File
Blob: src/workerd/api/tests/htmlrewriter-test.js
| 1 | // Copyright (c) 2024 Cloudflare, Inc. |
| 2 | // Licensed under the Apache 2.0 license found in the LICENSE file or at: |
| 3 | // https://opensource.org/licenses/Apache-2.0 |
| 4 | import { |
| 5 | strictEqual, |
| 6 | deepStrictEqual, |
| 7 | notStrictEqual, |
| 8 | rejects, |
| 9 | throws, |
| 10 | } from 'node:assert'; |
| 11 | |
| 12 | const arrayIterator = [][Symbol.iterator](); |
| 13 | const arrayIteratorPrototype = Object.getPrototypeOf( |
| 14 | Object.getPrototypeOf(arrayIterator) |
| 15 | ); |
| 16 | |
| 17 | export const passthroughWithContent = { |
| 18 | async test() { |
| 19 | const response = new Response('hello', { |
| 20 | headers: { |
| 21 | foo: 'bar', |
| 22 | }, |
| 23 | }); |
| 24 | strictEqual(response.headers.get('foo'), 'bar'); |
| 25 | |
| 26 | const rewriter = new HTMLRewriter(); |
| 27 | |
| 28 | const newResponse = rewriter.transform(response); |
| 29 | |
| 30 | notStrictEqual(response, newResponse); |
| 31 | strictEqual(newResponse.headers.get('foo'), 'bar'); |
| 32 | strictEqual(await newResponse.text(), 'hello'); |
| 33 | }, |
| 34 | }; |
| 35 | |
| 36 | export const passthroughWithContentAndHandler = { |
| 37 | async test() { |
| 38 | const response = new Response('<h3>hello</h3>', { |
| 39 | headers: { |
| 40 | foo: 'bar', |
| 41 | }, |
| 42 | }); |
| 43 | strictEqual(response.headers.get('foo'), 'bar'); |
| 44 | |
| 45 | const rewriter = new HTMLRewriter().on('h3', { |
| 46 | element() {}, |
| 47 | comments() {}, |
| 48 | text() {}, |
| 49 | }); |
| 50 | |
| 51 | const newResponse = rewriter.transform(response); |
| 52 | |
| 53 | notStrictEqual(response, newResponse); |
| 54 | strictEqual(newResponse.headers.get('foo'), 'bar'); |
| 55 | strictEqual(await newResponse.text(), '<h3>hello</h3>'); |
| 56 | }, |
| 57 | }; |
| 58 | |
| 59 | export const passthroughWithContentAndAsyncHandler = { |
| 60 | async test() { |
| 61 | const response = new Response('<h3>hello</h3>', { |
| 62 | headers: { |
| 63 | foo: 'bar', |
| 64 | }, |
| 65 | }); |
| 66 | strictEqual(response.headers.get('foo'), 'bar'); |
| 67 | |
| 68 | const rewriter = new HTMLRewriter().on('h3', { |
| 69 | async test() { |
| 70 | await scheduler.wait(10); |
| 71 | }, |
| 72 | async element() { |
| 73 | await scheduler.wait(10); |
| 74 | }, |
| 75 | async comments() { |
| 76 | await scheduler.wait(10); |
| 77 | }, |
| 78 | }); |
| 79 | |
| 80 | const newResponse = rewriter.transform(response); |
| 81 | |
| 82 | notStrictEqual(response, newResponse); |
| 83 | strictEqual(newResponse.headers.get('foo'), 'bar'); |
| 84 | strictEqual(await newResponse.text(), '<h3>hello</h3>'); |
| 85 | }, |
| 86 | }; |
| 87 | |
| 88 | export const passthroughWithContentAndAsyncHandler2 = { |
| 89 | async test() { |
| 90 | const response = new Response('<h3>hello</h3>', { |
| 91 | headers: { |
| 92 | foo: 'bar', |
| 93 | }, |
| 94 | }); |
| 95 | strictEqual(response.headers.get('foo'), 'bar'); |
| 96 | |
| 97 | const { promise, resolve } = Promise.withResolvers(); |
| 98 | |
| 99 | const rewriter = new HTMLRewriter().on('h3', { |
| 100 | async test() { |
| 101 | await promise; |
| 102 | }, |
| 103 | async element() { |
| 104 | await scheduler.wait(10); |
| 105 | }, |
| 106 | async comments() { |
| 107 | await scheduler.wait(10); |
| 108 | }, |
| 109 | }); |
| 110 | |
| 111 | const newResponse = rewriter.transform(response); |
| 112 | |
| 113 | // We can resolve the promise after calling transform and the transform |
| 114 | // will complete as expected. |
| 115 | resolve(); |
| 116 | |
| 117 | notStrictEqual(response, newResponse); |
| 118 | strictEqual(newResponse.headers.get('foo'), 'bar'); |
| 119 | strictEqual(await newResponse.text(), '<h3>hello</h3>'); |
| 120 | }, |
| 121 | }; |
| 122 | |
| 123 | export const passthroughWithContentStream = { |
| 124 | async test() { |
| 125 | const { readable, writable } = new TransformStream(); |
| 126 | |
| 127 | const rewriter = new HTMLRewriter().on('h3', { |
| 128 | async text(content) { |
| 129 | await scheduler.wait(10); |
| 130 | }, |
| 131 | }); |
| 132 | |
| 133 | const response = rewriter.transform(new Response(readable)); |
| 134 | |
| 135 | const writer = writable.getWriter(); |
| 136 | const enc = new TextEncoder(); |
| 137 | const results = await Promise.all([ |
| 138 | response.text(), |
| 139 | writer.write(enc.encode('<h3>hello</h3>')), |
| 140 | writer.close(), |
| 141 | ]); |
| 142 | |
| 143 | strictEqual(results[0], '<h3>hello</h3>'); |
| 144 | }, |
| 145 | }; |
| 146 | |
| 147 | export const passthroughWithEmptyStream = { |
| 148 | async test() { |
| 149 | const { readable, writable } = new TransformStream(); |
| 150 | |
| 151 | const rewriter = new HTMLRewriter().on('h3', { |
| 152 | async text(content) { |
| 153 | await scheduler.wait(10); |
| 154 | }, |
| 155 | }); |
| 156 | |
| 157 | const response = rewriter.transform(new Response(readable)); |
| 158 | |
| 159 | const writer = writable.getWriter(); |
| 160 | const results = await Promise.all([response.text(), writer.close()]); |
| 161 | |
| 162 | strictEqual(results[0], ''); |
| 163 | }, |
