@crawlee/core 4.0.0-beta.26 → 4.0.0-beta.261
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 +15 -15
- package/byte-utils.d.ts +17 -0
- package/byte-utils.js +42 -0
- package/configuration.d.ts +77 -235
- package/configuration.js +159 -227
- package/debug.d.ts +36 -0
- package/debug.js +70 -0
- package/errors.d.ts +45 -18
- package/errors.js +58 -27
- package/events/event-manager.d.ts +78 -0
- package/events/{event_manager.js → event-manager.js} +12 -13
- package/events/index.d.ts +3 -3
- package/events/index.js +2 -3
- package/events/local-event-manager.d.ts +32 -0
- package/events/local-event-manager.js +101 -0
- package/events/system-info.d.ts +38 -0
- package/events/system-info.js +1 -0
- package/index.d.ts +9 -12
- package/index.js +10 -11
- 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 +77 -1
- 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 +35 -0
- package/memory-storage/resource-clients/dataset.js +108 -0
- package/memory-storage/resource-clients/key-value-store.d.ts +54 -0
- package/memory-storage/resource-clients/key-value-store.js +179 -0
- package/memory-storage/resource-clients/request-queue.d.ts +55 -0
- package/memory-storage/resource-clients/request-queue.js +413 -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 +23 -23
- package/proxy-configuration.d.ts +118 -0
- package/proxy-configuration.js +141 -0
- package/recoverable-state.d.ts +177 -0
- package/recoverable-state.js +268 -0
- package/request.d.ts +48 -93
- package/request.js +139 -232
- 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 +39 -0
- package/storages/batched-adds.js +78 -0
- package/storages/dataset.d.ts +105 -58
- package/storages/dataset.js +292 -150
- package/storages/index.d.ts +11 -12
- package/storages/index.js +7 -12
- package/storages/key-value-store-codec.d.ts +32 -0
- package/storages/key-value-store-codec.js +108 -0
- package/storages/{key_value_store.d.ts → key-value-store.d.ts} +169 -72
- package/storages/key-value-store.js +767 -0
- package/storages/request-dedup-cache.d.ts +20 -0
- package/storages/request-dedup-cache.js +48 -0
- package/storages/{request_list.d.ts → request-list.d.ts} +33 -173
- package/storages/{request_list.js → request-list.js} +276 -255
- package/storages/request-loader.d.ts +127 -0
- package/storages/request-loader.js +36 -0
- package/storages/request-manager-tandem.d.ts +82 -0
- package/storages/request-manager-tandem.js +200 -0
- package/storages/request-manager.d.ts +119 -0
- package/storages/request-manager.js +1 -0
- package/storages/{request_provider.d.ts → request-queue.d.ts} +128 -159
- package/storages/request-queue.js +871 -0
- 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 +270 -0
- package/storages/transaction.js +296 -0
- package/storages/utils.d.ts +59 -11
- package/storages/utils.js +76 -16
- 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 -285
- 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 -111
- package/cookie_utils.js.map +0 -1
- package/crawlers/context_pipeline.d.ts +0 -61
- package/crawlers/context_pipeline.d.ts.map +0 -1
- package/crawlers/context_pipeline.js +0 -99
- package/crawlers/context_pipeline.js.map +0 -1
- package/crawlers/crawler_commons.d.ts +0 -191
- package/crawlers/crawler_commons.d.ts.map +0 -1
- package/crawlers/crawler_commons.js +0 -108
- package/crawlers/crawler_commons.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 -60
- 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 -55
- 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/internals/types.d.ts +0 -8
- package/crawlers/internals/types.d.ts.map +0 -1
- package/crawlers/internals/types.js +0 -2
- package/crawlers/internals/types.js.map +0 -1
- package/crawlers/statistics.d.ts +0 -213
- package/crawlers/statistics.d.ts.map +0 -1
- package/crawlers/statistics.js +0 -349
- package/crawlers/statistics.js.map +0 -1
- package/enqueue_links/enqueue_links.d.ts +0 -249
- package/enqueue_links/enqueue_links.d.ts.map +0 -1
- package/enqueue_links/enqueue_links.js +0 -269
- 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 -73
- package/enqueue_links/shared.d.ts.map +0 -1
- package/enqueue_links/shared.js +0 -212
- 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 +0 -52
- 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 +0 -27
- package/events/local_event_manager.d.ts.map +0 -1
- package/events/local_event_manager.js +0 -75
- package/events/local_event_manager.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 +0 -169
- package/proxy_configuration.d.ts.map +0 -1
- package/proxy_configuration.js +0 -247
- package/proxy_configuration.js.map +0 -1
- package/recoverable_state.d.ts +0 -123
- package/recoverable_state.d.ts.map +0 -1
- package/recoverable_state.js +0 -140
- package/recoverable_state.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 -186
- package/session_pool/session.d.ts.map +0 -1
- package/session_pool/session.js +0 -278
- package/session_pool/session.js.map +0 -1
- package/session_pool/session_pool.d.ts +0 -258
- package/session_pool/session_pool.d.ts.map +0 -1
- package/session_pool/session_pool.js +0 -387
- 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 +0 -512
- 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_list_adapter.d.ts +0 -58
- package/storages/request_list_adapter.d.ts.map +0 -1
- package/storages/request_list_adapter.js +0 -81
- package/storages/request_list_adapter.js.map +0 -1
- package/storages/request_manager_tandem.d.ts +0 -68
- package/storages/request_manager_tandem.d.ts.map +0 -1
- package/storages/request_manager_tandem.js +0 -124
- package/storages/request_manager_tandem.js.map +0 -1
- package/storages/request_provider.d.ts.map +0 -1
- package/storages/request_provider.js +0 -591
- package/storages/request_provider.js.map +0 -1
- package/storages/request_queue.d.ts +0 -128
- package/storages/request_queue.d.ts.map +0 -1
- package/storages/request_queue.js +0 -320
- 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 -250
- package/storages/sitemap_request_list.d.ts.map +0 -1
- package/storages/sitemap_request_list.js +0 -431
- package/storages/sitemap_request_list.js.map +0 -1
- package/storages/storage_manager.d.ts +0 -62
- 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/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
|
@@ -1,269 +0,0 @@
|
|
|
1
|
-
import ow from 'ow';
|
|
2
|
-
import { getDomain } from 'tldts';
|
|
3
|
-
import log from '@apify/log';
|
|
4
|
-
import { constructGlobObjectsFromGlobs, constructRegExpObjectsFromPseudoUrls, constructRegExpObjectsFromRegExps, createRequestOptions, createRequests, filterRequestsByPatterns, } from './shared.js';
|
|
5
|
-
/**
|
|
6
|
-
* The different enqueueing strategies available.
|
|
7
|
-
*
|
|
8
|
-
* Depending on the strategy you select, we will only check certain parts of the URLs found. Here is a diagram of each URL part and their name:
|
|
9
|
-
*
|
|
10
|
-
* ```md
|
|
11
|
-
* Protocol Domain
|
|
12
|
-
* ┌────┐ ┌─────────┐
|
|
13
|
-
* https://example.crawlee.dev/...
|
|
14
|
-
* │ └─────────────────┤
|
|
15
|
-
* │ Hostname │
|
|
16
|
-
* │ │
|
|
17
|
-
* └─────────────────────────┘
|
|
18
|
-
* Origin
|
|
19
|
-
*```
|
|
20
|
-
*
|
|
21
|
-
* - The `Protocol` is usually `http` or `https`
|
|
22
|
-
* - The `Domain` represents the path without any possible subdomains to a website. For example, `crawlee.dev` is the domain of `https://example.crawlee.dev/`
|
|
23
|
-
* - The `Hostname` is the full path to a website, including any subdomains. For example, `example.crawlee.dev` is the hostname of `https://example.crawlee.dev/`
|
|
24
|
-
* - The `Origin` is the combination of the `Protocol` and `Hostname`. For example, `https://example.crawlee.dev` is the origin of `https://example.crawlee.dev/`
|
|
25
|
-
*/
|
|
26
|
-
export var EnqueueStrategy;
|
|
27
|
-
(function (EnqueueStrategy) {
|
|
28
|
-
/**
|
|
29
|
-
* Matches any URLs found
|
|
30
|
-
*/
|
|
31
|
-
EnqueueStrategy["All"] = "all";
|
|
32
|
-
/**
|
|
33
|
-
* Matches any URLs that have the same hostname.
|
|
34
|
-
* For example, `https://wow.example.com/hello` will be matched for a base url of `https://wow.example.com/`, but
|
|
35
|
-
* `https://example.com/hello` will not be matched.
|
|
36
|
-
*
|
|
37
|
-
* > This strategy will match both `http` and `https` protocols regardless of the base URL protocol.
|
|
38
|
-
*/
|
|
39
|
-
EnqueueStrategy["SameHostname"] = "same-hostname";
|
|
40
|
-
/**
|
|
41
|
-
* Matches any URLs that have the same domain as the base URL.
|
|
42
|
-
* For example, `https://wow.an.example.com` and `https://example.com` will both be matched for a base url of
|
|
43
|
-
* `https://example.com`.
|
|
44
|
-
*
|
|
45
|
-
* > This strategy will match both `http` and `https` protocols regardless of the base URL protocol.
|
|
46
|
-
*/
|
|
47
|
-
EnqueueStrategy["SameDomain"] = "same-domain";
|
|
48
|
-
/**
|
|
49
|
-
* Matches any URLs that have the same hostname and protocol.
|
|
50
|
-
* For example, `https://wow.example.com/hello` will be matched for a base url of `https://wow.example.com/`, but
|
|
51
|
-
* `http://wow.example.com/hello` will not be matched.
|
|
52
|
-
*
|
|
53
|
-
* > This strategy will ensure the protocol of the base URL is the same as the protocol of the URL to be enqueued.
|
|
54
|
-
*/
|
|
55
|
-
EnqueueStrategy["SameOrigin"] = "same-origin";
|
|
56
|
-
})(EnqueueStrategy || (EnqueueStrategy = {}));
|
|
57
|
-
/**
|
|
58
|
-
* This function enqueues the urls provided to the {@link RequestQueue} provided. If you want to automatically find and enqueue links,
|
|
59
|
-
* you should use the context-aware `enqueueLinks` function provided on the crawler contexts.
|
|
60
|
-
*
|
|
61
|
-
* Optionally, the function allows you to filter the target links' URLs using an array of globs or regular expressions
|
|
62
|
-
* and override settings of the enqueued {@link Request} objects.
