Skip to content
rachel/workerd
File

Blob: src/workerd/api/data-url-test.c++

15.8 KB
1#include "data-url.h"
2 
3#include <workerd/jsg/url.h>
4 
5#include <kj/test.h>
6 
7namespace workerd::api {
8namespace {
9 
10KJ_TEST("DataUrl Basics") {
11  auto dataUrl =
12      KJ_ASSERT_NONNULL(DataUrl::tryParse("data:text/plain;base64,SGVsbG8sIFdvcmxkIQ=="_kj));
13  KJ_ASSERT(dataUrl.getMimeType() == MimeType::PLAINTEXT);
14  KJ_ASSERT(dataUrl.getData().asChars() == "Hello, World!"_kj);
15}
16 
17KJ_TEST("DataUrl Leading/Trailing Whitespace") {
18  auto dataUrl = KJ_ASSERT_NONNULL(
19      DataUrl::tryParse("    data: \t text/plain \t;base64\t\t ,SGVsbG8sIFdvcmxkIQ==    "_kj));
20  KJ_ASSERT(dataUrl.getMimeType() == MimeType::PLAINTEXT);
21  KJ_ASSERT(dataUrl.getData().asChars() == "Hello, World!"_kj);
22}
23 
24KJ_TEST("DataUrl base64 case-insensitive") {
25  auto dataUrl = KJ_ASSERT_NONNULL(
26      DataUrl::tryParse("    data: \t text/plain \t;BasE64\t\t ,SGVsbG8sIFdvcmxkIQ==    "_kj));
27  KJ_ASSERT(dataUrl.getMimeType() == MimeType::PLAINTEXT);
28  KJ_ASSERT(dataUrl.getData().asChars() == "Hello, World!"_kj);
29}
30 
31KJ_TEST("DataUrl no-base64") {
32  auto dataUrl = KJ_ASSERT_NONNULL(
33      DataUrl::tryParse("    data: \t text/plain \t;a=b\t\t ,SGVsbG8sIFdvcmxkIQ==    "_kj));
34  KJ_ASSERT(dataUrl.getMimeType() == MimeType::PLAINTEXT);
35 
36  auto& val = KJ_REQUIRE_NONNULL(dataUrl.getMimeType().params().find("a"_kj));
37  KJ_ASSERT(val == "b"_kj);
38 
39  KJ_ASSERT(dataUrl.getData().asChars() == "SGVsbG8sIFdvcmxkIQ=="_kj);
40}
41 
42KJ_TEST("DataUrl default mime type") {
43  auto dataUrl = KJ_ASSERT_NONNULL(DataUrl::tryParse("data:,Hello, World!"_kj));
44  KJ_ASSERT(dataUrl.getMimeType() == MimeType::PLAINTEXT);
45  KJ_ASSERT(dataUrl.getData().asChars() == "Hello, World!"_kj);
46}
47 
48KJ_TEST("DataUrl default mime type") {
49  auto dataUrl = KJ_ASSERT_NONNULL(DataUrl::tryParse("data:;,Hello, World!"_kj));
50  KJ_ASSERT(dataUrl.getMimeType() == MimeType::PLAINTEXT);
51  KJ_ASSERT(dataUrl.getData().asChars() == "Hello, World!"_kj);
52}
53 
54KJ_TEST("DataUrl default mime type") {
55  auto dataUrl = KJ_ASSERT_NONNULL(DataUrl::tryParse("data:;charset=UTF-8,Hello, World!"_kj));
56  KJ_ASSERT(dataUrl.getMimeType() == MimeType::PLAINTEXT);
57  KJ_ASSERT(dataUrl.getData().asChars() == "Hello, World!"_kj);
58 
59  auto& val = KJ_REQUIRE_NONNULL(dataUrl.getMimeType().params().find("charset"_kj));
60  KJ_ASSERT(val == "UTF-8"_kj);
61}
62 
63struct Test {
64  kj::StringPtr input;
65  kj::StringPtr mimeType;
66  kj::Array<const kj::byte> data;
67};
68 
69KJ_TEST("DataUrl Web Platform Tests") {
70 
71  Test tests[] = {{
72                    "data://test/,X"_kj,
73                    "text/plain;charset=US-ASCII"_kj,
74                    kj::heapArray<kj::byte>({88}),
75                  },
76    {
77      "data://test:test/,X"_kj,
78      nullptr,
79      kj::heapArray<kj::byte>(0),
80    },
81    {
82      "data:,X"_kj,
83      "text/plain;charset=US-ASCII"_kj,
84      kj::heapArray<kj::byte>({88}),
85    },
86    {
87      "data:"_kj,
88      nullptr,
89      kj::heapArray<kj::byte>(0),
90    },
91    {
92      "data:text/html"_kj,
93      nullptr,
94      kj::heapArray<kj::byte>(0),
