@crawlee/core 4.0.0-beta.13 → 4.0.0-beta.131

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (280) hide show
  1. package/README.md +17 -13
  2. package/autoscaling/autoscaled_pool.d.ts +67 -172
  3. package/autoscaling/autoscaled_pool.js +182 -329
  4. package/autoscaling/concurrency_system.d.ts +268 -0
  5. package/autoscaling/concurrency_system.js +362 -0
  6. package/autoscaling/cpu_load_signal.d.ts +43 -0
  7. package/autoscaling/cpu_load_signal.js +47 -0
  8. package/autoscaling/event_loop_load_signal.d.ts +51 -0
  9. package/autoscaling/event_loop_load_signal.js +60 -0
  10. package/autoscaling/index.d.ts +6 -1
  11. package/autoscaling/index.js +6 -1
  12. package/autoscaling/load_signal.d.ts +100 -0
  13. package/autoscaling/load_signal.js +105 -0
  14. package/autoscaling/memory_load_signal.d.ts +47 -0
  15. package/autoscaling/memory_load_signal.js +106 -0
  16. package/autoscaling/snapshotter.d.ts +58 -163
  17. package/autoscaling/snapshotter.js +45 -263
  18. package/autoscaling/storage_backend_load_signal.d.ts +56 -0
  19. package/autoscaling/storage_backend_load_signal.js +73 -0
  20. package/autoscaling/system_status.d.ts +67 -89
  21. package/autoscaling/system_status.js +92 -122
  22. package/autoscaling/weighted_avg.d.ts +5 -0
  23. package/autoscaling/weighted_avg.js +14 -0
  24. package/byte_utils.d.ts +17 -0
  25. package/byte_utils.js +42 -0
  26. package/configuration.d.ts +96 -223
  27. package/configuration.js +170 -222
  28. package/cookie_utils.d.ts +3 -2
  29. package/cookie_utils.js +18 -7
  30. package/crawlers/context_pipeline.d.ts +10 -1
  31. package/crawlers/context_pipeline.js +31 -8
  32. package/crawlers/crawler_commons.d.ts +90 -129
  33. package/crawlers/crawler_commons.js +1 -116
  34. package/crawlers/error_snapshotter.d.ts +2 -5
  35. package/crawlers/error_snapshotter.js +7 -8
  36. package/crawlers/error_tracker.d.ts +0 -1
  37. package/crawlers/error_tracker.js +0 -1
  38. package/crawlers/index.d.ts +1 -3
  39. package/crawlers/index.js +0 -3
  40. package/crawlers/internals/types.d.ts +0 -1
  41. package/crawlers/internals/types.js +0 -1
  42. package/crawlers/statistics.d.ts +187 -63
  43. package/crawlers/statistics.js +354 -164
  44. package/debug.d.ts +36 -0
  45. package/debug.js +70 -0
  46. package/enqueue_links/enqueue_links.d.ts +68 -154
  47. package/enqueue_links/enqueue_links.js +38 -210
  48. package/enqueue_links/index.d.ts +0 -1
  49. package/enqueue_links/index.js +0 -1
  50. package/enqueue_links/shared.d.ts +52 -26
  51. package/enqueue_links/shared.js +115 -66
  52. package/errors.d.ts +72 -4
  53. package/errors.js +89 -5
  54. package/events/event_manager.d.ts +35 -9
  55. package/events/event_manager.js +12 -13
  56. package/events/index.d.ts +0 -1
  57. package/events/index.js +0 -1
  58. package/events/local_event_manager.d.ts +15 -3
  59. package/events/local_event_manager.js +39 -13
  60. package/http.d.ts +9 -0
  61. package/http.js +28 -0
  62. package/index.d.ts +7 -4
  63. package/index.js +6 -3
  64. package/iterables.d.ts +79 -0
  65. package/iterables.js +134 -0
  66. package/log.d.ts +82 -3
  67. package/log.js +106 -1
  68. package/memory-storage/consts.d.ts +4 -0
  69. package/memory-storage/consts.js +4 -0
  70. package/memory-storage/index.d.ts +1 -0
  71. package/memory-storage/index.js +1 -0
  72. package/memory-storage/memory-storage.d.ts +38 -0
  73. package/memory-storage/memory-storage.js +130 -0
  74. package/memory-storage/resource-clients/common/base-client.d.ts +4 -0
  75. package/memory-storage/resource-clients/common/base-client.js +6 -0
  76. package/memory-storage/resource-clients/dataset.d.ts +40 -0
  77. package/memory-storage/resource-clients/dataset.js +108 -0
  78. package/memory-storage/resource-clients/key-value-store.d.ts +63 -0
  79. package/memory-storage/resource-clients/key-value-store.js +201 -0
  80. package/memory-storage/resource-clients/request-queue.d.ts +77 -0
  81. package/memory-storage/resource-clients/request-queue.js +409 -0
  82. package/memory-storage/utils.d.ts +16 -0
  83. package/memory-storage/utils.js +41 -0
  84. package/owned_or_injected.d.ts +58 -0
  85. package/owned_or_injected.js +98 -0
  86. package/package.json +14 -13
  87. package/proxy_configuration.d.ts +24 -132
  88. package/proxy_configuration.js +32 -147
  89. package/recoverable_state.d.ts +155 -0
  90. package/recoverable_state.js +231 -0
  91. package/request.d.ts +80 -16
  92. package/request.js +147 -68
  93. package/router.d.ts +193 -21
  94. package/router.js +188 -43
  95. package/serialization.d.ts +0 -1
  96. package/serialization.js +15 -15
  97. package/service_locator.d.ts +165 -0
  98. package/service_locator.js +253 -0
  99. package/session_pool/consts.d.ts +1 -2
