@crawlee/browser-pool 4.0.0-beta.18 → 4.0.0-beta.181
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.
- package/README.md +14 -14
- package/abstract-classes/browser-controller.d.ts +91 -14
- package/abstract-classes/browser-controller.js +61 -24
- package/abstract-classes/browser-plugin.d.ts +48 -17
- package/abstract-classes/browser-plugin.js +69 -13
- package/anonymize-proxy.d.ts +5 -2
- package/anonymize-proxy.js +9 -5
- package/browser-pool.d.ts +106 -29
- package/browser-pool.js +247 -86
- package/container-proxy-server.d.ts +0 -1
- package/container-proxy-server.js +0 -1
- package/events.d.ts +2 -3
- package/events.js +0 -1
- package/fingerprinting/hooks.d.ts +0 -1
- package/fingerprinting/hooks.js +26 -10
- package/fingerprinting/types.d.ts +3 -4
- package/fingerprinting/types.js +0 -1
- package/fingerprinting/utils.d.ts +0 -1
- package/fingerprinting/utils.js +5 -6
- package/index.d.ts +14 -7
- package/index.js +5 -3
- package/launch-context.d.ts +20 -5
- package/launch-context.js +21 -10
- package/package.json +9 -9
- package/playwright/playwright-browser.d.ts +2 -6
- package/playwright/playwright-browser.js +16 -17
- package/playwright/playwright-controller.d.ts +0 -1
- package/playwright/playwright-controller.js +9 -6
- package/playwright/playwright-plugin.d.ts +10 -7
- package/playwright/playwright-plugin.js +40 -16
- package/proxy-server.d.ts +0 -1
- package/proxy-server.js +0 -1
- package/puppeteer/puppeteer-controller.d.ts +0 -1
- package/puppeteer/puppeteer-controller.js +28 -20
- package/puppeteer/puppeteer-plugin.d.ts +5 -3
- package/puppeteer/puppeteer-plugin.js +83 -46
- package/remote-browser-pool.d.ts +170 -0
- package/remote-browser-pool.js +202 -0
- package/remote-browser-provider.d.ts +83 -0
- package/remote-browser-provider.js +67 -0
- package/utils.d.ts +7 -1
- package/utils.js +19 -1
- package/abstract-classes/browser-controller.d.ts.map +0 -1
- package/abstract-classes/browser-controller.js.map +0 -1
- package/abstract-classes/browser-plugin.d.ts.map +0 -1
- package/abstract-classes/browser-plugin.js.map +0 -1
- package/anonymize-proxy.d.ts.map +0 -1
- package/anonymize-proxy.js.map +0 -1
- package/browser-pool.d.ts.map +0 -1
- package/browser-pool.js.map +0 -1
- package/container-proxy-server.d.ts.map +0 -1
- package/container-proxy-server.js.map +0 -1
- package/events.d.ts.map +0 -1
- package/events.js.map +0 -1
- package/fingerprinting/hooks.d.ts.map +0 -1
- package/fingerprinting/hooks.js.map +0 -1
- package/fingerprinting/types.d.ts.map +0 -1
- package/fingerprinting/types.js.map +0 -1
- package/fingerprinting/utils.d.ts.map +0 -1
- package/fingerprinting/utils.js.map +0 -1
- package/index.d.ts.map +0 -1
- package/index.js.map +0 -1
- package/launch-context.d.ts.map +0 -1
- package/launch-context.js.map +0 -1
- package/logger.d.ts +0 -3
- package/logger.d.ts.map +0 -1
- package/logger.js +0 -5
- package/logger.js.map +0 -1
- package/playwright/playwright-browser.d.ts.map +0 -1
- package/playwright/playwright-browser.js.map +0 -1
- package/playwright/playwright-controller.d.ts.map +0 -1
- package/playwright/playwright-controller.js.map +0 -1
- package/playwright/playwright-plugin.d.ts.map +0 -1
- package/playwright/playwright-plugin.js.map +0 -1
- package/proxy-server.d.ts.map +0 -1
- package/proxy-server.js.map +0 -1
- package/puppeteer/puppeteer-controller.d.ts.map +0 -1
- package/puppeteer/puppeteer-controller.js.map +0 -1
- package/puppeteer/puppeteer-plugin.d.ts.map +0 -1
- package/puppeteer/puppeteer-plugin.js.map +0 -1
- package/utils.d.ts.map +0 -1
- package/utils.js.map +0 -1
package/browser-pool.js
CHANGED
|
@@ -1,15 +1,35 @@
|
|
|
1
|
+
import { AsyncResource } from 'node:async_hooks';
|
|
2
|
+
import { SessionError, serviceLocator } from '@crawlee/core';
|
|
3
|
+
import { parseArgument, schemas } from '@crawlee/utils/internal';
|
|
1
4
|
import { FingerprintGenerator } from 'fingerprint-generator';
|
|
2
5
|
import { FingerprintInjector } from 'fingerprint-injector';
|
|
3
6
|
import { nanoid } from 'nanoid';
|
|
4
|
-
import ow from 'ow';
|
|
5
7
|
import pLimit from 'p-limit';
|
|
6
8
|
import QuickLRU from 'quick-lru';
|
|
7
9
|
import { TypedEmitter } from 'tiny-typed-emitter';
|
|
10
|
+
import { z } from 'zod';
|
|
8
11
|
import { addTimeoutToPromise, tryCancel } from '@apify/timeout';
|
|
12
|
+
import { BROWSER_POOL_EVENTS } from './events.js';
|
|
9
13
|
import { createFingerprintPreLaunchHook, createPostPageCreateHook, createPrePageCreateHook, } from './fingerprinting/hooks.js';