| 164 | }; |
| 165 | |
| 166 | export const asyncElementHandler = { |
| 167 | async test() { |
| 168 | const rewriter = new HTMLRewriter().on('body', { |
| 169 | async element(e) { |
| 170 | await scheduler.wait(10); |
| 171 | e.setInnerContent('world'); |
| 172 | }, |
| 173 | }); |
| 174 | |
| 175 | const response = rewriter.transform(new Response('<body>hello</body>')); |
| 176 | |
| 177 | strictEqual(await response.text(), '<body>world</body>'); |
| 178 | }, |
| 179 | }; |
| 180 | |
| 181 | export const asyncCommentHandler = { |
| 182 | async test() { |
| 183 | const rewriter = new HTMLRewriter().on('body', { |
| 184 | async comments(comment) { |
| 185 | await scheduler.wait(10); |
| 186 | if (comment.text == 'hello') { |
| 187 | comment.text = 'world'; |
| 188 | } |
| 189 | }, |
| 190 | }); |
| 191 | |
| 192 | const response = rewriter.transform( |
| 193 | new Response('<body><!--hello--></body>') |
| 194 | ); |
| 195 | |
| 196 | strictEqual(await response.text(), '<body><!--world--></body>'); |
| 197 | }, |
| 198 | }; |
| 199 | |
| 200 | export const objectHandlers = { |
| 201 | async test() { |
| 202 | class DocumentContentHandlers { |
| 203 | deadTokens = {}; |
| 204 | doctypeCount = 0; |
| 205 | commentCount = 0; |
| 206 | textCount = 0; |
| 207 | expectedErrors = []; |
| 208 | |
| 209 | doctype(token) { |
| 210 | if (!this.deadTokens.doctype) { |
| 211 | this.deadTokens.doctype = token; |
| 212 | } |
| 213 | ++this.doctypeCount; |
| 214 | |
| 215 | this.sawDoctype = JSON.stringify(token); |
| 216 | } |
| 217 | |
| 218 | comments(token) { |
| 219 | if (!this.deadTokens.comment) { |
| 220 | this.deadTokens.comment = token; |
| 221 | } |
| 222 | ++this.commentCount; |
| 223 | } |
| 224 | |
| 225 | text(token) { |
| 226 | if (!this.deadTokens.text) { |
| 227 | this.deadTokens.text = token; |
| 228 | } |
| 229 | ++this.textCount; |
| 230 | } |
| 231 | |
| 232 | end(token) { |
| 233 | this.reachedEnd = true; |
| 234 | } |
| 235 | } |
| 236 | |
| 237 | class ElementContentHandlers { |
| 238 | deadTokens = {}; |
| 239 | elementCount = 0; |
| 240 | commentCount = 0; |
| 241 | textCount = 0; |
| 242 | expectedErrors = []; |
| 243 | |
| 244 | element(token) { |
| 245 | if (!this.deadTokens.element) { |
| 246 | this.deadTokens.element = token; |
| 247 | } |
| 248 | if (!this.deadTokens.attributesIterator) { |
| 249 | this.deadTokens.attributesIterator = token.attributes; |
| 250 | } |
| 251 | ++this.elementCount; |
| 252 | |
| 253 | // Exercise all the different methods on Element. |
| 254 | if ( |
| 255 | token.tagName === 'body' && |
| 256 | token.hasAttribute('foo') && |
| 257 | !token.hasAttribute('baz') && |
| 258 | token.getAttribute('foo') === 'bar' |
| 259 | ) { |
| 260 | token.removeAttribute('foo'); |
| 261 | token.setAttribute('baz', 'qux'); |
| 262 | |
| 263 | try { |
| 264 | token.tagName = 'should throw'; |
| 265 | throw new Error('should have thrown'); |
| 266 | } catch (e) { |
| 267 | this.expectedErrors.push(e.message); |
| 268 | } |
| 269 | |
| 270 | token.tagName = 'tail'; |
| 271 | |
| 272 | // These will show up in order in the response body. |
| 273 | token.before('<1>'); |
| 274 | token.before('<2>', { html: false }); |
| 275 | token.before('<3>\n', null); |
| 276 | token.before('<html>', { html: true }); |
| 277 | |
| 278 | // These will show up in reverse order in the response body. |
| 279 | token.prepend('<em>hello</em>, ', { html: true }); |
| 280 | token.prepend('<6>\n'); |
| 281 | token.prepend('<5>', { html: false }); |
| 282 | token.prepend('\n<4>', null); |
| 283 | |
| 284 | // Iterator tests. |
| 285 | |
| 286 | this.sawAttributes = JSON.stringify([...token.attributes]); |
| 287 | |
| 288 | let iterator = token.attributes; |
| 289 | let iteratorPrototype = Object.getPrototypeOf( |
| 290 | Object.getPrototypeOf(iterator) |
| 291 | ); |
| 292 | if (iteratorPrototype !== arrayIteratorPrototype) { |
| 293 | throw new Error( |
| 294 | 'attributes iterator does not have iterator prototype' |
| 295 | ); |
| 296 | } |
| 297 | |
| 298 | // Run the iterator down until it's done. |
| 299 | for (let [_k, _v] of iterator) { |
| 300 | // intentionally empty |
| 301 | } |
| 302 | // .next() should now be idempotent. |
| 303 | let result = iterator.next(); |
| 304 | let result2 = iterator.next(); |
| 305 | if ( |
| 306 | result.done !== result2.done || |
| 307 | result.value !== result2.value || |
| 308 | !result.done || |
| 309 | result.value |
| 310 | ) { |
| 311 | throw new Error( |
| 312 | 'exhausted iterator should continually return done' |
| 313 | ); |
| 314 | } |
| 315 | } else if (token.tagName === 'remove') { |
| 316 | let mode = token.getAttribute('mode'); |
| 317 | if (mode === null) { |
| 318 | throw new Error("missing attribute on 'remove' element"); |
| 319 | } |
| 320 | if (token.removed) { |
| 321 | throw new Error('element should not have been removed yet'); |
| 322 | } |
| 323 | |
| 324 | if (mode === 'all') { |
| 325 | token.remove(); |
| 326 | } else { |
| 327 | token.removeAndKeepContent(); |
| 328 | } |
| 329 | |
| 330 | if (!token.removed) { |
| 331 | throw new Error('element should have been removed now'); |
