pagerts 1.5.1 โ 1.5.4
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 +31 -9
- package/bin/main.js +64 -34
- package/bin/main.js.map +2 -2
- package/package.json +1 -1
package/README.md
CHANGED
|
@@ -14,6 +14,8 @@ PagerTS is a secure, modern command-line utility that transforms URLs into struc
|
|
|
14
14
|
- โก **Fast**: Efficient parsing with LinkeDOM and concurrent request handling
|
|
15
15
|
- ๐งช **Well-Tested**: Comprehensive test coverage with Jest
|
|
16
16
|
- ๐ฆ **Easy to Use**: Simple CLI interface with sensible defaults
|
|
17
|
+
- ๐๏ธ **Local File Support**: bare `pagerts` parses local file paths and `file:///...` inputs
|
|
18
|
+
- ๐งญ **Request Header Override**: Optional `--user-agent` flag for sites that behave differently by client
|
|
17
19
|
|
|
18
20
|
## Installation
|
|
19
21
|
|
|
@@ -21,13 +23,13 @@ PagerTS is a secure, modern command-line utility that transforms URLs into struc
|
|
|
21
23
|
|
|
22
24
|
```bash
|
|
23
25
|
npm install -g pagerts
|
|
24
|
-
pagerts
|
|
26
|
+
pagerts ./page.html
|
|
25
27
|
```
|
|
26
28
|
|
|
27
29
|
### Using npx (No Installation Required)
|
|
28
30
|
|
|
29
31
|
```bash
|
|
30
|
-
npx pagerts
|
|
32
|
+
npx pagerts ./page.html
|
|
31
33
|
```
|
|
32
34
|
|
|
33
35
|
### From Source
|
|
@@ -44,22 +46,34 @@ npm link
|
|
|
44
46
|
|
|
45
47
|
### Basic Usage
|
|
46
48
|
|
|
47
|
-
Extract resources from a
|
|
49
|
+
Extract resources from a local HTML file path:
|
|
48
50
|
|
|
49
51
|
```bash
|
|
50
|
-
pagerts
|
|
52
|
+
pagerts ./page.html
|
|
51
53
|
```
|
|
52
54
|
|
|
53
|
-
Extract from
|
|
55
|
+
Extract resources from a local file URL:
|
|
54
56
|
|
|
55
57
|
```bash
|
|
56
|
-
pagerts
|
|
58
|
+
pagerts file:///path/to/file.html
|
|
57
59
|
```
|
|
58
60
|
|
|
59
|
-
|
|
61
|
+
Fetch resources from a remote URL:
|
|
60
62
|
|
|
61
63
|
```bash
|
|
62
|
-
pagerts
|
|
64
|
+
pagerts fetch https://website.com
|
|
65
|
+
```
|
|
66
|
+
|
|
67
|
+
Override the HTTP user-agent for remote fetches:
|
|
68
|
+
|
|
69
|
+
```bash
|
|
70
|
+
pagerts fetch --user-agent "Mozilla/5.0 (X11; Linux x86_64; rv:139.0) Gecko/20100101 Firefox/139.0" https://example.com
|
|
71
|
+
```
|
|
72
|
+
|
|
73
|
+
Fetch from multiple remote URLs:
|
|
74
|
+
|
|
75
|
+
```bash
|
|
76
|
+
pagerts fetch https://example.com https://example.org
|
|
63
77
|
```
|
|
64
78
|
|
|
65
79
|
### Output Format
|
|
@@ -98,7 +112,8 @@ PagerTS takes security seriously. See [SECURITY.md](./SECURITY.md) for:
|
|
|
98
112
|
|
|
99
113
|
### Built-in Security Features
|
|
100
114
|
|
|
101
|
-
- โ
URL validation (only allows `http
|
|
115
|
+
- โ
URL validation for remote fetches (only allows `http://` and `https://`)
|
|
116
|
+
- โ
Local filesystem parsing through plain paths and `file://` inputs on the root command
|
|
102
117
|
- โ
Input sanitization to prevent XSS attacks
|
|
103
118
|
- โ
Rate limiting (50 requests/minute by default)
|
|
104
119
|
- โ
Request timeouts to prevent hanging
|
|
@@ -226,6 +241,13 @@ This project is licensed under the MIT License - see the [LICENSE](./LICENSE) fi
|
|
|
226
241
|
|
|
227
242
|
## Changelog
|
|
228
243
|
|
|
244
|
+
### v1.5.3
|
|
245
|
+
|
|
246
|
+
- Added `--user-agent` support to the `fetch` command so callers can override the HTTP User-Agent header for remote requests.
|
|
247
|
+
- Made the root `pagerts` command parse local file paths and `file:///` inputs directly, while keeping `fetch` remote-only.
|
|
248
|
+
- Improved CLI/runtime compatibility for locally resolved entrypoints and packaged builds.
|
|
249
|
+
- Updated focused tests to cover the new file-protocol validation and user-agent override behavior.
|
|
250
|
+
|
|
229
251
|
### v0.3.0 -> v1.4.3 summary
|
|
230
252
|
|
|
231
253
|
Key changes in this range:
|
package/bin/main.js
CHANGED
|
@@ -5,6 +5,7 @@ import { Command, createArgument, Option } from "commander";
|
|
|
5
5
|
import { createRequire } from "node:module";
|
|
6
6
|
import { resolve } from "node:path";
|
|
7
7
|
import { fileURLToPath } from "node:url";
|
|
8
|
+
import { realpathSync } from "node:fs";
|
|
8
9
|
|
|
9
10
|
// src/extractors/AbstractExtractor.ts
|
|
10
11
|
var AbstractExtractor = class {
|
|
@@ -81,9 +82,11 @@ import { parseHTML } from "linkedom";
|
|
|
81
82
|
var PageFetcher = class {
|
|
82
83
|
timeout;
|
|
83
84
|
maxRetries;
|
|
84
|
-
|
|
85
|
+
userAgent;
|
|
86
|
+
constructor(timeout = 1e4, maxRetries = 2, userAgent) {
|
|
85
87
|
this.timeout = timeout;
|
|
86
88
|
this.maxRetries = maxRetries;
|
|
89
|
+
this.userAgent = userAgent;
|
|
87
90
|
}
|
|
88
91
|
buildDOMResult(html, url) {
|
|
89
92
|
const { document } = parseHTML(html);
|
|
@@ -105,13 +108,16 @@ var PageFetcher = class {
|
|
|
105
108
|
controller.abort(new Error("Request timeout"));
|
|
106
109
|
}, this.timeout);
|
|
107
110
|
}
|
|
108
|
-
const
|
|
109
|
-
|
|
110
|
-
|
|
111
|
-
|
|
112
|
-
|
|
113
|
-
|
|
114
|
-
|
|
111
|
+
const headers = this.userAgent ? { "user-agent": this.userAgent } : void 0;
|
|
112
|
+
const content = await fetch(url, { headers, signal: controller.signal }).then(
|
|
113
|
+
async (response) => {
|
|
114
|
+
const buffer = await response.arrayBuffer();
|
|
115
|
+
const contentType = response.headers.get("content-type") ?? "";
|
|
116
|
+
const charsetMatch = /charset=([^\s;]+)/i.exec(contentType);
|
|
117
|
+
const html = this.decodeHtml(buffer, charsetMatch?.[1] ?? "utf-8");
|
|
118
|
+
return this.buildDOMResult(html, url);
|
|
119
|
+
}
|
|
120
|
+
);
|
|
115
121
|
return { url, content };
|
|
116
122
|
} catch (error) {
|
|
117
123
|
const abortTimeout = error instanceof Error && error.name === "AbortError";
|
|
@@ -276,21 +282,66 @@ async function buildPageMetadata(responses) {
|
|
|
276
282
|
}
|
|
277
283
|
return pageMetadatas;
|
|
278
284
|
}
|
|
285
|
+
function normalizeLocalPath(value) {
|
|
286
|
+
if (value.startsWith("file://")) {
|
|
287
|
+
return fileURLToPath(value);
|
|
288
|
+
}
|
|
289
|
+
return value;
|
|
290
|
+
}
|
|
291
|
+
async function runFileCommand(paths, options) {
|
|
292
|
+
if (options.failsafe && paths.length > MAX_FILES_FAILSAFE) {
|
|
293
|
+
console.error(
|
|
294
|
+
`
|
|
295
|
+
\u274C ${paths.length} files specified exceeds the safety limit of ${MAX_FILES_FAILSAFE}.`
|
|
296
|
+
);
|
|
297
|
+
console.error(` Pass --no-failsafe to bypass this check and process all files.`);
|
|
298
|
+
process.exit(1);
|
|
299
|
+
}
|
|
300
|
+
if (!options.failsafe && paths.length > MAX_FILES_FAILSAFE) {
|
|
301
|
+
console.error(
|
|
302
|
+
`
|
|
303
|
+
\u26A0\uFE0F Failsafe bypassed: processing ${paths.length} files (limit is ${MAX_FILES_FAILSAFE}).`
|
|
304
|
+
);
|
|
305
|
+
}
|
|
306
|
+
console.error(`
|
|
307
|
+
\u2705 Processing ${paths.length} file(s)...`);
|
|
308
|
+
const fileFetcher = new FileFetcher();
|
|
309
|
+
const normalizedPaths = paths.map((pathValue) => normalizeLocalPath(pathValue));
|
|
310
|
+
const responses = await fileFetcher.fetchAll(normalizedPaths);
|
|
311
|
+
const pageMetadatas = await buildPageMetadata(
|
|
312
|
+
responses.map(({ path, content, error }) => ({ path, content, error }))
|
|
313
|
+
);
|
|
314
|
+
await printer.print(...pageMetadatas);
|
|
315
|
+
}
|
|
279
316
|
function isCliEntrypoint() {
|
|
280
317
|
const invokedPath = process.argv[1];
|
|
281
318
|
if (!invokedPath) {
|
|
282
319
|
return false;
|
|
283
320
|
}
|
|
284
|
-
|
|
321
|
+
try {
|
|
322
|
+
return realpathSync(fileURLToPath(import.meta.url)) === realpathSync(resolve(invokedPath));
|
|
323
|
+
} catch {
|
|
324
|
+
return false;
|
|
325
|
+
}
|
|
285
326
|
}
|
|
286
327
|
async function runCli(argv = process.argv) {
|
|
287
328
|
program.name(name).version(version, "-v, --version").description(description);
|
|
288
|
-
program.