  100. package/session_pool/consts.js +1 -2
  101. package/session_pool/errors.d.ts +0 -1
  102. package/session_pool/errors.js +0 -1
  103. package/session_pool/fingerprint.d.ts +9 -0
  104. package/session_pool/fingerprint.js +30 -0
  105. package/session_pool/index.d.ts +0 -2
  106. package/session_pool/index.js +0 -2
  107. package/session_pool/session.d.ts +35 -88
  108. package/session_pool/session.js +101 -159
  109. package/session_pool/session_pool.d.ts +73 -91
  110. package/session_pool/session_pool.js +172 -165
  111. package/storages/batched_adds.d.ts +37 -0
  112. package/storages/batched_adds.js +73 -0
  113. package/storages/dataset.d.ts +114 -54
  114. package/storages/dataset.js +291 -148
  115. package/storages/index.d.ts +10 -8
  116. package/storages/index.js +8 -8
  117. package/storages/key_value_store.d.ts +185 -42
  118. package/storages/key_value_store.js +445 -169
  119. package/storages/key_value_store_codec.d.ts +32 -0
  120. package/storages/key_value_store_codec.js +113 -0
  121. package/storages/request_dedup_cache.d.ts +22 -0
  122. package/storages/request_dedup_cache.js +48 -0
  123. package/storages/request_list.d.ts +52 -115
  124. package/storages/request_list.js +183 -152
  125. package/storages/request_loader.d.ts +101 -0
  126. package/storages/request_loader.js +1 -0
  127. package/storages/request_manager.d.ts +33 -0
  128. package/storages/request_manager.js +1 -0
  129. package/storages/request_manager_tandem.d.ts +97 -0
  130. package/storages/request_manager_tandem.js +197 -0
  131. package/storages/request_queue.d.ts +290 -47
  132. package/storages/request_queue.js +762 -216
  133. package/storages/{sitemap_request_list.d.ts → sitemap_request_loader.d.ts} +45 -89
  134. package/storages/sitemap_request_loader.js +438 -0
  135. package/storages/storage_instance_manager.d.ts +87 -0
  136. package/storages/storage_instance_manager.js +256 -0
  137. package/storages/storage_stats.d.ts +48 -0
  138. package/storages/storage_stats.js +29 -0
  139. package/storages/throttling_request_manager.d.ts +239 -0
  140. package/storages/throttling_request_manager.js +646 -0
  141. package/storages/transaction.d.ts +252 -0
  142. package/storages/transaction.js +251 -0
  143. package/storages/utils.d.ts +59 -11
  144. package/storages/utils.js +75 -15
  145. package/system-info/cpu-info.d.ts +67 -0
  146. package/system-info/cpu-info.js +216 -0
  147. package/system-info/memory-info.d.ts +31 -0
  148. package/system-info/memory-info.js +115 -0
  149. package/system-info/ps-tree.d.ts +17 -0
  150. package/system-info/ps-tree.js +144 -0
  151. package/system-info/runtime.d.ts +14 -0
  152. package/system-info/runtime.js +80 -0
  153. package/typedefs.d.ts +0 -6
  154. package/typedefs.js +0 -1
  155. package/url.d.ts +9 -0
  156. package/url.js +11 -0
  157. package/validators.d.ts +22 -18
  158. package/validators.js +13 -18
  159. package/autoscaling/autoscaled_pool.d.ts.map +0 -1
  160. package/autoscaling/autoscaled_pool.js.map +0 -1
  161. package/autoscaling/index.d.ts.map +0 -1
  162. package/autoscaling/index.js.map +0 -1
  163. package/autoscaling/snapshotter.d.ts.map +0 -1
  164. package/autoscaling/snapshotter.js.map +0 -1
  165. package/autoscaling/system_status.d.ts.map +0 -1
  166. package/autoscaling/system_status.js.map +0 -1
  167. package/configuration.d.ts.map +0 -1
  168. package/configuration.js.map +0 -1
  169. package/cookie_utils.d.ts.map +0 -1
  170. package/cookie_utils.js.map +0 -1
  171. package/crawlers/context_pipeline.d.ts.map +0 -1
  172. package/crawlers/context_pipeline.js.map +0 -1
  173. package/crawlers/crawler_commons.d.ts.map +0 -1
  174. package/crawlers/crawler_commons.js.map +0 -1
  175. package/crawlers/crawler_utils.d.ts +0 -10
  176. package/crawlers/crawler_utils.d.ts.map +0 -1
  177. package/crawlers/crawler_utils.js +0 -12
  178. package/crawlers/crawler_utils.js.map +0 -1
  179. package/crawlers/error_snapshotter.d.ts.map +0 -1
  180. package/crawlers/error_snapshotter.js.map +0 -1
  181. package/crawlers/error_tracker.d.ts.map +0 -1
  182. package/crawlers/error_tracker.js.map +0 -1
  183. package/crawlers/index.d.ts.map +0 -1
  184. package/crawlers/index.js.map +0 -1
  185. package/crawlers/internals/types.d.ts.map +0 -1
  186. package/crawlers/internals/types.js.map +0 -1
  187. package/crawlers/statistics.d.ts.map +0 -1
  188. package/crawlers/statistics.js.map +0 -1
  189. package/enqueue_links/enqueue_links.d.ts.map +0 -1
  190. package/enqueue_links/enqueue_links.js.map +0 -1
  191. package/enqueue_links/index.d.ts.map +0 -1
  192. package/enqueue_links/index.js.map +0 -1
  193. package/enqueue_links/shared.d.ts.map +0 -1
  194. package/enqueue_links/shared.js.map +0 -1
  195. package/errors.d.ts.map +0 -1
