mirror of https://github.com/Mai-with-u/MaiBot.git
491 lines
21 KiB
Python
491 lines
21 KiB
Python
import json
|
||
import time
|
||
|
||
from typing import List, Dict, Optional, Any, Tuple
|
||
from json_repair import repair_json
|
||
|
||
from src.llm_models.utils_model import LLMRequest
|
||
from src.config.config import global_config, model_config
|
||
from src.common.logger import get_logger
|
||
from src.common.database.database_model import Expression
|
||
from src.chat.utils.prompt_builder import Prompt, global_prompt_manager
|
||
from src.bw_learner.learner_utils import weighted_sample
|
||
from src.chat.message_receive.chat_stream import get_chat_manager
|
||
|
||
logger = get_logger("expression_selector")
|
||
|
||
|
||
def init_prompt():
|
||
expression_evaluation_prompt = """{chat_observe_info}
|
||
|
||
你的名字是{bot_name}{target_message}
|
||
{reply_reason_block}
|
||
|
||
以下是可选的表达情境：
|
||
{all_situations}
|
||
|
||
请你分析聊天内容的语境、情绪、话题类型，从上述情境中选择最适合当前聊天情境的，最多{max_num}个情境。
|
||
考虑因素包括：
|
||
1.聊天的情绪氛围（轻松、严肃、幽默等）
|
||
2.话题类型（日常、技术、游戏、情感等）
|
||
3.情境与当前语境的匹配度
|
||
{target_message_extra_block}
|
||
|
||
请以JSON格式输出，只需要输出选中的情境编号：
|
||
例如：
|
||
{{
|
||
"selected_situations": [2, 3, 5, 7, 19]
|
||
}}
|
||
|
||
请严格按照JSON格式输出，不要包含其他内容：
|
||
"""
|
||
Prompt(expression_evaluation_prompt, "expression_evaluation_prompt")
|
||
|
||
|
||
class ExpressionSelector:
|
||
def __init__(self):
|
||
self.llm_model = LLMRequest(
|
||
model_set=model_config.model_task_config.tool_use, request_type="expression.selector"
|
||
)
|
||
|
||
def can_use_expression_for_chat(self, chat_id: str) -> bool:
|
||
"""
|
||
检查指定聊天流是否允许使用表达
|
||
|
||
Args:
|
||
chat_id: 聊天流ID
|
||
|
||
Returns:
|
||
bool: 是否允许使用表达
|
||
"""
|
||
try:
|
||
use_expression, _, _ = global_config.expression.get_expression_config_for_chat(chat_id)
|
||
return use_expression
|
||
except Exception as e:
|
||
logger.error(f"检查表达使用权限失败: {e}")
|
||
return False
|
||
|
||
@staticmethod
|
||
def _parse_stream_config_to_chat_id(stream_config_str: str) -> Optional[str]:
|
||
"""解析'platform:id:type'为chat_id，直接使用 ChatManager 提供的接口"""
|
||
try:
|
||
parts = stream_config_str.split(":")
|
||
if len(parts) != 3:
|
||
return None
|
||
platform = parts[0]
|
||
id_str = parts[1]
|
||
stream_type = parts[2]
|
||
is_group = stream_type == "group"
|
||
# 统一通过 chat_manager 生成 stream_id，避免各处自行实现哈希逻辑
|
||
return get_chat_manager().get_stream_id(platform, str(id_str), is_group=is_group)
|
||
except Exception:
|
||
return None
|
||
|
||
def get_related_chat_ids(self, chat_id: str) -> List[str]:
|
||
"""根据expression_groups配置，获取与当前chat_id相关的所有chat_id（包括自身）"""
|
||
groups = global_config.expression.expression_groups
|
||
|
||
# 检查是否存在全局共享组（包含"*"的组）
|
||
global_group_exists = any("*" in group for group in groups)
|
||
|
||
if global_group_exists:
|
||
# 如果存在全局共享组，则返回所有可用的chat_id
|
||
all_chat_ids = set()
|
||
for group in groups:
|
||
for stream_config_str in group:
|
||
if chat_id_candidate := self._parse_stream_config_to_chat_id(stream_config_str):
|
||
all_chat_ids.add(chat_id_candidate)
|
||
return list(all_chat_ids) if all_chat_ids else [chat_id]
|
||
|
||
# 否则使用现有的组逻辑
|
||
for group in groups:
|
||
group_chat_ids = []
|
||