|
|
63
|
-
*
|
|
64
|
-
* **Example usage**
|
|
65
|
-
*
|
|
66
|
-
* ```javascript
|
|
67
|
-
* await enqueueLinks({
|
|
68
|
-
* urls: aListOfFoundUrls,
|
|
69
|
-
* requestQueue,
|
|
70
|
-
* selector: 'a.product-detail',
|
|
71
|
-
* globs: [
|
|
72
|
-
* 'https://www.example.com/handbags/*',
|
|
73
|
-
* 'https://www.example.com/purses/*'
|
|
74
|
-
* ],
|
|
75
|
-
* });
|
|
76
|
-
* ```
|
|
77
|
-
*
|
|
78
|
-
* @param options All `enqueueLinks()` parameters are passed via an options object.
|
|
79
|
-
* @returns Promise that resolves to {@link BatchAddRequestsResult} object.
|
|
80
|
-
*/
|
|
81
|
-
export async function enqueueLinks(options) {
|
|
82
|
-
if (!options || Object.keys(options).length === 0) {
|
|
83
|
-
throw new RangeError([
|
|
84
|
-
'enqueueLinks() was called without the required options. You can only do that when you use the `crawlingContext.enqueueLinks()` method in request handlers.',
|
|
85
|
-
'Check out our guide on how to use enqueueLinks() here: https://crawlee.dev/js/docs/examples/crawl-relative-links',
|
|
86
|
-
].join('\n'));
|
|
87
|
-
}
|
|
88
|
-
ow(options, ow.object.exactShape({
|
|
89
|
-
urls: ow.array.ofType(ow.string),
|
|
90
|
-
requestQueue: ow.object.hasKeys('addRequestsBatched'),
|
|
91
|
-
robotsTxtFile: ow.optional.object.hasKeys('isAllowed'),
|
|
92
|
-
onSkippedRequest: ow.optional.function,
|
|
93
|
-
forefront: ow.optional.boolean,
|
|
94
|
-
skipNavigation: ow.optional.boolean,
|
|
95
|
-
limit: ow.optional.number,
|
|
96
|
-
selector: ow.optional.string,
|
|
97
|
-
baseUrl: ow.optional.string,
|
|
98
|
-
userData: ow.optional.object,
|
|
99
|
-
label: ow.optional.string,
|
|
100
|
-
pseudoUrls: ow.optional.array.ofType(ow.any(ow.string, ow.object.hasKeys('purl'))),
|
|
101
|
-
globs: ow.optional.array.ofType(ow.any(ow.string, ow.object.hasKeys('glob'))),
|
|
102
|
-
exclude: ow.optional.array.ofType(ow.any(ow.string, ow.regExp, ow.object.hasKeys('glob'), ow.object.hasKeys('regexp'))),
|
|
103
|
-
regexps: ow.optional.array.ofType(ow.any(ow.regExp, ow.object.hasKeys('regexp'))),
|
|
104
|
-
transformRequestFunction: ow.optional.function,
|
|
105
|
-
strategy: ow.optional.string.oneOf(Object.values(EnqueueStrategy)),
|
|
106
|
-
waitForAllRequestsToBeAdded: ow.optional.boolean,
|
|
107
|
-
}));
|
|
108
|
-
const { requestQueue, limit, urls, pseudoUrls, exclude, globs, regexps, transformRequestFunction, forefront, waitForAllRequestsToBeAdded, robotsTxtFile, onSkippedRequest, } = options;
|
|
109
|
-
const urlExcludePatternObjects = [];
|
|
110
|
-
const urlPatternObjects = [];
|
|
111
|
-
if (exclude?.length) {
|
|
112
|
-
for (const excl of exclude) {
|
|
113
|
-
if (typeof excl === 'string' || 'glob' in excl) {
|
|
114
|
-
urlExcludePatternObjects.push(...constructGlobObjectsFromGlobs([excl]));
|
|
115
|
-
}
|
|
116
|
-
else if (excl instanceof RegExp || 'regexp' in excl) {
|
|
117
|
-
urlExcludePatternObjects.push(...constructRegExpObjectsFromRegExps([excl]));
|
|
118
|
-
}
|
|
119
|
-
}
|
|
120
|
-
}
|
|
121
|
-
if (pseudoUrls?.length) {
|
|
122
|
-
log.deprecated('`pseudoUrls` option is deprecated, use `globs` or `regexps` instead');
|
|
123
|
-
urlPatternObjects.push(...constructRegExpObjectsFromPseudoUrls(pseudoUrls));
|
|
124
|
-
}
|
|
125
|
-
if (globs?.length) {
|
|
126
|
-
urlPatternObjects.push(...constructGlobObjectsFromGlobs(globs));
|
|
127
|
-
}
|
|
128
|
-
if (regexps?.length) {
|
|
129
|
-
urlPatternObjects.push(...constructRegExpObjectsFromRegExps(regexps));
|
|
130
|
-
}
|
|
131
|
-
if (!urlPatternObjects.length) {
|
|
132
|
-
options.strategy ??= EnqueueStrategy.SameHostname;
|
|
133
|
-
}
|
|
134
|
-
const enqueueStrategyPatterns = [];
|
|
135
|
-
if (options.baseUrl) {
|
|
136
|
-
const url = new URL(options.baseUrl);
|
|
137
|
-
switch (options.strategy) {
|
|
138
|
-
case EnqueueStrategy.SameHostname:
|
|
139
|
-
// We need to get the origin of the passed in domain in the event someone sets baseUrl
|
|
140
|
-
// to an url like https://example.com/deep/default/path and one of the found urls is an
|
|
141
|
-
// absolute relative path (/path/to/page)
|
|
142
|
-
enqueueStrategyPatterns.push({ glob: ignoreHttpSchema(`${url.origin}/**`) });
|
|
143
|
-
break;
|
|
144
|
-
case EnqueueStrategy.SameDomain: {
|
|
145
|
-
// Get the actual hostname from the base url
|
|
146
|
-
const baseUrlHostname = getDomain(url.hostname, { mixedInputs: false });
|
|
147
|
-
if (baseUrlHostname) {
|
|
148
|
-
// We have a hostname, so we can use it to match all links on the page that point to it and any subdomains of it
|
|
149
|
-
url.hostname = baseUrlHostname;
|
|
150
|
-
enqueueStrategyPatterns.push({ glob: ignoreHttpSchema(`${url.origin.replace(baseUrlHostname, `*.${baseUrlHostname}`)}/**`) }, { glob: ignoreHttpSchema(`${url.origin}/**`) });
|
|
151
|
-
}
|
|
152
|
-
else {
|
|
153
|
-
// We don't have a hostname (can happen for ips for instance), so reproduce the same behavior
|
|
154
|
-
// as SameDomainAndSubdomain
|
|
155
|
-
enqueueStrategyPatterns.push({ glob: ignoreHttpSchema(`${url.origin}/**`) });
|
|
156
|
-
}
|
|
157
|
-
break;
|
|
158
|
-
}
|
|
159
|
-
case EnqueueStrategy.SameOrigin: {
|
|
160
|
-
// The same behavior as SameHostname, but respecting the protocol of the URL
|
|
161
|
-
enqueueStrategyPatterns.push({ glob: `${url.origin}/**` });
|
|
162
|
-
break;
|
|
163
|
-
}
|
|
164
|
-
case EnqueueStrategy.All:
|
|
165
|
-
default:
|
|
166
|
-
enqueueStrategyPatterns.push({ glob: `http{s,}://**` });
|
|
167
|
-
break;
|
|
168
|
-
}
|
|
169
|
-
}
|
|
170
|
-
async function reportSkippedRequests(skippedRequests, reason) {
|
|
171
|
-
if (onSkippedRequest && skippedRequests.length > 0) {
|
|
172
|
-
await Promise.all(skippedRequests.map((request) => {
|
|
173
|
-
return onSkippedRequest({
|
|
174
|
-
url: request.url,
|
|
175
|
-
reason: request.skippedReason ?? reason,
|
|
176
|
-
});
|
|
177
|
-
}));
|
|
178
|
-
}
|
|
179
|
-
}
|
|
180
|
-
let requestOptions = createRequestOptions(urls, options);
|
|
181
|
-
if (robotsTxtFile) {
|
|
182
|
-
const skippedRequests = [];
|
|
183
|
-
requestOptions = requestOptions.filter((request) => {
|
|
184
|
-
if (robotsTxtFile.isAllowed(request.url)) {
|
|
185
|
-
return true;
|
|
186
|
-
}
|
|
187
|
-
skippedRequests.push(request);
|
|
188
|
-
return false;
|
|
189
|
-
});
|
|
190
|
-
await reportSkippedRequests(skippedRequests, 'robotsTxt');
|
|
191
|
-
}
|
|
192
|
-
if (transformRequestFunction) {
|
|
193
|
-
const skippedRequests = [];
|
|
194
|
-
requestOptions = requestOptions
|
|
195
|
-
.map((request) => {
|
|
196
|
-
const transformedRequest = transformRequestFunction(request);
|
|
197
|
-
if (!transformedRequest) {
|
|
198
|
-
skippedRequests.push(request);
|
|
199
|
-
}
|
|
200
|
-
return transformedRequest;
|
|
201
|
-
})
|
|
202
|
-
.filter((r) => Boolean(r));
|
|
203
|
-
await reportSkippedRequests(skippedRequests, 'filters');
|
|
204
|
-
}
|
|
205
|
-
async function createFilteredRequests() {
|
|
206
|
-
const skippedRequests = [];
|
|
207
|
-
// No user provided patterns means we can skip an extra filtering step
|
|
208
|
-
if (urlPatternObjects.length === 0) {
|
|
209
|
-
return createRequests(requestOptions, enqueueStrategyPatterns, urlExcludePatternObjects, options.strategy, (url) => skippedRequests.push(url));
|
|
210
|
-
}
|
|
211
|
-
// Generate requests based on the user patterns first
|
|
212
|
-
const generatedRequestsFromUserFilters = createRequests(requestOptions, urlPatternObjects, urlExcludePatternObjects, options.strategy, (url) => skippedRequests.push(url));
|
|
213
|
-
// ...then filter them by the enqueue links strategy (making this an AND check)
|
|
214
|
-
const filtered = filterRequestsByPatterns(generatedRequestsFromUserFilters, enqueueStrategyPatterns, (url) => skippedRequests.push(url));
|
|
215
|
-
await reportSkippedRequests(skippedRequests.map((url) => ({ url })), 'filters');
|
|
216
|
-
return filtered;
|
|
217
|
-
}
|
|
218
|
-
let requests = await createFilteredRequests();
|
|
219
|
-
if (typeof limit === 'number' && limit < requests.length) {
|
|
220
|
-
await reportSkippedRequests(requests.slice(limit), 'enqueueLimit');
|
|
221
|
-
requests = requests.slice(0, limit);
|
|
222
|
-
}
|
|
223
|
-
const { addedRequests } = await requestQueue.addRequestsBatched(requests, {
|
|
224
|
-
forefront,
|
|
225
|
-
waitForAllRequestsToBeAdded,
|
|
226
|
-
});
|
|
227
|
-
return { processedRequests: addedRequests, unprocessedRequests: [] };
|
|
228
|
-
}
|
|
229
|
-
/**
|
|
230
|
-
* @internal
|
|
231
|
-
* This method helps resolve the baseUrl that will be used for filtering in {@link enqueueLinks}.