  196. package/errors.js.map +0 -1
  197. package/events/event_manager.d.ts.map +0 -1
  198. package/events/event_manager.js.map +0 -1
  199. package/events/index.d.ts.map +0 -1
  200. package/events/index.js.map +0 -1
  201. package/events/local_event_manager.d.ts.map +0 -1
  202. package/events/local_event_manager.js.map +0 -1
  203. package/http_clients/base-http-client.d.ts +0 -140
  204. package/http_clients/base-http-client.d.ts.map +0 -1
  205. package/http_clients/base-http-client.js +0 -40
  206. package/http_clients/base-http-client.js.map +0 -1
  207. package/http_clients/form-data-like.d.ts +0 -67
  208. package/http_clients/form-data-like.d.ts.map +0 -1
  209. package/http_clients/form-data-like.js +0 -5
  210. package/http_clients/form-data-like.js.map +0 -1
  211. package/http_clients/got-scraping-http-client.d.ts +0 -20
  212. package/http_clients/got-scraping-http-client.d.ts.map +0 -1
  213. package/http_clients/got-scraping-http-client.js +0 -82
  214. package/http_clients/got-scraping-http-client.js.map +0 -1
  215. package/http_clients/index.d.ts +0 -3
  216. package/http_clients/index.d.ts.map +0 -1
  217. package/http_clients/index.js +0 -3
  218. package/http_clients/index.js.map +0 -1
  219. package/index.d.ts.map +0 -1
  220. package/index.js.map +0 -1
  221. package/log.d.ts.map +0 -1
  222. package/log.js.map +0 -1
  223. package/proxy_configuration.d.ts.map +0 -1
  224. package/proxy_configuration.js.map +0 -1
  225. package/request.d.ts.map +0 -1
  226. package/request.js.map +0 -1
  227. package/router.d.ts.map +0 -1
  228. package/router.js.map +0 -1
  229. package/serialization.d.ts.map +0 -1
  230. package/serialization.js.map +0 -1
  231. package/session_pool/consts.d.ts.map +0 -1
  232. package/session_pool/consts.js.map +0 -1
  233. package/session_pool/errors.d.ts.map +0 -1
  234. package/session_pool/errors.js.map +0 -1
  235. package/session_pool/events.d.ts +0 -3
  236. package/session_pool/events.d.ts.map +0 -1
  237. package/session_pool/events.js +0 -3
  238. package/session_pool/events.js.map +0 -1
  239. package/session_pool/index.d.ts.map +0 -1
  240. package/session_pool/index.js.map +0 -1
  241. package/session_pool/session.d.ts.map +0 -1
  242. package/session_pool/session.js.map +0 -1
  243. package/session_pool/session_pool.d.ts.map +0 -1
  244. package/session_pool/session_pool.js.map +0 -1
  245. package/storages/access_checking.d.ts +0 -13
  246. package/storages/access_checking.d.ts.map +0 -1
  247. package/storages/access_checking.js +0 -14
  248. package/storages/access_checking.js.map +0 -1
  249. package/storages/dataset.d.ts.map +0 -1
  250. package/storages/dataset.js.map +0 -1
  251. package/storages/index.d.ts.map +0 -1
  252. package/storages/index.js.map +0 -1
  253. package/storages/key_value_store.d.ts.map +0 -1
  254. package/storages/key_value_store.js.map +0 -1
  255. package/storages/request_list.d.ts.map +0 -1
  256. package/storages/request_list.js.map +0 -1
  257. package/storages/request_provider.d.ts +0 -308
  258. package/storages/request_provider.d.ts.map +0 -1
  259. package/storages/request_provider.js +0 -555
  260. package/storages/request_provider.js.map +0 -1
  261. package/storages/request_queue.d.ts.map +0 -1
  262. package/storages/request_queue.js.map +0 -1
  263. package/storages/request_queue_v2.d.ts +0 -87
  264. package/storages/request_queue_v2.d.ts.map +0 -1
  265. package/storages/request_queue_v2.js +0 -438
  266. package/storages/request_queue_v2.js.map +0 -1
  267. package/storages/sitemap_request_list.d.ts.map +0 -1
  268. package/storages/sitemap_request_list.js +0 -430
  269. package/storages/sitemap_request_list.js.map +0 -1
  270. package/storages/storage_manager.d.ts +0 -58
  271. package/storages/storage_manager.d.ts.map +0 -1
  272. package/storages/storage_manager.js +0 -105
  273. package/storages/storage_manager.js.map +0 -1
  274. package/storages/utils.d.ts.map +0 -1
  275. package/storages/utils.js.map +0 -1
  276. package/tsconfig.build.tsbuildinfo +0 -1
  277. package/typedefs.d.ts.map +0 -1
  278. package/typedefs.js.map +0 -1
  279. package/validators.d.ts.map +0 -1
  280. package/validators.js.map +0 -1
@@ -1,54 +1,12 @@
1
- import ow from 'ow';
2
- import log from '@apify/log';
3
- /**
4
- * Internal class for tracking the proxy tier history for a specific domain.
5
- *
6
- * Predicts the best proxy tier for the next request based on the error history for different proxy tiers.
7
- */
8
- class ProxyTierTracker {
9
- histogram;
10
- currentTier;
11
- constructor(tieredProxyUrls) {
12
- this.histogram = tieredProxyUrls.map(() => 0);
13
- this.currentTier = 0;
14
- }
15
- /**
16
- * Processes a single step of the algorithm and updates the current tier prediction based on the error history.
