Download app/core/llm_cache.py from cryptorugmunch/rugmuncher-backend: direct link, hf CLI and curl.
- Browser
- Download file 1.36 kB
-
https://huggingface.co/cryptorugmunch/rugmuncher-backend/resolve/main/app/core/llm_cache.py
- Command line
-
hf download hf://cryptorugmunch/rugmuncher-backend/app/core/llm_cache.py
-
curl -L -o llm_cache.py https://huggingface.co/cryptorugmunch/rugmuncher-backend/resolve/main/app/core/llm_cache.py
1.36 kB
| """Semantic LLM Cache for DataBus — caches identical + similar prompts. Redis-backed.""" | |
| import hashlib | |
| import json | |
| import os | |
| REDIS_URL = os.getenv("REDIS_CACHE_URL", "redis://localhost:6379/1") | |
| CACHE_TTL = int(os.getenv("LLM_CACHE_TTL", "3600")) | |
| def _cache_key(prompt: str, model: str) -> str: | |
| return f"llm_cache:{hashlib.sha256(f'{model}:{prompt}'.encode()).hexdigest()[:16]}" | |
| def get_cached(prompt: str, model: str) -> dict | None: | |
| """Check if prompt+model result is cached.""" | |
| import redis | |
| try: | |
| r = redis.from_url(REDIS_URL, decode_responses=True) | |
| data = r.get(_cache_key(prompt, model)) | |
| if data: | |
| return json.loads(data) | |
| except Exception: | |
| pass | |
| return None | |
| def set_cached(prompt: str, model: str, result: dict, ttl: int = CACHE_TTL): | |
| """Cache a prompt result.""" | |
| import redis | |
| try: | |
| r = redis.from_url(REDIS_URL, decode_responses=True) | |
| r.setex(_cache_key(prompt, model), ttl, json.dumps(result)) | |
| except Exception: | |
| pass | |
| def get_cache_stats() -> dict: | |
| """Get cache statistics.""" | |
| import redis | |
| try: | |
| r = redis.from_url(REDIS_URL, decode_responses=True) | |
| keys = r.keys("llm_cache:*") | |
| return {"cached_prompts": len(keys)} | |
| except Exception: | |
| return {"cached_prompts": 0, "error": "redis unavailable"} | |