test_parser.py 18 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619
  1. from django.test import TestCase
  2. from misago.markup.parser import parse
  3. from misago.users.testutils import create_test_user
  4. class MockRequest(object):
  5. scheme = "http"
  6. def __init__(self, user=None):
  7. self.user = user
  8. def get_host(self):
  9. return "test.com"
  10. class MockPoster(object):
  11. username = "LoremIpsum"
  12. slug = "loremipsum"
  13. class HTMLTests(TestCase):
  14. def test_html_escaped(self):
  15. """parser escapes all html"""
  16. test_text = """
  17. Lorem <strong>ipsum!</strong>
  18. """.strip()
  19. expected_result = """
  20. <p>Lorem &lt;strong&gt;ipsum!&lt;/strong&gt;</p>
  21. """.strip()
  22. result = parse(test_text, MockRequest(), MockPoster(), minify=True)
  23. self.assertEqual(expected_result, result["parsed_text"])
  24. self.assertEqual(result["internal_links"], [])
  25. self.assertEqual(result["images"], [])
  26. self.assertEqual(result["outgoing_links"], [])
  27. class BBCodeTests(TestCase):
  28. def test_inline_text(self):
  29. """inline elements are correctly parsed"""
  30. test_text = """
  31. Lorem **ipsum**, dolor met.
  32. Lorem [b]ipsum[/b], [i]dolor[/i] [u]met[/u].
  33. Lorem [b]**ipsum**[/b] [i]dolor[/i] [u]met[/u].
  34. Lorem [b]**ipsum[/b]** [i]dolor[/i] [u]met[/u].
  35. Lorem [b]__ipsum[/b]__ [i]dolor[/i] [u]met[/u].
  36. Lorem [b][i]ipsum[/i][/b].
  37. Lorem [b][i]ipsum[/b][/i].
  38. Lorem [b]ipsum[/B].
  39. """.strip()
  40. expected_result = """
  41. <p>Lorem <strong>ipsum</strong>, dolor met.</p>
  42. <p>Lorem <b>ipsum</b>, <i>dolor</i> <u>met</u>.</p>
  43. <p>Lorem <b><strong>ipsum</strong></b> <i>dolor</i> <u>met</u>.</p>
  44. <p>Lorem <b>**ipsum</b>** <i>dolor</i> <u>met</u>.</p>
  45. <p>Lorem <b>__ipsum</b>__ <i>dolor</i> <u>met</u>.</p>
  46. <p>Lorem <b><i>ipsum</i></b>.</p>
  47. <p>Lorem <b>[i]ipsum</b>[/i].</p>
  48. <p>Lorem <b>ipsum</b>.</p>
  49. """.strip()
  50. result = parse(test_text, MockRequest(), MockPoster(), minify=False)
  51. self.assertEqual(expected_result, result["parsed_text"])
  52. def test_hr(self):
  53. """hr bbcode is correctly parsed"""
  54. test_text = """
  55. Lorem ipsum.
  56. [hr]
  57. Dolor met.
  58. """.strip()
  59. expected_result = """
  60. <p>Lorem ipsum.</p>
  61. <hr/>
  62. <p>Dolor met.</p>
  63. """.strip()
  64. result = parse(test_text, MockRequest(), MockPoster(), minify=False)
  65. self.assertEqual(expected_result, result["parsed_text"])
  66. def test_img(self):
  67. """img bbcode is correctly parsed"""
  68. test_text = """
  69. Lorem ipsum [img]https://placekitten.com/g/1200/500[/img]
  70. Lorem ipsum [iMg]https://placekitten.com/g/1200/500[/ImG]
  71. Lorem ipsum !(https://placekitten.com/g/1200/500)
  72. """.strip()
  73. expected_result = """
  74. <p>Lorem ipsum <img alt="placekitten.com/g/1200/500" src="https://placekitten.com/g/1200/500"/></p>
  75. <p>Lorem ipsum <img alt="placekitten.com/g/1200/500" src="https://placekitten.com/g/1200/500"/></p>
  76. <p>Lorem ipsum <img alt="placekitten.com/g/1200/500" src="https://placekitten.com/g/1200/500"/></p>
  77. """.strip()
  78. result = parse(test_text, MockRequest(), MockPoster(), minify=False)
  79. self.assertEqual(expected_result, result["parsed_text"])
  80. def test_url(self):
  81. """url bbcode is correctly parsed"""
  82. test_text = """
  83. Lorem ipsum [url]placekitten.com/g/300/300[/url]
  84. Lorem ipsum [url]https://placekitten.com/g/600/600[/url]
  85. Lorem ipsum [uRL=https://placekitten.com/g/400/400"]Label text![/UrL]
  86. Lorem ipsum [Lorem ipsum](https://placekitten.com/g/1200/500)