17
- */
18
- processStep() {
19
- this.histogram.forEach((x, i) => {
20
- if (this.currentTier === i)
21
- return;
22
- if (x > 0)
23
- this.histogram[i]--;
24
- });
25
- const left = this.currentTier > 0 ? this.histogram[this.currentTier - 1] : Infinity;
26
- const right = this.currentTier < this.histogram.length - 1 ? this.histogram[this.currentTier + 1] : Infinity;
27
- if (this.histogram[this.currentTier] > Math.min(left, right)) {
28
- this.currentTier = left <= right ? this.currentTier - 1 : this.currentTier + 1;
29
- }
30
- else if (this.histogram[this.currentTier] === left) {
31
- this.currentTier--;
32
- }
33
- }
34
- /**
35
- * Increases the error score for the given proxy tier. This raises the chance of picking a different proxy tier for the subsequent requests.
36
- *
37
- * 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).
38
- * @param tier The proxy tier to mark as problematic.
39
- */
40
- addError(tier) {
41
- this.histogram[tier] += 10;
42
- }
43
- /**
44
- * Returns the best proxy tier for the next request based on the error history for different proxy tiers.
45
- * @returns The proxy tier prediction
46
- */
47
- predictTier() {
48
- this.processStep();
49
- return this.currentTier;
50
- }
51
- }
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
+ });
52
10
  /**
53
11
  * Configures connection to a proxy server with the provided options. Proxy servers are used to prevent target websites from blocking
54
12
  * your crawlers based on IP address rate limits or blacklists. Setting proxy configuration in your crawlers automatically configures
@@ -79,13 +37,9 @@ class ProxyTierTracker {
79
37
  */
80
38
  export class ProxyConfiguration {
81
39
  isManInTheMiddle = false;
82
- nextCustomUrlIndex = 0;
83
- proxyUrls;
84
- tieredProxyUrls;
85
- usedProxyUrls = new Map();
86
- newUrlFunction;
87
- log = log.child({ prefix: 'ProxyConfiguration' });
88
- domainTiers = new Map();
40
+ #nextCustomUrlIndex = 0;
41
+ #proxyUrls;
42
+ #newUrlFunction;
89
43
  /**
90
44
  * Creates a {@link ProxyConfiguration} instance based on the provided options. Proxy servers are used to prevent target websites from
91
45
  * blocking your crawlers based on IP address rate limits or blacklists. Setting proxy configuration in your crawlers automatically configures
@@ -108,19 +62,17 @@ export class ProxyConfiguration {
108
62
  */
109
63
  constructor(options = {}) {
110
64
  const { validateRequired, ...rest } = options;
111
- ow(rest, ow.object.exactShape({
112
- proxyUrls: ow.optional.array.nonEmpty.ofType(ow.string.url),
113
- newUrlFunction: ow.optional.function,
114
- tieredProxyUrls: ow.optional.array.nonEmpty.ofType(ow.array.nonEmpty.ofType(ow.any(ow.string.url, ow.null))),
115
- }));
116
- const { proxyUrls, newUrlFunction, tieredProxyUrls } = options;
117
- if ([proxyUrls, newUrlFunction, tieredProxyUrls].filter((x) => x).length > 1)
118
- this._throwCannotCombineCustomMethods();
65
+ if ('tieredProxyUrls' in rest) {
66
+ throw new Error('The `tieredProxyUrls` option has been removed in Crawlee v4. ' +
67
+ 'See the v4 upgrading guide for the recommended migration to named sessions.');
68
+ }
69
+ const { proxyUrls, newUrlFunction } = parseArgument(rest, proxyConfigurationOptionsSchema);
70
+ if (proxyUrls && newUrlFunction)
71
+ this.throwCannotCombineCustomMethods();
119
72
  if (!proxyUrls && !newUrlFunction && validateRequired)
120
- this._throwNoOptionsProvided();
121
- this.proxyUrls = proxyUrls;
122
- this.newUrlFunction = newUrlFunction;
123
- this.tieredProxyUrls = tieredProxyUrls;
73
+ this.throwNoOptionsProvided();
74
+ this.#proxyUrls = proxyUrls;
75
+ this.#newUrlFunction = newUrlFunction;
124
76
  }
125
77
  /**
126
78
  * This function creates a new {@link ProxyInfo} info object.
@@ -132,16 +84,7 @@ export class ProxyConfiguration {
132
84
  * @return Represents information about used proxy and its configuration.
133
85
  */
134
86
  async newProxyInfo(options) {
135
- let url;
136
- let tier;
137
- if (this.tieredProxyUrls) {
138
- const { proxyUrl, proxyTier } = this._handleTieredUrl(options);
139
- url = proxyUrl ?? undefined;
140
- tier = proxyTier;
141
- }
142
- else {
143
- url = await this.newUrl(options);
144
- }
87
+ const url = await this.newUrl(options);
145
88
  if (!url)
146
89
  return undefined;
147
90
  const { username, password, port, hostname } = new URL(url);
@@ -151,59 +94,8 @@ export class ProxyConfiguration {
151
94
  password: decodeURIComponent(password),
152
95
  hostname,
153
96
  port: port,
154
- proxyTier: tier,
155
97
  };
156
98
  }
157
- /**
158
- * Given a request / proxy tier, this function returns a new proxy URL based on the provided configuration options.
159
- * @param options Options for the tiered proxy rotation
160
- * @returns An object with the proxy URL and the proxy tier used.
161
- */
162
- _handleTieredUrl(options) {
163
- if (!this.tieredProxyUrls)
164
- throw new Error('Tiered proxy URLs are not set');
165
- if (!options || (!options?.request && options?.proxyTier === undefined)) {
166
- const allProxyUrls = this.tieredProxyUrls.flat();
167
- return {
168
- proxyUrl: allProxyUrls[this.nextCustomUrlIndex++ % allProxyUrls.length],
169
- };
170
- }
171
- let tierPrediction = options.proxyTier;
172
- if (typeof tierPrediction !== 'number') {
173
- tierPrediction = this.predictProxyTier(options.request);
174
- }
175
- const proxyTier = this.tieredProxyUrls[tierPrediction];
176
- return {
177
- proxyUrl: proxyTier[this.nextCustomUrlIndex++ % proxyTier.length],
178
- proxyTier: tierPrediction,
179
- };
180
- }
181
- /**
182
- * Given a `Request` object, this function returns the tier of the proxy that should be used for the request.
