better：优化表达方式和侧面人格
This commit is contained in:
SengokuCola
@@ -1,98 +1,18 @@
|
||||
import time
|
||||
import random
|
||||
from typing import List, Dict
|
||||
from typing import List
|
||||
from src.chat.heart_flow.observation.chatting_observation import ChattingObservation
|
||||
from src.chat.heart_flow.observation.observation import Observation
|
||||
from src.llm_models.utils_model import LLMRequest
|
||||
from src.config.config import global_config
|
||||
from src.common.logger import get_logger
|
||||
from src.chat.utils.prompt_builder import Prompt, global_prompt_manager
|
||||
from src.chat.message_receive.chat_stream import get_chat_manager
|
||||
from .base_processor import BaseProcessor
|
||||
from src.chat.focus_chat.info.info_base import InfoBase
|
||||
from src.chat.focus_chat.info.expression_selection_info import ExpressionSelectionInfo
|
||||
from src.chat.express.exprssion_learner import get_expression_learner
|
||||
from json_repair import repair_json
|
||||
import json
|
||||
from src.chat.express.expression_selector import expression_selector
|
||||
|
||||
logger = get_logger("processor")
|
||||
|
||||
|
||||
def weighted_sample_no_replacement(items, weights, k) -> list:
|
||||
"""
|
||||
加权随机抽样，不允许重复
|
||||
|
||||
Args:
|
||||
items: 待抽样的项目列表
|
||||
weights: 对应项目的权重列表
|
||||
k: 抽样数量
|
||||
|
||||
Returns:
|
||||
抽样结果列表
|
||||
"""
|
||||
if not items or k <= 0:
|
||||
return []
|
||||
|
||||
k = min(k, len(items))
|
||||
selected = []
|
||||
remaining_items = list(items)
|
||||
remaining_weights = list(weights)
|
||||
|
||||
for _ in range(k):
|
||||
if not remaining_items:
|
||||
break
|
||||
|
||||
# 计算累积权重
|
||||
total_weight = sum(remaining_weights)
|
||||
if total_weight <= 0:
|
||||
# 如果权重都为0或负数，则随机选择
|
||||
selected_index = random.randint(0, len(remaining_items) - 1)
|
||||
else:
|
||||
# 加权随机选择
|
||||
rand_val = random.uniform(0, total_weight)
|
||||
cumulative_weight = 0
|
||||
selected_index = 0
|
||||
for i, weight in enumerate(remaining_weights):
|
||||
cumulative_weight += weight
|
||||
if rand_val <= cumulative_weight:
|
||||
selected_index = i
|
||||
break
|
||||
|
||||
# 添加选中的项目
|
||||
selected.append(remaining_items[selected_index])
|
||||
# 移除已选中的项目
|
||||
remaining_items.pop(selected_index)
|
||||
remaining_weights.pop(selected_index)
|
||||
|
||||
return selected
|
||||
|
||||
|
||||
def init_prompt():
|
||||
expression_evaluation_prompt = """
|
||||
你的名字是{bot_name}
|
||||
|
||||
以下是正在进行的聊天内容：
|
||||
{chat_observe_info}
|
||||
|
||||
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10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
|
||||
|
||||
请以JSON格式输出，只需要输出选中的情境编号：
|
||||
{{
|
||||
"selected_situations": [1, 3, 5, 7, 9, 12, 15, 18, 21, 25]
|
||||
}}
|
||||
|
||||
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 ExpressionSelectorProcessor(BaseProcessor):
|
||||
log_prefix = "表达选择器"
|
||||
|
||||
@@ -101,16 +21,9 @@ class ExpressionSelectorProcessor(BaseProcessor):
|
||||
|
||||
self.subheartflow_id = subheartflow_id
|
||||
self.last_selection_time = 0
|
||||
self.selection_interval = 40 # 1分钟间隔
|
||||
self.selection_interval = 10 # 40秒间隔
|
||||
self.cached_expressions = [] # 缓存上一次选择的表达方式
|
||||
|
||||
# 表达方式选择模式
|
||||
self.selection_mode = getattr(global_config.expression, "selection_mode", "llm") # "llm" 或 "random"
|
||||
|
||||
self.llm_model = LLMRequest(
|
||||
model=global_config.model.utils_small,
|
||||
request_type="focus.processor.expression_selector",
|
||||
)
|
||||
|
||||
name = get_chat_manager().get_stream_name(self.subheartflow_id)
|
||||
self.log_prefix = f"[{name}] 表达选择器"
|
||||
@@ -158,26 +71,20 @@ class ExpressionSelectorProcessor(BaseProcessor):
|
||||
return []
|
||||
|
||||
try:
|
||||
# 根据模式选择表达方式
|
||||
# LLM模式：调用LLM选择15个，然后随机选5个
|
||||
selected_expressions = await self._select_suitable_expressions_llm(chat_info)
|
||||
# LLM模式：调用LLM选择5-10个，然后随机选5个
|
||||
selected_expressions = await expression_selector.select_suitable_expressions_llm(self.subheartflow_id, chat_info)
|
||||
cache_size = len(selected_expressions) if selected_expressions else 0
|
||||
mode_desc = f"LLM模式（已缓存{cache_size}个）"
|
||||
|
||||
if selected_expressions:
|
||||
# 缓存选择的表达方式
|
||||
self.cached_expressions = selected_expressions
|
||||
# 更新最后选择时间
|
||||
self.last_selection_time = current_time
|
||||
|
||||
# 从选择的表达方式中随机选5个
|
||||
final_expressions = random.sample(selected_expressions, min(4, len(selected_expressions)))
|
||||
|
||||
# 创建表达选择信息
|
||||
expression_info = ExpressionSelectionInfo()
