feat(data-models): 使用 __slots__ 优化内存占用和属性访问性能,更新多个数据模型

This commit is contained in:
Windpicker-owo
2025-12-01 19:57:33 +08:00
parent e72f39c42b
commit 943c2a6566
12 changed files with 175 additions and 63 deletions

View File

@@ -3,6 +3,7 @@ from typing import Any
class BaseDataModel:
__slots__ = ()
def deepcopy(self):
return copy.deepcopy(self)

View File

@@ -5,11 +5,12 @@ from typing import Any
from . import BaseDataModel
@dataclass
@dataclass(slots=True)
class DatabaseUserInfo(BaseDataModel):
"""
用户信息数据模型,用于存储用户的基本信息。
该类通过 dataclass 实现,继承自 BaseDataModel。
使用 __slots__ 优化内存占用和属性访问性能。
"""
platform: str = field(default_factory=str) # 用户所属平台如微信、QQ 等)
user_id: str = field(default_factory=str) # 用户唯一标识 ID
@@ -35,10 +36,12 @@ class DatabaseUserInfo(BaseDataModel):
"user_cardname": self.user_cardname,
}
@dataclass
@dataclass(slots=True)
class DatabaseGroupInfo(BaseDataModel):
"""
群组信息数据模型,用于存储群组的基本信息。
使用 __slots__ 优化内存占用和属性访问性能。
"""
group_id: str = field(default_factory=str) # 群组唯一标识 ID
group_name: str = field(default_factory=str) # 群组名称
@@ -52,7 +55,7 @@ class DatabaseGroupInfo(BaseDataModel):
group_name=data.get("group_name", ""),
platform=data.get("platform"),
)
def to_dict(self) -> dict:
"""将实例转换为字典"""
return {
@@ -60,12 +63,14 @@ class DatabaseGroupInfo(BaseDataModel):
"group_name": self.group_name,
"group_platform": self.platform,
}
@dataclass
@dataclass(slots=True)
class DatabaseChatInfo(BaseDataModel):
"""
聊天会话信息数据模型,用于描述一个聊天对话的上下文信息。
包括会话 ID、平台、创建时间、最后活跃时间以及关联的用户和群组信息。
使用 __slots__ 优化内存占用和属性访问性能。
"""
stream_id: str = field(default_factory=str) # 会话流 ID唯一标识一个聊天对话
platform: str = field(default_factory=str) # 所属平台如微信、QQ 等)
@@ -80,7 +85,50 @@ class DatabaseMessages(BaseDataModel):
"""
消息数据模型,用于存储每一条消息的完整信息,包括内容、元数据、用户、聊天上下文等。
使用 init=False 实现自定义初始化逻辑,通过 __init__ 手动设置字段。
使用 __slots__ 优化内存占用和属性访问性能。
"""
__slots__ = (
# 基础消息字段
"message_id",
"time",
"chat_id",
"reply_to",
"interest_value",
"key_words",
"key_words_lite",
"is_mentioned",
"is_at",
"reply_probability_boost",
"processed_plain_text",
"display_message",
"priority_mode",
"priority_info",
"additional_config",
"is_emoji",
"is_picid",
"is_command",
"is_notify",
"is_public_notice",
"notice_type",
"selected_expressions",
"is_read",
"actions",
"should_reply",
"should_act",
# 关联对象
"user_info",
"group_info",
"chat_info",
# 运行时扩展字段(固定)
"semantic_embedding",
"interest_calculated",
"is_voice",
"is_video",
"has_emoji",
"has_picid",
)
def __init__(
self,
message_id: str = "", # 消息唯一 ID
@@ -101,30 +149,38 @@ class DatabaseMessages(BaseDataModel):
is_emoji: bool = False, # 是否为表情消息
is_picid: bool = False, # 是否为图片消息(包含图片 ID
is_command: bool = False, # 是否为命令消息(如 /help
is_notify: bool = False, # 是否为notice消息如禁言、戳一戳等系统事件
is_public_notice: bool = False, # 是否为公共notice所有聊天可见
notice_type: str | None = None, # notice类型由适配器指定如 "group_ban", "poke" 等)
is_notify: bool = False, # 是否为 notice 消息(如禁言、戳一戳等系统事件)
is_public_notice: bool = False, # 是否为公共 notice所有聊天可见
notice_type: str | None = None, # notice 类型(由适配器指定,如 "group_ban", "poke" 等)
selected_expressions: str | None = None, # 选择的表情或响应模板
is_read: bool = False, # 是否已读
user_id: str = "", # 用户 ID
user_nickname: str = "", # 用户昵称
user_cardname: str | None = None, # 用户备注名或群名片
user_platform: str = "", # 用户所属平台
chat_info_group_id: str | None = None, # 所属群组 ID聊天上下文信息
chat_info_group_name: str | None = None, # 所属群组名称
chat_info_group_platform: str | None = None, # 所属群组平台
chat_info_user_id: str = "", # 聊天上下文中的用户 ID
chat_info_user_nickname: str = "", # 聊天上下文中的用户昵称
chat_info_user_cardname: str | None = None, # 聊天上下文中的用户备注名
chat_info_user_platform: str = "", # 聊天上下文中的用户平台
chat_info_stream_id: str = "", # 聊天上下文的会话流 ID
chat_info_platform: str = "", # 聊天上下文平台
chat_info_create_time: float = 0.0, # 聊天上下文创建时间
chat_info_last_active_time: float = 0.0, # 聊天上下文最后活跃时间
actions: list | None = None, # 与消息相关的动作列表(如回复、转发等)
should_reply: bool = False, # 是否应该自动回复
should_act: bool = False, # 是否应该执行动作(如发送消息)
**kwargs: Any, # 允许传入任意额外字段
# 用户信息(用于构建 user_info
user_id: str = "",
user_nickname: str = "",
user_cardname: str | None = None,
user_platform: str = "",
# 群组 / 聊天上下文信息(用于构建 group_info / chat_info
chat_info_group_id: str | None = None,
chat_info_group_name: str | None = None,
chat_info_group_platform: str | None = None,