  87. """.strip()
  88. expected_result = """
  89. <p>Lorem ipsum <a href="http://placekitten.com/g/300/300" rel="nofollow noopener">placekitten.com/g/300/300</a></p>
  90. <p>Lorem ipsum <a href="https://placekitten.com/g/600/600" rel="nofollow noopener">placekitten.com/g/600/600</a></p>
  91. <p>Lorem ipsum <a href="https://placekitten.com/g/400/400" rel="nofollow noopener">Label text!</a></p>
  92. <p>Lorem ipsum <a href="https://placekitten.com/g/1200/500" rel="nofollow noopener">Lorem ipsum</a></p>
  93. """.strip()
  94. result = parse(test_text, MockRequest(), MockPoster(), minify=False)
  95. self.assertEqual(expected_result, result["parsed_text"])
  96. class MinifyTests(TestCase):
  97. def test_minified_text(self):
  98. """parser minifies text successfully"""
  99. test_text = """
  100. Lorem ipsum.
  101. Lorem ipsum.
  102. """.strip()
  103. expected_result = """
  104. <p>Lorem ipsum.</p><p>Lorem ipsum.</p>
  105. """.strip()
  106. result = parse(test_text, MockRequest(), MockPoster(), minify=True)
  107. self.assertEqual(expected_result, result["parsed_text"])
  108. def test_minified_unicode_text(self):
  109. """parser minifies unicode text successfully"""
  110. test_text = """
  111. Bżęczyszczykiewłicz ipsum.
  112. Lorem ipsum.
  113. """.strip()
  114. expected_result = """
  115. <p>Bżęczyszczykiewłicz ipsum.</p><p>Lorem ipsum.</p>
  116. """.strip()
  117. result = parse(test_text, MockRequest(), MockPoster(), minify=True)
  118. self.assertEqual(expected_result, result["parsed_text"])
  119. def test_complex_paragraph(self):
  120. """parser minifies complex paragraph"""
  121. user = create_test_user("User", "user@example.com")
  122. test_text = (
  123. """
  124. Hey there @%s, how's going?
  125. """.strip()
  126. % user
  127. )
  128. expected_result = """
  129. <p>Hey there <a href="%s">@%s</a>, how's going?</p>
  130. """.strip() % (
  131. user.get_absolute_url(),
  132. user,
  133. )
  134. result = parse(test_text, MockRequest(user), user, minify=True)
  135. self.assertEqual(expected_result, result["parsed_text"])
  136. class CleanLinksTests(TestCase):
  137. def test_clean_current_link(self):
  138. """clean_links step cleans http://test.com"""
  139. test_text = """
  140. Lorem ipsum: http://test.com
  141. """.strip()
  142. expected_result = """
  143. <p>Lorem ipsum: <a href="/">test.com</a></p>
  144. """.strip()
  145. result = parse(test_text, MockRequest(), MockPoster(), minify=True)
  146. self.assertEqual(expected_result, result["parsed_text"])
  147. self.assertEqual(result["internal_links"], ["/"])
  148. self.assertEqual(result["images"], [])
  149. self.assertEqual(result["outgoing_links"], [])
  150. def test_clean_schemaless_link(self):
  151. """clean_links step cleans test.com"""
  152. test_text = """
  153. Lorem ipsum: test.com
  154. """.strip()
  155. expected_result = """
  156. <p>Lorem ipsum: <a href="/">test.com</a></p>
  157. """.strip()
  158. result = parse(test_text, MockRequest(), MockPoster(), minify=True)
  159. self.assertEqual(expected_result, result["parsed_text"])
  160. self.assertEqual(result["internal_links"], ["/"])
  161. self.assertEqual(result["images"], [])
  162. self.assertEqual(result["outgoing_links"], [])
  163. def test_trim_current_path(self):
  164. """clean_links step leaves http://test.com path"""
  165. test_text = """
  166. Lorem ipsum: http://test.com/somewhere-something/
  167. """.strip()
  168. expected_result = """
  169. <p>Lorem ipsum: <a href="/somewhere-something/">test.com/somewhere-something/</a></p>
  170. """.strip()
  171. result = parse(test_text, MockRequest(), MockPoster(), minify=True)
