···
1
+
import { describe, test, expect } from 'bun:test'
2
+
import { rewriteHtmlPaths, isHtmlContent } from './html-rewriter'
4
+
describe('rewriteHtmlPaths', () => {
5
+
const basePath = '/identifier/site/'
7
+
describe('absolute paths', () => {
8
+
test('rewrites absolute paths with leading slash', () => {
9
+
const html = '<img src="/image.png">'
10
+
const result = rewriteHtmlPaths(html, basePath, 'index.html')
11
+
expect(result).toBe('<img src="/identifier/site/image.png">')
14
+
test('rewrites nested absolute paths', () => {
15
+
const html = '<link href="/css/style.css">'
16
+
const result = rewriteHtmlPaths(html, basePath, 'index.html')
17
+
expect(result).toBe('<link href="/identifier/site/css/style.css">')
21
+
describe('relative paths from root document', () => {
22
+
test('rewrites relative paths with ./ prefix', () => {
23
+
const html = '<img src="./image.png">'
24
+
const result = rewriteHtmlPaths(html, basePath, 'index.html')
25
+
expect(result).toBe('<img src="/identifier/site/image.png">')
28
+
test('rewrites relative paths without prefix', () => {
29
+
const html = '<img src="image.png">'
30
+
const result = rewriteHtmlPaths(html, basePath, 'index.html')
31
+
expect(result).toBe('<img src="/identifier/site/image.png">')
34
+
test('rewrites relative paths with ../ (should stay at root)', () => {
35
+
const html = '<img src="../image.png">'
36
+
const result = rewriteHtmlPaths(html, basePath, 'index.html')
37
+
expect(result).toBe('<img src="/identifier/site/image.png">')
41
+
describe('relative paths from nested documents', () => {
42
+
test('rewrites relative path from nested document', () => {
43
+
const html = '<img src="./photo.jpg">'
44
+
const result = rewriteHtmlPaths(
47
+
'folder1/folder2/index.html'
49
+
expect(result).toBe(
50
+
'<img src="/identifier/site/folder1/folder2/photo.jpg">'
54
+
test('rewrites plain filename from nested document', () => {
55
+
const html = '<script src="app.js"></script>'
56
+
const result = rewriteHtmlPaths(
59
+
'folder1/folder2/index.html'
61
+
expect(result).toBe(
62
+
'<script src="/identifier/site/folder1/folder2/app.js"></script>'
66
+
test('rewrites ../ to go up one level', () => {
67
+
const html = '<img src="../image.png">'
68
+
const result = rewriteHtmlPaths(
71
+
'folder1/folder2/folder3/index.html'
73
+
expect(result).toBe(
74
+
'<img src="/identifier/site/folder1/folder2/image.png">'
78
+
test('rewrites multiple ../ to go up multiple levels', () => {
79
+
const html = '<link href="../../css/style.css">'
80
+
const result = rewriteHtmlPaths(
83
+
'folder1/folder2/folder3/index.html'
85
+
expect(result).toBe(
86
+
'<link href="/identifier/site/folder1/css/style.css">'
90
+
test('rewrites ../ with additional path segments', () => {
91
+
const html = '<img src="../assets/logo.png">'
92
+
const result = rewriteHtmlPaths(
95
+
'pages/about/index.html'
97
+
expect(result).toBe(
98
+
'<img src="/identifier/site/pages/assets/logo.png">'
102
+
test('handles complex nested relative paths', () => {
103
+
const html = '<script src="../../lib/vendor/jquery.js"></script>'
104
+
const result = rewriteHtmlPaths(
107
+
'pages/blog/post/index.html'
109
+
expect(result).toBe(
110
+
'<script src="/identifier/site/pages/lib/vendor/jquery.js"></script>'
114
+
test('handles ../ going past root (stays at root)', () => {
115
+
const html = '<img src="../../../image.png">'
116
+
const result = rewriteHtmlPaths(html, basePath, 'folder1/index.html')
117
+
expect(result).toBe('<img src="/identifier/site/image.png">')
121
+
describe('external URLs and special schemes', () => {
122
+
test('does not rewrite http URLs', () => {
123