|
|
10
|
-
|
|
14
|
+
const PAGE_CLOSE_TIMEOUT_MILLIS = 5000;
|
|
11
15
|
const PAGE_CLOSE_KILL_TIMEOUT_MILLIS = 1000;
|
|
12
16
|
const BROWSER_KILLER_INTERVAL_MILLIS = 10 * 1000;
|
|
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
|
+
retireBrowserAfterPageCount: schemas.anyNumber.default(100),
|
|
21
|
+
operationTimeoutSecs: schemas.anyNumber.default(15),
|
|
22
|
+
closeInactiveBrowserAfterSecs: schemas.anyNumber.default(300),
|
|
23
|
+
retireInactiveBrowserAfterSecs: schemas.anyNumber.default(10),
|
|
24
|
+
preLaunchHooks: schemas.anyArray.default(() => []),
|
|
25
|
+
postLaunchHooks: schemas.anyArray.default(() => []),
|
|
26
|
+
prePageCreateHooks: schemas.anyArray.default(() => []),
|
|
27
|
+
postPageCreateHooks: schemas.anyArray.default(() => []),
|
|
28
|
+
prePageCloseHooks: schemas.anyArray.default(() => []),
|
|
29
|
+
postPageCloseHooks: schemas.anyArray.default(() => []),
|
|
30
|
+
useFingerprints: z.boolean().default(true),
|
|
31
|
+
fingerprintOptions: schemas.anyObject.default(() => ({})),
|
|
32
|
+
});
|
|
13
33
|
/**
|
|
14
34
|
* The `BrowserPool` class is the most important class of the `browser-pool` module.
|
|
15
35
|
* It manages opening and closing of browsers and their pages and its constructor
|
|
@@ -60,6 +80,7 @@ const BROWSER_KILLER_INTERVAL_MILLIS = 10 * 1000;
|
|
|
60
80
|
export class BrowserPool extends TypedEmitter {
|
|
61
81
|
browserPlugins;
|
|
62
82
|
maxOpenPagesPerBrowser;
|
|
83
|
+
maxOpenBrowsers;
|
|
63
84
|
retireBrowserAfterPageCount;
|
|
64
85
|
operationTimeoutMillis;
|
|
65
86
|
closeInactiveBrowserAfterMillis;
|
|
@@ -81,29 +102,16 @@ export class BrowserPool extends TypedEmitter {
|
|
|
81
102
|
fingerprintInjector;
|
|
82
103
|
fingerprintGenerator;
|
|
83
104
|
fingerprintCache;
|
|
84
|
-
|
|
85
|
-
|
|
86
|
-
|
|
105
|
+
// kept as TS-private: tests replace this interval through bracket access
|
|
106
|
+
browserKillerInterval = setInterval(async () => this.closeInactiveRetiredBrowsers(), BROWSER_KILLER_INTERVAL_MILLIS);
|
|
107
|
+
#browserRetireInterval;
|
|
108
|
+
#limiter = pLimit(1);
|
|
109
|
+
#log;
|
|
87
110
|
constructor(options) {
|
|
88
111
|
super();
|
|
112
|
+
this.#log = serviceLocator.getLogger().child({ prefix: 'BrowserPool' });
|
|
89
113
|
this.browserKillerInterval.unref();
|
|
90
|
-
|
|
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;
|
|
114
|
+
const { browserPlugins, maxOpenPagesPerBrowser, retireBrowserAfterPageCount, operationTimeoutSecs, closeInactiveBrowserAfterSecs, retireInactiveBrowserAfterSecs, preLaunchHooks, postLaunchHooks, prePageCreateHooks, postPageCreateHooks, prePageCloseHooks, postPageCloseHooks, useFingerprints, fingerprintOptions, } = parseArgument(options, browserPoolOptionsSchema);
|
|
107
115
|
const firstPluginConstructor = browserPlugins[0].constructor;
|
|
108
116
|
for (let i = 1; i < browserPlugins.length; i++) {
|
|
109
117
|
const providedPlugin = browserPlugins[i];
|
|
@@ -115,18 +123,19 @@ export class BrowserPool extends TypedEmitter {
|
|
|
115
123
|
}
|
|
116
124
|
this.browserPlugins = browserPlugins;
|
|
117
125
|
this.maxOpenPagesPerBrowser = maxOpenPagesPerBrowser;
|
|
126
|
+
this.maxOpenBrowsers = Infinity;
|
|
118
127
|
this.retireBrowserAfterPageCount = retireBrowserAfterPageCount;
|
|
119
128
|
this.operationTimeoutMillis = operationTimeoutSecs * 1000;
|
|
120
129
|
this.closeInactiveBrowserAfterMillis = closeInactiveBrowserAfterSecs * 1000;
|
|
121
130
|
this.useFingerprints = useFingerprints;
|
|
122
131
|
this.fingerprintOptions = fingerprintOptions;
|
|
123
|
-
this
|
|
132
|
+
this.#browserRetireInterval = setInterval(async () => this.activeBrowserControllers.forEach((controller) => {
|
|
124
133
|
if (controller.activePages === 0 &&
|
|
125
134
|
controller.lastPageOpenedAt < Date.now() - retireInactiveBrowserAfterSecs * 1000) {
|
|
126
135
|
this.retireBrowserController(controller);
|
|
127
136
|
}
|
|
128
137
|
}), retireInactiveBrowserAfterSecs * 1000);
|
|
129
|
-
this
|
|
138
|
+
this.#browserRetireInterval.unref();
|
|
130
139
|
// hooks
|
|
131
140
|
this.preLaunchHooks = preLaunchHooks;
|
|
132
141
|
this.postLaunchHooks = postLaunchHooks;
|
|
@@ -136,30 +145,56 @@ export class BrowserPool extends TypedEmitter {
|
|
|
136
145
|
this.postPageCloseHooks = postPageCloseHooks;
|
|
137
146
|
// fingerprinting
|
|
138
147
|
if (this.useFingerprints) {
|
|
139
|
-
this.
