Spaces:
Sleeping
Sleeping
File size: 16,847 Bytes
ca1fd98 f5ba363 ca1fd98 f5ba363 ca1fd98 f5ba363 ca1fd98 f5ba363 ca1fd98 f5ba363 ca1fd98 02cc090 ca1fd98 02cc090 ca1fd98 02cc090 ca1fd98 02cc090 ca1fd98 | 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 | """Google AI provider implementation (Gemini models)."""
import json
import time
from typing import Any, AsyncIterator
import httpx
from app.models.providers.base import (
AuthenticationError,
BaseProvider,
CompletionResponse,
ModelInfo,
ModelNotFoundError,
ProviderError,
RateLimitError,
TokenUsage,
)
class GoogleProvider(BaseProvider):
"""Google AI API provider supporting Gemini models."""
PROVIDER_NAME = "google"
DEFAULT_BASE_URL = "https://generativelanguage.googleapis.com/v1beta"
# Model definitions with pricing (per 1K tokens)
MODELS = {
# Gemini 2.5 Series
"gemini-2.5-pro": ModelInfo(
id="gemini-2.5-pro",
name="Gemini 2.5 Pro",
provider="google",
context_window=2097152,
max_output_tokens=8192,
supports_functions=True,
supports_vision=True,
supports_streaming=True,
cost_per_1k_input=0.00125,
cost_per_1k_output=0.005,
),
"gemini-2.5-flash": ModelInfo(
id="gemini-2.5-flash",
name="Gemini 2.5 Flash",
provider="google",
context_window=1048576,
max_output_tokens=8192,
supports_functions=True,
supports_vision=True,
supports_streaming=True,
cost_per_1k_input=0.000075,
cost_per_1k_output=0.0003,
),
# Gemini 2.0 Series
"gemini-2.0-flash": ModelInfo(
id="gemini-2.0-flash",
name="Gemini 2.0 Flash",
provider="google",
context_window=1048576,
max_output_tokens=8192,
supports_functions=True,
supports_vision=True,
supports_streaming=True,
cost_per_1k_input=0.0,
cost_per_1k_output=0.0,
),
"gemini-2.0-flash-lite": ModelInfo(
id="gemini-2.0-flash-lite",
name="Gemini 2.0 Flash Lite",
provider="google",
context_window=524288,
max_output_tokens=8192,
supports_functions=True,
supports_vision=True,
supports_streaming=True,
cost_per_1k_input=0.0,
cost_per_1k_output=0.0,
),
# Gemini 3.0 Series (Preview)
"gemini-3-flash-preview": ModelInfo(
id="gemini-3-flash-preview",
name="Gemini 3 Flash Preview",
provider="google",
context_window=1048576,
max_output_tokens=8192,
supports_functions=True,
supports_vision=True,
supports_streaming=True,
cost_per_1k_input=0.0,
cost_per_1k_output=0.0,
),
"gemini-3.1-flash-lite-preview": ModelInfo(
id="gemini-3.1-flash-lite-preview",
name="Gemini 3.1 Flash Lite Preview",
provider="google",
context_window=524288,
max_output_tokens=8192,
supports_functions=True,
supports_vision=True,
supports_streaming=True,
cost_per_1k_input=0.0,
cost_per_1k_output=0.0,
),
# Gemini 1.5 Series (Stable)
"gemini-1.5-pro": ModelInfo(
id="gemini-1.5-pro",
name="Gemini 1.5 Pro",
provider="google",
context_window=2097152,
max_output_tokens=8192,
supports_functions=True,
supports_vision=True,
supports_streaming=True,
cost_per_1k_input=0.00125,
cost_per_1k_output=0.005,
),
"gemini-1.5-flash": ModelInfo(
id="gemini-1.5-flash",
name="Gemini 1.5 Flash",
provider="google",
context_window=1048576,
max_output_tokens=8192,
supports_functions=True,
supports_vision=True,
supports_streaming=True,
cost_per_1k_input=0.000075,
cost_per_1k_output=0.0003,
),
"gemini-pro": ModelInfo(
id="gemini-pro",
name="Gemini Pro",
provider="google",
context_window=32760,
max_output_tokens=8192,
supports_functions=True,
supports_vision=False,
supports_streaming=True,
cost_per_1k_input=0.0005,
cost_per_1k_output=0.0015,
),
}
# Aliases
MODEL_ALIASES = {
"gemini-flash": "gemini-2.5-flash",
"gemini-pro-latest": "gemini-2.5-pro",
"gemini-1.5": "gemini-1.5-pro",
}
def __init__(
self,
api_key: str,
base_url: str | None = None,
timeout: float = 60.0,
max_retries: int = 3,
rate_limit_rpm: int = 60,
):
super().__init__(
api_key=api_key,
base_url=base_url or self.DEFAULT_BASE_URL,
timeout=timeout,
max_retries=max_retries,
rate_limit_rpm=rate_limit_rpm,
)
self._client: httpx.AsyncClient | None = None
async def initialize(self) -> None:
"""Initialize the HTTP client."""