95    },
96    {
97      "data:text/html    ;charset=x   "_kj,
98      nullptr,
99      kj::heapArray<kj::byte>(0),
100    },
101    {
102      "data:,"_kj,
103      "text/plain;charset=US-ASCII"_kj,
104      kj::heapArray<kj::byte>(0),
105    },
106    {
107      "data:,X#X"_kj,
108      "text/plain;charset=US-ASCII"_kj,
109      kj::heapArray<kj::byte>({88}),
110    },
111    {
112      "data:,%FF"_kj,
113      "text/plain;charset=US-ASCII"_kj,
114      kj::heapArray<kj::byte>({255}),
115    },
116    {
117      "data:text/plain,X"_kj,
118      "text/plain"_kj,
119      kj::heapArray<kj::byte>({88}),
120    },
121    {
122      "data:text/plain ,X"_kj,
123      "text/plain"_kj,
124      kj::heapArray<kj::byte>({88}),
125    },
126    {
127      "data:text/plain%20,X"_kj,
128      "text/plain%20"_kj,
129      kj::heapArray<kj::byte>({88}),
130    },
131    {
132      "data:text/plain\f,X"_kj,
133      "text/plain%0c"_kj,
134      kj::heapArray<kj::byte>({88}),
135    },
136    {
137      "data:text/plain%0C,X"_kj,
138      "text/plain%0c"_kj,
139      kj::heapArray<kj::byte>({88}),
140    },
141    {
142      "data:text/plain;,X"_kj,
143      "text/plain"_kj,
144      kj::heapArray<kj::byte>({88}),
145    },
146    {
147      "data:;x=x;charset=x,X"_kj,
148      "text/plain;x=x;charset=x"_kj,
149      kj::heapArray<kj::byte>({88}),
150    },
151    {
152      "data:;x=x,X"_kj,
153      "text/plain;x=x"_kj,
154      kj::heapArray<kj::byte>({88}),
155    },
156    {
157      "data:text/plain;charset=windows-1252,%C2%B1"_kj,
158      "text/plain;charset=windows-1252"_kj,
159      kj::heapArray<kj::byte>({194, 177}),
160    },
161    {
162      "data:text/plain;Charset=UTF-8,%C2%B1"_kj,
163      "text/plain;charset=UTF-8"_kj,
164      kj::heapArray<kj::byte>({194, 177}),
165    },
166    {
167      "data:text/plain;charset=windows-1252,áñçə💩"_kj,
168      "text/plain;charset=windows-1252"_kj,
169      kj::heapArray<kj::byte>({195, 161, 195, 177, 195, 167, 201, 153, 240, 159, 146, 169}),
170    },
171    {
172      "data:text/plain;charset=UTF-8,áñçə💩"_kj,
173      "text/plain;charset=UTF-8"_kj,
174      kj::heapArray<kj::byte>({195, 161, 195, 177, 195, 167, 201, 153, 240, 159, 146, 169}),
175    },
176    {
177      "data:image/gif,%C2%B1"_kj,
178      "image/gif"_kj,
179      kj::heapArray<kj::byte>({194, 177}),
180    },
181    {
182      "data:IMAGE/gif,%C2%B1"_kj,
183      "image/gif"_kj,
184      kj::heapArray<kj::byte>({194, 177}),
185    },
186    {
187      "data:IMAGE/gif;hi=x,%C2%B1"_kj,
188      "image/gif;hi=x"_kj,
189      kj::heapArray<kj::byte>({194, 177}),
190    },
191    {
192      "data:IMAGE/gif;CHARSET=x,%C2%B1"_kj,
193      "image/gif;charset=x"_kj,
194      kj::heapArray<kj::byte>({194, 177}),
195    },
196    {
197      "data: ,%FF"_kj,
198      "text/plain;charset=US-ASCII"_kj,
199      kj::heapArray<kj::byte>({255}),
200    },
201    {
202      "data:%20,%FF"_kj,
203      "text/plain;charset=US-ASCII"_kj,
204      kj::heapArray<kj::byte>({255}),
205    },
206    {
207      "data:\f,%FF"_kj,
208      "text/plain;charset=US-ASCII"_kj,
209      kj::heapArray<kj::byte>({255}),
210    },
211    {
212      "data:%1F,%FF"_kj,
