@lobehub/chat 1.55.4 → 1.56.1

package/scripts/readmeWorkflow/syncProviderIndex.ts

@@ -47,5 +47,4 @@ const runProviderTable = async (lang?: string) => {
 export default async () => {
   await runProviderTable();
   await runProviderTable('zh-CN');
-  await runProviderTable('ja-JP');
 };

package/scripts/vercelIgnoredBuildStep.js

@@ -5,7 +5,7 @@ const branchName = process.env.VERCEL_GIT_COMMIT_REF || '';
 
 function shouldProceedBuild() {
   // 如果是 lighthouse 分支或以 testgru 开头的分支，取消构建
-  if (branchName === 'lighthouse' || branchName.startsWith('testgru')) {
+  if (branchName === 'lighthouse' || branchName.startsWith('gru/')) {
     return false;
   }
 

package/src/app/[variants]/(main)/chat/(workspace)/features/AgentSettings/CategoryContent/useCategory.tsx

@@ -1,10 +1,13 @@
 import { Icon } from '@lobehub/ui';
+import { MenuItemType } from 'antd/es/menu/interface';
 import { Blocks, Bot, BrainCog, MessagesSquare, Mic2, UserCircle } from 'lucide-react';
 import { useMemo } from 'react';
 import { useTranslation } from 'react-i18next';
 
 import type { MenuProps } from '@/components/Menu';
+import { INBOX_SESSION_ID } from '@/const/session';
 import { ChatSettingsTabs } from '@/store/global/initialState';
+import { useSessionStore } from '@/store/session';
 
 interface UseCategoryOptions {
   mobile?: boolean;
@@ -13,41 +16,44 @@ interface UseCategoryOptions {
 export const useCategory = ({ mobile }: UseCategoryOptions = {}) => {
   const { t } = useTranslation('setting');
   const iconSize = mobile ? { fontSize: 20 } : undefined;
+  const id = useSessionStore((s) => s.activeId);
+  const isInbox = id === INBOX_SESSION_ID;
 
   const cateItems: MenuProps['items'] = useMemo(
-    () => [
-      {
-        icon: <Icon icon={UserCircle} size={iconSize} />,
-        key: ChatSettingsTabs.Meta,
-        label: t('agentTab.meta'),
-      },
-      {
-        icon: <Icon icon={Bot} size={iconSize} />,
-        key: ChatSettingsTabs.Prompt,
-        label: t('agentTab.prompt'),
-      },
-      {
-        icon: <Icon icon={MessagesSquare} size={iconSize} />,
-        key: ChatSettingsTabs.Chat,
-        label: t('agentTab.chat'),
-      },
-      {
-        icon: <Icon icon={BrainCog} size={iconSize} />,
-        key: ChatSettingsTabs.Modal,
-        label: t('agentTab.modal'),
-      },
-      {
-        icon: <Icon icon={Mic2} size={iconSize} />,
-        key: ChatSettingsTabs.TTS,
-        label: t('agentTab.tts'),
-      },
-      {
-        icon: <Icon icon={Blocks} size={iconSize} />,
-        key: ChatSettingsTabs.Plugin,
-        label: t('agentTab.plugin'),
-      },
-    ],
-    [t],
+    () =>
+      [
+        (!isInbox && {
+          icon: <Icon icon={UserCircle} size={iconSize} />,
+          key: ChatSettingsTabs.Meta,
+          label: t('agentTab.meta'),
+        }) as MenuItemType,
+        (!isInbox && {
+          icon: <Icon icon={Bot} size={iconSize} />,
+          key: ChatSettingsTabs.Prompt,
+          label: t('agentTab.prompt'),
+        }) as MenuItemType,
+        {
+          icon: <Icon icon={MessagesSquare} size={iconSize} />,
+          key: ChatSettingsTabs.Chat,
+          label: t('agentTab.chat'),
+        },
+        {
+          icon: <Icon icon={BrainCog} size={iconSize} />,
+          key: ChatSettingsTabs.Modal,
+          label: t('agentTab.modal'),
+        },
+        {
+          icon: <Icon icon={Mic2} size={iconSize} />,
+          key: ChatSettingsTabs.TTS,
+          label: t('agentTab.tts'),
+        },
+        {
+          icon: <Icon icon={Blocks} size={iconSize} />,
+          key: ChatSettingsTabs.Plugin,
+          label: t('agentTab.plugin'),
+        },
+      ].filter(Boolean),
+    [t, isInbox],
   );
 
   return cateItems;

package/src/app/[variants]/(main)/chat/(workspace)/features/AgentSettings/index.tsx

