@d-zero/beholder 3.0.0 → 3.1.0
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/CHANGELOG.md +6 -0
- package/README.md +26 -0
- package/dist/dom-evaluation.js +20 -152
- package/dist/extract-meta.d.ts +98 -0
- package/dist/extract-meta.js +75 -0
- package/dist/index.d.ts +2 -0
- package/dist/index.js +1 -0
- package/dist/meta/collect-head.d.ts +63 -0
- package/dist/meta/collect-head.js +223 -0
- package/package.json +5 -3
- package/src/dom-evaluation.ts +20 -161
- package/src/extract-meta.spec.ts +247 -0
- package/src/extract-meta.ts +121 -0
- package/src/index.ts +2 -0
- package/src/meta/collect-head.ts +247 -0
- package/tsconfig.tsbuildinfo +1 -1
|
@@ -0,0 +1,223 @@
|
|
|
1
|
+
/**
|
|
2
|
+
* DOM-side raw `<head>` collector.
|
|
3
|
+
*
|
|
4
|
+
* `collectHeadFromDocument` walks a `Document` (Puppeteer page realm or jsdom realm
|
|
5
|
+
* alike) and produces a serializable {@link RawHeadEntry}[] that
|
|
6
|
+
* {@link ../meta/classify.ts | classify} can turn into a typed `Meta`.
|
|
7
|
+
*
|
|
8
|
+
* WHY this function is realm-agnostic:
|
|
9
|
+
*
|
|
10
|
+
* - The Puppeteer path stringifies this function via `Function.prototype.toString`
|
|
11
|
+
* and runs it as a `page.evaluate(string)` expression, so any closure over
|
|
12
|
+
* module-scope bindings would resolve to `undefined` in the browser realm.
|
|
13
|
+
* - The jsdom (Node) path calls it directly with the jsdom `Window`. Because
|
|
14
|
+
* `HTMLLinkElement` (etc.) in jsdom is a *different class instance* from the
|
|
15
|
+
* one in the page realm, `instanceof` only works when the constructor is read
|
|
16
|
+
* from the *passed* `window` rather than from bare globals.
|
|
17
|
+
*
|
|
18
|
+
* Together those constraints dictate that the function MUST:
|
|
19
|
+
*
|
|
20
|
+
* 1. Reference no module-level variables — only its own parameters and inner locals.
|
|
21
|
+
* 2. Take every HTML class constructor (`HTMLBaseElement`, …) from the passed
|
|
22
|
+
* `window` via destructuring instead of relying on ambient globals.
|
|
23
|
+
* 3. Stay in plain ES syntax (no TS-only constructs that need helper imports).
|
|
24
|
+
* @module
|
|
25
|
+
*/
|
|
26
|
+
/**
|
|
27
|
+
* Curated list of `window` globals whose presence indicates that a third-party
|
|
28
|
+
* tag library has been loaded on the page. Surfaced as a single
|
|
29
|
+
* `kind: 'window-global'` entry so that downstream consumers (e.g. tag-detection)
|
|
30
|
+
* can cross-reference the script/iframe signals.
|
|
31
|
+
*
|
|
32
|
+
* Kept here (rather than in `dom-evaluation.ts`) so the Puppeteer path and the
|
|
33
|
+
* jsdom path share one source of truth.