chat_info_user_id: str = "",
chat_info_user_nickname: str = "",
chat_info_user_cardname: str | None = None,
chat_info_user_platform: str = "",
chat_info_stream_id: str = "",
chat_info_platform: str = "",
chat_info_create_time: float = 0.0,
chat_info_last_active_time: float = 0.0,
# 运行时字段(固定)
semantic_embedding: Any | None = None,
interest_calculated: bool = False,
is_voice: bool = False, # 是否为语音消息
is_video: bool = False, # 是否为视频消息
has_emoji: bool = False, # 是否包含表情
has_picid: bool = False, # 是否包含图片 ID
):
# 初始化基础字段
self.message_id = message_id
@@ -186,21 +242,19 @@ class DatabaseMessages(BaseDataModel):
group_info=self.group_info,
)
# 扩展运行时字段
self.semantic_embedding = kwargs.pop("semantic_embedding", None)
self.interest_calculated = kwargs.pop("interest_calculated", False)
# 处理额外传入的字段kwargs
if kwargs:
for key, value in kwargs.items():
setattr(self, key, value)
# 运行时字段
self.semantic_embedding = semantic_embedding
self.interest_calculated = interest_calculated
self.is_voice = is_voice
self.is_video = is_video
self.has_emoji = has_emoji
self.has_picid = has_picid
# 注意: id 参数从数据库加载时会传入,但不存储(使用 message_id 作为业务主键)
def flatten(self) -> dict[str, Any]:
"""
将消息对象转换为字典格式,便于序列化存储或传输。
Returns:
包含所有字段的字典,其中嵌套对象(如 user_info、group_info已展开为扁平结构。
嵌套对象(如 user_info、group_info、chat_info展开为扁平结构。
"""
return {
"message_id": self.message_id,
@@ -228,13 +282,17 @@ class DatabaseMessages(BaseDataModel):
"is_read": self.is_read,
"actions": self.actions,
"should_reply": self.should_reply,
"should_act": self.should_act,
# user_info 展开
"user_id": self.user_info.user_id,
"user_nickname": self.user_info.user_nickname,
"user_cardname": self.user_info.user_cardname,
"user_platform": self.user_info.platform,
# group_info 展开(可能为 None
"chat_info_group_id": self.group_info.group_id if self.group_info else None,
"chat_info_group_name": self.group_info.group_name if self.group_info else None,
"chat_info_group_platform": self.group_info.platform if self.group_info else None,
# chat_info 展开
"chat_info_stream_id": self.chat_info.stream_id,
"chat_info_platform": self.chat_info.platform,
"chat_info_create_time": self.chat_info.create_time,
@@ -243,6 +301,9 @@ class DatabaseMessages(BaseDataModel):
"chat_info_user_nickname": self.chat_info.user_info.user_nickname,
"chat_info_user_cardname": self.chat_info.user_info.user_cardname,
"chat_info_user_platform": self.chat_info.user_info.platform,
# 运行时字段
"semantic_embedding": self.semantic_embedding,
"interest_calculated": self.interest_calculated,
}
def update_message_info(
@@ -301,13 +362,62 @@ class DatabaseMessages(BaseDataModel):
"display_message": self.display_message,
}
# DatabaseMessages 接受的所有参数名集合(用于 from_dict 过滤)
_VALID_INIT_PARAMS: frozenset[str] = frozenset({
"message_id", "time", "chat_id", "reply_to", "interest_value",
"key_words", "key_words_lite", "is_mentioned", "is_at",
"reply_probability_boost", "processed_plain_text", "display_message",
"priority_mode", "priority_info", "additional_config",
"is_emoji", "is_picid", "is_command", "is_notify", "is_public_notice",
"notice_type", "selected_expressions", "is_read", "actions",
"should_reply", "should_act",
"user_id", "user_nickname", "user_cardname", "user_platform",
"chat_info_group_id", "chat_info_group_name", "chat_info_group_platform",
"chat_info_user_id", "chat_info_user_nickname", "chat_info_user_cardname",
"chat_info_user_platform", "chat_info_stream_id", "chat_info_platform",
"chat_info_create_time", "chat_info_last_active_time",
"semantic_embedding", "interest_calculated",
"is_voice", "is_video", "has_emoji", "has_picid",
"id", # 数据库自增主键
})
@classmethod
def from_dict(cls, data: dict[str, Any]) -> "DatabaseMessages":
"""
从字典创建 DatabaseMessages 实例,自动过滤掉不支持的参数。
Args:
data: 包含消息数据的字典(如从数据库查询返回的结果)
Returns:
DatabaseMessages 实例
"""
# 只保留有效的参数
filtered_data = {k: v for k, v in data.items() if k in cls._VALID_INIT_PARAMS}
return cls(**filtered_data)
@dataclass(init=False)
class DatabaseActionRecords(BaseDataModel):
"""
动作记录数据模型,用于记录系统执行的某个操作或动作的详细信息。
用于审计、日志或调试。
使用 __slots__ 优化内存占用和属性访问性能。
"""
__slots__ = (
"action_id",
"time",
"action_name",
"action_data",
"action_done",
"action_build_into_prompt",
"action_prompt_display",
"chat_id",
"chat_info_stream_id",
"chat_info_platform",
)
def __init__(
self,
action_id: str, # 动作唯一 ID