for stream_config_str in group:
|
||
if chat_id_candidate := self._parse_stream_config_to_chat_id(stream_config_str):
|
||
group_chat_ids.append(chat_id_candidate)
|
||
if chat_id in group_chat_ids:
|
||
return group_chat_ids
|
||
return [chat_id]
|
||
|
||
def _select_expressions_simple(self, chat_id: str, max_num: int) -> Tuple[List[Dict[str, Any]], List[int]]:
|
||
"""
|
||
简单模式：只选择 count > 1 的项目，要求至少有10个才进行选择，随机选5个，不进行LLM选择
|
||
|
||
Args:
|
||
chat_id: 聊天流ID
|
||
max_num: 最大选择数量（此参数在此模式下不使用，固定选择5个）
|
||
|
||
Returns:
|
||
Tuple[List[Dict[str, Any]], List[int]]: 选中的表达方式列表和ID列表
|
||
"""
|
||
try:
|
||
# 支持多chat_id合并抽选
|
||
related_chat_ids = self.get_related_chat_ids(chat_id)
|
||
|
||
# 查询所有相关chat_id的表达方式，排除 rejected=1 的，且只选择 count > 1 的
|
||
# 如果 expression_checked_only 为 True，则只选择 checked=True 且 rejected=False 的
|
||
base_conditions = (Expression.chat_id.in_(related_chat_ids)) & (~Expression.rejected) & (Expression.count > 1)
|
||
if global_config.expression.expression_checked_only:
|
||
base_conditions = base_conditions & (Expression.checked)
|
||
style_query = Expression.select().where(base_conditions)
|
||
|
||
style_exprs = [
|
||
{
|
||
"id": expr.id,
|
||
"situation": expr.situation,
|
||
"style": expr.style,
|
||
"last_active_time": expr.last_active_time,
|
||
"source_id": expr.chat_id,
|
||
"create_date": expr.create_date if expr.create_date is not None else expr.last_active_time,
|
||
"count": expr.count if getattr(expr, "count", None) is not None else 1,
|
||
"checked": expr.checked if getattr(expr, "checked", None) is not None else False,
|
||
}
|
||
for expr in style_query
|
||
]
|
||
|
||
# 要求至少有一定数量的 count > 1 的表达方式才进行“完整简单模式”选择
|
||
min_required = 8
|
||
if len(style_exprs) < min_required:
|
||
# 高 count 样本不足：如果还有候选，就降级为随机选 3 个；如果一个都没有，则直接返回空
|
||
if not style_exprs:
|
||
logger.info(
|
||
f"聊天流 {chat_id} 没有满足 count > 1 且未被拒绝的表达方式，简单模式不进行选择"
|
||
)
|
||
# 完全没有高 count 样本时，退化为全量随机抽样（不进入LLM流程）
|
||
fallback_num = min(3, max_num) if max_num > 0 else 3
|
||
fallback_selected = self._random_expressions(chat_id, fallback_num)
|
||
if fallback_selected:
|
||
self.update_expressions_last_active_time(fallback_selected)
|
||
selected_ids = [expr["id"] for expr in fallback_selected]
|
||
logger.info(
|
||
f"聊天流 {chat_id} 使用简单模式降级随机抽选 {len(fallback_selected)} 个表达（无 count>1 样本）"
|
||
)
|
||
return fallback_selected, selected_ids
|
||
return [], []
|
||
logger.info(
|
||
f"聊天流 {chat_id} count > 1 的表达方式不足 {min_required} 个（实际 {len(style_exprs)} 个），"
|
||
f"简单模式降级为随机选择 3 个"
|
||
)
|
||
select_count = min(3, len(style_exprs))
|
||
else:
|
||
# 高 count 数量达标时，固定选择 5 个
|
||
select_count = 5
|
||
import random
|
||
|
||
selected_style = random.sample(style_exprs, select_count)
|
||
|
||
# 更新last_active_time
|
||
if selected_style:
|
||
self.update_expressions_last_active_time(selected_style)
|
||
|
||
selected_ids = [expr["id"] for expr in selected_style]
|
||
logger.debug(
|
||
f"think_level=0: 从 {len(style_exprs)} 个 count>1 的表达方式中随机选择了 {len(selected_style)} 个"
|
||
)
|
||
return selected_style, selected_ids
|
||
|
||
except Exception as e:
|
||
logger.error(f"简单模式选择表达方式失败: {e}")
|
||
return [], []
|
||
|
||
def _random_expressions(self, chat_id: str, total_num: int) -> List[Dict[str, Any]]:
|
||
"""
|
||
随机选择表达方式
|
||
|
||
Args:
|
||
chat_id: 聊天室ID
|
||
total_num: 需要选择的数量
|
||
|
||
Returns:
|
||
List[Dict[str, Any]]: 随机选择的表达方式列表
|
||
"""
|
||
try:
|
||
# 支持多chat_id合并抽选
|
||
related_chat_ids = self.get_related_chat_ids(chat_id)
|
||
|
||
# 优化：一次性查询所有相关chat_id的表达方式，排除 rejected=1 的表达
|
||
# 如果 expression_checked_only 为 True，则只选择 checked=True 且 rejected=False 的
|
||
base_conditions = (Expression.chat_id.in_(related_chat_ids)) & (~Expression.rejected)
|
||
if global_config.expression.expression_checked_only:
|
||
base_conditions = base_conditions & (Expression.checked)
|
||
style_query = Expression.select().where(base_conditions)
|
||
|
||
style_exprs = [
|
||
{
|
||
"id": expr.id,
|
||
"situation": expr.situation,
|
||
"style": expr.style,
|
||
"last_active_time": expr.last_active_time,
|
||
"source_id": expr.chat_id,
|
||
"create_date": expr.create_date if expr.create_date is not None else expr.last_active_time,
|
||
"count": expr.count if getattr(expr, "count", None) is not None else 1,
|
||
"checked": expr.checked if getattr(expr, "checked", None) is not None else False,
|
||
}
|
||
for expr in style_query
|
||
]
|
||
|
||
# 随机抽样
|
||
if style_exprs:
|
||
selected_style = weighted_sample(style_exprs, total_num)
|
||
else:
|
||
selected_style = []
|
||
|
||
return selected_style
|
||
|
||
except Exception as e:
|
||
logger.error(f"随机选择表达方式失败: {e}")
|
||
return []
|
||
|
||
async def select_suitable_expressions(
|
||
self,
|
||
chat_id: str,
|
||
chat_info: str,
|
||
max_num: int = 10,
|
||
target_message: Optional[str] = None,
|
||
reply_reason: Optional[str] = None,
|
||
think_level: int = 1,
|
||
) -> Tuple[List[Dict[str, Any]], List[int]]:
|
||
"""
|
||
选择适合的表达方式（使用classic模式：随机选择+LLM选择）
|
||
|
||
Args:
|
||
chat_id: 聊天流ID
|
||
chat_info: 聊天内容信息
|
||
max_num: 最大选择数量
|
||
target_message: 目标消息内容
|
||
reply_reason: planner给出的回复理由
|
||
think_level: 思考级别，0/1
|
||
|
||
Returns:
|
||
Tuple[List[Dict[str, Any]], List[int]]: 选中的表达方式列表和ID列表
|
||
"""
|
||
# 检查是否允许在此聊天流中使用表达
|
||
if not self.can_use_expression_for_chat(chat_id):
|
||
logger.debug(f"聊天流 {chat_id} 不允许使用表达，返回空列表")
|
||
return [], []
|
||
|
||
# 使用classic模式（随机选择+LLM选择）
|
||
logger.debug(f"使用classic模式为聊天流 {chat_id} 选择表达方式，think_level={think_level}")
|
||
return await self._select_expressions_classic(
|
||
chat_id, chat_info, max_num, target_message, reply_reason, think_level
|
||
)
|
||
|
||
async def _select_expressions_classic(
|
||
self,
|
||
chat_id: str,
|
||
chat_info: str,
|
||
max_num: int = 10,
|
||
target_message: Optional[str] = None,
|
||
reply_reason: Optional[str] = None,
|
||
think_level: int = 1,
|
||
) -> Tuple[List[Dict[str, Any]], List[int]]:
|
||
"""
|
||
classic模式：随机选择+LLM选择
|
||
|
||
Args:
|
||
chat_id: 聊天流ID
|
||
chat_info: 聊天内容信息
|
||
max_num: 最大选择数量
|
||
target_message: 目标消息内容
|
||
reply_reason: planner给出的回复理由
|
||
think_level: 思考级别，0/1
|
||
|
||
Returns:
|
||
Tuple[List[Dict[str, Any]], List[int]]: 选中的表达方式列表和ID列表
|
||
"""
|
||
try:
|
||
# think_level == 0: 只选择 count > 1 的项目，随机选10个，不进行LLM选择
|
||
if think_level == 0:
|
||
return self._select_expressions_simple(chat_id, max_num)
|
||
|
||
# think_level == 1: 先选高count，再从所有表达方式中随机抽样
|