|
|
232
|
-
* - If a user provides a base url, we always return it
|
|
233
|
-
* - If a user specifies {@link EnqueueStrategy.All} strategy, they do not care if the newly found urls are on the original
|
|
234
|
-
* request domain, or a redirected one
|
|
235
|
-
* - In all other cases, we return the domain of the original request as that's the one we need to use for filtering
|
|
236
|
-
*/
|
|
237
|
-
export function resolveBaseUrlForEnqueueLinksFiltering({ enqueueStrategy, finalRequestUrl, originalRequestUrl, userProvidedBaseUrl, }) {
|
|
238
|
-
// User provided base url takes priority
|
|
239
|
-
if (userProvidedBaseUrl) {
|
|
240
|
-
return userProvidedBaseUrl;
|
|
241
|
-
}
|
|
242
|
-
const originalUrlOrigin = new URL(originalRequestUrl).origin;
|
|
243
|
-
const finalUrlOrigin = new URL(finalRequestUrl ?? originalRequestUrl).origin;
|
|
244
|
-
// We can assume users want to go off the domain in this case
|
|
245
|
-
if (enqueueStrategy === EnqueueStrategy.All) {
|
|
246
|
-
return finalUrlOrigin;
|
|
247
|
-
}
|
|
248
|
-
// If the user wants to ensure the same domain is accessed, regardless of subdomains, we check to ensure the domains match
|
|
249
|
-
// Returning undefined here is intentional! If the domains don't match, having no baseUrl in enqueueLinks will cause it to not enqueue anything
|
|
250
|
-
// which is the intended behavior (since we went off domain)
|
|
251
|
-
if (enqueueStrategy === EnqueueStrategy.SameDomain) {
|
|
252
|
-
const originalHostname = getDomain(originalUrlOrigin, { mixedInputs: false });
|
|
253
|
-
const finalHostname = getDomain(finalUrlOrigin, { mixedInputs: false });
|
|
254
|
-
if (originalHostname === finalHostname) {
|
|
255
|
-
return finalUrlOrigin;
|
|
256
|
-
}
|
|
257
|
-
return undefined;
|
|
258
|
-
}
|
|
259
|
-
// Always enqueue urls that are from the same origin in all other cases, as the filtering happens on the original request url, even if there was a redirect
|
|
260
|
-
// before actually finding the urls
|
|
261
|
-
return originalUrlOrigin;
|
|
262
|
-
}
|
|
263
|
-
/**
|
|
264
|
-
* Internal function that changes the enqueue globs to match both http and https
|
|
265
|
-
*/
|
|
266
|
-
function ignoreHttpSchema(pattern) {
|
|
267
|
-
return pattern.replace(/^(https?):\/\//, 'http{s,}://');
|
|
268
|
-
}
|
|
269
|
-
//# sourceMappingURL=enqueue_links.js.map
|
|
@@ -1 +0,0 @@
|
|
|
1
|
-
{"version":3,"file":"enqueue_links.js","sourceRoot":"","sources":["../../src/enqueue_links/enqueue_links.ts"],"names":[],"mappings":"AAEA,OAAO,EAAE,MAAM,IAAI,CAAC;AACpB,OAAO,EAAE,SAAS,EAAE,MAAM,OAAO,CAAC;AAGlC,OAAO,GAAG,MAAM,YAAY,CAAC;AAkB7B,OAAO,EACH,6BAA6B,EAC7B,oCAAoC,EACpC,iCAAiC,EACjC,oBAAoB,EACpB,cAAc,EACd,wBAAwB,GAC3B,MAAM,aAAa,CAAC;AAmKrB;;;;;;;;;;;;;;;;;;;;GAoBG;AACH,MAAM,CAAN,IAAY,eAgCX;AAhCD,WAAY,eAAe;IACvB;;OAEG;IACH,8BAAW,CAAA;IAEX;;;;;;OAMG;IACH,iDAA8B,CAAA;IAE9B;;;;;;OAMG;IACH,6CAA0B,CAAA;IAE1B;;;;;;OAMG;IACH,6CAA0B,CAAA;AAC9B,CAAC,EAhCW,eAAe,KAAf,eAAe,QAgC1B;AAED;;;;;;;;;;;;;;;;;;;;;;;GAuBG;AACH,MAAM,CAAC,KAAK,UAAU,YAAY,CAC9B,OAOC;IAED,IAAI,CAAC,OAAO,IAAI,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;QAChD,MAAM,IAAI,UAAU,CAChB;YACI,4JAA4J;YAC5J,kHAAkH;SACrH,CAAC,IAAI,CAAC,IAAI,CAAC,CACf,CAAC;IACN,CAAC;IAED,EAAE,CACE,OAAc,EACd,EAAE,CAAC,MAAM,CAAC,UAAU,CAAC;QACjB,IAAI,EAAE,EAAE,CAAC,KAAK,CAAC,MAAM,CAAC,EAAE,CAAC,MAAM,CAAC;QAChC,YAAY,EAAE,EAAE,CAAC,MAAM,CAAC,OAAO,CAAC,oBAAoB,CAAC;QACrD,aAAa,EAAE,EAAE,CAAC,QAAQ,CAAC,MAAM,CAAC,OAAO,CAAC,WAAW,CAAC;QACtD,gBAAgB,EAAE,EAAE,CAAC,QAAQ,CAAC,QAAQ;QACtC,SAAS,EAAE,EAAE,CAAC,QAAQ,CAAC,OAAO;QAC9B,cAAc,EAAE,EAAE,CAAC,QAAQ,CAAC,OAAO;QACnC,KAAK,EAAE,EAAE,CAAC,QAAQ,CAAC,MAAM;QACzB,QAAQ,EAAE,EAAE,CAAC,QAAQ,CAAC,MAAM;QAC5B,OAAO,EAAE,EAAE,CAAC,QAAQ,CAAC,MAAM;QAC3B,QAAQ,EAAE,EAAE,CAAC,QAAQ,CAAC,MAAM;QAC5B,KAAK,EAAE,EAAE,CAAC,QAAQ,CAAC,MAAM;QACzB,UAAU,EAAE,EAAE,CAAC,QAAQ,CAAC,KAAK,CAAC,MAAM,CAAC,EAAE,CAAC,GAAG,CAAC,EAAE,CAAC,MAAM,EAAE,EAAE,CAAC,MAAM,CAAC,OAAO,CAAC,MAAM,CAAC,CAAC,CAAC;QAClF,KAAK,EAAE,EAAE,CAAC,QAAQ,CAAC,KAAK,CAAC,MAAM,CAAC,EAAE,CAAC,GAAG,CAAC,EAAE,CAAC,MAAM,EAAE,EAAE,CAAC,MAAM,CAAC,OAAO,CAAC,MAAM,CAAC,CAAC,CAAC;QAC7E,OAAO,EAAE,EAAE,CAAC,QAAQ,CAAC,KAAK,CAAC,MAAM,CAC7B,EAAE,CAAC,GAAG,CAAC,EAAE,CAAC,MAAM,EAAE,EAAE,CAAC,MAAM,EAAE,EAAE,CAAC,MAAM,CAAC,OAAO,CAAC,MAAM,CAAC,EAAE,EAAE,CAAC,MAAM,CAAC,OAAO,CAAC,QAAQ,CAAC,CAAC,CACvF;QACD,OAAO,EAAE,EAAE,CAAC,QAAQ,CAAC,KAAK,CAAC,MAAM,CAAC,EAAE,CAAC,GAAG,CAAC,EAAE,CAAC,MAAM,EAAE,EAAE,CAAC,MAAM,CAAC,OAAO,CAAC,QAAQ,CAAC,CAAC,CAAC;QACjF,wBAAwB,EAAE,EAAE,CAAC,QAAQ,CAAC,QAAQ;QAC9C,QAAQ,EAAE,EAAE,CAAC,QAAQ,CAAC,MAAM,CAAC,KAAK,CAAC,MAAM,CAAC,MAAM,CAAC,eAAe,CAAC,CAAC;QAClE,2BAA2B,EAAE,EAAE,CAAC,QAAQ,CAAC,OAAO;KACnD,CAAC,CACL,CAAC;IAEF,MAAM,EACF,YAAY,EACZ,KAAK,EACL,IAAI,EACJ,UAAU,EACV,OAAO,EACP,KAAK,EACL,OAAO,EACP,wBAAwB,EACxB,SAAS,EACT,2BAA2B,EAC3B,aAAa,EACb,gBAAgB,GACnB,GAAG,OAAO,CAAC;IAEZ,MAAM,wBAAwB,GAAuB,EAAE,CAAC;IACxD,MAAM,iBAAiB,GAAuB,EAAE,CAAC;IAEjD,IAAI,OAAO,EAAE,MAAM,EAAE,CAAC;QAClB,KAAK,MAAM,IAAI,IAAI,OAAO,EAAE,CAAC;YACzB,IAAI,OAAO,IAAI,KAAK,QAAQ,IAAI,MAAM,IAAI,IAAI,EAAE,CAAC;gBAC7C,wBAAwB,CAAC,IAAI,CAAC,GAAG,6BAA6B,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;YAC5E,CAAC;iBAAM,IAAI,IAAI,YAAY,MAAM,IAAI,QAAQ,IAAI,IAAI,EAAE,CAAC;gBACpD,wBAAwB,CAAC,IAAI,CAAC,GAAG,iCAAiC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;YAChF,CAAC;QACL,CAAC;IACL,CAAC;IAED,IAAI,UAAU,EAAE,MAAM,EAAE,CAAC;QACrB,GAAG,CAAC,UAAU,CAAC,qEAAqE,CAAC,CAAC;QACtF,iBAAiB,CAAC,IAAI,CAAC,GAAG,oCAAoC,CAAC,UAAU,CAAC,CAAC,CAAC;IAChF,CAAC;IAED,IAAI,KAAK,EAAE,MAAM,EAAE,CAAC;QAChB,iBAAiB,CAAC,IAAI,CAAC,GAAG,6BAA6B,CAAC,KAAK,CAAC,CAAC,CAAC;IACpE,CAAC;IAED,IAAI,OAAO,EAAE,MAAM,EAAE,CAAC;QAClB,iBAAiB,CAAC,IAAI,CAAC,GAAG,iCAAiC,CAAC,OAAO,CAAC,CAAC,CAAC;IAC1E,CAAC;IAED,IAAI,CAAC,iBAAiB,CAAC,MAAM,EAAE,CAAC;QAC5B,OAAO,CAAC,QAAQ,KAAK,eAAe,CAAC,YAAY,CAAC;IACtD,CAAC;IAED,MAAM,uBAAuB,GAAuB,EAAE,CAAC;IAEvD,IAAI,OAAO,CAAC,OAAO,EAAE,CAAC;QAClB,MAAM,GAAG,GAAG,IAAI,GAAG,CAAC,OAAO,CAAC,OAAO,CAAC,CAAC;QAErC,QAAQ,OAAO,CAAC,QAAQ,EAAE,CAAC;YACvB,KAAK,eAAe,CAAC,YAAY;gBAC7B,sFAAsF;gBACtF,uFAAuF;gBACvF,yCAAyC;gBACzC,uBAAuB,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,gBAAgB,CAAC,GAAG,GAAG,