183
- *
184
- * This returns `null` if `tieredProxyUrls` option is not set.
185
- */
186
- predictProxyTier(request) {
187
- if (!this.tieredProxyUrls)
188
- return null;
189
- const domain = new URL(request.url).hostname;
190
- if (!this.domainTiers.has(domain)) {
191
- this.domainTiers.set(domain, new ProxyTierTracker(this.tieredProxyUrls));
192
- }
193
- request.userData.__crawlee ??= {};
194
- const tracker = this.domainTiers.get(domain);
195
- if (typeof request.userData.__crawlee.lastProxyTier === 'number') {
196
- tracker.addError(request.userData.__crawlee.lastProxyTier);
197
- }
198
- const tierPrediction = tracker.predictTier();
199
- if (typeof request.userData.__crawlee.lastProxyTier === 'number' &&
200
- request.userData.__crawlee.lastProxyTier !== tierPrediction) {
201
- log.debug(`Changing proxy tier for domain "${domain}" from ${request.userData.__crawlee.lastProxyTier} to ${tierPrediction}.`);
202
- }
203
- request.userData.__crawlee.lastProxyTier = tierPrediction;
204
- request.userData.__crawlee.forefront = true;
205
- return tierPrediction;
206
- }
207
99
  /**
208
100
  * Returns a new proxy URL based on provided configuration options.
209
101
  *
@@ -211,25 +103,19 @@ export class ProxyConfiguration {
211
103
  * For example, `http://bob:password123@proxy.example.com:8000`
212
104
  */
213
105
  async newUrl(options) {
214
- if (this.newUrlFunction) {
215
- return (await this._callNewUrlFunction({ request: options?.request })) ?? undefined;
216
- }
217
- if (this.tieredProxyUrls) {
218
- return this._handleTieredUrl(options).proxyUrl ?? undefined;
106
+ if (this.#newUrlFunction) {
107
+ return (await this.callNewUrlFunction({ request: options?.request })) ?? undefined;
219
108
  }
220
- return this._handleProxyUrlsList();
109
+ return this.handleProxyUrlsList() ?? undefined;
221
110
  }
222
- /**
223
- * Handles custom url rotation
224
- */
225
- _handleProxyUrlsList() {
226
- return this.proxyUrls[this.nextCustomUrlIndex++ % this.proxyUrls.length];
111
+ handleProxyUrlsList() {
112
+ return this.#proxyUrls[this.#nextCustomUrlIndex++ % this.#proxyUrls.length];
227
113
  }
228
114
  /**
229
115
  * Calls the custom newUrlFunction and checks format of its return value
230
116
  */
231
- async _callNewUrlFunction(options) {
232
- const proxyUrl = await this.newUrlFunction(options);
117
+ async callNewUrlFunction(options) {
118
+ const proxyUrl = await this.#newUrlFunction(options);
233
119
  try {
234
120
  if (proxyUrl) {
235
121
  new URL(proxyUrl); // eslint-disable-line no-new
@@ -240,11 +126,10 @@ export class ProxyConfiguration {
240
126
  throw new Error(`The provided newUrlFunction did not return a valid URL.\nCause: ${err.message}`);
241
127
  }
242
128
  }
243
- _throwCannotCombineCustomMethods() {
129
+ throwCannotCombineCustomMethods() {
244
130
  throw new Error('Cannot combine custom proxies "options.proxyUrls" with custom generating function "options.newUrlFunction".');
245
131
  }
246
- _throwNoOptionsProvided() {
132
+ throwNoOptionsProvided() {
247
133
  throw new Error('One of "options.proxyUrls" or "options.newUrlFunction" needs to be provided.');
248
134
  }
249
135
  }
250
- //# sourceMappingURL=proxy_configuration.js.map
@@ -0,0 +1,155 @@
1
+ import type { Configuration, CrawleeLogger } from '@crawlee/core';
2
+ import { KeyValueStore } from '@crawlee/core';
3
+ import type { Awaitable } from '@crawlee/types';
4
+ import type { StandardSchemaV1 } from '@standard-schema/spec';
5
+ /**
6
+ * One direction of the conversion between the state model and its persisted form - either a plain function, or a
7
+ * [Standard Schema](https://standardschema.dev) whose validated output is the result.
8
+ *
9
+ * A schema that fails to validate makes {@link RecoverableState} throw a {@link StateValidationError}. Zod
10
+ * codecs work directly, as their validation *is* the decode direction; use `(state) => codec.encode(state)` for the
11
+ * other one.
12
+ */
13
+ export type StateConversion<TFrom, TTo> = ((value: TFrom) => Awaitable<TTo>) | StandardSchemaV1<TFrom, TTo>;
14
+ /**
15
+ * A {@link StateConversion} for a caller that cannot await one - {@link Statistics}, whose `toJSON()` is
16
+ * synchronous, being the reason this exists.
17
+ *
18
+ * Only the function arm can be narrowed here: a Standard Schema is free to validate asynchronously, so a schema
19
+ * that does is rejected when it runs rather than when it is passed.
20
+ */
21
+ export type SyncStateConversion<TFrom, TTo> = ((value: TFrom) => TTo) | StandardSchemaV1<TFrom, TTo>;
22
+ /**
23
+ * Applies a {@link SyncStateConversion}, throwing a {@link StateValidationError} for a schema that rejects
24
+ * the value.