|
|
34
|
+
*/
|
|
35
|
+
export const WINDOW_GLOBALS_TO_CHECK = [
|
|
36
|
+
'dataLayer',
|
|
37
|
+
'gtag',
|
|
38
|
+
'ga',
|
|
39
|
+
'_gaq',
|
|
40
|
+
'fbq',
|
|
41
|
+
'_fbq',
|
|
42
|
+
'clarity',
|
|
43
|
+
'_hjSettings',
|
|
44
|
+
'_hjid',
|
|
45
|
+
'twq',
|
|
46
|
+
'ttq',
|
|
47
|
+
'_linkedin_partner_id',
|
|
48
|
+
'pintrk',
|
|
49
|
+
'amplitude',
|
|
50
|
+
'mixpanel',
|
|
51
|
+
'analytics',
|
|
52
|
+
'heap',
|
|
53
|
+
'posthog',
|
|
54
|
+
'plausible',
|
|
55
|
+
'fathom',
|
|
56
|
+
'_paq',
|
|
57
|
+
's_account',
|
|
58
|
+
's',
|
|
59
|
+
'ym',
|
|
60
|
+
'UET',
|
|
61
|
+
'optimizely',
|
|
62
|
+
'_hsq',
|
|
63
|
+
'Sentry',
|
|
64
|
+
'Intercom',
|
|
65
|
+
'intercomSettings',
|
|
66
|
+
'drift',
|
|
67
|
+
'Tawk_API',
|
|
68
|
+
'zE',
|
|
69
|
+
'OneTrust',
|
|
70
|
+
'Cookiebot',
|
|
71
|
+
'Stripe',
|
|
72
|
+
'grecaptcha',
|
|
73
|
+
];
|
|
74
|
+
/**
|
|
75
|
+
* Walks the given window's `Document` and returns a serializable list of raw
|
|
76
|
+
* head entries.
|
|
77
|
+
*
|
|
78
|
+
* Two realms are supported:
|
|
79
|
+
*
|
|
80
|
+
* - Browser realm (Puppeteer): the function source is `.toString()`'d and run
|
|
81
|
+
* inside the page via `page.evaluate(string)`. Inside the page, `window`
|
|
82
|
+
* resolves to the page's global object, so destructured class constructors
|
|
83
|
+
* match `instanceof` checks against elements returned from `querySelectorAll`.
|
|
84
|
+
* - Node realm (jsdom et al.): the caller passes `dom.window` directly. jsdom's
|
|
85
|
+
* HTML element prototypes are distinct from the host Node's bare globals, so
|
|
86
|
+
* reading the constructors off the passed `window` is what makes `instanceof`
|
|
87
|
+
* succeed.
|
|
88
|
+
*
|
|
89
|
+
* The function MUST NOT close over any module-scope binding — all data it needs
|
|
90
|
+
* is reached through its two parameters.
|
|
91
|
+
* @param window - The window object whose `document` will be inspected. Provides
|
|
92
|
+
* both the DOM tree and the HTML element constructors used for
|
|
93
|
+
* `instanceof` narrowing.
|
|
94
|
+
* @param knownGlobals - Names of `window` properties that, when present,
|
|
95
|
+
* indicate a third-party tag library is loaded. Required
|
|
96
|
+
* (no default) so the Puppeteer-side string-eval path
|
|
97
|
+
* does not have to inline a default value list.
|
|
98
|
+
* @returns Serializable list of raw head entries for {@link ../meta/classify.ts | classify}.
|
|
99
|
+
*/
|
|
100
|
+
export function collectHeadFromDocument(window, knownGlobals) {
|
|
101
|
+
const document = window.document;
|
|
102
|
+
// TypeScript's `Window` interface in lib.dom does not directly expose the
|
|
103
|
+
// HTML element constructors (`HTMLLinkElement`, `HTMLScriptElement`, …)
|
|
104
|
+
// even though every real window object — browser realm AND jsdom realm —
|
|
105
|
+
// carries them at runtime. Widening the type here lets us destructure them
|
|
106
|
+
// uniformly; the runtime values come straight from the passed window, so
|
|
107
|
+
// the cast is purely cosmetic for TS and erased at compile time.
