@crawlee/browser-pool 4.0.0-beta.28 → 4.0.0-beta.280

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (86) hide show
  1. package/README.md +15 -15
  2. package/abstract-classes/browser-controller.d.ts +100 -29
  3. package/abstract-classes/browser-controller.js +70 -24
  4. package/abstract-classes/browser-plugin.d.ts +25 -22
  5. package/abstract-classes/browser-plugin.js +21 -13
  6. package/anonymize-proxy.d.ts +6 -2
  7. package/anonymize-proxy.js +9 -5
  8. package/browser-pool.d.ts +110 -51
  9. package/browser-pool.js +318 -147
  10. package/events.d.ts +2 -4
  11. package/events.js +0 -2
  12. package/fingerprinting/hooks.d.ts +0 -1
  13. package/fingerprinting/hooks.js +26 -10
  14. package/fingerprinting/types.d.ts +3 -34
  15. package/fingerprinting/types.js +0 -2
  16. package/fingerprinting/utils.d.ts +0 -1
  17. package/fingerprinting/utils.js +5 -6
  18. package/index.d.ts +15 -7
  19. package/index.js +6 -3
  20. package/launch-context.d.ts +15 -5
  21. package/launch-context.js +15 -10
  22. package/package.json +15 -15
  23. package/playwright/playwright-browser.d.ts +3 -6
  24. package/playwright/playwright-browser.js +17 -17
  25. package/playwright/playwright-controller.d.ts +1 -1
  26. package/playwright/playwright-controller.js +10 -6
  27. package/playwright/playwright-plugin.d.ts +3 -7
  28. package/playwright/playwright-plugin.js +18 -17
  29. package/playwright/remote-playwright-plugin.d.ts +29 -0
  30. package/playwright/remote-playwright-plugin.js +38 -0
  31. package/proxy-server.d.ts +0 -1
  32. package/proxy-server.js +0 -1
  33. package/puppeteer/puppeteer-controller.d.ts +1 -1
  34. package/puppeteer/puppeteer-controller.js +29 -20
  35. package/puppeteer/puppeteer-plugin.d.ts +12 -4
  36. package/puppeteer/puppeteer-plugin.js +73 -49
  37. package/puppeteer/remote-puppeteer-plugin.d.ts +22 -0
  38. package/puppeteer/remote-puppeteer-plugin.js +31 -0
  39. package/remote-browser-plugin.d.ts +49 -0
  40. package/remote-browser-plugin.js +63 -0
  41. package/remote-browser-provider.d.ts +81 -0
  42. package/remote-browser-provider.js +65 -0
  43. package/utils.d.ts +7 -1
  44. package/utils.js +19 -1
  45. package/abstract-classes/browser-controller.d.ts.map +0 -1
  46. package/abstract-classes/browser-controller.js.map +0 -1
  47. package/abstract-classes/browser-plugin.d.ts.map +0 -1
  48. package/abstract-classes/browser-plugin.js.map +0 -1
  49. package/anonymize-proxy.d.ts.map +0 -1
  50. package/anonymize-proxy.js.map +0 -1
  51. package/browser-pool.d.ts.map +0 -1
  52. package/browser-pool.js.map +0 -1
  53. package/container-proxy-server.d.ts +0 -12
  54. package/container-proxy-server.d.ts.map +0 -1
  55. package/container-proxy-server.js +0 -43
  56. package/container-proxy-server.js.map +0 -1
  57. package/events.d.ts.map +0 -1
  58. package/events.js.map +0 -1
  59. package/fingerprinting/hooks.d.ts.map +0 -1
  60. package/fingerprinting/hooks.js.map +0 -1
  61. package/fingerprinting/types.d.ts.map +0 -1
  62. package/fingerprinting/types.js.map +0 -1
  63. package/fingerprinting/utils.d.ts.map +0 -1
  64. package/fingerprinting/utils.js.map +0 -1
  65. package/index.d.ts.map +0 -1
  66. package/index.js.map +0 -1
  67. package/launch-context.d.ts.map +0 -1
  68. package/launch-context.js.map +0 -1
  69. package/logger.d.ts +0 -3
  70. package/logger.d.ts.map +0 -1
  71. package/logger.js +0 -5
  72. package/logger.js.map +0 -1
  73. package/playwright/playwright-browser.d.ts.map +0 -1
  74. package/playwright/playwright-browser.js.map +0 -1
  75. package/playwright/playwright-controller.d.ts.map +0 -1
  76. package/playwright/playwright-controller.js.map +0 -1
  77. package/playwright/playwright-plugin.d.ts.map +0 -1
  78. package/playwright/playwright-plugin.js.map +0 -1
  79. package/proxy-server.d.ts.map +0 -1
  80. package/proxy-server.js.map +0 -1
  81. package/puppeteer/puppeteer-controller.d.ts.map +0 -1
  82. package/puppeteer/puppeteer-controller.js.map +0 -1
  83. package/puppeteer/puppeteer-plugin.d.ts.map +0 -1
  84. package/puppeteer/puppeteer-plugin.js.map +0 -1
  85. package/utils.d.ts.map +0 -1
  86. package/utils.js.map +0 -1
package/browser-pool.js CHANGED
@@ -1,15 +1,36 @@
1
+ import { SessionError, serviceLocator } from '@crawlee/core';
2
+ import { parseArgument, schemas } from '@crawlee/utils/internal';
1
3
  import { FingerprintGenerator } from 'fingerprint-generator';
2
4
  import { FingerprintInjector } from 'fingerprint-injector';
3
5
  import { nanoid } from 'nanoid';
4
- import ow from 'ow';
5
6
  import pLimit from 'p-limit';
6
7
  import QuickLRU from 'quick-lru';
7
8
