feat(server): enable share og information for docs (#7794)

Author: forehalo

packages/backend/server/migrations/20240903033137_workspace_url_preview/migration.sql

@@ -0,0 +1,2 @@
+-- AlterTable
+ALTER TABLE "workspaces" ADD COLUMN     "enable_url_preview" BOOLEAN NOT NULL DEFAULT false;

packages/backend/server/package.json

@@ -94,6 +94,7 @@
     "ts-node": "^10.9.2",
     "typescript": "^5.4.5",
     "ws": "^8.16.0",
+    "xss": "^1.0.15",
     "yjs": "patch:yjs@npm%3A13.6.18#~/.yarn/patches/yjs-npm-13.6.18-ad0d5f7c43.patch",
     "zod": "^3.22.4"
   },

packages/backend/server/schema.prisma

@@ -97,9 +97,10 @@ model VerificationToken {
 }
 
 model Workspace {
-  id        String   @id @default(uuid()) @db.VarChar
-  public    Boolean
-  createdAt DateTime @default(now()) @map("created_at") @db.Timestamptz(3)
+  id               String   @id @default(uuid()) @db.VarChar
+  public           Boolean
+  enableUrlPreview Boolean  @default(false) @map("enable_url_preview")
+  createdAt        DateTime @default(now()) @map("created_at") @db.Timestamptz(3)
 
   pages           WorkspacePage[]
   permissions     WorkspaceUserPermission[]

packages/backend/server/src/app.module.ts

@@ -11,6 +11,7 @@ import { AppController } from './app.controller';
 import { AuthModule } from './core/auth';
 import { ADD_ENABLED_FEATURES, ServerConfigModule } from './core/config';
 import { DocStorageModule } from './core/doc';
+import { DocRendererModule } from './core/doc-renderer';
 import { FeatureModule } from './core/features';
 import { PermissionModule } from './core/permission';
 import { QuotaModule } from './core/quota';
@@ -42,7 +43,6 @@ import { ENABLED_PLUGINS } from './plugins/registry';
 
 export const FunctionalityModules = [
   ConfigModule.forRoot(),
-  ScheduleModule.forRoot(),
   EventModule,
   CacheModule,
   MutexModule,
@@ -156,24 +156,24 @@ export function buildAppModule() {
     .use(UserModule, AuthModule, PermissionModule)
 
     // business modules
-    .use(DocStorageModule)
+    .use(FeatureModule, QuotaModule, DocStorageModule)
 
     // sync server only
     .useIf(config => config.flavor.sync, SyncModule)
 
     // graphql server only
     .useIf(
       config => config.flavor.graphql,
+      ScheduleModule.forRoot(),
       GqlModule,
       StorageModule,
       ServerConfigModule,
-      WorkspaceModule,
-      FeatureModule,
-      QuotaModule
+      WorkspaceModule
     )
 
     // self hosted server only
-    .useIf(config => config.isSelfhosted, SelfhostModule);
+    .useIf(config => config.isSelfhosted, SelfhostModule)
+    .useIf(config => config.flavor.renderer, DocRendererModule);
 
   // plugin modules
   ENABLED_PLUGINS.forEach(name => {

packages/backend/server/src/core/doc-renderer/controller.ts

@@ -0,0 +1,95 @@
+import { Controller, Get, Param, Res } from '@nestjs/common';
+import type { Response } from 'express';
+import xss from 'xss';
+
+import { DocNotFound } from '../../fundamentals';
+import { PermissionService } from '../permission';
+import { PageDocContent } from '../utils/blocksuite';
+import { DocContentService } from './service';
+
+interface RenderOptions {
+  og: boolean;
+  content: boolean;
+}
+
+@Controller('/workspace/:workspaceId/:docId')
+export class DocRendererController {
+  constructor(
+    private readonly doc: DocContentService,
+    private readonly permission: PermissionService
+  ) {}
+
+  @Get()
+  async render(
+    @Res() res: Response,
+    @Param('workspaceId') workspaceId: string,
+    @Param('docId') docId: string
+  ) {
+    if (workspaceId === docId) {
+      throw new DocNotFound({ spaceId: workspaceId, docId });
+    }
+
+    // if page is public, show all
+    // if page is private, but workspace public og is on, show og but not content
+    const opts: RenderOptions = {
+      og: false,
+      content: false,
+    };
+    const isPagePublic = await this.permission.isPublicPage(workspaceId, docId);
+
+    if (isPagePublic) {
+      opts.og = true;
+      opts.content = true;
+    } else {
+      const allowPreview = await this.permission.allowUrlPreview(workspaceId);
+
+      if (allowPreview) {
+        opts.og = true;
+      }
+    }
+
+    let docContent = opts.og
+      ? await this.doc.getPageContent(workspaceId, docId)
+      : null;
+    if (!docContent) {
+      docContent = { title: 'untitled', summary: '' };
+    }
+
+    res.setHeader('Content-Type', 'text/html');
+    if (!opts.og) {
+      res.setHeader('X-Robots-Tag', 'noindex');
+    }
+    res.send(this._render(docContent, opts));
+  }
+
+  _render(doc: PageDocContent, { og }: RenderOptions): string {
+    const title = xss(doc.title);
+    const summary = xss(doc.summary);
+
+    return `
+      <!DOCTYPE html>
+      <html>
+        <head>
+          <title>${title} | AFFiNE</title>
+          <meta name="theme-color" content="#fafafa" />
+          <link rel="manifest" href="/manifest.json" />
+          <link rel="apple-touch-icon" sizes="180x180" href="/apple-touch-icon.png" />
+          <link rel="icon" sizes="192x192" href="/favicon-192.png" />
+          ${!og ? '<meta name="robots" content="noindex, nofollow" />' : ''}
+          <meta
+            name="twitter:title"
+            content="AFFiNE: There can be more than Notion and Miro."
+          />
+          <meta name="twitter:description" content="${title}" />
+          <meta name="twitter:site" content="@AffineOfficial" />
+          <meta name="twitter:image" content="https://affine.pro/og.jpeg" />
+          <meta property="og:title" content="${title}" />
+          <meta property="og:description" content="${summary}" />
+          <meta property="og:image" content="https://affine.pro/og.jpeg" />
+        </head>
+        <body>
+        </body>
+      </html>
+    `;
+  }
+}

packages/backend/server/src/core/doc-renderer/index.ts

@@ -0,0 +1,16 @@
+import { Module } from '@nestjs/common';
+
+import { DocStorageModule } from '../doc';
+import { PermissionModule } from '../permission';
+import { DocRendererController } from './controller';
+import { DocContentService } from './service';
+
+@Module({
+  imports: [DocStorageModule, PermissionModule],
+  providers: [DocContentService],
+  controllers: [DocRendererController],
+  exports: [DocContentService],
+})
+export class DocRendererModule {}
+
+export { DocContentService };

packages/backend/server/src/core/doc-renderer/service.ts

@@ -0,0 +1,88 @@
+import { Injectable } from '@nestjs/common';
+import { applyUpdate, Doc } from 'yjs';
+
+import { Cache } from '../../fundamentals';
+import { PgWorkspaceDocStorageAdapter } from '../doc';
+import {
+  type PageDocContent,
+  parsePageDoc,
+  parseWorkspaceDoc,
+  type WorkspaceDocContent,
+} from '../utils/blocksuite';
+
+@Injectable()
+export class DocContentService {
+  constructor(
+    private readonly cache: Cache,
+    private readonly workspace: PgWorkspaceDocStorageAdapter
+  ) {}
+
+  async getPageContent(
+    workspaceId: string,
+    guid: string
+  ): Promise<PageDocContent | null> {
+    const cacheKey = `workspace:${workspaceId}:doc:${guid}:content`;
+    const cachedResult = await this.cache.get<PageDocContent>(cacheKey);
+
+    if (cachedResult) {
+      return cachedResult;
+    }
+
+    const docRecord = await this.workspace.getDoc(workspaceId, guid);
+    if (!docRecord) {
+      return null;
+    }
+
+    const doc = new Doc();
+    applyUpdate(doc, docRecord.bin);
+
+    const content = parsePageDoc(doc);
+
+    if (content) {
+      await this.cache.set(cacheKey, content, {
+        ttl:
+          7 *
+          24 *
+          60 *
+          60 *
+          1000 /* TODO(@forehalo): we need time constants helper */,
+      });
+    }
+    return content;
+  }
+
+  async getWorkspaceContent(
+    workspaceId: string
+  ): Promise<WorkspaceDocContent | null> {
+    const cacheKey = `workspace:${workspaceId}:content`;
+    const cachedResult = await this.cache.get<WorkspaceDocContent>(cacheKey);
+
+    if (cachedResult) {
+      return cachedResult;
+    }
+
+    const docRecord = await this.workspace.getDoc(workspaceId, workspaceId);
+    if (!docRecord) {
+      return null;
+    }
+
+    const doc = new Doc();
+    applyUpdate(doc, docRecord.bin);
+
+    const content = parseWorkspaceDoc(doc);
+
+    if (content) {
+      await this.cache.set(cacheKey, content);
+    }
+
+    return content;
+  }
+
+  async markDocContentCacheStale(workspaceId: string, guid: string) {
+    const key =
+      workspaceId === guid
+        ? `workspace:${workspaceId}:content`
+        : `workspace:${workspaceId}:doc:${guid}:content`;
+    await this.cache.delete(key);
+  }
+}

packages/backend/server/src/core/doc/job.ts

@@ -1,4 +1,4 @@
-import { Injectable, Logger, OnModuleInit } from '@nestjs/common';
+import { Injectable, Logger, OnModuleInit, Optional } from '@nestjs/common';
 import { Cron, CronExpression, SchedulerRegistry } from '@nestjs/schedule';
 import { PrismaClient } from '@prisma/client';
 
@@ -11,14 +11,14 @@ export class DocStorageCronJob implements OnModuleInit {
   private busy = false;
 
   constructor(
-    private readonly registry: SchedulerRegistry,
     private readonly config: Config,
     private readonly db: PrismaClient,
-    private readonly workspace: PgWorkspaceDocStorageAdapter
+    private readonly workspace: PgWorkspaceDocStorageAdapter,
+    @Optional() private readonly registry?: SchedulerRegistry
   ) {}
 
   onModuleInit() {
-    if (this.config.doc.manager.enableUpdateAutoMerging) {
+    if (this.registry && this.config.doc.manager.enableUpdateAutoMerging) {
       this.registry.addInterval(
         this.autoMergePendingDocUpdates.name,
         // scheduler registry will clean up the interval when the app is stopped

packages/backend/server/src/core/permission/service.ts

@@ -212,7 +212,7 @@ export class PermissionService {
     const count = await this.prisma.workspace.count({
       where: {
         id: ws,
-        public: true,
+        enableUrlPreview: true,
       },
     });