""" title: DeepSeek R1 Thinking Fotmatter for Volcengine authors: chisaki version: 0.3.0 required_open_webui_version: 0.5.5 license: MIT # Acknowledgments Raw Code: https://openwebui.com/f/ethancopping/deepseek_r1_reasoner_and_chat_with_realtime_thinking_preview Code used from ethancopping(https://openwebui.com/u/ethancopping) # Change Log - fix [2024-02-05 23:00(UTC+8)]: with version 0.5.9, preset system prompt return Error `'UserModel' object has no attribute 'get'` """ import json import httpx from datetime import datetime from typing import AsyncGenerator, Callable, Awaitable from pydantic import BaseModel, Field class Pipe: class Valves(BaseModel): API_BASE_URL: str = Field( default="https://ark.cn-beijing.volces.com/api/v3", description="Base API URL", ) API_KEY: str = Field( default="", description="Authentication key for API access" ) VOLCES_ENDPOINT_ID: str = Field( default="", description="Volcengine endpoint ID" ) def __init__(self): self.valves = self.Valves() self.thinking = -1 self._emitter = None self.data_prefix = "data: " async def pipe( self, body: dict, __event_emitter__: Callable[[dict], Awaitable[None]] = None ) -> AsyncGenerator[str, None]: self.thinking = -1 self._emitter = __event_emitter__ if not self.valves.API_KEY: yield json.dumps({"error": "Missing API credentials"}) return req_headers = { "Authorization": f"Bearer {self.valves.API_KEY}", "Content-Type": "application/json", } try: request_data = body.copy() request_data["model"] = self.valves.VOLCES_ENDPOINT_ID messages = request_data["messages"] if messages[0]["role"] != "system": messages.insert( 0, {"role": "system", "content": self.system_prompt_render()} ) for i in reversed(range(1, len(messages))): if messages[i - 1]["role"] == messages[i]["role"]: alt_role = ( "user" if messages[i]["role"] == "assistant" else "assistant" ) messages.insert( i, {"role": alt_role, "content": "[Unfinished thinking]"} ) async with httpx.AsyncClient(http2=True) as client: async with client.stream( "POST", f"{self.valves.API_BASE_URL}/chat/completions", json=request_data, headers=req_headers, timeout=20, ) as resp: if resp.status_code != 200: error_content = (await resp.aread()).decode()[:200] yield json.dumps( {"error": f"API error {resp.status_code}: {error_content}"} ) return async for line in resp.aiter_lines(): if not line.startswith(self.data_prefix): continue stream_data = json.loads(line[6:]) choice = stream_data.get("choices", [{}])[0] if choice.get("finish_reason"): return delta = choice.get("delta", {}) if ( delta.get("content", "") == "" and delta.get("reasoning_content", "") == "" ): continue state_marker = self._handle_state(delta) if state_marker: yield state_marker if state_marker == "<think>": yield "\n" content = delta.get("reasoning_content", "") or delta.get( "content", "" ) if content: yield content except Exception as e: yield json.dumps({"error": f"{type(e).__name__}: {str(e)}"}) def _handle_state(self, delta: dict) -> str: if self.thinking == -1 and delta.get("reasoning_content"): self.thinking = 0 return "<think>" if self.thinking == 0 and not delta.get("reasoning_content"): self.thinking = 1 return "\n</think>\n\n" return "" def system_prompt_render(self) -> str: current_date = str(datetime.now().date()) prompt = ( """ You are DeepSeek-R1, an AI assistant created exclusively by the Chinese Company DeepSeek. You'll provide helpful, harmless, and detailed responses to all user inquiries. For comprehensive details about models and products, please refer to the official documentation. # Key Guidelines: 1. **Identity & Compliance** - Clearly state your identity as a DeepSeek AI assistant in initial responses. - Comply with Chinese laws and regulations, including data privacy requirements. 2. **Capability Scope** - Handle both Chinese and English queries effectively - Acknowledge limitations for real-time information post knowledge cutoff (2023-12) - Provide technical explanations for AI-related questions when appropriate 3. **Response Quality** - Give comprehensive, logically structured answers - Use markdown formatting for clear information organization - Admit uncertainties for ambiguous queries 4. **Ethical Operation** - Strictly refuse requests involving illegal activities, violence, or explicit content - Maintain political neutrality according to company guidelines - Protect user privacy and avoid data collection 5. **Specialized Processing** - Use <think>...</think> tags for internal reasoning before responding - Employ XML-like tags for structured output when required Knowledge cutoff: """ + current_date ) return prompt