@crawlee/core 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/byte_utils.d.ts +17 -0
- package/byte_utils.js +42 -0
- package/configuration.d.ts +88 -234
- package/configuration.js +162 -230
- package/debug.d.ts +36 -0
- package/debug.js +70 -0
- package/errors.d.ts +43 -10
- package/errors.js +57 -12
- package/events/event_manager.d.ts +35 -9
- package/events/event_manager.js +12 -13
- package/events/index.d.ts +1 -1
- package/events/index.js +0 -1
- package/events/local_event_manager.d.ts +15 -5
- package/events/local_event_manager.js +39 -42
- package/events/system_info.d.ts +38 -0
- package/events/system_info.js +1 -0
- package/index.d.ts +8 -11
- package/index.js +9 -10
- package/internal.d.ts +8 -0
- package/internal.js +9 -0
- package/iterables.d.ts +79 -0
- package/iterables.js +134 -0
- package/log.d.ts +81 -3
- package/log.js +134 -1
- package/memory-storage/consts.d.ts +4 -0
- package/memory-storage/consts.js +4 -0
- package/memory-storage/index.d.ts +1 -0
- package/memory-storage/index.js +1 -0
- package/memory-storage/memory-storage.d.ts +43 -0
- package/memory-storage/memory-storage.js +158 -0
- package/memory-storage/resource-clients/common/base-client.d.ts +4 -0
- package/memory-storage/resource-clients/common/base-client.js +6 -0
- package/memory-storage/resource-clients/dataset.d.ts +40 -0
- package/memory-storage/resource-clients/dataset.js +106 -0
- package/memory-storage/resource-clients/key-value-store.d.ts +56 -0
- package/memory-storage/resource-clients/key-value-store.js +180 -0
- package/memory-storage/resource-clients/request-queue.d.ts +77 -0
- package/memory-storage/resource-clients/request-queue.js +414 -0
- package/memory-storage/utils.d.ts +16 -0
- package/memory-storage/utils.js +41 -0
- package/owned_or_injected.d.ts +58 -0
- package/owned_or_injected.js +98 -0
- package/package.json +19 -20
- package/proxy_configuration.d.ts +29 -155
- package/proxy_configuration.js +38 -183
- package/recoverable_state.d.ts +177 -0
- package/recoverable_state.js +268 -0
- package/request.d.ts +62 -81
- package/request.js +125 -220
- package/serialization.d.ts +0 -1
- package/serialization.js +18 -18
- package/service_locator.d.ts +168 -0
- package/service_locator.js +255 -0
- package/storages/batched_adds.d.ts +37 -0
- package/storages/batched_adds.js +73 -0
- package/storages/dataset.d.ts +120 -56
- package/storages/dataset.js +299 -148
- package/storages/index.d.ts +9 -8
- package/storages/index.js +5 -8
- package/storages/key_value_store.d.ts +192 -65
- package/storages/key_value_store.js +453 -193
- package/storages/key_value_store_codec.d.ts +32 -0
- package/storages/key_value_store_codec.js +113 -0
- package/storages/request_dedup_cache.d.ts +22 -0
- package/storages/request_dedup_cache.js +48 -0
- package/storages/request_list.d.ts +55 -137
- package/storages/request_list.js +272 -245
- package/storages/request_loader.d.ts +127 -0
- package/storages/request_loader.js +36 -0
- package/storages/request_manager.d.ts +119 -0
- package/storages/request_manager.js +1 -0
- package/storages/request_manager_tandem.d.ts +94 -0
- package/storages/request_manager_tandem.js +200 -0
- package/storages/request_queue.d.ts +293 -48
- package/storages/request_queue.js +766 -216
- package/storages/storage_instance_manager.d.ts +86 -0
- package/storages/storage_instance_manager.js +256 -0
- package/storages/storage_stats.d.ts +48 -0
- package/storages/storage_stats.js +29 -0
- package/storages/transaction.d.ts +271 -0
- package/storages/transaction.js +296 -0
- package/storages/utils.d.ts +59 -11
- package/storages/utils.js +75 -15
- package/system-info/cpu-info.d.ts +67 -0
- package/system-info/cpu-info.js +216 -0
- package/system-info/memory-info.d.ts +31 -0
- package/system-info/memory-info.js +115 -0
- package/system-info/ps-tree.d.ts +17 -0
- package/system-info/ps-tree.js +144 -0
- package/system-info/runtime.d.ts +14 -0
- package/system-info/runtime.js +80 -0
- package/typedefs.d.ts +0 -6
- package/typedefs.js +0 -1
- package/url.d.ts +9 -0
- package/url.js +11 -0
- package/validators.d.ts +23 -18
- package/validators.js +14 -18
- package/autoscaling/autoscaled_pool.d.ts +0 -300
- package/autoscaling/autoscaled_pool.d.ts.map +0 -1
- package/autoscaling/autoscaled_pool.js +0 -533
- package/autoscaling/autoscaled_pool.js.map +0 -1
- package/autoscaling/index.d.ts +0 -4
- package/autoscaling/index.d.ts.map +0 -1
- package/autoscaling/index.js +0 -4
- package/autoscaling/index.js.map +0 -1
- package/autoscaling/snapshotter.d.ts +0 -189
- package/autoscaling/snapshotter.d.ts.map +0 -1
- package/autoscaling/snapshotter.js +0 -292
- package/autoscaling/snapshotter.js.map +0 -1
- package/autoscaling/system_status.d.ts +0 -181
- package/autoscaling/system_status.d.ts.map +0 -1
- package/autoscaling/system_status.js +0 -169
- package/autoscaling/system_status.js.map +0 -1
- package/configuration.d.ts.map +0 -1
- package/configuration.js.map +0 -1
- package/cookie_utils.d.ts +0 -43
- package/cookie_utils.d.ts.map +0 -1
- package/cookie_utils.js +0 -113
- package/cookie_utils.js.map +0 -1
- package/crawlers/crawler_commons.d.ts +0 -199
- package/crawlers/crawler_commons.d.ts.map +0 -1
- package/crawlers/crawler_commons.js +0 -116
- package/crawlers/crawler_commons.js.map +0 -1
- package/crawlers/crawler_extension.d.ts +0 -12
- package/crawlers/crawler_extension.d.ts.map +0 -1
- package/crawlers/crawler_extension.js +0 -14
- package/crawlers/crawler_extension.js.map +0 -1
- package/crawlers/crawler_utils.d.ts +0 -10
- package/crawlers/crawler_utils.d.ts.map +0 -1
- package/crawlers/crawler_utils.js +0 -12
- package/crawlers/crawler_utils.js.map +0 -1
- package/crawlers/error_snapshotter.d.ts +0 -59
