sanitizer_test.go 25 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613
  1. // SPDX-FileCopyrightText: Copyright The Miniflux Authors. All rights reserved.
  2. // SPDX-License-Identifier: Apache-2.0
  3. package sanitizer // import "miniflux.app/v2/internal/reader/sanitizer"
  4. import (
  5. "os"
  6. "testing"
  7. "miniflux.app/v2/internal/config"
  8. )
  9. func TestMain(m *testing.M) {
  10. config.Opts = config.NewOptions()
  11. exitCode := m.Run()
  12. os.Exit(exitCode)
  13. }
  14. func TestValidInput(t *testing.T) {
  15. input := `<p>This is a <strong>text</strong> with an image: <img src="http://example.org/" alt="Test" loading="lazy">.</p>`
  16. output := Sanitize("http://example.org/", input)
  17. if input != output {
  18. t.Errorf(`Wrong output: "%s" != "%s"`, input, output)
  19. }
  20. }
  21. func TestImgWithWidthAndHeightAttribute(t *testing.T) {
  22. input := `<img src="https://example.org/image.png" width="10" height="20">`
  23. expected := `<img src="https://example.org/image.png" width="10" height="20" loading="lazy">`
  24. output := Sanitize("http://example.org/", input)
  25. if output != expected {
  26. t.Errorf(`Wrong output: %s`, output)
  27. }
  28. }
  29. func TestImgWithWidthAndHeightAttributeLargerThanMinifluxLayout(t *testing.T) {
  30. input := `<img src="https://example.org/image.png" width="1200" height="675">`
  31. expected := `<img src="https://example.org/image.png" loading="lazy">`
  32. output := Sanitize("http://example.org/", input)
  33. if output != expected {
  34. t.Errorf(`Wrong output: %s`, output)
  35. }
  36. }
  37. func TestImgWithIncorrectWidthAndHeightAttribute(t *testing.T) {
  38. input := `<img src="https://example.org/image.png" width="10px" height="20px">`
  39. expected := `<img src="https://example.org/image.png" loading="lazy">`
  40. output := Sanitize("http://example.org/", input)
  41. if output != expected {
  42. t.Errorf(`Wrong output: %s`, output)
  43. }
  44. }
  45. func TestImgWithTextDataURL(t *testing.T) {
  46. input := `<img src="data:text/plain;base64,SGVsbG8sIFdvcmxkIQ==" alt="Example">`
  47. expected := ``
  48. output := Sanitize("http://example.org/", input)
  49. if output != expected {
  50. t.Errorf(`Wrong output: %s`, output)
  51. }
  52. }
  53. func TestImgWithDataURL(t *testing.T) {
  54. input := `<img src="data:image/gif;base64,test" alt="Example">`
  55. expected := `<img src="data:image/gif;base64,test" alt="Example" loading="lazy">`
  56. output := Sanitize("http://example.org/", input)
  57. if output != expected {
  58. t.Errorf(`Wrong output: %s`, output)
  59. }
  60. }
  61. func TestImgWithSrcset(t *testing.T) {
  62. input := `<img srcset="example-320w.jpg, example-480w.jpg 1.5x, example-640w.jpg 2x, example-640w.jpg 640w" src="example-640w.jpg" alt="Example">`
  63. expected := `<img srcset="http://example.org/example-320w.jpg, http://example.org/example-480w.jpg 1.5x, http://example.org/example-640w.jpg 2x, http://example.org/example-640w.jpg 640w" src="http://example.org/example-640w.jpg" alt="Example" loading="lazy">`
  64. output := Sanitize("http://example.org/", input)
  65. if output != expected {
  66. t.Errorf(`Wrong output: %s`, output)
  67. }
  68. }
  69. func TestSourceWithSrcsetAndMedia(t *testing.T) {
  70. input := `<picture><source media="(min-width: 800px)" srcset="elva-800w.jpg"></picture>`
  71. expected := `<picture><source media="(min-width: 800px)" srcset="http://example.org/elva-800w.jpg"></picture>`
  72. output := Sanitize("http://example.org/", input)
  73. if output != expected {
  74. t.Errorf(`Wrong output: %s`, output)
  75. }
  76. }
  77. func TestMediumImgWithSrcset(t *testing.T) {
  78. input := `<img alt="Image for post" class="t u v ef aj" src="https://miro.medium.com/max/5460/1*aJ9JibWDqO81qMfNtqgqrw.jpeg" srcset="https://miro.medium.com/max/552/1*aJ9JibWDqO81qMfNtqgqrw.jpeg 276w, https://miro.medium.com/max/1000/1*aJ9JibWDqO81qMfNtqgqrw.jpeg 500w" sizes="500px" width="2730" height="3407">`
