@crawlee/browser 4.0.0-beta.2 → 4.0.0-beta.200
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 +17 -13
- package/index.d.ts +0 -1
- package/index.js +0 -1
- package/internals/browser-crawler.d.ts +277 -179
- package/internals/browser-crawler.js +312 -276
- package/internals/browser-launcher.d.ts +71 -23
- package/internals/browser-launcher.js +73 -23
- package/package.json +10 -10
- package/index.d.ts.map +0 -1
- package/index.js.map +0 -1
- package/internals/browser-crawler.d.ts.map +0 -1
- package/internals/browser-crawler.js.map +0 -1
- package/internals/browser-launcher.d.ts.map +0 -1
- package/internals/browser-launcher.js.map +0 -1
- package/tsconfig.build.tsbuildinfo +0 -1
|
@@ -1,8 +1,21 @@
|
|
|
1
|
-
import {
|
|
2
|
-
import {
|
|
3
|
-
import {
|
|
4
|
-
import
|
|
5
|
-
import { addTimeoutToPromise, tryCancel } from '@apify/timeout';
|
|
1
|
+
import { BasicCrawler, browserPoolCookieToToughCookie, ContextPipeline, cookieStringToToughCookie, EnqueueStrategy, NavigationSkippedError, OwnedOrInjected, remainingNavigationWindowMillis, RequestState, RequestThrottledError, resolveBaseUrlForEnqueueLinksFiltering, SessionError, toughCookieToBrowserPoolCookie, validators, } from '@crawlee/basic';
|
|
2
|
+
import { assertBrowserPoolNotConfigured, CLOUDFLARE_RETRY_CSS_SELECTORS, parseArgument, RETRY_CSS_SELECTORS, schemas, tryAbsoluteURL, } from '@crawlee/utils/internal';
|
|
3
|
+
import { sleep } from '@crawlee/utils';
|
|
4
|
+
import { z } from 'zod';
|
|
5
|
+
import { addTimeoutToPromise, TimeoutError, tryCancel } from '@apify/timeout';
|
|
6
|
+
const COOKIES_BEFORE_HOOKS = Symbol('cookiesBeforeHooks');
|
|
7
|
+
const readContextField = (ctx, key) => ctx[key];
|
|
8
|
+
/**
|
|
9
|
+
* Whether an error thrown by `page.goto()` is a navigation timeout - either our own {@link TimeoutError}
|
|
10
|
+
* or the driver's, which Playwright/Puppeteer report with their own class and a `Timeout ... exceeded` message
|
|
11
|
+
* naming the raw millisecond value rather than the configured window.
|
|
12
|
+
*/
|
|
13
|
+
function isNavigationTimeoutError(error) {
|
|
14
|
+
return (error instanceof TimeoutError ||
|
|
15
|
+
error?.name === 'TimeoutError' ||
|
|
16
|
+
error?.constructor?.name === 'TimeoutError' ||
|
|
17
|
+
/timeout.*exceeded/i.test(error?.message ?? ''));
|
|
18
|
+
}
|
|
6
19
|
/**
|
|
7
20
|
* Provides a simple framework for parallel crawling of web pages
|
|
8
21
|
* using headless browsers with [Puppeteer](https://github.com/puppeteer/puppeteer)
|
|
@@ -15,119 +28,126 @@ import { addTimeoutToPromise, tryCancel } from '@apify/timeout';
|
|
|
15
28
|
* If the target website doesn't need JavaScript, we should consider using the {@link CheerioCrawler},
|
|
16
29
|
* which downloads the pages using raw HTTP requests and is about 10x faster.
|
|
17
30
|
*
|
|
18
|
-
* The source URLs are represented by the {@link Request} objects that are fed from the
|
|
19
|
-
*
|
|
20
|
-
* constructor
|
|
31
|
+
* The source URLs are represented by the {@link Request} objects that are fed from the
|
|
32
|
+
* {@link IRequestManager|request manager} provided via the {@link BrowserCrawlerOptions.requestManager|`requestManager`}
|
|
33
|
+
* constructor option (a {@link RequestQueue} is itself a request manager). If no `requestManager` is provided,
|
|
21
34
|
* the crawler will open the default request queue either when the {@link BrowserCrawler.addRequests|`crawler.addRequests()`} function is called,
|
|
22
35
|
* or if `requests` parameter (representing the initial requests) of the {@link BrowserCrawler.run|`crawler.run()`} function is provided.
|
|
23
36
|
*
|
|
24
|
-
*
|
|
25
|
-
*
|
|
26
|
-
*
|
|
37
|
+
* To read from a read-only source such as a {@link RequestList} while still being able to enqueue new requests,
|
|
38
|
+
* combine it with a queue into a {@link RequestManagerTandem} via {@link IRequestLoader.toTandem|`requestLoader.toTandem()`}
|
|
39
|
+
* and pass the result as `requestManager`.
|
|
40
|
+
*
|
|
41
|
+
* > The {@link BrowserCrawlerOptions.requestList|`requestList`} and {@link BrowserCrawlerOptions.requestQueue|`requestQueue`}
|
|
42
|
+
* > options are deprecated; they are still accepted and folded into a single `requestManager` for back-compat.
|
|
27
43
|
*
|
|
28
44
|
* The crawler finishes when there are no more {@link Request} objects to crawl.
