mirror of
https://github.com/xinnan-tech/xiaozhi-esp32-server.git
synced 2026-07-22 07:03:53 +08:00
Merge pull request #2944 from xinnan-tech/py_test_Memory_powermem
Py test memory powermem
This commit is contained in:
wengzh
@@ -242,7 +242,7 @@ Websocket接口地址: wss://2662r3426b.vicp.fun/xiaozhi/v1/
|
||||
| 智能对话 | 支持多种LLM(大语言模型)，实现智能对话 |
|
||||
| 视觉感知 | 支持多种VLLM(视觉大模型)，实现多模态交互 |
|
||||
| 意图识别 | 支持外挂的大模型意图识别、大模型自主函数调用，提供插件化意图处理机制 |
|
||||
| 记忆系统 | 支持本地短期记忆、mem0ai接口记忆，具备记忆总结功能 |
|
||||
| 记忆系统 | 支持本地短期记忆、mem0ai接口记忆、PowerMem智能记忆，具备记忆总结功能 |
|
||||
| 知识库 | 支持RAGFlow知识库，让大模型判断需要调度知识库后再回答 |
|
||||
| 工具调用 | 支持客户端IOT协议、客户MCP协议、服务端MCP协议、MCP接入点协议、自定义工具函数 |
|
||||
| 指令下发 | 依托MQTT协议，支持从智控台将MCP指令下发到ESP32设备 |
|
||||
@@ -330,6 +330,7 @@ Websocket接口地址: wss://2662r3426b.vicp.fun/xiaozhi/v1/
|
||||
| 类型 | 平台名称 | 使用方式 | 收费模式 | 备注 |
|
||||
|:------:|:---------------:|:----:|:---------:|:--:|
|
||||
| Memory | mem0ai | 接口调用 | 1000次/月额度 | |
|
||||
| Memory | [powermem](./docs/powermem-integration.md) | 本地总结 | 取决于LLM和DB | OceanBase开源，支持智能检索 |
|
||||
| Memory | mem_local_short | 本地总结 | 免费 | |
|
||||
| Memory | nomem | 无记忆模式 | 免费 | |
|
||||
|
||||
|
||||
+2
-1
@@ -240,7 +240,7 @@ Dieses Projekt bietet die folgenden Testwerkzeuge, um Ihnen bei der Überprüfun
|
||||
| Intelligenter Dialog | Unterstützt mehrere LLM (große Sprachmodelle), implementiert intelligenten Dialog |
|
||||
| Visuelle Wahrnehmung | Unterstützt mehrere VLLM (Vision Large Models), implementiert multimodale Interaktion |
|
||||
| Absichtserkennung | Unterstützt LLM-Absichtserkennung, Function Call-Funktionsaufruf, bietet plugin-basierten Absichtsverarbeitungsmechanismus |
|
||||
| Gedächtnissystem | Unterstützt lokales Kurzzeitgedächtnis, mem0ai-Schnittstellengedächtnis, mit Gedächtniszusammenfassungsfunktion |
|
||||
| Gedächtnissystem | Unterstützt lokales Kurzzeitgedächtnis, mem0ai-Schnittstellengedächtnis, PowerMem intelligentes Gedächtnis, mit Gedächtniszusammenfassungsfunktion |
|
||||
| Wissensdatenbank | Unterstützt RAGFlow-Wissensdatenbank, ermöglicht großem Modell die Bewertung, ob Wissensdatenbank benötigt wird, bevor geantwortet wird |
|
||||
| Werkzeugaufruf | Unterstützt Client-IOT-Protokoll, Client-MCP-Protokoll, Server-MCP-Protokoll, MCP-Endpunktprotokoll, benutzerdefinierte Werkzeugfunktionen |
|
||||
| Befehlsübermittlung | Basierend auf MQTT-Protokoll, unterstützt die Übermittlung von MCP-Befehlen von der intelligenten Steuerkonsole an ESP32-Geräte |
|
||||
@@ -328,6 +328,7 @@ Tatsächlich kann jedes VLLM, das OpenAI-Schnittstellenaufrufe unterstützt, int
|
||||
| Typ | Plattformname | Verwendungsmethode | Preismodell | Hinweise |
|
||||
|:------:|:---------------:|:----:|:---------:|:--:|
|
||||
| Memory | mem0ai | Schnittstellenaufrufe | 1000 Mal/Monat Kontingent | |
|
||||
| Memory | [powermem](./docs/powermem-integration.md) | Lokale Zusammenfassung | Abhängig von LLM und DB | OceanBase Open Source, unterstützt intelligente Abfrage |
|
||||
| Memory | mem_local_short | Lokale Zusammenfassung | Kostenlos | |
|
||||
| Memory | nomem | Kein Gedächtnismodus | Kostenlos | |
|
||||
|
||||
|
||||
+3
-1
@@ -238,7 +238,7 @@ This project provides the following testing tools to help you verify the system
|
||||
| Intelligent Dialogue | Supports multiple LLM(large language models), implements intelligent dialogue |
|
||||
| Visual Perception | Supports multiple VLLM(vision large models), implements multimodal interaction |
|
||||
| Intent Recognition | Supports LLM intent recognition, Function Call function calling, provides plugin-based intent processing mechanism |
|
||||
| Memory System | Supports local short-term memory, mem0ai interface memory, with memory summarization functionality |
|
||||
| Memory System | Supports local short-term memory, mem0ai interface memory, PowerMem intelligent memory, with memory summarization functionality |
|
||||
| Knowledge Base | Supports RAGFlow knowledge base, enabling LLM to judge whether to schedule the knowledge base after receiving the user's question, and then answer the question |
|
||||
| Command Delivery | Supports MCP command delivery to ESP32 devices via MQTT protocol from Smart Console |
|
||||
| Tool Calling | Supports client IOT protocol, client MCP protocol, server MCP protocol, MCP endpoint protocol, custom tool functions |
|
||||
@@ -330,7 +330,9 @@ In fact, any VLLM that supports OpenAI interface calls can be integrated and use
|
||||
| Type | Platform Name | Usage Method | Pricing Model | Notes |
|
||||
|:------:|:---------------:|:----:|:---------:|:--:|
|
||||
| Memory | mem0ai | Interface calls | 1000 times/month quota | |
|
||||