  79. expected := `<img alt="Image for post" src="https://miro.medium.com/max/5460/1*aJ9JibWDqO81qMfNtqgqrw.jpeg" srcset="https://miro.medium.com/max/552/1*aJ9JibWDqO81qMfNtqgqrw.jpeg 276w, https://miro.medium.com/max/1000/1*aJ9JibWDqO81qMfNtqgqrw.jpeg 500w" sizes="500px" loading="lazy">`
  80. output := Sanitize("http://example.org/", input)
  81. if output != expected {
  82. t.Errorf(`Wrong output: %s`, output)
  83. }
  84. }
  85. func TestSelfClosingTags(t *testing.T) {
  86. input := `<p>This <br> is a <strong>text</strong> <br/>with an image: <img src="http://example.org/" alt="Test" loading="lazy"/>.</p>`
  87. output := Sanitize("http://example.org/", input)
  88. if input != output {
  89. t.Errorf(`Wrong output: "%s" != "%s"`, input, output)
  90. }
  91. }
  92. func TestTable(t *testing.T) {
  93. input := `<table><tr><th>A</th><th colspan="2">B</th></tr><tr><td>C</td><td>D</td><td>E</td></tr></table>`
  94. output := Sanitize("http://example.org/", input)
  95. if input != output {
  96. t.Errorf(`Wrong output: "%s" != "%s"`, input, output)
  97. }
  98. }
  99. func TestRelativeURL(t *testing.T) {
  100. input := `This <a href="/test.html">link is relative</a> and this image: <img src="../folder/image.png"/>`
  101. expected := `This <a href="http://example.org/test.html" rel="noopener noreferrer" target="_blank" referrerpolicy="no-referrer">link is relative</a> and this image: <img src="http://example.org/folder/image.png" loading="lazy"/>`
  102. output := Sanitize("http://example.org/", input)
  103. if expected != output {
  104. t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
  105. }
  106. }
  107. func TestProtocolRelativeURL(t *testing.T) {
  108. input := `This <a href="//static.example.org/index.html">link is relative</a>.`
  109. expected := `This <a href="https://static.example.org/index.html" rel="noopener noreferrer" target="_blank" referrerpolicy="no-referrer">link is relative</a>.`
  110. output := Sanitize("http://example.org/", input)
  111. if expected != output {
  112. t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
  113. }
  114. }
  115. func TestInvalidTag(t *testing.T) {
  116. input := `<p>My invalid <b>tag</b>.</p>`
  117. expected := `<p>My invalid tag.</p>`
  118. output := Sanitize("http://example.org/", input)
  119. if expected != output {
  120. t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
  121. }
  122. }
  123. func TestVideoTag(t *testing.T) {
  124. input := `<p>My valid <video src="videofile.webm" autoplay poster="posterimage.jpg">fallback</video>.</p>`
  125. expected := `<p>My valid <video src="http://example.org/videofile.webm" poster="http://example.org/posterimage.jpg" controls>fallback</video>.</p>`
  126. output := Sanitize("http://example.org/", input)
  127. if expected != output {
  128. t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
  129. }
  130. }
  131. func TestAudioAndSourceTag(t *testing.T) {
  132. input := `<p>My music <audio controls="controls"><source src="foo.wav" type="audio/wav"></audio>.</p>`
  133. expected := `<p>My music <audio controls><source src="http://example.org/foo.wav" type="audio/wav"></audio>.</p>`
  134. output := Sanitize("http://example.org/", input)
  135. if expected != output {
  136. t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
  137. }
  138. }
  139. func TestUnknownTag(t *testing.T) {
  140. input := `<p>My invalid <unknown>tag</unknown>.</p>`
  141. expected := `<p>My invalid tag.</p>`
  142. output := Sanitize("http://example.org/", input)
  143. if expected != output {
  144. t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
  145. }
  146. }
  147. func TestInvalidNestedTag(t *testing.T) {
  148. input := `<p>My invalid <b>tag with some <em>valid</em> tag</b>.</p>`
  149. expected := `<p>My invalid tag with some <em>valid</em> tag.</p>`
  150. output := Sanitize("http://example.org/", input)
  151. if expected != output {