|
|
29
45
|
*
|
|
30
46
|
* `BrowserCrawler` opens a new browser page (i.e. tab or window) for each {@link Request} object to crawl
|
|
31
47
|
* and then calls the function provided by user as the {@link BrowserCrawlerOptions.requestHandler|`requestHandler`} option.
|
|
32
48
|
*
|
|
33
|
-
* New pages are only opened when there is enough free CPU and memory available,
|
|
34
|
-
*
|
|
35
|
-
*
|
|
36
|
-
*
|
|
37
|
-
*
|
|
38
|
-
* {@link AutoscaledPoolOptions.maxConcurrency|`maxConcurrency`} options of the
|
|
39
|
-
* underlying {@link AutoscaledPool} constructor are available directly in the `BrowserCrawler` constructor.
|
|
49
|
+
* New pages are only opened when there is enough free CPU and memory available, as judged by the crawler's
|
|
50
|
+
* {@link ConcurrencySystem}.
|
|
51
|
+
* Concurrency is tuned via the `minConcurrency`, `maxConcurrency` and `maxRequestsPerMinute` options of the
|
|
52
|
+
* `BrowserCrawler` constructor, or, for finer control, by injecting a pre-configured
|
|
53
|
+
* {@link ConcurrencySystem|`concurrencySystem`}.
|
|
40
54
|
*
|
|
41
55
|
* > *NOTE:* the pool of browser instances is internally managed by the {@link BrowserPool} class.
|
|
42
56
|
*
|
|
43
57
|
* @category Crawlers
|
|
44
58
|
*/
|
|
45
59
|
export class BrowserCrawler extends BasicCrawler {
|
|
46
|
-
|
|
60
|
+
/** Backs the {@link BrowserCrawler.browserPool|`browserPool`} getter. */
|
|
61
|
+
#browserPoolDep;
|
|
47
62
|
/**
|
|
48
|
-
* A reference to the underlying
|
|
49
|
-
*
|
|
63
|
+
* A reference to the underlying browser pool that manages the crawler's browsers. Typed as
|
|
64
|
+
* {@link IBrowserPool} so custom implementations can be plugged in via the `browserPool` constructor option.
|
|
50
65
|
*/
|
|
51
|
-
|
|
66
|
+
get browserPool() {
|
|
67
|
+
return this.#browserPoolDep.value;
|
|
68
|
+
}
|
|
69
|
+
ignoreShadowRoots;
|
|
70
|
+
ignoreIframes;
|
|
71
|
+
#navigationTimeoutMillis;
|
|
72
|
+
#preNavigationHooks;
|
|
73
|
+
#postNavigationHooks;
|
|
74
|
+
#saveResponseCookies;
|
|
52
75
|
/**
|
|
53
|
-
*
|
|
76
|
+
* @internal
|
|
54
77
|
*/
|
|
55
|
-
browserPool;
|
|
56
|
-
launchContext;
|
|
57
|
-
userProvidedRequestHandler;
|
|
58
|
-
navigationTimeoutMillis;
|
|
59
|
-
requestHandlerTimeoutInnerMillis;
|
|
60
|
-
preNavigationHooks;
|
|
61
|
-
postNavigationHooks;
|
|
62
|
-
persistCookiesPerSession;
|
|
63
78
|
static optionsShape = {
|
|
64
79
|
...BasicCrawler.optionsShape,
|
|
65
|
-
navigationTimeoutSecs:
|
|
66
|
-
|
|
67
|
-
|
|
68
|
-
|
|
69
|
-
|
|
70
|
-
|
|
71
|
-
|
|
72
|
-
|
|
73
|
-
|
|
74
|
-
proxyConfiguration:
|
|
75
|
-
|
|
76
|
-
|
|
80
|
+
navigationTimeoutSecs: schemas.anyNumber
|
|
81
|
+
.refine((value) => value > 0, 'Expected a number greater than 0')
|
|
82
|
+
.default(60),
|
|
83
|
+
preNavigationHooks: schemas.anyArray.default(() => []),
|
|
84
|
+
postNavigationHooks: schemas.anyArray.default(() => []),
|
|
85
|
+
browserPool: validators.browserPool.optional(),
|
|
86
|
+
browserPoolBuilder: schemas.anyFunction.optional(),
|
|
87
|
+
remoteBrowser: schemas.anyObject.optional(),
|
|
88
|
+
saveResponseCookies: z.boolean().default(true),
|
|
89
|
+
proxyConfiguration: validators.proxyConfiguration.optional(),
|
|
90
|
+
ignoreIframes: z.boolean().default(false),
|
|
91
|
+
ignoreShadowRoots: z.boolean().default(false),
|
|
77
92
|
};
|
|
93
|
+
/** @internal */
|
|
94
|
+
static optionsSchema = z.strictObject(BrowserCrawler.optionsShape);
|
|
78
95
|
/**
|
|
79
96
|
* All `BrowserCrawler` parameters are passed via an options object.