||
# 1. 获取所有表达方式并分离 count > 1 和 count <= 1 的
|
||
related_chat_ids = self.get_related_chat_ids(chat_id)
|
||
# 如果 expression_checked_only 为 True，则只选择 checked=True 且 rejected=False 的
|
||
base_conditions = (Expression.chat_id.in_(related_chat_ids)) & (~Expression.rejected)
|
||
if global_config.expression.expression_checked_only:
|
||
base_conditions = base_conditions & (Expression.checked)
|
||
style_query = Expression.select().where(base_conditions)
|
||
|
||
all_style_exprs = [
|
||
{
|
||
"id": expr.id,
|
||
"situation": expr.situation,
|
||
"style": expr.style,
|
||
"last_active_time": expr.last_active_time,
|
||
"source_id": expr.chat_id,
|
||
"create_date": expr.create_date if expr.create_date is not None else expr.last_active_time,
|
||
"count": expr.count if getattr(expr, "count", None) is not None else 1,
|
||
"checked": expr.checked if getattr(expr, "checked", None) is not None else False,
|
||
}
|
||
for expr in style_query
|
||
]
|
||
|
||
# 分离 count > 1 和 count <= 1 的表达方式
|
||
high_count_exprs = [expr for expr in all_style_exprs if (expr.get("count", 1) or 1) > 1]
|
||
|
||
# 根据 think_level 设置要求（仅支持 0/1，0 已在上方返回）
|
||
min_high_count = 10
|
||
min_total_count = 10
|
||
select_high_count = 5
|
||
select_random_count = 5
|
||
|
||
# 检查数量要求
|
||
# 对于高 count 表达：如果数量不足，不再直接停止，而是仅跳过“高 count 优先选择”
|
||
if len(high_count_exprs) < min_high_count:
|
||
logger.info(
|
||
f"聊天流 {chat_id} count > 1 的表达方式不足 {min_high_count} 个（实际 {len(high_count_exprs)} 个），"
|
||
f"将跳过高 count 优先选择，仅从全部表达中随机抽样"
|
||
)
|
||
high_count_valid = False
|
||
else:
|
||
high_count_valid = True
|
||
|
||
# 总量不足仍然直接返回，避免样本过少导致选择质量过低
|
||
if len(all_style_exprs) < min_total_count:
|
||
logger.info(
|
||
f"聊天流 {chat_id} 总表达方式不足 {min_total_count} 个（实际 {len(all_style_exprs)} 个），不进行选择"
|
||
)
|
||
return [], []
|
||
|
||
# 先选取高count的表达方式（如果数量达标）
|
||
if high_count_valid:
|
||
selected_high = weighted_sample(high_count_exprs, min(len(high_count_exprs), select_high_count))
|
||
else:
|
||
selected_high = []
|
||
|
||
# 然后从所有表达方式中随机抽样（使用加权抽样）
|
||
remaining_num = select_random_count
|
||
selected_random = weighted_sample(all_style_exprs, min(len(all_style_exprs), remaining_num))
|
||
|
||
# 合并候选池（去重，避免重复）
|
||
candidate_exprs = selected_high.copy()
|
||
candidate_ids = {expr["id"] for expr in candidate_exprs}
|
||
for expr in selected_random:
|
||
if expr["id"] not in candidate_ids:
|
||
candidate_exprs.append(expr)
|
||
candidate_ids.add(expr["id"])
|
||
|
||
# 打乱顺序，避免高count的都在前面
|
||
import random
|
||
|
||
random.shuffle(candidate_exprs)
|
||
|
||
# 2. 构建所有表达方式的索引和情境列表
|
||
all_expressions: List[Dict[str, Any]] = []
|
||
all_situations: List[str] = []
|
||
|
||
# 添加style表达方式
|
||
for expr in candidate_exprs:
|
||
expr = expr.copy()
|
||
all_expressions.append(expr)
|
||
all_situations.append(f"{len(all_expressions)}.当 {expr['situation']} 时，使用 {expr['style']}")
|
||
|
||
if not all_expressions:
|
||
logger.warning("没有找到可用的表达方式")
|
||
return [], []
|
||
|
||
all_situations_str = "\n".join(all_situations)
|
||
|
||
if target_message:
|
||
target_message_str = f'，现在你想要对这条消息进行回复："{target_message}"'
|
||
target_message_extra_block = "4.考虑你要回复的目标消息"
|
||
else:
|
||