@@ -8,6 +8,7 @@ import { useTranslation } from 'react-i18next';
 import { Flexbox } from 'react-layout-kit';
 
 import Header from '@/app/[variants]/(main)/settings/_layout/Desktop/Header';
+import { INBOX_SESSION_ID } from '@/const/session';
 import AgentChat from '@/features/AgentSetting/AgentChat';
 import AgentMeta from '@/features/AgentSetting/AgentMeta';
 import AgentModal from '@/features/AgentSetting/AgentModal';
@@ -38,8 +39,9 @@ const AgentSettings = memo(() => {
     s.updateSessionMeta,
     sessionMetaSelectors.currentAgentTitle(s),
   ]);
+  const isInbox = id === INBOX_SESSION_ID;
 
-  const [tab, setTab] = useState(ChatSettingsTabs.Meta);
+  const [tab, setTab] = useState(isInbox ? ChatSettingsTabs.Chat : ChatSettingsTabs.Meta);
 
   const ref = useRef<any>(null);
   const theme = useTheme();

package/src/app/[variants]/(main)/chat/(workspace)/features/SettingButton.tsx

@@ -8,6 +8,7 @@ import { useTranslation } from 'react-i18next';
 
 import { DESKTOP_HEADER_ICON_SIZE, MOBILE_HEADER_ICON_SIZE } from '@/const/layoutTokens';
 import { useOpenChatSettings } from '@/hooks/useInterceptingRoutes';
+import { useSessionStore } from '@/store/session';
 
 const AgentSettings = dynamic(() => import('./AgentSettings'), {
   ssr: false,
@@ -16,6 +17,7 @@ const AgentSettings = dynamic(() => import('./AgentSettings'), {
 const SettingButton = memo<{ mobile?: boolean }>(({ mobile }) => {
   const { t } = useTranslation('common');
   const openChatSettings = useOpenChatSettings();
+  const id = useSessionStore((s) => s.activeId);
 
   return (
     <>
@@ -25,7 +27,7 @@ const SettingButton = memo<{ mobile?: boolean }>(({ mobile }) => {
         size={mobile ? MOBILE_HEADER_ICON_SIZE : DESKTOP_HEADER_ICON_SIZE}
         title={t('header.session', { ns: 'setting' })}
       />
-      <AgentSettings />
+      <AgentSettings key={id} />
     </>
   );
 });

package/src/config/knowledge.ts

@@ -1,19 +1,17 @@
 import { createEnv } from '@t3-oss/env-nextjs';
 import { z } from 'zod';
 
-export const getKnowledgeConfig = () => {
-  return createEnv({
-    runtimeEnv: {
-      DEFAULT_FILES_CONFIG: process.env.DEFAULT_FILES_CONFIG,
-      UNSTRUCTURED_API_KEY: process.env.UNSTRUCTURED_API_KEY,
-      UNSTRUCTURED_SERVER_URL: process.env.UNSTRUCTURED_SERVER_URL,
-    },
-    server: {
-      DEFAULT_FILES_CONFIG: z.string().optional(),
-      UNSTRUCTURED_API_KEY: z.string().optional(),
-      UNSTRUCTURED_SERVER_URL: z.string().optional(),
-    },
-  });
-};
-
-export const knowledgeEnv = getKnowledgeConfig();
+export const knowledgeEnv = createEnv({
+  runtimeEnv: {
+    DEFAULT_FILES_CONFIG: process.env.DEFAULT_FILES_CONFIG,
+    FILE_TYPE_CHUNKING_RULES: process.env.FILE_TYPE_CHUNKING_RULES,
+    UNSTRUCTURED_API_KEY: process.env.UNSTRUCTURED_API_KEY,
+    UNSTRUCTURED_SERVER_URL: process.env.UNSTRUCTURED_SERVER_URL,
+  },
+  server: {
+    DEFAULT_FILES_CONFIG: z.string().optional(),
+    FILE_TYPE_CHUNKING_RULES: z.string().optional(),
+    UNSTRUCTURED_API_KEY: z.string().optional(),
+    UNSTRUCTURED_SERVER_URL: z.string().optional(),
+  },
+});