|
|
148
|
+
this.initializeFingerprinting();
|
|
140
149
|
}
|
|
141
150
|
}
|
|
142
151
|
/**
|
|
143
152
|
* Opens a new page in one of the running browsers or launches
|
|
144
153
|
* a new browser and opens a page there, if no browsers are active,
|
|
145
154
|
* or their page limits have been exceeded.
|
|
155
|
+
*
|
|
156
|
+
* **Session injection (best-effort):** When a {@link NewPageOptions.session|session} is
|
|
157
|
+
* provided, this implementation uses it as a cache key for browser fingerprints (when
|
|
158
|
+
* fingerprinting is enabled) and reads
|
|
159
|
+
* {@link ProxyInfo.url|session.proxyInfo.url} /
|
|
160
|
+
* {@link ProxyInfo.ignoreTlsErrors|session.proxyInfo.ignoreTlsErrors} as defaults
|
|
161
|
+
* for `proxyUrl` and `ignoreTlsErrors` respectively. Explicit `proxyUrl` /
|
|
162
|
+
* `ignoreTlsErrors` values in the options take precedence.
|
|
163
|
+
*
|
|
164
|
+
* Beyond fingerprint caching and proxy configuration, no other session
|
|
165
|
+
* properties are consumed — cookie and header injection remain the
|
|
166
|
+
* crawler's responsibility.
|
|
146
167
|
*/
|
|
147
168
|
async newPage(options = {}) {
|
|
148
|
-
const { id = nanoid(), pageOptions, browserPlugin = this.
|
|
169
|
+
const { id = nanoid(), pageOptions, browserPlugin = this.pickBrowserPlugin(), session, proxyUrl = session?.proxyInfo?.url, ignoreTlsErrors = session?.proxyInfo?.ignoreTlsErrors, } = options;
|
|
149
170
|
if (this.pages.has(id)) {
|
|
150
171
|
throw new Error(`Page with ID: ${id} already exists.`);
|
|
151
172
|
}
|
|
152
173
|
if (browserPlugin && !this.browserPlugins.includes(browserPlugin)) {
|
|
153
174
|
throw new Error('Provided browserPlugin is not one of the plugins used by BrowserPool.');
|
|
154
175
|
}
|
|
176
|
+
// Bind the limiter callback to the current async-hooks context. p-limit
|
|
177
|
+
// otherwise resumes queued callbacks in the previous task's
|
|
178
|
+
// AsyncLocalStorage context, leaking aborted cancelTasks across unrelated
|
|
179
|
+
// requests (https://github.com/apify/crawlee/issues/3670). Mirrors the
|
|
180
|
+
// fix p-limit landed upstream in v5 (sindresorhus/p-limit#71); v5 is an
|
|
181
|
+
// ESM-only rewrite, so we can't bump it in Crawlee v3.
|
|
182
|
+
// Besides the cancelTask leak, the wrapper also keeps the per-request *storage transaction*
|
|
183
|
+
// ALS-scoped: without it, a queued callback would resume in the previous request's async
|
|
184
|
+
// context and run request B's storage writes inside request A's transaction.
|
|
185
|
+
// TODO(crawlee@v4): bump p-limit to v5 and drop this AsyncResource.bind wrapper.
|
|
155
186
|
// Limiter is necessary - https://github.com/apify/crawlee/issues/1126
|
|
156
|
-
return this
|
|
157
|
-
let browserController = this.
|
|
187
|
+
return this.#limiter(AsyncResource.bind(async () => {
|
|
188
|
+
let browserController = this.pickBrowserWithFreeCapacity(browserPlugin, { proxyUrl });
|
|
158
189
|
if (!browserController)
|
|
159
|
-
browserController = await this.
|
|
190
|
+
browserController = await this.launchBrowser(id, {
|
|
191
|
+
browserPlugin,
|
|
192
|
+
proxyUrl,
|
|
193
|
+
ignoreTlsErrors,
|
|
194
|
+
});
|
|
160
195
|
tryCancel();
|
|
161
|
-
return await this.
|
|
162
|
-
});
|
|
196
|
+
return await this.createPageForBrowser(id, browserController, pageOptions, proxyUrl, ignoreTlsErrors);
|
|
197
|
+
}));
|
|
163
198
|
}
|
|
164
199
|
/**
|
|
165
200
|
* Unlike {@link newPage}, `newPageInNewBrowser` always launches a new
|
|
@@ -167,13 +202,13 @@ export class BrowserPool extends TypedEmitter {
|
|
|
167
202
|
* configure the new browser.