  152. t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
  153. }
  154. }
  155. func TestInvalidIFrame(t *testing.T) {
  156. input := `<iframe src="http://example.org/"></iframe>`
  157. expected := ``
  158. output := Sanitize("http://example.com/", input)
  159. if expected != output {
  160. t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
  161. }
  162. }
  163. func TestIFrameWithChildElements(t *testing.T) {
  164. input := `<iframe src="https://www.youtube.com/"><p>test</p></iframe>`
  165. expected := `<iframe src="https://www.youtube.com/" sandbox="allow-scripts allow-same-origin allow-popups allow-popups-to-escape-sandbox" loading="lazy"></iframe>`
  166. output := Sanitize("http://example.com/", input)
  167. if expected != output {
  168. t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
  169. }
  170. }
  171. func TestAnchorLink(t *testing.T) {
  172. input := `<p>This link is <a href="#some-anchor">an anchor</a></p>`
  173. expected := `<p>This link is <a href="#some-anchor">an anchor</a></p>`
  174. output := Sanitize("http://example.org/", input)
  175. if expected != output {
  176. t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
  177. }
  178. }
  179. func TestInvalidURLScheme(t *testing.T) {
  180. input := `<p>This link is <a src="file:///etc/passwd">not valid</a></p>`
  181. expected := `<p>This link is not valid</p>`
  182. output := Sanitize("http://example.org/", input)
  183. if expected != output {
  184. t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
  185. }
  186. }
  187. func TestAPTURIScheme(t *testing.T) {
  188. input := `<p>This link is <a href="apt:some-package?channel=test">valid</a></p>`
  189. expected := `<p>This link is <a href="apt:some-package?channel=test" rel="noopener noreferrer" target="_blank" referrerpolicy="no-referrer">valid</a></p>`
  190. output := Sanitize("http://example.org/", input)
  191. if expected != output {
  192. t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
  193. }
  194. }
  195. func TestBitcoinURIScheme(t *testing.T) {
  196. input := `<p>This link is <a href="bitcoin:175tWpb8K1S7NmH4Zx6rewF9WQrcZv245W">valid</a></p>`
  197. expected := `<p>This link is <a href="bitcoin:175tWpb8K1S7NmH4Zx6rewF9WQrcZv245W" rel="noopener noreferrer" target="_blank" referrerpolicy="no-referrer">valid</a></p>`
  198. output := Sanitize("http://example.org/", input)
  199. if expected != output {
  200. t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
  201. }
  202. }
  203. func TestCallToURIScheme(t *testing.T) {
  204. input := `<p>This link is <a href="callto:12345679">valid</a></p>`
  205. expected := `<p>This link is <a href="callto:12345679" rel="noopener noreferrer" target="_blank" referrerpolicy="no-referrer">valid</a></p>`
  206. output := Sanitize("http://example.org/", input)
  207. if expected != output {
  208. t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
  209. }
  210. }
  211. func TestFeedURIScheme(t *testing.T) {
  212. input := `<p>This link is <a href="feed://example.com/rss.xml">valid</a></p>`
  213. expected := `<p>This link is <a href="feed://example.com/rss.xml" rel="noopener noreferrer" target="_blank" referrerpolicy="no-referrer">valid</a></p>`
  214. output := Sanitize("http://example.org/", input)
  215. if expected != output {
  216. t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
  217. }
  218. input = `<p>This link is <a href="feed:https://example.com/rss.xml">valid</a></p>`
  219. expected = `<p>This link is <a href="feed:https://example.com/rss.xml" rel="noopener noreferrer" target="_blank" referrerpolicy="no-referrer">valid</a></p>`
  220. output = Sanitize("http://example.org/", input)
  221. if expected != output {
  222. t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
  223. }
  224. }
  225. func TestGeoURIScheme(t *testing.T) {
  226. input := `<p>This link is <a href="geo:13.4125,103.8667">valid</a></p>`