  import { TypedEmitter } from 'tiny-typed-emitter';
9
+ import { z } from 'zod';
8
10
  import { addTimeoutToPromise, tryCancel } from '@apify/timeout';
11
+ import { BROWSER_POOL_EVENTS } from './events.js';
9
12
  import { createFingerprintPreLaunchHook, createPostPageCreateHook, createPrePageCreateHook, } from './fingerprinting/hooks.js';
10
- import { log } from './logger.js';
13
+ const PAGE_CLOSE_TIMEOUT_MILLIS = 5000;
11
14
  const PAGE_CLOSE_KILL_TIMEOUT_MILLIS = 1000;
12
15
  const BROWSER_KILLER_INTERVAL_MILLIS = 10 * 1000;
16
+ const SLOT_POLL_INTERVAL_MILLIS = 500;
17
+ const browserPoolOptionsSchema = z.strictObject({
18
+ browserPlugins: schemas.anyArray.refine((value) => value.length >= 1, 'Expected a non-empty array'),
19
+ maxOpenPagesPerBrowser: schemas.anyNumber.default(20),
20
+ maxOpenBrowsers: schemas.anyNumber.default(Infinity),
21
+ retireBrowserAfterPageCount: schemas.anyNumber.default(100),
22
+ operationTimeoutSecs: schemas.anyNumber.default(15),
23
+ closeInactiveBrowserAfterSecs: schemas.anyNumber.default(300),
24
+ retireInactiveBrowserAfterSecs: schemas.anyNumber.default(10),
25
+ preLaunchHooks: schemas.anyArray.default(() => []),
26
+ postLaunchHooks: schemas.anyArray.default(() => []),
27
+ prePageCreateHooks: schemas.anyArray.default(() => []),
28
+ postPageCreateHooks: schemas.anyArray.default(() => []),
29
+ prePageCloseHooks: schemas.anyArray.default(() => []),
30
+ postPageCloseHooks: schemas.anyArray.default(() => []),
31
+ useFingerprints: z.boolean().default(true),
32
+ fingerprintOptions: schemas.anyObject.default(() => ({})),
33
+ });
13
34
  /**
14
35
  * The `BrowserPool` class is the most important class of the `browser-pool` module.
15
36
  * It manages opening and closing of browsers and their pages and its constructor
@@ -59,51 +80,45 @@ const BROWSER_KILLER_INTERVAL_MILLIS = 10 * 1000;
59
80
  */
60
81
  export class BrowserPool extends TypedEmitter {
61
82
  browserPlugins;
62
- maxOpenPagesPerBrowser;
63
- retireBrowserAfterPageCount;
64
- operationTimeoutMillis;
65
- closeInactiveBrowserAfterMillis;
66
- useFingerprints;
83
+ /** @internal */
67
84
  fingerprintOptions;
68
- preLaunchHooks;
69
- postLaunchHooks;
70
- prePageCreateHooks;
71
- postPageCreateHooks;
72
- prePageCloseHooks;
73
- postPageCloseHooks;
74
- pageCounter = 0;
75
- pages = new Map();
76
- pageIds = new WeakMap();
77
- startingBrowserControllers = new Set();
78
- activeBrowserControllers = new Set();
79
- retiredBrowserControllers = new Set();
80
- pageToBrowserController = new WeakMap();
85
+ /** @internal */
81
86
  fingerprintInjector;
82
87
  fingerprintGenerator;
88
+ /** @internal */
83
89
  fingerprintCache;
84
- browserKillerInterval = setInterval(async () => this._closeInactiveRetiredBrowsers(), BROWSER_KILLER_INTERVAL_MILLIS);
85
- browserRetireInterval;
86
- limiter = pLimit(1);
90
+ #maxOpenPagesPerBrowser;
91
+ #maxOpenBrowsers;
92
+ #retireBrowserAfterPageCount;
93
+ #operationTimeoutMillis;
94
+ #closeInactiveBrowserAfterMillis;
95
+ #preLaunchHooks;
96
+ #postLaunchHooks;
97
+ #prePageCreateHooks;
98
+ #postPageCreateHooks;
99
+ #prePageCloseHooks;
100
+ #postPageCloseHooks;
101
+ #pageCounter = 0;
102
+ // oxlint-disable-next-line crawlee/prefer-private-fields -- tests observe page tracking and controller retirement directly
103
+ pages = new Map();
104
+ #pageIds = new WeakMap();
105
+ #startingBrowserControllers = new Set();
106
+ // oxlint-disable-next-line crawlee/prefer-private-fields -- accessed by tests
107
+ activeBrowserControllers = new Set();
108
+ // oxlint-disable-next-line crawlee/prefer-private-fields -- accessed by tests
109
+ retiredBrowserControllers = new Set();
110
+ #pageToBrowserController = new WeakMap();
111
+ // oxlint-disable-next-line crawlee/prefer-private-fields -- tests replace this interval through bracket access
112
+ browserKillerInterval;
113
+ #browserRetireInterval;
114
+ #limiter = pLimit(1);
115
+ #log;
116
+ /** Shared by all `newPage` callers waiting for a free slot, so they don't each register their own listeners. */
117
+ #capacityChange;
87
118
  constructor(options) {
88
119
  super();
89
- this.browserKillerInterval.unref();
90
- ow(options, ow.object.exactShape({
91
- browserPlugins: ow.array.minLength(1),
92
- maxOpenPagesPerBrowser: ow.optional.number,
93
- retireBrowserAfterPageCount: ow.optional.number,
94
- operationTimeoutSecs: ow.optional.number,
95
- closeInactiveBrowserAfterSecs: ow.optional.number,
96
- retireInactiveBrowserAfterSecs: ow.optional.number,
97
- preLaunchHooks: ow.optional.array,
98
- postLaunchHooks: ow.optional.array,
99
