browserless 14.0.19 → 14.0.22

package/package.json

@@ -2,7 +2,7 @@
   "name": "browserless",
   "description": "The headless Chrome/Chromium driver on top of Puppeteer. Take screenshots, generate PDFs, extract text and HTML with a production-ready API.",
   "homepage": "https://browserless.js.org",
-  "version": "14.0.19",
+  "version": "14.0.22",
   "types": "index.d.ts",
   "main": "src/index.js",
   "author": {
@@ -35,10 +35,10 @@
     "web-scraping"
   ],
   "dependencies": {
-    "@browserless/errors": "14.0.0",
-    "@browserless/goto": "14.0.19",
-    "@browserless/pdf": "14.0.19",
-    "@browserless/screenshot": "14.0.19",
+    "@browserless/errors": "14.0.22",
+    "@browserless/goto": "14.0.22",
+    "@browserless/pdf": "14.0.22",
+    "@browserless/screenshot": "14.0.22",
     "debug-logfmt": "~1.4.15",
     "kill-process-group": "~1.0.16",
     "p-reflect": "~2.1.0",
@@ -48,7 +48,7 @@
     "superlock": "~1.3.5"
   },
   "devDependencies": {
-    "ava": "5",
+    "ava": "8",
     "ps-list": "7",
     "puppeteer": "25",
     "tinyspawn": "latest"
@@ -69,5 +69,5 @@
     "timeout": "2m",
     "workerThreads": false
   },
-  "gitHead": "1600991013bc8ed915b2effa9e74bbf1b97d1f02"
+  "gitHead": "22c8d447e55cfe1eb7b1e36af9a5cc4d9139bd23"
 }

package/src/driver.js

@@ -69,26 +69,32 @@ const getPid = subprocess => {
   return 'pid' in browserProcess ? browserProcess.pid : undefined
 }
 
-const close = async (subprocess, { signal = 'SIGKILL', ...debugOpts } = {}) => {
+const close = async (subprocess, { signal = 'SIGKILL', force = false, ...debugOpts } = {}) => {
   const pid = getPid(subprocess)
   const hasDisconnect = subprocess && typeof subprocess.disconnect === 'function'
   const hasClose = subprocess && typeof subprocess.close === 'function'
 
   if (pid === undefined && !hasDisconnect && !hasClose) return
 
-  // It's necessary to call `browser.close` for removing temporal files associated
-  // and remove listeners attached to the main process; check
-  // - https://github.com/puppeteer/puppeteer/blob/778ac92469d66c542c3c12fe0aa23703dd6315c2/src/node/BrowserRunner.ts#L146
-  // - https://github.com/puppeteer/puppeteer/blob/69d85e874416d62de6e821bef30e5cebcfd42f15/src/node/BrowserRunner.ts#L189
-  await pReflect(
-    pid === undefined
-      ? hasDisconnect
-        ? subprocess.disconnect()
-        : subprocess.close()
-      : hasClose
-        ? subprocess.close()
-        : killProcessGroup(subprocess, signal)
-  )
+  const child = typeof subprocess.process === 'function' ? subprocess.process() : undefined
+  const kill = () => {
+    const proc = child && child.pid ? child : subprocess
+    return proc && proc.pid ? killProcessGroup(proc, { signal }) : undefined
+  }
+
+  // `browser.close` removes temp files and listeners on the main process:
+  // https://github.com/puppeteer/puppeteer/blob/778ac92469d66c542c3c12fe0aa23703dd6315c2/src/node/BrowserRunner.ts#L146
+  // https://github.com/puppeteer/puppeteer/blob/69d85e874416d62de6e821bef30e5cebcfd42f15/src/node/BrowserRunner.ts#L189
+  const graceful = () => {
+    if (pid === undefined) {
+      return hasDisconnect ? subprocess.disconnect() : subprocess.close()
+    }
+    if (hasClose) return subprocess.close()
+    return kill()
+  }
+
+  // A browser that has stopped answering CDP will not finish `browser.close()`.
+  await pReflect(force ? kill() : graceful())
 
   debug('close', { pid, signal, ...debugOpts })
   return pid === undefined ? {} : { pid }

package/src/index.js

@@ -14,6 +14,19 @@ const { AbortError } = pRetry
 
 const driver = require('./driver')
 
+// Chrome can stop answering CDP while `connected` stays true, so the
+// disconnect respawn never runs. Puppeteer then waits out its 180s protocol
+// timeout on every context, and each wait holds a pool slot.
+const CREATE_CONTEXT_TIMEOUT = 10_000
+
+const closeLateContext = pending => pending.then(context => context.close()).catch(() => {})
+
+const isUnresponsiveBrowser = error =>
+  error?.name === 'TimeoutError' ||
+  (typeof error?.message === 'string' &&
+    error.message.includes('timed out') &&
+    error.message.includes('protocolTimeout'))
+
 module.exports = ({ timeout: globalTimeout = 30000, ...launchOpts } = {}) => {
   const lock = withLock()
   const goto = createGoto({ timeout: globalTimeout, ...launchOpts })
@@ -26,10 +39,10 @@ module.exports = ({ timeout: globalTimeout = 30000, ...launchOpts } = {}) => {
     return browserProcessPromise.then(browserProcess => driver.close(browserProcess, opts))
   }
 
-  const respawn = () =>
+  const respawn = ({ force = false } = {}) =>
     !isClosed &&
     Promise.all([
-      browserProcessPromise.then(driver.close),
+      browserProcessPromise.then(browser => driver.close(browser, { force })),
       (browserProcessPromise = spawn({ respawn: true }))
     ])
 
@@ -64,8 +77,24 @@ module.exports = ({ timeout: globalTimeout = 30000, ...launchOpts } = {}) => {
 
   let browserProcessPromise = spawn()
 
-  const createBrowserContext = contextOpts =>
-    getBrowser().then(browser => browser.createBrowserContext(contextOpts))
+  const createBrowserContext = async contextOpts => {
+    for (let attempt = 0; attempt < 2; attempt++) {
+      const browser = await getBrowser()
+      const pending = browser.createBrowserContext(contextOpts)
+
+      try {
+        return await pTimeout(pending, Math.min(globalTimeout, CREATE_CONTEXT_TIMEOUT))
+      } catch (error) {
+        closeLateContext(pending)
+        if (attempt > 0 || !isUnresponsiveBrowser(error)) throw error
+        debug('respawn', { reason: 'createBrowserContext', message: error.message })
+        await lock(async () => {
+          const current = await browserProcessPromise.catch(() => {})
+          if (current === browser) await respawn({ force: true })
+        })
+      }
+    }
+  }
 
   const getBrowser = async () => {
     if (isClosed) return browserProcessPromise