View File

@@ -533,7 +533,7 @@ class StreamContext(BaseDataModel):
loaded_count = 0
for msg_dict in db_messages:
try:
db_msg = DatabaseMessages(**msg_dict)
db_msg = DatabaseMessages.from_dict(msg_dict)
db_msg.is_read = True
self.history_messages.append(db_msg)
loaded_count += 1

View File

@@ -2,6 +2,7 @@
记忆图系统核心数据模型
定义节点、边、记忆等核心数据结构(包含三层记忆系统)
使用 __slots__ 优化内存占用和属性访问性能
"""
from __future__ import annotations
@@ -112,7 +113,7 @@ class MemoryStatus(Enum):
ARCHIVED = "archived" # 已归档(低价值,很少访问)
@dataclass
@dataclass(slots=True)
class MemoryNode:
"""记忆节点"""
@@ -168,7 +169,7 @@ class MemoryNode:
return f"Node({self.node_type.value}: {self.content})"
@dataclass
@dataclass(slots=True)
class MemoryEdge:
"""记忆边(节点之间的关系)"""
@@ -219,7 +220,7 @@ class MemoryEdge:
return f"Edge({self.source_id} --{self.relation}--> {self.target_id})"
@dataclass
@dataclass(slots=True)
class Memory:
"""完整记忆(由节点和边组成的子图)"""
@@ -342,7 +343,7 @@ class Memory:
return f"Memory({self.memory_type.value}: {self.to_text()})"
@dataclass
@dataclass(slots=True)
class StagedMemory:
"""临时记忆(未整理状态)"""
@@ -379,7 +380,7 @@ class StagedMemory:
# ============================================================================
@dataclass
@dataclass(slots=True)
class MemoryBlock:
"""
感知记忆块
@@ -439,7 +440,7 @@ class MemoryBlock:
return f"MemoryBlock({self.id[:8]}, messages={len(self.messages)}, recalls={self.recall_count})"
@dataclass
@dataclass(slots=True)
class PerceptualMemory:
"""
感知记忆(记忆堆的完整状态)
@@ -478,7 +479,7 @@ class PerceptualMemory:
)
@dataclass
@dataclass(slots=True)
class ShortTermMemory:
"""
短期记忆
@@ -558,7 +559,7 @@ class ShortTermMemory:
return f"ShortTermMemory({self.id[:8]}, content={self.content[:30]}..., importance={self.importance:.2f})"
@dataclass
@dataclass(slots=True)
class GraphOperation:
"""
图操作指令
@@ -604,7 +605,7 @@ class GraphOperation:
return f"GraphOperation({self.operation_type.value}, target={self.target_id}, confidence={self.confidence:.2f})"
@dataclass
@dataclass(slots=True)
class JudgeDecision:
"""
裁判模型决策结果
@@ -648,7 +649,7 @@ class JudgeDecision:
return f"JudgeDecision({status}, confidence={self.confidence:.2f}, extra_queries={len(self.additional_queries)})"
@dataclass
@dataclass(slots=True)
class ShortTermDecision:
"""
短期记忆决策结果