| Memory | [powermem](./docs/powermem-integration.md) | Local summarization | Depends on LLM and DB | OceanBase open source, supports intelligent retrieval |
|
||||
| Memory | mem_local_short | Local summarization | Free | |
|
||||
| Memory | nomem | No memory mode | Free | |
|
||||
|
||||
---
|
||||
|
||||
|
||||
+2
-1
@@ -241,7 +241,7 @@ Dự án này cung cấp các công cụ kiểm tra sau để giúp bạn xác m
|
||||
| Đối thoại thông minh | Hỗ trợ nhiều LLM(Mô hình ngôn ngữ lớn), thực hiện đối thoại thông minh |
|
||||
| Cảm nhận thị giác | Hỗ trợ nhiều VLLM(Mô hình lớn thị giác), thực hiện tương tác đa phương thức |
|
||||
| Nhận dạng ý định | Hỗ trợ nhận dạng ý định mô hình lớn gắn ngoài, gọi hàm tự chủ mô hình lớn, cung cấp cơ chế xử lý ý định dạng plugin |
|
||||
| Hệ thống bộ nhớ | Hỗ trợ bộ nhớ ngắn hạn cục bộ, bộ nhớ giao diện mem0ai, có chức năng tóm tắt bộ nhớ |
|
||||
| Hệ thống bộ nhớ | Hỗ trợ bộ nhớ ngắn hạn cục bộ, bộ nhớ giao diện mem0ai, bộ nhớ thông minh PowerMem, có chức năng tóm tắt bộ nhớ |
|
||||
| Kho tri thức | Hỗ trợ kho tri thức RAGFlow, cho phép mô hình lớn đánh giá cần gọi kho tri thức trước khi trả lời |
|
||||
| Gọi công cụ | Hỗ trợ giao thức IOT phía client, giao thức MCP phía client, giao thức MCP phía server, giao thức điểm truy cập MCP, hàm công cụ tùy chỉnh |
|
||||
| Gửi lệnh | Dựa vào giao thức MQTT, hỗ trợ gửi lệnh MCP từ bảng điều khiển thông minh xuống thiết bị ESP32 |
|
||||
@@ -329,6 +329,7 @@ Trên thực tế, bất kỳ VLLM nào hỗ trợ gọi giao diện openai đ
|
||||
| Loại | Tên nền tảng | Phương pháp sử dụng | Mô hình thu phí | Ghi chú |
|
||||
|:------:|:---------------:|:----:|:---------:|:--:|
|
||||
| Memory | mem0ai | Gọi giao diện | Hạn mức 1000 lần/tháng | |
|
||||
| Memory | [powermem](./docs/powermem-integration.md) | Tóm tắt cục bộ | Phụ thuộc vào LLM và DB | OceanBase mã nguồn mở, hỗ trợ tìm kiếm thông minh |
|
||||
| Memory | mem_local_short | Tóm tắt cục bộ | Miễn phí | |
|
||||
| Memory | nomem | Chế độ không có bộ nhớ | Miễn phí | |
|
||||
|
||||
|
||||
@@ -82,6 +82,7 @@ VAD:
|
||||
8、[新闻插件源配置指南](./newsnow_plugin_config.md)<br/>
|
||||
9、[知识库ragflow集成指南](./ragflow-integration.md)<br/>
|
||||
10、[如何部署上下文源](./context-provider-integration.md)<br/>
|
||||
11、[如何集成PowerMem智能记忆](./powermem-integration.md)<br/>
|
||||
|
||||
### 11、语音克隆、本地语音部署相关教程
|
||||
1、[如何在智控台克隆音色](./huoshan-streamTTS-voice-cloning.md)<br/>
|
||||
|
||||
@@ -0,0 +1,345 @@
|
||||
# PowerMem 记忆组件集成指南
|
||||
|
||||
## 简介
|
||||
|
||||
[PowerMem](https://www.powermem.ai/) 是由 OceanBase 开源的 Agent 记忆组件，通过本地 LLM 进行记忆总结和智能检索，为 AI 代理提供高效的记忆管理功能。
|
||||
|
||||
费用说明：PowerMem 本身开源免费，实际费用取决于您选择的 LLM 和数据库：
|
||||
- 使用 SQLite + 免费 LLM（如智谱 glm-4-flash）= **完全免费**
|
||||
- 使用云端 LLM 或云端数据库 = 按对应服务收费
|
||||
|
||||
> 💡 **最佳性能提示**：PowerMem 配合 OceanBase 使用可实现最大性能释放，SQLite 仅建议在资源不足的情况下使用。
|
||||
|
||||
- **GitHub**: https://github.com/oceanbase/powermem
|
||||
- **官网**: https://www.powermem.ai/
|
||||
- **使用示例**: https://github.com/oceanbase/powermem/tree/main/examples
|
||||
|
||||
## 功能特性
|
||||
|
||||
- **本地总结**：通过 LLM 在本地进行记忆总结和提取
|
||||
- **用户画像**：通过 `UserMemory` 自动提取用户信息（姓名、职业、兴趣等），持续更新用户画像
|
||||
- **智能遗忘**：基于艾宾浩斯遗忘曲线，自动"遗忘"过时噪声信息
|
||||
- **多种存储后端**：支持 OceanBase（推荐，最佳性能）、SeekDB（推荐，AI应用存储一体）、PostgreSQL、SQLite（轻量备选）
|
||||
- **多种 LLM 支持**：通义千问、智谱（glm-4-flash 免费）、OpenAI 等
|
||||
- **智能检索**：基于向量搜索的语义检索能力
|
||||
- **私有部署**：完全支持本地私有化部署
|
||||
- **异步操作**：高效的异步记忆管理
|
||||
|
||||
## 安装
|
||||
|
||||
PowerMem 已添加到项目依赖中，如果需要手动安装：
|
||||
|
||||
```bash
|
||||
pip install powermem
|
||||
```
|
||||
|
||||
## 配置说明
|
||||
|
||||
### 基础配置
|
||||
|
||||
在 `config.yaml` 中配置 PowerMem：
|
||||
|
||||
```yaml
|
||||
selected_module:
|
||||
Memory: powermem
|
||||
|
||||
Memory:
|
||||
powermem:
|
||||
type: powermem
|
||||
# 是否启用用户画像功能
|
||||
# 用户画像支持: oceanbase、seekdb、sqlite (powermem 0.3.0+)
|
||||
enable_user_profile: true
|
||||
|
||||
# ========== LLM 配置 ==========
|
||||
llm:
|
||||
provider: openai # 可选: qwen, openai, zhipu 等
|
||||
config:
|
||||
api_key: 你的LLM API密钥
|
||||
model: qwen-plus
|
||||
# openai_base_url: https://api.openai.com/v1 # 可选，自定义服务地址
|
||||
|
||||
# ========== Embedding 配置 ==========
|
||||
embedder:
|
||||
provider: openai # 可选: qwen, openai 等
|
||||
config:
|
||||
api_key: 你的嵌入模型API密钥
|
||||
model: text-embedding-v4
|
||||
openai_base_url: https://dashscope.aliyuncs.com/compatible-mode/v1
|
||||
# embedding_dims: 1024 # 向量维度，非1536时需配置
|
||||
|
||||
# ========== Database 配置 ==========
|
||||
vector_store:
|
||||
provider: sqlite # 可选: oceanbase(推荐), seekdb(推荐), postgres, sqlite(轻量)
|
||||
config: {} # SQLite 无需额外配置
|
||||
```
|
||||
|
||||
### 配置参数详解
|
||||
|
||||
#### LLM 配置
|
||||
|
||||
| 参数 | 说明 | 可选值 |
|
||||
|------|------|--------|