- prePageCreateHooks: ow.optional.array,
100
- postPageCreateHooks: ow.optional.array,
101
- prePageCloseHooks: ow.optional.array,
102
- postPageCloseHooks: ow.optional.array,
103
- useFingerprints: ow.optional.boolean,
104
- fingerprintOptions: ow.optional.object,
105
- }));
106
- const { browserPlugins, maxOpenPagesPerBrowser = 20, retireBrowserAfterPageCount = 100, operationTimeoutSecs = 15, closeInactiveBrowserAfterSecs = 300, retireInactiveBrowserAfterSecs = 10, preLaunchHooks = [], postLaunchHooks = [], prePageCreateHooks = [], postPageCreateHooks = [], prePageCloseHooks = [], postPageCloseHooks = [], useFingerprints = true, fingerprintOptions = {}, } = options;
120
+ this.#log = serviceLocator.getLogger().child({ prefix: 'BrowserPool' });
121
+ const { browserPlugins, maxOpenPagesPerBrowser, maxOpenBrowsers, retireBrowserAfterPageCount, operationTimeoutSecs, closeInactiveBrowserAfterSecs, retireInactiveBrowserAfterSecs, preLaunchHooks, postLaunchHooks, prePageCreateHooks, postPageCreateHooks, prePageCloseHooks, postPageCloseHooks, useFingerprints, fingerprintOptions, } = parseArgument(options, browserPoolOptionsSchema);
107
122
  const firstPluginConstructor = browserPlugins[0].constructor;
108
123
  for (let i = 1; i < browserPlugins.length; i++) {
109
124
  const providedPlugin = browserPlugins[i];
@@ -114,38 +129,62 @@ export class BrowserPool extends TypedEmitter {
114
129
  }
115
130
  }
116
131
  this.browserPlugins = browserPlugins;
117
- this.maxOpenPagesPerBrowser = maxOpenPagesPerBrowser;
118
- this.retireBrowserAfterPageCount = retireBrowserAfterPageCount;
119
- this.operationTimeoutMillis = operationTimeoutSecs * 1000;
120
- this.closeInactiveBrowserAfterMillis = closeInactiveBrowserAfterSecs * 1000;
121
- this.useFingerprints = useFingerprints;
132
+ this.#maxOpenBrowsers = maxOpenBrowsers;
122
133
  this.fingerprintOptions = fingerprintOptions;
123
- this.browserRetireInterval = setInterval(async () => this.activeBrowserControllers.forEach((controller) => {
134
+ this.#maxOpenPagesPerBrowser = maxOpenPagesPerBrowser;
135
+ this.#retireBrowserAfterPageCount = retireBrowserAfterPageCount;
136
+ this.#operationTimeoutMillis = operationTimeoutSecs * 1000;
137
+ this.#closeInactiveBrowserAfterMillis = closeInactiveBrowserAfterSecs * 1000;
138
+ // Sweeping slower than the window it enforces would round any sub-10s
139
+ // `closeInactiveBrowserAfterSecs` up to the sweep period.
140
+ this.browserKillerInterval = setInterval(async () => this.closeInactiveRetiredBrowsers(), Math.min(BROWSER_KILLER_INTERVAL_MILLIS, this.#closeInactiveBrowserAfterMillis));
141
+ this.browserKillerInterval.unref();
142
+ this.#browserRetireInterval = setInterval(async () => this.activeBrowserControllers.forEach((controller) => {
124
143
  if (controller.activePages === 0 &&
125
144
  controller.lastPageOpenedAt < Date.now() - retireInactiveBrowserAfterSecs * 1000) {
126
145
  this.retireBrowserController(controller);
127
146
  }
128
147
  }), retireInactiveBrowserAfterSecs * 1000);
129
- this.browserRetireInterval.unref();
148
+ this.#browserRetireInterval.unref();
130
149
  // hooks
131
- this.preLaunchHooks = preLaunchHooks;
132
- this.postLaunchHooks = postLaunchHooks;
133
- this.prePageCreateHooks = prePageCreateHooks;
134
- this.postPageCreateHooks = postPageCreateHooks;
135
- this.prePageCloseHooks = prePageCloseHooks;
136
- this.postPageCloseHooks = postPageCloseHooks;
150
+ this.#preLaunchHooks = preLaunchHooks;
151
+ this.#postLaunchHooks = postLaunchHooks;
152
+ this.#prePageCreateHooks = prePageCreateHooks;
153
+ this.#postPageCreateHooks = postPageCreateHooks;
154
+ this.#prePageCloseHooks = prePageCloseHooks;
155
+ this.#postPageCloseHooks = postPageCloseHooks;
137
156
  // fingerprinting
138
- if (this.useFingerprints) {
139
- this._initializeFingerprinting();
157
+ if (useFingerprints) {
158
+ this.#initializeFingerprinting();
159
+ // The fingerprint pre-launch hook goes last because of the fingerprint cache.
160
+ // It is usual to generate proxy per browser and we want to know the proxyUrl for the caching.
161
+ this.#preLaunchHooks = [...this.#preLaunchHooks, createFingerprintPreLaunchHook(this)];
162
+ this.#prePageCreateHooks = [createPrePageCreateHook(), ...this.#prePageCreateHooks];
163
+ this.#postPageCreateHooks = [
164
+ createPostPageCreateHook(this.fingerprintInjector),
165
+ ...this.#postPageCreateHooks,
166
+ ];
140
167
  }
141
168
  }
142
169
  /**
143
170
  * Opens a new page in one of the running browsers or launches
144
171
  * a new browser and opens a page there, if no browsers are active,
145
172
  * or their page limits have been exceeded.