+
const html = '<img src="http://example.com/image.png">'
124
+
const result = rewriteHtmlPaths(html, basePath, 'index.html')
125
+
expect(result).toBe('<img src="http://example.com/image.png">')
128
+
test('does not rewrite https URLs', () => {
129
+
const html = '<link href="https://cdn.example.com/style.css">'
130
+
const result = rewriteHtmlPaths(html, basePath, 'index.html')
131
+
expect(result).toBe(
132
+
'<link href="https://cdn.example.com/style.css">'
136
+
test('does not rewrite protocol-relative URLs', () => {
137
+
const html = '<script src="//cdn.example.com/script.js"></script>'
138
+
const result = rewriteHtmlPaths(html, basePath, 'index.html')
139
+
expect(result).toBe(
140
+
'<script src="//cdn.example.com/script.js"></script>'
144
+
test('does not rewrite data URIs', () => {
146
+
'<img src="data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAAUA">'
147
+
const result = rewriteHtmlPaths(html, basePath, 'index.html')
148
+
expect(result).toBe(
149
+
'<img src="data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAAUA">'
153
+
test('does not rewrite mailto links', () => {
154
+
const html = '<a href="mailto:test@example.com">Email</a>'
155
+
const result = rewriteHtmlPaths(html, basePath, 'index.html')
156
+
expect(result).toBe('<a href="mailto:test@example.com">Email</a>')
159
+
test('does not rewrite tel links', () => {
160
+
const html = '<a href="tel:+1234567890">Call</a>'
161
+
const result = rewriteHtmlPaths(html, basePath, 'index.html')
162
+
expect(result).toBe('<a href="tel:+1234567890">Call</a>')
166
+
describe('different HTML attributes', () => {
167
+
test('rewrites src attribute', () => {
168
+
const html = '<img src="/image.png">'
169
+
const result = rewriteHtmlPaths(html, basePath, 'index.html')
170
+
expect(result).toBe('<img src="/identifier/site/image.png">')
173
+
test('rewrites href attribute', () => {
174
+
const html = '<a href="/page.html">Link</a>'
175
+
const result = rewriteHtmlPaths(html, basePath, 'index.html')
176
+
expect(result).toBe('<a href="/identifier/site/page.html">Link</a>')
179
+
test('rewrites action attribute', () => {
180
+
const html = '<form action="/submit"></form>'
181
+
const result = rewriteHtmlPaths(html, basePath, 'index.html')
182
+
expect(result).toBe('<form action="/identifier/site/submit"></form>')
185
+
test('rewrites data attribute', () => {
186
+
const html = '<object data="/document.pdf"></object>'
187
+
const result = rewriteHtmlPaths(html, basePath, 'index.html')
188
+
expect(result).toBe(
189
+
'<object data="/identifier/site/document.pdf"></object>'
193
+
test('rewrites poster attribute', () => {
194
+
const html = '<video poster="/thumbnail.jpg"></video>'
195
+
const result = rewriteHtmlPaths(html, basePath, 'index.html')
196
+
expect(result).toBe(
197
+
'<video poster="/identifier/site/thumbnail.jpg"></video>'
201
+
test('rewrites srcset attribute with single URL', () => {
202
+
const html = '<img srcset="/image.png 1x">'
203
+
const result = rewriteHtmlPaths(html, basePath, 'index.html')
204
+
expect(result).toBe(
205
+
'<img srcset="/identifier/site/image.png 1x">'
209
+
test('rewrites srcset attribute with multiple URLs', () => {
210
+
const html = '<img srcset="/image-1x.png 1x, /image-2x.png 2x">'
211
+
const result = rewriteHtmlPaths(html, basePath, 'index.html')
212
+
expect(result).toBe(
213
+
'<img srcset="/identifier/site/image-1x.png 1x, /identifier/site/image-2x.png 2x">'
217
+
test('rewrites srcset with width descriptors', () => {
218
+
const html = '<img srcset="/small.jpg 320w, /large.jpg 1024w">'
219
+
const result = rewriteHtmlPaths(html, basePath, 'index.html')