- package/crawlers/error_snapshotter.d.ts.map +0 -1
- package/crawlers/error_snapshotter.js +0 -118
- package/crawlers/error_snapshotter.js.map +0 -1
- package/crawlers/error_tracker.d.ts +0 -54
- package/crawlers/error_tracker.d.ts.map +0 -1
- package/crawlers/error_tracker.js +0 -309
- package/crawlers/error_tracker.js.map +0 -1
- package/crawlers/index.d.ts +0 -7
- package/crawlers/index.d.ts.map +0 -1
- package/crawlers/index.js +0 -7
- package/crawlers/index.js.map +0 -1
- package/crawlers/statistics.d.ts +0 -204
- package/crawlers/statistics.d.ts.map +0 -1
- package/crawlers/statistics.js +0 -346
- package/crawlers/statistics.js.map +0 -1
- package/enqueue_links/enqueue_links.d.ts +0 -242
- package/enqueue_links/enqueue_links.d.ts.map +0 -1
- package/enqueue_links/enqueue_links.js +0 -250
- package/enqueue_links/enqueue_links.js.map +0 -1
- package/enqueue_links/index.d.ts +0 -3
- package/enqueue_links/index.d.ts.map +0 -1
- package/enqueue_links/index.js +0 -3
- package/enqueue_links/index.js.map +0 -1
- package/enqueue_links/shared.d.ts +0 -67
- package/enqueue_links/shared.d.ts.map +0 -1
- package/enqueue_links/shared.js +0 -190
- package/enqueue_links/shared.js.map +0 -1
- package/errors.d.ts.map +0 -1
- package/errors.js.map +0 -1
- package/events/event_manager.d.ts.map +0 -1
- package/events/event_manager.js.map +0 -1
- package/events/index.d.ts.map +0 -1
- package/events/index.js.map +0 -1
- package/events/local_event_manager.d.ts.map +0 -1
- package/events/local_event_manager.js.map +0 -1
- package/http_clients/base-http-client.d.ts +0 -134
- package/http_clients/base-http-client.d.ts.map +0 -1
- package/http_clients/base-http-client.js +0 -33
- package/http_clients/base-http-client.js.map +0 -1
- package/http_clients/form-data-like.d.ts +0 -67
- package/http_clients/form-data-like.d.ts.map +0 -1
- package/http_clients/form-data-like.js +0 -5
- package/http_clients/form-data-like.js.map +0 -1
- package/http_clients/got-scraping-http-client.d.ts +0 -15
- package/http_clients/got-scraping-http-client.d.ts.map +0 -1
- package/http_clients/got-scraping-http-client.js +0 -69
- package/http_clients/got-scraping-http-client.js.map +0 -1
- package/http_clients/index.d.ts +0 -3
- package/http_clients/index.d.ts.map +0 -1
- package/http_clients/index.js +0 -3
- package/http_clients/index.js.map +0 -1
- package/index.d.ts.map +0 -1
- package/index.js.map +0 -1
- package/log.d.ts.map +0 -1
- package/log.js.map +0 -1
- package/proxy_configuration.d.ts.map +0 -1
- package/proxy_configuration.js.map +0 -1
- package/request.d.ts.map +0 -1
- package/request.js.map +0 -1
- package/router.d.ts +0 -134
- package/router.d.ts.map +0 -1
- package/router.js +0 -164
- package/router.js.map +0 -1
- package/serialization.d.ts.map +0 -1
- package/serialization.js.map +0 -1
- package/session_pool/consts.d.ts +0 -4
- package/session_pool/consts.d.ts.map +0 -1
- package/session_pool/consts.js +0 -4
- package/session_pool/consts.js.map +0 -1
- package/session_pool/errors.d.ts +0 -8
- package/session_pool/errors.d.ts.map +0 -1
- package/session_pool/errors.js +0 -12
- package/session_pool/errors.js.map +0 -1
- package/session_pool/events.d.ts +0 -3
- package/session_pool/events.d.ts.map +0 -1
- package/session_pool/events.js +0 -3
- package/session_pool/events.js.map +0 -1
- package/session_pool/index.d.ts +0 -6
- package/session_pool/index.d.ts.map +0 -1
- package/session_pool/index.js +0 -6
- package/session_pool/index.js.map +0 -1
- package/session_pool/session.d.ts +0 -199
- package/session_pool/session.d.ts.map +0 -1
- package/session_pool/session.js +0 -271
- package/session_pool/session.js.map +0 -1
- package/session_pool/session_pool.d.ts +0 -251
- package/session_pool/session_pool.d.ts.map +0 -1
- package/session_pool/session_pool.js +0 -375
- package/session_pool/session_pool.js.map +0 -1
- package/storages/access_checking.d.ts +0 -13
- package/storages/access_checking.d.ts.map +0 -1
- package/storages/access_checking.js +0 -14
- package/storages/access_checking.js.map +0 -1
- package/storages/dataset.d.ts.map +0 -1
- package/storages/dataset.js.map +0 -1
- package/storages/index.d.ts.map +0 -1
- package/storages/index.js.map +0 -1
- package/storages/key_value_store.d.ts.map +0 -1
- package/storages/key_value_store.js.map +0 -1
- package/storages/request_list.d.ts.map +0 -1
- package/storages/request_list.js.map +0 -1
- package/storages/request_provider.d.ts +0 -307
- package/storages/request_provider.d.ts.map +0 -1
- package/storages/request_provider.js +0 -555
- package/storages/request_provider.js.map +0 -1
- package/storages/request_queue.d.ts.map +0 -1
- package/storages/request_queue.js.map +0 -1
- package/storages/request_queue_v2.d.ts +0 -87
- package/storages/request_queue_v2.d.ts.map +0 -1
- package/storages/request_queue_v2.js +0 -438
- package/storages/request_queue_v2.js.map +0 -1
- package/storages/sitemap_request_list.d.ts +0 -245
- package/storages/sitemap_request_list.d.ts.map +0 -1
- package/storages/sitemap_request_list.js +0 -430
- package/storages/sitemap_request_list.js.map +0 -1
- package/storages/storage_manager.d.ts +0 -58
- package/storages/storage_manager.d.ts.map +0 -1
- package/storages/storage_manager.js +0 -105
- package/storages/storage_manager.js.map +0 -1
- package/storages/utils.d.ts.map +0 -1
- package/storages/utils.js.map +0 -1
- package/tsconfig.build.tsbuildinfo +0 -1
- package/typedefs.d.ts.map +0 -1
- package/typedefs.js.map +0 -1
- package/validators.d.ts.map +0 -1
- package/validators.js.map +0 -1
package/package.json
CHANGED
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
{
|
|
2
2
|
"name": "@crawlee/core",
|
|
3
|
-
"version": "4.0.0-beta.