|
|
108
|
+
const w = window;
|
|
109
|
+
const { HTMLBaseElement, HTMLMetaElement, HTMLLinkElement, HTMLScriptElement, HTMLIFrameElement, } = w;
|
|
110
|
+
const entries = [];
|
|
111
|
+
const html = document.documentElement;
|
|
112
|
+
entries.push({
|
|
113
|
+
kind: 'html',
|
|
114
|
+
lang: html.lang || undefined,
|
|
115
|
+
dir: html.dir || undefined,
|
|
116
|
+
xmlns: html.getAttribute('xmlns') ?? undefined,
|
|
117
|
+
prefix: html.getAttribute('prefix') ?? undefined,
|
|
118
|
+
vocab: html.getAttribute('vocab') ?? undefined,
|
|
119
|
+
typeOf: html.getAttribute('typeof') ?? undefined,
|
|
120
|
+
itemscope: html.hasAttribute('itemscope') || undefined,
|
|
121
|
+
itemtype: html.getAttribute('itemtype') ?? undefined,
|
|
122
|
+
amp: html.hasAttribute('amp') || undefined,
|
|
123
|
+
lightning: html.hasAttribute('⚡') || undefined,
|
|
124
|
+
}, { kind: 'title', content: document.title });
|
|
125
|
+
for (const base of document.querySelectorAll('base')) {
|
|
126
|
+
if (!(base instanceof HTMLBaseElement))
|
|
127
|
+
continue;
|
|
128
|
+
entries.push({
|
|
129
|
+
kind: 'base',
|
|
130
|
+
href: base.getAttribute('href') ?? undefined,
|
|
131
|
+
target: base.getAttribute('target') ?? undefined,
|
|
132
|
+
});
|
|
133
|
+
}
|
|
134
|
+
for (const meta of document.querySelectorAll('meta')) {
|
|
135
|
+
if (!(meta instanceof HTMLMetaElement))
|
|
136
|
+
continue;
|
|
137
|
+
const name = meta.getAttribute('name');
|
|
138
|
+
const property = meta.getAttribute('property');
|
|
139
|
+
const httpEquiv = meta.getAttribute('http-equiv');
|
|
140
|
+
const itemprop = meta.getAttribute('itemprop');
|
|
141
|
+
const charset = meta.getAttribute('charset');
|
|
142
|
+
const content = meta.getAttribute('content');
|
|
143
|
+
const media = meta.getAttribute('media');
|
|
144
|
+
entries.push({
|
|
145
|
+
kind: 'meta',
|
|
146
|
+
name: name ? name.toLowerCase() : undefined,
|
|
147
|
+
property: property ? property.toLowerCase() : undefined,
|
|
148
|
+
httpEquiv: httpEquiv ? httpEquiv.toLowerCase() : undefined,
|
|
149
|
+
itemprop: itemprop ?? undefined,
|
|
150
|
+
charset: charset ?? undefined,
|
|
151
|
+
content: content ?? undefined,
|
|
152
|
+
media: media ?? undefined,
|
|
153
|
+
});
|
|
154
|
+
}
|
|
155
|
+
for (const link of document.querySelectorAll('link[href]')) {
|
|
156
|
+
if (!(link instanceof HTMLLinkElement))
|
|
157
|
+
continue;
|
|
158
|
+
const relRaw = link.getAttribute('rel') ?? '';
|
|
159
|
+
const rel = relRaw.toLowerCase().split(/\s+/u).filter(Boolean);
|
|
160
|
+
entries.push({
|
|
161
|
+
kind: 'link',
|
|
162
|
+
rel,
|
|
163
|
+
href: link.getAttribute('href') ?? '',
|
|
164
|
+
type: link.getAttribute('type') ?? undefined,
|
|
165
|
+
media: link.getAttribute('media') ?? undefined,
|
|
166
|
+
sizes: link.getAttribute('sizes') ?? undefined,
|
|
167
|
+
title: link.getAttribute('title') ?? undefined,
|
|
168
|
+
hreflang: link.getAttribute('hreflang') ?? undefined,
|
|
169
|
+
as: link.getAttribute('as') ?? undefined,