220
+
expect(result).toBe(
221
+
'<img srcset="/identifier/site/small.jpg 320w, /identifier/site/large.jpg 1024w">'
225
+
test('rewrites srcset with relative paths from nested document', () => {
226
+
const html = '<img srcset="../img1.png 1x, ../img2.png 2x">'
227
+
const result = rewriteHtmlPaths(
230
+
'folder1/folder2/index.html'
232
+
expect(result).toBe(
233
+
'<img srcset="/identifier/site/folder1/img1.png 1x, /identifier/site/folder1/img2.png 2x">'
238
+
describe('quote handling', () => {
239
+
test('handles double quotes', () => {
240
+
const html = '<img src="/image.png">'
241
+
const result = rewriteHtmlPaths(html, basePath, 'index.html')
242
+
expect(result).toBe('<img src="/identifier/site/image.png">')
245
+
test('handles single quotes', () => {
246
+
const html = "<img src='/image.png'>"
247
+
const result = rewriteHtmlPaths(html, basePath, 'index.html')
248
+
expect(result).toBe("<img src='/identifier/site/image.png'>")
251
+
test('handles mixed quotes in same document', () => {
252
+
const html = '<img src="/img1.png"><link href=\'/style.css\'>'
253
+
const result = rewriteHtmlPaths(html, basePath, 'index.html')
254
+
expect(result).toBe(
255
+
'<img src="/identifier/site/img1.png"><link href=\'/identifier/site/style.css\'>'
260
+
describe('multiple rewrites in same document', () => {
261
+
test('rewrites multiple attributes in complex HTML', () => {
266
+
<link href="/css/style.css" rel="stylesheet">
267
+
<script src="/js/app.js"></script>
270
+
<img src="/images/logo.png" alt="Logo">
271
+
<a href="/about.html">About</a>
272
+
<form action="/submit">
273
+
<button type="submit">Submit</button>
278
+
const result = rewriteHtmlPaths(html, basePath, 'index.html')
279
+
expect(result).toContain('href="/identifier/site/css/style.css"')
280
+
expect(result).toContain('src="/identifier/site/js/app.js"')
281
+
expect(result).toContain('src="/identifier/site/images/logo.png"')
282
+
expect(result).toContain('href="/identifier/site/about.html"')
283
+
expect(result).toContain('action="/identifier/site/submit"')
286
+
test('handles mix of relative and absolute paths', () => {
288
+
<img src="/abs/image.png">
289
+
<img src="./rel/image.png">
290
+
<img src="../parent/image.png">
291
+
<img src="https://external.com/image.png">
293
+
const result = rewriteHtmlPaths(
296
+
'folder1/folder2/page.html'
298
+
expect(result).toContain('src="/identifier/site/abs/image.png"')
299
+
expect(result).toContain(
300
+
'src="/identifier/site/folder1/folder2/rel/image.png"'
302
+
expect(result).toContain(
303
+
'src="/identifier/site/folder1/parent/image.png"'
305
+
expect(result).toContain('src="https://external.com/image.png"')
309
+
describe('edge cases', () => {
310
+
test('handles empty src attribute', () => {
311
+
const html = '<img src="">'
312
+
const result = rewriteHtmlPaths(html, basePath, 'index.html')
313
+
expect(result).toBe('<img src="">')
316
+
test('handles basePath without trailing slash', () => {
317
+
const html = '<img src="/image.png">'
318
+
const result = rewriteHtmlPaths(html, '/identifier/site', 'index.html')
319
+
expect(result).toBe('<img src="/identifier/site/image.png">')
322
+
test('handles basePath with trailing slash', () => {
323
+
const html = '<img src="/image.png">'
324
+
const result = rewriteHtmlPaths(
326
+
'/identifier/site/',
329
+
expect(result).toBe('<img src="/identifier/site/image.png">')
332
+
test('handles whitespace around equals sign', () => {
333
+
const html = '<img src = "/image.png">'
334
+
const result = rewriteHtmlPaths(html, basePath, 'index.html')
335
+
expect(result).toBe('<img src="/identifier/site/image.png">')
338
+