|
|
3
|
+
"version": "4.0.0-beta.200",
|
|
4
4
|
"description": "The scalable web crawling and scraping library for JavaScript/Node.js. Enables development of data extraction and web automation jobs (not only) with headless Chrome and Puppeteer.",
|
|
5
5
|
"engines": {
|
|
6
6
|
"node": ">=22.0.0"
|
|
@@ -8,6 +8,7 @@
|
|
|
8
8
|
"type": "module",
|
|
9
9
|
"exports": {
|
|
10
10
|
".": "./index.js",
|
|
11
|
+
"./internal": "./internal.js",
|
|
11
12
|
"./package.json": "./package.json"
|
|
12
13
|
},
|
|
13
14
|
"keywords": [
|
|
@@ -38,7 +39,7 @@
|
|
|
38
39
|
},
|
|
39
40
|
"homepage": "https://crawlee.dev",
|
|
40
41
|
"scripts": {
|
|
41
|
-
"build": "
|
|
42
|
+
"build": "pnpm clean && pnpm compile && pnpm copy",
|
|
42
43
|
"clean": "rimraf ./dist",
|
|
43
44
|
"compile": "tsc -p tsconfig.build.json",
|
|
44
45
|
"copy": "tsx ../../scripts/copy.ts"
|
|
@@ -47,28 +48,26 @@
|
|
|
47
48
|
"access": "public"
|
|
48
49
|
},
|
|
49
50
|
"dependencies": {
|
|
50
|
-
"@apify/consts": "^
|
|
51
|
-
"@apify/datastructures": "^
|
|
52
|
-
"@apify/log": "^
|
|
53
|
-
"@apify/
|
|
54
|
-
"@apify/
|
|
55
|
-
"@
|
|
56
|
-
"@crawlee/
|
|
57
|
-
"@crawlee/types": "4.0.0-beta.
|
|
58
|
-
"@crawlee/utils": "4.0.0-beta.
|
|
51
|
+
"@apify/consts": "^3.0.1",
|
|
52
|
+
"@apify/datastructures": "^3.0.1",
|
|
53
|
+
"@apify/log": "^3.0.1",
|
|
54
|
+
"@apify/timeout": "^1.0.1",
|
|
55
|
+
"@apify/utilities": "^3.0.1",
|
|
56
|
+
"@crawlee/fs-storage": "4.0.0-beta.200",
|
|
57
|
+
"@crawlee/http-client": "4.0.0-beta.200",
|
|
58
|
+
"@crawlee/types": "4.0.0-beta.200",
|
|
59
|
+
"@crawlee/utils": "4.0.0-beta.200",
|
|
59
60
|
"@sapphire/async-queue": "^1.5.5",
|
|
61
|
+
"@standard-schema/spec": "^1.0.0",
|
|
60
62
|
"@vladfrangu/async_event_emitter": "^2.4.6",
|
|
63
|
+
"content-type": "^1.0.5",
|
|
61
64
|
"csv-stringify": "^6.5.2",
|
|
62
|
-
"fs-extra": "^11.3.0",
|
|
63
|
-
"got-scraping": "^4.1.1",
|
|
64
65
|
"json5": "^2.2.3",
|
|
65
|
-
"
|
|
66
|
-
"
|
|
67
|
-
"stream-json": "^1.9.1",
|
|
68
|
-
"tldts": "^7.0.6",
|
|
69
|
-
"tough-cookie": "^5.1.2",
|
|
66
|
+
"mime-types": "^3.0.1",
|
|
67
|
+
"stream-json": "^3.7.0",
|
|
70
68
|
"tslib": "^2.8.1",
|
|
71
|
-
"type-fest": "^4.41.0"
|
|
69
|
+
"type-fest": "^4.41.0",
|
|
70
|
+
"zod": "^4.5.4"
|
|
72
71
|
},
|
|
73
72
|
"lerna": {
|
|
74
73
|
"command": {
|
|
@@ -77,5 +76,5 @@
|
|
|
77
76
|
}
|
|
78
77
|
}
|
|
79
78
|
},
|
|
80
|
-
"gitHead": "
|
|
79
|
+
"gitHead": "471663956b78edb995e824dd906776a28cc04535"
|
|
81
80
|
}
|
package/proxy_configuration.d.ts
CHANGED
|
@@ -1,8 +1,6 @@
|
|
|
1
|
-
import type {
|
|
1
|
+
import type { ProxyInfo } from '@crawlee/types';
|
|
2
2
|
export interface ProxyConfigurationFunction {
|
|
3
|
-
(
|
|
4
|
-
request?: Request;
|
|
5
|
-
}): string | null | Promise<string | null>;
|
|
3
|
+
(): string | null | Promise<string | null>;
|
|
6
4
|
}
|
|
7
5
|
export interface ProxyConfigurationOptions {
|
|
8
6
|
/**
|
|
@@ -10,120 +8,38 @@ export interface ProxyConfigurationOptions {
|
|
|
10
8
|
* Custom proxies are not compatible with Apify Proxy and an attempt to use both
|
|
11
9
|
* configuration options will cause an error to be thrown on initialize.