|
|
168
203
|
*/
|
|
169
204
|
async newPageInNewBrowser(options = {}) {
|
|
170
|
-
const { id = nanoid(), pageOptions, launchOptions, browserPlugin = this.
|
|
205
|
+
const { id = nanoid(), pageOptions, launchOptions, browserPlugin = this.pickBrowserPlugin() } = options;
|
|
171
206
|
if (this.pages.has(id)) {
|
|
172
207
|
throw new Error(`Page with ID: ${id} already exists.`);
|
|
173
208
|
}
|
|
174
|
-
const browserController = await this.
|
|
209
|
+
const browserController = await this.launchBrowser(id, { launchOptions, browserPlugin });
|
|
175
210
|
tryCancel();
|
|
176
|
-
return await this.
|
|
211
|
+
return await this.createPageForBrowser(id, browserController, pageOptions);
|
|
177
212
|
}
|
|
178
213
|
/**
|
|
179
214
|
* Opens new pages with all available plugins and returns an array
|
|
@@ -239,17 +274,22 @@ export class BrowserPool extends TypedEmitter {
|
|
|
239
274
|
getPageId(page) {
|
|
240
275
|
return this.pageIds.get(page);
|
|
241
276
|
}
|
|
242
|
-
async
|
|
277
|
+
async createPageForBrowser(pageId, browserController, pageOptions = {}, proxyUrl, ignoreTlsErrors) {
|
|
243
278
|
// This is needed for concurrent newPage calls to wait for the browser launch.
|
|
244
279
|
// It's not ideal though, we need to come up with a better API.
|
|
245
|
-
|
|
246
|
-
await browserController['isActivePromise'];
|
|
280
|
+
await browserController.waitForActive();
|
|
247
281
|
tryCancel();
|
|
248
282
|
const finalPageOptions = browserController.launchContext.useIncognitoPages ? pageOptions : undefined;
|
|
249
283
|
if (finalPageOptions) {
|
|
250
284
|
Object.assign(finalPageOptions, browserController.normalizeProxyOptions(proxyUrl, pageOptions));
|
|
285
|
+
if (ignoreTlsErrors) {
|
|
286
|
+
Object.assign(finalPageOptions, {
|
|
287
|
+
ignoreHTTPSErrors: true,
|
|
288
|
+
acceptInsecureCerts: true,
|
|
289
|
+
});
|
|
290
|
+
}
|
|
251
291
|
}
|
|
252
|
-
await this.
|
|
292
|
+
await this.executeHooks(this.prePageCreateHooks, pageId, browserController, finalPageOptions);
|
|
253
293
|
tryCancel();
|
|
254
294
|
let page;
|
|
255
295
|
try {
|
|
@@ -262,15 +302,15 @@ export class BrowserPool extends TypedEmitter {
|
|
|
262
302
|
if (browserController.totalPages >= this.retireBrowserAfterPageCount) {
|
|
263
303
|
this.retireBrowserController(browserController);
|
|
264
304
|
}
|
|
265
|
-
this.
|
|
305
|
+
this.overridePageClose(page);
|
|
266
306
|
}
|
|
267
307
|
catch (err) {
|
|
268
308
|
this.retireBrowserController(browserController);
|
|
269
309
|
throw new Error(`browserController.newPage() failed: ${browserController.id}\nCause:${err.message}.`);
|
|
270
310
|
}
|
|
271
|
-
await this.
|
|
311
|
+
await this.executeHooks(this.postPageCreateHooks, page, browserController);
|
|
272
312
|
tryCancel();
|
|
273
|
-
this.emit(
|
|
313
|
+
this.emit(BROWSER_POOL_EVENTS.PAGE_CREATED, page);
|
|
274
314
|
return page;
|
|
275
315
|
}
|
|
276
316
|
/**
|
|
@@ -285,7 +325,7 @@ export class BrowserPool extends TypedEmitter {
|
|
|
285
325
|
if (hasBeenRetiredOrKilled)
|
|
286
326
|
return;
|
|
287
327
|
this.retiredBrowserControllers.add(browserController);
|
|
288
|
-
this.emit(
|
|
328
|
+
this.emit(BROWSER_POOL_EVENTS.BROWSER_RETIRED, browserController);
|
|
289
329
|
this.startingBrowserControllers.delete(browserController);
|
|
290
330
|
this.activeBrowserControllers.delete(browserController);
|
|
291
331
|
}
|
|
@@ -298,6 +338,61 @@ export class BrowserPool extends TypedEmitter {
|
|
|
298
338
|
if (browserController)
|
|
299
339
|
this.retireBrowserController(browserController);
|
|
300
340
|
}
|
|
341
|
+
/**
|
|
342
|
+
* Releases a page back to the pool. The page is closed and, if the
|
|
343
|
+
* optional `error` is a {@link SessionError}, the browser controller
|
|
344
|
+
* that served the page is retired so that its tainted state (cookies,
|
|
345
|
+
* storage, etc.) cannot leak into future sessions.
|
|
346
|
+
*
|
|
347
|
+
* This is the primary way the crawler should return pages to the pool.
|
|
348
|
+
*
|
|
349
|
+
* @param page The page to release.
|
|
350
|
+
* @param options.error The error that caused the page to be released, if any.