|
|
170
|
+
crossorigin: link.getAttribute('crossorigin') ?? undefined,
|
|
171
|
+
color: link.getAttribute('color') ?? undefined,
|
|
172
|
+
blocking: link.getAttribute('blocking') ?? undefined,
|
|
173
|
+
imagesrcset: link.getAttribute('imagesrcset') ?? undefined,
|
|
174
|
+
});
|
|
175
|
+
}
|
|
176
|
+
const STRUCTURED_TYPES = new Set([
|
|
177
|
+
'application/ld+json',
|
|
178
|
+
'speculationrules',
|
|
179
|
+
'application/json+oembed',
|
|
180
|
+
'application/xml+oembed',
|
|
181
|
+
]);
|
|
182
|
+
for (const script of document.querySelectorAll('script[type]')) {
|
|
183
|
+
if (!(script instanceof HTMLScriptElement))
|
|
184
|
+
continue;
|
|
185
|
+
const scriptType = (script.getAttribute('type') ?? '').toLowerCase();
|
|
186
|
+
if (!STRUCTURED_TYPES.has(scriptType))
|
|
187
|
+
continue;
|
|
188
|
+
const src = script.getAttribute('src') ?? undefined;
|
|
189
|
+
const text = script.textContent ?? '';
|
|
190
|
+
const inHead = !!script.closest('head');
|
|
191
|
+
const inNoscript = !!script.closest('noscript');
|
|
192
|
+
const location = inHead ? 'head' : inNoscript ? 'noscript' : 'body';
|
|
193
|
+
entries.push({
|
|
194
|
+
kind: 'script',
|
|
195
|
+
scriptType,
|
|
196
|
+
content: text || undefined,
|
|
197
|
+
src,
|
|
198
|
+
location,
|
|
199
|
+
});
|
|
200
|
+
}
|
|
201
|
+
for (const iframe of document.querySelectorAll('iframe[src]')) {
|
|
202
|
+
if (!(iframe instanceof HTMLIFrameElement))
|
|
203
|
+
continue;
|
|
204
|
+
const src = iframe.getAttribute('src') ?? '';
|
|
205
|
+
if (!src)
|
|
206
|
+
continue;
|
|
207
|
+
const inHead = !!iframe.closest('head');
|
|
208
|
+
const inNoscript = !!iframe.closest('noscript');
|
|
209
|
+
const location = inHead ? 'head' : inNoscript ? 'noscript' : 'body';
|
|
210
|
+
entries.push({ kind: 'iframe', src, location });
|
|
211
|
+
}
|
|
212
|
+
const win = window;
|
|
213
|
+
const presentGlobals = [];
|
|
214
|
+
for (const name of knownGlobals) {
|
|
215
|
+
if (win[name] !== undefined) {
|
|
216
|
+
presentGlobals.push(name);
|
|
217
|
+
}
|
|
218
|
+
}
|
|
219
|
+
if (presentGlobals.length > 0) {
|
|
220
|
+
entries.push({ kind: 'window-global', names: presentGlobals });
|
|
221
|
+
}
|
|
222
|
+
return entries;
|
|
223
|
+
}
|
package/package.json
CHANGED
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
{
|
|
2
2
|
"name": "@d-zero/beholder",
|
|
3
|
-
"version": "3.
|
|
3
|
+
"version": "3.1.0",
|
|
4
4
|
"description": "Page-level scraper for web crawling and auditing",
|
|
5
5
|
"author": "D-ZERO",
|
|
6
6
|
"license": "MIT",
|
|
@@ -27,12 +27,14 @@
|
|
|
27
27
|
"simple-wappalyzer": "1.1.99"
|
|
28
28
|
},
|
|
29
29
|
"devDependencies": {
|
|
30
|
-
"@types/debug": "4.1.12"
|
|
30
|
+
"@types/debug": "4.1.12",
|
|
31
|
+
"@types/jsdom": "28.0.3",
|
|
32
|
+
"jsdom": "29.1.1"
|
|
31
33
|
},
|
|
32
34
|
"repository": {
|
|
33
35
|
"type": "git",
|
|
34
36
|
"url": "https://github.com/d-zero-dev/tools.git",
|
|
35
37
|
"directory": "packages/@d-zero/beholder"