|
|
329
|
+
program.addArgument(fileArg).addOption(
|
|
330
|
+
new Option("--no-failsafe", `bypass the ${MAX_FILES_FAILSAFE}-file limit safety check`)
|
|
331
|
+
).action(async (paths, options) => {
|
|
332
|
+
try {
|
|
333
|
+
await runFileCommand(paths, options);
|
|
334
|
+
} catch (error) {
|
|
335
|
+
console.error("\n\u274C An error occurred:", error instanceof Error ? error.message : error);
|
|
336
|
+
process.exit(1);
|
|
337
|
+
}
|
|
338
|
+
});
|
|
339
|
+
program.command("fetch").description("fetch and extract resources from remote URL(s)").addArgument(urlArg).addOption(
|
|
289
340
|
new Option(
|
|
290
341
|
"--watch",
|
|
291
342
|
"keep running: SIGWINCH re-fetches after resize, Ctrl-D releases in-flight requests, Ctrl-C exits"
|
|
292
343
|
)
|
|
293
|
-
).action(async (urls, options) => {
|
|
344
|
+
).addOption(new Option("-A, --user-agent <value>", "override the HTTP User-Agent header")).action(async (urls, options) => {
|
|
294
345
|
try {
|
|
295
346
|
const { validUrls, errors } = validateUrls(urls);
|
|
296
347
|
if (errors.length > 0) {
|
|
@@ -305,7 +356,7 @@ async function runCli(argv = process.argv) {
|
|
|
305
356
|
}
|
|
306
357
|
console.error(`
|
|
307
358
|
\u2705 Processing ${validUrls.length} valid URL(s)...`);
|
|
308
|
-
const pageFetcher = new PageFetcher(options.watch ? 0 : 1e4, 2);
|
|
359
|
+
const pageFetcher = new PageFetcher(options.watch ? 0 : 1e4, 2, options.userAgent);
|
|
309
360
|
const execute = async () => {
|
|
310
361
|
const responses = await pageFetcher.fetchAll(validUrls);
|
|
311
362
|
const pageMetadatas = await buildPageMetadata(responses);
|
|
@@ -342,28 +393,7 @@ async function runCli(argv = process.argv) {
|
|
|
342
393
|
new Option("--no-failsafe", `bypass the ${MAX_FILES_FAILSAFE}-file limit safety check`)
|
|
343
394
|
).action(async (paths, options) => {
|
|
344
395
|
try {
|
|
345
|
-
|
|
346
|
-
console.error(
|
|
347
|
-
`
|
|
348
|
-
\u274C ${paths.length} files specified exceeds the safety limit of ${MAX_FILES_FAILSAFE}.`
|
|
349
|
-
);
|
|
350
|
-
console.error(` Pass --no-failsafe to bypass this check and process all files.`);
|
|
351
|
-
process.exit(1);
|
|
352
|
-
}
|
|
353
|
-
if (!options.failsafe && paths.length > MAX_FILES_FAILSAFE) {
|
|
354
|
-
console.error(
|
|
355
|
-
`
|
|
356
|
-
\u26A0\uFE0F Failsafe bypassed: processing ${paths.length} files (limit is ${MAX_FILES_FAILSAFE}).`
|
|
357
|
-
);
|
|
358
|
-
}
|
|
359
|
-
console.error(`
|
|
360
|
-
\u2705 Processing ${paths.length} file(s)...`);
|
|
361
|
-
const fileFetcher = new FileFetcher();
|
|
362
|
-
const responses = await fileFetcher.fetchAll(paths);
|
|
363
|
-
const pageMetadatas = await buildPageMetadata(
|
|
364
|
-
responses.map(({ path, content, error }) => ({ path, content, error }))
|
|
365
|
-
);
|
|
366
|
-
await printer.print(...pageMetadatas);
|
|
396
|
+
await runFileCommand(paths, options);
|
|
367
397
|
} catch (error) {
|
|
368
398
|
console.error("\n\u274C An error occurred:", error instanceof Error ? error.message : error);
|
|
369
399
|
process.exit(1);
|
package/bin/main.js.map
CHANGED
|
@@ -1,7 +1,7 @@
|
|
|
1
1
|
{
|
|
2
2
|
"version": 3,
|
|
3
3
|
"sources": ["../src/main.ts", "../src/extractors/AbstractExtractor.ts", "../src/extractors/PageExtractor.ts", "../src/resource.ts", "../src/extractors/ResourceExtractor.ts", "../src/page/PageFetcher.ts", "../src/page/FileFetcher.ts", "../src/printers/AbstractResourcePrinter.ts", "../src/printers/JSONStylePrinter.ts", "../src/security.ts"],
|
|
4
|
-
"sourcesContent": ["#!/usr/bin/env node\nimport { Command, createArgument, Option } from 'commander';\nimport { createRequire } from 'node:module';\nimport { resolve } from 'node:path';\nimport { fileURLToPath } from 'node:url';\n\nimport { PageExtractor, ResourceExtractor } from './extractors/index.js';\nimport { FileFetcher, MAX_FILES_FAILSAFE, PageFetcher, type PageMetadata } from './page/index.js';\nimport { JSONStylePrinter } from './printers/index.js';\nimport { validateUrls } from './security.js';\n\nconst require = createRequire(import.meta.url);\nconst pkg = require('../package.json') as {\n description: string;\n name: string;\n version: string;\n};\n\nconst { description, name, version } = pkg;\n\nconst program = new Command();\n\nconst urlArg = createArgument('<url...>', 'remote https://URL to extract from');\nconst fileArg = createArgument('<paths...>', 'local file paths to extract from');\n\n// Shared extractor instances.\nconst pageExtractor = new PageExtractor();\nconst resourceExtractor = new ResourceExtractor(['a', 'meta', 'link', 'embed', 'script']);\nconst printer = new JSONStylePrinter();\n\nasync function buildPageMetadata(\n responses: Array<{\n url?: string;\n path?: string;\n content?: import('./page/index.js').DOMResult;\n error?: string;\n }>\n): Promise<PageMetadata[]> {\n const pageMetadatas: PageMetadata[] = [];\n\n for (const { content, url: responseUrl, path, error } of responses) {\n const resolvedUrl = responseUrl ?? path ?? '';\n const resources =\n error !== undefined || !content ? [] : await resourceExtractor.extract(content);\n const descriptor =\n error !== undefined || !content\n ? { url: resolvedUrl, error: error ?? 'Unknown error', resources }\n : await pageExtractor.extract(content);\n pageMetadatas.push({ ...descriptor, resources });\n }\n\n return pageMetadatas;\n}\n\nfunction isCliEntrypoint(): boolean {\n const invokedPath = process.argv[1];\n if (!invokedPath) {\n return false;\n }\n\n return fileURLToPath(import.meta.url) === resolve(invokedPath);\n}\n\nexport async function runCli(argv: string[] = process.argv): Promise<void> {\n program.name(name).version(version, '-v, --version').description(description);\n\n // \u2500\u2500 fetch subcommand (default remote URL mode) \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n program\n .command('fetch', { isDefault: true })\n .description('fetch and extract resources from remote URL(s)')\n .addArgument(urlArg)\n .addOption(\n new Option(\n '--watch',\n 'keep running: SIGWINCH re-fetches after resize, Ctrl-D releases in-flight requests, Ctrl-C exits'\n )\n )\n .action(async (urls: string[], options: { watch: boolean }) => {\n try {\n const { validUrls, errors } = validateUrls(urls);\n\n if (errors.length > 0) {\n console.error('\\n\u274C URL Validation Errors:');\n errors.forEach(({ url: invalidUrl, error }) => {\n console.error(` - ${invalidUrl}: ${error}`);\n });\n }\n\n if (validUrls.length === 0) {\n console.error('\\n\u274C No valid URLs to process. Exiting.');\n process.exit(1);\n }\n\n console.error(`\\n\u2705 Processing ${validUrls.length} valid URL(s)...`);\n\n const pageFetcher = new PageFetcher(options.watch ? 0 : 10000, 2);\n\n const execute = async (): Promise<void> => {\n const responses = await pageFetcher.fetchAll(validUrls);\n const pageMetadatas = await buildPageMetadata(responses);\n await printer.print(...pageMetadatas);\n };\n\n if (options.watch) {\n process.stdin.resume();\n process.on('SIGINT', () => process.exit(0));\n\n let activeExecution: Promise<void> | null = null;\n process.stdin.on('end', () => {\n activeExecution = null;\n });\n\n let winchTimer: ReturnType<typeof setTimeout> | null = null;\n process.on('SIGWINCH', () => {\n if (winchTimer !