173
+ *
174
+ * **Session injection (best-effort):** When a {@link NewPageOptions.session|session} is
175
+ * provided, this implementation uses it as a cache key for browser fingerprints (when
176
+ * fingerprinting is enabled) and reads
177
+ * {@link ProxyInfo.url|session.proxyInfo.url} /
178
+ * {@link ProxyInfo.ignoreTlsErrors|session.proxyInfo.ignoreTlsErrors} as defaults
179
+ * for `proxyUrl` and `ignoreTlsErrors` respectively. Explicit `proxyUrl` /
180
+ * `ignoreTlsErrors` values in the options take precedence.
181
+ *
182
+ * Beyond fingerprint caching and proxy configuration, no other session
183
+ * properties are consumed — cookie and header injection remain the
184
+ * crawler's responsibility.
146
185
  */
147
186
  async newPage(options = {}) {
148
- const { id = nanoid(), pageOptions, browserPlugin = this._pickBrowserPlugin(), proxyUrl, proxyTier } = options;
187
+ const { id = nanoid(), pageOptions, browserPlugin = this.#pickBrowserPlugin(), session, proxyUrl = session?.proxyInfo?.url, ignoreTlsErrors = session?.proxyInfo?.ignoreTlsErrors, } = options;
149
188
  if (this.pages.has(id)) {
150
189
  throw new Error(`Page with ID: ${id} already exists.`);
151
190
  }
@@ -153,12 +192,16 @@ export class BrowserPool extends TypedEmitter {
153
192
  throw new Error('Provided browserPlugin is not one of the plugins used by BrowserPool.');
154
193
  }
155
194
  // Limiter is necessary - https://github.com/apify/crawlee/issues/1126
156
- return this.limiter(async () => {
157
- let browserController = this._pickBrowserWithFreeCapacity(browserPlugin, { proxyTier, proxyUrl });
158
- if (!browserController)
159
- browserController = await this._launchBrowser(id, { browserPlugin, proxyTier, proxyUrl });
195
+ return this.#limiter(async () => {
196
+ let browserController = this.#pickBrowserWithFreeCapacity(browserPlugin, { proxyUrl });
197
+ // Checked under the limiter so concurrent waiters can't all pass on the same freed slot.
198
+ while (!browserController && !this.hasFreeBrowserSlot()) {
199
+ await this.#nextCapacityChange();
200
+ browserController = this.#pickBrowserWithFreeCapacity(browserPlugin, { proxyUrl });
201
+ }
202
+ browserController ??= await this.#launchBrowser(id, { browserPlugin, proxyUrl, ignoreTlsErrors });
160
203
  tryCancel();
161
- return await this._createPageForBrowser(id, browserController, pageOptions, proxyUrl);
204
+ return await this.#createPageForBrowser(id, browserController, pageOptions, proxyUrl, ignoreTlsErrors);
162
205
  });
163
206
  }
164
207
  /**
@@ -167,13 +210,13 @@ export class BrowserPool extends TypedEmitter {
167
210
  * configure the new browser.
168
211
  */
169
212
  async newPageInNewBrowser(options = {}) {
170
- const { id = nanoid(), pageOptions, launchOptions, browserPlugin = this._pickBrowserPlugin() } = options;
213
+ const { id = nanoid(), pageOptions, launchOptions, browserPlugin = this.#pickBrowserPlugin() } = options;
171
214
  if (this.pages.has(id)) {
172
215
  throw new Error(`Page with ID: ${id} already exists.`);
173
216
  }
174
- const browserController = await this._launchBrowser(id, { launchOptions, browserPlugin });
217
+ const browserController = await this.#launchBrowser(id, { launchOptions, browserPlugin });
175
218
  tryCancel();
176
- return await this._createPageForBrowser(id, browserController, pageOptions);
219
+ return await this.#createPageForBrowser(id, browserController, pageOptions);
177
220
  }
178
221
  /**
179
222
  * Opens new pages with all available plugins and returns an array
@@ -219,7 +262,7 @@ export class BrowserPool extends TypedEmitter {
219
262
  * @param page - Browser plugin page
220
263
  */
221
264
  getBrowserControllerByPage(page) {
222
- return this.pageToBrowserController.get(page);
265
+ return this.#pageToBrowserController.get(page);
223
266
  }
224
267
  /**
225
268
  * If you provided a custom ID to one of your pages or saved the
@@ -237,40 +280,45 @@ export class BrowserPool extends TypedEmitter {
237
280
  * until it's closed.
238
281
  */
239
282
  getPageId(page) {
240
- return this.pageIds.get(page);
283
+ return this.#pageIds.get(page);
241
284
  }
242
- async _createPageForBrowser(pageId, browserController, pageOptions = {}, proxyUrl) {
285
+ async #createPageForBrowser(pageId, browserController, pageOptions = {}, proxyUrl, ignoreTlsErrors) {
243
286
  // This is needed for concurrent newPage calls to wait for the browser launch.
244
287
  // It's not ideal though, we need to come up with a better API.