|
||||
| `llm.provider` | LLM 提供商 | `qwen`, `openai`, `zhipu` 等 |
|
||||
| `llm.config.api_key` | API 密钥 | - |
|
||||
| `llm.config.model` | 模型名称 | 根据提供商选择 |
|
||||
| `llm.config.openai_base_url` | 自定义服务地址（可选） | - |
|
||||
|
||||
#### Embedding 配置
|
||||
|
||||
| 参数 | 说明 | 可选值 |
|
||||
|------|------|--------|
|
||||
| `embedder.provider` | 嵌入模型提供商 | `qwen`, `openai` 等 |
|
||||
| `embedder.config.api_key` | API 密钥 | - |
|
||||
| `embedder.config.model` | 模型名称 | 根据提供商选择 |
|
||||
| `embedder.config.openai_base_url` | 自定义服务地址（可选） | - |
|
||||
|
||||
#### Database 配置
|
||||
|
||||
| 参数 | 说明 | 可选值 |
|
||||
|------|------|--------|
|
||||
| `vector_store.provider` | 存储后端类型 | `oceanbase`(推荐), `seekdb`(推荐), `postgres`, `sqlite`(轻量) |
|
||||
| `vector_store.config` | 数据库连接配置 | 根据 provider 设置 |
|
||||
|
||||
### 记忆模式说明
|
||||
|
||||
PowerMem 支持两种记忆模式：
|
||||
|
||||
| 模式 | 配置 | 功能 | 存储要求 |
|
||||
|------|------|------|----------|
|
||||
| **普通记忆** | `enable_user_profile: false` | 对话记忆存储与检索 | 支持所有数据库 |
|
||||
| **用户画像** | `enable_user_profile: true` | 记忆 + 自动提取用户画像 | oceanbase、seekdb、sqlite |
|
||||
|
||||
> 📌 **版本说明**：PowerMem 0.3.0+ 版本，用户画像功能支持 OceanBase、SeekDB、SQLite 三种存储后端。
|
||||
|
||||
### 使用通义千问（推荐）
|
||||
|
||||
1. 访问 [阿里云百炼平台](https://bailian.console.aliyun.com/) 注册账号
|
||||
2. 在 [API Key 管理](https://bailian.console.aliyun.com/?apiKey=1#/api-key) 页面获取 API 密钥
|
||||
3. 配置如下：
|
||||
|
||||
```yaml
|
||||
Memory:
|
||||
powermem:
|
||||
type: powermem
|
||||
enable_user_profile: true
|
||||
llm:
|
||||
provider: qwen
|
||||
config:
|
||||
api_key: sk-xxxxxxxxxxxxxxxx
|
||||
model: qwen-plus
|
||||
embedder:
|
||||
provider: openai
|
||||
config:
|
||||
api_key: sk-xxxxxxxxxxxxxxxx
|
||||
model: text-embedding-v4
|
||||
openai_base_url: https://dashscope.aliyuncs.com/compatible-mode/v1
|
||||
vector_store:
|
||||
provider: sqlite
|
||||
config: {}
|
||||
```
|
||||
|
||||
### 使用智谱免费 LLM（完全免费方案）
|
||||
|
||||
智谱提供免费的 glm-4-flash 模型，配合 SQLite 可实现完全免费使用：
|
||||
|
||||
1. 访问 [智谱AI开放平台](https://bigmodel.cn/) 注册账号
|
||||
2. 在 [API Keys](https://bigmodel.cn/usercenter/proj-mgmt/apikeys) 页面获取 API 密钥
|
||||
3. 配置如下：
|
||||
|
||||
```yaml
|
||||
Memory:
|
||||
powermem:
|
||||
type: powermem
|
||||
enable_user_profile: true
|
||||
llm:
|
||||
provider: openai # 使用 openai 兼容模式
|
||||
config:
|
||||
api_key: xxxxxxxxxxxxxxxx.xxxxxxxxxxxxxxxx
|
||||
model: glm-4-flash
|
||||
openai_base_url: https://open.bigmodel.cn/api/paas/v4/
|
||||
embedder:
|
||||
provider: openai
|
||||
config:
|
||||
api_key: xxxxxxxxxxxxxxxx.xxxxxxxxxxxxxxxx
|
||||
model: embedding-3
|
||||
openai_base_url: https://open.bigmodel.cn/api/paas/v4/
|
||||
vector_store:
|
||||
provider: sqlite
|
||||
config: {}
|
||||
```
|
||||
|
||||
### 使用 OpenAI
|
||||
|
||||
```yaml
|
||||
Memory:
|
||||
powermem:
|
||||
type: powermem
|
||||
enable_user_profile: true
|
||||
llm:
|
||||
provider: openai
|
||||
config:
|
||||
api_key: sk-xxxxxxxxxxxxxxxx
|
||||
model: gpt-4o-mini
|
||||
openai_base_url: https://api.openai.com/v1
|
||||
embedder:
|
||||
provider: openai
|
||||
config:
|
||||
api_key: sk-xxxxxxxxxxxxxxxx
|
||||
model: text-embedding-3-small
|
||||
openai_base_url: https://api.openai.com/v1
|
||||
vector_store:
|
||||
provider: sqlite
|
||||
config: {}
|
||||
```
|
||||
|
||||
### 使用 OceanBase（最佳性能方案）
|
||||
|
||||
OceanBase 是 PowerMem 的最佳搭档，可实现最大性能释放：
|
||||
|
||||
1. 部署 OceanBase 数据库（支持开源本地部署或使用云服务）
|
||||
- 开源部署：https://github.com/oceanbase/oceanbase
|
||||
- 云服务：https://www.oceanbase.com/
|
||||
2. 配置如下：
|
||||
|
||||
```yaml
|
||||
Memory:
|
||||
powermem:
|
||||
type: powermem
|
||||
enable_user_profile: true
|
||||
llm:
|
||||
provider: qwen
|
||||
config:
|
||||
api_key: sk-xxxxxxxxxxxxxxxx
|
||||
model: qwen-plus
|
||||
embedder:
|
||||
provider: openai
|
||||
config:
|
||||
api_key: sk-xxxxxxxxxxxxxxxx
|
||||
model: text-embedding-v4
|
||||
openai_base_url: https://dashscope.aliyuncs.com/compatible-mode/v1
|
||||
vector_store:
|
||||
provider: oceanbase
|
||||
config:
|
||||
host: 127.0.0.1
|
||||
port: 2881
|
||||
user: root@test
|
||||
password: your_password
|
||||
db_name: powermem
|
||||
collection_name: memories # 默认值
|
||||
embedding_model_dims: 1536 # 嵌入向量维度，必需参数
|
||||
```
|
||||
|
||||
## 设备记忆隔离
|
||||
|
||||
PowerMem 会自动使用设备 ID（`device_id`）作为 `user_id` 进行记忆隔离。这意味着：
|
||||
|
||||
- 每个设备拥有独立的记忆空间
|
||||
- 不同设备之间的记忆完全隔离
|
||||
- 同一设备的多次对话可以共享记忆上下文
|
||||
|
||||
## 用户画像（UserMemory）
|
||||
|
||||
PowerMem 提供 `UserMemory` 类，可自动从对话中提取用户画像信息。
|
||||
|
||||
> 📌 **版本说明**：PowerMem 0.3.0+ 版本，用户画像功能支持 OceanBase、SeekDB、SQLite 三种存储后端。
|
||||
|
||||
### 启用用户画像
|
||||
|
||||
在配置中设置 `enable_user_profile: true` 即可启用：
|
||||