== null) clearTimeout(winchTimer);\n winchTimer = setTimeout(() => {\n winchTimer = null;\n activeExecution = execute().catch((err: unknown) => {\n console.error('\\n\u274C An error occurred:', err instanceof Error ? err.message : err);\n });\n }, 150);\n });\n\n activeExecution = execute();\n await activeExecution;\n } else {\n await execute();\n }\n } catch (error) {\n console.error('\\n\u274C An error occurred:', error instanceof Error ? error.message : error);\n process.exit(1);\n }\n });\n\n // \u2500\u2500 file subcommand (local filesystem access) \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n program\n .command('file')\n .description('extract resources from local file(s) via direct filesystem access')\n .addArgument(fileArg)\n .addOption(\n new Option('--no-failsafe', `bypass the ${MAX_FILES_FAILSAFE}-file limit safety check`)\n )\n .action(async (paths: string[], options: { failsafe: boolean }) => {\n try {\n if (options.failsafe && paths.length > MAX_FILES_FAILSAFE) {\n console.error(\n `\\n\u274C ${paths.length} files specified exceeds the safety limit of ${MAX_FILES_FAILSAFE}.`\n );\n console.error(` Pass --no-failsafe to bypass this check and process all files.`);\n process.exit(1);\n }\n\n if (!options.failsafe && paths.length > MAX_FILES_FAILSAFE) {\n console.error(\n `\\n\u26A0\uFE0F Failsafe bypassed: processing ${paths.length} files (limit is ${MAX_FILES_FAILSAFE}).`\n );\n }\n\n console.error(`\\n\u2705 Processing ${paths.length} file(s)...`);\n\n const fileFetcher = new FileFetcher();\n const responses = await fileFetcher.fetchAll(paths);\n const pageMetadatas = await buildPageMetadata(\n responses.map(({ path, content, error }) => ({ path, content, error }))\n );\n\n await printer.print(...pageMetadatas);\n } catch (error) {\n console.error('\\n\u274C An error occurred:', error instanceof Error ? error.message : error);\n process.exit(1);\n }\n });\n\n await program.parseAsync(argv);\n}\n\nif (isCliEntrypoint()) {\n runCli().catch((error: unknown) => {\n console.error('\\n\u274C An error occurred:', error instanceof Error ? error.message : error);\n process.exit(1);\n });\n}\n", "export abstract class AbstractExtractor<V, R> {\n constructor(readonly name: string) {}\n abstract extract(value: V): Promise<R>;\n}\n", "import type { Page } from '../page/index.js';\nimport type { DOMResult } from '../page/index.js';\nimport { AbstractExtractor } from './AbstractExtractor.js';\n\nexport class PageExtractor extends AbstractExtractor<DOMResult, Page> {\n constructor() {\n super('page-extractor');\n }\n\n async extract(value: DOMResult): Promise<Page> {\n const {\n window: { document },\n url,\n } = value;\n return { title: document.title, url };\n }\n}\n", "/**\n * @license MIT\n * We are interested in visualising a page as a collection of tags.\n *\n * We wish to work with tags that can be compactly previewed on a webpage.\n * Here we must declare all of the element types that can be used to represent\n * a resource that can be hyperlinked off a webpage.\n */\ntype Tags = HTMLElementTagNameMap;\n\nexport const RESOURCE_DISPLAYABLE_KEYS = [\n 'id',\n 'innerText',\n 'textContent',\n 'class',\n 'ariaLabel',\n 'ariaDescription',\n 'alt',\n] as const;\n\nexport type DisplayableKey = (typeof RESOURCE_DISPLAYABLE_KEYS)[number];\n\nexport const RESOURCE_LINK_KEYS = ['href', 'data-src', 'target', 'action', 'src', 'url'] as const;\n\nexport type LinkKey = (typeof RESOURCE_LINK_KEYS)[number];\n\nexport type AttributeKey = DisplayableKey | LinkKey;\n\nexport type ResourceKey = { key: AttributeKey; value: string };\nexport type ResourceLink = { key: LinkKey; value: string };\n\nexport type ExternalResource = {\n text: ResourceKey;\n link: ResourceLink;\n};\n\nexport type Tag = keyof Tags;\n\nexport type Resource = HTMLElement & {\n [K in AttributeKey]?: string | null;\n};\n\nexport type ResourceByName<T extends keyof Tags> = Tags[T];\n\n// --- adapters ---\n\nconst readAttr = (element: Resource, key: AttributeKey): string | undefined => {\n const v = element.getAttribute(key);\n return v != null && v.trim() !== '' ? v : undefined;\n};\n\nexport function findResourceText(element: Resource): ResourceKey | undefined {\n for (const key of RESOURCE_DISPLAYABLE_KEYS) {\n const value = readAttr(element, key);\n if (value !== undefined) return { key, value };\n }\n return undefined;\n}\n\nexport function findResourceLink(element: Resource): ResourceLink | undefined {\n for (const key of RESOURCE_LINK_KEYS) {\n const value = readAttr(element, key);\n if (value !== undefined) return { key, value };\n }\n return undefined;\n}\n\nexport const isResourceKey = (key: string): key is AttributeKey =>\n (RESOURCE_DISPLAYABLE_KEYS as readonly string[]).includes(key) ||\n (RESOURCE_LINK_KEYS as readonly string[]).includes(key);\n", "import type { DOMResult } from '../page/index.js';\nimport {\n findResourceLink,\n findResourceText,\n type ExternalResource,\n type Resource,\n type Tag,\n} from '../resource.js';\nimport { AbstractExtractor } from './AbstractExtractor.js';\n\nexport class ResourceExtractor extends AbstractExtractor<DOMResult, ExternalResource[]> {\n constructor(private readonly tags: Tag[]) {\n super('page-extractor');\n }\n async extract(value: DOMResult): Promise<ExternalResource[]> {\n const { document } = value.window;\n return this.tags.flatMap((tag) =>\n Array.from(document.querySelectorAll<Resource>(tag)).flatMap((element) => {\n const link = findResourceLink(element);\n if (!link) return [];\n const text = findResourceText(element) ?? { key: 'src' as const, value: link.value };\n return [{ text, link }];\n })\n );\n }\n}\n", "import { parseHTML } from 'linkedom';\n\ntype ParseHTMLResult = {\n document: Document;\n};\n\nexport interface DOMResult {\n window: { document: Document };\n url: string;\n}\n\ninterface PageResponse {\n url: string;\n content?: DOMResult;\n error?: string;\n}\n\nexport class PageFetcher {\n private readonly timeout: number;\n private readonly maxRetries: number;\n\n constructor(timeout = 10000, maxRetries = 2) {\n this.timeout = timeout;\n this.maxRetries = maxRetries;\n }\n\n private buildDOMResult(html: string, url: string): DOMResult {\n const { document } = parseHTML(html) as ParseHTMLResult;\n return { window: { document }, url };\n }\n\n private decodeHtml(buffer: ArrayBuffer, charset: string): string {\n try {\n return new TextDecoder(charset).decode(new Uint8Array(buffer));\n } catch {\n return new TextDecoder('utf-8').decode(new Uint8Array(buffer));\n }\n }\n\n private async fetchPage(url: string, retryCount = 0): Promise<PageResponse> {\n const controller = new AbortController();\n let timeoutId: ReturnType<typeof setTimeout> | null = null;\n\n try {\n if (this.timeout > 0) {\n timeoutId = setTimeout(() => {\n controller.abort(new Error('Request timeout'));\n }, this.timeout);\n }\n\n const content = await fetch(url, { signal: controller.signal }).then(async (response) => {\n const buffer = await response.arrayBuffer();\n const contentType = response.headers.get('content-type') ?? '';\n const charsetMatch = /charset=([^\\s;]+)/i.exec(contentType);\n const html = this.decodeHtml(buffer, charsetMatch?.