245
- // eslint-disable-next-line dot-notation -- accessing private property
246
- await browserController['isActivePromise'];
288
+ await browserController.waitForActive();
247
289
  tryCancel();
248
290
  const finalPageOptions = browserController.launchContext.useIncognitoPages ? pageOptions : undefined;
249
291
  if (finalPageOptions) {
250
292
  Object.assign(finalPageOptions, browserController.normalizeProxyOptions(proxyUrl, pageOptions));
293
+ if (ignoreTlsErrors) {
294
+ Object.assign(finalPageOptions, {
295
+ ignoreHTTPSErrors: true,
296
+ acceptInsecureCerts: true,
297
+ });
298
+ }
251
299
  }
252
- await this._executeHooks(this.prePageCreateHooks, pageId, browserController, finalPageOptions);
300
+ await this.#executeHooks(this.#prePageCreateHooks, pageId, browserController, finalPageOptions);
253
301
  tryCancel();
254
302
  let page;
255
303
  try {
256
- page = (await addTimeoutToPromise(async () => browserController.newPage(finalPageOptions), this.operationTimeoutMillis, 'browserController.newPage() timed out.'));
304
+ page = (await addTimeoutToPromise(async () => browserController.newPage(finalPageOptions), this.#operationTimeoutMillis, 'browserController.newPage() timed out.'));
257
305
  tryCancel();
258
306
  this.pages.set(pageId, page);
259
- this.pageIds.set(page, pageId);
260
- this.pageToBrowserController.set(page, browserController);
307
+ this.#pageIds.set(page, pageId);
308
+ this.#pageToBrowserController.set(page, browserController);
261
309
  // if you synchronously trigger a lot of page launches, browser will not get retired soon enough. Not sure if it's a problem, let's monitor it.
262
- if (browserController.totalPages >= this.retireBrowserAfterPageCount) {
310
+ if (browserController.totalPages >= this.#retireBrowserAfterPageCount) {
263
311
  this.retireBrowserController(browserController);
264
312
  }
265
- this._overridePageClose(page);
313
+ this.#overridePageClose(page);
266
314
  }
267
315
  catch (err) {
268
316
  this.retireBrowserController(browserController);
269
317
  throw new Error(`browserController.newPage() failed: ${browserController.id}\nCause:${err.message}.`);
270
318
  }
271
- await this._executeHooks(this.postPageCreateHooks, page, browserController);
319
+ await this.#executeHooks(this.#postPageCreateHooks, page, browserController);
272
320
  tryCancel();
273
- this.emit("pageCreated" /* BROWSER_POOL_EVENTS.PAGE_CREATED */, page);
321
+ this.emit(BROWSER_POOL_EVENTS.PAGE_CREATED, page);
274
322
  return page;
275
323
  }
276
324
  /**
@@ -279,14 +327,14 @@ export class BrowserPool extends TypedEmitter {
279
327
  *
280
328
  */
281
329
  retireBrowserController(browserController) {
282
- const isStarting = this.startingBrowserControllers.has(browserController);
330
+ const isStarting = this.#startingBrowserControllers.has(browserController);
283
331
  const isActive = this.activeBrowserControllers.has(browserController);
284
332
  const hasBeenRetiredOrKilled = !isStarting && !isActive;
285
333
  if (hasBeenRetiredOrKilled)
286
334
  return;
287
335
  this.retiredBrowserControllers.add(browserController);
288
- this.emit("browserRetired" /* BROWSER_POOL_EVENTS.BROWSER_RETIRED */, browserController);
289
- this.startingBrowserControllers.delete(browserController);
336
+ this.emit(BROWSER_POOL_EVENTS.BROWSER_RETIRED, browserController);
337
+ this.#startingBrowserControllers.delete(browserController);
290
338
  this.activeBrowserControllers.delete(browserController);
291
339
  }
292
340
  /**
@@ -298,12 +346,68 @@ export class BrowserPool extends TypedEmitter {
298
346
  if (browserController)
299
347
  this.retireBrowserController(browserController);
300
348
  }
349
+ /**
350
+ * Releases a page back to the pool. The page is closed and, if the
351
+ * optional `error` is a {@link SessionError} (including
352
+ * {@link SessionRetiredError}), the browser controller that served
353
+ * the page is retired so that the finished session's state (cookies,
354
+ * storage, etc.) cannot leak into future sessions.
355
+ *
356
+ * This is the primary way the crawler should return pages to the pool.
357
+ *
358
+ * @param page The page to release.
359
+ * @param options.error The error that caused the page to be released, if any.
360
+ */
361
+ async closePage(page, options) {
362
+ if (options?.error instanceof SessionError) {
363
+ this.retireBrowserByPage(page);
364
+ }
365
+ // The bound on this lives in `overridePageClose`, where it covers the pool's own
366
+ // bookkeeping as well. Wrapping it here instead would time out the override and abandon
367
+ // that bookkeeping, leaving the browser holding a slot it can never get back.
368
+ await page.close();
369
+ }
370
+ /**
371
+ * Extracts the relevant state (currently just cookies) from a page via its
372
+ * owning {@link BrowserController}. Returns empty state when the page is
373
+ * no longer associated with a controller.
374
+ *
375
+ * As with {@link BrowserPool.injectPageState}, cookies are isolated per
376
+ * page only when the pool is configured with `useIncognitoPages: true`.
377
+ * With the default `useIncognitoPages: false`, the extracted cookies
378
+ * include those set by any sibling page sharing the same browser.
379
+ */
380
+ async extractPageState(page) {
381
+ const controller = this.getBrowserControllerByPage(page);
382
+ if (!controller) {
383
+ return { cookies: [] };
384
+ }
385
+ return { cookies: await controller.getCookies(page) };
386
+ }
387
+ /**
388
+ * Injects state into a page via its owning {@link BrowserController}.
389
+ *
390
+ * No-op when the page is no longer associated with a controller.
391
+ *
392
+ * Note that cookies are isolated per page only when the pool is configured
393
+ * with `useIncognitoPages: true` — each page then gets its own browser
394
+ * context. With the default `useIncognitoPages: false`, all pages in a
395
+ * browser share a single context, so injected cookies are visible to every
396
+ * page served by that browser.