  227. expected := `<p>This link is <a href="geo:13.4125,103.8667" rel="noopener noreferrer" target="_blank" referrerpolicy="no-referrer">valid</a></p>`
  228. output := Sanitize("http://example.org/", input)
  229. if expected != output {
  230. t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
  231. }
  232. }
  233. func TestItunesURIScheme(t *testing.T) {
  234. input := `<p>This link is <a href="itms://itunes.com/apps/my-app-name">valid</a></p>`
  235. expected := `<p>This link is <a href="itms://itunes.com/apps/my-app-name" rel="noopener noreferrer" target="_blank" referrerpolicy="no-referrer">valid</a></p>`
  236. output := Sanitize("http://example.org/", input)
  237. if expected != output {
  238. t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
  239. }
  240. input = `<p>This link is <a href="itms-apps://itunes.com/apps/my-app-name">valid</a></p>`
  241. expected = `<p>This link is <a href="itms-apps://itunes.com/apps/my-app-name" rel="noopener noreferrer" target="_blank" referrerpolicy="no-referrer">valid</a></p>`
  242. output = Sanitize("http://example.org/", input)
  243. if expected != output {
  244. t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
  245. }
  246. }
  247. func TestMagnetURIScheme(t *testing.T) {
  248. input := `<p>This link is <a href="magnet:?xt.1=urn:sha1:YNCKHTQCWBTRNJIV4WNAE52SJUQCZO5C&amp;xt.2=urn:sha1:TXGCZQTH26NL6OUQAJJPFALHG2LTGBC7">valid</a></p>`
  249. expected := `<p>This link is <a href="magnet:?xt.1=urn:sha1:YNCKHTQCWBTRNJIV4WNAE52SJUQCZO5C&amp;xt.2=urn:sha1:TXGCZQTH26NL6OUQAJJPFALHG2LTGBC7" rel="noopener noreferrer" target="_blank" referrerpolicy="no-referrer">valid</a></p>`
  250. output := Sanitize("http://example.org/", input)
  251. if expected != output {
  252. t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
  253. }
  254. }
  255. func TestMailtoURIScheme(t *testing.T) {
  256. input := `<p>This link is <a href="mailto:jsmith@example.com?subject=A%20Test&amp;body=My%20idea%20is%3A%20%0A">valid</a></p>`
  257. expected := `<p>This link is <a href="mailto:jsmith@example.com?subject=A%20Test&amp;body=My%20idea%20is%3A%20%0A" rel="noopener noreferrer" target="_blank" referrerpolicy="no-referrer">valid</a></p>`
  258. output := Sanitize("http://example.org/", input)
  259. if expected != output {
  260. t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
  261. }
  262. }
  263. func TestNewsURIScheme(t *testing.T) {
  264. input := `<p>This link is <a href="news://news.server.example/*">valid</a></p>`
  265. expected := `<p>This link is <a href="news://news.server.example/*" rel="noopener noreferrer" target="_blank" referrerpolicy="no-referrer">valid</a></p>`
  266. output := Sanitize("http://example.org/", input)
  267. if expected != output {
  268. t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
  269. }
  270. input = `<p>This link is <a href="news:example.group.this">valid</a></p>`
  271. expected = `<p>This link is <a href="news:example.group.this" rel="noopener noreferrer" target="_blank" referrerpolicy="no-referrer">valid</a></p>`
  272. output = Sanitize("http://example.org/", input)
  273. if expected != output {
  274. t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
  275. }
  276. input = `<p>This link is <a href="nntp://news.server.example/example.group.this">valid</a></p>`
  277. expected = `<p>This link is <a href="nntp://news.server.example/example.group.this" rel="noopener noreferrer" target="_blank" referrerpolicy="no-referrer">valid</a></p>`
  278. output = Sanitize("http://example.org/", input)
  279. if expected != output {
  280. t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
  281. }
  282. }
  283. func TestRTMPURIScheme(t *testing.T) {
  284. input := `<p>This link is <a href="rtmp://mycompany.com/vod/mp4:mycoolvideo.mov">valid</a></p>`
  285. expected := `<p>This link is <a href="rtmp://mycompany.com/vod/mp4:mycoolvideo.mov" rel="noopener noreferrer" target="_blank" referrerpolicy="no-referrer">valid</a></p>`
  286. output := Sanitize("http://example.org/", input)