  172. self.assertEqual(expected_result, result["parsed_text"])
  173. self.assertEqual(result["internal_links"], ["/somewhere-something/"])
  174. self.assertEqual(result["images"], [])
  175. self.assertEqual(result["outgoing_links"], [])
  176. def test_clean_outgoing_link_domain(self):
  177. """clean_links step leaves outgoing domain link"""
  178. test_text = """
  179. Lorem ipsum: http://somewhere.com
  180. """.strip()
  181. expected_result = """
  182. <p>Lorem ipsum: <a href="http://somewhere.com" rel="nofollow noopener">somewhere.com</a></p>
  183. """.strip()
  184. result = parse(test_text, MockRequest(), MockPoster(), minify=True)
  185. self.assertEqual(expected_result, result["parsed_text"])
  186. self.assertEqual(result["outgoing_links"], ["somewhere.com"])
  187. self.assertEqual(result["images"], [])
  188. self.assertEqual(result["internal_links"], [])
  189. def test_trim_outgoing_path(self):
  190. """clean_links step leaves outgoing link domain and path"""
  191. test_text = """
  192. Lorem ipsum: http://somewhere.com/somewhere-something/
  193. """.strip()
  194. expected_result = """
  195. <p>Lorem ipsum: <a href="http://somewhere.com/somewhere-something/" rel="nofollow noopener">somewhere.com/somewhere-something/</a></p>
  196. """.strip()
  197. result = parse(test_text, MockRequest(), MockPoster(), minify=True)
  198. self.assertEqual(expected_result, result["parsed_text"])
  199. self.assertEqual(
  200. result["outgoing_links"], ["somewhere.com/somewhere-something/"]
  201. )
  202. self.assertEqual(result["images"], [])
  203. self.assertEqual(result["internal_links"], [])
  204. def test_clean_local_image_src(self):
  205. """clean_links step cleans local image src"""
  206. test_text = """
  207. !(http://test.com/image.jpg)
  208. """.strip()
  209. expected_result = """
  210. <p><img alt="test.com/image.jpg" src="/image.jpg"/></p>
  211. """.strip()
  212. result = parse(test_text, MockRequest(), MockPoster(), minify=True)
  213. self.assertEqual(expected_result, result["parsed_text"])
  214. self.assertEqual(result["images"], ["/image.jpg"])
  215. self.assertEqual(result["internal_links"], [])
  216. self.assertEqual(result["outgoing_links"], [])
  217. def test_clean_remote_image_src(self):
  218. """clean_links step cleans remote image src"""
  219. test_text = """
  220. !(http://somewhere.com/image.jpg)
  221. """.strip()
  222. expected_result = """
  223. <p><img alt="somewhere.com/image.jpg" src="http://somewhere.com/image.jpg"/></p>
  224. """.strip()
  225. result = parse(test_text, MockRequest(), MockPoster(), minify=True)
  226. self.assertEqual(expected_result, result["parsed_text"])
  227. self.assertEqual(result["images"], ["somewhere.com/image.jpg"])
  228. self.assertEqual(result["internal_links"], [])
  229. self.assertEqual(result["outgoing_links"], [])
  230. def test_clean_linked_image(self):
  231. """parser handles image element nested in link"""
  232. test_text = """
  233. [![3.png](http://test.com/a/thumb/test/43/)](http://test.com/a/test/43/)
  234. """.strip()
  235. expected_result = """
  236. <p><a href="/a/test/43/"><img alt="3.png" src="/a/thumb/test/43/"/></a></p>
  237. """.strip()
  238. result = parse(test_text, MockRequest(), MockPoster(), minify=True)
  239. self.assertEqual(expected_result, result["parsed_text"])
  240. self.assertEqual(result["images"], ["/a/thumb/test/43/"])
  241. self.assertEqual(result["internal_links"], ["/a/test/43/"])
  242. self.assertEqual(result["outgoing_links"], [])
  243. def test_force_shva(self):
  244. """parser appends ?shva=1 bit to attachment links if flag is present"""