397
+ */
398
+ async injectPageState(page, state) {
399
+ const controller = this.getBrowserControllerByPage(page);
400
+ if (!controller) {
401
+ return;
402
+ }
403
+ await controller.setCookies(page, state.cookies);
404
+ }
301
405
  /**
302
406
  * Removes all active browsers from the pool. The browsers will be
303
407
  * closed after all their pages are closed.
304
408
  */
305
409
  retireAllBrowsers() {
306
- [...this.startingBrowserControllers, ...this.activeBrowserControllers].forEach((controller) => {
410
+ [...this.#startingBrowserControllers, ...this.activeBrowserControllers].forEach((controller) => {
307
411
  this.retireBrowserController(controller);
308
412
  });
309
413
  }
@@ -312,71 +416,86 @@ export class BrowserPool extends TypedEmitter {
312
416
  * @return {Promise<void>}
313
417
  */
314
418
  async closeAllBrowsers() {
315
- const controllers = this._getAllBrowserControllers();
419
+ const controllers = this.#getAllBrowserControllers();
316
420
  const promises = [...controllers]
317
421
  .filter((controller) => controller.isActive)
318
422
  .map(async (controller) => controller.close());
319
423
  await Promise.all(promises);
320
424
  }
425
+ async [Symbol.asyncDispose]() {
426
+ await this.destroy();
427
+ }
321
428
  /**
322
- * Closes all managed browsers and tears down the pool.
429
+ * Closes every managed browser and empties the pool, which stays usable afterwards — a crawler releases its
430
+ * browsers when a run ends and may start another run on the same pool.
323
431
  */
324
- async destroy() {
325
- clearInterval(this.browserKillerInterval);
326
- clearInterval(this.browserRetireInterval);
327
- this.browserKillerInterval = undefined;
328
- this.browserRetireInterval = undefined;
432
+ async releaseAllBrowsers() {
329
433
  await this.closeAllBrowsers();
330
- this._teardown();
331
- }
332
- _teardown() {
333
- this.startingBrowserControllers.clear();
434
+ this.#startingBrowserControllers.clear();
334
435
  this.activeBrowserControllers.clear();
335
436
  this.retiredBrowserControllers.clear();
437
+ }
438
+ /**
439
+ * Closes all managed browsers and tears the pool down for good: its intervals are cleared without being
440
+ * re-armed and its listeners are dropped, so it cannot be used again.
441
+ */
442
+ async destroy() {
443
+ clearInterval(this.browserKillerInterval);
444
+ clearInterval(this.#browserRetireInterval);
445
+ this.browserKillerInterval = undefined;
446
+ this.#browserRetireInterval = undefined;
447
+ await this.releaseAllBrowsers();
336
448
  this.removeAllListeners();
337
449
  }
338
- _getAllBrowserControllers() {
450
+ #getAllBrowserControllers() {
339
451
  return new Set([
340
- ...this.startingBrowserControllers,
452
+ ...this.#startingBrowserControllers,
341
453
  ...this.activeBrowserControllers,
342
454
  ...this.retiredBrowserControllers,
343
455
  ]);
344
456
  }
345
- async _launchBrowser(pageId, options) {
346
- const { browserPlugin, launchOptions, proxyTier, proxyUrl } = options;
457
+ async #launchBrowser(pageId, options) {
458
+ const { browserPlugin, launchOptions, proxyUrl, ignoreTlsErrors } = options;
347
459
  const browserController = browserPlugin.createController();
348
- this.startingBrowserControllers.add(browserController);
460
+ this.#startingBrowserControllers.add(browserController);
349
461
  const launchContext = browserPlugin.createLaunchContext({
350
462
  id: pageId,
351
463
  launchOptions,
352
- proxyTier,
353
464
  proxyUrl,
354
465
  });
466
+ // Disable SSL verification for MITM proxies
467
+ if (ignoreTlsErrors) {
468
+ /**
469
+ * @see https://playwright.dev/docs/api/class-browser/#browser-new-context
470
+ * @see https://github.com/puppeteer/puppeteer/blob/main/docs/api.md
471
+ */
472
+ launchContext.launchOptions.ignoreHTTPSErrors = true;
473
+ launchContext.launchOptions.acceptInsecureCerts = true;
474
+ }
355
475
  try {
356
476
  // If the hooks or the launch fails, we need to delete the controller,
357
477
  // because otherwise it would be stuck in limbo without a browser.
358
- await this._executeHooks(this.preLaunchHooks, pageId, launchContext);
478
+ await this.#executeHooks(this.#preLaunchHooks, pageId, launchContext);
359
479
  tryCancel();
360
480
  const browser = await browserPlugin.launch(launchContext);
361
481
  tryCancel();
362
482
  browserController.assignBrowser(browser, launchContext);
363
483
  }
364
484
  catch (err) {
365
- this.startingBrowserControllers.delete(browserController);
485
+ this.#startingBrowserControllers.delete(browserController);
366
486
  throw err;
367
487
  }
368
- log.debug('Launched new browser.', { id: browserController.id });
369
- browserController.proxyTier = proxyTier;
488
+ this.#log.debug('Launched new browser.', { id: browserController.id });
370
489
  browserController.proxyUrl = proxyUrl;
371
490
  try {
372
491
  // If the launch fails on the post-launch hooks, we need to clean up
373
492
  // both the controller and the browser before throwing.