[1] ?? 'utf-8');\n return this.buildDOMResult(html, url);\n });\n\n return { url, content };\n } catch (error) {\n const abortTimeout = error instanceof Error && error.name === 'AbortError';\n const message = abortTimeout\n ? 'Request timeout'\n : error instanceof Error\n ? error.message\n : 'Unknown error';\n\n // Retry logic for transient errors\n if (retryCount < this.maxRetries && this.isRetryableError(message)) {\n process.stderr.write(`Retrying ${url} (attempt ${retryCount + 1}/${this.maxRetries})...\\n`);\n await this.delay(1000 * (retryCount + 1)); // Exponential backoff\n return this.fetchPage(url, retryCount + 1);\n }\n\n return { url, error: `Failed to fetch: ${message}` };\n } finally {\n if (timeoutId !== null) {\n clearTimeout(timeoutId);\n }\n }\n }\n\n private isRetryableError(message: string): boolean {\n const retryablePatterns = [/timeout/i, /ECONNRESET/i, /ETIMEDOUT/i, /ENOTFOUND/i, /network/i];\n return retryablePatterns.some((pattern) => pattern.test(message));\n }\n\n private delay(ms: number): Promise<void> {\n return new Promise((resolve) => setTimeout(resolve, ms));\n }\n\n async fetchAll(urls: string[]): Promise<PageResponse[]> {\n const responses = await Promise.all(urls.map((url) => this.fetchPage(url)));\n return responses.filter((response) => response.content !== undefined || response.error);\n }\n}\n", "import { readFile } from 'node:fs/promises';\nimport { parseHTML } from 'linkedom';\n\nimport type { DOMResult } from './PageFetcher.js';\n\nexport const MAX_FILES_FAILSAFE = 254;\n\ntype ParseHTMLResult = {\n document: Document;\n};\n\nexport interface FileResponse {\n path: string;\n content?: DOMResult;\n error?: string;\n}\n\nexport class FileFetcher {\n private buildDOMResult(html: string, filePath: string): DOMResult {\n const { document } = parseHTML(html) as ParseHTMLResult;\n return { window: { document }, url: `file://${filePath}` };\n }\n\n async fetchFile(filePath: string): Promise<FileResponse> {\n try {\n // filePath is supplied directly by the CLI user, not derived from network input.\n // eslint-disable-next-line security/detect-non-literal-fs-filename\n const html = await readFile(filePath, 'utf-8');\n return { path: filePath, content: this.buildDOMResult(html, filePath) };\n } catch (error) {\n return {\n path: filePath,\n error: error instanceof Error ? error.message : 'Unknown error',\n };\n }\n }\n\n async fetchAll(filePaths: string[]): Promise<FileResponse[]> {\n return Promise.all(filePaths.map((p) => this.fetchFile(p)));\n }\n}\n", "import type { PageMetadata } from '../page/index.js';\n\nexport abstract class AbstractResourcePrinter {\n constructor() {}\n abstract print(...pages: PageMetadata[]): void | Promise<void>;\n}\n", "import type { PageMetadata } from '../page/index.js';\nimport { AbstractResourcePrinter } from './AbstractResourcePrinter.js';\n\nexport class JSONStylePrinter extends AbstractResourcePrinter {\n print(...pages: PageMetadata[]): void | Promise<void> {\n const json = JSON.stringify(pages);\n process.stdout.write(json + '\\n');\n }\n}\n", "/**\n * Security utilities for URL validation and sanitization\n */\n\nconst ALLOWED_PROTOCOLS = ['http:', 'https:'];\nconst MAX_URL_LENGTH = 2048;\nconst SUSPICIOUS_PATTERNS = [\n /javascript:/i,\n /data:/i,\n /vbscript:/i,\n /<script/i,\n /on\\w+=/i, // Event handlers like onclick=\n];\n\nexport interface ValidationResult {\n isValid: boolean;\n error?: string;\n sanitizedUrl?: string;\n}\n\n/**\n * Validates a URL for security concerns\n * @param url - The URL to validate\n * @returns ValidationResult object with validation status\n */\nexport function validateUrl(url: string): ValidationResult {\n // Check if URL is empty or whitespace\n if (!url || !url.trim()) {\n return {\n isValid: false,\n error: 'URL cannot be empty',\n };\n }\n\n const trimmedUrl = url.trim();\n\n // Check URL length to prevent DoS\n if (trimmedUrl.length > MAX_URL_LENGTH) {\n return {\n isValid: false,\n error: `URL exceeds maximum length of ${MAX_URL_LENGTH} characters`,\n };\n }\n\n // Check for suspicious patterns\n for (const pattern of SUSPICIOUS_PATTERNS) {\n if (pattern.test(trimmedUrl)) {\n return {\n isValid: false,\n error: 'URL contains suspicious patterns',\n };\n }\n }\n\n // Parse the URL\n let parsedUrl: URL;\n try {\n parsedUrl = new URL(trimmedUrl);\n } catch {\n return {\n isValid: false,\n error: 'Invalid URL format',\n };\n }\n\n // Check protocol\n if (!ALLOWED_PROTOCOLS.includes(parsedUrl.protocol)) {\n return {\n isValid: false,\n error: `Protocol ${parsedUrl.protocol} is not allowed. Allowed protocols: ${ALLOWED_PROTOCOLS.join(', ')}`,\n };\n }\n\n // Check for localhost/internal IPs in production (security consideration)\n const hostname = parsedUrl.hostname.toLowerCase();\n const isLocalhost =\n hostname === 'localhost' ||\n hostname === '127.0.0.1' ||\n hostname === '::1' ||\n hostname.startsWith('192.168.') ||\n hostname.startsWith('10.') ||\n /^172\\.(1[6-9]|2\\d|3[01])\\./.test(hostname);\n\n if (isLocalhost) {\n // Allow but warn about localhost URLs\n console.warn(`Warning: Accessing local network resource: ${trimmedUrl}`);\n }\n\n return {\n isValid: true,\n sanitizedUrl: parsedUrl.toString(),\n };\n}\n\n/**\n * Validates an array of URLs\n * @param urls - Array of URLs to validate\n * @returns Object with valid URLs and errors\n */\nexport function validateUrls(urls: string[]): {\n validUrls: string[];\n errors: Array<{ url: string; error: string }>;\n} {\n const validUrls: string[] = [];\n const errors: Array<{ url: string; error: string }> = [];\n\n for (const url of urls) {\n const result = validateUrl(url);\n if (result.isValid && result.sanitizedUrl) {\n validUrls.push(result.sanitizedUrl);\n } else {\n errors.push({\n url,\n error: result.error || 'Unknown validation error',\n });\n }\n }\n\n return { validUrls, errors };\n}\n\n/**\n * Rate limiter to prevent abuse\n */\nexport class RateLimiter {\n private requests: number[] = [];\n private readonly maxRequests: number;\n private readonly windowMs: number;\n\n constructor(maxRequests = 10, windowMs = 60000) {\n this.maxRequests = maxRequests;\n this.windowMs = windowMs;\n }\n\n /**\n * Check if a request is allowed under rate limiting\n * @returns true if request is allowed, false otherwise\n */\n public isAllowed(): boolean {\n const now = Date.now();\n\n // Remove old requests outside the time window\n this.requests = this.requests.filter((time) => now - time < this.windowMs);\n\n if (this.requests.length >= this.maxRequests) {\n return false;\n }\n\n this.requests.push(now);\n return true;\n }\n\n /**\n * Get remaining requests in current window\n */\n public getRemainingRequests(): number {\n const now = Date.now();\n this.requests = this.requests.filter((time) => now - time < this.windowMs);\n return Math.max(0, this.maxRequests - this.requests.length);\n }\n}\n\n/**\n * Sanitizes HTML content to prevent XSS attacks\n * @param text - Text to sanitize\n * @returns Sanitized text\n */\nexport function sanitizeText(text: string): string {\n if (!text) return '';\n\n return text\n .replace(/</g, '<')\n .replace(/>/g, '>')\n .replace(/\"/g, '"')\n .replace(/'/g, ''')\n .replace(/\\//g, '/');\n}\n"],
