Spaces:
Sleeping
Sleeping
File size: 10,357 Bytes
f866820 | 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 |
# src/llm_providers.py
# Priority: GEMINI β GROQ β OPENROUTER β fallback
import os
import json
import time
from typing import Optional, Dict, Any
try:
import requests
_HAS_REQUESTS = True
except Exception:
import urllib.request as _urllib_request
import urllib.error as _urllib_error
_HAS_REQUESTS = False
def _http_post(url: str, headers: dict, payload: dict, timeout: int = 30):
"""
Perform HTTP POST request with JSON payload.
Args:
url: Target URL
headers: HTTP headers
payload: JSON-serializable payload
timeout: Request timeout in seconds
Returns:
Parsed JSON response
Raises:
requests.RequestException: If using requests and request fails
urllib.error.URLError: If using urllib and request fails
json.JSONDecodeError: If response is not valid JSON
ValueError: If url is empty or payload is not serializable
"""
if not url:
raise ValueError("URL cannot be empty")
try:
data = json.dumps(payload).encode("utf-8")
except Exception as e:
raise ValueError(f"Failed to serialize payload to JSON: {str(e)}")
if _HAS_REQUESTS:
try:
r = requests.post(url, headers=headers, data=json.dumps(payload), timeout=timeout)
r.raise_for_status()
return r.json()
except requests.RequestException:
raise
except json.JSONDecodeError as e:
raise json.JSONDecodeError(f"Failed to decode JSON response: {str(e)}")
else:
try:
req = _urllib_request.Request(url, data=data, headers=headers, method="POST")
with _urllib_request.urlopen(req, timeout=timeout) as resp:
response_data = resp.read().decode("utf-8")
return json.loads(response_data)
except _urllib_error.URLError:
raise
except json.JSONDecodeError as e:
raise json.JSONDecodeError(f"Failed to decode JSON response: {str(e)}")
# GEMINI ------------------------------------------------------------
def _call_gemini(prompt: str, temperature: float, max_tokens: int, context: Optional[str]):
"""
Call Gemini API with prompt and context.
Args:
prompt: User prompt
temperature: Sampling temperature (0.0-1.0)
max_tokens: Maximum tokens to generate
context: Additional context for the prompt
Returns:
Dict with 'text' and 'meta' keys
Raises:
RuntimeError: If GEMINI_API_KEY is not set
Exception: If API call fails
"""
api_key = os.getenv("GEMINI_API_KEY")
if not api_key:
raise RuntimeError("GEMINI_API_KEY missing")
model = os.getenv("GEMINI_MODEL", "gemini-1.5-flash")
url = f"https://generativelanguage.googleapis.com/v1beta/models/{model}:generateContent?key={api_key}"
content = prompt
if context:
content = f"Context:\n{context}\n\nUser question:\n{prompt}"
payload = {
"contents": [{"parts": [{"text": content}]}],
"generationConfig": {
"temperature": float(temperature),
"maxOutputTokens": int(max_tokens)
}
}
start = time.time()
try:
j = _http_post(url, {"Content-Type": "application/json"}, payload)
except Exception as e:
raise RuntimeError(f"Gemini API call failed: {str(e)}")
elapsed = time.time() - start
try:
text = j["candidates"][0]["content"]["parts"][0]["text"]
except (KeyError, IndexError, TypeError) as e:
text = json.dumps(j)[:1000]
raise RuntimeError(f"Unexpected Gemini API response format: {str(e)}. Response: {text}")
return {"text": text, "meta": {"provider": "gemini", "model": model, "elapsed_s": elapsed}}
# GROQ --------------------------------------------------------------
def _call_groq(prompt: str, temperature: float, max_tokens: int, context: Optional[str]):
"""
Call Groq API with prompt and context.
Args:
prompt: User prompt
temperature: Sampling temperature (0.0-1.0)
max_tokens: Maximum tokens to generate
context: Additional context for the prompt
Returns:
Dict with 'text' and 'meta' keys
Raises:
RuntimeError: If GROQ_API_KEY is not set
Exception: If API call fails
"""
api_key = os.getenv("GROQ_API_KEY")
if not api_key:
raise RuntimeError("GROQ_API_KEY missing")
url = "https://api.groq.com/openai/v1/chat/completions"
model = os.getenv("GROQ_MODEL", "mixtral-8x7b-32768")
system_msg = "You are a concise assistant. Include citations if context is provided."