|
|
97
|
+
* @internal
|
|
80
98
|
*/
|
|
81
|
-
constructor(options
|
|
82
|
-
|
|
83
|
-
|
|
99
|
+
constructor(options) {
|
|
100
|
+
const { navigationTimeoutSecs, saveResponseCookies, browserPool, remoteBrowser, preNavigationHooks, postNavigationHooks, ignoreIframes, ignoreShadowRoots, contextPipelineBuilder, browserPoolBuilder, extendContext, ...basicCrawlerOptions } = parseArgument(options, BrowserCrawler.optionsSchema, 'BrowserCrawlerOptions');
|
|
101
|
+
if (browserPool) {
|
|
102
|
+
assertBrowserPoolNotConfigured(new.target.name, { remoteBrowser });
|
|
103
|
+
}
|
|
104
|
+
const skipGuard = (action) => async (ctx) => ctx.request.skipNavigation ? {} : ((await action(ctx)) ?? {});
|
|
84
105
|
super({
|
|
85
106
|
...basicCrawlerOptions,
|
|
86
|
-
|
|
87
|
-
|
|
88
|
-
|
|
89
|
-
|
|
90
|
-
|
|
91
|
-
|
|
92
|
-
|
|
93
|
-
|
|
94
|
-
|
|
95
|
-
|
|
96
|
-
|
|
97
|
-
|
|
98
|
-
|
|
99
|
-
|
|
100
|
-
|
|
101
|
-
|
|
102
|
-
|
|
103
|
-
|
|
104
|
-
|
|
105
|
-
|
|
106
|
-
|
|
107
|
-
|
|
108
|
-
|
|
109
|
-
|
|
110
|
-
|
|
111
|
-
|
|
112
|
-
}
|
|
113
|
-
if (launchContext?.userAgent) {
|
|
114
|
-
if (browserPoolOptions.useFingerprints)
|
|
115
|
-
this.log.info('Custom user agent provided, disabling automatic browser fingerprint injection!');
|
|
116
|
-
browserPoolOptions.useFingerprints = false;
|
|
117
|
-
}
|
|
118
|
-
const { preLaunchHooks = [], postLaunchHooks = [], ...rest } = browserPoolOptions;
|
|
119
|
-
this.browserPool = new BrowserPool({
|
|
120
|
-
...rest,
|
|
121
|
-
preLaunchHooks: [this._extendLaunchContext.bind(this), ...preLaunchHooks],
|
|
122
|
-
postLaunchHooks: [this._maybeAddSessionRetiredListener.bind(this), ...postLaunchHooks],
|
|
107
|
+
contextPipelineBuilder: () => {
|
|
108
|
+
// A single navigation window covers the pre-navigation hooks, the navigation, and the
|
|
109
|
+
// post-navigation hooks: the whole phase shares one `navigationTimeoutSecs` budget, so a slow
|
|
110
|
+
// hook eats into the same window the navigation uses. The navigation itself is bounded by
|
|
111
|
+
// capping its `gotoOptions.timeout` to the remaining budget.
|
|
112
|
+
const windowGuard = (step) => skipGuard(async (ctx) => {
|
|
113
|
+
const remaining = remainingNavigationWindowMillis(ctx, this.#navigationTimeoutMillis);
|
|
114
|
+
if (remaining <= 0) {
|
|
115
|
+
throw new TimeoutError(`Navigation timed out after ${this.#navigationTimeoutMillis / 1000} seconds.`);
|
|
116
|
+
}
|
|
117
|
+
return addTimeoutToPromise(async () => step(ctx), remaining, `Navigation timed out after ${this.#navigationTimeoutMillis / 1000} seconds.`);
|
|
118
|
+
});
|
|
119
|
+
let pipeline = contextPipelineBuilder().compose(this.prepareNavigation.bind(this));
|
|
120
|
+
for (const hook of this.#preNavigationHooks) {
|
|
121
|
+
pipeline = pipeline.compose(windowGuard(hook));
|
|
122
|
+
}
|
|
123
|
+
pipeline = pipeline.compose(skipGuard(this.navigate.bind(this)));
|
|
124
|
+
for (const hook of this.#postNavigationHooks) {
|
|
125
|
+
pipeline = pipeline.compose(windowGuard(hook));
|
|
126
|
+
}
|
|
127
|
+
return pipeline
|
|
128
|
+
.compose(skipGuard(this.finalizeNavigation.bind(this)))
|
|
129
|
+
.compose(this.handleBlockedRequestByContent.bind(this))
|
|
130
|
+
.compose(this.restoreRequestState.bind(this));
|
|
131
|
+
},
|
|
132
|
+
extendContext,
|
|
123
133
|
});
|
|
134
|
+
this.#navigationTimeoutMillis = navigationTimeoutSecs * 1000;
|
|
135
|
+
// The public option hooks are extension-aware; internal storage uses the base context type
|
|
136
|
+
// (the pipeline composes hooks against the concrete context, which does not statically carry
|
|
137
|
+
// `ContextExtension`). The extension members are present at runtime regardless.
