Merge branch 'main' of https://github.com/lobehub/lobe-chat

CHANGELOG.md

@@ -2,6 +2,81 @@
 
 # Changelog
 
+### [Version 1.32.2](https://github.com/lobehub/lobe-chat/compare/v1.32.1...v1.32.2)
+
+<sup>Released on **2024-11-19**</sup>
+
+<br/>
+
+<details>
+<summary><kbd>Improvements and Fixes</kbd></summary>
+
+</details>
+
+<div align="right">
+
+[![](https://img.shields.io/badge/-BACK_TO_TOP-151515?style=flat-square)](#readme-top)
+
+</div>
+
+### [Version 1.32.1](https://github.com/lobehub/lobe-chat/compare/v1.32.0...v1.32.1)
+
+<sup>Released on **2024-11-19**</sup>
+
+#### 🐛 Bug Fixes
+
+- **misc**: Keyword search for chat history & sessions.
+
+#### 💄 Styles
+
+- **misc**: Support o1 models using streaming.
+
+<br/>
+
+<details>
+<summary><kbd>Improvements and Fixes</kbd></summary>
+
+#### What's fixed
+
+- **misc**: Keyword search for chat history & sessions, closes [#4725](https://github.com/lobehub/lobe-chat/issues/4725) ([415d772](https://github.com/lobehub/lobe-chat/commit/415d772))
+
+#### Styles
+
+- **misc**: Support o1 models using streaming, closes [#4732](https://github.com/lobehub/lobe-chat/issues/4732) ([7e9e71a](https://github.com/lobehub/lobe-chat/commit/7e9e71a))
+
+</details>
+
+<div align="right">
+
+[![](https://img.shields.io/badge/-BACK_TO_TOP-151515?style=flat-square)](#readme-top)
+
+</div>
+
+## [Version 1.32.0](https://github.com/lobehub/lobe-chat/compare/v1.31.11...v1.32.0)
+
+<sup>Released on **2024-11-19**</sup>
+
+#### ✨ Features
+
+- **misc**: Add support InternLM (书生浦语) provider.
+
+<br/>
+
+<details>
+<summary><kbd>Improvements and Fixes</kbd></summary>
+
+#### What's improved
+
+- **misc**: Add support InternLM (书生浦语) provider, closes [#4711](https://github.com/lobehub/lobe-chat/issues/4711) ([aaae059](https://github.com/lobehub/lobe-chat/commit/aaae059))
+
+</details>
+
+<div align="right">
+
+[![](https://img.shields.io/badge/-BACK_TO_TOP-151515?style=flat-square)](#readme-top)
+
+</div>
+
 ### [Version 1.31.11](https://github.com/lobehub/lobe-chat/compare/v1.31.10...v1.31.11)
 
 <sup>Released on **2024-11-18**</sup>

Dockerfile

@@ -168,6 +168,8 @@ ENV \
     HUGGINGFACE_API_KEY="" HUGGINGFACE_MODEL_LIST="" HUGGINGFACE_PROXY_URL="" \
     # Hunyuan
     HUNYUAN_API_KEY="" HUNYUAN_MODEL_LIST="" \
+    # InternLM
+    INTERNLM_API_KEY="" INTERNLM_MODEL_LIST="" \
     # Minimax
     MINIMAX_API_KEY="" MINIMAX_MODEL_LIST="" \
     # Mistral

Dockerfile.database

@@ -203,6 +203,8 @@ ENV \
     HUGGINGFACE_API_KEY="" HUGGINGFACE_MODEL_LIST="" HUGGINGFACE_PROXY_URL="" \
     # Hunyuan
     HUNYUAN_API_KEY="" HUNYUAN_MODEL_LIST="" \
+    # InternLM
+    INTERNLM_API_KEY="" INTERNLM_MODEL_LIST="" \
     # Minimax
     MINIMAX_API_KEY="" MINIMAX_MODEL_LIST="" \
     # Mistral

package.json

@@ -1,6 +1,6 @@
 {
   "name": "@lobehub/chat",
-  "version": "1.31.11",
+  "version": "1.32.2",
   "description": "Lobe Chat - an open-source, high-performance chatbot framework that supports speech synthesis, multimodal, and extensible Function Call plugin system. Supports one-click free deployment of your private ChatGPT/LLM web application.",
   "keywords": [
     "framework",

src/app/(main)/settings/llm/ProviderList/providers.tsx

@@ -10,6 +10,7 @@ import {
   GoogleProviderCard,
   GroqProviderCard,
   HunyuanProviderCard,
+  InternLMProviderCard,
   MinimaxProviderCard,
   MistralProviderCard,
   MoonshotProviderCard,
@@ -85,6 +86,7 @@ export const useProviderList = (): ProviderItem[] => {
       MinimaxProviderCard,
       Ai360ProviderCard,
       TaichuProviderCard,
+      InternLMProviderCard,
       SiliconCloudProviderCard,
     ],
     [

src/config/llm.ts

@@ -124,6 +124,9 @@ export const getLLMConfig = () => {
 
       ENABLED_XAI: z.boolean(),
       XAI_API_KEY: z.string().optional(),
+
+      ENABLED_INTERNLM: z.boolean(),
+      INTERNLM_API_KEY: z.string().optional(),
     },
     runtimeEnv: {
       API_KEY_SELECT_MODE: process.env.API_KEY_SELECT_MODE,
@@ -246,6 +249,9 @@ export const getLLMConfig = () => {
 