test('preserves query strings in URLs', () => {
339
+
const html = '<img src="/image.png?v=123">'
340
+
const result = rewriteHtmlPaths(html, basePath, 'index.html')
341
+
expect(result).toBe('<img src="/identifier/site/image.png?v=123">')
344
+
test('preserves hash fragments in URLs', () => {
345
+
const html = '<a href="/page.html#section">Link</a>'
346
+
const result = rewriteHtmlPaths(html, basePath, 'index.html')
347
+
expect(result).toBe(
348
+
'<a href="/identifier/site/page.html#section">Link</a>'
352
+
test('handles paths with special characters', () => {
353
+
const html = '<img src="/folder-name/file_name.png">'
354
+
const result = rewriteHtmlPaths(html, basePath, 'index.html')
355
+
expect(result).toBe(
356
+
'<img src="/identifier/site/folder-name/file_name.png">'
361
+
describe('real-world scenario', () => {
362
+
test('handles the example from the bug report', () => {
363
+
// HTML file at: /folder1/folder2/folder3/index.html
364
+
// Image at: /folder1/folder2/img.png
365
+
// Reference: src="../img.png"
366
+
const html = '<img src="../img.png">'
367
+
const result = rewriteHtmlPaths(
370
+
'folder1/folder2/folder3/index.html'
372
+
expect(result).toBe(
373
+
'<img src="/identifier/site/folder1/folder2/img.png">'
377
+
test('handles deeply nested static site structure', () => {
378
+
// A typical static site with nested pages and shared assets
383
+
<link href="../../css/style.css" rel="stylesheet">
384
+
<link href="../../css/theme.css" rel="stylesheet">
385
+
<script src="../../js/main.js"></script>
388
+
<img src="../../images/logo.png" alt="Logo">
389
+
<img src="./post-image.jpg" alt="Post">
390
+
<a href="../index.html">Back to Blog</a>
391
+
<a href="../../index.html">Home</a>
395
+
const result = rewriteHtmlPaths(
398
+
'blog/posts/my-post.html'
401
+
// Assets two levels up
402
+
expect(result).toContain('href="/identifier/site/css/style.css"')
403
+
expect(result).toContain('href="/identifier/site/css/theme.css"')
404
+
expect(result).toContain('src="/identifier/site/js/main.js"')
405
+
expect(result).toContain('src="/identifier/site/images/logo.png"')
408
+
expect(result).toContain(
409
+
'src="/identifier/site/blog/posts/post-image.jpg"'
413
+
expect(result).toContain('href="/identifier/site/blog/index.html"')
416
+
expect(result).toContain('href="/identifier/site/index.html"')
421
+
describe('isHtmlContent', () => {
422
+
test('identifies HTML by content type', () => {
423
+
expect(isHtmlContent('file.txt', 'text/html')).toBe(true)
424
+
expect(isHtmlContent('file.txt', 'text/html; charset=utf-8')).toBe(
429
+
test('identifies HTML by .html extension', () => {
430
+
expect(isHtmlContent('index.html')).toBe(true)
431
+
expect(isHtmlContent('page.html', undefined)).toBe(true)
432
+
expect(isHtmlContent('/path/to/file.html')).toBe(true)
435
+
test('identifies HTML by .htm extension', () => {
436
+
expect(isHtmlContent('index.htm')).toBe(true)
437
+
expect(isHtmlContent('page.htm', undefined)).toBe(true)
440
+
test('handles case-insensitive extensions', () => {
441
+
expect(isHtmlContent('INDEX.HTML')).toBe(true)
442
+
expect(isHtmlContent('page.HTM')).toBe(true)
443
+
expect(isHtmlContent('File.HtMl')).toBe(true)
446
+
test('returns false for non-HTML files', () => {
447
+
expect(isHtmlContent('script.js')).toBe(false)
448
+
expect(isHtmlContent('style.css')).toBe(false)
449
+
expect(isHtmlContent('image.png')).toBe(false)
450
+
expect(isHtmlContent('data.json')).toBe(false)
453
+
test('returns false for files with no extension', () => {
454
+
expect(isHtmlContent('README')).toBe(false)
455
+
expect(isHtmlContent('Makefile')).toBe(false)
ptr.pet