""" ProfileAgent：用户资料收集 Specialist 负责提取资料、资料追问、资料收集开场白，不负责 Redis 持久化（由 Orchestrator 统一处理） """ import json import time from typing import Any, Dict, List, Optional from langchain_core.messages import AIMessage, HumanMessage, SystemMessage from app.agents.chat.helpers import format_history_string, get_history_with_window from app.agents.chat.prompts_profile import ( get_profile_extraction_prompt, get_profile_followup_prompt, get_profile_greeting_prompt, ) from app.core.agent_logging import agent_span, log_agent_payload, log_agent_summary from app.core.config import settings from app.core.dependencies import get_llm_provider from app.core.json_utils import extract_json_payload from app.core.langchain_llm import ainvoke_json_object from app.core.logging import get_logger from app.agents.chat.reply_limits import ( nonempty_segments_or_fallback, segments_from_llm_response, truncate_chat_segments, ) logger = get_logger(__name__) def _get_langchain_llm(): try: provider = get_llm_provider() return getattr(provider, "langchain_llm", None) except Exception: return None def _message_contents_char_count(messages: List[Any]) -> int: n = 0 for m in messages: c = getattr(m, "content", None) if isinstance(c, str): n += len(c) return n class ProfileAgent: """用户资料收集 Specialist Agent""" def __init__(self): self.llm = _get_langchain_llm() async def extract_profile_from_message( self, user_message: str, missing_fields: List[str], conversation_id: Optional[str] = None, ) -> Dict[str, Any]: """从用户消息中提取资料字段，不持久化""" if not self.llm or not missing_fields: return {} recent_dialogue = "" if conversation_id: hw = await get_history_with_window( conversation_id, max_pairs=settings.chat_history_max_pairs, max_chars=settings.chat_history_max_chars, ) recent = hw.window[-4:] if len(hw.window) > 4 else hw.window parts = [] for msg in recent: if isinstance(msg, HumanMessage): parts.append(f"用户: {msg.content}") elif isinstance(msg, AIMessage): parts.append(f"助手: {msg.content}") recent_dialogue = "\n".join(parts) if parts else "" try: prompt = get_profile_extraction_prompt( user_message, missing_fields, recent_dialogue=recent_dialogue or None ) content = await ainvoke_json_object( self.llm, prompt, max_tokens=512, agent="ProfileAgent.extract_profile_from_message", ) parsed = json.loads(extract_json_payload(content)) result = {} if "birth_year" in parsed and parsed["birth_year"] is not None: raw = parsed["birth_year"] if isinstance(raw, int) and 1900 <= raw <= 2100: result["birth_year"] = raw elif isinstance(raw, str) and raw.isdigit(): y = int(raw) if y < 100: y = 1900 + y if y >= 50 else 2000 + y if 1900 <= y <= 2100: result["birth_year"] = y if "birth_place" in parsed and parsed["birth_place"]: result["birth_place"] = str(parsed["birth_place"]) if "grew_up_place" in parsed and parsed["grew_up_place"]: result["grew_up_place"] = str(parsed["grew_up_place"]) if "occupation" in parsed and parsed["occupation"]: result["occupation"] = str(parsed["occupation"]) return result except (json.JSONDecodeError, Exception) as e: logger.error("提取资料信息失败: {}", e) return {} async def generate_profile_followup( self, conversation_id: str, user_message: str, missing_fields: List[str], filled_fields: Dict[str, str], nickname: str = "", interview_stage_hint: str = "", ) -> List[str]: """生成资料追问回复，不持久化（由 Orchestrator 负责）""" if not self.llm: return ["谢谢！