File size: 6,700 Bytes
aca8ab4 |
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 |
"""
Configuration management for model pricing and settings.
"""
import os
import json
import logging
from pathlib import Path
from typing import Dict, Optional
logger = logging.getLogger(__name__)
class LangFuseConfig:
"""Manage LangFuse observability configuration."""
def __init__(self):
"""Initialize LangFuse configuration from environment variables."""
self.enabled = os.getenv("LANGFUSE_ENABLED", "true").lower() == "true"
self.public_key = os.getenv("LANGFUSE_PUBLIC_KEY", "")
self.secret_key = os.getenv("LANGFUSE_SECRET_KEY", "")
self.host = os.getenv("LANGFUSE_HOST", "https://cloud.langfuse.com")
# Optional: custom settings
self.trace_all_llm_calls = os.getenv("LANGFUSE_TRACE_ALL_LLM", "true").lower() == "true"
self.trace_rag = os.getenv("LANGFUSE_TRACE_RAG", "true").lower() == "true"
self.flush_at = int(os.getenv("LANGFUSE_FLUSH_AT", "15")) # Flush after N observations
self.flush_interval = int(os.getenv("LANGFUSE_FLUSH_INTERVAL", "10")) # Seconds
def is_configured(self) -> bool:
"""Check if LangFuse is properly configured."""
if not self.enabled:
return False
if not self.public_key or not self.secret_key:
logger.warning("LangFuse is enabled but API keys are missing")
return False
return True
def get_init_params(self) -> Dict:
"""Get initialization parameters for LangFuse client."""
return {
"public_key": self.public_key,
"secret_key": self.secret_key,
"host": self.host,
"flush_at": self.flush_at,
"flush_interval": self.flush_interval,
}
class PricingConfig:
"""Manage model pricing configuration with JSON + env override support."""
def __init__(self, config_path: Optional[str] = None):
"""
Initialize pricing configuration.
Args:
config_path: Path to pricing JSON file (optional)
"""
if config_path is None:
# Default to config/pricing.json relative to project root
project_root = Path(__file__).parent.parent
config_path = project_root / "config" / "pricing.json"
self.config_path = Path(config_path)
self.pricing_data = self._load_pricing_config()
def _load_pricing_config(self) -> Dict:
"""Load pricing configuration from JSON file."""
try:
if not self.config_path.exists():
logger.warning(f"Pricing config not found at {self.config_path}, using defaults")
return self._get_default_pricing()
with open(self.config_path, 'r') as f:
data = json.load(f)
logger.info(f"Loaded pricing config from {self.config_path}")
return data
except Exception as e:
logger.error(f"Error loading pricing config: {e}, using defaults")
return self._get_default_pricing()
def _get_default_pricing(self) -> Dict:
"""Return default pricing if config file not found."""
return {
"models": {
"gpt-4o-mini": {
"input_price_per_1m": 0.15,
"output_price_per_1m": 0.60
},
"phi-4-multimodal-instruct": {
"input_price_per_1m": 0.08,
"output_price_per_1m": 0.32
}
},
"embeddings": {
"text-embedding-3-small": {
"price_per_1m": 0.02
}
}
}
def get_model_pricing(self, model_name: str) -> Dict[str, float]:
"""
Get pricing for a specific model.
Args:
model_name: Model deployment name
Returns:
Dict with input_price_per_1m, output_price_per_1m
"""
# Check for environment variable overrides first
env_input = os.getenv("PRICING_INPUT_PER_1M")
env_output = os.getenv("PRICING_OUTPUT_PER_1M")
if env_input and env_output:
logger.info(f"Using pricing from environment variables: "
f"input=${env_input}, output=${env_output}")
return {
"input_price_per_1m": float(env_input),
"output_price_per_1m": float(env_output)
}
# Fall back to JSON config
model_pricing = self.pricing_data.get("models", {}).get(model_name)
if model_pricing:
logger.info(f"Using pricing for {model_name} from config: "
f"input=${model_pricing['input_price_per_1m']}, "
f"output=${model_pricing['output_price_per_1m']}")
return {
"input_price_per_1m": model_pricing["input_price_per_1m"],
"output_price_per_1m": model_pricing["output_price_per_1m"]
}
# Default fallback
logger.warning(f"Model {model_name} not found in config, using gpt-4o-mini defaults")
return {
"input_price_per_1m": 0.15,
"output_price_per_1m": 0.60
}
def get_embedding_pricing(self, embedding_model: str) -> float:
"""
Get pricing for embedding model.
Args:
embedding_model: Embedding model name
Returns:
Price per 1M tokens
"""
# Check environment variable override
env_embedding = os.getenv("PRICING_EMBEDDING_PER_1M")
if env_embedding:
logger.info(f"Using embedding pricing from env: ${env_embedding}")
return float(env_embedding)
# Fall back to JSON config
embedding_pricing = self.pricing_data.get("embeddings", {}).get(embedding_model)
if embedding_pricing:
price = embedding_pricing["price_per_1m"]
logger.info(f"Using embedding pricing for {embedding_model}: ${price}")
return price
# Default fallback
logger.warning(f"Embedding model {embedding_model} not found, using default $0.02")
return 0.02
# Global instances (lazy loaded)
_pricing_config = None
_langfuse_config = None
def get_pricing_config() -> PricingConfig:
"""Get or create global pricing config instance."""
global _pricing_config
if _pricing_config is None:
_pricing_config = PricingConfig()
return _pricing_config
def get_langfuse_config() -> LangFuseConfig:
"""Get or create global LangFuse config instance."""
global _langfuse_config
if _langfuse_config is None:
_langfuse_config = LangFuseConfig()
return _langfuse_config
|