213      "text/plain;charset=US-ASCII"_kj,
214      kj::heapArray<kj::byte>({255}),
215    },
216    {
217      "data:\u0000,%FF"_kj,
218      "text/plain;charset=US-ASCII"_kj,
219      kj::heapArray<kj::byte>({255}),
220    },
221    {
222      "data:%00,%FF"_kj,
223      "text/plain;charset=US-ASCII"_kj,
224      kj::heapArray<kj::byte>({255}),
225    },
226    {
227      "data:text/html  ,X"_kj,
228      "text/html"_kj,
229      kj::heapArray<kj::byte>({88}),
230    },
231    {
232      "data:text / html,X"_kj,
233      "text/plain;charset=US-ASCII"_kj,
234      kj::heapArray<kj::byte>({88}),
235    },
236    {
237      "data:†,X"_kj,
238      "text/plain;charset=US-ASCII"_kj,
239      kj::heapArray<kj::byte>({88}),
240    },
241    {
242      "data:†/†,X"_kj,
243      "%e2%80%a0/%e2%80%a0"_kj,
244      kj::heapArray<kj::byte>({88}),
245    },
246    {
247      "data:X,X"_kj,
248      "text/plain;charset=US-ASCII"_kj,
249      kj::heapArray<kj::byte>({88}),
250    },
251    {
252      "data:image/png,X X"_kj,
253      "image/png"_kj,
254      kj::heapArray<kj::byte>({88, 32, 88}),
255    },
256    {
257      "data:application/javascript,X X"_kj,
258      "application/javascript"_kj,
259      kj::heapArray<kj::byte>({88, 32, 88}),
260    },
261    {
262      "data:application/xml,X X"_kj,
263      "application/xml"_kj,
264      kj::heapArray<kj::byte>({88, 32, 88}),
265    },
266    {
267      "data:text/javascript,X X"_kj,
268      "text/javascript"_kj,
269      kj::heapArray<kj::byte>({88, 32, 88}),
270    },
271    {
272      "data:text/plain,X X"_kj,
273      "text/plain"_kj,
274      kj::heapArray<kj::byte>({88, 32, 88}),
275    },
276    {
277      "data:unknown/unknown,X X"_kj,
278      "unknown/unknown"_kj,
279      kj::heapArray<kj::byte>({88, 32, 88}),
280    },
281    {
282      "data:text/plain;a=\",\",X"_kj,
283      "text/plain;a=\"\""_kj,
284      kj::heapArray<kj::byte>({34, 44, 88}),
285    },
286    {
287      "data:text/plain;a=%2C,X"_kj,
288      "text/plain;a=%2C"_kj,
289      kj::heapArray<kj::byte>({88}),
290    },
291    {
292      "data:;base64;base64,WA"_kj,
293      "text/plain"_kj,
294      kj::heapArray<kj::byte>({88}),
295    },
296    {
297      "data:x/x;base64;base64,WA"_kj,
298      "x/x"_kj,
299      kj::heapArray<kj::byte>({88}),
300    },
301    {
302      "data:x/x;base64;charset=x,WA"_kj,
303      "x/x;charset=x"_kj,
304      kj::heapArray<kj::byte>({87, 65}),
305    },
306    {
307      "data:x/x;base64;charset=x;base64,WA"_kj,
308      "x/x;charset=x"_kj,
309      kj::heapArray<kj::byte>({88}),
310    },
311    {
312      "data:x/x;base64;base64x,WA"_kj,
313      "x/x"_kj,
314      kj::heapArray<kj::byte>({87, 65}),
315    },
316    {
317      "data:;base64,W%20A"_kj,
318      "text/plain;charset=US-ASCII"_kj,
319      kj::heapArray<kj::byte>({88}),
320    },
321    {
322      "data:;base64,W%0CA"_kj,
323      "text/plain;charset=US-ASCII"_kj,
324      kj::heapArray<kj::byte>({88}),
325    },
326    {
327      "data:x;base64x,WA"_kj,
328      "text/plain;charset=US-ASCII"_kj,
329      kj::heapArray<kj::byte>({87, 65}),
330    },
331    {
332      "data:x;base64;x,WA"_kj,
333      "text/plain;charset=US-ASCII"_kj,