target_message_str = ""
|
||
target_message_extra_block = ""
|
||
|
||
chat_context = f"以下是正在进行的聊天内容：{chat_info}"
|
||
|
||
# 构建reply_reason块
|
||
if reply_reason:
|
||
reply_reason_block = f"你的回复理由是：{reply_reason}"
|
||
chat_context = ""
|
||
else:
|
||
reply_reason_block = ""
|
||
|
||
# 3. 构建prompt（只包含情境，不包含完整的表达方式）
|
||
prompt = (await global_prompt_manager.get_prompt_async("expression_evaluation_prompt")).format(
|
||
bot_name=global_config.bot.nickname,
|
||
chat_observe_info=chat_context,
|
||
all_situations=all_situations_str,
|
||
max_num=max_num,
|
||
target_message=target_message_str,
|
||
target_message_extra_block=target_message_extra_block,
|
||
reply_reason_block=reply_reason_block,
|
||
)
|
||
|
||
# 4. 调用LLM
|
||
content, (reasoning_content, model_name, _) = await self.llm_model.generate_response_async(prompt=prompt)
|
||
|
||
# print(prompt)
|
||
# print(content)
|
||
|
||
if not content:
|
||
logger.warning("LLM返回空结果")
|
||
return [], []
|
||
|
||
# 5. 解析结果
|
||
result = repair_json(content)
|
||
if isinstance(result, str):
|
||
result = json.loads(result)
|
||
|
||
if not isinstance(result, dict) or "selected_situations" not in result:
|
||
logger.error("LLM返回格式错误")
|
||
logger.info(f"LLM返回结果: \n{content}")
|
||
return [], []
|
||
|
||
selected_indices = result["selected_situations"]
|
||
|
||
# 根据索引获取完整的表达方式
|
||
valid_expressions: List[Dict[str, Any]] = []
|
||
selected_ids = []
|
||
for idx in selected_indices:
|
||
if isinstance(idx, int) and 1 <= idx <= len(all_expressions):
|
||
expression = all_expressions[idx - 1] # 索引从1开始
|
||
selected_ids.append(expression["id"])
|
||
valid_expressions.append(expression)
|
||
|
||
# 对选中的所有表达方式，更新last_active_time
|
||
if valid_expressions:
|
||
self.update_expressions_last_active_time(valid_expressions)
|
||
|
||
logger.debug(f"从{len(all_expressions)}个情境中选择了{len(valid_expressions)}个")
|
||
return valid_expressions, selected_ids
|
||
|
||
except Exception as e:
|
||
logger.error(f"classic模式处理表达方式选择时出错: {e}")
|
||
return [], []
|
||
|
||
def update_expressions_last_active_time(self, expressions_to_update: List[Dict[str, Any]]):
|
||
"""对一批表达方式更新last_active_time"""
|
||
if not expressions_to_update:
|
||
return
|
||
updates_by_key = {}
|
||
for expr in expressions_to_update:
|
||
source_id: str = expr.get("source_id") # type: ignore
|
||
situation: str = expr.get("situation") # type: ignore
|
||
style: str = expr.get("style") # type: ignore
|
||
if not source_id or not situation or not style:
|
||
logger.warning(f"表达方式缺少必要字段，无法更新: {expr}")
|
||
continue
|
||
key = (source_id, situation, style)
|
||
if key not in updates_by_key:
|
||
updates_by_key[key] = expr
|
||
for chat_id, situation, style in updates_by_key:
|
||
query = Expression.select().where(
|
||
(Expression.chat_id == chat_id) & (Expression.situation == situation) & (Expression.style == style)
|
||
)
|
||
if query.exists():
|
||
expr_obj = query.get()
|
||
expr_obj.last_active_time = time.time()
|
||
expr_obj.save()
|
||
logger.debug("表达方式激活: 更新last_active_time in db")
|
||
|
||
|
||
init_prompt()
|
||
|
||
try:
|
||
expression_selector = ExpressionSelector()
|
||
except Exception as e:
|
||
logger.error(f"ExpressionSelector初始化失败: {e}")
|