|
|
5
|
-
"mappings": ";;;AACA,SAAS,SAAS,gBAAgB,cAAc;AAChD,SAAS,qBAAqB;AAC9B,SAAS,eAAe;AACxB,SAAS,qBAAqB;;;
|
|
4
|
+
"sourcesContent": ["#!/usr/bin/env node\nimport { Command, createArgument, Option } from 'commander';\nimport { createRequire } from 'node:module';\nimport { resolve } from 'node:path';\nimport { fileURLToPath } from 'node:url';\nimport { realpathSync } from 'node:fs';\n\nimport { PageExtractor, ResourceExtractor } from './extractors/index.js';\nimport { FileFetcher, MAX_FILES_FAILSAFE, PageFetcher, type PageMetadata } from './page/index.js';\nimport { JSONStylePrinter } from './printers/index.js';\nimport { validateUrls } from './security.js';\n\nconst require = createRequire(import.meta.url);\nconst pkg = require('../package.json') as {\n description: string;\n name: string;\n version: string;\n};\n\nconst { description, name, version } = pkg;\n\nconst program = new Command();\n\nconst urlArg = createArgument('<url...>', 'remote https://URL to extract from');\nconst fileArg = createArgument('<paths...>', 'local file paths to extract from');\n\n// Shared extractor instances.\nconst pageExtractor = new PageExtractor();\nconst resourceExtractor = new ResourceExtractor(['a', 'meta', 'link', 'embed', 'script']);\nconst printer = new JSONStylePrinter();\n\nasync function buildPageMetadata(\n responses: Array<{\n url?: string;\n path?: string;\n content?: import('./page/index.js').DOMResult;\n error?: string;\n }>\n): Promise<PageMetadata[]> {\n const pageMetadatas: PageMetadata[] = [];\n\n for (const { content, url: responseUrl, path, error } of responses) {\n const resolvedUrl = responseUrl ?? path ?? '';\n const resources =\n error !== undefined || !content ? [] : await resourceExtractor.extract(content);\n const descriptor =\n error !== undefined || !content\n ? { url: resolvedUrl, error: error ?? 'Unknown error', resources }\n : await pageExtractor.extract(content);\n pageMetadatas.push({ ...descriptor, resources });\n }\n\n return pageMetadatas;\n}\n\nfunction normalizeLocalPath(value: string): string {\n if (value.startsWith('file://')) {\n return fileURLToPath(value);\n }\n\n return value;\n}\n\nasync function runFileCommand(paths: string[], options: { failsafe: boolean }): Promise<void> {\n if (options.failsafe && paths.length > MAX_FILES_FAILSAFE) {\n console.error(\n `\\n\u274C ${paths.length} files specified exceeds the safety limit of ${MAX_FILES_FAILSAFE}.`\n );\n console.error(` Pass --no-failsafe to bypass this check and process all files.`);\n process.exit(1);\n }\n\n if (!options.failsafe && paths.length > MAX_FILES_FAILSAFE) {\n console.error(\n `\\n\u26A0\uFE0F Failsafe bypassed: processing ${paths.length} files (limit is ${MAX_FILES_FAILSAFE}).`\n );\n }\n\n console.error(`\\n\u2705 Processing ${paths.length} file(s)...`);\n\n const fileFetcher = new FileFetcher();\n const normalizedPaths = paths.map((pathValue) => normalizeLocalPath(pathValue));\n const responses = await fileFetcher.fetchAll(normalizedPaths);\n const pageMetadatas = await buildPageMetadata(\n responses.map(({ path, content, error }) => ({ path, content, error }))\n );\n\n await printer.print(...pageMetadatas);\n}\n\nfunction isCliEntrypoint(): boolean {\n const invokedPath = process.argv[1];\n if (!invokedPath) {\n return false;\n }\n\n try {\n return realpathSync(fileURLToPath(import.meta.url)) === realpathSync(resolve(invokedPath));\n } catch {\n return false;\n }\n}\n\nexport async function runCli(argv: string[] = process.argv): Promise<void> {\n program.name(name).version(version, '-v, --version').description(description);\n\n program\n .addArgument(fileArg)\n .addOption(\n new Option('--no-failsafe', `bypass the ${MAX_FILES_FAILSAFE}-file limit safety check`)\n )\n .action(async (paths: string[], options: { failsafe: boolean }) => {\n try {\n await runFileCommand(paths, options);\n } catch (error) {\n console.error('\\n\u274C An error occurred:', error instanceof Error ? error.message : error);\n process.exit(1);\n }\n });\n\n // \u2500\u2500 fetch subcommand (remote URL mode only) \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n program\n .command('fetch')\n .description('fetch and extract resources from remote URL(s)')\n .addArgument(urlArg)\n .addOption(\n new Option(\n '--watch',\n 'keep running: SIGWINCH re-fetches after resize, Ctrl-D releases in-flight requests, Ctrl-C exits'\n )\n )\n .addOption(new Option('-A, --user-agent <value>', 'override the HTTP User-Agent header'))\n .action(async (urls: string[], options: { watch: boolean; userAgent?: string }) => {\n try {\n const { validUrls, errors } = validateUrls(urls);\n\n if (errors.length > 0) {\n console.error('\\n\u274C URL Validation Errors:');\n errors.forEach(({ url: invalidUrl, error }) => {\n console.error(` - ${invalidUrl}: ${error}`);\n });\n }\n\n if (validUrls.length === 0) {\n console.error('\\n\u274C No valid URLs to process. Exiting.');\n process.exit(1);\n }\n\n console.error(`\\n\u2705 Processing ${validUrls.length} valid URL(s)...`);\n\n const pageFetcher = new PageFetcher(options.watch ? 0 : 10000, 2, options.userAgent);\n\n const execute = async (): Promise<void> => {\n const responses = await pageFetcher.fetchAll(validUrls);\n const pageMetadatas = await buildPageMetadata(responses);\n await printer.print(...pageMetadatas);\n };\n\n if (options.watch) {\n process.stdin.resume();\n process.on('SIGINT', () => process.exit(0));\n\n let activeExecution: Promise<void> | null = null;\n process.stdin.on('end', () => {\n activeExecution = null;\n });\n\n let winchTimer: ReturnType<typeof setTimeout> | null = null;\n process.on('SIGWINCH', () => {\n if (winchTimer !== null) clearTimeout(winchTimer);\n winchTimer = setTimeout(() => {\n winchTimer = null;\n activeExecution = execute().catch((err: unknown) => {\n console.error('\\n\u274C An error occurred:', err instanceof Error ? err.message : err);\n });\n }, 150);\n });\n\n activeExecution = execute();\n await activeExecution;\n } else {\n await execute();\n }\n } catch (error) {\n console.error('\\n\u274C An error occurred:', error instanceof Error ? error.message : error);\n process.exit(1);\n }\n });\n\n // \u2500\u2500 file subcommand (local filesystem access) \u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\u2500\n program\n .command('file')\n .description('extract resources from local file(s) via direct filesystem access')\n .addArgument(fileArg)\n .addOption(\n new Option('--no-failsafe', `bypass the ${MAX_FILES_FAILSAFE}-file limit safety check`)\n )\n .action(async (paths: string[], options: { failsafe: boolean }) => {\n try {\n await runFileCommand(paths, options);\n } catch (error) {\n console.error('\\n\u274C An error occurred:', error instanceof Error ? error.message : error);\n process.exit(1);\n }\n });\n\n await program.parseAsync(argv);\n}\n\nif (isCliEntrypoint()) {\n runCli().catch((error: unknown) => {\n console.error('\\n\u274C An error occurred:', error instanceof Error ? error.message : error);\n process.exit(1);\n });\n}\n", "export abstract class AbstractExtractor<V, R> {\n constructor(readonly name: string) {}\n abstract extract(value: V): Promise<R>;\n}\n", "import type { Page } from '../page/index.js';\nimport type { DOMResult } from '../page/index.js';\nimport { AbstractExtractor } from './AbstractExtractor.js';\n\nexport class PageExtractor extends AbstractExtractor<DOMResult, Page> {\n constructor() {\n super('page-extractor');\n }\n\n async extract(value: DOMResult): Promise<Page> {\n const {\n window: { document },\n url,\n } = value;\n return { title: document.title, url };\n }\n}\n", "/**\n * @license MIT\n * We are interested in visualising a page as a collection of tags.