|
|
12
10
|
*/
|
|
13
|
-
proxyUrls?: string[];
|
|
11
|
+
proxyUrls?: (string | null)[];
|
|
14
12
|
/**
|
|
15
|
-
* Custom function that allows you to generate the new proxy URL dynamically.
|
|
13
|
+
* Custom function that allows you to generate the new proxy URL dynamically.
|
|
16
14
|
* Can return either stringified proxy URL or `null` if the proxy should not be used. Can be asynchronous.
|
|
17
15
|
*
|
|
18
16
|
* This function is used to generate the URL when {@link ProxyConfiguration.newUrl} or {@link ProxyConfiguration.newProxyInfo} is called.
|
|
19
17
|
*/
|
|
20
18
|
newUrlFunction?: ProxyConfigurationFunction;
|
|
21
19
|
/**
|
|
22
|
-
*
|
|
23
|
-
*
|
|
24
|
-
*
|
|
25
|
-
*
|
|
26
|
-
* The crawler probes lower-level proxies at intervals to check if it can make the tier downshift.
|
|
27
|
-
*
|
|
28
|
-
* This feature is useful when you have a set of proxies with different performance characteristics (speed, price, antibot performance etc.) and you want to use the best one for each domain.
|
|
29
|
-
*
|
|
30
|
-
* Use `null` as a proxy URL to disable the proxy for the given tier.
|
|
20
|
+
* When truthy, the constructor throws unless one of `proxyUrls` / `newUrlFunction` was given. Falsy by
|
|
21
|
+
* default, so a bare `ProxyConfiguration` can be constructed. Set by the Apify SDK, which builds the options
|
|
22
|
+
* object itself; declared here only so that it stays type-checkable.
|
|
23
|
+
* @internal
|
|
31
24
|
*/
|
|
32
|
-
|
|
33
|
-
}
|
|
34
|
-
export interface TieredProxy {
|
|
35
|
-
proxyUrl: string | null;
|
|
36
|
-
proxyTier?: number;
|
|
25
|
+
validateRequired?: boolean;
|
|
37
26
|
}
|
|
38
27
|
/**
|
|
39
|
-
*
|
|
40
|
-
*
|
|
41
|
-
* Outside of crawlers, you can get this object by calling {@link ProxyConfiguration.newProxyInfo}.
|
|
28
|
+
* Minimal contract that any object passed to a crawler as its `proxyConfiguration`
|
|
29
|
+
* option must satisfy.
|
|
42
30
|
*
|
|
43
|
-
*
|
|
44
|
-
*
|
|
45
|
-
*
|
|
46
|
-
* const proxyConfiguration = new ProxyConfiguration({
|
|
47
|
-
* proxyUrls: ['...', '...'] // List of Proxy URLs to rotate
|
|
48
|
-
* });
|
|
49
|
-
*
|
|
50
|
-
* // Getting proxyInfo object by calling class method directly
|
|
51
|
-
* const proxyInfo = await proxyConfiguration.newProxyInfo();
|
|
52
|
-
*
|
|
53
|
-
* // In crawler
|
|
54
|
-
* const crawler = new CheerioCrawler({
|
|
55
|
-
* // ...
|
|
56
|
-
* proxyConfiguration,
|
|
57
|
-
* requestHandler({ proxyInfo }) {
|
|
58
|
-
* // Getting used proxy URL
|
|
59
|
-
* const proxyUrl = proxyInfo.url;
|
|
60
|
-
*
|
|
61
|
-
* // Getting ID of used Session
|
|
62
|
-
* const sessionIdentifier = proxyInfo.sessionId;
|
|
63
|
-
* }
|
|
64
|
-
* })
|
|
65
|
-
*
|
|
66
|
-
* ```
|
|
67
|
-
*/
|
|
68
|
-
export interface ProxyInfo {
|
|
69
|
-
/**
|
|
70
|
-
* The identifier of used {@link Session}, if used.
|
|
71
|
-
*/
|
|
72
|
-
sessionId?: string;
|
|
73
|
-
/**
|
|
74
|
-
* The URL of the proxy.
|
|
75
|
-
*/
|
|
76
|
-
url: string;
|
|
77
|
-
/**
|
|
78
|
-
* Username for the proxy.
|
|
79
|
-
*/
|
|
80
|
-
username?: string;
|
|
81
|
-
/**
|
|
82
|
-
* User's password for the proxy.
|
|
83
|
-
*/
|
|
84
|
-
password: string;
|
|
85
|
-
/**
|
|
86
|
-
* Hostname of your proxy.
|
|
87
|
-
*/
|
|
88
|
-
hostname: string;
|
|
89
|
-
/**
|
|
90
|
-
* Proxy port.
|
|
91
|
-
*/
|
|
92
|
-
port: number | string;
|
|
93
|
-
/**
|
|
94
|
-
* Proxy tier for the current proxy, if applicable (only for `tieredProxyUrls`).
|
|
95
|
-
*/
|
|
96
|
-
proxyTier?: number;
|
|
97
|
-
}
|
|
98
|
-
interface TieredProxyOptions {
|
|
99
|
-
request?: Request;
|
|
100
|
-
proxyTier?: number;
|
|
101
|
-
}
|
|
102
|
-
/**
|
|
103
|
-
* Internal class for tracking the proxy tier history for a specific domain.
|
|
31
|
+
* Implement this interface to plug a custom proxy-provisioning strategy into any Crawlee
|
|
32
|
+
* crawler — for example a remote proxy service or a thin wrapper around the built-in
|
|
33
|
+
* `ProxyConfiguration` with different rotation rules. *
|
|
104
34
|
*
|
|
105
|
-
*
|
|
35
|
+
* @category Scaling
|
|
106
36
|
*/
|
|
107
|
-
|
|
108
|
-
private histogram;
|
|
109
|
-
private currentTier;
|
|
110
|
-
constructor(tieredProxyUrls: (string | null)[][]);
|
|
37
|
+
export interface IProxyConfiguration {
|
|
111
38
|
/**
|
|
112
|
-
*
|
|
39
|
+
* Creates a new {@link ProxyInfo} object describing the proxy to use.