334      kj::heapArray<kj::byte>({87, 65}),
335    },
336    {
337      "data:x;base64=x,WA"_kj,
338      "text/plain;charset=US-ASCII"_kj,
339      kj::heapArray<kj::byte>({87, 65}),
340    },
341    {
342      "data:; base64,WA"_kj,
343      "text/plain;charset=US-ASCII"_kj,
344      kj::heapArray<kj::byte>({88}),
345    },
346    {
347      "data:;  base64,WA"_kj,
348      "text/plain;charset=US-ASCII"_kj,
349      kj::heapArray<kj::byte>({88}),
350    },
351    {
352      "data:  ;charset=x   ;  base64,WA"_kj,
353      "text/plain;charset=x"_kj,
354      kj::heapArray<kj::byte>({88}),
355    },
356    {
357      "data:;base64;,WA"_kj,
358      "text/plain"_kj,
359      kj::heapArray<kj::byte>({87, 65}),
360    },
361    {
362      "data:;base64 ,WA"_kj,
363      "text/plain;charset=US-ASCII"_kj,
364      kj::heapArray<kj::byte>({88}),
365    },
366    {
367      "data:;base64   ,WA"_kj,
368      "text/plain;charset=US-ASCII"_kj,
369      kj::heapArray<kj::byte>({88}),
370    },
371    {
372      "data:;base 64,WA"_kj,
373      "text/plain"_kj,
374      kj::heapArray<kj::byte>({87, 65}),
375    },
376    {
377      "data:;BASe64,WA"_kj,
378      "text/plain;charset=US-ASCII"_kj,
379      kj::heapArray<kj::byte>({88}),
380    },
381    {
382      "data:;%62ase64,WA"_kj,
383      "text/plain"_kj,
384      kj::heapArray<kj::byte>({87, 65}),
385    },
386    {
387      "data:%3Bbase64,WA"_kj,
388      "text/plain;charset=US-ASCII"_kj,
389      kj::heapArray<kj::byte>({87, 65}),
390    },
391    {
392      "data:;charset=x,X"_kj,
393      "text/plain;charset=x"_kj,
394      kj::heapArray<kj::byte>({88}),
395    },
396    {
397      "data:; charset=x,X"_kj,
398      "text/plain;charset=x"_kj,
399      kj::heapArray<kj::byte>({88}),
400    },
401    {
402      "data:;charset =x,X"_kj,
403      "text/plain"_kj,
404      kj::heapArray<kj::byte>({88}),
405    },
406    {
407      "data:;charset= x,X"_kj,
408      "text/plain;charset=\" x\""_kj,
409      kj::heapArray<kj::byte>({88}),
410    },
411    {
412      "data:;charset=,X"_kj,
413      "text/plain"_kj,
414      kj::heapArray<kj::byte>({88}),
415    },
416    {
417      "data:;charset,X"_kj,
418      "text/plain"_kj,
419      kj::heapArray<kj::byte>({88}),
420    },
421    {
422      "data:;charset=\"x\",X"_kj,
423      "text/plain;charset=x"_kj,
424      kj::heapArray<kj::byte>({88}),
425    },
426    {
427      "data:;CHARSET=\"X\",X"_kj,
428      "text/plain;charset=X"_kj,
429      kj::heapArray<kj::byte>({88}),
430    }};
431 
432  auto testPtr = kj::arrayPtr<Test>(tests, 72);
433 
434  for (auto& test: testPtr) {
435    if (test.mimeType == nullptr) {
436      KJ_ASSERT(DataUrl::tryParse(test.input) == kj::none);
437    } else {
438      auto parsed = KJ_ASSERT_NONNULL(DataUrl::tryParse(test.input));
439      KJ_ASSERT(parsed.getMimeType().toString() == test.mimeType);
440      KJ_ASSERT(parsed.getData() == test.data);
441    }
442  }
443}
444 
445struct Base64Test {
446  kj::StringPtr input;
447  kj::Array<kj::byte> expected;
448};
449 
450KJ_TEST("DataUrl base64") {
451  // Our base64 decoder is not very strict and way more forgiving than the
452  // web platform's forgiving base64 decoder. That's just fine for us.