       ENABLED_XAI: !!process.env.XAI_API_KEY,
       XAI_API_KEY: process.env.XAI_API_KEY,
+
+      ENABLED_INTERNLM: !!process.env.INTERNLM_API_KEY,
+      INTERNLM_API_KEY: process.env.INTERNLM_API_KEY,
     },
   });
 };

src/config/modelProviders/index.ts

@@ -14,6 +14,7 @@ import GoogleProvider from './google';
 import GroqProvider from './groq';
 import HuggingFaceProvider from './huggingface';
 import HunyuanProvider from './hunyuan';
+import InternLMProvider from './internlm';
 import MinimaxProvider from './minimax';
 import MistralProvider from './mistral';
 import MoonshotProvider from './moonshot';
@@ -69,6 +70,7 @@ export const LOBE_DEFAULT_MODEL_LIST: ChatModelCard[] = [
   HunyuanProvider.chatModels,
   WenxinProvider.chatModels,
   SenseNovaProvider.chatModels,
+  InternLMProvider.chatModels,
 ].flat();
 
 export const DEFAULT_MODEL_PROVIDER_LIST = [
@@ -105,6 +107,7 @@ export const DEFAULT_MODEL_PROVIDER_LIST = [
   MinimaxProvider,
   Ai360Provider,
   TaichuProvider,
+  InternLMProvider,
   SiliconCloudProvider,
 ];
 
@@ -131,6 +134,7 @@ export { default as GoogleProviderCard } from './google';
 export { default as GroqProviderCard } from './groq';
 export { default as HuggingFaceProviderCard } from './huggingface';
 export { default as HunyuanProviderCard } from './hunyuan';
+export { default as InternLMProviderCard } from './internlm';
 export { default as MinimaxProviderCard } from './minimax';
 export { default as MistralProviderCard } from './mistral';
 export { default as MoonshotProviderCard } from './moonshot';

src/config/modelProviders/internlm.ts

@@ -0,0 +1,42 @@
+import { ModelProviderCard } from '@/types/llm';
+
+const InternLM: ModelProviderCard = {
+  chatModels: [
+    {
+      description: '我们最新的模型系列，有着卓越的推理性能，支持 1M 的上下文长度以及更强的指令跟随和工具调用能力。',
+      displayName: 'InternLM2.5',
+      enabled: true,
+      functionCall: true,
+      id: 'internlm2.5-latest',
+      maxOutput: 4096,
+      pricing: {
+        input: 0,
+        output: 0,
+      },
+      tokens: 32_768,
+    },
+    {
+      description: '我们仍在维护的老版本模型，有 7B、20B 多种模型参数量可选。',
+      displayName: 'InternLM2 Pro Chat',
+      functionCall: true,
+      id: 'internlm2-pro-chat',
+      maxOutput: 4096,
+      pricing: {
+        input: 0,
+        output: 0,
+      },
+      tokens: 32_768,
+    },
+  ],
+  checkModel: 'internlm2.5-latest',
+  description:
+    '致力于大模型研究与开发工具链的开源组织。为所有 AI 开发者提供高效、易用的开源平台，让最前沿的大模型与算法技术触手可及',
+  disableBrowserRequest: true,
+  id: 'internlm',
+  modelList: { showModelFetcher: true },
+  modelsUrl: 'https://internlm.intern-ai.org.cn/doc/docs/Models#%E8%8E%B7%E5%8F%96%E6%A8%A1%E5%9E%8B%E5%88%97%E8%A1%A8',
+  name: 'InternLM',
+  url: 'https://internlm.intern-ai.org.cn',
+};
+
+export default InternLM;

src/database/server/models/__tests__/session.test.ts

@@ -231,8 +231,18 @@ describe('SessionModel', () => {
 
     it('should return sessions with matching title', async () => {
       await serverDB.insert(sessions).values([
-        { id: '1', userId, title: 'Hello World', description: 'Some description' },
-        { id: '2', userId, title: 'Another Session', description: 'Another description' },
+        { id: '1', userId },
+        { id: '2', userId },
+      ]);
+
+      await serverDB.insert(agents).values([
+        { id: 'agent-1', userId, model: 'gpt-3.5-turbo', title: 'Hello, Agent 1' },
+        { id: 'agent-2', userId, model: 'gpt-4', title: 'Agent 2' },
+      ]);
+
+      await serverDB.insert(agentsToSessions).values([
+        { agentId: 'agent-1', sessionId: '1' },
+        { agentId: 'agent-2', sessionId: '2' },
       ]);
 