\n *\n * We wish to work with tags that can be compactly previewed on a webpage.\n * Here we must declare all of the element types that can be used to represent\n * a resource that can be hyperlinked off a webpage.\n */\ntype Tags = HTMLElementTagNameMap;\n\nexport const RESOURCE_DISPLAYABLE_KEYS = [\n 'id',\n 'innerText',\n 'textContent',\n 'class',\n 'ariaLabel',\n 'ariaDescription',\n 'alt',\n] as const;\n\nexport type DisplayableKey = (typeof RESOURCE_DISPLAYABLE_KEYS)[number];\n\nexport const RESOURCE_LINK_KEYS = ['href', 'data-src', 'target', 'action', 'src', 'url'] as const;\n\nexport type LinkKey = (typeof RESOURCE_LINK_KEYS)[number];\n\nexport type AttributeKey = DisplayableKey | LinkKey;\n\nexport type ResourceKey = { key: AttributeKey; value: string };\nexport type ResourceLink = { key: LinkKey; value: string };\n\nexport type ExternalResource = {\n text: ResourceKey;\n link: ResourceLink;\n};\n\nexport type Tag = keyof Tags;\n\nexport type Resource = HTMLElement & {\n [K in AttributeKey]?: string | null;\n};\n\nexport type ResourceByName<T extends keyof Tags> = Tags[T];\n\n// --- adapters ---\n\nconst readAttr = (element: Resource, key: AttributeKey): string | undefined => {\n const v = element.getAttribute(key);\n return v != null && v.trim() !== '' ? v : undefined;\n};\n\nexport function findResourceText(element: Resource): ResourceKey | undefined {\n for (const key of RESOURCE_DISPLAYABLE_KEYS) {\n const value = readAttr(element, key);\n if (value !== undefined) return { key, value };\n }\n return undefined;\n}\n\nexport function findResourceLink(element: Resource): ResourceLink | undefined {\n for (const key of RESOURCE_LINK_KEYS) {\n const value = readAttr(element, key);\n if (value !== undefined) return { key, value };\n }\n return undefined;\n}\n\nexport const isResourceKey = (key: string): key is AttributeKey =>\n (RESOURCE_DISPLAYABLE_KEYS as readonly string[]).includes(key) ||\n (RESOURCE_LINK_KEYS as readonly string[]).includes(key);\n", "import type { DOMResult } from '../page/index.js';\nimport {\n findResourceLink,\n findResourceText,\n type ExternalResource,\n type Resource,\n type Tag,\n} from '../resource.js';\nimport { AbstractExtractor } from './AbstractExtractor.js';\n\nexport class ResourceExtractor extends AbstractExtractor<DOMResult, ExternalResource[]> {\n constructor(private readonly tags: Tag[]) {\n super('page-extractor');\n }\n async extract(value: DOMResult): Promise<ExternalResource[]> {\n const { document } = value.window;\n return this.tags.flatMap((tag) =>\n Array.from(document.querySelectorAll<Resource>(tag)).flatMap((element) => {\n const link = findResourceLink(element);\n if (!link) return [];\n const text = findResourceText(element) ?? { key: 'src' as const, value: link.value };\n return [{ text, link }];\n })\n );\n }\n}\n", "import { parseHTML } from 'linkedom';\n\ntype ParseHTMLResult = {\n document: Document;\n};\n\nexport interface DOMResult {\n window: { document: Document };\n url: string;\n}\n\ninterface PageResponse {\n url: string;\n content?: DOMResult;\n error?: string;\n}\n\nexport class PageFetcher {\n private readonly timeout: number;\n private readonly maxRetries: number;\n private readonly userAgent?: string;\n\n constructor(timeout = 10000, maxRetries = 2, userAgent?: string) {\n this.timeout = timeout;\n this.maxRetries = maxRetries;\n this.userAgent = userAgent;\n }\n\n private buildDOMResult(html: string, url: string): DOMResult {\n const { document } = parseHTML(html) as ParseHTMLResult;\n return { window: { document }, url };\n }\n\n private decodeHtml(buffer: ArrayBuffer, charset: string): string {\n try {\n return new TextDecoder(charset).decode(new Uint8Array(buffer));\n } catch {\n return new TextDecoder('utf-8').decode(new Uint8Array(buffer));\n }\n }\n\n private async fetchPage(url: string, retryCount = 0): Promise<PageResponse> {\n const controller = new AbortController();\n let timeoutId: ReturnType<typeof setTimeout> | null = null;\n\n try {\n if (this.timeout > 0) {\n timeoutId = setTimeout(() => {\n controller.abort(new Error('Request timeout'));\n }, this.timeout);\n }\n\n const headers = this.userAgent ? { 'user-agent': this.userAgent } : undefined;\n const content = await fetch(url, { headers, signal: controller.signal }).then(\n async (response) => {\n const buffer = await response.arrayBuffer();\n const contentType = response.headers.get('content-type') ?? '';\n const charsetMatch = /charset=([^\\s;]+)/i.exec(contentType);\n const html = this.decodeHtml(buffer, charsetMatch?.[1] ?? 'utf-8');\n return this.buildDOMResult(html, url);\n }\n );\n\n return { url, content };\n } catch (error) {\n const abortTimeout = error instanceof Error && error.name === 'AbortError';\n const message = abortTimeout\n ? 'Request timeout'\n : error instanceof Error\n ? error.message\n : 'Unknown error';\n\n // Retry logic for transient errors\n if (retryCount < this.maxRetries && this.isRetryableError(message)) {\n process.stderr.write(`Retrying ${url} (attempt ${retryCount + 1}/${this.maxRetries})...\\n`);\n await this.delay(1000 * (retryCount + 1)); // Exponential backoff\n return this.fetchPage(url, retryCount + 1);\n }\n\n return { url, error: `Failed to fetch: ${message}` };\n } finally {\n if (timeoutId !== null) {\n clearTimeout(timeoutId);\n }\n }\n }\n\n private isRetryableError(message: string): boolean {\n const retryablePatterns = [/timeout/i, /ECONNRESET/i, /ETIMEDOUT/i, /ENOTFOUND/i, /network/i];\n return retryablePatterns.some((pattern) => pattern.test(message));\n }\n\n private delay(ms: number): Promise<void> {\n return new Promise((resolve) => setTimeout(resolve, ms));\n }\n\n async fetchAll(urls: string[]): Promise<PageResponse[]> {\n const responses = await Promise.all(urls.map((url) => this.fetchPage(url)));\n return responses.filter((response) => response.content !== undefined || response.error);\n }\n}\n", "import { readFile } from 'node:fs/promises';\nimport { parseHTML } from 'linkedom';\n\nimport type { DOMResult } from './PageFetcher.js';\n\nexport const MAX_FILES_FAILSAFE = 254;\n\ntype ParseHTMLResult = {\n document: Document;\n};\n\nexport interface FileResponse {\n path: string;\n content?: DOMResult;\n error?: string;\n}\n\nexport class FileFetcher {\n private buildDOMResult(html: string, filePath: string): DOMResult {\n const { document } = parseHTML(html) as ParseHTMLResult;\n return { window: { document }, url: `file://${filePath}` };\n }\n\n async fetchFile(filePath: string): Promise<FileResponse> {\n try {\n // filePath is supplied directly by the CLI user, not derived from network input.