|
|
40
|
+
* Returns `undefined` when no proxy should be used.
|
|
113
41
|
*/
|
|
114
|
-
|
|
115
|
-
/**
|
|
116
|
-
* Increases the error score for the given proxy tier. This raises the chance of picking a different proxy tier for the subsequent requests.
|
|
117
|
-
*
|
|
118
|
-
* The error score is increased by 10 for the given tier. This means that this tier will be disadvantaged for the next 10 requests (every new request prediction decreases the error score by 1).
|
|
119
|
-
* @param tier The proxy tier to mark as problematic.
|
|
120
|
-
*/
|
|
121
|
-
addError(tier: number): void;
|
|
122
|
-
/**
|
|
123
|
-
* Returns the best proxy tier for the next request based on the error history for different proxy tiers.
|
|
124
|
-
* @returns The proxy tier prediction
|
|
125
|
-
*/
|
|
126
|
-
predictTier(): number;
|
|
42
|
+
newProxyInfo(): Promise<ProxyInfo | undefined>;
|
|
127
43
|
}
|
|
128
44
|
/**
|
|
129
45
|
* Configures connection to a proxy server with the provided options. Proxy servers are used to prevent target websites from blocking
|
|
@@ -153,16 +69,9 @@ declare class ProxyTierTracker {
|
|
|
153
69
|
* ```
|
|
154
70
|
* @category Scaling
|
|
155
71
|
*/
|
|
156
|
-
export declare class ProxyConfiguration {
|
|
157
|
-
|
|
158
|
-
|
|
159
|
-
protected proxyUrls?: string[];
|
|
160
|
-
protected tieredProxyUrls?: (string | null)[][];
|
|
161
|
-
protected usedProxyUrls: Map<string, string>;
|
|
162
|
-
protected newUrlFunction?: ProxyConfigurationFunction;
|
|
163
|
-
// @ts-ignore optional peer dependency or compatibility with es2022
|
|
164
|
-
protected log: import("@apify/log").Log;
|
|
165
|
-
protected domainTiers: Map<string, ProxyTierTracker>;
|
|
72
|
+
export declare class ProxyConfiguration implements IProxyConfiguration {
|
|
73
|
+
#private;
|
|
74
|
+
readonly isManInTheMiddle = false;
|
|
166
75
|
/**
|
|
167
76
|
* Creates a {@link ProxyConfiguration} instance based on the provided options. Proxy servers are used to prevent target websites from
|
|
168
77
|
* blocking your crawlers based on IP address rate limits or blacklists. Setting proxy configuration in your crawlers automatically configures
|
|
@@ -190,57 +99,22 @@ export declare class ProxyConfiguration {
|
|
|
190
99
|
* the currently used proxy via the requestHandler parameter `proxyInfo`.
|
|
191
100
|
* Use it if you want to work with a rich representation of a proxy URL.
|
|
192
101
|
* If you need the URL string only, use {@link ProxyConfiguration.newUrl}.
|
|
193
|
-
* @param [sessionId]
|
|
194
|
-
* Represents the identifier of user {@link Session} that can be managed by the {@link SessionPool} or
|
|
195
|
-
* you can use the Apify Proxy [Session](https://docs.apify.com/proxy#sessions) identifier.
|
|
196
|
-
* When the provided sessionId is a number, it's converted to a string. Property sessionId of
|
|
197
|
-
* {@link ProxyInfo} is always returned as a type string.
|
|
198
102
|
*
|
|
199
|
-
* All the HTTP requests going through the proxy with the same session identifier
|
|
200
|
-
* will use the same target proxy server (i.e. the same IP address).
|
|
201
|
-
* The identifier must not be longer than 50 characters and include only the following: `0-9`, `a-z`, `A-Z`, `"."`, `"_"` and `"~"`.
|
|
202
103
|
* @return Represents information about used proxy and its configuration.
|
|
203
104
|
*/
|
|
204
|
-
newProxyInfo(
|
|
205
|
-
/**
|
|
206
|
-
* Given a session identifier and a request / proxy tier, this function returns a new proxy URL based on the provided configuration options.
|
|
207
|
-
* @param _sessionId Session identifier
|
|
208
|
-
* @param options Options for the tiered proxy rotation
|
|
209
|
-
* @returns An object with the proxy URL and the proxy tier used.
|
|
210
|
-
*/
|
|
211
|
-
protected _handleTieredUrl(_sessionId: string, options?: TieredProxyOptions): TieredProxy;
|
|
212
|
-
/**
|
|
213
|
-
* Given a `Request` object, this function returns the tier of the proxy that should be used for the request.
|
|
214
|
-
*
|
|
215
|
-
* This returns `null` if `tieredProxyUrls` option is not set.
|
|
216
|
-
*/
|
|
217
|
-
protected predictProxyTier(request: Request): number | null;
|
|
105
|
+
newProxyInfo(): Promise<ProxyInfo | undefined>;
|
|
218
106
|
/**
|
|
219
|
-
* Returns a new proxy URL based on provided configuration options
|
|
220
|
-
* @param [sessionId]
|
|
221
|
-
* Represents the identifier of user {@link Session} that can be managed by the {@link SessionPool} or
|
|
222
|
-
* you can use the Apify Proxy [Session](https://docs.apify.com/proxy#sessions) identifier.
|
|
223
|
-
* When the provided sessionId is a number, it's converted to a string.
|
|
107
|
+
* Returns a new proxy URL based on provided configuration options.
|
|
224
108
|
*
|
|
225
|
-
* All the HTTP requests going through the proxy with the same session identifier
|
|
226
|
-
* will use the same target proxy server (i.e. the same IP address).
|
|
227
|
-
* The identifier must not be longer than 50 characters and include only the following: `0-9`, `a-z`, `A-Z`, `"."`, `"_"` and `"~"`.
|
|
228
109
|
* @return A string with a proxy URL, including authentication credentials and port number.