  287. if expected != output {
  288. t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
  289. }
  290. }
  291. func TestSIPURIScheme(t *testing.T) {
  292. input := `<p>This link is <a href="sip:+1-212-555-1212:1234@gateway.com;user=phone">valid</a></p>`
  293. expected := `<p>This link is <a href="sip:+1-212-555-1212:1234@gateway.com;user=phone" rel="noopener noreferrer" target="_blank" referrerpolicy="no-referrer">valid</a></p>`
  294. output := Sanitize("http://example.org/", input)
  295. if expected != output {
  296. t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
  297. }
  298. input = `<p>This link is <a href="sips:alice@atlanta.com?subject=project%20x&amp;priority=urgent">valid</a></p>`
  299. expected = `<p>This link is <a href="sips:alice@atlanta.com?subject=project%20x&amp;priority=urgent" rel="noopener noreferrer" target="_blank" referrerpolicy="no-referrer">valid</a></p>`
  300. output = Sanitize("http://example.org/", input)
  301. if expected != output {
  302. t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
  303. }
  304. }
  305. func TestSkypeURIScheme(t *testing.T) {
  306. input := `<p>This link is <a href="skype:echo123?call">valid</a></p>`
  307. expected := `<p>This link is <a href="skype:echo123?call" rel="noopener noreferrer" target="_blank" referrerpolicy="no-referrer">valid</a></p>`
  308. output := Sanitize("http://example.org/", input)
  309. if expected != output {
  310. t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
  311. }
  312. }
  313. func TestSpotifyURIScheme(t *testing.T) {
  314. input := `<p>This link is <a href="spotify:track:2jCnn1QPQ3E8ExtLe6INsx">valid</a></p>`
  315. expected := `<p>This link is <a href="spotify:track:2jCnn1QPQ3E8ExtLe6INsx" rel="noopener noreferrer" target="_blank" referrerpolicy="no-referrer">valid</a></p>`
  316. output := Sanitize("http://example.org/", input)
  317. if expected != output {
  318. t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
  319. }
  320. }
  321. func TestSteamURIScheme(t *testing.T) {
  322. input := `<p>This link is <a href="steam://settings/account">valid</a></p>`
  323. expected := `<p>This link is <a href="steam://settings/account" rel="noopener noreferrer" target="_blank" referrerpolicy="no-referrer">valid</a></p>`
  324. output := Sanitize("http://example.org/", input)
  325. if expected != output {
  326. t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
  327. }
  328. }
  329. func TestSubversionURIScheme(t *testing.T) {
  330. input := `<p>This link is <a href="svn://example.org">valid</a></p>`
  331. expected := `<p>This link is <a href="svn://example.org" rel="noopener noreferrer" target="_blank" referrerpolicy="no-referrer">valid</a></p>`
  332. output := Sanitize("http://example.org/", input)
  333. if expected != output {
  334. t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
  335. }
  336. input = `<p>This link is <a href="svn+ssh://example.org">valid</a></p>`
  337. expected = `<p>This link is <a href="svn+ssh://example.org" rel="noopener noreferrer" target="_blank" referrerpolicy="no-referrer">valid</a></p>`
  338. output = Sanitize("http://example.org/", input)
  339. if expected != output {
  340. t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
  341. }
  342. }
  343. func TestTelURIScheme(t *testing.T) {
  344. input := `<p>This link is <a href="tel:+1-201-555-0123">valid</a></p>`
  345. expected := `<p>This link is <a href="tel:+1-201-555-0123" rel="noopener noreferrer" target="_blank" referrerpolicy="no-referrer">valid</a></p>`
  346. output := Sanitize("http://example.org/", input)
  347. if expected != output {
  348. t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
  349. }
  350. }
  351. func TestWebcalURIScheme(t *testing.T) {
  352. input := `<p>This link is <a href="webcal://example.com/calendar.ics">valid</a></p>`
  353. expected := `<p>This link is <a href="webcal://example.com/calendar.ics" rel="noopener noreferrer" target="_blank" referrerpolicy="no-referrer">valid</a></p>`