|
|
138
|
+
this.#preNavigationHooks = preNavigationHooks;
|
|
139
|
+
this.#postNavigationHooks = postNavigationHooks;
|
|
140
|
+
this.ignoreIframes = ignoreIframes;
|
|
141
|
+
this.ignoreShadowRoots = ignoreShadowRoots;
|
|
142
|
+
this.#saveResponseCookies = saveResponseCookies;
|
|
143
|
+
this.#browserPoolDep = OwnedOrInjected.resolve(browserPool, () => browserPoolBuilder(remoteBrowser));
|
|
124
144
|
}
|
|
125
|
-
|
|
126
|
-
|
|
127
|
-
|
|
128
|
-
|
|
129
|
-
|
|
130
|
-
|
|
145
|
+
/** @internal */
|
|
146
|
+
getNavigationTimeoutMillis() {
|
|
147
|
+
return this.#navigationTimeoutMillis;
|
|
148
|
+
}
|
|
149
|
+
buildContextPipeline() {
|
|
150
|
+
return ContextPipeline.create().compose(this.preparePage.bind(this));
|
|
131
151
|
}
|
|
132
152
|
async containsSelectors(page, selectors) {
|
|
133
153
|
const foundSelectors = (await Promise.all(selectors.map((selector) => page.$(selector))))
|
|
@@ -138,12 +158,6 @@ export class BrowserCrawler extends BasicCrawler {
|
|
|
138
158
|
}
|
|
139
159
|
async isRequestBlocked(crawlingContext) {
|
|
140
160
|
const { page, response } = crawlingContext;
|
|
141
|
-
const blockedStatusCodes =
|
|
142
|
-
// eslint-disable-next-line dot-notation
|
|
143
|
-
(this.sessionPool?.['blockedStatusCodes'].length ?? 0) > 0
|
|
144
|
-
? // eslint-disable-next-line dot-notation
|
|
145
|
-
this.sessionPool['blockedStatusCodes']
|
|
146
|
-
: DEFAULT_BLOCKED_STATUS_CODES;
|
|
147
161
|
// Cloudflare specific heuristic - wait 5 seconds if we get a 403 for the JS challenge to load / resolve.
|
|
148
162
|
if ((await this.containsSelectors(page, CLOUDFLARE_RETRY_CSS_SELECTORS)) && response?.status() === 403) {
|
|
149
163
|
await sleep(5000);
|
|
@@ -154,170 +168,244 @@ export class BrowserCrawler extends BasicCrawler {
|
|
|
154
168
|
return `Cloudflare challenge failed, found selectors: ${foundSelectors.join(', ')}`;
|
|
155
169
|
}
|
|
156
170
|
const foundSelectors = await this.containsSelectors(page, RETRY_CSS_SELECTORS);
|
|
157
|
-
const
|
|
171
|
+
const statusCode = response?.status() ?? 0;
|
|
158
172
|
if (foundSelectors)
|
|
159
173
|
return `Found selectors: ${foundSelectors.join(', ')}`;
|
|
160
|
-
if (
|
|
161
|
-
return `Received blocked status code: ${
|
|
174
|
+
if (this.blockedStatusCodes.has(statusCode))
|
|
175
|
+
return `Received blocked status code: ${statusCode}`;
|
|
162
176
|
return false;
|
|
163
177
|
}
|
|
164
|
-
|
|
165
|
-
|
|
166
|
-
*/
|
|
167
|
-
async _runRequestHandler(crawlingContext) {
|
|
168
|
-
const newPageOptions = {
|
|
178
|
+
async preparePage(crawlingContext, onCleanup) {
|
|
179
|
+
const page = await this.browserPool.newPage({
|
|
169
180
|
id: crawlingContext.id,
|
|
170
|
-
|
|
171
|
-
|
|
172
|
-
|
|
173
|
-
|
|
174
|
-
|
|
175
|
-
|
|
181
|
+
session: crawlingContext.session,
|
|
182
|
+
});
|
|
183
|
+
tryCancel();
|
|
184
|
+
onCleanup(() => {
|
|
185
|
+
crawlingContext.registerDeferredCleanup(async () => {
|
|
186
|
+
const error = !crawlingContext.session.isUsable()
|
|
187
|
+
? new SessionError('Session is no longer usable')
|
|
188
|
+
: undefined;
|
|
189
|
+
await this.browserPool
|
|
190
|
+
.closePage(page, { error })
|
|
191
|
+
.catch((closeError) => this.log.debug('Error while closing page', { error: closeError }));
|
|
176
192
|
});
|
|
177
|
-
|
|
178
|
-
|
|
179
|
-
|
|
180
|
-
|
|
181
|
-
|
|
182
|
-
|
|
183
|
-
|
|
184
|
-
|
|
185
|
-
|
|
186
|
-
|
|
187
|
-
|
|
188
|
-
|
|
189
|
-
}
|
|
193
|
+
});
|
|
194
|
+
const addRequests = crawlingContext.addRequests;
|
|
195
|
+
const extractLinks = async (options) => {
|
|
196
|
+
return extractUrlsFromPage(page, options?.selector ?? 'a', options?.baseUrl ?? crawlingContext.request.loadedUrl ?? crawlingContext.request.url);
|
|
197
|
+
};
|
|
198
|
+
return {
|
|
199
|
+
page,
|
|
200
|
+
get response() {
|
|
201
|
+
throw new Error("The `response` property is not available. This might mean that you're trying to access it before navigation or that navigation resulted in `null` (this should only happen with `about:` URLs)");
|
|
202
|
+
},
|
|
203
|
+
get gotoOptions() {
|
|
204
|
+
throw new Error('The `gotoOptions` property is not available until `prepareNavigation` runs.');
|
|
205
|
+
},
|
|
206
|
+
extractLinks,
|
|
207
|
+
enqueueLinks: async (options = {}) => {
|
|
208
|
+
const baseUrl = resolveBaseUrlForEnqueueLinksFiltering({
|
|
209
|
+
enqueueStrategy: options.strategy,
|
|
210
|
+
finalRequestUrl: crawlingContext.request.loadedUrl,
|
|
211
|
+
originalRequestUrl: crawlingContext.request.url,
|
|
212
|
+
userProvidedBaseUrl: options.baseUrl,
|
|
213
|
+
});
|
|
214
|
+
const urls = await extractLinks(options);
|
|
215
|
+
return addRequests(urls, {
|
|
216
|
+
...options,
|
|
217
|
+
baseUrl,
|
|
218
|
+
strategy: options.strategy ?? EnqueueStrategy.SameHostname,
|
|
219
|
+
});
|
|
220
|
+
},
|
|
221
|
+
};
|
|
222
|
+
}
|
|
223
|
+
async prepareNavigation(crawlingContext) {
|
|
224
|
+
if (crawlingContext.request.skipNavigation) {
|
|
225
|
+
return {
|
|
226
|
+
request: new Proxy(crawlingContext.request, {
|
|
227
|
+
get(target, propertyName, receiver) {
|
|
228
|
+
if (propertyName === 'loadedUrl') {
|
|
229
|
+
throw new NavigationSkippedError('The `request.loadedUrl` property is not available - `skipNavigation` was used');
|
|
230
|
+
}
|
|
231
|
+
return Reflect.get(target, propertyName, receiver);
|
|
232
|
+
},
|
|
233
|
+
}),
|
|
234
|
+
get response() {
|
|
235
|
+
throw new NavigationSkippedError('The `response` property is not available - `skipNavigation` was used');
|
|
236
|
+
},
|
|
237
|
+
};
|
|
190
238
|
}
|
|
191
|
-
|
|
239
|
+
crawlingContext.request.state = RequestState.BEFORE_NAV;
|
|
240
|
+
return {
|
|
241
|
+
// Default to the full navigation timeout so a pre-navigation hook can read it; `navigate` narrows it
|
|
242
|
+
// to the remaining shared window unless a hook overrode it (see there).