package/src/hooks/useInterceptingRoutes.ts

@@ -2,6 +2,7 @@ import { useMemo } from 'react';
 import urlJoin from 'url-join';
 
 import { INBOX_SESSION_ID } from '@/const/session';
+import { isDeprecatedEdition } from '@/const/version';
 import { useIsMobile } from '@/hooks/useIsMobile';
 import { useQueryRoute } from '@/hooks/useQueryRoute';
 import { useAgentStore } from '@/store/agent';
@@ -15,7 +16,7 @@ export const useOpenChatSettings = (tab: ChatSettingsTabs = ChatSettingsTabs.Met
   const router = useQueryRoute();
 
   return useMemo(() => {
-    if (activeId === INBOX_SESSION_ID) {
+    if (isDeprecatedEdition && activeId === INBOX_SESSION_ID) {
       return () => router.push(urlJoin('/settings', SettingsTabs.Agent));
     }
 

package/src/server/modules/ContentChunk/index.ts

@@ -1,9 +1,13 @@
 import { ChunkingLoader } from 'src/libs/langchain';
 import { Strategy } from 'unstructured-client/sdk/models/shared';
 
-import { NewChunkItem, NewUnstructuredChunkItem } from '@/database/schemas';
+import { knowledgeEnv } from '@/config/knowledge';
+import type { NewChunkItem, NewUnstructuredChunkItem } from '@/database/schemas';
 import { ChunkingStrategy, Unstructured } from '@/libs/unstructured';
 
+import { ChunkingRuleParser } from './rules';
+import type { ChunkingService } from './rules';
+
 export interface ChunkContentParams {
   content: Uint8Array;
   fileType: string;
@@ -19,23 +23,57 @@ interface ChunkResult {
 export class ContentChunk {
   private unstructuredClient: Unstructured;
   private langchainClient: ChunkingLoader;
+  private chunkingRules: Record<string, ChunkingService[]>;
 
   constructor() {
     this.unstructuredClient = new Unstructured();
     this.langchainClient = new ChunkingLoader();
+    this.chunkingRules = ChunkingRuleParser.parse(knowledgeEnv.FILE_TYPE_CHUNKING_RULES || '');
   }
 
-  isUsingUnstructured(params: ChunkContentParams) {
-    return params.fileType === 'application/pdf' && params.mode === 'hi-res';
+  private getChunkingServices(fileType: string): ChunkingService[] {
+    const ext = fileType.split('/').pop()?.toLowerCase() || '';
+    return this.chunkingRules[ext] || ['default'];
   }
 
   async chunkContent(params: ChunkContentParams): Promise<ChunkResult> {
-    if (this.isUsingUnstructured(params))
-      return await this.chunkByUnstructured(params.filename, params.content);
-
+    const services = this.getChunkingServices(params.fileType);
+
+    for (const service of services) {
+      try {
+        switch (service) {
+          case 'unstructured': {
+            if (this.canUseUnstructured()) {
+              return await this.chunkByUnstructured(params.filename, params.content);
+            }
+            break;
+          }
+
+          case 'doc2x': {
+            // Future implementation
+            break;
+          }
+
+          default: {
+            return await this.chunkByLangChain(params.filename, params.content);
+          }
+        }
+      } catch (error) {
+        // If this is the last service, throw the error
+        if (service === services.at(-1)) throw error;
+        // Otherwise continue to next service
+        console.error(`Chunking failed with service ${service}:`, error);
+      }
+    }
+
+    // Fallback to langchain if no service succeeded
     return await this.chunkByLangChain(params.filename, params.content);
   }
 