|
||||
```yaml
|
||||
Memory:
|
||||
powermem:
|
||||
type: powermem
|
||||
enable_user_profile: true # 启用用户画像
|
||||
llm:
|
||||
provider: qwen
|
||||
config:
|
||||
api_key: sk-xxxxxxxxxxxxxxxx
|
||||
model: qwen-plus
|
||||
embedder:
|
||||
provider: openai
|
||||
config:
|
||||
api_key: sk-xxxxxxxxxxxxxxxx
|
||||
model: text-embedding-v4
|
||||
openai_base_url: https://dashscope.aliyuncs.com/compatible-mode/v1
|
||||
vector_store:
|
||||
provider: sqlite # 用户画像支持: oceanbase、seekdb、sqlite
|
||||
config: {}
|
||||
```
|
||||
|
||||
### 用户画像能力
|
||||
|
||||
| 能力 | 说明 |
|
||||
|------|------|
|
||||
| **信息提取** | 自动从对话中提取姓名、年龄、职业、兴趣等 |
|
||||
| **持续更新** | 随着对话进行，不断完善用户画像 |
|
||||
| **画像检索** | 将用户画像与记忆搜索结合，提升检索相关性 |
|
||||
| **智能遗忘** | 基于艾宾浩斯遗忘曲线，淡化过时信息 |
|
||||
|
||||
### 工作原理
|
||||
|
||||
启用用户画像后，小智在查询记忆时会自动返回：
|
||||
1. **用户画像**：用户的基本信息、兴趣爱好等
|
||||
2. **相关记忆**：与当前对话相关的历史记忆
|
||||
|
||||
> ✅ **版本说明**：PowerMem 0.3.0+ 版本，用户画像功能支持 OceanBase、SeekDB、SQLite 三种存储后端。
|
||||
|
||||
## 与其他记忆组件的对比
|
||||
|
||||
| 特性 | PowerMem | mem0ai | mem_local_short |
|
||||
|------|----------|--------|-----------------|
|
||||
| 工作方式 | 本地总结 | 云端接口 | 本地总结 |
|
||||
| 存储位置 | 本地/云端DB | 云端 | 本地YAML |
|
||||
| 费用 | 取决于LLM和DB | 1000次/月免费 | 完全免费 |
|
||||
| 智能检索 | ✅ 向量搜索 | ✅ 向量搜索 | ❌ 全量返回 |
|
||||
| 用户画像 | ✅ UserMemory | ❌ | ❌ |
|
||||
| 智能遗忘 | ✅ 遗忘曲线 | ❌ | ❌ |
|
||||
| 私有部署 | ✅ 支持 | ❌ 仅云端 | ✅ 支持 |
|
||||
| 数据库支持 | OceanBase(推荐)/SeekDB/PostgreSQL/SQLite | - | YAML 文件 |
|
||||
|
||||
## 常见问题
|
||||
|
||||
### 1. API 密钥错误
|
||||
|
||||
如果出现 `API key is required` 错误，请检查：
|
||||
- `llm_api_key` 和 `embedding_api_key` 是否正确填写
|
||||
- API 密钥是否有效
|
||||
|
||||
### 2. 模型不存在
|
||||
|
||||
如果出现模型不存在的错误，请确认：
|
||||
- `llm_model` 和 `embedding_model` 名称是否正确
|
||||
- 对应的模型服务是否已开通
|
||||
|
||||
### 3. 连接超时
|
||||
|
||||
如果出现连接超时，可以尝试：
|
||||
- 检查网络连接
|
||||
- 如果使用代理，配置 `llm_base_url` 和 `embedding_base_url`
|
||||
|
||||
## 测试验证
|
||||
|
||||
可以在虚拟环境中测试 PowerMem 是否正常工作：
|
||||
|
||||
```bash
|
||||
# 激活虚拟环境
|
||||
source .venv/bin/activate
|
||||
|
||||
# 测试 PowerMem 导入
|
||||
python -c "from powermem import AsyncMemory; print('PowerMem 导入成功')"
|
||||
|
||||
# 测试 UserMemory 导入（用户画像功能）
|
||||
python -c "from powermem import UserMemory; print('UserMemory 导入成功')"
|
||||
```
|
||||
|
||||
## 更多资源
|
||||
|
||||
- [PowerMem 官方文档](https://www.powermem.ai/)
|
||||
- [PowerMem GitHub 仓库](https://github.com/oceanbase/powermem)
|
||||
- [PowerMem 使用示例](https://github.com/oceanbase/powermem/tree/main/examples)
|
||||
- [OceanBase 官网](https://www.oceanbase.com/)
|
||||
- [OceanBase GitHub](https://github.com/oceanbase/oceanbase)
|
||||
- [SeekDB GitHub](https://github.com/oceanbase/seekdb)（AI原生搜索数据库）
|
||||
- [阿里云百炼平台](https://bailian.console.aliyun.com/)
|
||||
|
||||
@@ -0,0 +1,115 @@
|
||||
-- 新增powermem记忆模型供应器
|
||||
INSERT INTO `ai_model_provider` (`id`, `model_type`, `provider_code`, `name`, `fields`, `sort`, `creator`, `create_date`, `updater`, `update_date`)
|
||||
VALUES ('SYSTEM_Memory_powermem', 'Memory', 'powermem', 'PowerMem记忆', '[
|
||||
{"key":"enable_user_profile","label":"启用用户画像","type":"boolean"},
|
||||
{"key":"llm_provider","label":"LLM提供商","type":"string"},
|
||||
{"key":"llm_api_key","label":"LLM API密钥","type":"string"},
|
||||
{"key":"llm_model","label":"LLM模型","type":"string"},
|
||||
{"key":"openai_base_url","label":"OpenAI基础URL","type":"string"},
|
||||
{"key":"embedding_provider","label":"Embedding提供商","type":"string"},
|
||||
{"key":"embedding_api_key","label":"Embedding API密钥","type":"string"},
|
||||
{"key":"embedding_model","label":"Embedding模型","type":"string"},
|
||||
{"key":"embedding_openai_base_url","label":"Embedding OpenAI基础URL","type":"string"},
|
||||
{"key":"embedding_dims","label":"Embedding维度","type":"integer"},
|
||||
{"key":"vector_store","label":"向量存储配置(JSON)","type":"dict"}
|
||||
]', 4, 1, NOW(), 1, NOW());
|
||||
|
||||
-- 新增PowerMem记忆模型配置
|
||||
INSERT INTO `ai_model_config` VALUES (
|
||||
'Memory_powermem',
|
||||
'Memory',
|
||||
'powermem',
|
||||
'PowerMem记忆',
|
||||
0,
|
||||
1,
|
||||
'{\"type\": \"powermem\", \"enable_user_profile\": true, \"llm_provider\": \"openai\", \"llm_api_key\": \"你的LLM API密钥\", \"llm_model\": \"qwen-plus\", \"openai_base_url\": \"\", \"embedding_provider\": \"openai\", \"embedding_api_key\": \"你的嵌入模型API密钥\", \"embedding_model\": \"text-embedding-v4\", \"embedding_openai_base_url\": \"https://api.openai.com/v1\", \"embedding_dims\": \"\", \"vector_store\": {\"provider\": \"sqlite\", \"config\": {}}}',
|
||||
NULL,
|
||||
NULL,
|
||||
4,
|
||||
NULL,
|
||||
NULL,
|
||||
NULL,
|
||||
NULL
|
||||
);
|
||||
|
||||
|
||||
-- PowerMem记忆配置说明
|
||||
UPDATE `ai_model_config` SET
