53d9e003af
- 回忆录：事实边界补充允许清单；传记文体示例与 JSON 叙事要求对齐 - default 职业提示 occupation_context；cadre/military 退休语境 - GET 章节读路径零写入，prepare_chapter_read_view + markdown_for_response - 文本归一抽到 core/text_normalize；移除弃用 reply 策略与 recompose_chapters_for_story - ConversationService：WS 连接/用户段落/结束对话；对外错误固定文案 - 测试：HTTP 脱敏契约、章节读视图、occupation 与 background_voice
55 lines
1.7 KiB
Python
55 lines
1.7 KiB
Python
"""
|
||
聊天输入归一：供访谈 Agent / 编排层对 ASR 与键盘输入做可控预处理（规则 / 可选 LLM）。
|
||
|
||
不改变 segment 落库原文；仅作为模型侧派生净稿。
|
||
与 memoir 共用同一套确定性规则，避免聊天与回忆录对同一句理解割裂。
|
||
"""
|
||
|
||
from __future__ import annotations
|
||
|
||
from typing import Any
|
||
|
||
from app.core.config import settings
|
||
from app.core.text_normalize import apply_oral_rules, llm_normalize_text
|
||
from app.core.logging import get_logger
|
||
|
||
logger = get_logger(__name__)
|
||
|
||
apply_conversation_input_rules = apply_oral_rules
|
||
|
||
|
||
def _llm_normalize_chat_input(text: str, llm: Any) -> str | None:
|
||
"""仅修正明显错字与同音字，不增事实；失败返回 None。"""
|
||
return llm_normalize_text(
|
||
text,
|
||
llm,
|
||
max_input_chars=int(settings.chat_input_normalize_llm_max_input_chars),
|
||
max_tokens=int(settings.chat_input_normalize_llm_max_tokens),
|
||
agent_name="chat_input_normalize.llm",
|
||
)
|
||
|
||
|
||
def normalize_chat_input_for_agent(text: str, *, llm: Any | None = None) -> str:
|
||
"""
|
||
聊天侧单一出口：编排层与 InterviewAgent 共用。
|
||
|
||
- 全局关闭：原文
|
||
- off：原文
|
||
- rules：仅规则
|
||
- llm：先规则，再（可选）LLM；无 llm 或失败则保留规则结果
|
||
"""
|
||
if not settings.chat_input_normalize_enabled:
|
||
return text or ""
|
||
mode = (settings.chat_input_normalize_mode or "rules").strip().lower()
|
||
if mode == "off":
|
||
return text or ""
|
||
|
||
base = apply_conversation_input_rules(text or "")
|
||
if mode != "llm":
|
||
return base
|
||
|
||
refined = _llm_normalize_chat_input(base, llm)
|
||
if refined is not None:
|
||
return refined
|
||
return base
|