| 332 | } |
| 333 | } else if (token.tagName === 'after') { |
| 334 | let isHtml = token.getAttribute('is-html'); |
| 335 | let html = isHtml === 'true' ? true : false; |
| 336 | token.after('<after>', { html }); |
| 337 | } else if (token.tagName === 'append') { |
| 338 | let isHtml = token.getAttribute('is-html'); |
| 339 | let html = isHtml === 'true' ? true : false; |
| 340 | token.append('<append>', { html }); |
| 341 | } else if (token.tagName === 'replace') { |
| 342 | let isHtml = token.getAttribute('is-html'); |
| 343 | let html = isHtml === 'true' ? true : false; |
| 344 | token.replace('<replace>', { html }); |
| 345 | } else if (token.tagName === 'set-inner-content') { |
| 346 | let isHtml = token.getAttribute('is-html'); |
| 347 | let html = isHtml === 'true' ? true : false; |
| 348 | token.setInnerContent('<set-inner-content>', { html }); |
| 349 | } else if (token.tagName === 'set-attribute') { |
| 350 | if (!token.hasAttribute('foo')) { |
| 351 | throw new Error('element should have had attribute'); |
| 352 | } |
| 353 | let attr = token.getAttribute('foo'); |
| 354 | if (attr !== '') { |
| 355 | throw new Error('element attribute should have been empty'); |
| 356 | } |
| 357 | token.setAttribute('foo', 'bar'); |
| 358 | |
| 359 | if (token.getAttribute('nonexistent')) { |
| 360 | throw new Error('attribute should not exist'); |
| 361 | } |
| 362 | } |
| 363 | } |
| 364 | |
| 365 | comments(token) { |
| 366 | if (!this.deadTokens.comment) { |
| 367 | this.deadTokens.comment = token; |
| 368 | } |
| 369 | ++this.commentCount; |
| 370 | |
| 371 | // Exercise all the different methods on Comment. |
| 372 | if (token.text === ' SET TEXT PROPERTY ') { |
| 373 | token.text = ' text property has been set '; |
| 374 | } else if (token.text === ' REMOVE ME ') { |
| 375 | if (token.removed) { |
| 376 | throw new Error("Shouldn't be removed yet"); |
| 377 | } |
| 378 | |
| 379 | token.remove(); |
| 380 | |
| 381 | if (!token.removed) { |
| 382 | throw new Error('Should be removed now'); |
| 383 | } |
| 384 | } else if (token.text === ' REPLACE ME ') { |
| 385 | if (token.removed) { |
| 386 | throw new Error("Shouldn't be removed yet"); |
| 387 | } |
| 388 | |
| 389 | token.replace('this will get overwritten'); |
| 390 | |
| 391 | if (!token.removed) { |
| 392 | throw new Error('Should be removed now'); |
| 393 | } |
| 394 | |
| 395 | token.replace('<REPLACED>', null); |
| 396 | |
| 397 | if (!token.removed) { |
| 398 | throw new Error('Should still be removed'); |
| 399 | } |
| 400 | |
| 401 | token.before('<!-- ', { html: true }); |
| 402 | token.after(' -->', { html: true }); |
| 403 | } |
| 404 | } |
| 405 | |
| 406 | text(token) { |
| 407 | if (!this.deadTokens.text) { |
| 408 | this.deadTokens.text = token; |
| 409 | } |
| 410 | ++this.textCount; |
| 411 | |
| 412 | if (token.lastInTextNode && token.text.length > 0) { |
| 413 | throw new Error('last text chunk has non-zero length'); |
| 414 | } else if (!token.lastInTextNode && token.text.length === 0) { |
| 415 | throw new Error('non-last text chunk has zero length'); |
| 416 | } |
| 417 | |
| 418 | if (token.text === 'world') { |
| 419 | token.before('again, '); |
| 420 | token.after('...'); |
| 421 | |
| 422 | if (token.removed) { |
| 423 | throw new Error("Shouldn't be removed yet"); |
| 424 | } |
| 425 | |
| 426 | token.replace('this will get overwritten'); |
| 427 | |
| 428 | if (!token.removed) { |
| 429 | throw new Error('Should be removed now'); |
| 430 | } |
| 431 | |
| 432 | token.replace('<WORLD>', { html: true }); |
| 433 | |
| 434 | if (!token.removed) { |
| 435 | throw new Error('Should still be removed'); |
| 436 | } |
| 437 | } else if (token.text === 'REMOVE ME\n') { |
| 438 | if (token.removed) { |
| 439 | throw new Error("Shouldn't be removed yet"); |
| 440 | } |
| 441 | |
| 442 | token.remove(); |
| 443 | |
| 444 | if (!token.removed) { |
| 445 | throw new Error('Should be removed now'); |
| 446 | } |
| 447 | } |
| 448 | } |
| 449 | } |
| 450 | |
| 451 | let documentHandlers = new DocumentContentHandlers(); |
| 452 | let elementHandlers = new ElementContentHandlers(); |
| 453 | |
| 454 | const rewriter = new HTMLRewriter() |
| 455 | .onDocument(documentHandlers) |
| 456 | .on('*', elementHandlers); |
| 457 | |
| 458 | let _count = 0; |
| 459 | |
| 460 | const enc = new TextEncoder(); |
| 461 | const kInput = [ |
| 462 | '<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN">', |
| 463 | '<!-- document-level comment -->', |
| 464 | 'document-level text', |
| 465 | '<body foo="bar" an-attr="ibute?">world<p>REMOVE ME', |
| 466 | '<remove mode="all">inner content</remove>', |
| 467 | '<remove mode="keep">inner content</remove>', |
| 468 | '<after is-html="true">inner content</after>', |
| 469 | '<after is-html="false">inner content</after>', |
| 470 | '<append is-html="true">inner content</append>', |
| 471 | '<append is-html="false">inner content</append>', |
| 472 | '<replace is-html="true">inner content</replace>', |
| 473 | '<replace is-html="false">inner content</replace>', |
| 474 | '<set-inner-content is-html="true">inner content</set-inner-content>', |
