{"repo": "balizero1987/ayo", "files": {"backend/services/llm_clients/deepseek_client.py": {"23cd129904123772abdfe94e41758c4e74cb2b80": {"t": 1785745440.041311, "n": 6588, "h": "734ad4a71f2313f4", "c": "\"\"\"\nDeepSeek V3 Client - Cheapest LLM API fallback\n\nPricing: $0.27/1M input, $1.10/1M output (December 2025)\nAPI: OpenAI-compatible\n\nFallback chain: Gemini 2.0 Flash → DeepSeek V3 → OpenRouter (free models)\n\"\"\"\n\nimport logging\nfrom collections.abc import AsyncGenerator\nfrom dataclasses import dataclass\n\nimport httpx\n\nfrom app.core.config import settings\n\nlogger = logging.getLogger(__name__)\n\n\n@dataclass\nclass DeepSeekResponse:\n    \"\"\"Response from DeepSeek API\"\"\"\n\n    content: str\n    model_name: str\n    input_tokens: int\n    output_tokens: int\n    finish_reason: str\n\n\nclass DeepSeekClient:\n    \"\"\"\n    DeepSeek V3 Client - Cheapest available LLM API.\n\n    Pricing (December 2025):\n    - Input: $0.27 per 1M tokens\n    - Output: $1.10 per 1M tokens\n    - Context: 64k tokens\n\n    Use as fallback when Gemini quota is exhausted.\n    \"\"\"\n\n    BASE_URL = \"https://api.deepseek.com/v1\"\n    DEFAULT_MODEL = \"deepseek-chat\"  # DeepSeek V3\n\n    def __init__(self, api_key: str | None = None):\n        \"\"\"\n        Initialize DeepSeek client.\n\n        Args:\n            api_key: DeepSeek API key (defaults to settings.deepseek_api_key)\n        \"\"\"\n        self.api_key = api_key or settings.deepseek_api_key\n        if not self.api_key:\n            logger.warning(\"⚠️ DeepSeek API key not set. Set DEEPSEEK_API_KEY env var.\")\n\n    @property\n    def is_available(self) -> bool:\n        \"\"\"Check if DeepSeek is configured and available\"\"\"\n        return bool(self.api_key)\n\n    async def complete(\n        self,\n        messages: list[dict],\n        model: str = DEFAULT_MODEL,\n        temperature: float = 0.7,\n        max_tokens: int = 8192,\n    ) -> DeepSeekResponse:\n        \"\"\"\n        Generate completion from DeepSeek.\n\n        Args:\n            messages: List of messages in OpenAI format\n            model: Model to use (default: deepseek-chat)\n            temperature: Sampling temperature (0.0-2.0)\n            max_tokens: Maximum tokens to generate\n\n        Returns:\n            DeepSeekResponse with content and metadata\n        \"\"\"\n        if not self.api_key:\n            raise RuntimeError(\"DeepSeek API key not configured\")\n\n        async with httpx.AsyncClient(timeout=60.0) as client:\n            response = await client.post(\n                f\"{self.BASE_URL}/chat/completions\",\n                headers={\n                    \"Authorization\": f\"Bearer {self.api_key}\",\n                    \"Content-Type\": \"application/json\",\n                },\n                json={\n                    \"model\": model,\n                    \"messages\": messages,\n                    \"temperature\": temperature,\n                    \"max_tokens\": max_tokens,\n                },\n            )\n\n            if response.status_code != 200:\n                error_text = response.text\n                logger.error(f\"DeepSeek API error: {response.status_code} - {error_text}\")\n                raise RuntimeError(f\"DeepSeek API error: {response.status_code}\")\n\n            data = response.json()\n\n            return DeepSeekResponse(\n                content=data[\"choices\"][0][\"message\"][\"content\"],\n                model_name=data.get(\"model\", model),\n                input_tokens=data.get(\"usage\", {}).get(\"prompt_tokens\", 0),\n                output_tokens=data.get(\"usage\", {}).get(\"completion_tokens\", 0),\n                finish_reason=data[\"choices\"][0].get(\"finish_reason\", \"stop\"),\n            )\n\n    async def complete_stream(\n        self,\n        messages: list[dict],\n        model: str = DEFAULT_MODEL,\n        temperature: float = 0.7,\n        max_tokens: int = 8192,\n    ) -> AsyncGenerator[str, None]:\n        \"\"\"\n        Stream completion from DeepSeek.\n\n        Args:\n            messages: List of messages in OpenAI format\n            model: Model to use\n            temperature: Sampling temperature\n            max_tokens: Maximum tokens to generate\n\n        Yields:\n            Text chunks as they are generated\n        \"\"\"\n        if not self.api_key:\n            raise RuntimeError(\"DeepSeek API key not configured\")\n\n        async with (\n            httpx.AsyncClient(timeout=120.0) as client,\n            client.stream(\n                \"POST\",\n                f\"{self.BASE_URL}/chat/completions\",\n                headers={\n                    \"Authorization\": f\"Bearer {self.api_key}\",\n                    \"Content-Type\": \"application/json\",\n                },\n                json={\n                    \"model\": model,\n                    \"messages\": messages,\n                    \"temperature\": temperature,\n                    \"max_tokens\": max_tokens,\n                    \"stream\": True,\n                },\n            ) as response,\n        ):\n            if response.status_code != 200:\n                error_text = await response.aread()\n                logger.error(f\"DeepSeek streaming error: {response.status_code}\")\n                raise RuntimeError(f\"DeepSeek API error: {response.status_code}\")\n\n            async for line in response.aiter_lines():\n                if line.startswith(\"data: \"):\n                    data_str = line[6:]\n                    if data_str == \"[DONE]\":\n                        break\n\n                    try:\n                        import json\n\n                        data = json.loads(data_str)\n                        delta = data.get(\"choices\", [{}])[0].get(\"delta\", {})\n                        content = delta.get(\"content\", \"\")\n                        if content:\n                            yield content\n                    except Exception as e:\n                        logger.warning(f\"Failed to parse DeepSeek stream: {e}\")\n                        continue\n\n\ndeepseek_client = DeepSeekClient()\n\n\nif __name__ == \"__main__\":\n    import asyncio\n\n    async def test():\n        logger.info(\"🚀 Testing DeepSeek V3 Client...\")\n        logger.info(f\"   API Key: {'✅ Set' if settings.deepseek_api_key else '❌ Not set'}\")\n\n        if not settings.deepseek_api_key:\n            logger.info(\"Set DEEPSEEK_API_KEY to test\")\n            return\n\n        client = DeepSeekClient()\n\n        messages = [\n            {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n            {\"role\": \"user\", \"content\": \"What is 2+2? Answer in one word.\"},\n        ]\n\n        try:\n            result = await client.complete(messages)\n            logger.info(f\"✅ Response: {result.content}\")\n            logger.info(f\"   Model: {result.model_name}\")\n            logger.info(f\"   Tokens: {result.input_tokens} in, {result.output_tokens} out\")\n        except Exception as e:\n            logger.error(f\"❌ Error: {e}\")\n\n    asyncio.run(test())\n"}}}}