|
||||
`doc_link` = 'https://github.com/oceanbase/powermem',
|
||||
`remark` = 'PowerMem是OceanBase开源的agent记忆组件，通过本地LLM进行记忆总结
|
||||
GitHub: https://github.com/oceanbase/powermem
|
||||
官网: https://www.powermem.ai/
|
||||
使用示例: https://github.com/oceanbase/powermem/tree/main/examples
|
||||
|
||||
【费用说明】
|
||||
PowerMem本身免费，实际费用取决于所选LLM和数据库：
|
||||
- 使用sqlite + 免费LLM(如glm-4-flash) = 完全免费
|
||||
- 使用云端LLM或云端数据库 = 按对应服务收费
|
||||
|
||||
【enable_user_profile】用户画像功能
|
||||
- false: 使用普通记忆模式(AsyncMemory)
|
||||
- true: 使用用户画像模式(UserMemory)，自动提取用户信息
|
||||
- 用户画像功能支持: oceanbase、seekdb、sqlite (powermem 0.3.0+)
|
||||
|
||||
【llm】LLM配置 - 用于记忆总结和用户画像提取
|
||||
provider: LLM提供商，可选值：
|
||||
- qwen: 通义千问 (https://bailian.console.aliyun.com/?apiKey=1#/api-key)
|
||||
- openai: OpenAI兼容接口
|
||||
- zhipu: 智谱AI (https://bigmodel.cn/usercenter/proj-mgmt/apikeys) - 推荐使用免费的glm-4-flash
|
||||
config: LLM配置参数
|
||||
- api_key: API密钥 (必填)
|
||||
- model: 模型名称，如 qwen-plus、glm-4-flash 等
|
||||
- openai_base_url: 自定义服务地址 (可选)，如 https://api.openai.com/v1
|
||||
示例：
|
||||
{"provider": "zhipu", "config": {"api_key": "your_key", "model": "glm-4-flash"}}
|
||||
{"provider": "qwen", "config": {"api_key": "your_key", "model": "qwen-plus"}}
|
||||
|
||||
【embedder】Embedding配置 - 用于向量化记忆内容
|
||||
provider: 嵌入模型提供商，可选值：
|
||||
- qwen: 通义千问
|
||||
- openai: OpenAI兼容接口
|
||||
config: Embedding配置参数
|
||||
- api_key: API密钥 (必填)
|
||||
- model: 模型名称，如 text-embedding-v4、text-embedding-3-small 等
|
||||
- openai_base_url: 自定义服务地址 (可选)
|
||||
- embedding_dims: 向量维度 (可选)，非1536时需配置
|
||||
示例：
|
||||
{"provider": "openai", "config": {"api_key": "your_key", "model": "text-embedding-v4", "openai_base_url": "https://dashscope.aliyuncs.com/compatible-mode/v1"}}
|
||||
|
||||
【vector_store】数据库存储配置 - 用于存储向量化的记忆
|
||||
provider: 数据库类型，可选值：
|
||||
- sqlite: 轻量级本地数据库 (推荐入门使用，无需额外配置)
|
||||
- oceanbase: OceanBase数据库 (推荐生产使用，最佳性能)
|
||||
- seekdb: SeekDB (推荐，AI应用存储一体)
|
||||
- postgres: PostgreSQL数据库
|
||||
|
||||
SQLite配置 (无需额外配置):
|
||||
{"provider": "sqlite", "config": {}}
|
||||
|
||||
OceanBase配置示例:
|
||||
{"provider": "oceanbase", "config": {
|
||||
"host": "127.0.0.1",
|
||||
"port": 2881,
|
||||
"user": "root@test",
|
||||
"password": "your_password",
|
||||
"db_name": "powermem",
|
||||
"collection_name": "memories",
|
||||
"embedding_model_dims": 1024
|
||||
}}
|
||||
注意：
|
||||
- collection_name: 默认表名，如创建维度错误请删除此表或更改名称
|
||||
- embedding_model_dims: 嵌入向量维度，需与embedder的模型维度匹配
|
||||
例如智谱：embedding-2维度是1024，embedding-3维度是2048
|
||||
|
||||
【推荐配置组合】
|
||||
1. 完全免费方案：
|
||||
- LLM: zhipu + glm-4-flash (免费)
|
||||
- Embedder: 通义千问 text-embedding-v4
|
||||
- Database: sqlite
|
||||
|
||||
2. 生产环境方案：
|
||||
- LLM: qwen-plus 或其他商业模型
|
||||
- Embedder: text-embedding-v4
|
||||
- Database: oceanbase 或 seekdb
|
||||
'
|
||||
WHERE `id` = 'Memory_powermem';
|
||||
@@ -515,6 +515,13 @@ databaseChangeLog:
|
||||
- sqlFile:
|
||||
encoding: utf8
|
||||
path: classpath:db/changelog/202602021555.sql
|
||||
- changeSet:
|
||||
id: 202602051017
|
||||
author: RanChen
|
||||
changes:
|
||||
- sqlFile:
|
||||
encoding: utf8
|
||||
path: classpath:db/changelog/202602051017.sql
|
||||
- changeSet:
|
||||
id: 202602051125
|
||||
author: DaGou12138
|
||||
|
||||
@@ -279,6 +279,54 @@ Memory:
|
||||
# https://app.mem0.ai/dashboard/api-keys
|
||||
# 每月有1000次免费调用
|
||||
api_key: 你的mem0ai api key
|
||||
powermem:
|
||||
# PowerMem是OceanBase开源的agent记忆组件，通过本地LLM进行记忆总结
|
||||
# 费用说明：PowerMem本身免费，实际费用取决于所选LLM和数据库
|
||||
# - 使用sqlite + 免费LLM(如glm-4-flash) = 完全免费
|
||||
# - 使用云端LLM或云端数据库 = 按对应服务收费
|
||||
# GitHub: https://github.com/oceanbase/powermem
|
||||
# 官网: https://www.powermem.ai/
|
||||
# 使用示例: https://github.com/oceanbase/powermem/tree/main/examples
|
||||
type: powermem
|
||||
# 是否启用用户画像功能
|
||||
# - false: 使用普通记忆模式(AsyncMemory)
|
||||
# - true: 使用用户画像模式(UserMemory)，自动提取用户信息
|
||||
# 用户画像功能支持: oceanbase、seekdb、sqlite (powermem 0.3.0+)
|
||||
enable_user_profile: true
|
||||
# ========== LLM 配置 ==========
|
||||
# 通义千问: https://bailian.console.aliyun.com/?apiKey=1#/api-key
|
||||
# 智谱AI(免费): https://bigmodel.cn/usercenter/proj-mgmt/apikeys
|
||||
llm:
|
||||
provider: openai # 可选: qwen, openai, zhipu 等
|
||||
config:
|
||||
api_key: 你的LLM API密钥
|
||||
model: qwen-plus
|
||||
# openai_base_url: https://api.openai.com/v1 # 可选，自定义服务地址
|
||||
# ========== Embedding 配置 ==========
|
||||
embedder:
|
||||
provider: openai # 可选: qwen, openai 等
|
||||
config:
|
||||
api_key: 你的嵌入模型API密钥
|
||||
model: text-embedding-v4
|
||||
openai_base_url: https://dashscope.aliyuncs.com/compatible-mode/v1
|
||||
# embedding_dims: 1024 # 向量维度，非1536时需配置
|
||||
|
||||