  354. output := Sanitize("http://example.org/", input)
  355. if expected != output {
  356. t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
  357. }
  358. }
  359. func TestXMPPURIScheme(t *testing.T) {
  360. input := `<p>This link is <a href="xmpp:user@host?subscribe&amp;type=subscribed">valid</a></p>`
  361. expected := `<p>This link is <a href="xmpp:user@host?subscribe&amp;type=subscribed" rel="noopener noreferrer" target="_blank" referrerpolicy="no-referrer">valid</a></p>`
  362. output := Sanitize("http://example.org/", input)
  363. if expected != output {
  364. t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
  365. }
  366. }
  367. func TestBlacklistedLink(t *testing.T) {
  368. input := `<p>This image is not valid <img src="https://stats.wordpress.com/some-tracker"></p>`
  369. expected := `<p>This image is not valid </p>`
  370. output := Sanitize("http://example.org/", input)
  371. if expected != output {
  372. t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
  373. }
  374. }
  375. func TestPixelTracker(t *testing.T) {
  376. input := `<p><img src="https://tracker1.example.org/" height="1" width="1"> and <img src="https://tracker2.example.org/" height="1" width="1"/></p>`
  377. expected := `<p> and </p>`
  378. output := Sanitize("http://example.org/", input)
  379. if expected != output {
  380. t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
  381. }
  382. }
  383. func TestXmlEntities(t *testing.T) {
  384. input := `<pre>echo "test" &gt; /etc/hosts</pre>`
  385. expected := `<pre>echo &#34;test&#34; &gt; /etc/hosts</pre>`
  386. output := Sanitize("http://example.org/", input)
  387. if expected != output {
  388. t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
  389. }
  390. }
  391. func TestEspaceAttributes(t *testing.T) {
  392. input := `<td rowspan="<b>test</b>">test</td>`
  393. expected := `<td rowspan="&lt;b&gt;test&lt;/b&gt;">test</td>`
  394. output := Sanitize("http://example.org/", input)
  395. if expected != output {
  396. t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
  397. }
  398. }
  399. func TestReplaceYoutubeURL(t *testing.T) {
  400. input := `<iframe src="http://www.youtube.com/embed/test123?version=3&#038;rel=1&#038;fs=1&#038;autohide=2&#038;showsearch=0&#038;showinfo=1&#038;iv_load_policy=1&#038;wmode=transparent"></iframe>`
  401. expected := `<iframe src="https://www.youtube-nocookie.com/embed/test123?version=3&amp;rel=1&amp;fs=1&amp;autohide=2&amp;showsearch=0&amp;showinfo=1&amp;iv_load_policy=1&amp;wmode=transparent" sandbox="allow-scripts allow-same-origin allow-popups allow-popups-to-escape-sandbox" loading="lazy"></iframe>`
  402. output := Sanitize("http://example.org/", input)
  403. if expected != output {
  404. t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
  405. }
  406. }
  407. func TestReplaceSecureYoutubeURL(t *testing.T) {
  408. input := `<iframe src="https://www.youtube.com/embed/test123"></iframe>`
  409. expected := `<iframe src="https://www.youtube-nocookie.com/embed/test123" sandbox="allow-scripts allow-same-origin allow-popups allow-popups-to-escape-sandbox" loading="lazy"></iframe>`
  410. output := Sanitize("http://example.org/", input)
  411. if expected != output {
  412. t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
  413. }
  414. }
  415. func TestReplaceSecureYoutubeURLWithParameters(t *testing.T) {
  416. input := `<iframe src="https://www.youtube.com/embed/test123?rel=0&amp;controls=0"></iframe>`
  417. expected := `<iframe src="https://www.youtube-nocookie.com/embed/test123?rel=0&amp;controls=0" sandbox="allow-scripts allow-same-origin allow-popups allow-popups-to-escape-sandbox" loading="lazy"></iframe>`
  418. output := Sanitize("http://example.org/", input)
  419. if expected != output {
  420. t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
  421. }
  422. }
  423. func TestReplaceYoutubeURLAlreadyReplaced(t *testing.T) {