还能告诉我更多吗？"] try: prompt = get_profile_followup_prompt( missing_fields, filled_fields, user_message, nickname, interview_stage_hint=interview_stage_hint, ) hw = await get_history_with_window( conversation_id, max_pairs=settings.chat_history_max_pairs, max_chars=settings.chat_history_max_chars, ) messages: List[Any] = [SystemMessage(content=prompt)] messages.extend(hw.window) messages.append(HumanMessage(content=user_message)) log_agent_payload( logger, "ProfileAgent.followup.prompt", format_history_string(messages), ) chat_llm = self.llm.bind( max_tokens=settings.chat_profile_followup_max_tokens ) llm_t0 = time.perf_counter() with agent_span( logger, "ProfileAgent.followup.llm", conversation_id=conversation_id, ): logger.info( "event=chat_prompt_built agent=ProfileAgent.generate_profile_followup " "prompt_chars={} history_pairs_total={} history_pairs_windowed={}", _message_contents_char_count(messages), hw.turn_total, len(hw.window) // 2, ) response = await chat_llm.ainvoke(messages) logger.info( "event=chat_llm_done agent=ProfileAgent.generate_profile_followup " "response_latency_ms={:.2f}", (time.perf_counter() - llm_t0) * 1000, ) response_text = ( response.content if hasattr(response, "content") else str(response) ) log_agent_payload( logger, "ProfileAgent.followup.raw_response", response_text ) raw_list = segments_from_llm_response(response_text, max_segments=3) if not raw_list: raw_list = [response_text.strip()] out = truncate_chat_segments( raw_list, max_segments=3, max_chars_per_segment=settings.chat_interview_max_chars_per_segment, ) log_agent_summary( logger, "ProfileAgent.followup segments={} conversation_id={}", len(out), conversation_id, ) segments = ( out if out else [ response_text.strip()[ : settings.chat_interview_max_chars_per_segment ] ] ) return nonempty_segments_or_fallback( segments, fallback="谢谢分享！能再告诉我一些吗？", ) except Exception as e: logger.error("生成资料跟进回复失败: {}", e) return ["谢谢分享！能再告诉我一些吗？"] async def generate_profile_greeting( self, conversation_id: str, missing_fields: List[str], nickname: str = "", ) -> List[str]: """生成资料收集开场白，不持久化（由 Orchestrator 负责）""" if not self.llm: return ["你好！在开始之前，能告诉我你是哪一年出生的吗？"] try: prompt = get_profile_greeting_prompt(missing_fields, nickname) hw = await get_history_with_window( conversation_id, max_pairs=settings.chat_history_max_pairs, max_chars=settings.chat_history_max_chars, ) messages: List[Any] = [SystemMessage(content=prompt)] messages.extend(hw.window) if hw.window: messages.append( HumanMessage(content="（请根据上文自然接话，继续资料收集开场。）") ) else: messages.append(HumanMessage(content="（请说出资料收集开场白。）")) log_agent_payload( logger, "ProfileAgent.greeting.prompt", format_history_string(messages) ) chat_llm = self.llm.bind( max_tokens=settings.chat_profile_followup_max_tokens ) llm_t0 = time.perf_counter() with agent_span( logger, "ProfileAgent.greeting.llm", conversation_id=conversation_id, ): logger.info( "event=chat_prompt_built agent=ProfileAgent.generate_profile_greeting " "prompt_chars={} history_pairs_total={} history_pairs_windowed={}", _message_contents_char_count(messages), hw.turn_total, len(hw.window) // 2, ) response = await chat_llm.ainvoke(messages) logger.info( "event=chat_llm_done agent=ProfileAgent.generate_profile_greeting " "response_latency_ms={:.2f}", (time.perf_counter() - llm_t0) * 1000, ) response_text = ( response.content if hasattr(response, "content") else str(response) ) log_agent_payload( logger, "ProfileAgent.greeting.raw_response", response_text ) raw_list = segments_from_llm_response(response_text, max_segments=2) if not raw_list: raw_list = [response_text.strip()] out = truncate_chat_segments( raw_list, max_segments=2, max_chars_per_segment=settings.chat_interview_max_chars_per_segment, ) log_agent_summary( logger, "ProfileAgent.greeting segments={} conversation_id={}", len(out), conversation_id, ) segments = ( out if out else [ response_text.strip()[ : settings.chat_interview_max_chars_per_segment ] ] ) return nonempty_segments_or_fallback( segments, fallback="你好！在开始之前，能告诉我你是哪一年出生的吗？", ) except Exception as e: logger.error("生成资料收集开场白失败: {}", e) return [ "你好！在我们开始聊人生故事之前，能先简单介绍一下你自己吗？比如你是哪一年出生的？" ]