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC,CAAC;gBAC7E,MAAM;YACV,KAAK,eAAe,CAAC,UAAU,CAAC,CAAC,CAAC;gBAC9B,4CAA4C;gBAC5C,MAAM,eAAe,GAAG,SAAS,CAAC,GAAG,CAAC,QAAQ,EAAE,EAAE,WAAW,EAAE,KAAK,EAAE,CAAC,CAAC;gBAExE,IAAI,eAAe,EAAE,CAAC;oBAClB,gHAAgH;oBAChH,GAAG,CAAC,QAAQ,GAAG,eAAe,CAAC;oBAC/B,uBAAuB,CAAC,IAAI,CACxB,EAAE,IAAI,EAAE,gBAAgB,CAAC,GAAG,GAAG,CAAC,MAAM,CAAC,OAAO,CAAC,eAAe,EAAE,KAAK,eAAe,EAAE,CAAC,KAAK,CAAC,EAAE,EAC/F,EAAE,IAAI,EAAE,gBAAgB,CAAC,GAAG,GAAG,CAAC,MAAM,KAAK,CAAC,EAAE,CACjD,CAAC;gBACN,CAAC;qBAAM,CAAC;oBACJ,6FAA6F;oBAC7F,4BAA4B;oBAC5B,uBAAuB,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,gBAAgB,CAAC,GAAG,GAAG,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC,CAAC;gBACjF,CAAC;gBAED,MAAM;YACV,CAAC;YACD,KAAK,eAAe,CAAC,UAAU,CAAC,CAAC,CAAC;gBAC9B,4EAA4E;gBAC5E,uBAAuB,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,GAAG,GAAG,CAAC,MAAM,KAAK,EAAE,CAAC,CAAC;gBAC3D,MAAM;YACV,CAAC;YACD,KAAK,eAAe,CAAC,GAAG,CAAC;YACzB;gBACI,uBAAuB,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,eAAe,EAAE,CAAC,CAAC;gBACxD,MAAM;QACd,CAAC;IACL,CAAC;IAED,KAAK,UAAU,qBAAqB,CAChC,eAAwE,EACxE,MAA4B;QAE5B,IAAI,gBAAgB,IAAI,eAAe,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YACjD,MAAM,OAAO,CAAC,GAAG,CACb,eAAe,CAAC,GAAG,CAAC,CAAC,OAAO,EAAE,EAAE;gBAC5B,OAAO,gBAAgB,CAAC;oBACpB,GAAG,EAAE,OAAO,CAAC,GAAG;oBAChB,MAAM,EAAE,OAAO,CAAC,aAAa,IAAI,MAAM;iBAC1C,CAAkB,CAAC;YACxB,CAAC,CAAC,CACL,CAAC;QACN,CAAC;IACL,CAAC;IAED,IAAI,cAAc,GAAG,oBAAoB,CAAC,IAAI,EAAE,OAAO,CAAC,CAAC;IAEzD,IAAI,aAAa,EAAE,CAAC;QAChB,MAAM,eAAe,GAAqB,EAAE,CAAC;QAE7C,cAAc,GAAG,cAAc,CAAC,MAAM,CAAC,CAAC,OAAO,EAAE,EAAE;YAC/C,IAAI,aAAa,CAAC,SAAS,CAAC,OAAO,CAAC,GAAG,CAAC,EAAE,CAAC;gBACvC,OAAO,IAAI,CAAC;YAChB,CAAC;YAED,eAAe,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;YAC9B,OAAO,KAAK,CAAC;QACjB,CAAC,CAAC,CAAC;QAEH,MAAM,qBAAqB,CAAC,eAAe,EAAE,WAAW,CAAC,CAAC;IAC9D,CAAC;IAED,IAAI,wBAAwB,EAAE,CAAC;QAC3B,MAAM,eAAe,GAAqB,EAAE,CAAC;QAE7C,cAAc,GAAG,cAAc;aAC1B,GAAG,CAAC,CAAC,OAAO,EAAE,EAAE;YACb,MAAM,kBAAkB,GAAG,wBAAwB,CAAC,OAAO,CAAC,CAAC;YAC7D,IAAI,CAAC,kBAAkB,EAAE,CAAC;gBACtB,eAAe,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;YAClC,CAAC;YACD,OAAO,kBAAkB,CAAC;QAC9B,CAAC,CAAC;aACD,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,OAAO,CAAC,CAAC,CAAC,CAAqB,CAAC;QAEnD,MAAM,qBAAqB,CAAC,eAAe,EAAE,SAAS,CAAC,CAAC;IAC5D,CAAC;IAED,KAAK,UAAU,sBAAsB;QACjC,MAAM,eAAe,GAAa,EAAE,CAAC;QAErC,sEAAsE;QACtE,IAAI,iBAAiB,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;YACjC,OAAO,cAAc,CACjB,cAAc,EACd,uBAAuB,EACvB,wBAAwB,EACxB,OAAO,CAAC,QAAQ,EAChB,CAAC,GAAG,EAAE,EAAE,CAAC,eAAe,CAAC,IAAI,CAAC,GAAG,CAAC,CACrC,CAAC;QACN,CAAC;QAED,qDAAqD;QACrD,MAAM,gCAAgC,GAAG,cAAc,CACnD,cAAc,EACd,iBAAiB,EACjB,wBAAwB,EACxB,OAAO,CAAC,QAAQ,EAChB,CAAC,GAAG,EAAE,EAAE,CAAC,eAAe,CAAC,IAAI,CAAC,GAAG,CAAC,CACrC,CAAC;QACF,+EAA+E;QAC/E,MAAM,QAAQ,GAAG,wBAAwB,CAAC,gCAAgC,EAAE,uBAAuB,EAAE,CAAC,GAAG,EAAE,EAAE,CACzG,eAAe,CAAC,IAAI,CAAC,GAAG,CAAC,CAC5B,CAAC;QAEF,MAAM,qBAAqB,CACvB,eAAe,CAAC,GAAG,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,CAAC,EAAE,GAAG,EAAE,CAAC,CAAC,EACvC,SAAS,CACZ,CAAC;QAEF,OAAO,QAAQ,CAAC;IACpB,CAAC;IAED,IAAI,QAAQ,GAAG,MAAM,sBAAsB,EAAE,CAAC;IAC9C,IAAI,OAAO,KAAK,KAAK,QAAQ,IAAI,KAAK,GAAG,QAAQ,CAAC,MAAM,EAAE,CAAC;QACvD,MAAM,qBAAqB,CAAC,QAAQ,CAAC,KAAK,CAAC,KAAK,CAAC,EAAE,cAAc,CAAC,CAAC;QACnE,QAAQ,GAAG,QAAQ,CAAC,KAAK,CAAC,CAAC,EAAE,KAAK,CAAC,CAAC;IACxC,CAAC;IAED,MAAM,EAAE,aAAa,EAAE,GAAG,MAAM,YAAY,CAAC,kBAAkB,CAAC,QAAQ,EAAE;QACtE,SAAS;QACT,2BAA2B;KAC9B,CAAC,CAAC;IAEH,OAAO,EAAE,iBAAiB,EAAE,aAAa,EAAE,mBAAmB,EAAE,EAAE,EAAE,CAAC;AACzE,CAAC;AAED;;;;;;;GAOG;AACH,MAAM,UAAU,sCAAsC,CAAC,EACnD,eAAe,EACf,eAAe,EACf,kBAAkB,EAClB,mBAAmB,GACN;IACb,wCAAwC;IACxC,IAAI,mBAAmB,EAAE,CAAC;QACtB,OAAO,mBAAmB,CAAC;IAC/B,CAAC;IAED,MAAM,iBAAiB,GAAG,IAAI,GAAG,CAAC,kBAAkB,CAAC,CAAC,MAAM,CAAC;IAC7D,MAAM,cAAc,GAAG,IAAI,GAAG,CAAC,eAAe,IAAI,kBAAkB,CAAC,CAAC,MAAM,CAAC;IAE7E,6DAA6D;IAC7D,IAAI,eAAe,KAAK,eAAe,CAAC,GAAG,EAAE,CAAC;QAC1C,OAAO,cAAc,CAAC;IAC1B,CAAC;IAED,0HAA0H;IAC1H,+IAA+I;IAC/I,4DAA4D;IAC5D,IAAI,eAAe,KAAK,eAAe,CAAC,UAAU,EAAE,CAAC;QACjD,MAAM,gBAAgB,GAAG,SAAS,CAAC,iBAAiB,EAAE,EAAE,WAAW,EAAE,KAAK,EAAE,CAAE,CAAC;QAC/E,MAAM,aAAa,GAAG,SAAS,CAAC,cAAc,EAAE,EAAE,WAAW,EAAE,KAAK,EAAE,CAAE,CAAC;QAEzE,IAAI,gBAAgB,KAAK,aAAa,EAAE,CAAC;YACrC,OAAO,cAAc,CAAC;QAC1B,CAAC;QAED,OAAO,SAAS,CAAC;IACrB,CAAC;IAED,2JAA2J;IAC3J,mCAAmC;IACnC,OAAO,iBAAiB,CAAC;AAC7B,CAAC;AAYD;;GAEG;AACH,SAAS,gBAAgB,CAAC,OAAe;IACrC,OAAO,OAAO,CAAC,OAAO,CAAC,gBAAgB,EAAE,aAAa,CAAC,CAAC;AAC5D,CAAC"}
|
package/enqueue_links/index.d.ts
DELETED
|
@@ -1 +0,0 @@
|
|
|
1
|
-
{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../../src/enqueue_links/index.ts"],"names":[],"mappings":"AAAA,cAAc,oBAAoB,CAAC;AACnC,cAAc,aAAa,CAAC"}
|
package/enqueue_links/index.js
DELETED
|
@@ -1 +0,0 @@
|
|
|
1
|
-
{"version":3,"file":"index.js","sourceRoot":"","sources":["../../src/enqueue_links/index.ts"],"names":[],"mappings":"AAAA,cAAc,oBAAoB,CAAC;AACnC,cAAc,aAAa,CAAC"}
|
|
@@ -1,73 +0,0 @@
|
|
|
1
|
-
import type { Awaitable } from '@crawlee/types';
|
|
2
|
-
import type { RequestOptions } from '../request.js';
|
|
3
|
-
import { Request } from '../request.js';
|
|
4
|
-
import type { EnqueueLinksOptions } from './enqueue_links.js';
|
|
5
|
-
export { tryAbsoluteURL } from '@crawlee/utils';
|
|
6
|
-
export type UrlPatternObject = {
|
|
7
|
-
glob?: string;
|
|
8
|
-
regexp?: RegExp;
|
|
9
|
-
} & Pick<RequestOptions, 'method' | 'payload' | 'label' | 'userData' | 'headers'>;
|
|
10
|
-
export type PseudoUrlObject = {
|
|
11
|
-
purl: string;
|
|
12
|
-
} & Pick<RequestOptions, 'method' | 'payload' | 'label' | 'userData' | 'headers'>;
|
|
13
|
-
export type PseudoUrlInput = string | PseudoUrlObject;
|
|
14
|
-
export type GlobObject = {
|
|
15
|
-
glob: string;
|
|
16
|
-
} & Pick<RequestOptions, 'method' | 'payload' | 'label' | 'userData' | 'headers'>;
|
|
17
|
-
export type GlobInput = string | GlobObject;
|
|
18
|
-
export type RegExpObject = {
|
|
19
|
-
regexp: RegExp;
|
|
20
|
-
} & Pick<RequestOptions, 'method' | 'payload' | 'label' | 'userData' | 'headers'>;
|
|
21
|
-
export type RegExpInput = RegExp | RegExpObject;
|
|
22
|
-