|
|
36
38
|
},
|
|
37
|
-
"gitHead": "
|
|
39
|
+
"gitHead": "e69344a9d4d45b0ec0ee942f920b84bbd0fb77ae"
|
|
38
40
|
}
|
package/src/dom-evaluation.ts
CHANGED
|
@@ -22,6 +22,7 @@ import { raceWithTimeout } from '@d-zero/shared/race-with-timeout';
|
|
|
22
22
|
|
|
23
23
|
import { domDetailsLog, domLog } from './debug.js';
|
|
24
24
|
import { classify, emptyMeta } from './meta/classify.js';
|
|
25
|
+
import { WINDOW_GLOBALS_TO_CHECK, collectHeadFromDocument } from './meta/collect-head.js';
|
|
25
26
|
import { detectTags } from './meta/tag-detection.js';
|
|
26
27
|
import { parseUrl } from './parse-url.js';
|
|
27
28
|
|
|
@@ -515,46 +516,6 @@ export type GetMetaContext = {
|
|
|
515
516
|
readonly includeRaw?: boolean;
|
|
516
517
|
};
|
|
517
518
|
|
|
518
|
-
const WINDOW_GLOBALS_TO_CHECK: readonly string[] = [
|
|
519
|
-
'dataLayer',
|
|
520
|
-
'gtag',
|
|
521
|
-
'ga',
|
|
522
|
-
'_gaq',
|
|
523
|
-
'fbq',
|
|
524
|
-
'_fbq',
|
|
525
|
-
'clarity',
|
|
526
|
-
'_hjSettings',
|
|
527
|
-
'_hjid',
|
|
528
|
-
'twq',
|
|
529
|
-
'ttq',
|
|
530
|
-
'_linkedin_partner_id',
|
|
531
|
-
'pintrk',
|
|
532
|
-
'amplitude',
|
|
533
|
-
'mixpanel',
|
|
534
|
-
'analytics',
|
|
535
|
-
'heap',
|
|
536
|
-
'posthog',
|
|
537
|
-
'plausible',
|
|
538
|
-
'fathom',
|
|
539
|
-
'_paq',
|
|
540
|
-
's_account',
|
|
541
|
-
's',
|
|
542
|
-
'ym',
|
|
543
|
-
'UET',
|
|
544
|
-
'optimizely',
|
|
545
|
-
'_hsq',
|
|
546
|
-
'Sentry',
|
|
547
|
-
'Intercom',
|
|
548
|
-
'intercomSettings',
|
|
549
|
-
'drift',
|
|
550
|
-
'Tawk_API',
|
|
551
|
-
'zE',
|
|
552
|
-
'OneTrust',
|
|
553
|
-
'Cookiebot',
|
|
554
|
-
'Stripe',
|
|
555
|
-
'grecaptcha',
|
|
556
|
-
];
|
|
557
|
-
|
|
558
519
|
/**
|
|
559
520
|
* Extracts comprehensive metadata from the page.
|
|
560
521
|
*
|
|
@@ -639,129 +600,27 @@ async function runGetMeta(page: Page, context: GetMetaContext): Promise<Meta | n
|
|
|
639
600
|
}
|
|
640
601
|
|
|
641
602
|
/**
|
|
603
|
+
* Collects raw `<head>` entries from a Puppeteer page by injecting
|
|
604
|
+
* {@link collectHeadFromDocument} into the page realm.
|
|
642
605
|
*
|
|
643
|
-
*
|
|
606
|
+
* WHY string-eval instead of `page.evaluate(fn, args)`: the shared
|
|
607
|
+
* implementation lives in this module (`collectHeadFromDocument`), and a
|
|
608
|
+
* `page.evaluate(() => collectHeadFromDocument(window, …))` wrapper cannot
|
|
609
|
+
* reach that module-scope binding inside the page realm — only the wrapper's
|
|
610
|
+
* own source crosses the CDP boundary. Serializing the implementation via
|
|
611
|
+
* `Function.prototype.toString` and invoking it through
|
|
612
|
+
* `page.evaluate(string)` is what keeps the Puppeteer path and the
|
|
613
|
+
* jsdom path on one source of truth.
|
|
614
|
+
*
|
|
615
|
+
* The same {@link collectHeadFromDocument} function is also exposed via
|
|
616
|
+
* {@link ../extract-meta.ts | extractMetaFromDocument} for jsdom/Node callers,
|
|
617
|
+
* so the two paths cannot drift apart.