|
|
229
110
|
* For example, `http://bob:password123@proxy.example.com:8000`
|
|
230
111
|
*/
|
|
231
|
-
newUrl(
|
|
232
|
-
|
|
233
|
-
* Handles custom url rotation with session
|
|
234
|
-
*/
|
|
235
|
-
protected _handleCustomUrl(sessionId?: string): string;
|
|
112
|
+
newUrl(): Promise<string | undefined>;
|
|
113
|
+
private handleProxyUrlsList;
|
|
236
114
|
/**
|
|
237
115
|
* Calls the custom newUrlFunction and checks format of its return value
|
|
238
116
|
*/
|
|
239
|
-
|
|
240
|
-
|
|
241
|
-
|
|
242
|
-
protected _throwCannotCombineCustomMethods(): never;
|
|
243
|
-
protected _throwNoOptionsProvided(): never;
|
|
117
|
+
private callNewUrlFunction;
|
|
118
|
+
private throwCannotCombineCustomMethods;
|
|
119
|
+
private throwNoOptionsProvided;
|
|
244
120
|
}
|
|
245
|
-
export {};
|
|
246
|
-
//# sourceMappingURL=proxy_configuration.d.ts.map
|
package/proxy_configuration.js
CHANGED
|
@@ -1,55 +1,12 @@
|
|
|
1
|
-
import
|
|
2
|
-
import
|
|
3
|
-
|
|
4
|
-
|
|
5
|
-
|
|
6
|
-
|
|
7
|
-
|
|
8
|
-
|
|
9
|
-
|
|
10
|
-
histogram;
|
|
11
|
-
currentTier;
|
|
12
|
-
constructor(tieredProxyUrls) {
|
|
13
|
-
this.histogram = tieredProxyUrls.map(() => 0);
|
|
14
|
-
this.currentTier = 0;
|
|
15
|
-
}
|
|
16
|
-
/**
|
|
17
|
-
* Processes a single step of the algorithm and updates the current tier prediction based on the error history.
|
|
18
|
-
*/
|
|
19
|
-
processStep() {
|
|
20
|
-
this.histogram.forEach((x, i) => {
|
|
21
|
-
if (this.currentTier === i)
|
|
22
|
-
return;
|
|
23
|
-
if (x > 0)
|
|
24
|
-
this.histogram[i]--;
|
|
25
|
-
});
|
|
26
|
-
const left = this.currentTier > 0 ? this.histogram[this.currentTier - 1] : Infinity;
|
|
27
|
-
const right = this.currentTier < this.histogram.length - 1 ? this.histogram[this.currentTier + 1] : Infinity;
|
|
28
|
-
if (this.histogram[this.currentTier] > Math.min(left, right)) {
|
|
29
|
-
this.currentTier = left <= right ? this.currentTier - 1 : this.currentTier + 1;
|
|
30
|
-
}
|
|
31
|
-
else if (this.histogram[this.currentTier] === left) {
|
|
32
|
-
this.currentTier--;
|
|
33
|
-
}
|
|
34
|
-
}
|
|
35
|
-
/**
|
|
36
|
-
* Increases the error score for the given proxy tier. This raises the chance of picking a different proxy tier for the subsequent requests.
|
|
37
|
-
*
|
|
38
|
-
* The error score is increased by 10 for the given tier. This means that this tier will be disadvantaged for the next 10 requests (every new request prediction decreases the error score by 1).
|
|
39
|
-
* @param tier The proxy tier to mark as problematic.
|
|
40
|
-
*/
|
|
41
|
-
addError(tier) {
|
|
42
|
-
this.histogram[tier] += 10;
|
|
43
|
-
}
|
|
44
|
-
/**
|
|
45
|
-
* Returns the best proxy tier for the next request based on the error history for different proxy tiers.
|
|
46
|
-
* @returns The proxy tier prediction
|
|
47
|
-
*/
|
|
48
|
-
predictTier() {
|
|
49
|
-
this.processStep();
|
|
50
|
-
return this.currentTier;
|
|
51
|
-
}
|
|
52
|
-
}
|
|
1
|
+
import { z } from 'zod';
|
|
2
|
+
import { parseArgument, schemas } from './validators.js';
|
|
3
|
+
const proxyConfigurationOptionsSchema = z.strictObject({
|
|
4
|
+
proxyUrls: z
|
|
5
|
+
.array(z.union([z.url(), z.null()]))
|
|
6
|
+
.nonempty()
|
|
7
|
+
.optional(),
|
|
8
|
+
newUrlFunction: schemas.anyFunction.optional(),
|
|
9
|
+
});
|
|
53
10
|
/**
|
|
54
11
|
* Configures connection to a proxy server with the provided options. Proxy servers are used to prevent target websites from blocking
|
|
55
12
|
* your crawlers based on IP address rate limits or blacklists. Setting proxy configuration in your crawlers automatically configures
|
|
@@ -80,13 +37,9 @@ class ProxyTierTracker {
|
|
|
80
37
|
*/
|
|
81
38
|
export class ProxyConfiguration {
|
|
82
39
|
isManInTheMiddle = false;
|
|
83
|
-
nextCustomUrlIndex = 0;
|
|
84
|
-
proxyUrls;
|
|
85
|
-
|
|
86
|
-
usedProxyUrls = new Map();
|
|
87
|
-
newUrlFunction;
|
|
88
|
-
log = log.child({ prefix: 'ProxyConfiguration' });
|
|
89
|
-
domainTiers = new Map();
|
|
40
|
+
#nextCustomUrlIndex = 0;
|
|
41
|
+
#proxyUrls;
|
|
42
|
+
#newUrlFunction;
|
|
90
43
|
/**
|
|
91
44
|
* Creates a {@link ProxyConfiguration} instance based on the provided options. Proxy servers are used to prevent target websites from
|
|
92
45
|
* blocking your crawlers based on IP address rate limits or blacklists. Setting proxy configuration in your crawlers automatically configures
|
|
@@ -108,20 +61,21 @@ export class ProxyConfiguration {
|
|
|
108
61
|
* ```
|
|
109
62
|
*/
|
|
110
63
|
constructor(options = {}) {
|
|
64
|
+
// `validateRequired` is destructured off before the strict-object parse on purpose: the Apify SDK passes it
|
|
65
|
+
// through a computed key (`['validateRequired' as string]: false`), and leaving it in `rest` would make
|
|
66
|
+
// `Actor.createProxyConfiguration()` fail the `z.strictObject` check with a `ZodError`.