| 475 | '<set-inner-content is-html="false">inner content</set-inner-content>', |
| 476 | '<set-attribute foo></set-attribute>', |
| 477 | '<set-attribute foo=></set-attribute>', |
| 478 | '<set-attribute foo=""></set-attribute>', |
| 479 | '<!-- REMOVE ME -->', |
| 480 | '<!-- REPLACE ME -->', |
| 481 | '<!-- SET TEXT PROPERTY -->', |
| 482 | '</body>', |
| 483 | ]; |
| 484 | |
| 485 | const kResult = `<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"><!-- document-level comment -->document-level text<1><2><3> |
| 486 | <html><tail an-attr="ibute?" baz="qux"> |
| 487 | <4><5><6> |
| 488 | <em>hello</em>, again, <WORLD>...<p>REMOVE MEinner content<after is-html="true">inner content</after><after><after is-html="false">inner content</after><after><append is-html="true">inner content<append></append><append is-html="false">inner content<append></append><replace><replace><set-inner-content is-html="true"><set-inner-content></set-inner-content><set-inner-content is-html="false"><set-inner-content></set-inner-content><set-attribute foo="bar"></set-attribute><set-attribute foo="bar"></set-attribute><set-attribute foo="bar"></set-attribute><!-- <REPLACED> --><!-- text property has been set --></tail>`; |
| 489 | |
| 490 | const readable = new ReadableStream({ |
| 491 | async pull(controller) { |
| 492 | await scheduler.wait(1); |
| 493 | if (kInput.length > 0) { |
| 494 | controller.enqueue(enc.encode(kInput.shift())); |
| 495 | } else { |
| 496 | controller.close(); |
| 497 | } |
| 498 | }, |
| 499 | }); |
| 500 | |
| 501 | const response = rewriter.transform(new Response(readable)); |
| 502 | |
| 503 | // At this point, we should not have seen any tokens. |
| 504 | strictEqual(elementHandlers.deadTokens.element, undefined); |
| 505 | |
| 506 | strictEqual(await response.text(), kResult); |
| 507 | |
| 508 | // Now we've seen tokens |
| 509 | notStrictEqual(elementHandlers.deadTokens.element, undefined); |
| 510 | |
| 511 | // Verify that tokens are invalidated outside handler execution scope. |
| 512 | throws(() => documentHandlers.deadTokens.doctype.publicId, { |
| 513 | message: |
| 514 | 'This content token is no longer valid. Content tokens are only ' + |
| 515 | 'valid during the execution of the relevant content handler.', |
| 516 | }); |
| 517 | throws(() => documentHandlers.deadTokens.comment.text, { |
| 518 | message: |
| 519 | 'This content token is no longer valid. Content tokens are only ' + |
| 520 | 'valid during the execution of the relevant content handler.', |
| 521 | }); |
| 522 | throws(() => documentHandlers.deadTokens.text.text, { |
| 523 | message: |
| 524 | 'This content token is no longer valid. Content tokens are only ' + |
| 525 | 'valid during the execution of the relevant content handler.', |
| 526 | }); |
| 527 | throws(() => elementHandlers.deadTokens.element.getAttribute('foo'), { |
| 528 | message: |
| 529 | 'This content token is no longer valid. Content tokens are only ' + |
| 530 | 'valid during the execution of the relevant content handler.', |
| 531 | }); |
| 532 | throws(() => elementHandlers.deadTokens.attributesIterator.next(), { |
| 533 | message: |
| 534 | 'This content token is no longer valid. Content tokens are only ' + |
| 535 | 'valid during the execution of the relevant content handler.', |
| 536 | }); |
| 537 | throws(() => elementHandlers.deadTokens.comment.text, { |
| 538 | message: |
| 539 | 'This content token is no longer valid. Content tokens are only ' + |
| 540 | 'valid during the execution of the relevant content handler.', |
| 541 | }); |
| 542 | throws(() => elementHandlers.deadTokens.text.text, { |
| 543 | message: |
| 544 | 'This content token is no longer valid. Content tokens are only ' + |
| 545 | 'valid during the execution of the relevant content handler.', |
| 546 | }); |
| 547 | |
| 548 | strictEqual(documentHandlers.doctypeCount, 1); |
| 549 | strictEqual(documentHandlers.commentCount, 4); |
| 550 | strictEqual(documentHandlers.textCount, 26); |
| 551 | strictEqual(documentHandlers.reachedEnd, true); |
| 552 | strictEqual(elementHandlers.elementCount, 15); |
| 553 | strictEqual(elementHandlers.commentCount, 3); |
| 554 | strictEqual(elementHandlers.textCount, 24); |
| 555 | deepStrictEqual(elementHandlers.expectedErrors, [ |
| 556 | 'Parser error: ` ` character is forbidden in the tag name', |
| 557 | ]); |
| 558 | }, |
| 559 | }; |
| 560 | |
| 561 | export const manualWriting = { |
| 562 | async test() { |
| 563 | const { readable, writable } = new IdentityTransformStream(); |
| 564 | |
| 565 | const response = new HTMLRewriter() |
| 566 | .on('*', { |
| 567 | element(element) { |
| 568 | element.prepend('foo '); |
| 569 | }, |
| 570 | }) |
| 571 | .transform(new Response(readable)); |
| 572 | |
| 573 | const writer = writable.getWriter(); |
| 574 | const encoder = new TextEncoder(); |
| 575 | |
| 576 | // Because this variation uses IdentityTransformStream, we must |
| 577 | // initiate the read before doing the writes. |
| 578 | const promise = response.text(); |
| 579 | |
| 580 | await writer.write(encoder.encode('<html>')); |
