350 lines
12 KiB
Python
350 lines
12 KiB
Python
"""
|
|
S4 Custom Grok Text Node
|
|
Author: S4MUEL
|
|
GitHub: https://github.com/S4MUEL-404/ComfyUI-S4API
|
|
|
|
Grok text generation node with custom API key input.
|
|
"""
|
|
|
|
import requests
|
|
import json
|
|
import os
|
|
import hashlib
|
|
import base64
|
|
import io
|
|
from PIL import Image
|
|
from typing import Optional
|
|
import torch
|
|
|
|
|
|
class S4TextWithGrok:
|
|
"""
|
|
Generates text via Grok API endpoint.
|
|
"""
|
|
|
|
# Fixed API configuration
|
|
API_URL = "https://api.x.ai/v1/chat/completions"
|
|
|
|
def __init__(self):
|
|
# Ensure cache directory exists
|
|
self.cache_dir = os.path.join(os.path.dirname(os.path.dirname(__file__)), "cache")
|
|
os.makedirs(self.cache_dir, exist_ok=True)
|
|
|
|
@classmethod
|
|
def INPUT_TYPES(cls):
|
|
return {
|
|
"required": {
|
|
"system_prompt": (
|
|
"STRING",
|
|
{
|
|
"multiline": True,
|
|
"default": "",
|
|
"tooltip": "System instruction for the AI",
|
|
},
|
|
),
|
|
"user_prompt": (
|
|
"STRING",
|
|
{
|
|
"multiline": True,
|
|
"default": "",
|
|
"tooltip": "User prompt for Grok",
|
|
},
|
|
),
|
|
"api_key": (
|
|
"STRING",
|
|
{
|
|
"multiline": False,
|
|
"default": "",
|
|
"tooltip": "Your Grok API Key"
|
|
}
|
|
),
|
|
"model": (
|
|
["grok-4-latest", "grok-4-base", "grok-2-latest", "grok-2-base", "grok-2-1212"],
|
|
{
|
|
"default": "grok-4-latest",
|
|
"tooltip": "Choose Grok model",
|
|
},
|
|
),
|
|
"max_tokens": (
|
|
"INT",
|
|
{
|
|
"default": 2048,
|
|
"min": 1,
|
|
"max": 32768,
|
|
"step": 1,
|
|
"display": "number",
|
|
"tooltip": "Maximum number of tokens to generate",
|
|
},
|
|
),
|
|
"temperature": (
|
|
"FLOAT",
|
|
{
|
|
"default": 0.7,
|
|
"min": 0.0,
|
|
"max": 2.0,
|
|
"step": 0.1,
|
|
"display": "number",
|
|
"tooltip": "Sampling temperature (0.0 = deterministic, 2.0 = very random)",
|
|
},
|
|
),
|
|
"top_p": (
|
|
"FLOAT",
|
|
{
|
|
"default": 0.95,
|
|
"min": 0.0,
|
|
"max": 1.0,
|
|
"step": 0.05,
|
|
"display": "number",
|
|
"tooltip": "Top-p sampling threshold",
|
|
},
|
|
),
|
|
"seed": (
|
|
"INT",
|
|
{
|
|
"default": 0,
|
|
"min": 0,
|
|
"max": 2147483647,
|
|
"step": 1,
|
|
"display": "number",
|
|
"control_after_generate": True,
|
|
"tooltip": "Random seed for reproducible results (32-bit limit for Grok API)",
|
|
},
|
|
),
|
|
},
|
|
"optional": {
|
|
"images": ("IMAGE", {"tooltip": "Optional reference images"}),
|
|
},
|
|
}
|
|
|
|
RETURN_TYPES = ("STRING",)
|
|
RETURN_NAMES = ("text",)
|
|
FUNCTION = "generate_text"
|
|
CATEGORY = "💀PromptsO"
|
|
DESCRIPTION = "Generates text using Grok API with custom API key."
