fix: prefer using system chrome install of playwright install

Fixes #7

Author: harlan-zw

packages/crawl/src/cli.ts

@@ -7,7 +7,7 @@ import { dirname, join, resolve } from 'pathe'
 import { withHttps } from 'ufo'
 import { crawlAndGenerate } from './crawl.ts'
 import { parseUrlPattern, validateGlobPattern } from './glob-utils.ts'
-import { ensurePlaywrightInstalled } from './playwright-utils.ts'
+import { ensurePlaywrightInstalled, isUseChromeSupported } from './playwright-utils.ts'
 
 // Read version from package.json
 const __dirname = dirname(fileURLToPath(import.meta.url))
@@ -464,10 +464,19 @@ async function main() {
 
   // Check playwright installation if needed
   if (options.driver === 'playwright') {
-    const playwrightInstalled = await ensurePlaywrightInstalled()
-    if (!playwrightInstalled) {
-      p.log.error('Cannot proceed without Playwright. Please install it manually or use the HTTP driver instead.')
-      process.exit(1)
+    // Check Chrome support and configure if available
+    const chromeSupported = await isUseChromeSupported()
+    if (chromeSupported) {
+      options.useChrome = true
+      p.log.info('System Chrome detected and enabled.')
+    }
+    else {
+      const playwrightInstalled = await ensurePlaywrightInstalled()
+      if (!playwrightInstalled) {
+        p.log.error('Cannot proceed without Playwright. Please install it manually or use the HTTP driver instead.')
+        process.exit(1)
+      }
+      p.log.info('Using global playwright instance.')
     }
   }
 

packages/crawl/src/crawl.ts

@@ -63,6 +63,7 @@ export async function crawlAndGenerate(options: CrawlOptions, onProgress?: (prog
     generateIndividualMd = true,
     origin,
     driver = 'http',
+    useChrome,
     followLinks = false,
     maxDepth = 1,
     globPatterns = [],
@@ -469,7 +470,17 @@ export async function crawlAndGenerate(options: CrawlOptions, onProgress?: (prog
 
   if (driver === 'playwright') {
     // PlaywrightCrawler - installation check should happen in CLI layer
-    crawler = new PlaywrightCrawler(crawlerOptions as PlaywrightCrawlerOptions)
+    const playwrightOptions = crawlerOptions as PlaywrightCrawlerOptions
+
+    // Add useChrome option if specified
+    if (useChrome) {
+      playwrightOptions.launchContext = {
+        ...playwrightOptions.launchContext,
+        useChrome,
+      }
+    }
+
+    crawler = new PlaywrightCrawler(playwrightOptions)
   }
   else {
     crawler = new HttpCrawler(crawlerOptions as HttpCrawlerOptions)

packages/crawl/src/playwright-utils.ts

@@ -1,4 +1,5 @@
 import * as p from '@clack/prompts'
+import { PlaywrightCrawler } from 'crawlee'
 import { addDependency } from 'nypm'
 
 export async function checkPlaywrightInstallation(): Promise<boolean> {
@@ -56,3 +57,30 @@ export async function ensurePlaywrightInstalled(): Promise<boolean> {
 
   return true
 }
+
+export async function isUseChromeSupported(): Promise<boolean> {
+  try {
+    // Create a minimal crawler instance with useChrome option
+    const crawler = new PlaywrightCrawler({
+      launchContext: {
+        useChrome: true,
+      },
+      requestHandler: async () => {
+        // Minimal no-op handler
+      },
+      maxRequestsPerCrawl: 1,
+    })
+    const page = await crawler.browserPool.newPage()
+    await page.evaluate(() => {
+      // Simple check to ensure page is functional
+      return window.navigator.userAgent
+    })
+    await page.close()
+    await crawler.browserPool.closeAllBrowsers()
+    crawler.stop()
+    return true
+  }
+  catch {
+  }
+  return false
+}

packages/crawl/src/types.ts

@@ -8,6 +8,7 @@ export interface CrawlOptions {
   origin?: string
   chunkSize?: number
   driver?: 'http' | 'playwright'
+  useChrome?: boolean
   followLinks?: boolean
   maxDepth?: number
   globPatterns?: ParsedUrlPattern[]