| 581 | await writer.write(encoder.encode('bar')); |
| 582 | await writer.write(encoder.encode('</html>')); |
| 583 | await writer.close(); |
| 584 | |
| 585 | strictEqual(await promise, '<html>foo bar</html>'); |
| 586 | }, |
| 587 | }; |
| 588 | |
| 589 | export const manualWriting2 = { |
| 590 | async test() { |
| 591 | const { readable, writable } = new TransformStream(); |
| 592 | |
| 593 | const response = new HTMLRewriter() |
| 594 | .on('*', { |
| 595 | element(element) { |
| 596 | element.prepend('foo '); |
| 597 | }, |
| 598 | }) |
| 599 | .transform(new Response(readable)); |
| 600 | |
| 601 | const writer = writable.getWriter(); |
| 602 | const encoder = new TextEncoder(); |
| 603 | |
| 604 | await writer.write(encoder.encode('<html>')); |
| 605 | await writer.write(encoder.encode('bar')); |
| 606 | await writer.write(encoder.encode('</html>')); |
| 607 | await writer.close(); |
| 608 | |
| 609 | // This variation uses the JavaScript TransformStream, so we can |
| 610 | // initiate the read after doing the writes. |
| 611 | const promise = response.text(); |
| 612 | |
| 613 | strictEqual(await promise, '<html>foo bar</html>'); |
| 614 | }, |
| 615 | }; |
| 616 | |
| 617 | export const streamingReplacement = { |
| 618 | async test() { |
| 619 | const { readable, writable } = new TransformStream(); |
| 620 | |
| 621 | const response = new HTMLRewriter() |
| 622 | .on('*', { |
| 623 | async element(element) { |
| 624 | const dataStream = ( |
| 625 | await fetch('data:,the quick brown fox jumped over the lazy dog%20') |
| 626 | ).body; |
| 627 | element.prepend(dataStream, { html: false }); |
| 628 | }, |
| 629 | }) |
| 630 | .transform(new Response(readable)); |
| 631 | |
| 632 | const writer = writable.getWriter(); |
| 633 | const encoder = new TextEncoder(); |
| 634 | |
| 635 | await writer.write(encoder.encode('<html>')); |
| 636 | await writer.write(encoder.encode('bar')); |
| 637 | await writer.write(encoder.encode('</html>')); |
| 638 | await writer.close(); |
| 639 | |
| 640 | // This variation uses the JavaScript TransformStream, so we can |
| 641 | // initiate the read after doing the writes. |
| 642 | const promise = response.text(); |
| 643 | strictEqual( |
| 644 | await promise, |
| 645 | `<html>the quick brown fox jumped over the lazy dog bar</html>` |
| 646 | ); |
| 647 | }, |
| 648 | }; |
| 649 | |
| 650 | export const streamingReplacementHTML = { |
| 651 | async test() { |
| 652 | const { readable, writable } = new TransformStream(); |
| 653 | |
| 654 | const response = new HTMLRewriter() |
| 655 | .on('*', { |
| 656 | async element(element) { |
| 657 | const dataStream = ( |
| 658 | await fetch('data:,<b>such markup <i>much wow</i></b> ') |
| 659 | ).body; |
| 660 | element.prepend(dataStream, { html: true }); |
| 661 | }, |
| 662 | }) |
| 663 | .transform(new Response(readable)); |
| 664 | |
| 665 | const writer = writable.getWriter(); |
| 666 | const encoder = new TextEncoder(); |
| 667 | |
| 668 | await writer.write(encoder.encode('<html>')); |
| 669 | await writer.write(encoder.encode('bar')); |
| 670 | await writer.write(encoder.encode('</html>')); |
| 671 | await writer.close(); |
| 672 | |
| 673 | // This variation uses the JavaScript TransformStream, so we can |
| 674 | // initiate the read after doing the writes. |
| 675 | const promise = response.text(); |
| 676 | strictEqual( |
| 677 | await promise, |
| 678 | `<html><b>such markup <i>much wow</i></b>bar</html>` |
| 679 | ); |
| 680 | }, |
| 681 | }; |
| 682 | |
| 683 | export const streamingReplacementReplace = { |
| 684 | async test() { |
| 685 | const { readable, writable } = new TransformStream(); |
| 686 | |
| 687 | const response = new HTMLRewriter() |
| 688 | .on('.dinosaur', { |
| 689 | async element(element) { |
| 690 | const dataStream = (await fetch('data:,goodbye world')).body; |
| 691 | element.replace(dataStream, { html: false }); |
| 692 | }, |
| 693 | }) |
| 694 | .transform(new Response(readable)); |
| 695 | |
| 696 | const writer = writable.getWriter(); |
| 697 | const encoder = new TextEncoder(); |
| 698 | |
| 699 | await writer.write(encoder.encode('<html>')); |
| 700 | await writer.write( |
| 701 | encoder.encode('<div class="dinosaur">hello world</div>') |
| 702 | ); |
| 703 | await writer.write(encoder.encode('</html>')); |
| 704 | await writer.close(); |
| 705 | |
| 706 | // This variation uses the JavaScript TransformStream, so we can |
| 707 | // initiate the read after doing the writes. |
| 708 | const promise = response.text(); |
| 709 | strictEqual(await promise, `<html>goodbye world</html>`); |
| 710 | }, |
| 711 | }; |
| 712 | |
| 713 | export const streamingReplacementMultiple = { |
| 714 | async test() { |
| 715 | const { readable, writable } = new TransformStream(); |
| 716 | |
| 717 | const response = new HTMLRewriter() |
| 718 | .on('*', { |
| 719 | async element(element) { |
| 720 | element.prepend(await fetch('data:,alpha%20')); |
| 721 | element.append(await fetch('data:,%20gamma')); |
| 722 | }, |
| 723 | }) |
| 724 | .transform(new Response(readable)); |
| 725 | |
| 726 | const writer = writable.getWriter(); |