453  // These cases were extracted from the Web Platform Tests for data urls
454  // See: https://github.com/web-platform-tests/wpt/blob/master/fetch/data-urls/resources/
455  Base64Test tests[] = {{""_kj, nullptr}, {"abcd"_kj, kj::heapArray<kj::byte>({105, 183, 29})},
456    {" abcd"_kj, kj::heapArray<kj::byte>({105, 183, 29})},
457    {"abcd "_kj, kj::heapArray<kj::byte>({105, 183, 29})},
458    {" abcd==="_kj, kj::heapArray<kj::byte>({105, 183, 29})},
459    {"abcd=== "_kj, kj::heapArray<kj::byte>({105, 183, 29})},
460    {"abcd ==="_kj, kj::heapArray<kj::byte>({105, 183, 29})}, {"a"_kj, nullptr},
461    {"ab"_kj, kj::heapArray<kj::byte>({105})}, {"abc"_kj, kj::heapArray<kj::byte>({105, 183})},
462    {"abcde"_kj, kj::heapArray<kj::byte>({105, 183, 29})}, {"𐀀"_kj, nullptr}, {"="_kj, nullptr},
463    {"=="_kj, nullptr}, {"==="_kj, nullptr}, {"===="_kj, nullptr}, {"====="_kj, nullptr},
464    {"a="_kj, nullptr}, {"a=="_kj, nullptr}, {"a==="_kj, nullptr}, {"a===="_kj, nullptr},
465    {"a====="_kj, nullptr}, {"ab="_kj, kj::heapArray<kj::byte>({105})},
466    {"ab=="_kj, kj::heapArray<kj::byte>({105})}, {"ab==="_kj, kj::heapArray<kj::byte>({105})},
467    {"ab===="_kj, kj::heapArray<kj::byte>({105})}, {"ab====="_kj, kj::heapArray<kj::byte>({105})},
468    {"abc="_kj, kj::heapArray<kj::byte>({105, 183})},
469    {"abc=="_kj, kj::heapArray<kj::byte>({105, 183})},
470    {"abc==="_kj, kj::heapArray<kj::byte>({105, 183})},
471    {"abc===="_kj, kj::heapArray<kj::byte>({105, 183})},
472    {"abc====="_kj, kj::heapArray<kj::byte>({105, 183})},
473    {"abcd="_kj, kj::heapArray<kj::byte>({105, 183, 29})},
474    {"abcd=="_kj, kj::heapArray<kj::byte>({105, 183, 29})},
475    {"abcd==="_kj, kj::heapArray<kj::byte>({105, 183, 29})},
476    {"abcd===="_kj, kj::heapArray<kj::byte>({105, 183, 29})},
477    {"abcd====="_kj, kj::heapArray<kj::byte>({105, 183, 29})},
478    {"abcde="_kj, kj::heapArray<kj::byte>({105, 183, 29})},
479    {"abcde=="_kj, kj::heapArray<kj::byte>({105, 183, 29})},
480    {"abcde==="_kj, kj::heapArray<kj::byte>({105, 183, 29})},
481    {"abcde===="_kj, kj::heapArray<kj::byte>({105, 183, 29})},
482    {"abcde====="_kj, kj::heapArray<kj::byte>({105, 183, 29})}, {"=a"_kj, nullptr},
483    {"=a="_kj, nullptr}, {"a=b"_kj, kj::heapArray<kj::byte>({105})},
484    {"a=b="_kj, kj::heapArray<kj::byte>({105})}, {"ab=c"_kj, kj::heapArray<kj::byte>({105, 183})},