25
+ *
26
+ * @internal
27
+ */
28
+ export declare function convertStateSync<TFrom, TTo>(conversion: SyncStateConversion<TFrom, TTo>, value: TFrom, persistStateKey: string): TTo;
29
+ export interface RecoverableStatePersistenceOptions {
30
+ /**
31
+ * The key under which the state is stored in the KeyValueStore
32
+ */
33
+ persistStateKey: string;
34
+ /**
35
+ * Flag to enable or disable state persistence
36
+ */
37
+ persistenceEnabled?: boolean;
38
+ /**
39
+ * The KeyValueStore to persist into, defaulting to the default store. Accepts a pending
40
+ * {@link KeyValueStore.open} so that callers do not have to be async to point at a specific store.
41
+ */
42
+ keyValueStore?: KeyValueStore | PromiseLike<KeyValueStore>;
43
+ /**
44
+ * Time limit for a single load or save of the state, in milliseconds.
45
+ * @default 60_000
46
+ */
47
+ persistenceTimeoutMillis?: number;
48
+ }
49
+ /**
50
+ * Options for configuring the RecoverableState
51
+ */
52
+ export interface RecoverableStateOptions<TStateModel = Record<string, unknown>, TPersistedState = TStateModel> extends RecoverableStatePersistenceOptions {
53
+ /**
54
+ * The state used when no persisted state is found, and the state {@link RecoverableState.reset} restores.
55
+ *
56
+ * A plain value is deep-copied with `structuredClone` each time it is used, so pass a factory for a state
57
+ * that `structuredClone` cannot rebuild - one holding class instances, say, or one derived from a schema.
58
+ */
59
+ defaultState: TStateModel | (() => TStateModel);
60
+ /**
61
+ * A logger instance for logging operations related to state persistence
62
+ */
63
+ logger?: CrawleeLogger;
64
+ /**
65
+ * Configuration instance to use when opening the KeyValueStore
66
+ */
67
+ configuration?: Configuration;
68
+ /**
69
+ * Optional conversion of the state to a plain JSON-serializable value before it is persisted.
70
+ * If not provided, the state is persisted as is.
71
+ */
72
+ serialize?: StateConversion<TStateModel, TPersistedState>;
73
+ /**
74
+ * Optional conversion of a persisted value back to the state model, and the place to validate a record before
75
+ * trusting it. If not provided, the persisted value is used as is.
76
+ */
77
+ deserialize?: StateConversion<TPersistedState, TStateModel>;
78
+ }
79
+ /**
80
+ * A class for managing persistent recoverable state using a plain JavaScript object.
81
+ *
82
+ * This class facilitates state persistence to a `KeyValueStore`, allowing data to be saved and retrieved
83
+ * across migrations or restarts. It manages the loading, saving, and resetting of state data,
84
+ * with optional persistence capabilities.
85
+ *
86
+ * The state is represented by a plain JavaScript object that can be serialized to and deserialized from JSON.
87
+ * The class automatically hooks into the event system to persist state when needed.
88
+ */
89
+ export declare class RecoverableState<TStateModel = Record<string, unknown>, TPersistedState = TStateModel> {
90
+ #private;
91
+ /**
92
+ * Initialize a new recoverable state object.
93
+ *
94
+ * @param options Configuration options for the recoverable state
95
+ */
96
+ constructor(options: RecoverableStateOptions<TStateModel, TPersistedState>);
97
+ /**
98
+ * Initialize the recoverable state.
99
+ *
100
+ * If persistence is enabled, this method loads the saved state and registers the object to listen for
101
+ * PERSIST_STATE events. A state established beforehand by {@link RecoverableState.reset} survives if there
102
+ * is no record to restore.
103
+ *
104
+ * Calling this again after a {@link RecoverableState.teardown} starts a new persistence window - the
105
+ * listener is registered again and the record reloaded.
106
+ *
107
+ * @returns The loaded state object
108
+ */
109
+ initialize(): Promise<TStateModel>;
110
+ /**
111
+ * Clean up resources used by the recoverable state.
112
+ *
113
+ * If persistence is enabled, this method deregisters the object from PERSIST_STATE events
114
+ * and persists the current state one last time, warning rather than throwing if that write fails - cleanup
115
+ * runs when the work is already done, and failing it would bury whatever the caller was doing. The in-memory
116
+ * state is left alone, and {@link RecoverableState.initialize} can be called again to open a new
117
+ * persistence window.
118
+ */
119
+ teardown(): Promise<void>;
120
+ /**
121
+ * Get the current state.
122
+ *
123
+ * Throws until the state has been established, by either {@link RecoverableState.initialize} or the
124
+ * synchronous {@link RecoverableState.reset} - the latter being how a caller that cannot await in its
125
+ * constructor gets a usable state right away.
126
+ */
127
+ get currentValue(): TStateModel;
128
+ /**
129
+ * Reset the in-memory state to the default values, leaving any persisted record alone.
130
+ *
131
+ * Use {@link RecoverableState.resetStore} to clear the persisted record as well.
132
+ */
133
+ reset(): void;
134
+ /**
135
+ * Clear the persisted state record, leaving the in-memory state alone.
136
+ *
137
+ * This is a between-lifecycles operation - its point is to stop the next {@link RecoverableState.initialize}
138
+ * from restoring the record, so it throws while PERSIST_STATE events are still being handled, where the next
139
+ * one would write the record straight back. Use {@link RecoverableState.reset} to reset the state itself,
140
+ * or {@link RecoverableState.teardown} before clearing the record.
141
+ *
142
+ * A no-op if persistence is disabled or no KeyValueStore is available yet.
143
+ */
144
+ resetStore(): Promise<void>;
145
+ /**
146
+ * Persist the current state to the KeyValueStore.