|
|
351
|
+
*/
|
|
352
|
+
async closePage(page, options) {
|
|
353
|
+
if (options?.error instanceof SessionError) {
|
|
354
|
+
this.retireBrowserByPage(page);
|
|
355
|
+
}
|
|
356
|
+
// The bound on this lives in `overridePageClose`, where it covers the pool's own
|
|
357
|
+
// bookkeeping as well. Wrapping it here instead would time out the override and abandon
|
|
358
|
+
// that bookkeeping, leaving the browser holding a slot it can never get back.
|
|
359
|
+
await page.close();
|
|
360
|
+
}
|
|
361
|
+
/**
|
|
362
|
+
* Extracts the relevant state (currently just cookies) from a page via its
|
|
363
|
+
* owning {@link BrowserController}. Returns empty state when the page is
|
|
364
|
+
* no longer associated with a controller.
|
|
365
|
+
*
|
|
366
|
+
* As with {@link BrowserPool.injectPageState}, cookies are isolated per
|
|
367
|
+
* page only when the pool is configured with `useIncognitoPages: true`.
|
|
368
|
+
* With the default `useIncognitoPages: false`, the extracted cookies
|
|
369
|
+
* include those set by any sibling page sharing the same browser.
|
|
370
|
+
*/
|
|
371
|
+
async extractPageState(page) {
|
|
372
|
+
const controller = this.getBrowserControllerByPage(page);
|
|
373
|
+
if (!controller) {
|
|
374
|
+
return { cookies: [] };
|
|
375
|
+
}
|
|
376
|
+
return { cookies: await controller.getCookies(page) };
|
|
377
|
+
}
|
|
378
|
+
/**
|
|
379
|
+
* Injects state into a page via its owning {@link BrowserController}.
|
|
380
|
+
*
|
|
381
|
+
* No-op when the page is no longer associated with a controller.
|
|
382
|
+
*
|
|
383
|
+
* Note that cookies are isolated per page only when the pool is configured
|
|
384
|
+
* with `useIncognitoPages: true` — each page then gets its own browser
|
|
385
|
+
* context. With the default `useIncognitoPages: false`, all pages in a
|
|
386
|
+
* browser share a single context, so injected cookies are visible to every
|
|
387
|
+
* page served by that browser.
|
|
388
|
+
*/
|
|
389
|
+
async injectPageState(page, state) {
|
|
390
|
+
const controller = this.getBrowserControllerByPage(page);
|
|
391
|
+
if (!controller) {
|
|
392
|
+
return;
|
|
393
|
+
}
|
|
394
|
+
await controller.setCookies(page, state.cookies);
|
|
395
|
+
}
|
|
301
396
|
/**
|
|
302
397
|
* Removes all active browsers from the pool. The browsers will be
|
|
303
398
|
* closed after all their pages are closed.
|
|
@@ -312,50 +407,66 @@ export class BrowserPool extends TypedEmitter {
|
|
|
312
407
|
* @return {Promise<void>}
|
|
313
408
|
*/
|
|
314
409
|
async closeAllBrowsers() {
|
|
315
|
-
const controllers = this.
|
|
410
|
+
const controllers = this.getAllBrowserControllers();
|
|
316
411
|
const promises = [...controllers]
|
|
317
412
|
.filter((controller) => controller.isActive)
|
|
318
413
|
.map(async (controller) => controller.close());
|
|
319
414
|
await Promise.all(promises);
|
|
320
415
|
}
|
|
416
|
+
async [Symbol.asyncDispose]() {
|
|
417
|
+
await this.destroy();
|
|
418
|
+
}
|
|
321
419
|
/**
|
|
322
|
-
* Closes
|
|
420
|
+
* Closes every managed browser and empties the pool, which stays usable afterwards — a crawler releases its
|
|
421
|
+
* browsers when a run ends and may start another run on the same pool.
|
|
323
422
|
*/
|
|
324
|
-
async
|
|
325
|
-
clearInterval(this.browserKillerInterval);
|
|
326
|
-
clearInterval(this.browserRetireInterval);
|
|
327
|
-
this.browserKillerInterval = undefined;
|
|
328
|
-
this.browserRetireInterval = undefined;
|
|
423
|
+
async releaseAllBrowsers() {
|
|
329
424
|
await this.closeAllBrowsers();
|
|
330
|
-
this._teardown();
|
|
331
|
-
}
|
|
332
|
-
_teardown() {
|
|
333
425
|
this.startingBrowserControllers.clear();
|
|
334
426
|
this.activeBrowserControllers.clear();
|
|
335
427
|
this.retiredBrowserControllers.clear();
|
|
428
|
+
}
|
|
429
|
+
/**
|
|
430
|
+
* Closes all managed browsers and tears the pool down for good: its intervals are cleared without being
|
|
431
|
+
* re-armed and its listeners are dropped, so it cannot be used again.