\n // eslint-disable-next-line security/detect-non-literal-fs-filename\n const html = await readFile(filePath, 'utf-8');\n return { path: filePath, content: this.buildDOMResult(html, filePath) };\n } catch (error) {\n return {\n path: filePath,\n error: error instanceof Error ? error.message : 'Unknown error',\n };\n }\n }\n\n async fetchAll(filePaths: string[]): Promise<FileResponse[]> {\n return Promise.all(filePaths.map((p) => this.fetchFile(p)));\n }\n}\n", "import type { PageMetadata } from '../page/index.js';\n\nexport abstract class AbstractResourcePrinter {\n constructor() {}\n abstract print(...pages: PageMetadata[]): void | Promise<void>;\n}\n", "import type { PageMetadata } from '../page/index.js';\nimport { AbstractResourcePrinter } from './AbstractResourcePrinter.js';\n\nexport class JSONStylePrinter extends AbstractResourcePrinter {\n print(...pages: PageMetadata[]): void | Promise<void> {\n const json = JSON.stringify(pages);\n process.stdout.write(json + '\\n');\n }\n}\n", "/**\n * Security utilities for URL validation and sanitization\n */\n\nconst ALLOWED_PROTOCOLS = ['http:', 'https:'];\nconst MAX_URL_LENGTH = 2048;\nconst SUSPICIOUS_PATTERNS = [\n /javascript:/i,\n /data:/i,\n /vbscript:/i,\n /<script/i,\n /on\\w+=/i, // Event handlers like onclick=\n];\n\nexport interface ValidationResult {\n isValid: boolean;\n error?: string;\n sanitizedUrl?: string;\n}\n\n/**\n * Validates a URL for security concerns\n * @param url - The URL to validate\n * @returns ValidationResult object with validation status\n */\nexport function validateUrl(url: string): ValidationResult {\n // Check if URL is empty or whitespace\n if (!url || !url.trim()) {\n return {\n isValid: false,\n error: 'URL cannot be empty',\n };\n }\n\n const trimmedUrl = url.trim();\n\n // Check URL length to prevent DoS\n if (trimmedUrl.length > MAX_URL_LENGTH) {\n return {\n isValid: false,\n error: `URL exceeds maximum length of ${MAX_URL_LENGTH} characters`,\n };\n }\n\n // Check for suspicious patterns\n for (const pattern of SUSPICIOUS_PATTERNS) {\n if (pattern.test(trimmedUrl)) {\n return {\n isValid: false,\n error: 'URL contains suspicious patterns',\n };\n }\n }\n\n // Parse the URL\n let parsedUrl: URL;\n try {\n parsedUrl = new URL(trimmedUrl);\n } catch {\n return {\n isValid: false,\n error: 'Invalid URL format',\n };\n }\n\n // Check protocol\n if (!ALLOWED_PROTOCOLS.includes(parsedUrl.protocol)) {\n return {\n isValid: false,\n error: `Protocol ${parsedUrl.protocol} is not allowed. Allowed protocols: ${ALLOWED_PROTOCOLS.join(', ')}`,\n };\n }\n\n // Check for localhost/internal IPs in production (security consideration)\n const hostname = parsedUrl.hostname.toLowerCase();\n const isLocalhost =\n hostname === 'localhost' ||\n hostname === '127.0.0.1' ||\n hostname === '::1' ||\n hostname.startsWith('192.168.') ||\n hostname.startsWith('10.') ||\n /^172\\.(1[6-9]|2\\d|3[01])\\./.test(hostname);\n\n if (isLocalhost) {\n // Allow but warn about localhost URLs\n console.warn(`Warning: Accessing local network resource: ${trimmedUrl}`);\n }\n\n return {\n isValid: true,\n sanitizedUrl: parsedUrl.toString(),\n };\n}\n\n/**\n * Validates an array of URLs\n * @param urls - Array of URLs to validate\n * @returns Object with valid URLs and errors\n */\nexport function validateUrls(urls: string[]): {\n validUrls: string[];\n errors: Array<{ url: string; error: string }>;\n} {\n const validUrls: string[] = [];\n const errors: Array<{ url: string; error: string }> = [];\n\n for (const url of urls) {\n const result = validateUrl(url);\n if (result.isValid && result.sanitizedUrl) {\n validUrls.push(result.sanitizedUrl);\n } else {\n errors.push({\n url,\n error: result.error || 'Unknown validation error',\n });\n }\n }\n\n return { validUrls, errors };\n}\n\n/**\n * Rate limiter to prevent abuse\n */\nexport class RateLimiter {\n private requests: number[] = [];\n private readonly maxRequests: number;\n private readonly windowMs: number;\n\n constructor(maxRequests = 10, windowMs = 60000) {\n this.maxRequests = maxRequests;\n this.windowMs = windowMs;\n }\n\n /**\n * Check if a request is allowed under rate limiting\n * @returns true if request is allowed, false otherwise\n */\n public isAllowed(): boolean {\n const now = Date.now();\n\n // Remove old requests outside the time window\n this.requests = this.requests.filter((time) => now - time < this.windowMs);\n\n if (this.requests.length >= this.maxRequests) {\n return false;\n }\n\n this.requests.push(now);\n return true;\n }\n\n /**\n * Get remaining requests in current window\n */\n public getRemainingRequests(): number {\n const now = Date.now();\n this.requests = this.requests.filter((time) => now - time < this.windowMs);\n return Math.max(0, this.maxRequests - this.requests.length);\n }\n}\n\n/**\n * Sanitizes HTML content to prevent XSS attacks\n * @param text - Text to sanitize\n * @returns Sanitized text\n */\nexport function sanitizeText(text: string): string {\n if (!text) return '';\n\n return text\n .replace(/</g, '<')\n .replace(/>/g, '>')\n .replace(/\"/g, '"')\n .replace(/'/g, ''')\n .replace(/\\//g, '/');\n}\n"],
|
|
5
|
+
"mappings": ";;;AACA,SAAS,SAAS,gBAAgB,cAAc;AAChD,SAAS,qBAAqB;AAC9B,SAAS,eAAe;AACxB,SAAS,qBAAqB;AAC9B,SAAS,oBAAoB;;;ACLtB,IAAe,oBAAf,MAAuC;AAAA,EAC5C,YAAqBA,OAAc;AAAd,gBAAAA;AAAA,EAAe;AAEtC;;;ACCO,IAAM,gBAAN,cAA4B,kBAAmC;AAAA,EACpE,cAAc;AACZ,UAAM,gBAAgB;AAAA,EACxB;AAAA,EAEA,MAAM,QAAQ,OAAiC;AAC7C,UAAM;AAAA,MACJ,QAAQ,EAAE,SAAS;AAAA,MACnB;AAAA,IACF,IAAI;AACJ,WAAO,EAAE,OAAO,SAAS,OAAO,IAAI;AAAA,EACtC;AACF;;;ACNO,IAAM,4BAA4B;AAAA,EACvC;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AACF;AAIO,IAAM,qBAAqB,CAAC,QAAQ,YAAY,UAAU,UAAU,OAAO,KAAK;AAwBvF,IAAM,WAAW,CAAC,SAAmB,QAA0C;AAC7E,QAAM,IAAI,QAAQ,aAAa,GAAG;AAClC,SAAO,KAAK,QAAQ,EAAE,KAAK,MAAM,KAAK,IAAI;AAC5C;AAEO,SAAS,iBAAiB,SAA4C;AAC3E,aAAW,OAAO,2BAA2B;AAC3C,UAAM,QAAQ,SAAS,SAAS,GAAG;AACnC,QAAI,UAAU,OAAW,QAAO,EAAE,KAAK,MAAM;AAAA,EAC/C;AACA,SAAO;AACT;AAEO,SAAS,iBAAiB,SAA6C;AAC5E,aAAW,OAAO,oBAAoB;AACpC,UAAM,QAAQ,SAAS,SAAS,GAAG;AACnC,QAAI,UAAU,OAAW,QAAO,EAAE,KAAK,MAAM;AAAA,EAC/C;AACA,SAAO;AACT;;;ACvDO,IAAM,oBAAN,cAAgC,kBAAiD;AAAA,EACtF,YAA6B,MAAa;AACxC,UAAM,gBAAgB;AADK;AAAA,EAE7B;AAAA,EACA,MAAM,QAAQ,OAA+C;AAC3D,UAAM,EAAE,SAAS,IAAI,MAAM;AAC3B,WAAO,KAAK,KAAK;AAAA,MAAQ,CAAC,QACxB,MAAM,KAAK,SAAS,iBAA2B,GAAG,CAAC,EAAE,QAAQ,CAAC,YAAY;AACxE,cAAM,OAAO,iBAAiB,OAAO;AACrC,YAAI,CAAC,KAAM,QAAO,CAAC;AACnB,cAAM,OAAO,iBAAiB,OAAO,KAAK,EAAE,KAAK,OAAgB,OAAO,KAAK,MAAM;AACnF,eAAO,CAAC,EAAE,MAAM,KAAK,CAAC;AAAA,MACxB,CAAC;AAAA,IACH;AAAA,EACF;AACF;;;ACzBA,SAAS,iBAAiB;AAiBnB,IAAM,cAAN,MAAkB;AAAA,EACN;AAAA,EACA;AAAA,EACA;AAAA,EAEjB,YAAY,UAAU,KAAO,aAAa,GAAG,WAAoB;AAC/D,SAAK,UAAU;AACf,SAAK,aAAa;AAClB,SAAK,YAAY;AAAA,EACnB;AAAA,EAEQ,eAAe,MAAc,KAAwB;AAC3D,UAAM,EAAE,SAAS,IAAI,UAAU,IAAI;AACnC,WAAO,EAAE,QAAQ,EAAE,SAAS,GAAG,IAAI;AAAA,EACrC;AAAA,EAEQ,WAAW,QAAqB,SAAyB;AAC/D,QAAI;AACF,aAAO,IAAI,YAAY,OAAO,EAAE,OAAO,IAAI,WAAW,MAAM,CAAC;AAAA,IAC/D,QAAQ;AACN,aAAO,IAAI,YAAY,OAAO,EAAE,OAAO,IAAI,WAAW,MAAM,CAAC;AAAA,IAC/D;AAAA,EACF;AAAA,EAEA,MAAc,UAAU,KAAa,aAAa,GAA0B;AAC1E,UAAM,aAAa,IAAI,gBAAgB;AACvC,QAAI,YAAkD;AAEtD,QAAI;AACF,UAAI,KAAK,UAAU,GAAG;AACpB,oBAAY,WAAW,MAAM;AAC3B,qBAAW,MAAM,IAAI,MAAM,iBAAiB,CAAC;AAAA,QAC/C,GAAG,KAAK,OAAO;AAAA,MACjB;AAEA,YAAM,UAAU,KAAK,YAAY,EAAE,cAAc,KAAK,UAAU,IAAI;AACpE,YAAM,UAAU,MAAM,MAAM,KAAK,EAAE,SAAS,QAAQ,WAAW,OAAO,CAAC,EAAE;AAAA,QACvE,OAAO,aAAa;AAClB,gBAAM,SAAS,MAAM,SAAS,YAAY;AAC1C,gBAAM,cAAc,SAAS,QAAQ,IAAI,cAAc,KAAK;AAC5D,gBAAM,eAAe,qBAAqB,KAAK,WAAW;AAC1D,gBAAM,OAAO,KAAK,WAAW,QAAQ,eAAe,CAAC,KAAK,OAAO;AACjE,iBAAO,KAAK,eAAe,MAAM,GAAG;AAAA,QACtC;AAAA,MACF;AAEA,aAAO,EAAE,KAAK,QAAQ;AAAA,IACxB,SAAS,OAAO;AACd,YAAM,eAAe,iBAAiB,SAAS,MAAM,SAAS;AAC9D,YAAM,UAAU,eACZ,oBACA,iBAAiB,QACf,MAAM,UACN;AAGN,UAAI,aAAa,KAAK,cAAc,KAAK,iBAAiB,OAAO,GAAG;AAClE,gBAAQ,OAAO,MAAM,YAAY,GAAG,aAAa,aAAa,CAAC,IAAI,KAAK,UAAU;AAAA,CAAQ;AAC1F,cAAM,KAAK,MAAM,OAAQ,aAAa,EAAE;AACxC,eAAO,KAAK,UAAU,KAAK,aAAa,CAAC;AAAA,MAC3C;AAEA,aAAO,EAAE,KAAK,OAAO,oBAAoB,OAAO,GAAG;AAAA,IACrD,UAAE;AACA,UAAI,cAAc,MAAM;AACtB,qBAAa,SAAS;AAAA,MACxB;AAAA,IACF;AAAA,EACF;AAAA,EAEQ,iBAAiB,SAA0B;AACjD,UAAM,oBAAoB,CAAC,YAAY,eAAe,cAAc,cAAc,UAAU;AAC5F,WAAO,kBAAkB,KAAK,CAAC,YAAY,QAAQ,KAAK,OAAO,CAAC;AAAA,EAClE;AAAA,EAEQ,MAAM,IAA2B;AACvC,WAAO,IAAI,QAAQ,CAACC,aAAY,WAAWA,UAAS,EAAE,CAAC;AAAA,EACzD;AAAA,EAEA,MAAM,SAAS,MAAyC;AACtD,UAAM,YAAY,MAAM,QAAQ,IAAI,KAAK,IAAI,CAAC,QAAQ,KAAK,UAAU,GAAG,CAAC,CAAC;AAC1E,WAAO,UAAU,OAAO,CAAC,aAAa,SAAS,YAAY,UAAa,SAAS,KAAK;AAAA,EACxF;AACF;;;ACpGA,SAAS,gBAAgB;AACzB,SAAS,aAAAC,kBAAiB;AAInB,IAAM,qBAAqB;AAY3B,IAAM,cAAN,MAAkB;AAAA,EACf,eAAe,MAAc,UAA6B;AAChE,UAAM,EAAE,SAAS,IAAIA,WAAU,IAAI;AACnC,WAAO,EAAE,QAAQ,EAAE,SAAS,GAAG,KAAK,UAAU,QAAQ,GAAG;AAAA,EAC3D;AAAA,EAEA,MAAM,UAAU,UAAyC;AACvD,QAAI;AAGF,YAAM,OAAO,MAAM,SAAS,UAAU,OAAO;AAC7C,aAAO,EAAE,MAAM,UAAU,SAAS,KAAK,eAAe,MAAM,QAAQ,EAAE;AAAA,IACxE,SAAS,OAAO;AACd,aAAO;AAAA,QACL,MAAM;AAAA,QACN,OAAO,iBAAiB,QAAQ,MAAM,UAAU;AAAA,MAClD;AAAA,IACF;AAAA,EACF;AAAA,EAEA,MAAM,SAAS,WAA8C;AAC3D,WAAO,QAAQ,IAAI,UAAU,IAAI,CAAC,MAAM,KAAK,UAAU,CAAC,CAAC,CAAC;AAAA,EAC5D;AACF;;;ACtCO,IAAe,0BAAf,MAAuC;AAAA,EAC5C,cAAc;AAAA,EAAC;AAEjB;;;ACFO,IAAM,mBAAN,cAA+B,wBAAwB;AAAA,EAC5D,SAAS,OAA6C;AACpD,UAAM,OAAO,KAAK,UAAU,KAAK;AACjC,YAAQ,OAAO,MAAM,OAAO,IAAI;AAAA,EAClC;AACF;;;ACJA,IAAM,oBAAoB,CAAC,SAAS,QAAQ;AAC5C,IAAM,iBAAiB;AACvB,IAAM,sBAAsB;AAAA,EAC1B;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA;AACF;AAaO,SAAS,YAAY,KAA+B;AAEzD,MAAI,CAAC,OAAO,CAAC,IAAI,KAAK,GAAG;AACvB,WAAO;AAAA,MACL,SAAS;AAAA,MACT,OAAO;AAAA,IACT;AAAA,EACF;AAEA,QAAM,aAAa,IAAI,KAAK;AAG5B,MAAI,WAAW,SAAS,gBAAgB;AACtC,WAAO;AAAA,MACL,SAAS;AAAA,MACT,OAAO,iCAAiC,cAAc;AAAA,IACxD;AAAA,EACF;AAGA,aAAW,WAAW,qBAAqB;AACzC,QAAI,QAAQ,KAAK,UAAU,GAAG;AAC5B,aAAO;AAAA,QACL,SAAS;AAAA,QACT,OAAO;AAAA,MACT;AAAA,IACF;AAAA,EACF;AAGA,MAAI;AACJ,MAAI;AACF,gBAAY,IAAI,IAAI,UAAU;AAAA,EAChC,QAAQ;AACN,WAAO;AAAA,MACL,SAAS;AAAA,MACT,OAAO;AAAA,IACT;AAAA,EACF;AAGA,MAAI,CAAC,kBAAkB,SAAS,UAAU,QAAQ,GAAG;AACnD,WAAO;AAAA,MACL,SAAS;AAAA,MACT,OAAO,YAAY,UAAU,QAAQ,uCAAuC,kBAAkB,KAAK,IAAI,CAAC;AAAA,IAC1G;AAAA,EACF;AAGA,QAAM,WAAW,UAAU,SAAS,YAAY;AAChD,QAAM,cACJ,aAAa,eACb,aAAa,eACb,aAAa,SACb,SAAS,WAAW,UAAU,KAC9B,SAAS,WAAW,KAAK,KACzB,6BAA6B,KAAK,QAAQ;AAE5C,MAAI,aAAa;AAEf,YAAQ,KAAK,8CAA8C,UAAU,EAAE;AAAA,EACzE;AAEA,SAAO;AAAA,IACL,SAAS;AAAA,IACT,cAAc,UAAU,SAAS;AAAA,EACnC;AACF;AAOO,SAAS,aAAa,MAG3B;AACA,QAAM,YAAsB,CAAC;AAC7B,QAAM,SAAgD,CAAC;AAEvD,aAAW,OAAO,MAAM;AACtB,UAAM,SAAS,YAAY,GAAG;AAC9B,QAAI,OAAO,WAAW,OAAO,cAAc;AACzC,gBAAU,KAAK,OAAO,YAAY;AAAA,IACpC,OAAO;AACL,aAAO,KAAK;AAAA,QACV;AAAA,QACA,OAAO,OAAO,SAAS;AAAA,MACzB,CAAC;AAAA,IACH;AAAA,EACF;AAEA,SAAO,EAAE,WAAW,OAAO;AAC7B;;;AT3GA,IAAMC,WAAU,cAAc,YAAY,GAAG;AAC7C,IAAM,MAAMA,SAAQ,iBAAiB;AAMrC,IAAM,EAAE,aAAa,MAAM,QAAQ,IAAI;AAEvC,IAAM,UAAU,IAAI,QAAQ;AAE5B,IAAM,SAAS,eAAe,YAAY,oCAAoC;AAC9E,IAAM,UAAU,eAAe,cAAc,kCAAkC;AAG/E,IAAM,gBAAgB,IAAI,cAAc;AACxC,IAAM,oBAAoB,IAAI,kBAAkB,CAAC,KAAK,QAAQ,QAAQ,SAAS,QAAQ,CAAC;AACxF,IAAM,UAAU,IAAI,iBAAiB;AAErC,eAAe,kBACb,WAMyB;AACzB,QAAM,gBAAgC,CAAC;AAEvC,aAAW,EAAE,SAAS,KAAK,aAAa,MAAM,MAAM,KAAK,WAAW;AAClE,UAAM,cAAc,eAAe,QAAQ;AAC3C,UAAM,YACJ,UAAU,UAAa,CAAC,UAAU,CAAC,IAAI,MAAM,kBAAkB,QAAQ,OAAO;AAChF,UAAM,aACJ,UAAU,UAAa,CAAC,UACpB,EAAE,KAAK,aAAa,OAAO,SAAS,iBAAiB,UAAU,IAC/D,MAAM,cAAc,QAAQ,OAAO;AACzC,kBAAc,KAAK,EAAE,GAAG,YAAY,UAAU,CAAC;AAAA,EACjD;AAEA,SAAO;AACT;AAEA,SAAS,mBAAmB,OAAuB;AACjD,MAAI,MAAM,WAAW,SAAS,GAAG;AAC/B,WAAO,cAAc,KAAK;AAAA,EAC5B;AAEA,SAAO;AACT;AAEA,eAAe,eAAe,OAAiB,SAA+C;AAC5F,MAAI,QAAQ,YAAY,MAAM,SAAS,oBAAoB;AACzD,YAAQ;AAAA,MACN;AAAA,SAAO,MAAM,MAAM,gDAAgD,kBAAkB;AAAA,IACvF;AACA,YAAQ,MAAM,mEAAmE;AACjF,YAAQ,KAAK,CAAC;AAAA,EAChB;AAEA,MAAI,CAAC,QAAQ,YAAY,MAAM,SAAS,oBAAoB;AAC1D,YAAQ;AAAA,MACN;AAAA,8CAAuC,MAAM,MAAM,oBAAoB,kBAAkB;AAAA,IAC3F;AAAA,EACF;AAEA,UAAQ,MAAM;AAAA,oBAAkB,MAAM,MAAM,aAAa;AAEzD,QAAM,cAAc,IAAI,YAAY;AACpC,QAAM,kBAAkB,MAAM,IAAI,CAAC,cAAc,mBAAmB,SAAS,CAAC;AAC9E,QAAM,YAAY,MAAM,YAAY,SAAS,eAAe;AAC5D,QAAM,gBAAgB,MAAM;AAAA,IAC1B,UAAU,IAAI,CAAC,EAAE,MAAM,SAAS,MAAM,OAAO,EAAE,MAAM,SAAS,MAAM,EAAE;AAAA,EACxE;AAEA,QAAM,QAAQ,MAAM,GAAG,aAAa;AACtC;AAEA,SAAS,kBAA2B;AAClC,QAAM,cAAc,QAAQ,KAAK,CAAC;AAClC,MAAI,CAAC,aAAa;AAChB,WAAO;AAAA,EACT;AAEA,MAAI;AACF,WAAO,aAAa,cAAc,YAAY,GAAG,CAAC,MAAM,aAAa,QAAQ,WAAW,CAAC;AAAA,EAC3F,QAAQ;AACN,WAAO;AAAA,EACT;AACF;AAEA,eAAsB,OAAO,OAAiB,QAAQ,MAAqB;AACzE,UAAQ,KAAK,IAAI,EAAE,QAAQ,SAAS,eAAe,EAAE,YAAY,WAAW;AAE5E,UACG,YAAY,OAAO,EACnB;AAAA,IACC,IAAI,OAAO,iBAAiB,cAAc,kBAAkB,0BAA0B;AAAA,EACxF,EACC,OAAO,OAAO,OAAiB,YAAmC;AACjE,QAAI;AACF,YAAM,eAAe,OAAO,OAAO;AAAA,IACrC,SAAS,OAAO;AACd,cAAQ,MAAM,+BAA0B,iBAAiB,QAAQ,MAAM,UAAU,KAAK;AACtF,cAAQ,KAAK,CAAC;AAAA,IAChB;AAAA,EACF,CAAC;AAGH,UACG,QAAQ,OAAO,EACf,YAAY,gDAAgD,EAC5D,YAAY,MAAM,EAClB;AAAA,IACC,IAAI;AAAA,MACF;AAAA,MACA;AAAA,IACF;AAAA,EACF,EACC,UAAU,IAAI,OAAO,4BAA4B,qCAAqC,CAAC,EACvF,OAAO,OAAO,MAAgB,YAAoD;AACjF,QAAI;AACF,YAAM,EAAE,WAAW,OAAO,IAAI,aAAa,IAAI;AAE/C,UAAI,OAAO,SAAS,GAAG;AACrB,gBAAQ,MAAM,iCAA4B;AAC1C,eAAO,QAAQ,CAAC,EAAE,KAAK,YAAY,MAAM,MAAM;AAC7C,kBAAQ,MAAM,OAAO,UAAU,KAAK,KAAK,EAAE;AAAA,QAC7C,CAAC;AAAA,MACH;AAEA,UAAI,UAAU,WAAW,GAAG;AAC1B,gBAAQ,MAAM,6CAAwC;AACtD,gBAAQ,KAAK,CAAC;AAAA,MAChB;AAEA,cAAQ,MAAM;AAAA,oBAAkB,UAAU,MAAM,kBAAkB;AAElE,YAAM,cAAc,IAAI,YAAY,QAAQ,QAAQ,IAAI,KAAO,GAAG,QAAQ,SAAS;AAEnF,YAAM,UAAU,YAA2B;AACzC,cAAM,YAAY,MAAM,YAAY,SAAS,SAAS;AACtD,cAAM,gBAAgB,MAAM,kBAAkB,SAAS;AACvD,cAAM,QAAQ,MAAM,GAAG,aAAa;AAAA,MACtC;AAEA,UAAI,QAAQ,OAAO;AACjB,gBAAQ,MAAM,OAAO;AACrB,gBAAQ,GAAG,UAAU,MAAM,QAAQ,KAAK,CAAC,CAAC;AAE1C,YAAI,kBAAwC;AAC5C,gBAAQ,MAAM,GAAG,OAAO,MAAM;AAC5B,4BAAkB;AAAA,QACpB,CAAC;AAED,YAAI,aAAmD;AACvD,gBAAQ,GAAG,YAAY,MAAM;AAC3B,cAAI,eAAe,KAAM,cAAa,UAAU;AAChD,uBAAa,WAAW,MAAM;AAC5B,yBAAa;AACb,8BAAkB,QAAQ,EAAE,MAAM,CAAC,QAAiB;AAClD,sBAAQ,MAAM,+BAA0B,eAAe,QAAQ,IAAI,UAAU,GAAG;AAAA,YAClF,CAAC;AAAA,UACH,GAAG,GAAG;AAAA,QACR,CAAC;AAED,0BAAkB,QAAQ;AAC1B,cAAM;AAAA,MACR,OAAO;AACL,cAAM,QAAQ;AAAA,MAChB;AAAA,IACF,SAAS,OAAO;AACd,cAAQ,MAAM,+BAA0B,iBAAiB,QAAQ,MAAM,UAAU,KAAK;AACtF,cAAQ,KAAK,CAAC;AAAA,IAChB;AAAA,EACF,CAAC;AAGH,UACG,QAAQ,MAAM,EACd,YAAY,mEAAmE,EAC/E,YAAY,OAAO,EACnB;AAAA,IACC,IAAI,OAAO,iBAAiB,cAAc,kBAAkB,0BAA0B;AAAA,EACxF,EACC,OAAO,OAAO,OAAiB,YAAmC;AACjE,QAAI;AACF,YAAM,eAAe,OAAO,OAAO;AAAA,IACrC,SAAS,OAAO;AACd,cAAQ,MAAM,+BAA0B,iBAAiB,QAAQ,MAAM,UAAU,KAAK;AACtF,cAAQ,KAAK,CAAC;AAAA,IAChB;AAAA,EACF,CAAC;AAEH,QAAM,QAAQ,WAAW,IAAI;AAC/B;AAEA,IAAI,gBAAgB,GAAG;AACrB,SAAO,EAAE,MAAM,CAAC,UAAmB;AACjC,YAAQ,MAAM,+BAA0B,iBAAiB,QAAQ,MAAM,UAAU,KAAK;AACtF,YAAQ,KAAK,CAAC;AAAA,EAChB,CAAC;AACH;",
|
|
6
6
|
"names": ["name", "resolve", "parseHTML", "require"]
|
|
7
7
|
}
|