|
|
243
|
+
gotoOptions: { timeout: this.#navigationTimeoutMillis },
|
|
244
|
+
[COOKIES_BEFORE_HOOKS]: this.getCookieHeaderFromRequest(crawlingContext.request),
|
|
245
|
+
};
|
|
246
|
+
}
|
|
247
|
+
// oxlint-disable-next-line crawlee/prefer-private-fields -- patched by @crawlee/otel
|
|
248
|
+
async navigate(crawlingContext) {
|
|
192
249
|
tryCancel();
|
|
193
|
-
|
|
194
|
-
|
|
195
|
-
|
|
196
|
-
|
|
197
|
-
// So we must not save the session prior to making sure it was used only once, otherwise we would use it twice.
|
|
198
|
-
const { request, session } = crawlingContext;
|
|
199
|
-
if (!request.skipNavigation) {
|
|
200
|
-
await this._handleNavigation(crawlingContext);
|
|
201
|
-
tryCancel();
|
|
202
|
-
await this._responseHandler(crawlingContext);
|
|
203
|
-
tryCancel();
|
|
204
|
-
// save cookies
|
|
205
|
-
// TODO: Should we save the cookies also after/only the handle page?
|
|
206
|
-
if (this.persistCookiesPerSession) {
|
|
207
|
-
const cookies = await crawlingContext.browserController.getCookies(page);
|
|
208
|
-
tryCancel();
|
|
209
|
-
session?.setCookies(cookies, request.loadedUrl);
|
|
210
|
-
}
|
|
211
|
-
}
|
|
212
|
-
if (!this.requestMatchesEnqueueStrategy(request)) {
|
|
213
|
-
this.log.debug(
|
|
214
|
-
// eslint-disable-next-line dot-notation
|
|
215
|
-
`Skipping request ${request.id} (starting url: ${request.url} -> loaded url: ${request.loadedUrl}) because it does not match the enqueue strategy (${request['enqueueStrategy']}).`);
|
|
216
|
-
request.noRetry = true;
|
|
217
|
-
request.state = RequestState.SKIPPED;
|
|
218
|
-
return;
|
|
250
|
+
const gotoOptions = crawlingContext.gotoOptions;
|
|
251
|
+
const remaining = remainingNavigationWindowMillis(crawlingContext, this.#navigationTimeoutMillis);
|
|
252
|
+
if (remaining <= 0) {
|
|
253
|
+
throw new TimeoutError(`Navigation timed out after ${this.#navigationTimeoutMillis / 1000} seconds.`);
|
|
219
254
|
}
|
|
220
|
-
|
|
221
|
-
|
|
222
|
-
|
|
223
|
-
|
|
255
|
+
// If a hook left the default `navigationTimeoutMillis` in place, bound the goto to whatever is left of the
|
|
256
|
+
// shared navigation window. If it overrode the value - including `0`, Playwright's "no timeout" - honour
|
|
257
|
+
// that verbatim as the goto's own timeout. The driver enforces this natively (so a timed-out goto is
|
|
258
|
+
// aborted, not left lingering) and `handleNavigationTimeout` turns its error into our own message.