|
|
432
|
+
*/
|
|
433
|
+
async destroy() {
|
|
434
|
+
clearInterval(this.browserKillerInterval);
|
|
435
|
+
clearInterval(this.#browserRetireInterval);
|
|
436
|
+
this.browserKillerInterval = undefined;
|
|
437
|
+
this.#browserRetireInterval = undefined;
|
|
438
|
+
await this.releaseAllBrowsers();
|
|
336
439
|
this.removeAllListeners();
|
|
337
440
|
}
|
|
338
|
-
|
|
441
|
+
getAllBrowserControllers() {
|
|
339
442
|
return new Set([
|
|
340
443
|
...this.startingBrowserControllers,
|
|
341
444
|
...this.activeBrowserControllers,
|
|
342
445
|
...this.retiredBrowserControllers,
|
|
343
446
|
]);
|
|
344
447
|
}
|
|
345
|
-
async
|
|
346
|
-
const { browserPlugin, launchOptions,
|
|
448
|
+
async launchBrowser(pageId, options) {
|
|
449
|
+
const { browserPlugin, launchOptions, proxyUrl, ignoreTlsErrors } = options;
|
|
347
450
|
const browserController = browserPlugin.createController();
|
|
348
451
|
this.startingBrowserControllers.add(browserController);
|
|
349
452
|
const launchContext = browserPlugin.createLaunchContext({
|
|
350
453
|
id: pageId,
|
|
351
454
|
launchOptions,
|
|
352
|
-
proxyTier,
|
|
353
455
|
proxyUrl,
|
|
354
456
|
});
|
|
457
|
+
// Disable SSL verification for MITM proxies
|
|
458
|
+
if (ignoreTlsErrors) {
|
|
459
|
+
/**
|
|
460
|
+
* @see https://playwright.dev/docs/api/class-browser/#browser-new-context
|
|
461
|
+
* @see https://github.com/puppeteer/puppeteer/blob/main/docs/api.md
|
|
462
|
+
*/
|
|
463
|
+
launchContext.launchOptions.ignoreHTTPSErrors = true;
|
|
464
|
+
launchContext.launchOptions.acceptInsecureCerts = true;
|
|
465
|
+
}
|
|
355
466
|
try {
|
|
356
467
|
// If the hooks or the launch fails, we need to delete the controller,
|
|
357
468
|
// because otherwise it would be stuck in limbo without a browser.
|
|
358
|
-
await this.
|
|
469
|
+
await this.executeHooks(this.preLaunchHooks, pageId, launchContext);
|
|
359
470
|
tryCancel();
|
|
360
471
|
const browser = await browserPlugin.launch(launchContext);
|
|
361
472
|
tryCancel();
|
|
@@ -365,18 +476,17 @@ export class BrowserPool extends TypedEmitter {
|
|
|
365
476
|
this.startingBrowserControllers.delete(browserController);
|
|
366
477
|
throw err;
|
|
367
478
|
}
|
|
368
|
-
log.debug('Launched new browser.', { id: browserController.id });
|
|
369
|
-
browserController.proxyTier = proxyTier;
|
|
479
|
+
this.#log.debug('Launched new browser.', { id: browserController.id });
|
|
370
480
|
browserController.proxyUrl = proxyUrl;
|
|
371
481
|
try {
|
|
372
482
|
// If the launch fails on the post-launch hooks, we need to clean up
|
|
373
483
|
// both the controller and the browser before throwing.
|
|
374
|
-
await this.
|
|
484
|
+
await this.executeHooks(this.postLaunchHooks, pageId, browserController);
|
|
375
485
|
}
|
|
376
486
|
catch (err) {
|
|
377
487
|
this.startingBrowserControllers.delete(browserController);
|
|
378
488
|
browserController.close().catch((closeErr) => {
|
|
379
|
-
log.error(`Could not close browser whose post-launch hooks failed.\nCause:${closeErr.message}`, {
|
|
489
|
+
this.#log.error(`Could not close browser whose post-launch hooks failed.\nCause:${closeErr.message}`, {
|
|
380
490
|
id: browserController.id,
|
|
381
491
|
});
|
|
382
492
|
});
|
|
@@ -386,33 +496,31 @@ export class BrowserPool extends TypedEmitter {
|
|
|
386
496
|
browserController.activate();
|
|
387
497
|
this.startingBrowserControllers.delete(browserController);
|
|
388
498
|
this.activeBrowserControllers.add(browserController);
|
|
389
|
-
this.emit(
|
|
499
|
+
this.emit(BROWSER_POOL_EVENTS.BROWSER_LAUNCHED, browserController);
|
|
390
500
|
return browserController;
|
|
391
501
|
}
|
|
392
502
|
/**
|
|
393
503
|
* Picks plugins round robin.
|
|
394
504
|
* @private
|
|
395
505
|
*/
|
|
396
|
-
|
|
506
|
+
pickBrowserPlugin() {
|
|
397
507
|
const pluginIndex = this.pageCounter % this.browserPlugins.length;
|
|
398
508
|
this.pageCounter++;
|
|
399
509
|
return this.browserPlugins[pluginIndex];
|
|
400
510
|
}
|
|
401
|
-
|
|
511
|
+
pickBrowserWithFreeCapacity(browserPlugin, options) {
|
|
402
512
|
return [...this.activeBrowserControllers].find((controller) => {
|
|
403
513
|
const hasCapacity = controller.activePages < this.maxOpenPagesPerBrowser;
|
|
404
514
|
const isCorrectPlugin = controller.browserPlugin === browserPlugin;
|
|
405
515
|
const isSameProxyUrl = controller.proxyUrl === options?.proxyUrl;
|
|
406
|
-
const isCorrectProxyTier = controller.proxyTier === options?.proxyTier;
|
|
407
516
|
return (isCorrectPlugin &&
|
|
408
517
|
hasCapacity &&
|
|
409
|
-
(
|
|
410
|
-
(options?.proxyTier && isCorrectProxyTier) ||
|
|
518
|
+
(!controller.launchContext.browserPerProxy ||
|
|
411
519
|
(options?.proxyUrl && isSameProxyUrl) ||
|
|
412
|
-
(!options?.proxyUrl && !