  245. test_text = """
  246. ![3.png](http://test.com/a/thumb/test/43/)
  247. """.strip()
  248. expected_result = """
  249. <p><img alt="3.png" src="/a/thumb/test/43/?shva=1"/></p>
  250. """.strip()
  251. result = parse(
  252. test_text, MockRequest(), MockPoster(), minify=True, force_shva=True
  253. )
  254. self.assertEqual(expected_result, result["parsed_text"])
  255. self.assertEqual(result["images"], ["/a/thumb/test/43/"])
  256. self.assertEqual(result["internal_links"], [])
  257. self.assertEqual(result["outgoing_links"], [])
  258. def test_remove_shva(self):
  259. """parser removes ?shva=1 bit from attachment links if flag is absent"""
  260. test_text = """
  261. ![3.png](http://test.com/a/thumb/test/43/?shva=1)
  262. """.strip()
  263. expected_result = """
  264. <p><img alt="3.png" src="/a/thumb/test/43/"/></p>
  265. """.strip()
  266. result = parse(test_text, MockRequest(), MockPoster(), minify=True)
  267. self.assertEqual(expected_result, result["parsed_text"])
  268. self.assertEqual(result["images"], ["/a/thumb/test/43/?shva=1"])
  269. self.assertEqual(result["internal_links"], [])
  270. self.assertEqual(result["outgoing_links"], [])
  271. class LinkifyTests(TestCase):
  272. def test_clean_current_link(self):
  273. """clean_links step cleans http://test.com"""
  274. test_text = """
  275. Lorem ipsum: `<http://test.com>`
  276. """.strip()
  277. expected_result = """
  278. <p>Lorem ipsum: <code>&lt;http://test.com&gt;</code></p>
  279. """.strip()
  280. result = parse(test_text, MockRequest(), MockPoster(), minify=True)
  281. self.assertEqual(expected_result, result["parsed_text"])
  282. self.assertEqual(result["internal_links"], [])
  283. self.assertEqual(result["images"], [])
  284. self.assertEqual(result["outgoing_links"], [])
  285. class StriketroughTests(TestCase):
  286. def test_striketrough(self):
  287. """striketrough markdown deletes test"""
  288. test_text = """
  289. Lorem ~~ipsum, dolor~~ met.
  290. """.strip()
  291. expected_result = """
  292. <p>Lorem <del>ipsum, dolor</del> met.</p>
  293. """.strip()
  294. result = parse(test_text, MockRequest(), MockPoster(), minify=False)
  295. self.assertEqual(expected_result, result["parsed_text"])
  296. class QuoteTests(TestCase):
  297. def test_quotes(self):
  298. """bbcode for quote is supported"""
  299. test_text = """
  300. Lorem ipsum.
  301. [quote]Dolor met[/quote]
  302. [quote]Dolor <b>met</b>[/quote]
  303. [quote]Dolor **met**[quote]Dolor met[/quote][/quote]
  304. """.strip()
  305. expected_result = """
  306. <p>Lorem ipsum.</p>
  307. <aside class="quote-block">
  308. <div class="quote-heading"></div>
  309. <blockquote class="quote-body">
  310. <p>Dolor met</p>
  311. </blockquote>
  312. </aside>
  313. <aside class="quote-block">
  314. <div class="quote-heading"></div>
  315. <blockquote class="quote-body">
  316. <p>Dolor &lt;b&gt;met&lt;/b&gt;</p>
  317. </blockquote>
  318. </aside>
  319. <aside class="quote-block">
  320. <div class="quote-heading"></div>
  321. <blockquote class="quote-body">
  322. <p>Dolor <strong>met</strong></p>
  323. <aside class="quote-block">
  324. <div class="quote-heading"></div>
  325. <blockquote class="quote-body">
  326. <p>Dolor met</p>
  327. </blockquote>
  328. </aside>
  329. </blockquote>
  330. </aside>
  331. """.strip()
  332. result = parse(test_text, MockRequest(), MockPoster(), minify=False)
  333. self.assertEqual(expected_result, result["parsed_text"])
  334. def test_authored_quotes(self):
  335. """bbcode for authored quote is supported and handles mentions as well"""
  336. test_text = """