# ========== Database 配置 ==========
|
||||
# oceanbase(推荐,最佳性能), seekdb(推荐,AI应用存储一体), postgres, sqlite(轻量备选)
|
||||
# 用户画像功能支持: oceanbase、seekdb、sqlite
|
||||
vector_store:
|
||||
provider: sqlite
|
||||
config: {} # SQLite 无需额外配置
|
||||
# OceanBase 配置示例:
|
||||
# provider: oceanbase
|
||||
# config:
|
||||
# host: 127.0.0.1
|
||||
# port: 2881
|
||||
# user: root@test
|
||||
# password: your_password
|
||||
# db_name: powermem
|
||||
# collection_name: memories # 默认表名，如创建维度错误请删除此表或更改名称
|
||||
# embedding_model_dims: 1024 # 嵌入向量维度，注意跟使用模型适配，以智普为例：embedding-2的维度是1024，embedding-3的维度是2048
|
||||
nomem:
|
||||
# 不想使用记忆功能，可以使用nomem
|
||||
type: nomem
|
||||
|
||||
@@ -841,7 +841,8 @@ class ConnectionHandler:
|
||||
try:
|
||||
# 使用带记忆的对话
|
||||
memory_str = None
|
||||
if self.memory is not None:
|
||||
# 仅当query非空（代表用户询问）时查询记忆
|
||||
if self.memory is not None and query:
|
||||
future = asyncio.run_coroutine_threadsafe(
|
||||
self.memory.query_memory(query), self.loop
|
||||
)
|
||||
|
||||
@@ -1,3 +1,4 @@
|
||||
import json
|
||||
import traceback
|
||||
|
||||
from ..base import MemoryProviderBase, logger
|
||||
@@ -36,11 +37,26 @@ class MemoryProvider(MemoryProviderBase):
|
||||
|
||||
try:
|
||||
# Format the content as a message list for mem0
|
||||
messages = [
|
||||
{"role": message.role, "content": message.content}
|
||||
for message in msgs
|
||||
if message.role != "system"
|
||||
]
|
||||
messages = []
|
||||
for message in msgs:
|
||||
if message.role == "system":
|
||||
continue
|
||||
|
||||
content = message.content
|
||||
|
||||
# Extract content from JSON format if present (for ASR with emotion/language tags)
|
||||
# Same logic as in query_memory method
|
||||
try:
|
||||
if content and content.strip().startswith("{") and content.strip().endswith("}"):
|
||||
data = json.loads(content)
|
||||
if "content" in data:
|
||||
content = data["content"]
|
||||
except (json.JSONDecodeError, KeyError, TypeError):
|
||||
# If parsing fails, use original content
|
||||
pass
|
||||
|
||||
messages.append({"role": message.role, "content": content})
|
||||
|
||||
result = self.client.add(messages, user_id=self.role_id)
|
||||
logger.bind(tag=TAG).debug(f"Save memory result: {result}")
|
||||
except Exception as e:
|
||||
@@ -56,7 +72,16 @@ class MemoryProvider(MemoryProviderBase):
|
||||
|
||||
filters = {"user_id": self.role_id}
|
||||
|
||||
results = self.client.search(query, filters=filters)
|
||||
search_query = query
|
||||
try:
|
||||
if query.strip().startswith("{") and query.strip().endswith("}"):
|
||||
data = json.loads(query)
|
||||
if "content" in data:
|
||||
search_query = data["content"]
|
||||
except (json.JSONDecodeError, KeyError):
|
||||
pass
|
||||
|
||||
results = self.client.search(search_query, filters=filters)
|
||||
if not results or "results" not in results:
|
||||
return ""
|
||||
|
||||
|
||||
@@ -149,10 +149,22 @@ class MemoryProvider(MemoryProviderBase):
|
||||
|
||||
msgStr = ""
|
||||
for msg in msgs:
|
||||
content = msg.content
|
||||
|
||||
# Extract content from JSON format if present (for ASR with emotion/language tags)
|
||||
try:
|
||||
if content and content.strip().startswith("{") and content.strip().endswith("}"):
|
||||
data = json.loads(content)
|
||||
if "content" in data:
|
||||
content = data["content"]
|
||||
except (json.JSONDecodeError, KeyError, TypeError):
|
||||
# If parsing fails, use original content
|
||||
pass
|
||||
|
||||
if msg.role == "user":
|
||||
msgStr += f"User: {msg.content}\n"
|
||||
msgStr += f"User: {content}\n"
|
||||
elif msg.role == "assistant":
|
||||
msgStr += f"Assistant: {msg.content}\n"
|
||||
msgStr += f"Assistant: {content}\n"
|
||||
if self.short_memory and len(self.short_memory) > 0:
|
||||
msgStr += "历史记忆：\n"
|
||||
msgStr += self.short_memory
|
||||
|
||||
@@ -0,0 +1,346 @@
|
||||
#!/usr/bin/env python
|
||||
# -*- coding: UTF-8 -*-
|
||||
"""
|
||||
@time: 2026/01/08
|
||||
@file: powermem.py
|
||||
@desc: PowerMem memory provider for xiaozhi-esp32-server
|
||||
PowerMem is an open-source agent memory component from OceanBase
|
||||
GitHub: https://github.com/oceanbase/powermem
|
||||
Website: https://www.powermem.ai/
|
||||
@Author: wayyoungboy
|
||||
"""
|
||||
|
||||
import asyncio
|
||||
import json
|
||||
import traceback
|
||||
from typing import Optional, Dict, Any
|
||||
|
||||
from ..base import MemoryProviderBase, logger
|
||||
|
||||
TAG = __name__
|
||||
|
||||
|
||||
class MemoryProvider(MemoryProviderBase):
|
||||
"""
|
||||
PowerMem memory provider implementation.
|
||||
|
||||
PowerMem is an open-source agent memory component that provides
|
||||
efficient memory management for AI agents.