       const result = await sessionModel.queryByKeyword('hello');
@@ -241,9 +251,21 @@ describe('SessionModel', () => {
     });
 
     it('should return sessions with matching description', async () => {
+      // The sessions has no title and desc,
+      // see: https://github.com/lobehub/lobe-chat/pull/4725
       await serverDB.insert(sessions).values([
-        { id: '1', userId, title: 'Session 1', description: 'Description with keyword' },
-        { id: '2', userId, title: 'Session 2', description: 'Another description' },
+        { id: '1', userId },
+        { id: '2', userId },
+      ]);
+
+      await serverDB.insert(agents).values([
+        { id: 'agent-1', userId, model: 'gpt-3.5-turbo', title: 'Agent 1', description: 'Description with Keyword' },
+        { id: 'agent-2', userId, model: 'gpt-4', title: 'Agent 2' },
+      ]);
+
+      await serverDB.insert(agentsToSessions).values([
+        { agentId: 'agent-1', sessionId: '1' },
+        { agentId: 'agent-2', sessionId: '2' },
       ]);
 
       const result = await sessionModel.queryByKeyword('keyword');
@@ -253,11 +275,23 @@ describe('SessionModel', () => {
 
     it('should return sessions with matching title or description', async () => {
       await serverDB.insert(sessions).values([
+        { id: '1', userId },
+        { id: '2', userId },
+        { id: '3', userId },
+      ]);
+
+      await serverDB.insert(agents).values([
         { id: '1', userId, title: 'Title with keyword', description: 'Some description' },
         { id: '2', userId, title: 'Another Session', description: 'Description with keyword' },
         { id: '3', userId, title: 'Third Session', description: 'Third description' },
       ]);
 
+      await serverDB.insert(agentsToSessions).values([
+        { agentId: '1', sessionId: '1' },
+        { agentId: '2', sessionId: '2' },
+        { agentId: '3', sessionId: '3' },
+      ]);
+
       const result = await sessionModel.queryByKeyword('keyword');
       expect(result).toHaveLength(2);
       expect(result.map((s) => s.id)).toEqual(['1', '2']);

src/database/server/models/file.ts

@@ -1,5 +1,6 @@
 import { asc, count, eq, ilike, inArray, notExists, or, sum } from 'drizzle-orm';
 import { and, desc, like } from 'drizzle-orm/expressions';
+import type { PgTransaction } from 'drizzle-orm/pg-core';
 
 import { serverDBEnv } from '@/config/db';
 import { serverDB } from '@/database/server/core/db';
@@ -9,6 +10,9 @@ import {
   FileItem,
   NewFile,
   NewGlobalFile,
+  chunks,
+  embeddings,
+  fileChunks,
   files,
   globalFiles,
   knowledgeBaseFiles,
@@ -68,6 +72,10 @@ export class FileModel {
     const fileHash = file.fileHash!;
 
     return await serverDB.transaction(async (trx) => {
+      // 1. 删除相关的 chunks
+      await this.deleteFileChunks(trx as any, [id]);
+
+      // 2. 删除文件记录
       await trx.delete(files).where(and(eq(files.id, id), eq(files.userId, this.userId)));
 
       const result = await trx
@@ -107,6 +115,9 @@ export class FileModel {
     const hashList = fileList.map((file) => file.fileHash!);
 
     return await serverDB.transaction(async (trx) => {
+      // 1. 删除相关的 chunks
+      await this.deleteFileChunks(trx as any, ids);
+
       // delete the files
       await trx.delete(files).where(and(inArray(files.id, ids), eq(files.userId, this.userId)));
 
@@ -289,4 +300,30 @@ export class FileModel {
       ),
     });
   }
+
+  // 抽象出通用的删除 chunks 方法
+  private async deleteFileChunks(trx: PgTransaction<any>, fileIds: string[]) {
+    const BATCH_SIZE = 1000; // 每批处理的数量
+
+    // 1. 获取所有关联的 chunk IDs
+    const relatedChunks = await trx
+      .select({ chunkId: fileChunks.chunkId })
+      .from(fileChunks)
+      .where(inArray(fileChunks.fileId, fileIds));
+
+    const chunkIds = relatedChunks.map((c) => c.chunkId).filter(Boolean) as string[];
+
+    if (chunkIds.length === 0) return;
+
+    // 2. 分批处理删除
+    for (let i = 0; i < chunkIds.length; i += BATCH_SIZE) {
+      const batchChunkIds = chunkIds.slice(i, i + BATCH_SIZE);
+
+      await trx.delete(embeddings).where(inArray(embeddings.chunkId, batchChunkIds));
+
+      await trx.delete(chunks).where(inArray(chunks.id, batchChunkIds));
+    }
+
+    return chunkIds;
+  }
 }