self._client = httpx.AsyncClient(
base_url=self.base_url,
headers={"Content-Type": "application/json"},
timeout=self.timeout,
)
async def shutdown(self) -> None:
"""Close the HTTP client."""
if self._client:
await self._client.aclose()
self._client = None
async def _ensure_client(self) -> httpx.AsyncClient:
"""Ensure client is initialized."""
if not self._client:
await self.initialize()
return self._client # type: ignore
def _resolve_model(self, model: str) -> str:
"""Resolve model alias to full model ID."""
return self.MODEL_ALIASES.get(model, model)
def get_models(self) -> list[ModelInfo]:
"""Get available Google AI models."""
return list(self.MODELS.values())
def _convert_messages(
self, messages: list[dict[str, Any]]
) -> tuple[str | None, list[dict[str, Any]]]:
"""Convert OpenAI-style messages to Gemini format.
Returns:
Tuple of (system_instruction, contents)
"""
system_instruction: str | None = None
contents: list[dict[str, Any]] = []
for msg in messages:
role = msg["role"]
content = msg["content"]
if role == "system":
system_instruction = content
elif role == "assistant":
contents.append({
"role": "model",
"parts": [{"text": content}] if isinstance(content, str) else content,
})
elif role == "user":
contents.append({
"role": "user",
"parts": [{"text": content}] if isinstance(content, str) else content,
})
elif role == "function":
# Function response
contents.append({
"role": "function",
"parts": [{
"functionResponse": {
"name": msg.get("name", "function"),
"response": {"result": content},
}
}],
})
elif role == "tool":
# Tool response
contents.append({
"role": "function",
"parts": [{
"functionResponse": {
"name": msg.get("tool_call_id", "tool"),
"response": {"result": content},
}
}],
})
return system_instruction, contents
def _convert_tools(
self, tools: list[dict[str, Any]] | None
) -> list[dict[str, Any]] | None:
"""Convert OpenAI-style tools to Gemini format."""
if not tools:
return None
function_declarations = []
for tool in tools:
if tool.get("type") == "function":
func = tool["function"]
function_declarations.append({
"name": func["name"],
"description": func.get("description", ""),
"parameters": func.get("parameters", {"type": "object", "properties": {}}),
})
return [{"functionDeclarations": function_declarations}] if function_declarations else None
async def complete(
self,
messages: list[dict[str, Any]],
model: str,
temperature: float = 0.7,
max_tokens: int | None = None,
functions: list[dict[str, Any]] | None = None,
function_call: str | dict[str, str] | None = None,
tools: list[dict[str, Any]] | None = None,
tool_choice: str | dict[str, Any] | None = None,
stop: list[str] | None = None,
**kwargs: Any,
) -> CompletionResponse:
"""Generate a completion using Google AI API."""