| 727 | const encoder = new TextEncoder(); |
| 728 | |
| 729 | await writer.write(encoder.encode('<html>')); |
| 730 | await writer.write(encoder.encode('beta')); |
| 731 | await writer.write(encoder.encode('</html>')); |
| 732 | await writer.close(); |
| 733 | |
| 734 | // This variation uses the JavaScript TransformStream, so we can |
| 735 | // initiate the read after doing the writes. |
| 736 | const promise = response.text(); |
| 737 | strictEqual(await promise, `<html>alpha beta gamma</html>`); |
| 738 | }, |
| 739 | }; |
| 740 | |
| 741 | export const streamingReplacementBadUTF8 = { |
| 742 | async test() { |
| 743 | const { readable, writable } = new TransformStream(); |
| 744 | |
| 745 | const response = new HTMLRewriter() |
| 746 | .on('*', { |
| 747 | async element(element) { |
| 748 | element.prepend(await fetch('data:,garbage%e2%28%a1')); |
| 749 | }, |
| 750 | }) |
| 751 | .transform(new Response(readable)); |
| 752 | |
| 753 | const writer = writable.getWriter(); |
| 754 | const encoder = new TextEncoder(); |
| 755 | |
| 756 | await writer.write(encoder.encode('<html>')); |
| 757 | await writer.write(encoder.encode('bar')); |
| 758 | await writer.write(encoder.encode('</html>')); |
| 759 | await writer.close(); |
| 760 | |
| 761 | // This variation uses the JavaScript TransformStream, so we can |
| 762 | // initiate the read after doing the writes. |
| 763 | await rejects(response.text(), { message: 'Parser error: Invalid UTF-8' }); |
| 764 | }, |
| 765 | }; |
| 766 | |
| 767 | export const appendOnEnd = { |
| 768 | async test() { |
| 769 | const kInput = |
| 770 | '<!doctype html><html><head></head><body><!----></body></html>'; |
| 771 | const kSuffix = '<!-- an end comment -->'; |
| 772 | const result = new HTMLRewriter() |
| 773 | .onDocument({ |
| 774 | end(end) { |
| 775 | end.append(kSuffix, { html: true }); |
| 776 | }, |
| 777 | }) |
| 778 | .transform(new Response(kInput)); |
| 779 | strictEqual(await result.text(), kInput + kSuffix); |
| 780 | }, |
| 781 | }; |
| 782 | |
| 783 | export const interleavedAsyncHandlers = { |
| 784 | async test() { |
| 785 | class OneTimeBarrier { |
| 786 | constructor(limit) { |
| 787 | this.limit = limit; |
| 788 | this.current = 0; |
| 789 | let resolve; |
| 790 | this.promise = new Promise((r) => (resolve = r)); |
| 791 | this.resolver = resolve; |
| 792 | } |
| 793 | |
| 794 | async wait() { |
| 795 | this.current += 1; |
| 796 | if (this.current >= this.limit) { |
| 797 | this.resolver(); |
| 798 | } else { |
| 799 | await this.promise; |
| 800 | } |
| 801 | } |
| 802 | } |
| 803 | |
| 804 | const barrier = new OneTimeBarrier(2); |
| 805 | const responses = await Promise.all([ |
| 806 | new HTMLRewriter() |
| 807 | .on('body', { |
| 808 | async element(e) { |
| 809 | await barrier.wait(); |
| 810 | e.setInnerContent('foo bar'); |
| 811 | }, |
| 812 | }) |
| 813 | .transform(new Response('<body>body value</body>')) |
| 814 | .text(), |
| 815 | new HTMLRewriter() |
| 816 | .on('body', { |
| 817 | async element(e) { |
| 818 | await barrier.wait(); |
| 819 | e.remove(); |
| 820 | }, |
| 821 | }) |
| 822 | .transform( |
| 823 | new Response('<!doctype html><html><head></head><body></body></html>') |
| 824 | ) |
| 825 | .arrayBuffer(), |
| 826 | ]); |
| 827 | |
| 828 | const body = responses[0]; |
| 829 | const blank = responses[1]; |
| 830 | |
| 831 | const inserted = new HTMLRewriter() |
| 832 | .on('html', { |
| 833 | element(e) { |
| 834 | e.append(body, { html: true }); |
| 835 | }, |
| 836 | }) |
| 837 | .transform(new Response(blank)); |
| 838 | |
| 839 | strictEqual( |
| 840 | await inserted.text(), |
| 841 | '<!doctype html><html><head></head><body>foo bar</body></html>' |
| 842 | ); |
| 843 | }, |
| 844 | }; |
| 845 | |
| 846 | export const exceptionInHandler = { |
| 847 | async test() { |
| 848 | const response = new HTMLRewriter() |
| 849 | .on('*', { |
| 850 | text() { |
| 851 | throw new Error('boom'); |
| 852 | }, |
| 853 | }) |
| 854 | .transform(new Response('<body>hello</body>')); |
| 855 | |
| 856 | await rejects(response.text(), { |
| 857 | message: 'boom', |
| 858 | }); |
| 859 | }, |
| 860 | }; |
| 861 | |
| 862 | export const exceptionInAsyncHandler = { |
| 863 | async test() { |
| 864 | const response = new HTMLRewriter() |
| 865 | .on('*', { |
| 866 | async text() { |
| 867 | throw new Error('boom'); |
| 868 | }, |
| 869 | }) |
| 870 | .transform(new Response('<body>hello</body>')); |
| 871 | |
| 872 | await rejects(response.text(), { |
| 873 | message: 'boom', |
| 874 | }); |
| 875 | }, |
| 876 | }; |
| 877 | |
| 878 | export const invalidEncoding = { |
| 879 | async test() { |
| 880 | const response = new Response('hello', { |
| 881 | headers: { |
| 882 | 'content-type': 'text/html; charset=invalid', |
| 883 | }, |
| 884 | }); |
| 885 | |
| 886 | throws( |
| 887 | () => { |
| 888 | new HTMLRewriter().on('*', {}).transform(response); |
| 889 | }, |
| 890 | { |
| 891 | message: 'Parser error: Unknown character encoding has been provided.', |
| 892 | } |
| 893 | ); |
| 894 | }, |
| 895 | }; |
| 896 | |
| 897 | export const exceptionPropagation = { |