|
|
520
|
+
(!options?.proxyUrl && !controller.proxyUrl)));
|
|
413
521
|
});
|
|
414
522
|
}
|
|
415
|
-
async
|
|
523
|
+
async closeInactiveRetiredBrowsers() {
|
|
416
524
|
const closedBrowserIds = [];
|
|
417
525
|
for (const controller of this.retiredBrowserControllers) {
|
|
418
526
|
const millisSinceLastPageOpened = Date.now() - controller.lastPageOpenedAt;
|
|
@@ -420,61 +528,115 @@ export class BrowserPool extends TypedEmitter {
|
|
|
420
528
|
const isBrowserEmpty = controller.activePages === 0;
|
|
421
529
|
if (isBrowserIdle || isBrowserEmpty) {
|
|
422
530
|
const { id } = controller;
|
|
423
|
-
log.debug('Closing retired browser.', { id });
|
|
531
|
+
this.#log.debug('Closing retired browser.', { id });
|
|
424
532
|
await controller.close();
|
|
425
533
|
this.retiredBrowserControllers.delete(controller);
|
|
426
534
|
closedBrowserIds.push(id);
|
|
427
535
|
}
|
|
428
536
|
}
|
|
429
537
|
if (closedBrowserIds.length) {
|
|
430
|
-
log.debug('Closed retired browsers.', {
|
|
538
|
+
this.#log.debug('Closed retired browsers.', {
|
|
431
539
|
count: closedBrowserIds.length,
|
|
432
540
|
closedBrowserIds,
|
|
433
541
|
});
|
|
434
542
|
}
|
|
435
543
|
}
|
|
436
|
-
|
|
544
|
+
overridePageClose(page) {
|
|
437
545
|
const originalPageClose = page.close;
|
|
438
546
|
const browserController = this.pageToBrowserController.get(page);
|
|
439
547
|
const pageId = this.getPageId(page);
|
|
440
548
|
page.close = async (...args) => {
|
|
441
|
-
|
|
442
|
-
|
|
443
|
-
|
|
444
|
-
|
|
445
|
-
|
|
549
|
+
// A browser can acknowledge the close and then never destroy the target, in which case
|
|
550
|
+
// this never settles and the page's own `close` event never fires either (Chromium
|
|
551
|
+
// 536385539). Bound the whole sequence so nothing here can hang the caller, then
|
|
552
|
+
// reconcile the pool regardless of the outcome, so a page that refuses to close cannot
|
|
553
|
+
// leave the pool believing it is still open. `Promise.race` rather than
|
|
554
|
+
// `addTimeoutToPromise`: the latter inherits its AbortController from the calling frame,
|
|
555
|
+
// so a timeout here would cancel the task of whoever awaited `page.close()`.
|
|
556
|
+
let pageClosed = false;
|
|
557
|
+
const closing = (async () => {
|
|
558
|
+
await this.executeHooks(this.prePageCloseHooks, page, browserController);
|
|
559
|
+
await originalPageClose.apply(page, args).catch((err) => {
|
|
560
|
+
this.#log.debug(`Could not close page.\nCause:${err.message}`, { id: browserController.id });
|
|
561
|
+
});
|
|
562
|
+
// Whether the page itself is gone. Tracked separately from the sequence finishing,
|
|
563
|
+
// so that a slow hook does not get the browser retired.
|
|
564
|
+
pageClosed = true;
|
|
565
|
+
await this.executeHooks(this.postPageCloseHooks, pageId, browserController);
|
|
566
|
+
})();
|
|
567
|
+
let timeout;
|
|
568
|
+
const finished = await Promise.race([
|
|
569
|
+
closing.then(() => true, (err) => {
|
|
570
|
+
this.#log.warning(`Closing a page failed, releasing it from the pool anyway.\nCause:${err.message}`, { id: browserController.id, pageId });
|
|
571
|
+
return true;
|
|
572
|
+
}),
|
|
573
|
+
new Promise((resolve) => {
|
|
574
|
+
timeout = setTimeout(() => resolve(false), PAGE_CLOSE_TIMEOUT_MILLIS);
|
|
575
|
+
}),
|
|
576
|
+
]);
|
|
577
|
+
clearTimeout(timeout);
|
|
578
|
+
if (!finished) {
|
|
579
|
+
this.#log.warning(`Closing a page did not finish within ${PAGE_CLOSE_TIMEOUT_MILLIS / 1000} seconds, ` +
|
|
580
|
+
'releasing it from the pool anyway.', { id: browserController.id, pageId });
|
|
581
|
+
}
|
|
582
|
+
if (!pageClosed) {
|
|
583
|
+
// The page is still attached, so this browser cannot be trusted with more work.
|
|
584
|
+
// Retiring it lets the reclamation below close the process once its pages drain.
|
|
585
|
+
this.retireBrowserController(browserController);
|
|
586
|
+
}
|
|
587
|
+
browserController.registerPageClosed(page);
|
|
446
588
|
this.pages.delete(pageId);
|
|
447
|
-
this.