|
|
618
|
+
* @param page - The Puppeteer page whose document will be inspected.
|
|
644
619
|
*/
|
|
645
620
|
async function collectHeadOnPage(page: Page): Promise<RawHeadEntry[]> {
|
|
646
|
-
const
|
|
647
|
-
|
|
648
|
-
|
|
649
|
-
|
|
650
|
-
type Out = unknown;
|
|
651
|
-
const entries: Out[] = [];
|
|
652
|
-
|
|
653
|
-
const html = document.documentElement;
|
|
654
|
-
entries.push(
|
|
655
|
-
{
|
|
656
|
-
kind: 'html',
|
|
657
|
-
lang: html.lang || undefined,
|
|
658
|
-
dir: html.dir || undefined,
|
|
659
|
-
xmlns: html.getAttribute('xmlns') ?? undefined,
|
|
660
|
-
prefix: html.getAttribute('prefix') ?? undefined,
|
|
661
|
-
vocab: html.getAttribute('vocab') ?? undefined,
|
|
662
|
-
typeOf: html.getAttribute('typeof') ?? undefined,
|
|
663
|
-
itemscope: html.hasAttribute('itemscope') || undefined,
|
|
664
|
-
itemtype: html.getAttribute('itemtype') ?? undefined,
|
|
665
|
-
amp: html.hasAttribute('amp') || undefined,
|
|
666
|
-
lightning: html.hasAttribute('⚡') || undefined,
|
|
667
|
-
},
|
|
668
|
-
{ kind: 'title', content: document.title },
|
|
669
|
-
);
|
|
670
|
-
|
|
671
|
-
for (const base of document.querySelectorAll('base')) {
|
|
672
|
-
if (!(base instanceof HTMLBaseElement)) continue;
|
|
673
|
-
entries.push({
|
|
674
|
-
kind: 'base',
|
|
675
|
-
href: base.getAttribute('href') ?? undefined,
|
|
676
|
-
target: base.getAttribute('target') ?? undefined,
|
|
677
|
-
});
|
|
678
|
-
}
|
|
679
|
-
|
|
680
|
-
for (const meta of document.querySelectorAll('meta')) {
|
|
681
|
-
if (!(meta instanceof HTMLMetaElement)) continue;
|
|
682
|
-
const name = meta.getAttribute('name');
|
|
683
|
-
const property = meta.getAttribute('property');
|
|
684
|
-
const httpEquiv = meta.getAttribute('http-equiv');
|
|
685
|
-
const itemprop = meta.getAttribute('itemprop');
|
|
686
|
-
const charset = meta.getAttribute('charset');
|
|
687
|
-
const content = meta.getAttribute('content');
|
|
688
|
-
const media = meta.getAttribute('media');
|
|
689
|
-
entries.push({
|
|
690
|
-
kind: 'meta',
|
|
691
|
-
name: name ? name.toLowerCase() : undefined,
|
|
692
|
-
property: property ? property.toLowerCase() : undefined,
|
|
693
|
-
httpEquiv: httpEquiv ? httpEquiv.toLowerCase() : undefined,
|
|
694
|
-
itemprop: itemprop ?? undefined,
|
|
695
|
-
charset: charset ?? undefined,
|
|
696
|
-
content: content ?? undefined,
|
|
697
|
-
media: media ?? undefined,
|
|
698
|
-
});
|
|
699
|
-
}
|
|
700
|
-
|
|
701
|
-
for (const link of document.querySelectorAll('link[href]')) {
|
|
702
|
-
if (!(link instanceof HTMLLinkElement)) continue;
|
|
703
|
-
const relRaw = link.getAttribute('rel') ?? '';
|
|
704
|
-
const rel = relRaw.toLowerCase().split(/\s+/u).filter(Boolean);
|
|
705
|
-
entries.push({
|
|
706
|
-
kind: 'link',
|
|
707
|
-
rel,
|
|
708
|
-
href: link.getAttribute('href') ?? '',
|
|
709
|
-
type: link.getAttribute('type') ?? undefined,
|
|