374
- await this._executeHooks(this.postLaunchHooks, pageId, browserController);
493
+ await this.#executeHooks(this.#postLaunchHooks, pageId, browserController);
375
494
  }
376
495
  catch (err) {
377
- this.startingBrowserControllers.delete(browserController);
496
+ this.#startingBrowserControllers.delete(browserController);
378
497
  browserController.close().catch((closeErr) => {
379
- log.error(`Could not close browser whose post-launch hooks failed.\nCause:${closeErr.message}`, {
498
+ this.#log.error(`Could not close browser whose post-launch hooks failed.\nCause:${closeErr.message}`, {
380
499
  id: browserController.id,
381
500
  });
382
501
  });
@@ -384,105 +503,157 @@ export class BrowserPool extends TypedEmitter {
384
503
  }
385
504
  tryCancel();
386
505
  browserController.activate();
387
- this.startingBrowserControllers.delete(browserController);
506
+ this.#startingBrowserControllers.delete(browserController);
388
507
  this.activeBrowserControllers.add(browserController);
389
- this.emit("browserLaunched" /* BROWSER_POOL_EVENTS.BROWSER_LAUNCHED */, browserController);
508
+ this.emit(BROWSER_POOL_EVENTS.BROWSER_LAUNCHED, browserController);
390
509
  return browserController;
391
510
  }
392
511
  /**
393
512
  * Picks plugins round robin.
394
- * @private
395
513
  */
396
- _pickBrowserPlugin() {
397
- const pluginIndex = this.pageCounter % this.browserPlugins.length;
398
- this.pageCounter++;
514
+ #pickBrowserPlugin() {
515
+ const pluginIndex = this.#pageCounter % this.browserPlugins.length;
516
+ this.#pageCounter++;
399
517
  return this.browserPlugins[pluginIndex];
400
518
  }
401
- _pickBrowserWithFreeCapacity(browserPlugin, options) {
519
+ #pickBrowserWithFreeCapacity(browserPlugin, options) {
402
520
  return [...this.activeBrowserControllers].find((controller) => {
403
- const hasCapacity = controller.activePages < this.maxOpenPagesPerBrowser;
521
+ const hasCapacity = controller.activePages < this.#maxOpenPagesPerBrowser;
404
522
  const isCorrectPlugin = controller.browserPlugin === browserPlugin;
405
523
  const isSameProxyUrl = controller.proxyUrl === options?.proxyUrl;
406
- const isCorrectProxyTier = controller.proxyTier === options?.proxyTier;
407
524
  return (isCorrectPlugin &&
408
525
  hasCapacity &&
409
- ((!controller.launchContext.browserPerProxy && !options?.proxyTier) ||
410
- (options?.proxyTier && isCorrectProxyTier) ||
526
+ (!controller.launchContext.browserPerProxy ||
411
527
  (options?.proxyUrl && isSameProxyUrl) ||
412
- (!options?.proxyUrl && !options?.proxyTier && !controller.proxyUrl && !controller.proxyTier)));
528
+ (!options?.proxyUrl && !controller.proxyUrl)));
413
529
  });
414
530
  }
415
- async _closeInactiveRetiredBrowsers() {
531
+ // oxlint-disable-next-line crawlee/prefer-private-fields -- accessed by tests
532
+ async closeInactiveRetiredBrowsers() {
416
533
  const closedBrowserIds = [];
417
534
  for (const controller of this.retiredBrowserControllers) {
418
535
  const millisSinceLastPageOpened = Date.now() - controller.lastPageOpenedAt;
419
- const isBrowserIdle = millisSinceLastPageOpened >= this.closeInactiveBrowserAfterMillis;
536
+ const isBrowserIdle = millisSinceLastPageOpened >= this.#closeInactiveBrowserAfterMillis;
420
537
  const isBrowserEmpty = controller.activePages === 0;
421
538
  if (isBrowserIdle || isBrowserEmpty) {
422
539
  const { id } = controller;
423
- log.debug('Closing retired browser.', { id });
540
+ this.#log.debug('Closing retired browser.', { id });
424
541
  await controller.close();
425
542
  this.retiredBrowserControllers.delete(controller);
426
543
  closedBrowserIds.push(id);
427
544
  }
428
545
  }
429
546
  if (closedBrowserIds.length) {
430
- log.debug('Closed retired browsers.', {
547
+ this.#log.debug('Closed retired browsers.', {
431
548
  count: closedBrowserIds.length,
432
549
  closedBrowserIds,
433
550
  });
434
551
  }
435
552
  }
436
- _overridePageClose(page) {
553
+ #overridePageClose(page) {
437
554
  const originalPageClose = page.close;
438
- const browserController = this.pageToBrowserController.get(page);
555
+ const browserController = this.#pageToBrowserController.get(page);
439
556
  const pageId = this.getPageId(page);
440
557
  page.close = async (...args) => {
441
- await this._executeHooks(this.prePageCloseHooks, page, browserController);
442
- await originalPageClose.apply(page, args).catch((err) => {
443
- log.debug(`Could not close page.\nCause:${err.message}`, { id: browserController.id });
444
- });
445
- await this._executeHooks(this.postPageCloseHooks, pageId, browserController);
558
+ // A browser can acknowledge the close and then never destroy the target, in which case
559
+ // this never settles and the page's own `close` event never fires either (Chromium
560
+ // 536385539). Bound the whole sequence so nothing here can hang the caller, then
561
+ // reconcile the pool regardless of the outcome, so a page that refuses to close cannot
562
+ // leave the pool believing it is still open. `Promise.race` rather than
563
+ // `addTimeoutToPromise`: the latter inherits its AbortController from the calling frame,
564
+ // so a timeout here would cancel the task of whoever awaited `page.close()`.
565
+ let pageClosed = false;
566
+ const closing = (async () => {
567
+ await this.#executeHooks(this.#prePageCloseHooks, page, browserController);
568
+ await originalPageClose.apply(page, args).catch((err) => {
569
+ this.#log.debug(`Could not close page.\nCause:${err.message}`, { id: browserController.id });
570
+ });
571
+ // Whether the page itself is gone. Tracked separately from the sequence finishing,
572
+ // so that a slow hook does not get the browser retired.