|
|
448
|
-
this.emit(
|
|
589
|
+
this.closeRetiredBrowserWithNoPages(browserController);
|
|
590
|
+
this.emit(BROWSER_POOL_EVENTS.PAGE_CLOSED, page);
|
|
449
591
|
};
|
|
450
592
|
}
|
|
451
|
-
async
|
|
593
|
+
async executeHooks(hooks, ...args) {
|
|
452
594
|
for (const hook of hooks) {
|
|
453
595
|
await hook(...args);
|
|
454
596
|
}
|
|
455
597
|
}
|
|
456
|
-
|
|
598
|
+
closeRetiredBrowserWithNoPages(browserController) {
|
|
457
599
|
if (browserController.activePages === 0 && this.retiredBrowserControllers.has(browserController)) {
|
|
458
600
|
// Run this with a delay, otherwise page.close()
|
|
459
601
|
// might fail with "Protocol error (Target.closeTarget): Target closed."
|
|
460
602
|
setTimeout(() => {
|
|
461
|
-
log.debug('Closing retired browser because it has no active pages', { id: browserController.id });
|
|
603
|
+
this.#log.debug('Closing retired browser because it has no active pages', { id: browserController.id });
|
|
462
604
|
void browserController.close().finally(() => {
|
|
463
605
|
this.retiredBrowserControllers.delete(browserController);
|
|
464
606
|
});
|
|
465
607
|
}, PAGE_CLOSE_KILL_TIMEOUT_MILLIS);
|
|
466
608
|
}
|
|
467
609
|
}
|
|
468
|
-
|
|
610
|
+
/**
|
|
611
|
+
* Returns `true` if the pool can accept a new browser launch without exceeding
|
|
612
|
+
* {@link BrowserPoolOptions.maxOpenBrowsers}. Counts starting, active, and retired browsers.
|
|
613
|
+
*/
|
|
614
|
+
hasFreeBrowserSlot() {
|
|
615
|
+
const total = this.startingBrowserControllers.size +
|
|
616
|
+
this.activeBrowserControllers.size +
|
|
617
|
+
this.retiredBrowserControllers.size;
|
|
618
|
+
return total < this.maxOpenBrowsers;
|
|
619
|
+
}
|
|
620
|
+
/**
|
|
621
|
+
* Returns `true` if any active browser has room for another page.
|
|
622
|
+
*/
|
|
623
|
+
hasActiveBrowserWithFreeCapacity() {
|
|
624
|
+
for (const controller of this.activeBrowserControllers) {
|
|
625
|
+
if (controller.activePages < this.maxOpenPagesPerBrowser)
|
|
626
|
+
return true;
|
|
627
|
+
}
|
|
628
|
+
return false;
|
|
629
|
+
}
|
|
630
|
+
initializeFingerprinting() {
|
|
469
631
|
const { useFingerprintCache = true, fingerprintCacheSize = 10_000 } = this.fingerprintOptions;
|
|
470
632
|
this.fingerprintGenerator = new FingerprintGenerator(this.fingerprintOptions.fingerprintGeneratorOptions);
|
|
471
633
|
this.fingerprintInjector = new FingerprintInjector();
|
|
472
634
|
if (useFingerprintCache) {
|
|
473
635
|
this.fingerprintCache = new QuickLRU({ maxSize: fingerprintCacheSize });
|
|
474
636
|
}
|
|
475
|
-
this.
|
|
637
|
+
this.addFingerprintHooks();
|
|
476
638
|
}
|
|
477
|
-
|
|
639
|
+
addFingerprintHooks() {
|
|
478
640
|
this.preLaunchHooks = [
|
|
479
641
|
...this.preLaunchHooks,
|
|
480
642
|
// This is flipped because of the fingerprint cache.
|
|
@@ -485,4 +647,3 @@ export class BrowserPool extends TypedEmitter {
|
|
|
485
647
|
this.postPageCreateHooks = [createPostPageCreateHook(this.fingerprintInjector), ...this.postPageCreateHooks];
|
|
486
648
|
}
|
|
487
649
|
}
|
|
488
|
-
//# sourceMappingURL=browser-pool.js.map
|
package/events.d.ts
CHANGED
|
@@ -1,11 +1,10 @@
|
|
|
1
|
-
export declare
|
|
1
|
+
export declare enum BROWSER_POOL_EVENTS {
|
|
2
2
|
BROWSER_LAUNCHED = "browserLaunched",
|
|
3
3
|
BROWSER_RETIRED = "browserRetired",
|
|
4
4
|
BROWSER_CLOSED = "browserClosed",
|
|
5
5
|
PAGE_CREATED = "pageCreated",
|
|
6
6
|
PAGE_CLOSED = "pageClosed"
|
|
7
7
|
}
|
|
8
|
-
export declare
|
|
8
|
+
export declare enum BROWSER_CONTROLLER_EVENTS {
|
|
9
9
|
BROWSER_CLOSED = "browserClosed"
|
|
10
10
|
}
|
|
11
|
-
//# sourceMappingURL=events.d.ts.map
|
package/events.js
CHANGED
|
@@ -14,4 +14,3 @@ export declare function createPrePageCreateHook(): (_pageId: string, browserCont
|
|
|
14
14
|
* @internal
|
|
15
15
|
*/
|
|
16
16
|
export declare function createPostPageCreateHook(fingerprintInjector: FingerprintInjector): (page: any, browserController: BrowserController) => Promise<void>;
|
|
17
|
-
//# sourceMappingURL=hooks.d.ts.map
|