import logging
logger = logging.getLogger(__name__)
logger.info(f"GoogleProvider.complete called with model={model}")
await self._acquire_rate_limit()
model = self._resolve_model(model)
logger.info(f"GoogleProvider after resolve: model={model}")
model_info = self.get_model_info(model)
logger.info(f"GoogleProvider model_info: {model_info}")
if not model_info:
raise ModelNotFoundError(self.PROVIDER_NAME, model)
client = await self._ensure_client()
# Convert messages
system_instruction, contents = self._convert_messages(messages)
# Build request payload
payload: dict[str, Any] = {
"contents": contents,
"generationConfig": {
"temperature": temperature,
},
}
if max_tokens:
payload["generationConfig"]["maxOutputTokens"] = max_tokens
if stop:
payload["generationConfig"]["stopSequences"] = stop
if system_instruction:
payload["systemInstruction"] = {"parts": [{"text": system_instruction}]}
# Convert tools
gemini_tools = self._convert_tools(tools)
if not gemini_tools and functions:
gemini_tools = [{
"functionDeclarations": [
{
"name": f["name"],
"description": f.get("description", ""),
"parameters": f.get("parameters", {"type": "object", "properties": {}}),
}
for f in functions
]
}]
if gemini_tools:
payload["tools"] = gemini_tools
start_time = time.time()
url = f"/models/{model}:generateContent?key={self.api_key}"
try:
response = await self._retry_with_backoff(
self._make_request, client, url, payload
)
except httpx.HTTPStatusError as e:
self._handle_http_error(e)
latency_ms = (time.time() - start_time) * 1000
# Parse response
candidates = response.get("candidates", [])
if not candidates:
raise ProviderError("No candidates in response", self.PROVIDER_NAME)
candidate = candidates[0]
content_parts = candidate.get("content", {}).get("parts", [])
# Extract text content and function calls
text_content = ""
tool_calls = []
for part in content_parts:
if "text" in part:
text_content += part["text"]
elif "functionCall" in part:
fc = part["functionCall"]
tool_calls.append({
"id": f"call_{fc['name']}",
"type": "function",
"function": {
"name": fc["name"],
"arguments": json.dumps(fc.get("args", {})),
},
})
# Parse usage
usage_data = response.get("usageMetadata", {})
usage = TokenUsage(
prompt_tokens=usage_data.get("promptTokenCount", 0),
completion_tokens=usage_data.get("candidatesTokenCount", 0),
total_tokens=usage_data.get("totalTokenCount", 0),
)
cost = self.calculate_cost(model, usage)
self._track_usage(usage, cost)
# Map finish reason
finish_reason_map = {
"STOP": "stop",
"MAX_TOKENS": "length",
"SAFETY": "content_filter",
"RECITATION": "content_filter",
}
finish_reason = finish_reason_map.get(
candidate.get("finishReason", ""), candidate.get("finishReason")
)
return CompletionResponse(
content=text_content,
model=model,
provider=self.PROVIDER_NAME,
usage=usage,
finish_reason=finish_reason,
function_call=None,
tool_calls=tool_calls if tool_calls else None,
raw_response=response,
latency_ms=latency_ms,
cost=cost,
)
async def _make_request(
self, client: httpx.AsyncClient, url: str, payload: dict[str, Any]
) -> dict[str, Any]:
"""Make the API request."""
response = await client.post(url, json=payload)
response.raise_for_status()
return response.json()
def _handle_http_error(self, error: httpx.HTTPStatusError) -> None:
"""Handle HTTP errors from Google AI."""
status = error.response.status_code
try:
body = error.response.json()
message = body.get("error", {}).get("message", str(error))
except Exception:
message = str(error)
if status == 401 or status == 403:
raise AuthenticationError(self.PROVIDER_NAME, message)
elif status == 429:
retry_after = error.response.headers.get("retry-after")
raise RateLimitError(
self.PROVIDER_NAME,
retry_after=float(retry_after) if retry_after else None,
message=message,
)
elif status == 404:
# Extract model name from URL if possible
model_name = "unknown"
url = str(error.request.url)
if "/models/" in url:
try:
model_name = url.split("/models/")[1].split(":")[0]
except Exception:
pass
raise ModelNotFoundError(self.PROVIDER_NAME, model_name)
else:
raise ProviderError(message, self.PROVIDER_NAME, status)
async def stream(
self,
messages: list[dict[str, Any]],
model: str,
temperature: float = 0.7,
max_tokens: int | None = None,
**kwargs: Any,
) -> AsyncIterator[str]:
"""Stream a completion from Google AI."""
await self._acquire_rate_limit()
model = self._resolve_model(model)
model_info = self.get_model_info(model)
if not model_info:
raise ModelNotFoundError(self.PROVIDER_NAME, model)
client = await self._ensure_client()
system_instruction, contents = self._convert_messages(messages)
payload: dict[str, Any] = {
"contents": contents,
"generationConfig": {
"temperature": temperature,
},
}
if max_tokens:
payload["generationConfig"]["maxOutputTokens"] = max_tokens
if system_instruction:
payload["systemInstruction"] = {"parts": [{"text": system_instruction}]}
url = f"/models/{model}:streamGenerateContent?key={self.api_key}&alt=sse"
try:
async with client.stream("POST", url, json=payload) as response:
response.raise_for_status()
async for line in response.aiter_lines():
if line.startswith("data: "):
data = line[6:]
try:
chunk = json.loads(data)
candidates = chunk.get("candidates", [])
if candidates:
parts = candidates[0].get("content", {}).get("parts", [])
for part in parts:
if "text" in part:
yield part["text"]
except json.JSONDecodeError:
continue
except httpx.HTTPStatusError as e:
self._handle_http_error(e)
|