573
+ pageClosed = true;
574
+ await this.#executeHooks(this.#postPageCloseHooks, pageId, browserController);
575
+ })();
576
+ let timeout;
577
+ const finished = await Promise.race([
578
+ closing.then(() => true, (err) => {
579
+ this.#log.warning(`Closing a page failed, releasing it from the pool anyway.\nCause:${err.message}`, { id: browserController.id, pageId });
580
+ return true;
581
+ }),
582
+ new Promise((resolve) => {
583
+ timeout = setTimeout(() => resolve(false), PAGE_CLOSE_TIMEOUT_MILLIS);
584
+ }),
585
+ ]);
586
+ clearTimeout(timeout);
587
+ if (!finished) {
588
+ this.#log.warning(`Closing a page did not finish within ${PAGE_CLOSE_TIMEOUT_MILLIS / 1000} seconds, ` +
589
+ 'releasing it from the pool anyway.', { id: browserController.id, pageId });
590
+ }
591
+ if (!pageClosed) {
592
+ // The page is still attached, so this browser cannot be trusted with more work.
593
+ // Retiring it lets the reclamation below close the process once its pages drain.
594
+ this.retireBrowserController(browserController);
595
+ }
596
+ browserController.registerPageClosed(page);
446
597
  this.pages.delete(pageId);
447
- this._closeRetiredBrowserWithNoPages(browserController);
448
- this.emit("pageClosed" /* BROWSER_POOL_EVENTS.PAGE_CLOSED */, page);
598
+ this.#closeRetiredBrowserWithNoPages(browserController);
599
+ this.emit(BROWSER_POOL_EVENTS.PAGE_CLOSED, page);
449
600
  };
450
601
  }
451
- async _executeHooks(hooks, ...args) {
602
+ async #executeHooks(hooks, ...args) {
452
603
  for (const hook of hooks) {
453
604
  await hook(...args);
454
605
  }
455
606
  }
456
- _closeRetiredBrowserWithNoPages(browserController) {
607
+ #closeRetiredBrowserWithNoPages(browserController) {
457
608
  if (browserController.activePages === 0 && this.retiredBrowserControllers.has(browserController)) {
458
609
  // Run this with a delay, otherwise page.close()
459
610
  // might fail with "Protocol error (Target.closeTarget): Target closed."
460
611
  setTimeout(() => {
461
- log.debug('Closing retired browser because it has no active pages', { id: browserController.id });
612
+ this.#log.debug('Closing retired browser because it has no active pages', { id: browserController.id });
462
613
  void browserController.close().finally(() => {
463
614
  this.retiredBrowserControllers.delete(browserController);
464
615
  });
465
616
  }, PAGE_CLOSE_KILL_TIMEOUT_MILLIS);
466
617
  }
467
618
  }
468
- _initializeFingerprinting() {
619
+ /**
620
+ * Returns `true` if the pool can accept a new browser launch without exceeding `maxOpenBrowsers`.
621
+ * Counts starting, active, and retired browsers.
622
+ *
623
+ * @internal
624
+ */
625
+ hasFreeBrowserSlot() {
626
+ const total = this.#startingBrowserControllers.size +
627
+ this.activeBrowserControllers.size +
628
+ this.retiredBrowserControllers.size;
629
+ return total < this.#maxOpenBrowsers;
630
+ }
631
+ /**
632
+ * Resolves on the next browser-retired / page-closed event, or after a short fallback interval. All
633
+ * concurrently-waiting `newPage` calls share a single promise (and a single pair of event listeners) per tick.
634
+ */
635
+ #nextCapacityChange() {
636
+ this.#capacityChange ??= new Promise((resolve) => {
637
+ const done = () => {
638
+ clearTimeout(timer);
639
+ this.off(BROWSER_POOL_EVENTS.BROWSER_RETIRED, done);
640
+ this.off(BROWSER_POOL_EVENTS.PAGE_CLOSED, done);
641
+ this.#capacityChange = undefined;
642
+ resolve();
643
+ };
644
+ const timer = setTimeout(done, SLOT_POLL_INTERVAL_MILLIS);
645
+ timer.unref?.();
646
+ this.once(BROWSER_POOL_EVENTS.BROWSER_RETIRED, done);
647
+ this.once(BROWSER_POOL_EVENTS.PAGE_CLOSED, done);
648
+ });
649
+ return this.#capacityChange;
650
+ }
651
+ #initializeFingerprinting() {
469
652
  const { useFingerprintCache = true, fingerprintCacheSize = 10_000 } = this.fingerprintOptions;
470
653
  this.fingerprintGenerator = new FingerprintGenerator(this.fingerprintOptions.fingerprintGeneratorOptions);
471
654
  this.fingerprintInjector = new FingerprintInjector();
472
655
  if (useFingerprintCache) {
473
656
  this.fingerprintCache = new QuickLRU({ maxSize: fingerprintCacheSize });
474
657
  }
475
- this._addFingerprintHooks();
476
- }
477
- _addFingerprintHooks() {
478
- this.preLaunchHooks = [
479
- ...this.preLaunchHooks,
480
- // This is flipped because of the fingerprint cache.
481
- // It is usual to generate proxy per browser and we want to know the proxyUrl for the caching.
482
- createFingerprintPreLaunchHook(this),
483
- ];
484
- this.prePageCreateHooks = [createPrePageCreateHook(), ...this.prePageCreateHooks];
485
- this.postPageCreateHooks = [createPostPageCreateHook(this.fingerprintInjector), ...this.postPageCreateHooks];
486
658
  }
487
659
  }
488
- //# sourceMappingURL=browser-pool.js.map