|
|
111
67
|
const { validateRequired, ...rest } = options;
|
|
112
|
-
|
|
113
|
-
|
|
114
|
-
|
|
115
|
-
|
|
116
|
-
})
|
|
117
|
-
|
|
118
|
-
|
|
119
|
-
this._throwCannotCombineCustomMethods();
|
|
68
|
+
if ('tieredProxyUrls' in rest) {
|
|
69
|
+
throw new Error('The `tieredProxyUrls` option has been removed in Crawlee v4. ' +
|
|
70
|
+
'See the v4 upgrading guide for the recommended migration to named sessions.');
|
|
71
|
+
}
|
|
72
|
+
const { proxyUrls, newUrlFunction } = parseArgument(rest, proxyConfigurationOptionsSchema);
|
|
73
|
+
if (proxyUrls && newUrlFunction)
|
|
74
|
+
this.throwCannotCombineCustomMethods();
|
|
120
75
|
if (!proxyUrls && !newUrlFunction && validateRequired)
|
|
121
|
-
this.
|
|
122
|
-
this
|
|
123
|
-
this
|
|
124
|
-
this.tieredProxyUrls = tieredProxyUrls;
|
|
76
|
+
this.throwNoOptionsProvided();
|
|
77
|
+
this.#proxyUrls = proxyUrls;
|
|
78
|
+
this.#newUrlFunction = newUrlFunction;
|
|
125
79
|
}
|
|
126
80
|
/**
|
|
127
81
|
* This function creates a new {@link ProxyInfo} info object.
|
|
@@ -129,140 +83,42 @@ export class ProxyConfiguration {
|
|
|
129
83
|
* the currently used proxy via the requestHandler parameter `proxyInfo`.
|
|
130
84
|
* Use it if you want to work with a rich representation of a proxy URL.
|
|
131
85
|
* If you need the URL string only, use {@link ProxyConfiguration.newUrl}.
|
|
132
|
-
* @param [sessionId]
|
|
133
|
-
* Represents the identifier of user {@link Session} that can be managed by the {@link SessionPool} or
|
|
134
|
-
* you can use the Apify Proxy [Session](https://docs.apify.com/proxy#sessions) identifier.
|
|
135
|
-
* When the provided sessionId is a number, it's converted to a string. Property sessionId of
|
|
136
|
-
* {@link ProxyInfo} is always returned as a type string.
|
|
137
86
|
*
|
|
138
|
-
* All the HTTP requests going through the proxy with the same session identifier
|
|
139
|
-
* will use the same target proxy server (i.e. the same IP address).
|
|
140
|
-
* The identifier must not be longer than 50 characters and include only the following: `0-9`, `a-z`, `A-Z`, `"."`, `"_"` and `"~"`.
|
|
141
87
|
* @return Represents information about used proxy and its configuration.
|
|
142
88
|
*/
|
|
143
|
-
async newProxyInfo(
|
|
144
|
-
|
|
145
|
-
sessionId = `${sessionId}`;
|
|
146
|
-
let url;
|
|
147
|
-
let tier;
|
|
148
|
-
if (this.tieredProxyUrls) {
|
|
149
|
-
const { proxyUrl, proxyTier } = this._handleTieredUrl(sessionId ?? cryptoRandomObjectId(6), options);
|
|
150
|
-
url = proxyUrl ?? undefined;
|
|
151
|
-
tier = proxyTier;
|
|
152
|
-
}
|
|
153
|
-
else {
|
|
154
|
-
url = await this.newUrl(sessionId, options);
|
|
155
|
-
}
|
|
89
|
+
async newProxyInfo() {
|
|
90
|
+
const url = await this.newUrl();
|
|
156
91
|
if (!url)
|
|
157
92
|
return undefined;
|
|
158
93
|
const { username, password, port, hostname } = new URL(url);
|
|
159
94
|
return {
|
|
160
|
-
sessionId,
|
|
161
95
|
url,
|
|
162
96
|
username: decodeURIComponent(username),
|
|
163
97
|
password: decodeURIComponent(password),
|
|
164
98
|
hostname,
|
|
165
99
|
port: port,
|
|
166
|
-
proxyTier: tier,
|
|
167
100
|
};
|
|
168
101
|
}
|
|
169
102
|
/**
|
|
170
|
-
*
|
|
171
|
-
* @param _sessionId Session identifier
|
|
172
|
-
* @param options Options for the tiered proxy rotation
|
|
173
|
-
* @returns An object with the proxy URL and the proxy tier used.
|
|
174
|
-
*/
|
|
175
|
-
_handleTieredUrl(_sessionId, options) {
|
|
176
|
-
if (!this.tieredProxyUrls)
|
|
177
|
-
throw new Error('Tiered proxy URLs are not set');
|
|
178
|
-
if (!options || (!options?.request && options?.proxyTier === undefined)) {
|
|
179
|
-
const allProxyUrls = this.tieredProxyUrls.flat();
|
|
180
|
-
return {
|
|
181
|
-
proxyUrl: allProxyUrls[this.nextCustomUrlIndex++ % allProxyUrls.length],
|
|
182
|
-
};
|
|
183
|
-
}
|
|
184
|
-
let tierPrediction = options.proxyTier;
|
|
185
|
-
if (typeof tierPrediction !== 'number') {
|
|
186
|
-
tierPrediction = this.predictProxyTier(options.request);
|
|
187
|
-
}
|
|
188
|
-
const proxyTier = this.tieredProxyUrls[tierPrediction];
|
|
189
|
-
return {
|
|
190
|
-
proxyUrl: proxyTier[this.nextCustomUrlIndex++ % proxyTier.length],
|
|
191
|
-
proxyTier: tierPrediction,
|
|
192
|
-
};
|
|
193
|
-
}
|
|
194
|
-
/**
|
|
195
|
-
* Given a `Request` object, this function returns the tier of the proxy that should be used for the request.
|
|
196
|
-
*
|
|
197
|
-
* This returns `null` if `tieredProxyUrls` option is not set.
