0]
url = getattr(first, "url", str(first))
return self._fetch_image(url), url
# Handle string URL
if isinstance(output, str):
return self._fetch_image(output), output
print(f"Unknown output type: {type(output)}")
return None, None
except Exception as e:
print(f"Error extracting image from output: {e}")
return None, None
async def generate(
self,
prompt: str,
width: int = 1024,
height: int = 1024,
seed: Optional[int] = None,
model_key: Optional[str] = None,
image_url: Optional[str] = None,
) -> Tuple[bytes, str, Optional[str]]:
"""
Generate an image using Replicate API (official library).
Args:
prompt: Image generation prompt
width: Image width
height: Image height
seed: Random seed for uniqueness (if None, generates random)
model_key: Which model to use (default from config)
image_url: Optional image URL for image-to-image generation
Returns:
Tuple of (image_bytes, model_used, image_url)
"""
# Use random seed if not provided (ensures unique images)
if seed is None:
seed = random.randint(1, 2147483647)
# Get models to try (fallback chain)
model_key = model_key or self.default_model
# Check if using OpenAI image generation API (gpt-image-1.5)
if model_key == "gpt-image-1.5":
if not self.openai_client:
raise ValueError("OpenAI API key not configured for gpt-image-1.5")
try:
print("Generating image with gpt-image-1.5")
size_str = f"{width}x{height}"
# Use a timeout for OpenAI image generation
result = await asyncio.wait_for(
self.openai_client.images.generate(
model="gpt-image-1.5",
prompt=prompt,
quality="auto",
background="auto",
moderation="auto",
size=size_str,
output_format="jpeg",
output_compression=90,
),
timeout=120.0
)
if result.data and len(result.data) > 0:
image_base64 = result.data[0].b64_json
if image_base64:
image_bytes = base64.b64decode(image_base64)
print("Successfully generated image with gpt-image-1.5")
return image_bytes, "gpt-image-1.5", None
raise Exception("No image data returned from OpenAI API")
except asyncio.TimeoutError:
print("Timed out generating image with gpt-image-1.5")
raise Exception("Timeout: Image generation with gpt-image-1.5 took too long")
except Exception as e:
print(f"OpenAI image generation failed: {e}")
print("Falling back to Replicate models...")
model_key = None # Will use default fallback chain
# Build fallback chain for Replicate models
if model_key and model_key in MODEL_REGISTRY:
models_to_try = [model_key] + [m for m in DEFAULT_FALLBACK_MODELS if m != model_key]
else:
models_to_try = DEFAULT_FALLBACK_MODELS
last_error = None
for current_model in models_to_try:
cfg = MODEL_REGISTRY.get(current_model)
if not cfg:
continue
# Build input parameters
input_data = {"prompt": prompt}
# Add image URL for image-to-image if provided (for nano-banana and nano-banana-pro)
# Google Nano Banana models expect image_input as an array
if image_url and current_model in ["nano-banana", "nano-banana-pro"]:
input_data["image_input"] = [image_url]
# Note: guidance_scale may not be supported by nano-banana on Replicate
# Relying on minimal prompts to preserve the original image
# Add seed if supported
input_data["seed"] = seed
# Some models use width/height, others use aspect_ratio
if cfg.get("uses_dimensions"):
input_data["width"] = width
input_data["height"] = height
else:
aspect_ratio = convert_dimensions_to_aspect_ratio(width, height)
input_data[cfg["param_name"]] = aspect_ratio
# Retry logic
for attempt in range(RETRY_ATTEMPTS):
try:
print(f"Generating image with {current_model} (attempt {attempt + 1})")
# Use official Replicate client - offload blocking call to thread and add timeout
try:
# Wrap the blocking call in a thread and add a 3-minute timeout
output = await asyncio.wait_for(
asyncio.to_thread(self.client.run, cfg["id"], input=input_data),
timeout=180.0 # 3 minutes timeout
)
except asyncio.TimeoutError:
print(f"Timed out generating image with {current_model} after 180 seconds")
raise Exception(f"Timeout: Image generation with {current_model} took too long")
# Extract image bytes and URL
image_bytes, image_url = self._extract_image_from_output(output)
if image_bytes:
print(f"Successfully generated image with {current_model}")
return image_bytes, current_model, image_url
except Exception as e:
last_error = e
if attempt < RETRY_ATTEMPTS - 1:
print(f"Attempt {attempt + 1} failed: {e}, retrying...")
await asyncio.sleep(2 ** attempt) # Exponential backoff (use await for async sleep)
continue
# Model failed, try next in fallback chain
print(f"Model {current_model} failed, trying next...")
# All models failed
raise Exception(f"All image generation models failed. Last error: {last_error}")
async def generate_with_retry(
self,
prompt: str,
width: int = 1024,
height: int = 1024,
max_retries: int = 2,
) -> Tuple[bytes, str, Optional[str]]:
"""
Generate image with automatic retry on failure.
Uses different random seed each attempt for variety.
Returns:
Tuple of (image_bytes, model_used, image_url)
"""
last_error = None
for attempt in range(max_retries + 1):
try:
seed = random.randint(1, 2147483647)
return await self.generate(
prompt=prompt,
width=width,
height=height,
seed=seed,
)
except Exception as e:
last_error = e
if attempt < max_retries:
await asyncio.sleep(2)
continue
raise last_error
# Global instance
image_service = ImageService()
|