|
|
259
|
+
const gotoTimeout = gotoOptions;
|
|
260
|
+
if (gotoTimeout.timeout === this.#navigationTimeoutMillis) {
|
|
261
|
+
gotoTimeout.timeout = remaining;
|
|
224
262
|
}
|
|
225
|
-
|
|
263
|
+
const cookiesBeforeHooks = readContextField(crawlingContext, COOKIES_BEFORE_HOOKS);
|
|
264
|
+
const cookiesAfterHooks = this.getCookieHeaderFromRequest(crawlingContext.request);
|
|
265
|
+
await this.applyCookies(crawlingContext, cookiesBeforeHooks, cookiesAfterHooks);
|
|
266
|
+
let response;
|
|
226
267
|
try {
|
|
227
|
-
|
|
228
|
-
request.state = RequestState.DONE;
|
|
268
|
+
response = (await this.navigationHandler(crawlingContext, gotoOptions)) ?? undefined;
|
|
229
269
|
}
|
|
230
|
-
catch (
|
|
231
|
-
|
|
232
|
-
|
|
270
|
+
catch (error) {
|
|
271
|
+
await this.handleNavigationTimeout(crawlingContext, error);
|
|
272
|
+
crawlingContext.request.state = RequestState.ERROR;
|
|
273
|
+
this.throwIfProxyError(error);
|
|
274
|
+
throw error;
|
|
233
275
|
}
|
|
234
276
|
tryCancel();
|
|
277
|
+
crawlingContext.request.state = RequestState.AFTER_NAV;
|
|
278
|
+
return { response };
|
|
235
279
|
}
|
|
236
|
-
|
|
237
|
-
|
|
238
|
-
|
|
239
|
-
|
|
240
|
-
|
|
241
|
-
// which configures proxy and session for every new request
|
|
242
|
-
const browserControllerInstance = this.browserPool.getBrowserControllerByPage(page);
|
|
243
|
-
crawlingContext.browserController = browserControllerInstance;
|
|
244
|
-
if (!createNewSession) {
|
|
245
|
-
crawlingContext.session = browserControllerInstance.launchContext.session;
|
|
280
|
+
async finalizeNavigation(crawlingContext) {
|
|
281
|
+
tryCancel();
|
|
282
|
+
let response;
|
|
283
|
+
try {
|
|
284
|
+
response = crawlingContext.response;
|
|
246
285
|
}
|
|
247
|
-
|
|
248
|
-
|
|
286
|
+
catch {
|
|
287
|
+
// `preparePage` installs a throwing getter for `response`; reaching this branch means
|
|
288
|
+
// navigation produced no response and no hook overrode it. Treat as undefined.
|
|
249
289
|
}
|
|
250
|
-
|
|
251
|
-
|
|
252
|
-
|
|
253
|
-
|
|
254
|
-
|
|
255
|
-
|
|
256
|
-
onSkippedRequest: this.onSkippedRequest,
|
|
257
|
-
originalRequestUrl: crawlingContext.request.url,
|
|
258
|
-
finalRequestUrl: crawlingContext.request.loadedUrl,
|
|
259
|
-
});
|
|
260
|
-
};
|
|
290
|
+
await this.processResponse(response, crawlingContext);
|
|
291
|
+
tryCancel();
|
|
292
|
+
// Persist cookies from the navigation response before the user handler runs.
|
|
293
|
+
// Cookies set during `requestHandler` are saved again afterwards.
|
|
294
|
+
await this.persistCookiesFromPage(crawlingContext);
|
|
295
|
+
return { request: crawlingContext.request };
|
|
261
296
|
}
|
|
262
|
-
|
|
263
|
-
|
|
264
|
-
|
|
265
|
-
|
|
266
|
-
|
|
297
|
+
/**
|
|
298
|
+
* Copies cookies from the live browser page into the session cookie jar.
|
|
299
|
+
*/
|
|
300
|
+
async persistCookiesFromPage(crawlingContext) {
|
|
301
|
+
if (!this.#saveResponseCookies || !crawlingContext.session) {
|
|
302
|
+
return;
|
|
303
|
+
}
|
|
304
|
+
const { cookies } = await this.browserPool.extractPageState(crawlingContext.page);
|
|
267
305
|
tryCancel();
|
|
268
|
-
|
|
269
|
-
await
|
|
306
|
+
// Prefer the live page URL — the handler may have navigated after the initial load.
|
|
307
|
+
const url = (await crawlingContext.page.url()) || crawlingContext.request.loadedUrl || crawlingContext.request.url;
|
|
308
|
+
for (const cookie of cookies) {
|
|
309
|
+
try {
|
|
310
|
+
await crawlingContext.session.cookieJar.setCookie(browserPoolCookieToToughCookie(cookie), url, {
|
|
311
|
+
ignoreError: false,
|
|
312
|
+
});
|
|
313
|
+
}
|
|
314
|
+
catch (e) {
|
|
315
|
+
this.log.debug(`Could not set cookie: ${e.message}`);
|
|
316
|
+
}
|
|
317
|
+
}
|
|
318
|
+
}
|
|
319
|
+
/**
|
|
320
|
+
* Runs the user request handler, then re-reads browser cookies so login flows /
|
|
321
|
+
* `page.setCookie` / XHR `Set-Cookie` updates are stored for later requests.
|
|
322
|
+
* @internal
|
|
323
|
+
*/
|
|
324
|
+
async runRequestHandler(crawlingContext) {
|
|
270
325
|
try {
|
|
271
|
-
|
|
326
|
+
await super.runRequestHandler(crawlingContext);
|
|
272
327
|
}
|
|
273
|
-
|
|
274
|
-
|
|
275
|
-
|
|
276
|
-
|
|
277
|
-
|
|
328
|
+
finally {
|
|
329
|
+
if (!crawlingContext.request.skipNavigation) {
|
|
330
|
+
try {
|
|
331
|
+
await this.persistCookiesFromPage(crawlingContext);
|
|
332
|
+
}
|
|
333
|
+
catch {
|
|
334
|
+
// Page may already be closed on some failure paths; ignore.