| 898 | async test() { |
| 899 | const { readable, writable } = new IdentityTransformStream(); |
| 900 | const response = new HTMLRewriter().transform(new Response(readable)); |
| 901 | response.body.cancel(new Error('boom')); |
| 902 | |
| 903 | const writer = writable.getWriter(); |
| 904 | |
| 905 | const enc = new TextEncoder(); |
| 906 | |
| 907 | await writer.write(enc.encode('test')); |
| 908 | |
| 909 | rejects(writer.write(enc.encode('test')), { |
| 910 | message: 'boom', |
| 911 | }); |
| 912 | }, |
| 913 | }; |
| 914 | |
| 915 | // handled HTMLRewriter errors must not poison the IoContext. |
| 916 | export const continueAfterException = { |
| 917 | async test() { |
| 918 | const errorResponse = new HTMLRewriter() |
| 919 | .on('*', { |
| 920 | element() { |
| 921 | throw new Error('intentional error for testing'); |
| 922 | }, |
| 923 | }) |
| 924 | .transform(new Response('<div>test</div>')); |
| 925 | |
| 926 | await rejects(errorResponse.text(), { |
| 927 | message: 'intentional error for testing', |
| 928 | }); |
| 929 | |
| 930 | const successResponse = new HTMLRewriter() |
| 931 | .on('div', { |
| 932 | element(el) { |
| 933 | el.setInnerContent('success'); |
| 934 | }, |
| 935 | }) |
| 936 | .transform(new Response('<div>original</div>')); |
| 937 | |
| 938 | strictEqual(await successResponse.text(), '<div>success</div>'); |
| 939 | }, |
| 940 | }; |
| 941 | |
| 942 | export const continueAfterException2 = { |
| 943 | async test() { |
| 944 | const errorResponse = new HTMLRewriter() |
| 945 | .on('*', { |
| 946 | element() { |
| 947 | throw new Error('intentional error for pipeTo testing'); |
| 948 | }, |
| 949 | }) |
| 950 | .transform(new Response('<div>test</div>')); |
| 951 | |
| 952 | const { writable } = new TransformStream(); |
| 953 | |
| 954 | await rejects(errorResponse.body.pipeTo(writable), { |
| 955 | message: 'intentional error for pipeTo testing', |
| 956 | }); |
| 957 | |
| 958 | const successResponse = new HTMLRewriter() |
| 959 | .on('div', { |
| 960 | element(el) { |
| 961 | el.setInnerContent('success after pipeTo'); |
| 962 | }, |
| 963 | }) |
| 964 | .transform(new Response('<div>original</div>')); |
| 965 | |
| 966 | strictEqual( |
| 967 | await successResponse.text(), |
| 968 | '<div>success after pipeTo</div>' |
| 969 | ); |
| 970 | }, |
| 971 | }; |
| 972 | |
| 973 | export const sameToken = { |
| 974 | async test() { |
| 975 | const obj = {}; |
| 976 | let element; |
| 977 | |
| 978 | const r = new HTMLRewriter() |
| 979 | .on('*', { |
| 980 | element(e) { |
| 981 | element = e; |
| 982 | strictEqual(e.hi, undefined); |
| 983 | |
| 984 | e.hi = 'test'; |
| 985 | strictEqual(e.hi, 'test'); |
| 986 | |
| 987 | e.hi = 'hi'; |
| 988 | e.obj = obj; |
| 989 | |
| 990 | e.replace('foo'); |
| 991 | }, |
| 992 | }) |
| 993 | .on('img', { |
| 994 | element(e) { |
| 995 | notStrictEqual(e, element); |
| 996 | notStrictEqual(e.hi, 'hi'); |
| 997 | notStrictEqual(e.obj, obj); |
| 998 | |
| 999 | // The HTMLRewriter creates a fresh new Element/Doctype/Text |
| 1000 | // object for each handler, thus `e.hi` will yield undefined even if we |
| 1001 | // assigned it in the first handler. |
| 1002 | // See https://jira.cfdata.org/browse/EW-2200. |
| 1003 | e.replace(e.hi); |
| 1004 | }, |
| 1005 | }) |
| 1006 | .transform(new Response('<img />')) |
| 1007 | .text(); |
| 1008 | |
| 1009 | await r; |
| 1010 | }, |
| 1011 | }; |
| 1012 | |
| 1013 | // Regression test for VULN-122672: HTMLRewriter AttributesIterator UAF. |
| 1014 | // When element attributes are modified during iteration (adding new attributes |
| 1015 | // that cause the underlying Vec to reallocate), the iterator's stale pointers |
| 1016 | // would read from freed memory. The fix invalidates all live iterators when |
| 1017 | // setAttribute() or removeAttribute() is called. |
| 1018 | export const attributesIteratorInvalidatedOnSetAttribute = { |
| 1019 | async test() { |
| 1020 | const html = `<div a="1" b="2" c="3">test</div>`; |
| 1021 | |
| 1022 | const rewriter = new HTMLRewriter().on('div', { |
| 1023 | element(el) { |
| 1024 | const iter = el.attributes[Symbol.iterator](); |
| 1025 | |
| 1026 | // Read first attribute - valid. |
| 1027 | const first = iter.next(); |
| 1028 | strictEqual(first.done, false); |
| 1029 | |
| 1030 | // Mutate attributes — this must invalidate the iterator. |
| 1031 | el.setAttribute('newattr', 'value'); |
| 1032 | |
| 1033 | // Subsequent next() must throw, not read freed memory. |
| 1034 | throws(() => iter.next(), { |
| 1035 | message: |
| 1036 | 'The attributes of this element have been modified during iteration. ' + |
| 1037 | 'You must create a new iterator after modifying attributes.', |
| 1038 | }); |
| 1039 | }, |
| 1040 | }); |
| 1041 | |
| 1042 | await rewriter.transform(new Response(html)).text(); |
| 1043 | }, |
| 1044 | }; |
| 1045 | |
| 1046 | export const attributesIteratorInvalidatedOnRemoveAttribute = { |
| 1047 | async test() { |
| 1048 | const html = `<div a="1" b="2" c="3">test</div>`; |
| 1049 | |
| 1050 | const rewriter = new HTMLRewriter().on('div', { |
| 1051 | element(el) { |