485    {"ab=c="_kj, kj::heapArray<kj::byte>({105, 183})},
486    {"abc=d"_kj, kj::heapArray<kj::byte>({105, 183, 29})},
487    {"abc=d="_kj, kj::heapArray<kj::byte>({105, 183, 29})},
488    {"ab\u000Bcd"_kj, kj::heapArray<kj::byte>({105, 183, 29})},
489    {"ab\u3000cd"_kj, kj::heapArray<kj::byte>({105, 183, 29})},
490    {"ab\u3001cd"_kj, kj::heapArray<kj::byte>({105, 183, 29})},
491    {"ab\tcd"_kj, kj::heapArray<kj::byte>({105, 183, 29})},
492    {"ab\ncd"_kj, kj::heapArray<kj::byte>({105, 183, 29})},
493    {"ab\fcd"_kj, kj::heapArray<kj::byte>({105, 183, 29})},
494    {"ab\rcd"_kj, kj::heapArray<kj::byte>({105, 183, 29})},
495    {"ab cd"_kj, kj::heapArray<kj::byte>({105, 183, 29})},
496    {"ab\u00a0cd"_kj, kj::heapArray<kj::byte>({105, 183, 29})},
497    {"ab\t\n\f\r cd"_kj, kj::heapArray<kj::byte>({105, 183, 29})},
498    {" \t\n\f\r ab\t\n\f\r cd\t\n\f\r "_kj, kj::heapArray<kj::byte>({105, 183, 29})},
499    {"ab\t\n\f\r =\t\n\f\r =\t\n\f\r "_kj, kj::heapArray<kj::byte>({105})}, {"A"_kj, nullptr},
500    {"/A"_kj, kj::heapArray<kj::byte>({252})}, {"//A"_kj, kj::heapArray<kj::byte>({255, 240})},
501    {"///A"_kj, kj::heapArray<kj::byte>({255, 255, 192})},
502    {"////A"_kj, kj::heapArray<kj::byte>({255, 255, 255})}, {"/"_kj, nullptr},
503    {"A/"_kj, kj::heapArray<kj::byte>({3})}, {"AA/"_kj, kj::heapArray<kj::byte>({0, 15})},
504    {"AAAA/"_kj, kj::heapArray<kj::byte>({0, 0, 0})},
505    {"AAA/"_kj, kj::heapArray<kj::byte>({0, 0, 63})},
506    {"\u0000nonsense"_kj, kj::heapArray<kj::byte>({158, 137, 236, 122, 123, 30})},
507    {"abcd\u0000nonsense"_kj, kj::heapArray<kj::byte>({105, 183, 29, 158, 137, 236, 122, 123, 30})},
508    {"YQ"_kj, kj::heapArray<kj::byte>({97})}, {"YR"_kj, kj::heapArray<kj::byte>({97})},
509    {"~~"_kj, nullptr}, {".."_kj, nullptr}, {"--"_kj, nullptr}, {"__"_kj, nullptr}};
510 
511  auto testPtr = kj::arrayPtr<Base64Test>(tests, 80);
512 
513  for (auto& test: testPtr) {
514    auto input = kj::str("data:;base64,", test.input);
515 
516    auto url = KJ_ASSERT_NONNULL(DataUrl::tryParse(input));
517 
518    KJ_ASSERT(url.getData() == test.expected, test.input);
519  }
520}
521 
522KJ_TEST("Large Data URL") {
523  auto str = kj::str(kj::repeat('a', 6000));
524  auto url = kj::str("data:,", str);
525  auto parsed = KJ_ASSERT_NONNULL(DataUrl::tryParse(url));
526  KJ_ASSERT(parsed.getData().asChars() == str);
527}
528 
529}  // namespace
530}  // namespace workerd::api