export type SkippedRequestReason = 'robotsTxt' | 'limit' | 'enqueueLimit' | 'filters' | 'redirect' | 'depth';
|
|
23
|
-
export type SkippedRequestCallback = (args: {
|
|
24
|
-
url: string;
|
|
25
|
-
reason: SkippedRequestReason;
|
|
26
|
-
}) => Awaitable<void>;
|
|
27
|
-
/**
|
|
28
|
-
* @ignore
|
|
29
|
-
*/
|
|
30
|
-
export declare function updateEnqueueLinksPatternCache(item: GlobInput | RegExpInput | PseudoUrlInput, pattern: RegExpObject | GlobObject): void;
|
|
31
|
-
/**
|
|
32
|
-
* Helper factory used in the `enqueueLinks()` and enqueueLinksByClickingElements() function
|
|
33
|
-
* to construct RegExps from PseudoUrl strings.
|
|
34
|
-
* @ignore
|
|
35
|
-
*/
|
|
36
|
-
export declare function constructRegExpObjectsFromPseudoUrls(pseudoUrls: readonly PseudoUrlInput[]): RegExpObject[];
|
|
37
|
-
/**
|
|
38
|
-
* Helper factory used in the `enqueueLinks()` and enqueueLinksByClickingElements() function
|
|
39
|
-
* to construct Glob objects from Glob pattern strings.
|
|
40
|
-
* @ignore
|
|
41
|
-
*/
|
|
42
|
-
export declare function constructGlobObjectsFromGlobs(globs: readonly GlobInput[]): GlobObject[];
|
|
43
|
-
/**
|
|
44
|
-
* @internal
|
|
45
|
-
*/
|
|
46
|
-
export declare function validateGlobPattern(glob: string): string;
|
|
47
|
-
/**
|
|
48
|
-
* Helper factory used in the `enqueueLinks()` and enqueueLinksByClickingElements() function
|
|
49
|
-
* to check RegExps input and return valid RegExps.
|
|
50
|
-
* @ignore
|
|
51
|
-
*/
|
|
52
|
-
export declare function constructRegExpObjectsFromRegExps(regexps: readonly RegExpInput[]): RegExpObject[];
|
|
53
|
-
/**
|
|
54
|
-
* @ignore
|
|
55
|
-
*/
|
|
56
|
-
export declare function createRequests(requestOptions: (string | RequestOptions)[], urlPatternObjects?: UrlPatternObject[], excludePatternObjects?: UrlPatternObject[], strategy?: EnqueueLinksOptions['strategy'], onSkippedUrl?: (url: string) => void): Request[];
|
|
57
|
-
export declare function filterRequestsByPatterns(requests: Request[], patterns?: UrlPatternObject[], onSkippedUrl?: (url: string) => void): Request[];
|
|
58
|
-
/**
|
|
59
|
-
* @ignore
|
|
60
|
-
*/
|
|
61
|
-
export declare function createRequestOptions(sources: readonly (string | Record<string, unknown>)[], options?: Pick<EnqueueLinksOptions, 'label' | 'userData' | 'baseUrl' | 'skipNavigation' | 'strategy'>): RequestOptions[];
|
|
62
|
-
/**
|
|
63
|
-
* Takes an Apify {@link RequestOptions} object and changes its attributes in a desired way. This user-function is used
|
|
64
|
-
* {@link enqueueLinks} to modify requests before enqueuing them.
|
|
65
|
-
*/
|
|
66
|
-
export interface RequestTransform {
|
|
67
|
-
/**
|
|
68
|
-
* @param original Request options to be modified.
|
|
69
|
-
* @returns The modified request options to enqueue.
|
|
70
|
-
*/
|
|
71
|
-
(original: RequestOptions): RequestOptions | false | undefined | null;
|
|
72
|
-
}
|
|
73
|
-
//# sourceMappingURL=shared.d.ts.map
|
|
@@ -1 +0,0 @@
|
|
|
1
|
-
{"version":3,"file":"shared.d.ts","sourceRoot":"","sources":["../../src/enqueue_links/shared.ts"],"names":[],"mappings":"AAEA,OAAO,KAAK,EAAE,SAAS,EAAE,MAAM,gBAAgB,CAAC;AAKhD,OAAO,KAAK,EAAE,cAAc,EAAE,MAAM,eAAe,CAAC;AACpD,OAAO,EAAE,OAAO,EAAE,MAAM,eAAe,CAAC;AACxC,OAAO,KAAK,EAAE,mBAAmB,EAAE,MAAM,oBAAoB,CAAC;AAE9D,OAAO,EAAE,cAAc,EAAE,MAAM,gBAAgB,CAAC;AAYhD,MAAM,MAAM,gBAAgB,GAAG;IAC3B,IAAI,CAAC,EAAE,MAAM,CAAC;IACd,MAAM,CAAC,EAAE,MAAM,CAAC;CACnB,GAAG,IAAI,CAAC,cAAc,EAAE,QAAQ,GAAG,SAAS,GAAG,OAAO,GAAG,UAAU,GAAG,SAAS,CAAC,CAAC;AAElF,MAAM,MAAM,eAAe,GAAG;IAAE,IAAI,EAAE,MAAM,CAAA;CAAE,GAAG,IAAI,CACjD,cAAc,EACd,QAAQ,GAAG,SAAS,GAAG,OAAO,GAAG,UAAU,GAAG,SAAS,CAC1D,CAAC;AAEF,MAAM,MAAM,cAAc,GAAG,MAAM,GAAG,eAAe,CAAC;AAEtD,MAAM,MAAM,UAAU,GAAG;IAAE,IAAI,EAAE,MAAM,CAAA;CAAE,GAAG,IAAI,CAC5C,cAAc,EACd,QAAQ,GAAG,SAAS,GAAG,OAAO,GAAG,UAAU,GAAG,SAAS,CAC1D,CAAC;AAEF,MAAM,MAAM,SAAS,GAAG,MAAM,GAAG,UAAU,CAAC;AAE5C,MAAM,MAAM,YAAY,GAAG;IAAE,MAAM,EAAE,MAAM,CAAA;CAAE,GAAG,IAAI,CAChD,cAAc,EACd,QAAQ,GAAG,SAAS,GAAG,OAAO,GAAG,UAAU,GAAG,SAAS,CAC1D,CAAC;AAEF,MAAM,MAAM,WAAW,GAAG,MAAM,GAAG,YAAY,CAAC;AAEhD,MAAM,MAAM,oBAAoB,GAAG,WAAW,GAAG,OAAO,GAAG,cAAc,GAAG,SAAS,GAAG,UAAU,GAAG,OAAO,CAAC;AAE7G,MAAM,MAAM,sBAAsB,GAAG,CAAC,IAAI,EAAE;IAAE,GAAG,EAAE,MAAM,CAAC;IAAC,MAAM,EAAE,oBAAoB,CAAA;CAAE,KAAK,SAAS,CAAC,IAAI,CAAC,CAAC;AAE9G;;GAEG;AACH,wBAAgB,8BAA8B,CAC1C,IAAI,EAAE,SAAS,GAAG,WAAW,GAAG,cAAc,EAC9C,OAAO,EAAE,YAAY,GAAG,UAAU,GACnC,IAAI,CAMN;AAED;;;;GAIG;AACH,wBAAgB,oCAAoC,CAAC,UAAU,EAAE,SAAS,cAAc,EAAE,GAAG,YAAY,EAAE,CAiB1G;AAED;;;;GAIG;AACH,wBAAgB,6BAA6B,CAAC,KAAK,EAAE,SAAS,SAAS,EAAE,GAAG,UAAU,EAAE,CAkCvF;AAED;;GAEG;AACH,wBAAgB,mBAAmB,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,CAKxD;AAED;;;;GAIG;AACH,wBAAgB,iCAAiC,CAAC,OAAO,EAAE,SAAS,WAAW,EAAE,GAAG,YAAY,EAAE,CAgBjG;AAED;;GAEG;AACH,wBAAgB,cAAc,CAC1B,cAAc,EAAE,CAAC,MAAM,GAAG,cAAc,CAAC,EAAE,EAC3C,iBAAiB,CAAC,EAAE,gBAAgB,EAAE,EACtC,qBAAqB,GAAE,gBAAgB,EAAO,EAC9C,QAAQ,CAAC,EAAE,mBAAmB,CAAC,UAAU,CAAC,EAC1C,YAAY,CAAC,EAAE,CAAC,GAAG,EAAE,MAAM,KAAK,IAAI,GACrC,OAAO,EAAE,CAqCX;AAED,wBAAgB,wBAAwB,CACpC,QAAQ,EAAE,OAAO,EAAE,EACnB,QAAQ,CAAC,EAAE,gBAAgB,EAAE,EAC7B,YAAY,CAAC,EAAE,CAAC,GAAG,EAAE,MAAM,KAAK,IAAI,GACrC,OAAO,EAAE,CAmBX;AAED;;GAEG;AACH,wBAAgB,oBAAoB,CAChC,OAAO,EAAE,SAAS,CAAC,MAAM,GAAG,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,CAAC,EAAE,EACtD,OAAO,GAAE,IAAI,CAAC,mBAAmB,EAAE,OAAO,GAAG,UAAU,GAAG,SAAS,GAAG,gBAAgB,GAAG,UAAU,CAAM,GAC1G,cAAc,EAAE,CA+BlB;AAmBD;;;GAGG;AACH,MAAM,WAAW,gBAAgB;IAC7B;;;OAGG;IACH,CAAC,QAAQ,EAAE,cAAc,GAAG,cAAc,GAAG,KAAK,GAAG,SAAS,GAAG,IAAI,CAAC;CACzE"}
|
package/enqueue_links/shared.js
DELETED
|
@@ -1,212 +0,0 @@
|
|
|
1
|
-
import { URL } from 'node:url';
|
|
2
|
-
import { Minimatch } from 'minimatch';
|
|
3
|
-
import { purlToRegExp } from '@apify/pseudo_url';
|
|
4
|
-
import { Request } from '../request.js';
|
|
5
|
-
export { tryAbsoluteURL } from '@crawlee/utils';
|
|
6
|
-
const MAX_ENQUEUE_LINKS_CACHE_SIZE = 1000;
|
|
7
|
-
/**
|
|
8
|
-
* To enable direct use of the Actor UI `globs`/`regexps`/`pseudoUrls` output while keeping high performance,
|
|
9
|
-
* all the regexps from the output are only constructed once and kept in a cache
|
|
10
|
-
* by the `enqueueLinks()` function.