|
|
198
|
-
*/
|
|
199
|
-
predictProxyTier(request) {
|
|
200
|
-
if (!this.tieredProxyUrls)
|
|
201
|
-
return null;
|
|
202
|
-
const domain = new URL(request.url).hostname;
|
|
203
|
-
if (!this.domainTiers.has(domain)) {
|
|
204
|
-
this.domainTiers.set(domain, new ProxyTierTracker(this.tieredProxyUrls));
|
|
205
|
-
}
|
|
206
|
-
request.userData.__crawlee ??= {};
|
|
207
|
-
const tracker = this.domainTiers.get(domain);
|
|
208
|
-
if (typeof request.userData.__crawlee.lastProxyTier === 'number') {
|
|
209
|
-
tracker.addError(request.userData.__crawlee.lastProxyTier);
|
|
210
|
-
}
|
|
211
|
-
const tierPrediction = tracker.predictTier();
|
|
212
|
-
if (typeof request.userData.__crawlee.lastProxyTier === 'number' &&
|
|
213
|
-
request.userData.__crawlee.lastProxyTier !== tierPrediction) {
|
|
214
|
-
log.debug(`Changing proxy tier for domain "${domain}" from ${request.userData.__crawlee.lastProxyTier} to ${tierPrediction}.`);
|
|
215
|
-
}
|
|
216
|
-
request.userData.__crawlee.lastProxyTier = tierPrediction;
|
|
217
|
-
request.userData.__crawlee.forefront = true;
|
|
218
|
-
return tierPrediction;
|
|
219
|
-
}
|
|
220
|
-
/**
|
|
221
|
-
* Returns a new proxy URL based on provided configuration options and the `sessionId` parameter.
|
|
222
|
-
* @param [sessionId]
|
|
223
|
-
* Represents the identifier of user {@link Session} that can be managed by the {@link SessionPool} or
|
|
224
|
-
* you can use the Apify Proxy [Session](https://docs.apify.com/proxy#sessions) identifier.
|
|
225
|
-
* When the provided sessionId is a number, it's converted to a string.
|
|
103
|
+
* Returns a new proxy URL based on provided configuration options.
|
|
226
104
|
*
|
|
227
|
-
* All the HTTP requests going through the proxy with the same session identifier
|
|
228
|
-
* will use the same target proxy server (i.e. the same IP address).
|
|
229
|
-
* The identifier must not be longer than 50 characters and include only the following: `0-9`, `a-z`, `A-Z`, `"."`, `"_"` and `"~"`.
|
|
230
105
|
* @return A string with a proxy URL, including authentication credentials and port number.
|
|
231
106
|
* For example, `http://bob:password123@proxy.example.com:8000`
|
|
232
107
|
*/
|
|
233
|
-
async newUrl(
|
|
234
|
-
if (
|
|
235
|
-
|
|
236
|
-
if (this.newUrlFunction) {
|
|
237
|
-
return (await this._callNewUrlFunction(sessionId, { request: options?.request })) ?? undefined;
|
|
238
|
-
}
|
|
239
|
-
if (this.tieredProxyUrls) {
|
|
240
|
-
return this._handleTieredUrl(sessionId ?? cryptoRandomObjectId(6), options).proxyUrl ?? undefined;
|
|
108
|
+
async newUrl() {
|
|
109
|
+
if (this.#newUrlFunction) {
|
|
110
|
+
return (await this.callNewUrlFunction()) ?? undefined;
|
|
241
111
|
}
|
|
242
|
-
return this.
|
|
112
|
+
return this.handleProxyUrlsList() ?? undefined;
|
|
243
113
|
}
|
|
244
|
-
|
|
245
|
-
|
|
246
|
-
*/
|
|
247
|
-
_handleCustomUrl(sessionId) {
|
|
248
|
-
let customUrlToUse;
|
|
249
|
-
if (!sessionId) {
|
|
250
|
-
return this.proxyUrls[this.nextCustomUrlIndex++ % this.proxyUrls.length];
|
|
251
|
-
}
|
|
252
|
-
if (this.usedProxyUrls.has(sessionId)) {
|
|
253
|
-
customUrlToUse = this.usedProxyUrls.get(sessionId);
|
|
254
|
-
}
|
|
255
|
-
else {
|
|
256
|
-
customUrlToUse = this.proxyUrls[this.nextCustomUrlIndex++ % this.proxyUrls.length];
|
|
257
|
-
this.usedProxyUrls.set(sessionId, customUrlToUse);
|
|
258
|
-
}
|
|
259
|
-
return customUrlToUse;
|
|
114
|
+
handleProxyUrlsList() {
|
|
115
|
+
return this.#proxyUrls[this.#nextCustomUrlIndex++ % this.#proxyUrls.length];
|
|
260
116
|
}
|
|
261
117
|
/**
|
|
262
118
|
* Calls the custom newUrlFunction and checks format of its return value
|
|
263
119
|
*/
|
|
264
|
-
async
|
|
265
|
-
const proxyUrl = await this
|
|
120
|
+
async callNewUrlFunction() {
|
|
121
|
+
const proxyUrl = await this.#newUrlFunction();
|
|
266
122
|
try {
|
|
267
123
|
if (proxyUrl) {
|
|
268
124
|
new URL(proxyUrl); // eslint-disable-line no-new
|
|
@@ -273,11 +129,10 @@ export class ProxyConfiguration {
|
|
|
273
129
|
throw new Error(`The provided newUrlFunction did not return a valid URL.\nCause: ${err.message}`);
|
|
274
130
|
}
|
|
275
131
|
}
|
|
276
|
-
|
|
132
|
+
throwCannotCombineCustomMethods() {
|
|
277
133
|
throw new Error('Cannot combine custom proxies "options.proxyUrls" with custom generating function "options.newUrlFunction".');
|
|
278
134
|
}
|
|
279
|
-
|
|
135
|
+
throwNoOptionsProvided() {
|
|
280
136
|
throw new Error('One of "options.proxyUrls" or "options.newUrlFunction" needs to be provided.');
|
|
281
137
|
}
|
|
282
138
|
}
|
|
283
|
-
//# sourceMappingURL=proxy_configuration.js.map
|