147
+ *
148
+ * This method is typically called in response to a PERSIST_STATE event, but can also be called
149
+ * directly when needed. It is a no-op if persistence is disabled, if no KeyValueStore is available yet, or if
150
+ * there is no state to write. A failed write only rejects here - the periodic and teardown ones warn instead.
151
+ *
152
+ * @param eventData Optional data associated with a PERSIST_STATE event
153
+ */
154
+ persistState(eventData?: Record<string, unknown>): Promise<void>;
155
+ }
@@ -0,0 +1,231 @@
1
+ import { addTimeoutToPromise, storage as timeoutStorage } from '@apify/timeout';
2
+ import { EventType, KeyValueStore, serviceLocator, StateValidationError } from '@crawlee/core';
3
+ const DEFAULT_PERSISTENCE_TIMEOUT_MILLIS = 60_000;
4
+ /**
5
+ * Applies a {@link SyncStateConversion}, throwing a {@link StateValidationError} for a schema that rejects
6
+ * the value.
7
+ *
8
+ * @internal
9
+ */
10
+ export function convertStateSync(conversion, value, persistStateKey) {
11
+ if (typeof conversion === 'function') {
12
+ return conversion(value);
13
+ }
14
+ const result = conversion['~standard'].validate(value);
15
+ if ('then' in result) {
16
+ throw new Error(`The state conversion for '${persistStateKey}' validated asynchronously, which this caller cannot await.`);
17
+ }
18
+ if (result.issues) {
19
+ throw new StateValidationError(persistStateKey, result.issues);
20
+ }
21
+ return result.value;
22
+ }
23
+ /**
24
+ * A class for managing persistent recoverable state using a plain JavaScript object.
25
+ *
26
+ * This class facilitates state persistence to a `KeyValueStore`, allowing data to be saved and retrieved
27
+ * across migrations or restarts. It manages the loading, saving, and resetting of state data,
28
+ * with optional persistence capabilities.
29
+ *
30
+ * The state is represented by a plain JavaScript object that can be serialized to and deserialized from JSON.
31
+ * The class automatically hooks into the event system to persist state when needed.
32
+ */
33
+ export class RecoverableState {
34
+ #defaultState;
35
+ #state = null;
36
+ #initialized = false;
37
+ #listening = false;
38
+ #persistenceEnabled;
39
+ #persistStateKey;
40
+ #persistenceTimeoutMillis;
41
+ #configuration;
42
+ #keyValueStore;
43
+ #log;
44
+ #serialize;
45
+ #deserialize;
46
+ #persistStateQuietly;
47
+ /**
48
+ * Initialize a new recoverable state object.
49
+ *
50
+ * @param options Configuration options for the recoverable state
51
+ */
52
+ constructor(options) {
53
+ const { defaultState } = options;
54
+ this.#defaultState =
55
+ typeof defaultState === 'function'
56
+ ? defaultState
57
+ : () => structuredClone(defaultState);
58
+ this.#persistStateKey = options.persistStateKey;
59
+ this.#persistenceEnabled = options.persistenceEnabled ?? false;
60
+ this.#persistenceTimeoutMillis = options.persistenceTimeoutMillis ?? DEFAULT_PERSISTENCE_TIMEOUT_MILLIS;
61
+ this.#configuration = options.configuration;
62
+ this.#keyValueStore = options.keyValueStore ?? null;
63
+ this.#log = options.logger ?? serviceLocator.getLogger().child({ prefix: 'RecoverableState' });
64
+ this.#serialize = this.#toConversion(options.serialize);
65
+ this.#deserialize = this.#toConversion(options.deserialize);
66
+ // The automatic persists, where a rejection has nowhere useful to go - the event manager does not catch
67
+ // listener errors, and throwing from teardown would bury the outcome of the work it cleans up after.
68
+ this.#persistStateQuietly = async (eventData) => this.persistState(eventData).catch((error) => this.#log.warning(`Failed to persist the state under key '${this.#persistStateKey}'.`, { error }));
69
+ }
70
+ /** Normalizes a conversion option into a function. Absent conversions pass the value through unchanged. */
71
+ #toConversion(conversion) {
72
+ if (conversion === undefined) {
73
+ return async (value) => value;
74
+ }
75
+ if (typeof conversion === 'function') {
76
+ return async (value) => conversion(value);
77
+ }
78
+ return async (value) => {
79
+ const result = await conversion['~standard'].validate(value);
80
+ if (result.issues) {
81
+ throw new StateValidationError(this.#persistStateKey, result.issues);
82
+ }
83
+ return result.value;
84
+ };
85
+ }
86
+ /**
87
+ * Initialize the recoverable state.
88
+ *
89
+ * If persistence is enabled, this method loads the saved state and registers the object to listen for
90
+ * PERSIST_STATE events. A state established beforehand by {@link RecoverableState.reset} survives if there
91
+ * is no record to restore.
92
+ *
93
+ * Calling this again after a {@link RecoverableState.teardown} starts a new persistence window - the
94
+ * listener is registered again and the record reloaded.
95
+ *
96
+ * @returns The loaded state object
97
+ */
98
+ async initialize() {
99
+ if (this.#initialized) {
100
+ return this.currentValue;
101
+ }
102
+ if (this.#persistenceEnabled) {
103
+ this.#keyValueStore ??= KeyValueStore.open(null, {
104
+ configuration: this.#configuration ?? serviceLocator.getConfiguration(),
105
+ });
106
+ await this.#resolveKeyValueStore();
107
+ serviceLocator.getEventManager().on(EventType.PERSIST_STATE, this.#persistStateQuietly);
108
+ this.#listening = true;
109
+ }
110
+ // Flipped before the record is loaded, so that a caller catching a `StateValidationError` is left with a
111
+ // fully wired object running on the default state rather than a half-initialized one.