|
|
11
|
-
* @ignore
|
|
12
|
-
*/
|
|
13
|
-
const enqueueLinksPatternCache = new Map();
|
|
14
|
-
/**
|
|
15
|
-
* @ignore
|
|
16
|
-
*/
|
|
17
|
-
export function updateEnqueueLinksPatternCache(item, pattern) {
|
|
18
|
-
enqueueLinksPatternCache.set(item, pattern);
|
|
19
|
-
if (enqueueLinksPatternCache.size > MAX_ENQUEUE_LINKS_CACHE_SIZE) {
|
|
20
|
-
const key = enqueueLinksPatternCache.keys().next().value;
|
|
21
|
-
enqueueLinksPatternCache.delete(key);
|
|
22
|
-
}
|
|
23
|
-
}
|
|
24
|
-
/**
|
|
25
|
-
* Helper factory used in the `enqueueLinks()` and enqueueLinksByClickingElements() function
|
|
26
|
-
* to construct RegExps from PseudoUrl strings.
|
|
27
|
-
* @ignore
|
|
28
|
-
*/
|
|
29
|
-
export function constructRegExpObjectsFromPseudoUrls(pseudoUrls) {
|
|
30
|
-
return pseudoUrls.map((item) => {
|
|
31
|
-
// Get pseudoUrl object from cache.
|
|
32
|
-
let regexpObject = enqueueLinksPatternCache.get(item);
|
|
33
|
-
if (regexpObject)
|
|
34
|
-
return regexpObject;
|
|
35
|
-
if (typeof item === 'string') {
|
|
36
|
-
regexpObject = { regexp: purlToRegExp(item) };
|
|
37
|
-
}
|
|
38
|
-
else {
|
|
39
|
-
const { purl, ...requestOptions } = item;
|
|
40
|
-
regexpObject = { regexp: purlToRegExp(purl), ...requestOptions };
|
|
41
|
-
}
|
|
42
|
-
updateEnqueueLinksPatternCache(item, regexpObject);
|
|
43
|
-
return regexpObject;
|
|
44
|
-
});
|
|
45
|
-
}
|
|
46
|
-
/**
|
|
47
|
-
* Helper factory used in the `enqueueLinks()` and enqueueLinksByClickingElements() function
|
|
48
|
-
* to construct Glob objects from Glob pattern strings.
|
|
49
|
-
* @ignore
|
|
50
|
-
*/
|
|
51
|
-
export function constructGlobObjectsFromGlobs(globs) {
|
|
52
|
-
return globs
|
|
53
|
-
.filter((glob) => {
|
|
54
|
-
// Skip possibly nullish, empty strings
|
|
55
|
-
if (!glob) {
|
|
56
|
-
return false;
|
|
57
|
-
}
|
|
58
|
-
if (typeof glob === 'string') {
|
|
59
|
-
return glob.trim().length > 0;
|
|
60
|
-
}
|
|
61
|
-
if (glob.glob) {
|
|
62
|
-
return glob.glob.trim().length > 0;
|
|
63
|
-
}
|
|
64
|
-
return false;
|
|
65
|
-
})
|
|
66
|
-
.map((item) => {
|
|
67
|
-
// Get glob object from cache.
|
|
68
|
-
let globObject = enqueueLinksPatternCache.get(item);
|
|
69
|
-
if (globObject)
|
|
70
|
-
return globObject;
|
|
71
|
-
if (typeof item === 'string') {
|
|
72
|
-
globObject = { glob: validateGlobPattern(item) };
|
|
73
|
-
}
|
|
74
|
-
else {
|
|
75
|
-
const { glob, ...requestOptions } = item;
|
|
76
|
-
globObject = { glob: validateGlobPattern(glob), ...requestOptions };
|
|
77
|
-
}
|
|
78
|
-
updateEnqueueLinksPatternCache(item, globObject);
|
|
79
|
-
return globObject;
|
|
80
|
-
});
|
|
81
|
-
}
|
|
82
|
-
/**
|
|
83
|
-
* @internal
|
|
84
|
-
*/
|
|
85
|
-
export function validateGlobPattern(glob) {
|
|
86
|
-
const globTrimmed = glob.trim();
|
|
87
|
-
if (globTrimmed.length === 0)
|
|
88
|
-
throw new Error(`Cannot parse Glob pattern '${globTrimmed}': it must be an non-empty string`);
|
|
89
|
-
return globTrimmed;
|
|
90
|
-
}
|
|
91
|
-
/**
|
|
92
|
-
* Helper factory used in the `enqueueLinks()` and enqueueLinksByClickingElements() function
|
|
93
|
-
* to check RegExps input and return valid RegExps.
|
|
94
|
-
* @ignore
|
|
95
|
-
*/
|
|
96
|
-
export function constructRegExpObjectsFromRegExps(regexps) {
|
|
97
|
-
return regexps.map((item) => {
|
|
98
|
-
// Get regexp object from cache.
|
|
99
|
-
let regexpObject = enqueueLinksPatternCache.get(item);
|
|
100
|
-
if (regexpObject)
|
|
101
|
-
return regexpObject;
|
|
102
|
-
if (item instanceof RegExp) {
|
|
103
|
-
regexpObject = { regexp: item };
|
|
104
|
-
}
|
|
105
|
-
else {
|
|
106
|
-
regexpObject = item;
|
|
107
|
-
}
|
|
108
|
-
updateEnqueueLinksPatternCache(item, regexpObject);
|
|
109
|
-
return regexpObject;
|
|
110
|
-
});
|
|
111
|
-
}
|
|
112
|
-
/**
|
|
113
|
-
* @ignore
|
|
114
|
-
*/
|
|
115
|
-
export function createRequests(requestOptions, urlPatternObjects, excludePatternObjects = [], strategy, onSkippedUrl) {
|
|
116
|
-
const excludePatternObjectMatchers = excludePatternObjects.map(createPatternObjectMatcher);
|
|
117
|
-
const urlPatternObjectMatchers = urlPatternObjects?.map(createPatternObjectMatcher);
|
|
118
|
-
return requestOptions
|
|
119
|
-
.map((opts) => ({ url: typeof opts === 'string' ? opts : opts.url, opts }))
|
|
120
|
-
.filter(({ url }) => {
|
|
121
|
-
const matchesExcludePatterns = excludePatternObjectMatchers.some(({ match }) => match(url));
|
|
122
|
-
if (matchesExcludePatterns) {
|
|
123
|
-
onSkippedUrl?.(url);
|
|
124
|
-
}
|
|
125
|
-
return !matchesExcludePatterns;
|
|
126
|
-
})
|
|
127
|
-
.map(({ url, opts }) => {
|
|
128
|
-
if (!urlPatternObjectMatchers || !urlPatternObjectMatchers.length) {
|
|
129
|
-
return new Request(typeof opts === 'string' ? { url: opts, enqueueStrategy: strategy } : { ...opts });
|
|
130
|
-
}
|
|
131
|
-
for (const urlPatternObject of urlPatternObjectMatchers) {
|
|
132
|
-
const { match, glob, regexp, ...requestRegExpOptions } = urlPatternObject;
|
|
133
|
-
if (match(url)) {
|
|
134
|
-
const request = typeof opts === 'string'
|
|
135
|
-
? { url: opts, ...requestRegExpOptions, enqueueStrategy: strategy }
|
|
136
|
-
: { ...opts, ...requestRegExpOptions, enqueueStrategy: strategy };
|
|
137
|
-
return new Request(request);
|
|
138
|
-
}
|
|
139
|
-
}
|
|
140
|
-
// didn't match any positive pattern
|
|
141
|
-
onSkippedUrl?.(url);
|
|
142
|
-
return null;
|
|
143
|
-
})
|
|
144
|
-
.filter((request) => request);
|
|
145
|
-
}
|
|
146
|
-
export function filterRequestsByPatterns(requests, patterns, onSkippedUrl) {