|
|
|
|
def encode_image_to_base64(self, image_tensor):
|
|
"""Convert image tensor to base64 string"""
|
|
try:
|
|
# Convert tensor to PIL Image
|
|
if len(image_tensor.shape) == 4:
|
|
# Batch of images, take first one
|
|
img_tensor = image_tensor[0]
|
|
else:
|
|
img_tensor = image_tensor
|
|
|
|
# Convert from torch tensor to PIL Image
|
|
img_array = (img_tensor.cpu().numpy() * 255).astype('uint8')
|
|
pil_image = Image.fromarray(img_array)
|
|
|
|
# Convert to base64
|
|
buffer = io.BytesIO()
|
|
pil_image.save(buffer, format="JPEG", quality=95)
|
|
img_base64 = base64.b64encode(buffer.getvalue()).decode('utf-8')
|
|
|
|
return img_base64
|
|
except Exception as e:
|
|
print(f"⚠️ Error encoding image: {e}")
|
|
return None
|
|
|
|
def validate_inputs(self, user_prompt: str, api_key: str) -> None:
|
|
"""Validate input parameters"""
|
|
if not user_prompt or not user_prompt.strip():
|
|
raise ValueError("User prompt cannot be empty")
|
|
if not api_key or not api_key.strip():
|
|
raise ValueError("API key cannot be empty")
|
|
|
|
def generate_cache_key(self, system_prompt: str, user_prompt: str, model: str,
|
|
max_tokens: int, temperature: float, top_p: float, seed: int, images=None) -> str:
|
|
"""Generate cache key based on input parameters"""
|
|
cache_data = {
|
|
"system_prompt": system_prompt.strip(),
|
|
"user_prompt": user_prompt.strip(),
|
|
"model": model,
|
|
"max_tokens": max_tokens,
|
|
"temperature": temperature,
|
|
"top_p": top_p,
|
|
"seed": seed
|
|
}
|
|
|
|
# Include image hash if images are provided
|
|
if images is not None:
|
|
# Create a simple hash of the image tensor
|
|
image_hash = hashlib.md5(str(images.shape).encode() + str(images.sum().item()).encode()).hexdigest()[:8]
|
|
cache_data["images_hash"] = image_hash
|
|
|
|
cache_str = json.dumps(cache_data, sort_keys=True)
|
|
return hashlib.md5(cache_str.encode()).hexdigest()
|
|
|
|
def load_from_cache(self, cache_key: str) -> Optional[str]:
|
|
"""Load result from cache if exists"""
|
|
cache_file = os.path.join(self.cache_dir, f"{cache_key}.json")
|
|
if os.path.exists(cache_file):
|
|
try:
|
|
with open(cache_file, 'r', encoding='utf-8') as f:
|
|
cache_data = json.load(f)
|
|
return cache_data.get("result")
|
|
except Exception as e:
|
|
print(f"⚠️ Failed to load cache: {e}")
|
|
return None
|
|
|
|
def save_to_cache(self, cache_key: str, result: str) -> None:
|
|
"""Save result to cache"""
|
|
cache_file = os.path.join(self.cache_dir, f"{cache_key}.json")
|
|
try:
|
|
cache_data = {
|
|
"result": result,
|
|
"timestamp": json.dumps({"time": "cached"})
|
|
}
|
|
with open(cache_file, 'w', encoding='utf-8') as f:
|
|
json.dump(cache_data, f, ensure_ascii=False, indent=2)
|
|
except Exception as e:
|
|
print(f"⚠️ Failed to save cache: {e}")
|
|
|
|
def generate_text(
|
|
self,
|
|
system_prompt,
|
|
user_prompt,
|
|
api_key,
|
|
model="grok-4-latest",
|
|
max_tokens=2048,
|
|
temperature=0.7,
|
|
top_p=0.95,
|
|
seed=0,
|
|
images=None,
|
|
):
|
|
self.validate_inputs(user_prompt, api_key)
|
|
|
|
# Generate cache key
|
|
cache_key = self.generate_cache_key(
|
|
system_prompt, user_prompt, model, max_tokens, temperature, top_p, seed, images
|
|
)
|
|
|
|
# Check cache first
|
|
cached_result = self.load_from_cache(cache_key)
|
|
if cached_result:
|
|
print(f"💾 Using cached result for seed {seed}")
|
|
return (cached_result,)
|