|
|
335
|
+
}
|
|
336
|
+
}
|
|
278
337
|
}
|
|
279
|
-
tryCancel();
|
|
280
|
-
crawlingContext.request.state = RequestState.AFTER_NAV;
|
|
281
|
-
await this._executeHooks(this.postNavigationHooks, crawlingContext, gotoOptions);
|
|
282
338
|
}
|
|
283
|
-
async
|
|
284
|
-
|
|
339
|
+
async handleBlockedRequestByContent(crawlingContext) {
|
|
340
|
+
if (this.retryOnBlocked) {
|
|
341
|
+
const error = await this.isRequestBlocked(crawlingContext);
|
|
342
|
+
if (error)
|
|
343
|
+
throw new SessionError(error);
|
|
344
|
+
}
|
|
345
|
+
return {};
|
|
346
|
+
}
|
|
347
|
+
async restoreRequestState(crawlingContext) {
|
|
348
|
+
crawlingContext.request.state = RequestState.REQUEST_HANDLER;
|
|
349
|
+
return {};
|
|
350
|
+
}
|
|
351
|
+
async applyCookies({ session, request, page }, preHooksCookies, postHooksCookies) {
|
|
352
|
+
const sessionCookie = session
|
|
353
|
+
? (await session.cookieJar.getCookies(request.url)).map(toughCookieToBrowserPoolCookie)
|
|
354
|
+
: [];
|
|
285
355
|
const parsedPreHooksCookies = preHooksCookies.split(/ *; */).map((c) => cookieStringToToughCookie(c));
|
|
286
356
|
const parsedPostHooksCookies = postHooksCookies.split(/ *; */).map((c) => cookieStringToToughCookie(c));
|
|
287
|
-
|
|
357
|
+
const cookies = [...sessionCookie, ...parsedPreHooksCookies, ...parsedPostHooksCookies]
|
|
288
358
|
.filter((c) => typeof c !== 'undefined' && c !== null)
|
|
289
|
-
.map((c) => ({ ...c, url: c.domain ? undefined : request.url }))
|
|
359
|
+
.map((c) => ({ ...c, url: c.domain ? undefined : request.url }));
|
|
360
|
+
await this.browserPool.injectPageState(page, { cookies });
|
|
290
361
|
}
|
|
291
362
|
/**
|
|
292
|
-
* Marks session bad in
|
|
363
|
+
* Marks session bad on navigation timeout, and stops in-flight page loading on any navigation error.
|
|
293
364
|
*/
|
|
294
|
-
async
|
|
295
|
-
const { session } = crawlingContext;
|
|
296
|
-
|
|
297
|
-
|
|
365
|
+
async handleNavigationTimeout(crawlingContext, error) {
|
|
366
|
+
const { session, page } = crawlingContext;
|
|
367
|
+
// Fire-and-forget: no user code will run on this page after a failed navigation.
|
|
368
|
+
// Swallow rejections: the page may already be detached.
|
|
369
|
+
void page.evaluate(() => window.stop()).catch(() => { });
|
|
370
|
+
if (isNavigationTimeoutError(error)) {
|
|
371
|
+
session?.markBad();
|
|
372
|
+
// The driver was handed the remaining window (usually shorter than `navigationTimeoutSecs` once the
|
|
373
|
+
// hooks have run), so it names that value in its own error; report the configured window instead.
|
|
374
|
+
throw new TimeoutError(`Navigation timed out after ${this.#navigationTimeoutMillis / 1000} seconds.`);
|
|
298
375
|
}
|
|
299
|
-
await crawlingContext.page.close();
|
|
300
376
|
}
|
|
301
377
|
/**
|
|
302
378
|
* Transforms proxy-related errors to `SessionError`.
|
|
303
379
|
*/
|
|
304
|
-
|
|
380
|
+
throwIfProxyError(error) {
|
|
305
381
|
if (this.isProxyError(error)) {
|
|
306
|
-
throw new SessionError(this.
|
|
382
|
+
throw new SessionError(this.getMessageFromError(error));
|
|
307
383
|
}
|
|
308
384
|
}
|
|
309
|
-
|
|
310
|
-
|
|
311
|
-
*/
|
|
312
|
-
async _responseHandler(crawlingContext) {
|
|
313
|
-
const { response, session, request, page } = crawlingContext;
|
|
385
|
+
async processResponse(response, crawlingContext) {
|
|
386
|
+
const { session, request, page } = crawlingContext;
|
|
314
387
|
if (typeof response === 'object' && typeof response.status === 'function') {
|
|
315
388
|
const status = response.status();
|
|
316
|
-
this.
|
|
389
|
+
this.statistics.registerStatusCode(status);
|
|
390
|
+
// Ahead of the error-status throw below: a 429 the user opted into treating as an error is still a
|
|
391
|
+
// rate limit the domain should back off from.
|
|
392
|
+
if (status === 429) {
|
|
393
|
+
// Both drivers lower-case header names and join duplicates, so a plain lookup is enough.
|
|
394
|
+
const retryAfter = response.headers?.()['retry-after'];
|
|
395
|
+
if (this.recordDomainRateLimit(crawlingContext.request.url, retryAfter)) {
|
|
396
|
+
throw new RequestThrottledError(`${crawlingContext.request.url} responded with 429.`);
|
|
397
|
+
}
|
|
398
|
+
}
|
|
399
|
+
if (this.isErrorStatusCode(status)) {
|
|
400
|
+
if (this.additionalHttpErrorStatusCodes.has(status)) {
|
|
401
|
+
throw new Error(`${status} - Error status code was set by user.`);
|
|
402
|
+
}
|
|
403
|
+
throw new Error(`${status} - Internal Server Error`);
|
|
404
|
+
}
|
|
317
405
|
}
|
|
318
406
|
if (this.sessionPool && response && session) {
|
|
319
407
|
if (typeof response === 'object' && typeof response.status === 'function') {
|
|
320
|
-
this.