710
|
-
media: link.getAttribute('media') ?? undefined,
|
|
711
|
-
sizes: link.getAttribute('sizes') ?? undefined,
|
|
712
|
-
title: link.getAttribute('title') ?? undefined,
|
|
713
|
-
hreflang: link.getAttribute('hreflang') ?? undefined,
|
|
714
|
-
as: link.getAttribute('as') ?? undefined,
|
|
715
|
-
crossorigin: link.getAttribute('crossorigin') ?? undefined,
|
|
716
|
-
color: link.getAttribute('color') ?? undefined,
|
|
717
|
-
blocking: link.getAttribute('blocking') ?? undefined,
|
|
718
|
-
imagesrcset: link.getAttribute('imagesrcset') ?? undefined,
|
|
719
|
-
});
|
|
720
|
-
}
|
|
721
|
-
|
|
722
|
-
const STRUCTURED_TYPES = new Set([
|
|
723
|
-
'application/ld+json',
|
|
724
|
-
'speculationrules',
|
|
725
|
-
'application/json+oembed',
|
|
726
|
-
'application/xml+oembed',
|
|
727
|
-
]);
|
|
728
|
-
for (const script of document.querySelectorAll('script[type]')) {
|
|
729
|
-
if (!(script instanceof HTMLScriptElement)) continue;
|
|
730
|
-
const scriptType = (script.getAttribute('type') ?? '').toLowerCase();
|
|
731
|
-
if (!STRUCTURED_TYPES.has(scriptType)) continue;
|
|
732
|
-
const src = script.getAttribute('src') ?? undefined;
|
|
733
|
-
const text = script.textContent ?? '';
|
|
734
|
-
const inHead = !!script.closest('head');
|
|
735
|
-
const inNoscript = !!script.closest('noscript');
|
|
736
|
-
const location = inHead ? 'head' : inNoscript ? 'noscript' : 'body';
|
|
737
|
-
entries.push({
|
|
738
|
-
kind: 'script',
|
|
739
|
-
scriptType,
|
|
740
|
-
content: text || undefined,
|
|
741
|
-
src,
|
|
742
|
-
location,
|
|
743
|
-
});
|
|
744
|
-
}
|
|
745
|
-
|
|
746
|
-
for (const iframe of document.querySelectorAll('iframe[src]')) {
|
|
747
|
-
if (!(iframe instanceof HTMLIFrameElement)) continue;
|
|
748
|
-
const src = iframe.getAttribute('src') ?? '';
|
|
749
|
-
if (!src) continue;
|
|
750
|
-
const inHead = !!iframe.closest('head');
|
|
751
|
-
const inNoscript = !!iframe.closest('noscript');
|
|
752
|
-
const location = inHead ? 'head' : inNoscript ? 'noscript' : 'body';
|
|
753
|
-
entries.push({ kind: 'iframe', src, location });
|
|
754
|
-
}
|
|
755
|
-
|
|
756
|
-
const win = window as unknown as Record<string, unknown>;
|
|
757
|
-
const presentGlobals = knownGlobals.filter((name) => win[name] !== undefined);
|
|
758
|
-
if (presentGlobals.length > 0) {
|
|
759
|
-
entries.push({ kind: 'window-global', names: presentGlobals });
|
|
760
|
-
}
|
|
761
|
-
|
|
762
|
-
return entries;
|
|
763
|
-
}, WINDOW_GLOBALS_TO_CHECK)
|
|
764
|
-
.catch(() => [] as unknown[]);
|
|
765
|
-
|
|
621
|
+
const fnSource = collectHeadFromDocument.toString();
|
|
622
|
+
const globalsLiteral = JSON.stringify(WINDOW_GLOBALS_TO_CHECK);
|
|
623
|
+
const expr = `(${fnSource})(window, ${globalsLiteral})`;
|
|
624
|
+
const raw = await page.evaluate(expr).catch(() => [] as unknown[]);
|
|
766
625
|
return raw as RawHeadEntry[];
|
|
767
626
|
}
|