View File

@@ -33,7 +33,7 @@ if TYPE_CHECKING:
logger = get_logger(__name__)
@dataclass
@dataclass(slots=True)
class Path:
"""表示一条路径"""
@@ -58,7 +58,7 @@ class Path:
return node_id in self.nodes
@dataclass
@dataclass(slots=True)
class PathExpansionConfig:
"""路径扩展配置"""

View File

@@ -16,7 +16,7 @@ class InjectionType(Enum):
return self.value
@dataclass
@dataclass(slots=True)
class InjectionRule:
"""Prompt注入规则"""
@@ -118,7 +118,7 @@ class EventType(Enum):
return self.value
@dataclass
@dataclass(slots=True)
class PythonDependency:
"""Python包依赖信息"""
@@ -139,7 +139,7 @@ class PythonDependency:
return self.install_name
@dataclass
@dataclass(slots=True)
class PermissionNodeField:
"""权限节点声明字段"""
@@ -147,7 +147,7 @@ class PermissionNodeField:
description: str # 权限描述
@dataclass
@dataclass(slots=True)
class AdapterInfo:
"""适配器组件信息"""

View File

@@ -6,7 +6,7 @@ from dataclasses import dataclass, field
from typing import Any
@dataclass
@dataclass(slots=True)
class ConfigField:
"""配置字段定义"""

View File

@@ -4,7 +4,7 @@ from typing import Any
from src.plugin_system.base.component_types import PythonDependency
@dataclass
@dataclass(slots=True)
class PluginMetadata:
"""
插件元数据,用于存储插件的开发者信息和用户帮助信息。

View File

@@ -15,7 +15,7 @@ from src.common.logger import get_logger
logger = get_logger("stream_tool_history")
@dataclass
@dataclass(slots=True)
class ToolCallRecord:
"""工具调用记录"""
tool_name: str

View File

@@ -17,7 +17,7 @@ from src.plugin_system.core.stream_tool_history import ToolCallRecord, get_strea
logger = get_logger("tool_use")
@dataclass
@dataclass(slots=True)
class ToolExecutionConfig:
"""工具执行配置"""
max_concurrent_tools: int = 5 # 最大并发工具数量
@@ -25,7 +25,7 @@ class ToolExecutionConfig:
enable_dependency_check: bool = True # 是否启用依赖检查
@dataclass
@dataclass(slots=True)
class ToolExecutionResult:
"""工具执行结果"""
tool_call: ToolCall

View File

@@ -481,7 +481,7 @@ class ChatterPlanFilter:
)
# 将字典转换为DatabaseMessages对象
read_messages = [
DatabaseMessages(**msg_dict) for msg_dict in fallback_messages_dicts
DatabaseMessages.from_dict(msg_dict) for msg_dict in fallback_messages_dicts
]
unread_messages = stream_context.get_unread_messages() # 获取未读消息
@@ -646,8 +646,8 @@ class ChatterPlanFilter:
target_message_obj["message_id"] = target_message_obj["id"]
try:
# 使用 ** 解包字典传入构造函数
action_message_obj = DatabaseMessages(**target_message_obj)
# 使用 from_dict 工厂方法创建对象(自动过滤无效参数)
action_message_obj = DatabaseMessages.from_dict(target_message_obj)
logger.debug(
f"[{action}] 成功转换目标消息为 DatabaseMessages 对象: {action_message_obj.message_id}"
)
@@ -670,7 +670,7 @@ class ChatterPlanFilter:
if latest_message_dict:
from src.common.data_models.database_data_model import DatabaseMessages
try:
action_message_obj = DatabaseMessages(**latest_message_dict)
action_message_obj = DatabaseMessages.from_dict(latest_message_dict)
logger.info(f"[{action}] 成功使用最新消息: {action_message_obj.message_id}")
except Exception as e:
logger.error(f"[{action}] 无法转换最新消息: {e}")
@@ -689,7 +689,7 @@ class ChatterPlanFilter:
if target_message_dict:
from src.common.data_models.database_data_model import DatabaseMessages
try:
action_message_obj = DatabaseMessages(**target_message_dict)
action_message_obj = DatabaseMessages.from_dict(target_message_dict)
except Exception as e:
logger.error(
f"[{action}] 无法将默认的最新消息转换为 DatabaseMessages 对象: {e}",

View File

@@ -309,7 +309,7 @@ class PromptBuilder:
limit=30, # 限制数量,私聊不需要太多
)
history_messages = [
DatabaseMessages(**msg_dict) for msg_dict in fallback_messages_dicts
DatabaseMessages.from_dict(msg_dict) for msg_dict in fallback_messages_dicts
]
if not history_messages: