Merge pull request #597 from marp-team/browser-manager

Allow using Firefox / WebDriver BiDi protocol during conversion

CHANGELOG.md

@@ -13,6 +13,8 @@
 
 ### Added
 
+<!-- Allow using Firefox / WebDriver BiDi protocol during conversion ([#597](https://github.com/marp-team/marp-cli/pull/597)) -->
+
 - CI testing against Node.js v22 ([#591](https://github.com/marp-team/marp-cli/pull/591))
 
 ### Changed

jest.config.js

@@ -35,6 +35,7 @@ module.exports = {
   coverageProvider: 'v8',
   coverageThreshold: { global: { lines: 95 } },
   moduleFileExtensions: ['ts', 'tsx', 'js', 'jsx', 'json', 'node'],
+  prettierPath: null,
   setupFiles: ['./jest.setup.js'],
   transform: {
     ...jsWithBabel.transform,

package.json

@@ -40,9 +40,7 @@
   },
   "pkg": {
     "scripts": "lib/**/*.js",
-    "assets": [
-      "node_modules/vm2/**/*"
-    ]
+    "assets": ["tmp/icu/icudt*.dat", "node_modules/sharp/**/*", "node_modules/@img/**/*"]
   },
   "browserslist": [
     "> 1% and last 3 versions",
@@ -61,7 +59,7 @@
     "lint:css": "stylelint \"src/**/*.{css,scss}\"",
     "prepack": "npm-run-all --parallel check:* lint:* test:coverage --parallel build types",
     "preversion": "run-p check:* lint:* test:coverage",
-    "standalone": "node -e 'fs.rmSync(`bin`,{recursive:true,force:true})' && pkg --out-path ./bin .",
+    "standalone": "node -e 'fs.rmSync(`bin`,{recursive:true,force:true})' && pkg --options \"icu-data-dir=$(node ./scripts/icu.mjs)\" -C gzip --out-path ./bin .",
     "standalone:pack": "node ./scripts/pack.js",
     "test": "jest",
     "test:coverage": "jest --coverage",
@@ -82,11 +80,12 @@
     "@tsconfig/node20": "^20.1.4",
     "@tsconfig/recommended": "^1.0.7",
     "@types/cheerio": "^0.22.35",
+    "@types/debug": "^4.1.12",
     "@types/dom-view-transitions": "^1.0.5",
     "@types/express": "^4.17.21",
     "@types/jest": "^29.5.13",
     "@types/markdown-it": "^14.1.2",
-    "@types/node": "~16.18.108",
+    "@types/node": "~18.19.50",
     "@types/pug": "^2.0.10",
     "@types/supertest": "^6.0.2",
     "@types/which": "^3.0.4",
@@ -162,6 +161,7 @@
     "cosmiconfig": "^9.0.0",
     "puppeteer-core": "23.3.0",
     "serve-index": "^1.9.1",
+    "sharp": "^0.33.5",
     "tmp": "^0.2.3",
     "ws": "^8.18.0",
     "yargs": "^17.7.2"

scripts/icu.mjs

@@ -0,0 +1,53 @@
+/* For baking the ICU data into the standalone binary, we need to download the compatible ICU data from the ICU repository. */
+import fs from 'node:fs'
+import path from 'node:path'
+import { fileURLToPath } from 'node:url'
+import { promisify } from 'node:util'
+import yauzl from 'yauzl'
+
+const __dirname = path.dirname(fileURLToPath(import.meta.url))
+const icuDir = path.join(__dirname, '../tmp/icu')
+await fs.promises.mkdir(icuDir, { recursive: true })
+
+const zipFromBuffer = promisify(yauzl.fromBuffer)
+
+// Get the ICU version and endianness
+const [icuMajor, icuMinor] = process.versions.icu.split('.')
+const icuEndianness = process.config.variables.icu_endianness.toLowerCase()
+
+// Download the ICU data
+const response = await fetch(
+  `https://github.com/unicode-org/icu/releases/download/release-${icuMajor}-${icuMinor}/icu4c-${icuMajor}_${icuMinor}-data-bin-${icuEndianness}.zip`
+)
+
+if (!response.ok) {
+  throw new Error(`Failed to download ICU data: ${response.statusText}`)
+}
+
+// Extract the ICU data
+const zip = await zipFromBuffer(Buffer.from(await response.arrayBuffer()), {
+  lazyEntries: true,
+})
+
+const icuDat = await new Promise((res, rej) => {
+  zip.on('error', (err) => rej(err))
+  zip.on('entry', async (entry) => {
+    if (/icudt\d+.\.dat/.test(entry.fileName)) {
+      zip.openReadStream(entry, (err, readStream) => {
+        if (err) return rej(err)
+
+        const output = path.join(icuDir, entry.fileName)
+
+        readStream.pipe(fs.createWriteStream(output))
+        res(output)
+      })
+    } else {
+      zip.readEntry()
+    }
+  })
+  zip.on('end', () => rej(new Error('Failed to find ICU data in the archive')))
+  zip.readEntry()
+})
+
+// Print the relative path to the ICU data from the project root
+console.log(path.relative(path.join(__dirname, '../'), icuDat))

src/browser/browser.ts

@@ -1,21 +1,50 @@
+import { EventEmitter } from 'node:events'
+import { launch } from 'puppeteer-core'
+import type {
+  Browser as PuppeteerBrowser,
+  ProtocolType,
+  PuppeteerLaunchOptions,
+  Page,
+} from 'puppeteer-core'
+import type TypedEventEmitter from 'typed-emitter'
+import { isWSL } from '../utils/wsl'
+
 export type BrowserKind = 'chrome' | 'firefox'
-export type BrowserProtocol = 'webdriver-bidi' | 'cdp'
-export type BrowserPurpose = 'convert' | 'preview'
+export type BrowserProtocol = ProtocolType
 
 export interface BrowserOptions {
-  purpose: BrowserPurpose
+  path: string
+  timeout?: number
+}
+
+// eslint-disable-next-line @typescript-eslint/consistent-type-definitions -- TypedEventEmitter is only compatible with type
+type BrowserEvents = {
+  close: (browser: PuppeteerBrowser) => void
+  disconnect: (browser: PuppeteerBrowser) => void
+  launch: (browser: PuppeteerBrowser) => void
 }
 
-export abstract class Browser {
+const wslHostMatcher = /^\/mnt\/[a-z]\//
+
+export abstract class Browser
+  extends (EventEmitter as new () => TypedEventEmitter<BrowserEvents>)
+  implements AsyncDisposable
+{
   static readonly kind: BrowserKind
   static readonly protocol: BrowserProtocol
 
-  // ---
-
-  purpose: BrowserPurpose
+  path: string
+  protocolTimeout: number
+  puppeteer: PuppeteerBrowser | undefined
+  timeout: number
 
   constructor(opts: BrowserOptions) {
-    this.purpose = opts.purpose
+    super()
+
+    this.path = opts.path
+    this.timeout = opts.timeout ?? 30000
+    this.protocolTimeout =
+      this.timeout === 0 ? 0 : Math.max(180_000, this.timeout)
   }
 
   get kind() {
@@ -25,4 +54,81 @@ export abstract class Browser {
   get protocol() {
     return (this.constructor as typeof Browser).protocol
   }
+
+  async launch(opts: PuppeteerLaunchOptions = {}): Promise<PuppeteerBrowser> {
+    if (!this.puppeteer) {
+      const puppeteer = await this.launchPuppeteer(opts)
+
+      puppeteer.once('disconnected', () => {
+        this.emit('disconnect', puppeteer)
+        this.puppeteer = undefined
+      })
+
+      this.puppeteer = puppeteer
+      this.emit('launch', puppeteer)
+
+      return puppeteer
+    }
+    return this.puppeteer
+  }
+
+  async withPage<T>(fn: (page: Page) => T) {
+    const puppeteer = await this.launch()
+    const page = await puppeteer.newPage()
+
+    page.setDefaultTimeout(this.timeout)
+    page.setDefaultNavigationTimeout(this.timeout)
+
+    try {
+      return await fn(page)
+    } finally {
+      await page.close()
+    }
+  }
+
+  async close() {
+    if (this.puppeteer) {
+      const { puppeteer } = this
+
+      if (puppeteer.connected) {
+        await puppeteer.close()
+        this.emit('close', puppeteer)
+      }
+
+      this.puppeteer = undefined
+    }
+  }
+
+  async [Symbol.asyncDispose]() {
+    await this.close()
+  }
+
+  async browserInWSLHost(): Promise<boolean> {
+    return (
+      !!(await isWSL()) &&
+      wslHostMatcher.test(this.puppeteer?.process()?.spawnfile ?? this.path)
+    )
+  }
+
+  /** @internal Overload in subclass to customize launch behavior */
+  protected async launchPuppeteer(
+    opts: PuppeteerLaunchOptions
+  ): Promise<PuppeteerBrowser> {
+    return await launch(this.generateLaunchOptions(opts))
+  }
+
+  /** @internal */
+  protected generateLaunchOptions(
+    mergeOptions: PuppeteerLaunchOptions = {}
+  ): PuppeteerLaunchOptions {
+    return {
+      browser: this.kind,
+      executablePath: this.path,
+      headless: true,
+      protocol: this.protocol,
+      protocolTimeout: this.protocolTimeout,
+      timeout: this.timeout,
+      ...mergeOptions,
+    }
+  }
 }

src/browser/browsers/chrome-cdp.ts

@@ -1,6 +1,26 @@
-import { Browser } from '../browser'
+import type { PuppeteerLaunchOptions } from 'puppeteer-core'
+import macDockIcon from '../../assets/mac-dock-icon.png'
+import { BrowserProtocol } from '../browser'
+import { ChromeBrowser } from './chrome'
 
-export class ChromeCdpBrowser extends Browser {
-  static readonly kind = 'chrome' as const
-  static readonly protocol = 'cdp' as const
+export class ChromeCdpBrowser extends ChromeBrowser {
+  static readonly protocol: BrowserProtocol = 'cdp'
+
+  protected async launchPuppeteer(opts: PuppeteerLaunchOptions) {
+    const puppeteer = await super.launchPuppeteer(opts)
+
+    // macOS specific: Set Marp icon asynchrnously
+    if (process.platform === 'darwin') {
+      puppeteer
+        .target()
+        .createCDPSession()
+        .then((session) => {
+          session
+            .send('Browser.setDockTile', { image: macDockIcon.slice(22) })
+            .catch(() => void 0)
+        })
+    }
+
+    return puppeteer
+  }
 }