|
||||
|
||||
Supports multiple storage backends (sqlite, oceanbase, postgres),
|
||||
LLM providers (qwen, openai, etc.) and embedding providers.
|
||||
|
||||
Config options:
|
||||
- enable_user_profile: bool - Enable UserMemory for user profiling (requires OceanBase)
|
||||
- database_provider: str - Storage backend (sqlite, oceanbase, postgres)
|
||||
- llm_provider: str - LLM provider (qwen, openai, etc.)
|
||||
- embedding_provider: str - Embedding provider (qwen, openai, etc.)
|
||||
"""
|
||||
|
||||
def __init__(self, config: Dict[str, Any], summary_memory: Optional[str] = None):
|
||||
super().__init__(config)
|
||||
self.use_powermem = False
|
||||
self.memory_client = None
|
||||
self.enable_user_profile = False
|
||||
self.last_profile_content = "" # Cache for user profile from UserMemory
|
||||
|
||||
try:
|
||||
# Check if user profile mode is enabled
|
||||
self.enable_user_profile = config.get("enable_user_profile", False)
|
||||
|
||||
# Get configuration parameters
|
||||
database_provider = config.get("database_provider", "sqlite")
|
||||
llm_provider = config.get("llm_provider", "qwen")
|
||||
embedding_provider = config.get("embedding_provider", "qwen")
|
||||
|
||||
# Build powermem configuration dict
|
||||
# PowerMem supports two config styles:
|
||||
# 1. powermem style: database, llm, embedding
|
||||
# 2. mem0 style: vector_store, llm, embedder
|
||||
powermem_config = {}
|
||||
|
||||
# Configure vector store / database
|
||||
if "vector_store" in config:
|
||||
powermem_config["vector_store"] = config["vector_store"]
|
||||
elif "database" in config:
|
||||
powermem_config["database"] = config["database"]
|
||||
else:
|
||||
powermem_config["vector_store"] = {
|
||||
"provider": database_provider,
|
||||
"config": {}
|
||||
}
|
||||
|
||||
# Configure LLM
|
||||
if "llm" in config:
|
||||
powermem_config["llm"] = config["llm"]
|
||||
else:
|
||||
llm_config = {}
|
||||
if "llm_api_key" in config:
|
||||
llm_config["api_key"] = config["llm_api_key"]
|
||||
if "llm_model" in config:
|
||||
llm_config["model"] = config["llm_model"]
|
||||
# Handle base_url based on provider type
|
||||
# - qwen provider uses dashscope_base_url
|
||||
# - openai provider uses openai_base_url
|
||||
if "llm_base_url" in config:
|
||||
if llm_provider == "qwen":
|
||||
llm_config["dashscope_base_url"] = config["llm_base_url"]
|
||||
else:
|
||||
llm_config["openai_base_url"] = config["llm_base_url"]
|
||||
if "openai_base_url" in config:
|
||||
llm_config["openai_base_url"] = config["openai_base_url"]
|
||||
if "dashscope_base_url" in config:
|
||||
llm_config["dashscope_base_url"] = config["dashscope_base_url"]
|
||||
powermem_config["llm"] = {
|
||||
"provider": llm_provider,
|
||||
"config": llm_config
|
||||
}
|
||||
|
||||
# Configure embedder
|
||||
if "embedder" in config:
|
||||
powermem_config["embedder"] = config["embedder"]
|
||||
else:
|
||||
embedder_config = {}
|
||||
if "embedding_api_key" in config:
|
||||
embedder_config["api_key"] = config["embedding_api_key"]
|
||||
if "embedding_model" in config:
|
||||
embedder_config["model"] = config["embedding_model"]
|
||||
# Handle base_url based on provider type
|
||||
# - qwen provider uses dashscope_base_url
|
||||
# - openai provider uses openai_base_url
|
||||
# Priority: embedding_xxx_base_url > embedding_base_url > xxx_base_url
|
||||
if "embedding_base_url" in config:
|
||||
if embedding_provider == "qwen":
|
||||
embedder_config["dashscope_base_url"] = config["embedding_base_url"]
|
||||
else:
|
||||
embedder_config["openai_base_url"] = config["embedding_base_url"]
|
||||
# Embedding-specific base_url (higher priority)
|
||||
if "embedding_openai_base_url" in config:
|
||||
embedder_config["openai_base_url"] = config["embedding_openai_base_url"]
|
||||
if "embedding_dashscope_base_url" in config:
|
||||
embedder_config["dashscope_base_url"] = config["embedding_dashscope_base_url"]
|
||||
powermem_config["embedder"] = {
|
||||
"provider": embedding_provider,
|
||||
"config": embedder_config
|
||||
}
|
||||
|
||||
# Initialize memory client based on mode
|
||||
if self.enable_user_profile:
|
||||
from powermem import UserMemory
|
||||
self.memory_client = UserMemory(config=powermem_config)
|
||||
memory_mode = "UserMemory (用户画像模式)"
|
||||
else:
|
||||
from powermem import AsyncMemory
|
||||
self.memory_client = AsyncMemory(config=powermem_config)
|
||||
memory_mode = "AsyncMemory (普通记忆模式)"
|
||||
|
||||
self.use_powermem = True
|
||||
|
||||
logger.bind(tag=TAG).info(
|
||||
f"PowerMem initialized successfully: mode={memory_mode}, "
|
||||
f"database={powermem_config['vector_store']['provider']}, llm={powermem_config['llm']['provider']}, embedding={powermem_config['embedder']['provider']}"
|
||||
)
|
||||
|
||||
except ImportError as e:
|
||||
logger.bind(tag=TAG).error(
|
||||
f"PowerMem not installed. Please install with: pip install powermem. Error: {e}"
|
||||
)
|
||||
self.use_powermem = False
|
||||
except Exception as e:
|
||||
logger.bind(tag=TAG).error(f"Failed to initialize PowerMem: {str(e)}")
|
||||
logger.bind(tag=TAG).debug(f"Detailed error: {traceback.format_exc()}")
|
||||
self.use_powermem = False
|
||||
|
||||
async def save_memory(self, msgs, session_id=None):
|
||||
"""
|
||||
Save conversation messages to PowerMem.