|
|
|
# Prepare messages
|
|
messages = []
|
|
|
|
# Add system message if provided
|
|
if system_prompt and system_prompt.strip():
|
|
messages.append({
|
|
"role": "system",
|
|
"content": system_prompt.strip()
|
|
})
|
|
|
|
# Build user message content
|
|
user_content = []
|
|
|
|
# Add text content
|
|
user_content.append({
|
|
"type": "text",
|
|
"text": user_prompt
|
|
})
|
|
|
|
# Add image content if provided
|
|
if images is not None:
|
|
if len(images.shape) == 4:
|
|
# Multiple images
|
|
for i in range(images.shape[0]):
|
|
img_base64 = self.encode_image_to_base64(images[i])
|
|
if img_base64:
|
|
user_content.append({
|
|
"type": "image_url",
|
|
"image_url": {
|
|
"url": f"data:image/jpeg;base64,{img_base64}"
|
|
}
|
|
})
|
|
else:
|
|
# Single image
|
|
img_base64 = self.encode_image_to_base64(images)
|
|
if img_base64:
|
|
user_content.append({
|
|
"type": "image_url",
|
|
"image_url": {
|
|
"url": f"data:image/jpeg;base64,{img_base64}"
|
|
}
|
|
})
|
|
|
|
# Add user message
|
|
messages.append({
|
|
"role": "user",
|
|
"content": user_content
|
|
})
|
|
|
|
# Prepare request data
|
|
data = {
|
|
"model": model,
|
|
"messages": messages,
|
|
"max_tokens": max_tokens,
|
|
"temperature": temperature,
|
|
"top_p": top_p,
|
|
"stream": False
|
|
}
|
|
|
|
# Add seed for reproducible results
|
|
if seed > 0:
|
|
data["seed"] = seed
|
|
|
|
# Make API call
|
|
headers = {
|
|
"Authorization": f"Bearer {api_key}",
|
|
"Content-Type": "application/json"
|
|
}
|
|
|
|
print(f"🤖 Generating text with Grok...")
|
|
print(f" • Model: {model}")
|
|
print(f" • User prompt: {user_prompt[:50]}{'...' if len(user_prompt) > 50 else ''}")
|
|
if system_prompt.strip():
|
|
print(f" • System prompt: {system_prompt[:30]}{'...' if len(system_prompt) > 30 else ''}")
|
|
if images is not None:
|
|
print(f" • Reference images: {images.shape[0] if len(images.shape) == 4 else 1}")
|
|
print(f" • Max tokens: {max_tokens}")
|
|
print(f" • Temperature: {temperature}")
|
|
print(f" • Top-p: {top_p}")
|
|
print(f" • Seed: {seed}")
|
|
print(f" • Cache key: {cache_key[:8]}...")
|
|
|
|
try:
|
|
response = requests.post(
|
|
self.API_URL,
|
|
headers=headers,
|
|
json=data,
|
|
timeout=120
|
|
)
|
|
|
|
if response.status_code != 200:
|
|
error_data = response.json() if response.content else {}
|
|
error_msg = error_data.get("error", {}).get("message", response.text)
|
|
raise Exception(f"API request failed with status {response.status_code}: {error_msg}")
|
|
|
|
# Parse response
|
|
result = response.json()
|
|
|
|
if "choices" not in result or not result["choices"]:
|
|
raise Exception("No response generated from API")
|
|
|
|
# Extract generated text
|
|
generated_text = result["choices"][0]["message"]["content"]
|
|
|
|
# Log usage information
|
|
if "usage" in result:
|
|
usage = result["usage"]
|
|
print(f"📊 Token usage:")
|
|
print(f" • Prompt tokens: {usage.get('prompt_tokens', 0)}")
|
|
print(f" • Completion tokens: {usage.get('completion_tokens', 0)}")
|
|
print(f" • Total tokens: {usage.get('total_tokens', 0)}")
|
|
|
|
print(f"✅ Text generated successfully!")
|
|
print(f" • Response length: {len(generated_text)} characters")
|
|
|
|
# Save to cache
|
|
self.save_to_cache(cache_key, generated_text)
|
|
|
|
return (generated_text,)
|
|
|
|
except Exception as e:
|
|
print(f"❌ Error generating text: {str(e)}")
|
|
raise e |