|
|
408
|
+
this.throwOnBlockedRequest(response.status());
|
|
321
409
|
}
|
|
322
410
|
else {
|
|
323
411
|
this.log.debug('Got a malformed Browser response.', { request, response });
|
|
@@ -325,69 +413,18 @@ export class BrowserCrawler extends BasicCrawler {
|
|
|
325
413
|
}
|
|
326
414
|
request.loadedUrl = await page.url();
|
|
327
415
|
}
|
|
328
|
-
async _extendLaunchContext(_pageId, launchContext) {
|
|
329
|
-
const launchContextExtends = {};
|
|
330
|
-
if (this.sessionPool) {
|
|
331
|
-
launchContextExtends.session = await this.sessionPool.getSession();
|
|
332
|
-
}
|
|
333
|
-
if (this.proxyConfiguration && !launchContext.proxyUrl) {
|
|
334
|
-
const proxyInfo = await this.proxyConfiguration.newProxyInfo(launchContextExtends.session?.id, {
|
|
335
|
-
proxyTier: launchContext.proxyTier ?? undefined,
|
|
336
|
-
});
|
|
337
|
-
launchContext.proxyUrl = proxyInfo?.url;
|
|
338
|
-
launchContextExtends.proxyInfo = proxyInfo;
|
|
339
|
-
// Disable SSL verification for MITM proxies
|
|
340
|
-
if (this.proxyConfiguration.isManInTheMiddle) {
|
|
341
|
-
/**
|
|
342
|
-
* @see https://playwright.dev/docs/api/class-browser/#browser-new-context
|
|
343
|
-
* @see https://github.com/puppeteer/puppeteer/blob/main/docs/api.md
|
|
344
|
-
*/
|
|
345
|
-
launchContext.launchOptions.ignoreHTTPSErrors = true;
|
|
346
|
-
launchContext.launchOptions.acceptInsecureCerts = true;
|
|
347
|
-
}
|
|
348
|
-
}
|
|
349
|
-
launchContext.extend(launchContextExtends);
|
|
350
|
-
}
|
|
351
|
-
_maybeAddSessionRetiredListener(_pageId, browserController) {
|
|
352
|
-
if (this.sessionPool) {
|
|
353
|
-
const listener = (session) => {
|
|
354
|
-
const { launchContext } = browserController;
|
|
355
|
-
if (session.id === launchContext.session.id) {
|
|
356
|
-
this.browserPool.retireBrowserController(browserController);
|
|
357
|
-
}
|
|
358
|
-
};
|
|
359
|
-
this.sessionPool.on(EVENT_SESSION_RETIRED, listener);
|
|
360
|
-
browserController.on("browserClosed" /* BROWSER_CONTROLLER_EVENTS.BROWSER_CLOSED */, () => {
|
|
361
|
-
return this.sessionPool.removeListener(EVENT_SESSION_RETIRED, listener);
|
|
362
|
-
});
|
|
363
|
-
}
|
|
364
|
-
}
|
|
365
416
|
/**
|
|
366
|
-
*
|
|
367
|
-
*
|
|
417
|
+
* Closes the browsers of a pool the crawler owns, so a finished run leaves none behind. The pool itself is
|
|
418
|
+
* crawler-lifetime and survives — destroying it here would hand a repeated `run()` a dead pool.
|
|
368
419
|
*/
|
|
369
420
|
async teardown() {
|
|
370
|
-
await this.
|
|
421
|
+
await this.#browserPoolDep.ifOwned((pool) => pool.releaseAllBrowsers());
|
|
371
422
|
await super.teardown();
|
|
372
423
|
}
|
|
373
|
-
|
|
374
|
-
|
|
375
|
-
|
|
376
|
-
|
|
377
|
-
enqueueStrategy: options?.strategy,
|
|
378
|
-
finalRequestUrl,
|
|
379
|
-
originalRequestUrl,
|
|
380
|
-
userProvidedBaseUrl: options?.baseUrl,
|
|
381
|
-
});
|
|
382
|
-
const urls = await extractUrlsFromPage(page, options?.selector ?? 'a', options?.baseUrl ?? finalRequestUrl ?? originalRequestUrl);
|
|
383
|
-
return enqueueLinks({
|
|
384
|
-
requestQueue,
|
|
385
|
-
robotsTxtFile,
|
|
386
|
-
onSkippedRequest,
|
|
387
|
-
urls,
|
|
388
|
-
baseUrl,
|
|
389
|
-
...options,
|
|
390
|
-
});
|
|
424
|
+
async destroy() {
|
|
425
|
+
await super.destroy();
|
|
426
|
+
await this.#browserPoolDep.ifOwned((pool) => pool.destroy());
|
|
427
|
+
}
|
|
391
428
|
}
|
|
392
429
|
/**
|
|
393
430
|
* Extracts URLs from a given page.
|
|
@@ -414,4 +451,3 @@ page, selector, baseUrl) {
|
|
|
414
451
|
})
|
|
415
452
|
.filter((href) => !!href);
|
|
416
453
|
}
|
|
417
|
-
//# sourceMappingURL=browser-crawler.js.map
|