|
||||
|
||||
Args:
|
||||
msgs: List of message objects with 'role' and 'content' attributes
|
||||
|
||||
session_id: Session identifier (optional, for compatibility)
|
||||
|
||||
Returns:
|
||||
Result from PowerMem API or None if failed
|
||||
"""
|
||||
if not self.use_powermem or self.memory_client is None:
|
||||
logger.bind(tag=TAG).warning("PowerMem is not available, skipping save_memory")
|
||||
return None
|
||||
|
||||
if len(msgs) < 2:
|
||||
logger.bind(tag=TAG).debug("Not enough messages to save (need at least 2)")
|
||||
return None
|
||||
|
||||
try:
|
||||
# Format the content as a message list for PowerMem
|
||||
messages = []
|
||||
for message in msgs:
|
||||
if message.role == "system":
|
||||
continue
|
||||
|
||||
content = message.content
|
||||
|
||||
# Extract content from JSON format if present (for ASR with emotion/language tags)
|
||||
# Same logic as in query_memory method
|
||||
try:
|
||||
if content and content.strip().startswith("{") and content.strip().endswith("}"):
|
||||
data = json.loads(content)
|
||||
if "content" in data:
|
||||
content = data["content"]
|
||||
except (json.JSONDecodeError, KeyError, TypeError):
|
||||
# If parsing fails, use original content
|
||||
pass
|
||||
|
||||
messages.append({"role": message.role, "content": content})
|
||||
|
||||
# Add memory using PowerMem SDK
|
||||
result = self.memory_client.add(
|
||||
messages=messages,
|
||||
user_id=self.role_id
|
||||
)
|
||||
# Handle both sync and async returns
|
||||
if asyncio.iscoroutine(result):
|
||||
result = await result
|
||||
|
||||
logger.bind(tag=TAG).debug(f"Save memory result: {result}")
|
||||
|
||||
# Cache user profile if UserMemory mode and profile was extracted
|
||||
if self.enable_user_profile and result:
|
||||
if result.get('profile_extracted'):
|
||||
self.last_profile_content = result.get('profile_content', '')
|
||||
logger.bind(tag=TAG).debug(f"User profile extracted: {self.last_profile_content}")
|
||||
|
||||
return result
|
||||
|
||||
except Exception as e:
|
||||
logger.bind(tag=TAG).error(f"Error saving memory: {str(e)}")
|
||||
logger.bind(tag=TAG).debug(f"Detailed error: {traceback.format_exc()}")
|
||||
return None
|
||||
|
||||
async def query_memory(self, query: str) -> str:
|
||||
"""
|
||||
Query memories from PowerMem based on similarity search.
|
||||
|
||||
Args:
|
||||
query: The search query string (may be JSON format with metadata)
|
||||
|
||||
Returns:
|
||||
Formatted string of relevant memories or empty string if none found
|
||||
"""
|
||||
if not self.use_powermem or self.memory_client is None:
|
||||
logger.bind(tag=TAG).warning("PowerMem is not available, skipping query_memory")
|
||||
return ""
|
||||
|
||||
try:
|
||||
if not getattr(self, "role_id", None):
|
||||
logger.bind(tag=TAG).debug("No role_id set, returning empty memory")
|
||||
return ""
|
||||
|
||||
# Extract content from JSON format if present (for ASR with emotion/language tags)
|
||||
search_query = query
|
||||
try:
|
||||
if query.strip().startswith("{") and query.strip().endswith("}"):
|
||||
data = json.loads(query)
|
||||
if "content" in data:
|
||||
search_query = data["content"]
|
||||
except (json.JSONDecodeError, KeyError):
|
||||
# If parsing fails, use original query
|
||||
pass
|
||||
|
||||
result_parts = []
|
||||
|
||||
# If user profile mode is enabled, include user profile in results
|
||||
if self.enable_user_profile:
|
||||
profile = await self.get_user_profile()
|
||||
if profile:
|
||||
result_parts.append(f"【用户画像】\n{profile}")
|
||||
|
||||
# Search memories using PowerMem SDK
|
||||
if self.enable_user_profile:
|
||||
# UserMemory uses sync search
|
||||
results = await asyncio.to_thread(
|
||||
self.memory_client.search,
|
||||
query=search_query,
|
||||
user_id=self.role_id,
|
||||
limit=30
|
||||
)
|
||||
else:
|
||||
# AsyncMemory uses async search
|
||||
results = await self.memory_client.search(
|
||||
query=search_query,
|
||||
user_id=self.role_id,
|
||||
limit=30
|
||||
)
|
||||
|
||||
if results and "results" in results:
|
||||
# Format each memory entry with its update time
|
||||
memories = []
|
||||
for entry in results.get("results", []):
|
||||
# Get timestamp from updated_at or created_at
|
||||
timestamp = ""
|
||||
if "updated_at" in entry and entry["updated_at"]:
|
||||
timestamp = str(entry["updated_at"])
|
||||
elif "created_at" in entry and entry["created_at"]:
|
||||
timestamp = str(entry["created_at"])
|
||||
|
||||
if timestamp:
|
||||
try:
|
||||
# Parse and reformat the timestamp (remove milliseconds if present)
|
||||
if "." in timestamp:
|
||||
dt = timestamp.split(".")[0]
|
||||
else:
|
||||
dt = timestamp
|
||||
formatted_time = dt.replace("T", " ")
|
||||
except Exception:
|
||||
formatted_time = timestamp
|
||||
else:
|
||||
formatted_time = ""
|
||||
|
||||
memory = entry.get("memory", "") or entry.get("content", "")
|
||||
if memory:
|
||||
if formatted_time:
|
||||
# Store tuple of (timestamp, formatted_string) for sorting
|
||||
memories.append((timestamp, f"[{formatted_time}] {memory}"))
|
||||
else:
|
||||
memories.append(("", memory))
|
||||
|
||||
# Sort by timestamp in descending order (newest first)
|
||||
memories.sort(key=lambda x: x[0], reverse=True)
|
||||
|
||||
# Extract only the formatted strings
|
||||
if memories:
|
||||
memories_str = "\n".join(f"- {memory[1]}" for memory in memories)
|
||||
result_parts.append(f"【相关记忆】\n{memories_str}")
|
||||
|
||||
final_result = "\n\n".join(result_parts)
|
||||
logger.bind(tag=TAG).debug(f"Query results: {final_result}")
|
||||
return final_result
|
||||
|
||||
except Exception as e:
|
||||
logger.bind(tag=TAG).error(f"Error querying memory: {str(e)}")
|
||||
logger.bind(tag=TAG).debug(f"Detailed error: {traceback.format_exc()}")
|
||||
return ""
|
||||
|
||||
async def get_user_profile(self) -> str:
|
||||
"""
|
||||
Get user profile from PowerMem (only available in UserMemory mode).
|
||||
|
||||
In PowerMem 0.3.0+, user profile is automatically extracted during add()
|
||||
and cached in last_profile_content.
|
||||
|
||||
Returns:
|
||||
Formatted user profile string or empty string if not available
|
||||
"""
|
||||
if not self.use_powermem or self.memory_client is None:
|
||||
return ""
|
||||
|
||||
if not self.enable_user_profile:
|
||||
logger.bind(tag=TAG).debug("User profile mode is not enabled")
|
||||
return ""
|
||||
|
||||
# Return cached profile content from last add() operation
|
||||
if self.last_profile_content:
|
||||
return self.last_profile_content
|
||||
|
||||
return ""
|
||||
|
||||
@@ -23,6 +23,7 @@ loguru==0.7.3
|
||||
requests==2.32.5
|
||||
cozepy==0.20.0
|
||||
mem0ai==1.0.0
|
||||
powermem>=0.3.1
|
||||
bs4==0.0.2
|
||||
modelscope==1.32.0
|
||||
sherpa_onnx==1.12.17
|
||||
|
||||
Reference in New Issue
Block a user