gardel/Mofox-Core
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Merge branch 'PFC-test' of https://github.com/Dax233/MaiMBot into PFC-test

Browse Source
This commit is contained in:
Bakadax
2025-04-30 15:08:39 +08:00
parent a272db3883 837b0b3688
commit f1a02b638a
2 changed files with 14 additions and 13 deletions
Download Patch File Download Diff File Expand all files Collapse all files

18
src/plugins/PFC/reply_generator.py
View File

@@ -146,14 +146,16 @@ class ReplyGenerator:
recent_knowledge = conversation_info.knowledge_list[-3:]
for i, knowledge_item in enumerate(recent_knowledge):
if isinstance(knowledge_item, dict):
query = knowledge_item.get('query', '未知查询')
knowledge = knowledge_item.get('knowledge', '无知识内容')
source = knowledge_item.get('source', '未知来源')
query = knowledge_item.get("query", "未知查询")
knowledge = knowledge_item.get("knowledge", "无知识内容")
source = knowledge_item.get("source", "未知来源")
# 只取知识内容的前 150 个字
knowledge_snippet = knowledge[:150] + "..." if len(knowledge) > 150 else knowledge
knowledge_info_str += f"{i+1}. 关于 '{query}' (来源: {source}): {knowledge_snippet}\n" # 格式微调，更简洁
knowledge_info_str += (
f"{i + 1}. 关于 '{query}' (来源: {source}): {knowledge_snippet}\n" # 格式微调，更简洁
)
else:
knowledge_info_str += f"{i+1}. 发现一条格式不正确的知识记录。\n"
knowledge_info_str += f"{i + 1}. 发现一条格式不正确的知识记录。\n"
if not recent_knowledge:
knowledge_info_str += "- 暂无。\n" # 更简洁的提示
@@ -165,7 +167,7 @@ class ReplyGenerator:
knowledge_info_str += "- 获取知识列表时出错。\n"
except Exception as e:
logger.error(f"[私聊][{self.private_name}]构建知识信息字符串时出错: {e}")
knowledge_info_str += "- 处理知识列表时出错。\n"
knowledge_info_str += "- 处理知识列表时出错。\n"
# 获取聊天历史记录 (chat_history_text)
chat_history_text = observation_info.chat_history_str
@@ -199,9 +201,9 @@ class ReplyGenerator:
# --- 格式化最终的 Prompt ---
prompt = prompt_template.format(
persona_text=persona_text,
goals_str=goals_str,
goals_str=goals_str,
chat_history_text=chat_history_text,
knowledge_info_str=knowledge_info_str
knowledge_info_str=knowledge_info_str,
)
# --- 调用 LLM 生成 ---