|
|
147
|
-
if (!patterns?.length) {
|
|
148
|
-
return requests;
|
|
149
|
-
}
|
|
150
|
-
const filtered = [];
|
|
151
|
-
const patternMatchers = patterns?.map(createPatternObjectMatcher);
|
|
152
|
-
for (const request of requests) {
|
|
153
|
-
const matchingPattern = patternMatchers.find(({ match }) => match(request.url));
|
|
154
|
-
if (matchingPattern !== undefined) {
|
|
155
|
-
filtered.push(request);
|
|
156
|
-
}
|
|
157
|
-
else {
|
|
158
|
-
onSkippedUrl?.(request.url);
|
|
159
|
-
}
|
|
160
|
-
}
|
|
161
|
-
return filtered;
|
|
162
|
-
}
|
|
163
|
-
/**
|
|
164
|
-
* @ignore
|
|
165
|
-
*/
|
|
166
|
-
export function createRequestOptions(sources, options = {}) {
|
|
167
|
-
return sources
|
|
168
|
-
.map((src) => typeof src === 'string'
|
|
169
|
-
? { url: src, enqueueStrategy: options.strategy }
|
|
170
|
-
: { ...src, enqueueStrategy: options.strategy })
|
|
171
|
-
.filter(({ url }) => {
|
|
172
|
-
try {
|
|
173
|
-
return new URL(url, options.baseUrl).href;
|
|
174
|
-
}
|
|
175
|
-
catch (err) {
|
|
176
|
-
return false;
|
|
177
|
-
}
|
|
178
|
-
})
|
|
179
|
-
.map((requestOptions) => {
|
|
180
|
-
requestOptions.url = new URL(requestOptions.url, options.baseUrl).href;
|
|
181
|
-
requestOptions.userData ??= options.userData ?? {};
|
|
182
|
-
if (typeof options.label === 'string') {
|
|
183
|
-
requestOptions.userData = {
|
|
184
|
-
...requestOptions.userData,
|
|
185
|
-
label: options.label,
|
|
186
|
-
};
|
|
187
|
-
}
|
|
188
|
-
if (options.skipNavigation) {
|
|
189
|
-
requestOptions.skipNavigation = true;
|
|
190
|
-
}
|
|
191
|
-
return requestOptions;
|
|
192
|
-
});
|
|
193
|
-
}
|
|
194
|
-
/**
|
|
195
|
-
* @ignore
|
|
196
|
-
*/
|
|
197
|
-
function createPatternObjectMatcher(urlPatternObject) {
|
|
198
|
-
const { regexp, glob } = urlPatternObject;
|
|
199
|
-
let match;
|
|
200
|
-
if (regexp) {
|
|
201
|
-
match = (url) => regexp.test(url);
|
|
202
|
-
}
|
|
203
|
-
else if (glob) {
|
|
204
|
-
const m = new Minimatch(glob, { nocase: true });
|
|
205
|
-
match = (url) => m.match(url);
|
|
206
|
-
}
|
|
207
|
-
else {
|
|
208
|
-
match = () => false;
|
|
209
|
-
}
|
|
210
|
-
return { ...urlPatternObject, match };
|
|
211
|
-
}
|
|
212
|
-
//# sourceMappingURL=shared.js.map
|
|
@@ -1 +0,0 @@
|
|
|
1
|
-
{"version":3,"file":"shared.js","sourceRoot":"","sources":["../../src/enqueue_links/shared.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,GAAG,EAAE,MAAM,UAAU,CAAC;AAG/B,OAAO,EAAE,SAAS,EAAE,MAAM,WAAW,CAAC;AAEtC,OAAO,EAAE,YAAY,EAAE,MAAM,mBAAmB,CAAC;AAGjD,OAAO,EAAE,OAAO,EAAE,MAAM,eAAe,CAAC;AAGxC,OAAO,EAAE,cAAc,EAAE,MAAM,gBAAgB,CAAC;AAEhD,MAAM,4BAA4B,GAAG,IAAI,CAAC;AAE1C;;;;;GAKG;AACH,MAAM,wBAAwB,GAAG,IAAI,GAAG,EAAE,CAAC;AAgC3C;;GAEG;AACH,MAAM,UAAU,8BAA8B,CAC1C,IAA8C,EAC9C,OAAkC;IAElC,wBAAwB,CAAC,GAAG,CAAC,IAAI,EAAE,OAAO,CAAC,CAAC;IAC5C,IAAI,wBAAwB,CAAC,IAAI,GAAG,4BAA4B,EAAE,CAAC;QAC/D,MAAM,GAAG,GAAG,wBAAwB,CAAC,IAAI,EAAE,CAAC,IAAI,EAAE,CAAC,KAAK,CAAC;QACzD,wBAAwB,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC;IACzC,CAAC;AACL,CAAC;AAED;;;;GAIG;AACH,MAAM,UAAU,oCAAoC,CAAC,UAAqC;IACtF,OAAO,UAAU,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE;QAC3B,mCAAmC;QACnC,IAAI,YAAY,GAAG,wBAAwB,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;QACtD,IAAI,YAAY;YAAE,OAAO,YAAY,CAAC;QAEtC,IAAI,OAAO,IAAI,KAAK,QAAQ,EAAE,CAAC;YAC3B,YAAY,GAAG,EAAE,MAAM,EAAE,YAAY,CAAC,IAAI,CAAC,EAAE,CAAC;QAClD,CAAC;aAAM,CAAC;YACJ,MAAM,EAAE,IAAI,EAAE,GAAG,cAAc,EAAE,GAAG,IAAI,CAAC;YACzC,YAAY,GAAG,EAAE,MAAM,EAAE,YAAY,CAAC,IAAI,CAAC,EAAE,GAAG,cAAc,EAAE,CAAC;QACrE,CAAC;QAED,8BAA8B,CAAC,IAAI,EAAE,YAAY,CAAC,CAAC;QAEnD,OAAO,YAAY,CAAC;IACxB,CAAC,CAAC,CAAC;AACP,CAAC;AAED;;;;GAIG;AACH,MAAM,UAAU,6BAA6B,CAAC,KAA2B;IACrE,OAAO,KAAK;SACP,MAAM,CAAC,CAAC,IAAI,EAAE,EAAE;QACb,uCAAuC;QACvC,IAAI,CAAC,IAAI,EAAE,CAAC;YACR,OAAO,KAAK,CAAC;QACjB,CAAC;QAED,IAAI,OAAO,IAAI,KAAK,QAAQ,EAAE,CAAC;YAC3B,OAAO,IAAI,CAAC,IAAI,EAAE,CAAC,MAAM,GAAG,CAAC,CAAC;QAClC,CAAC;QAED,IAAI,IAAI,CAAC,IAAI,EAAE,CAAC;YACZ,OAAO,IAAI,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC,MAAM,GAAG,CAAC,CAAC;QACvC,CAAC;QAED,OAAO,KAAK,CAAC;IACjB,CAAC,CAAC;SACD,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE;QACV,8BAA8B;QAC9B,IAAI,UAAU,GAAG,wBAAwB,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;QACpD,IAAI,UAAU;YAAE,OAAO,UAAU,CAAC;QAElC,IAAI,OAAO,IAAI,KAAK,QAAQ,EAAE,CAAC;YAC3B,UAAU,GAAG,EAAE,IAAI,EAAE,mBAAmB,CAAC,IAAI,CAAC,EAAE,CAAC;QACrD,CAAC;aAAM,CAAC;YACJ,MAAM,EAAE,IAAI,EAAE,GAAG,cAAc,EAAE,GAAG,IAAI,CAAC;YACzC,UAAU,GAAG,EAAE,IAAI,EAAE,mBAAmB,CAAC,IAAI,CAAC,EAAE,GAAG,cAAc,EAAE,CAAC;QACxE,CAAC;QAED,8BAA8B,CAAC,IAAI,EAAE,UAAU,CAAC,CAAC;QAEjD,OAAO,UAAU,CAAC;IACtB,CAAC,CAAC,CAAC;AACX,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,mBAAmB,CAAC,IAAY;IAC5C,MAAM,WAAW,GAAG,IAAI,CAAC,IAAI,EAAE,CAAC;IAChC,IAAI,WAAW,CAAC,MAAM,KAAK,CAAC;QACxB,MAAM,IAAI,KAAK,CAAC,8BAA8B,WAAW,mCAAmC,CAAC,CAAC;IAClG,OAAO,WAAW,CAAC;AACvB,CAAC;AAED;;;;GAIG;AACH,MAAM,UAAU,iCAAiC,CAAC,OAA+B;IAC7E,OAAO,OAAO,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE;QACxB,gCAAgC;QAChC,IAAI,YAAY,GAAG,wBAAwB,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;QACtD,IAAI,YAAY;YAAE,OAAO,YAAY,CAAC;QAEtC,IAAI,IAAI,YAAY,MAAM,EAAE,CAAC;YACzB,YAAY,GAAG,EAAE,MAAM,EAAE,IAAI,EAAE,CAAC;QACpC,CAAC;aAAM,CAAC;YACJ,YAAY,GAAG,IAAI,CAAC;QACxB,CAAC;QAED,8BAA8B,CAAC,IAAI,EAAE,YAAY,CAAC,CAAC;QAEnD,OAAO,YAAY,CAAC;IACxB,CAAC,CAAC,CAAC;AACP,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,cAAc,CAC1B,cAA2C,EAC3C,iBAAsC,EACtC,wBAA4C,EAAE,EAC9C,QAA0C,EAC1C,YAAoC;IAEpC,MAAM,4BAA4B,GAAG,qBAAqB,CAAC,GAAG,CAAC,0BAA0B,CAAC,CAAC;IAC3F,MAAM,wBAAwB,GAAG