|
||||
expression_info.set_selected_expressions(final_expressions)
|
||||
expression_info.set_selected_expressions(selected_expressions)
|
||||
|
||||
logger.info(f"{self.log_prefix} 为当前聊天选择了{len(final_expressions)}个表达方式（{mode_desc}）")
|
||||
logger.info(f"{self.log_prefix} 为当前聊天选择了{len(selected_expressions)}个表达方式（{mode_desc}）")
|
||||
return [expression_info]
|
||||
else:
|
||||
logger.debug(f"{self.log_prefix} 未选择任何表达方式")
|
||||
@@ -187,104 +94,3 @@ class ExpressionSelectorProcessor(BaseProcessor):
|
||||
logger.error(f"{self.log_prefix} 处理表达方式选择时出错: {e}")
|
||||
return []
|
||||
|
||||
async def _get_random_expressions(self) -> tuple[List[Dict], List[Dict], List[Dict]]:
|
||||
"""随机获取表达方式：20个style，20个grammar，20个personality"""
|
||||
expression_learner = get_expression_learner()
|
||||
|
||||
# 获取所有表达方式
|
||||
(
|
||||
learnt_style_expressions,
|
||||
learnt_grammar_expressions,
|
||||
personality_expressions,
|
||||
) = await expression_learner.get_expression_by_chat_id(self.subheartflow_id)
|
||||
|
||||
# 随机选择
|
||||
selected_style = random.sample(learnt_style_expressions, min(15, len(learnt_style_expressions)))
|
||||
selected_grammar = random.sample(learnt_grammar_expressions, min(15, len(learnt_grammar_expressions)))
|
||||
selected_personality = random.sample(personality_expressions, min(5, len(personality_expressions)))
|
||||
|
||||
return selected_style, selected_grammar, selected_personality
|
||||
|
||||
async def _select_suitable_expressions_llm(self, chat_info: str) -> List[Dict[str, str]]:
|
||||
"""使用LLM选择适合的表达方式"""
|
||||
|
||||
# 1. 获取35个随机表达方式
|
||||
style_exprs, grammar_exprs, personality_exprs = await self._get_random_expressions()
|
||||
|
||||
# 2. 构建所有表达方式的索引和情境列表
|
||||
all_expressions = []
|
||||
all_situations = []
|
||||
|
||||
# 添加style表达方式
|
||||
for expr in style_exprs:
|
||||
if isinstance(expr, dict) and "situation" in expr and "style" in expr:
|
||||
expr_with_type = expr.copy()
|
||||
expr_with_type["type"] = "style"
|
||||
all_expressions.append(expr_with_type)
|
||||
all_situations.append(f"{len(all_expressions)}. [语言风格] {expr['situation']}")
|
||||
|
||||
# 添加grammar表达方式
|
||||
for expr in grammar_exprs:
|
||||
if isinstance(expr, dict) and "situation" in expr and "style" in expr:
|
||||
expr_with_type = expr.copy()
|
||||
expr_with_type["type"] = "grammar"
|
||||
all_expressions.append(expr_with_type)
|
||||
all_situations.append(f"{len(all_expressions)}. [句法语法] {expr['situation']}")
|
||||
|
||||
# 添加personality表达方式
|
||||
for expr in personality_exprs:
|
||||
if isinstance(expr, dict) and "situation" in expr and "style" in expr:
|
||||
expr_with_type = expr.copy()
|
||||
expr_with_type["type"] = "personality"
|
||||
all_expressions.append(expr_with_type)
|
||||
all_situations.append(f"{len(all_expressions)}. [个性表达] {expr['situation']}")
|
||||
|
||||
if not all_expressions:
|
||||
logger.warning(f"{self.log_prefix} 没有找到可用的表达方式")
|
||||
return []
|
||||
|
||||
all_situations_str = "\n".join(all_situations)
|
||||
|
||||
# 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_info,
|
||||
all_situations=all_situations_str,
|
||||
)
|
||||
|
||||
# 4. 调用LLM
|
||||
try:
|
||||
content, _ = await self.llm_model.generate_response_async(prompt=prompt)
|
||||
|
||||
# logger.info(f"{self.log_prefix} LLM返回结果: {content}")
|
||||
|
||||
if not content:
|
||||
logger.warning(f"{self.log_prefix} 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(f"{self.log_prefix} LLM返回格式错误")
|
||||
return []
|
||||
|
||||
selected_indices = result["selected_situations"]
|
||||
|
||||
# 根据索引获取完整的表达方式
|
||||
valid_expressions = []
|
||||
for idx in selected_indices:
|
||||
if isinstance(idx, int) and 1 <= idx <= len(all_expressions):
|
||||
valid_expressions.append(all_expressions[idx - 1]) # 索引从1开始
|
||||
|
||||
logger.info(f"{self.log_prefix} LLM从{len(all_expressions)}个情境中选择了{len(valid_expressions)}个")
|
||||
return valid_expressions
|
||||
|
||||
except Exception as e:
|
||||
logger.error(f"{self.log_prefix} LLM处理表达方式选择时出错: {e}")
|
||||
return []
|
||||
|
||||
|
||||
init_prompt()
|
||||
|
||||
@@ -490,7 +490,7 @@ class DefaultReplyer:
|
||||
learnt_style_expressions,
|
||||
learnt_grammar_expressions,
|
||||
personality_expressions,
|
||||
) = await expression_learner.get_expression_by_chat_id(chat_stream.stream_id)
|
||||
) = expression_learner.get_expression_by_chat_id(chat_stream.stream_id)
|
||||
|
||||
style_habbits = []
|
||||
grammar_habbits = []
|
||||
|
||||
Reference in New Issue
Block a user