  424. input := `<iframe src="https://www.youtube-nocookie.com/embed/test123?rel=0&amp;controls=0" sandbox="allow-scripts allow-same-origin"></iframe>`
  425. expected := `<iframe src="https://www.youtube-nocookie.com/embed/test123?rel=0&amp;controls=0" sandbox="allow-scripts allow-same-origin allow-popups allow-popups-to-escape-sandbox" loading="lazy"></iframe>`
  426. output := Sanitize("http://example.org/", input)
  427. if expected != output {
  428. t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
  429. }
  430. }
  431. func TestReplaceProtocolRelativeYoutubeURL(t *testing.T) {
  432. input := `<iframe src="//www.youtube.com/embed/Bf2W84jrGqs" width="560" height="314" allowfullscreen="allowfullscreen"></iframe>`
  433. expected := `<iframe src="https://www.youtube-nocookie.com/embed/Bf2W84jrGqs" width="560" height="314" allowfullscreen="allowfullscreen" sandbox="allow-scripts allow-same-origin allow-popups allow-popups-to-escape-sandbox" loading="lazy"></iframe>`
  434. output := Sanitize("http://example.org/", input)
  435. if expected != output {
  436. t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
  437. }
  438. }
  439. func TestReplaceYoutubeURLWithCustomURL(t *testing.T) {
  440. os.Clearenv()
  441. os.Setenv("YOUTUBE_EMBED_URL_OVERRIDE", "https://invidious.custom/embed/")
  442. var err error
  443. parser := config.NewParser()
  444. config.Opts, err = parser.ParseEnvironmentVariables()
  445. if err != nil {
  446. t.Fatalf(`Parsing failure: %v`, err)
  447. }
  448. input := `<iframe src="https://www.youtube.com/embed/test123?version=3&#038;rel=1&#038;fs=1&#038;autohide=2&#038;showsearch=0&#038;showinfo=1&#038;iv_load_policy=1&#038;wmode=transparent"></iframe>`
  449. expected := `<iframe src="https://invidious.custom/embed/test123?version=3&amp;rel=1&amp;fs=1&amp;autohide=2&amp;showsearch=0&amp;showinfo=1&amp;iv_load_policy=1&amp;wmode=transparent" sandbox="allow-scripts allow-same-origin allow-popups allow-popups-to-escape-sandbox" loading="lazy"></iframe>`
  450. output := Sanitize("http://example.org/", input)
  451. if expected != output {
  452. t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
  453. }
  454. }
  455. func TestReplaceIframeURL(t *testing.T) {
  456. input := `<iframe src="https://player.vimeo.com/video/123456?title=0&amp;byline=0"></iframe>`
  457. expected := `<iframe src="https://player.vimeo.com/video/123456?title=0&amp;byline=0" sandbox="allow-scripts allow-same-origin allow-popups allow-popups-to-escape-sandbox" loading="lazy"></iframe>`
  458. output := Sanitize("http://example.org/", input)
  459. if expected != output {
  460. t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
  461. }
  462. }
  463. func TestReplaceNoScript(t *testing.T) {
  464. input := `<p>Before paragraph.</p><noscript>Inside <code>noscript</code> tag with an image: <img src="http://example.org/" alt="Test" loading="lazy"></noscript><p>After paragraph.</p>`
  465. expected := `<p>Before paragraph.</p><p>After paragraph.</p>`
  466. output := Sanitize("http://example.org/", input)
  467. if expected != output {
  468. t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
  469. }
  470. }
  471. func TestReplaceScript(t *testing.T) {
  472. input := `<p>Before paragraph.</p><script type="text/javascript">alert("1");</script><p>After paragraph.</p>`
  473. expected := `<p>Before paragraph.</p><p>After paragraph.</p>`
  474. output := Sanitize("http://example.org/", input)
  475. if expected != output {
  476. t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
  477. }
  478. }
  479. func TestReplaceStyle(t *testing.T) {
  480. input := `<p>Before paragraph.</p><style>body { background-color: #ff0000; }</style><p>After paragraph.</p>`
  481. expected := `<p>Before paragraph.</p><p>After paragraph.</p>`
  482. output := Sanitize("http://example.org/", input)
  483. if expected != output {
  484. t.Errorf(`Wrong output: "%s" != "%s"`, expected, output)
  485. }
  486. }