,iBAAiB,EAAE,GAAG,CAAC,0BAA0B,CAAC,CAAC;IAEpF,OAAO,cAAc;SAChB,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,CAAC,EAAE,GAAG,EAAE,OAAO,IAAI,KAAK,QAAQ,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC,GAAG,EAAE,IAAI,EAAE,CAAC,CAAC;SAC1E,MAAM,CAAC,CAAC,EAAE,GAAG,EAAE,EAAE,EAAE;QAChB,MAAM,sBAAsB,GAAG,4BAA4B,CAAC,IAAI,CAAC,CAAC,EAAE,KAAK,EAAE,EAAE,EAAE,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC;QAE5F,IAAI,sBAAsB,EAAE,CAAC;YACzB,YAAY,EAAE,CAAC,GAAG,CAAC,CAAC;QACxB,CAAC;QAED,OAAO,CAAC,sBAAsB,CAAC;IACnC,CAAC,CAAC;SACD,GAAG,CAAC,CAAC,EAAE,GAAG,EAAE,IAAI,EAAE,EAAE,EAAE;QACnB,IAAI,CAAC,wBAAwB,IAAI,CAAC,wBAAwB,CAAC,MAAM,EAAE,CAAC;YAChE,OAAO,IAAI,OAAO,CAAC,OAAO,IAAI,KAAK,QAAQ,CAAC,CAAC,CAAC,EAAE,GAAG,EAAE,IAAI,EAAE,eAAe,EAAE,QAAQ,EAAE,CAAC,CAAC,CAAC,EAAE,GAAG,IAAI,EAAE,CAAC,CAAC;QAC1G,CAAC;QAED,KAAK,MAAM,gBAAgB,IAAI,wBAAwB,EAAE,CAAC;YACtD,MAAM,EAAE,KAAK,EAAE,IAAI,EAAE,MAAM,EAAE,GAAG,oBAAoB,EAAE,GAAG,gBAAgB,CAAC;YAC1E,IAAI,KAAK,CAAC,GAAG,CAAC,EAAE,CAAC;gBACb,MAAM,OAAO,GACT,OAAO,IAAI,KAAK,QAAQ;oBACpB,CAAC,CAAC,EAAE,GAAG,EAAE,IAAI,EAAE,GAAG,oBAAoB,EAAE,eAAe,EAAE,QAAQ,EAAE;oBACnE,CAAC,CAAC,EAAE,GAAG,IAAI,EAAE,GAAG,oBAAoB,EAAE,eAAe,EAAE,QAAQ,EAAE,CAAC;gBAE1E,OAAO,IAAI,OAAO,CAAC,OAAO,CAAC,CAAC;YAChC,CAAC;QACL,CAAC;QAED,oCAAoC;QACpC,YAAY,EAAE,CAAC,GAAG,CAAC,CAAC;QACpB,OAAO,IAAI,CAAC;IAChB,CAAC,CAAC;SACD,MAAM,CAAC,CAAC,OAAO,EAAE,EAAE,CAAC,OAAO,CAAc,CAAC;AACnD,CAAC;AAED,MAAM,UAAU,wBAAwB,CACpC,QAAmB,EACnB,QAA6B,EAC7B,YAAoC;IAEpC,IAAI,CAAC,QAAQ,EAAE,MAAM,EAAE,CAAC;QACpB,OAAO,QAAQ,CAAC;IACpB,CAAC;IAED,MAAM,QAAQ,GAAc,EAAE,CAAC;IAC/B,MAAM,eAAe,GAAG,QAAQ,EAAE,GAAG,CAAC,0BAA0B,CAAC,CAAC;IAElE,KAAK,MAAM,OAAO,IAAI,QAAQ,EAAE,CAAC;QAC7B,MAAM,eAAe,GAAG,eAAe,CAAC,IAAI,CAAC,CAAC,EAAE,KAAK,EAAE,EAAE,EAAE,CAAC,KAAK,CAAC,OAAO,CAAC,GAAG,CAAC,CAAC,CAAC;QAEhF,IAAI,eAAe,KAAK,SAAS,EAAE,CAAC;YAChC,QAAQ,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;QAC3B,CAAC;aAAM,CAAC;YACJ,YAAY,EAAE,CAAC,OAAO,CAAC,GAAG,CAAC,CAAC;QAChC,CAAC;IACL,CAAC;IAED,OAAO,QAAQ,CAAC;AACpB,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,oBAAoB,CAChC,OAAsD,EACtD,UAAuG,EAAE;IAEzG,OAAO,OAAO;SACT,GAAG,CAAC,CAAC,GAAG,EAAE,EAAE,CACT,OAAO,GAAG,KAAK,QAAQ;QACnB,CAAC,CAAC,EAAE,GAAG,EAAE,GAAG,EAAE,eAAe,EAAE,OAAO,CAAC,QAAQ,EAAE;QACjD,CAAC,CAAE,EAAE,GAAG,GAAG,EAAE,eAAe,EAAE,OAAO,CAAC,QAAQ,EAAqB,CAC1E;SACA,MAAM,CAAC,CAAC,EAAE,GAAG,EAAE,EAAE,EAAE;QAChB,IAAI,CAAC;YACD,OAAO,IAAI,GAAG,CAAC,GAAG,EAAE,OAAO,CAAC,OAAO,CAAC,CAAC,IAAI,CAAC;QAC9C,CAAC;QAAC,OAAO,GAAG,EAAE,CAAC;YACX,OAAO,KAAK,CAAC;QACjB,CAAC;IACL,CAAC,CAAC;SACD,GAAG,CAAC,CAAC,cAAc,EAAE,EAAE;QACpB,cAAc,CAAC,GAAG,GAAG,IAAI,GAAG,CAAC,cAAc,CAAC,GAAG,EAAE,OAAO,CAAC,OAAO,CAAC,CAAC,IAAI,CAAC;QACvE,cAAc,CAAC,QAAQ,KAAK,OAAO,CAAC,QAAQ,IAAI,EAAE,CAAC;QAEnD,IAAI,OAAO,OAAO,CAAC,KAAK,KAAK,QAAQ,EAAE,CAAC;YACpC,cAAc,CAAC,QAAQ,GAAG;gBACtB,GAAG,cAAc,CAAC,QAAQ;gBAC1B,KAAK,EAAE,OAAO,CAAC,KAAK;aACvB,CAAC;QACN,CAAC;QAED,IAAI,OAAO,CAAC,cAAc,EAAE,CAAC;YACzB,cAAc,CAAC,cAAc,GAAG,IAAI,CAAC;QACzC,CAAC;QAED,OAAO,cAAc,CAAC;IAC1B,CAAC,CAAC,CAAC;AACX,CAAC;AAED;;GAEG;AACH,SAAS,0BAA0B,CAAC,gBAAkC;IAClE,MAAM,EAAE,MAAM,EAAE,IAAI,EAAE,GAAG,gBAAgB,CAAC;IAC1C,IAAI,KAAK,CAAC;IACV,IAAI,MAAM,EAAE,CAAC;QACT,KAAK,GAAG,CAAC,GAAW,EAAE,EAAE,CAAC,MAAM,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;IAC9C,CAAC;SAAM,IAAI,IAAI,EAAE,CAAC;QACd,MAAM,CAAC,GAAG,IAAI,SAAS,CAAC,IAAI,EAAE,EAAE,MAAM,EAAE,IAAI,EAAE,CAAC,CAAC;QAChD,KAAK,GAAG,CAAC,GAAW,EAAE,EAAE,CAAC,CAAC,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC;IAC1C,CAAC;SAAM,CAAC;QACJ,KAAK,GAAG,GAAG,EAAE,CAAC,KAAK,CAAC;IACxB,CAAC;IACD,OAAO,EAAE,GAAG,gBAAgB,EAAE,KAAK,EAAE,CAAC;AAC1C,CAAC"}
|
package/errors.d.ts.map
DELETED
|
@@ -1 +0,0 @@
|
|
|
1
|
-
{"version":3,"file":"errors.d.ts","sourceRoot":"","sources":["../src/errors.ts"],"names":[],"mappings":"AAAA;;GAEG;AACH,qBAAa,iBAAkB,SAAQ,KAAK;CAAG;AAE/C;;;GAGG;AACH,qBAAa,aAAc,SAAQ,iBAAiB;CAAG;AAEvD;;GAEG;AACH,qBAAa,iBAAkB,SAAQ,aAAa;CAAG;AAEvD;;;;GAIG;AACH,qBAAa,iBAAkB,SAAQ,KAAK;gBAC5B,OAAO,CAAC,EAAE,MAAM;CAG/B;AAED;;;;GAIG;AACH,qBAAa,YAAa,SAAQ,iBAAiB;gBACnC,OAAO,CAAC,EAAE,MAAM;CAG/B;AAED,qBAAa,+BAAgC,SAAQ,KAAK;gBAC1C,OAAO,CAAC,EAAE,MAAM;CAG/B;AAED,qBAAa,kCAAmC,SAAQ,KAAK;gBAC7C,KAAK,EAAE,OAAO,EAAE,OAAO,CAAC,EAAE,YAAY;CAGrD;AAED,qBAAa,2BAA4B,SAAQ,aAAa;gBAC9C,KAAK,EAAE,OAAO,EAAE,OAAO,CAAC,EAAE,YAAY;CAGrD;AAED,qBAAa,mBAAoB,SAAQ,KAAK;gBAC9B,KAAK,EAAE,OAAO,EAAE,OAAO,CAAC,EAAE,YAAY;CAGrD"}
|
package/errors.js.map
DELETED
|
@@ -1 +0,0 @@
|
|
|
1
|
-
{"version":3,"file":"errors.js","sourceRoot":"","sources":["../src/errors.ts"],"names":[],"mappings":"AAAA;;GAEG;AACH,MAAM,OAAO,iBAAkB,SAAQ,KAAK;CAAG;AAE/C;;;GAGG;AACH,MAAM,OAAO,aAAc,SAAQ,iBAAiB;CAAG;AAEvD;;GAEG;AACH,MAAM,OAAO,iBAAkB,SAAQ,aAAa;CAAG;AAEvD;;;;GAIG;AACH,MAAM,OAAO,iBAAkB,SAAQ,KAAK;IACxC,YAAY,OAAgB;QACxB,KAAK,CAAC,OAAO,IAAI,gDAAgD,CAAC,CAAC;IACvE,CAAC;CACJ;AAED;;;;GAIG;AACH,MAAM,OAAO,YAAa,SAAQ,iBAAiB;IAC/C,YAAY,OAAgB;QACxB,KAAK,CAAC,iDAAiD,OAAO,CAAC,CAAC,CAAC,KAAK,OAAO,EAAE,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC;IAC5F,CAAC;CACJ;AAED,MAAM,OAAO,+BAAgC,SAAQ,KAAK;IACtD,YAAY,OAAgB;QACxB,KAAK,CAAC,kEAAkE,OAAO,CAAC,CAAC,CAAC,MAAM,OAAO,EAAE,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC;IAC9G,CAAC;CACJ;AAED,MAAM,OAAO,kCAAmC,SAAQ,KAAK;IACzD,YAAY,KAAc,EAAE,OAAsB;QAC9C,KAAK,CAAC,SAAS,EAAE,EAAE,KAAK,EAAE,KAAK,EAAE,GAAG,OAAO,EAAE,CAAC,CAAC;IACnD,CAAC;CACJ;AAED,MAAM,OAAO,2BAA4B,SAAQ,aAAa;IAC1D,YAAY,KAAc,EAAE,OAAsB;QAC9C,KAAK,CAAC,SAAS,EAAE,EAAE,KAAK,EAAE,KAAK,EAAE,GAAG,OAAO,EAAE,CAAC,CAAC;IACnD,CAAC;CACJ;AAED,MAAM,OAAO,mBAAoB,SAAQ,KAAK;IAC1C,YAAY,KAAc,EAAE,OAAsB;QAC9C,KAAK,CAAC,SAAS,EAAE,EAAE,KAAK,EAAE,KAAK,EAAE,GAAG,OAAO,EAAE,CAAC,CAAC;IACnD,CAAC;CACJ"}
|