| 1052 | const iter = el.attributes[Symbol.iterator](); |
| 1053 | iter.next(); // consume first |
| 1054 | |
| 1055 | // removeAttribute also must invalidate the iterator. |
| 1056 | el.removeAttribute('b'); |
| 1057 | |
| 1058 | throws(() => iter.next(), { |
| 1059 | message: |
| 1060 | 'The attributes of this element have been modified during iteration. ' + |
| 1061 | 'You must create a new iterator after modifying attributes.', |
| 1062 | }); |
| 1063 | }, |
| 1064 | }); |
| 1065 | |
| 1066 | await rewriter.transform(new Response(html)).text(); |
| 1067 | }, |
| 1068 | }; |
| 1069 | |
| 1070 | // After mutation, creating a fresh iterator must work normally. |
| 1071 | export const attributesIteratorFreshAfterMutation = { |
| 1072 | async test() { |
| 1073 | const html = `<div a="1">test</div>`; |
| 1074 | let attrs = []; |
| 1075 | |
| 1076 | const rewriter = new HTMLRewriter().on('div', { |
| 1077 | element(el) { |
| 1078 | el.setAttribute('b', '2'); |
| 1079 | |
| 1080 | // A new iterator created after mutation should work fine. |
| 1081 | for (const [name, value] of el.attributes) { |
| 1082 | attrs.push([name, value]); |
| 1083 | } |
| 1084 | }, |
| 1085 | }); |
| 1086 | |
| 1087 | await rewriter.transform(new Response(html)).text(); |
| 1088 | |
| 1089 | // Should see both original and newly-set attribute. |
| 1090 | strictEqual(attrs.length, 2); |
| 1091 | deepStrictEqual(attrs[0], ['a', '1']); |
| 1092 | deepStrictEqual(attrs[1], ['b', '2']); |
| 1093 | }, |
| 1094 | }; |
| 1095 | |
| 1096 | // Iterating without mutation must still work as before. |
| 1097 | export const attributesIteratorNormalIteration = { |
| 1098 | async test() { |
| 1099 | const html = `<div x="1" y="2" z="3">test</div>`; |
| 1100 | let attrs = []; |
| 1101 | |
| 1102 | const rewriter = new HTMLRewriter().on('div', { |
| 1103 | element(el) { |
| 1104 | for (const [name, value] of el.attributes) { |
| 1105 | attrs.push([name, value]); |
| 1106 | } |
| 1107 | }, |
| 1108 | }); |
| 1109 | |
| 1110 | await rewriter.transform(new Response(html)).text(); |
| 1111 | |
| 1112 | deepStrictEqual(attrs, [ |
| 1113 | ['x', '1'], |
| 1114 | ['y', '2'], |
| 1115 | ['z', '3'], |
| 1116 | ]); |
| 1117 | }, |
| 1118 | }; |
| 1119 | |
| 1120 | export const svgNamespace = { |
| 1121 | async test() { |
| 1122 | const response = new Response(` |
| 1123 | <svg><a /></svg> |
| 1124 | `); |
| 1125 | let namespace; |
| 1126 | |
| 1127 | await new HTMLRewriter() |
| 1128 | .on('a', { |
| 1129 | element(e) { |
| 1130 | namespace = e.namespaceURI; |
| 1131 | }, |
| 1132 | }) |
| 1133 | .transform(response) |
| 1134 | .text(); |
| 1135 | |
| 1136 | strictEqual(namespace, 'http://www.w3.org/2000/svg'); |
| 1137 | }, |
| 1138 | }; |
| 1139 | |
| 1140 | export const handlerPerformingManySmallWrites = { |
| 1141 | async test() { |
| 1142 | const promiseDepth = 128 * 100; |
| 1143 | const numReads = 2; |
| 1144 | const input = new Response('<div></div>'); |
| 1145 | |
| 1146 | const output = new HTMLRewriter() |
| 1147 | .onDocument({ |
| 1148 | end(e) { |
| 1149 | for (let i = 0; i < promiseDepth; i++) { |
| 1150 | e.append('<div></div>', { html: true }); |
| 1151 | } |
| 1152 | }, |
| 1153 | }) |
| 1154 | .transform(input); |
| 1155 | |
| 1156 | // Begin reading but do not fully consume |
| 1157 | const reader = output.body.getReader(); |
| 1158 | for (let i = 0; i < numReads; i++) { |
| 1159 | await reader.read(); |
| 1160 | } |
| 1161 | }, |
| 1162 | }; |
| 1163 | |
| 1164 | export const hugeNumberOfHandlersForAnElement = { |
| 1165 | // Many small handlers |
| 1166 | async test() { |
| 1167 | const promiseDepth = 128 * 100; |
| 1168 | const numReads = promiseDepth; |
| 1169 | |
| 1170 | const input = new Response('<div></div>'.repeat(promiseDepth)); |
| 1171 | |
| 1172 | const output = new HTMLRewriter() |
| 1173 | .on('div', { |
| 1174 | element(e) { |
| 1175 | e.append('<div></div>', { html: true }); |
| 1176 | }, |
| 1177 | }) |
| 1178 | .transform(input); |
| 1179 | |
| 1180 | // Begin reading but do not fully consume |
| 1181 | const reader = output.body.getReader(); |
| 1182 | for (let i = 0; i < numReads; i++) { |
| 1183 | await reader.read(); |
| 1184 | } |
| 1185 | }, |
| 1186 | }; |
| 1187 | |
| 1188 | // Test HTMLRewriter with JS-backed ReadableStream |
| 1189 | // This test was moved from streams-respond-test.js because it triggers |
| 1190 | // a flaky ASAN failure related to V8's cppgc memory validation. |
| 1191 | export const htmlRewriterStream = { |
| 1192 | async test() { |
| 1193 | const enc = new TextEncoder(); |
| 1194 | |
| 1195 | const readable = new ReadableStream({ |
| 1196 | pull(controller) { |
| 1197 | controller.enqueue(enc.encode('Hello World!')); |
| 1198 | controller.close(); |
| 1199 | }, |
| 1200 | }); |
| 1201 | |
| 1202 | let response = new Response(readable, { |
| 1203 | status: 200, |
| 1204 | headers: { |
| 1205 | 'Content-Type': 'text/html; charset=utf-8', |
| 1206 | }, |
| 1207 | }); |
| 1208 | |
| 1209 | response = new HTMLRewriter().transform(response); |
| 1210 | |
| 1211 | strictEqual(await response.text(), 'Hello World!'); |
| 1212 | }, |
| 1213 | }; |