112
+ this.#initialized = true;
113
+ this.#state ??= this.#defaultState();
114
+ await this.#loadSavedState();
115
+ return this.currentValue;
116
+ }
117
+ /**
118
+ * Clean up resources used by the recoverable state.
119
+ *
120
+ * If persistence is enabled, this method deregisters the object from PERSIST_STATE events
121
+ * and persists the current state one last time, warning rather than throwing if that write fails - cleanup
122
+ * runs when the work is already done, and failing it would bury whatever the caller was doing. The in-memory
123
+ * state is left alone, and {@link RecoverableState.initialize} can be called again to open a new
124
+ * persistence window.
125
+ */
126
+ async teardown() {
127
+ this.#initialized = false;
128
+ if (!this.#persistenceEnabled) {
129
+ return;
130
+ }
131
+ serviceLocator.getEventManager().off(EventType.PERSIST_STATE, this.#persistStateQuietly);
132
+ this.#listening = false;
133
+ await this.#persistStateQuietly();
134
+ }
135
+ /**
136
+ * Get the current state.
137
+ *
138
+ * Throws until the state has been established, by either {@link RecoverableState.initialize} or the
139
+ * synchronous {@link RecoverableState.reset} - the latter being how a caller that cannot await in its
140
+ * constructor gets a usable state right away.
141
+ */
142
+ get currentValue() {
143
+ if (this.#state === null) {
144
+ throw new Error('Recoverable state has not yet been loaded - call initialize() or reset() first');
145
+ }
146
+ return this.#state;
147
+ }
148
+ /**
149
+ * Reset the in-memory state to the default values, leaving any persisted record alone.
150
+ *
151
+ * Use {@link RecoverableState.resetStore} to clear the persisted record as well.
152
+ */
153
+ reset() {
154
+ this.#state = this.#defaultState();
155
+ }
156
+ /**
157
+ * Clear the persisted state record, leaving the in-memory state alone.
158
+ *
159
+ * This is a between-lifecycles operation - its point is to stop the next {@link RecoverableState.initialize}
160
+ * from restoring the record, so it throws while PERSIST_STATE events are still being handled, where the next
161
+ * one would write the record straight back. Use {@link RecoverableState.reset} to reset the state itself,
162
+ * or {@link RecoverableState.teardown} before clearing the record.
163
+ *
164
+ * A no-op if persistence is disabled or no KeyValueStore is available yet.
165
+ */
166
+ async resetStore() {
167
+ if (this.#listening) {
168
+ throw new Error(`Cannot clear the state persisted under key '${this.#persistStateKey}' while it is still being persisted periodically - the next PERSIST_STATE event would write it straight back. Use reset() to reset the state itself, or teardown() before clearing the record.`);
169
+ }
170
+ if (!this.#persistenceEnabled) {
171
+ return;
172
+ }
173
+ const keyValueStore = await this.#resolveKeyValueStore();
174
+ if (keyValueStore === null) {
175
+ return;
176
+ }
177
+ await this.#withTimeout(async () => keyValueStore.setValue(this.#persistStateKey, null), 'Clearing the persisted state');
178
+ }
179
+ /**
180
+ * Persist the current state to the KeyValueStore.
181
+ *
182
+ * This method is typically called in response to a PERSIST_STATE event, but can also be called
183
+ * directly when needed. It is a no-op if persistence is disabled, if no KeyValueStore is available yet, or if
184
+ * there is no state to write. A failed write only rejects here - the periodic and teardown ones warn instead.
185
+ *
186
+ * @param eventData Optional data associated with a PERSIST_STATE event
187
+ */
188
+ async persistState(eventData) {
189
+ if (!this.#persistenceEnabled || this.#state === null) {
190
+ return;
191
+ }
192
+ const keyValueStore = await this.#resolveKeyValueStore();
193
+ if (keyValueStore === null) {
194
+ return;
195
+ }
196
+ this.#log.debug(`Persisting state of the RecoverableState (eventData=${JSON.stringify(eventData)}).`);
197
+ const serializedState = await this.#serialize(this.currentValue);
198
+ await this.#withTimeout(async () => keyValueStore.setValue(this.#persistStateKey, serializedState), 'Persisting the state');
199
+ }
200
+ /** Awaits a store handed over as a pending `open()`, keeping the resolved instance for later calls. */
201
+ async #resolveKeyValueStore() {
202
+ if (this.#keyValueStore === null) {
203
+ return null;
204
+ }
205
+ this.#keyValueStore = await this.#keyValueStore;
206
+ return this.#keyValueStore;
207
+ }
208
+ /**
209
+ * Load the saved state from the KeyValueStore. Leaves the current state alone if there is no record to load.
210
+ */
211
+ async #loadSavedState() {
212
+ if (!this.#persistenceEnabled) {
213
+ return;
214
+ }
215
+ const keyValueStore = await this.#resolveKeyValueStore();
216
+ if (keyValueStore === null) {
217
+ return;
218
+ }
219
+ const storedState = await this.#withTimeout(async () => keyValueStore.getValue(this.#persistStateKey), 'Loading the persisted state');
220
+ if (storedState === null || storedState === undefined) {
221
+ return;
222
+ }
223
+ this.#state = await this.#deserialize(storedState);
224
+ }
225
+ async #withTimeout(operation, description) {
226
+ // `@apify/timeout` shares one `AbortController` across nested frames and `KeyValueStore` checks it on
227
+ // every operation, so a teardown-time persist running inside an already-expired request handler timeout
228
+ // would be aborted before it started. Hence a fresh timeout context.
229
+ return timeoutStorage.exit(async () => addTimeoutToPromise(operation, this.#persistenceTimeoutMillis, `${description} under key '${this.#persistStateKey}' timed out after ${this.#persistenceTimeoutMillis / 1000} seconds.`));
230
+ }
231
+ }