+  private canUseUnstructured(): boolean {
+    return !!(knowledgeEnv.UNSTRUCTURED_API_KEY && knowledgeEnv.UNSTRUCTURED_SERVER_URL);
+  }
+
   private chunkByUnstructured = async (
     filename: string,
     content: Uint8Array,

package/src/server/modules/ContentChunk/rules.test.ts

@@ -0,0 +1,81 @@
+import { describe, expect, it } from 'vitest';
+import { ChunkingRuleParser } from './rules';
+
+describe('ChunkingRuleParser', () => {
+  describe('parse', () => {
+    it('should parse a single file type rule correctly', () => {
+      const input = 'pdf=unstructured,default';
+      const result = ChunkingRuleParser.parse(input);
+
+      expect(result).toEqual({
+        pdf: ['unstructured', 'default'],
+      });
+    });
+
+    it('should parse multiple file type rules correctly', () => {
+      const input = 'pdf=unstructured,default;doc=doc2x,default;txt=default';
+      const result = ChunkingRuleParser.parse(input);
+
+      expect(result).toEqual({
+        pdf: ['unstructured', 'default'],
+        doc: ['doc2x', 'default'],
+        txt: ['default'],
+      });
+    });
+
+    it('should convert file types to lowercase', () => {
+      const input = 'PDF=unstructured;DOC=doc2x';
+      const result = ChunkingRuleParser.parse(input);
+
+      expect(result).toEqual({
+        pdf: ['unstructured'],
+        doc: ['doc2x'],
+      });
+    });
+
+    it('should filter out invalid service names', () => {
+      const input = 'pdf=unstructured,invalid,default,wrongservice';
+      const result = ChunkingRuleParser.parse(input);
+
+      expect(result).toEqual({
+        pdf: ['unstructured', 'default'],
+      });
+    });
+
+    it('should handle empty string input', () => {
+      const input = '';
+      const result = ChunkingRuleParser.parse(input);
+
+      expect(result).toEqual({});
+    });
+
+    it('should skip invalid rule formats', () => {
+      const input = 'pdf=unstructured;invalid;doc=doc2x;=default;txt';
+      const result = ChunkingRuleParser.parse(input);
+
+      expect(result).toEqual({
+        pdf: ['unstructured'],
+        doc: ['doc2x'],
+      });
+    });
+
+    it('should handle whitespace in service names', () => {
+      const input = 'pdf= unstructured , default ;doc=doc2x';
+      const result = ChunkingRuleParser.parse(input);
+
+      expect(result).toEqual({
+        pdf: ['unstructured', 'default'],
+        doc: ['doc2x'],
+      });
+    });
+
+    it('should handle duplicate services for same file type', () => {
+      const input = 'pdf=unstructured,default,unstructured';
+      const result = ChunkingRuleParser.parse(input);
+
+      expect(result).toEqual({
+        pdf: ['unstructured', 'default', 'unstructured'],
+      });
+    });
+  });
+});

package/src/server/modules/ContentChunk/rules.ts

@@ -0,0 +1,23 @@
+export type ChunkingService = 'unstructured' | 'doc2x' | 'default';
+
+export const ChunkingRuleParser = {
+  parse(rulesStr: string): Record<string, ChunkingService[]> {
+    const rules: Record<string, ChunkingService[]> = {};
+
+    // Split by semicolon for different file types
+    const fileTypeRules = rulesStr.split(';');
+
+    for (const rule of fileTypeRules) {
+      const [fileType, services] = rule.split('=');
+      if (!fileType || !services) continue;
+
+      // Split services by comma and validate each service
+      rules[fileType.toLowerCase()] = services
+        .split(',')
+        .map((s) => s.trim().toLowerCase())
+        .filter((s): s is ChunkingService => ['unstructured', 'doc2x', 'default'].includes(s));
+    }
+
+    return rules;
+  },
+} as const;