messages = [
{"role": "system", "content": system_msg},
{"role": "user", "content": f"Context:\n{context}\n\n{prompt}" if context else prompt}
]
payload = {
"model": model,
"messages": messages,
"temperature": float(temperature),
"max_tokens": int(max_tokens)
}
headers = {"Authorization": f"Bearer {api_key}", "Content-Type": "application/json"}
start = time.time()
try:
j = _http_post(url, headers, payload)
except Exception as e:
raise RuntimeError(f"Groq API call failed: {str(e)}")
elapsed = time.time() - start
try:
text = j["choices"][0]["message"]["content"]
except (KeyError, IndexError, TypeError) as e:
text = json.dumps(j)[:1000]
raise RuntimeError(f"Unexpected Groq API response format: {str(e)}. Response: {text}")
return {"text": text, "meta": {"provider": "groq", "model": model, "elapsed_s": elapsed}}
# OPENROUTER --------------------------------------------------------
def _call_openrouter(prompt: str, temperature: float, max_tokens: int, context: Optional[str]):
"""
Call OpenRouter API with prompt and context.
Args:
prompt: User prompt
temperature: Sampling temperature (0.0-1.0)
max_tokens: Maximum tokens to generate
context: Additional context for the prompt
Returns:
Dict with 'text' and 'meta' keys
Raises:
RuntimeError: If OPENROUTER_API_KEY is not set
Exception: If API call fails
"""
api_key = os.getenv("OPENROUTER_API_KEY")
if not api_key:
raise RuntimeError("OPENROUTER_API_KEY missing")
url = os.getenv("OPENROUTER_URL", "https://api.openrouter.ai/v1/chat/completions")
model = os.getenv("OPENROUTER_MODEL", "gpt-4o-mini")
system_msg = "You are a concise assistant. Use supplied context."
user_content = f"Context:\n{context}\n\n{prompt}" if context else prompt
messages = [
{"role": "system", "content": system_msg},
{"role": "user", "content": user_content}
]
payload = {
"model": model,
"messages": messages,
"temperature": float(temperature),
"max_tokens": int(max_tokens)
}
headers = {"Authorization": f"Bearer {api_key}", "Content-Type": "application/json"}
start = time.time()
try:
j = _http_post(url, headers, payload)
except Exception as e:
raise RuntimeError(f"OpenRouter API call failed: {str(e)}")
elapsed = time.time() - start
try:
text = j["choices"][0]["message"]["content"]
except (KeyError, IndexError, TypeError) as e:
text = json.dumps(j)[:1000]
raise RuntimeError(f"Unexpected OpenRouter API response format: {str(e)}. Response: {text}")
return {"text": text, "meta": {"provider": "openrouter", "model": model, "elapsed_s": elapsed}}
# FALLBACK ----------------------------------------------------------
def _fallback(prompt: str, context: Optional[str]):
"""
Simple fallback responder when all LLM providers fail.
Args:
prompt: User prompt
context: Additional context
Returns:
Dict with 'text' and 'meta' keys
"""
ctx = (context or "")[:800]
text = f"[offline] {prompt}\n\nContext snippet:\n{ctx}"
return {"text": text, "meta": {"provider": "local-fallback"}}
# PUBLIC ENTRYPOINT -------------------------------------------------
def call_llm(prompt: str, temperature: float = 0.0, max_tokens: int = 512, context: Optional[str] = None, **kwargs):
"""
Call LLM with automatic fallback cascade: Gemini β Groq β OpenRouter β Local.
If one provider fails, automatically tries the next one.
Args:
prompt: User prompt
temperature: Sampling temperature (0.0-1.0)
max_tokens: Maximum tokens to generate
context: Additional context for the prompt
**kwargs: Additional arguments passed to provider functions
Returns:
Dict with 'text' and 'meta' keys containing the response and metadata
Raises:
Exception: If all providers fail
"""
if not prompt or not isinstance(prompt, str):
raise ValueError("prompt must be a non-empty string")
# Validate temperature and max_tokens
temperature = max(0.0, min(1.0, float(temperature))) # Clamp to [0.0, 1.0]
max_tokens = max(1, int(max_tokens)) # Ensure positive
errors = []
# Try Gemini first
if os.getenv("GEMINI_API_KEY"):
try:
return _call_gemini(prompt, temperature, max_tokens, context)
except Exception as e:
errors.append(f"gemini: {str(e)}")
# Continue to next provider
# Try Groq second
if os.getenv("GROQ_API_KEY"):
try:
return _call_groq(prompt, temperature, max_tokens, context)
except Exception as e:
errors.append(f"groq: {str(e)}")
# Continue to next provider
# Try OpenRouter third
if os.getenv("OPENROUTER_API_KEY"):
try:
return _call_openrouter(prompt, temperature, max_tokens, context)
except Exception as e:
errors.append(f"openrouter: {str(e)}")
# Continue to fallback
# All providers failed, use local fallback
error_summary = "; ".join(errors) if errors else "No API keys configured"
return {
"text": f"[All providers failed: {error_summary}] Using local fallback.",
"meta": {"provider": "local-fallback", "errors": errors}
} |