  337. Lorem ipsum.
  338. [quote]Dolor met[/quote]
  339. [quote=\"@Bob\"]Dolor <b>met</b>[/quote]
  340. [quote]Dolor **met**[quote=@Bob]Dolor met[/quote][/quote]
  341. """.strip()
  342. expected_result = """
  343. <p>Lorem ipsum.</p>
  344. <aside class="quote-block">
  345. <div class="quote-heading"></div>
  346. <blockquote class="quote-body">
  347. <p>Dolor met</p>
  348. </blockquote>
  349. </aside>
  350. <aside class="quote-block">
  351. <div class="quote-heading">@Bob</div>
  352. <blockquote class="quote-body">
  353. <p>Dolor &lt;b&gt;met&lt;/b&gt;</p>
  354. </blockquote>
  355. </aside>
  356. <aside class="quote-block">
  357. <div class="quote-heading"></div>
  358. <blockquote class="quote-body">
  359. <p>Dolor <strong>met</strong></p>
  360. <aside class="quote-block">
  361. <div class="quote-heading">@Bob</div>
  362. <blockquote class="quote-body">
  363. <p>Dolor met</p>
  364. </blockquote>
  365. </aside>
  366. </blockquote>
  367. </aside>
  368. """.strip()
  369. request = MockRequest(user=MockPoster())
  370. result = parse(test_text, request, MockPoster(), minify=False)
  371. self.assertEqual(expected_result, result["parsed_text"])
  372. def test_custom_quote_title(self):
  373. """parser handles custom quotetitle"""
  374. test_text = """
  375. Lorem ipsum.
  376. [quote=\"Lorem ipsum very test\"]Dolor <b>met</b>[/quote]
  377. """.strip()
  378. expected_result = """
  379. <p>Lorem ipsum.</p>
  380. <aside class="quote-block">
  381. <div class="quote-heading">Lorem ipsum very test</div>
  382. <blockquote class="quote-body">
  383. <p>Dolor &lt;b&gt;met&lt;/b&gt;</p>
  384. </blockquote>
  385. </aside>
  386. """.strip()
  387. result = parse(test_text, MockRequest(), MockPoster(), minify=False)
  388. self.assertEqual(expected_result, result["parsed_text"])
  389. def test_hr_edge_case(self):
  390. """test for weird edge case in which hr gets moved outside of quote"""
  391. test_text = """
  392. Lorem ipsum.
  393. [quote]
  394. Dolor met
  395. - - - - -
  396. Amet elit
  397. [/quote]
  398. """.strip()
  399. expected_result = """
  400. <p>Lorem ipsum.</p>
  401. <aside class="quote-block">
  402. <div class="quote-heading"></div>
  403. <blockquote class="quote-body">
  404. <p>Dolor met</p>
  405. <hr/>
  406. <p>Amet elit</p>
  407. </blockquote>
  408. </aside>
  409. """.strip()
  410. result = parse(test_text, MockRequest(), MockPoster(), minify=False)
  411. self.assertEqual(expected_result, result["parsed_text"])
  412. class CodeTests(TestCase):
  413. def test_code(self):
  414. """code bbcode is correctly parsed"""
  415. test_text = """
  416. Lorem ipsum.
  417. [code]
  418. Dolor [b]met.[/b]
  419. [/code]
  420. """.strip()
  421. expected_result = """
  422. <p>Lorem ipsum.</p>
  423. <pre><code>Dolor [b]met.[/b]</code></pre>
  424. """.strip()
  425. result = parse(test_text, MockRequest(), MockPoster(), minify=False)
  426. self.assertEqual(expected_result, result["parsed_text"])
  427. def test_inline_code(self):
  428. """inline code bbcode is correctly parsed"""
  429. test_text = """
  430. Lorem ipsum.
  431. [code]Dolor [b]met.[/b][/code]
  432. """.strip()
  433. expected_result = """
  434. <p>Lorem ipsum.</p>
  435. <pre><code>Dolor [b]met.[/b]</code></pre>
  436. """.strip()
  437. result = parse(test_text, MockRequest(), MockPoster(), minify=False)
  438. self.assertEqual(expected_result, result["parsed_text"])
  439. def test_code_strip(self):
  440. """code bbcode trims its content"""
  441. test_text = """
  442. Lorem ipsum.
  443. [code]
  444. Dolor [b]met.[/b]
  445. [/code]
  446. """.strip()
  447. expected_result = """
  448. <p>Lorem ipsum.</p>
  449. <pre><code> Dolor [b]met.[/b]</code></pre>
  450. """.strip()
  451. result = parse(test_text, MockRequest(), MockPoster(), minify=False)
  452. self.assertEqual(expected_result, result["parsed_text"])
  453. def test_code_language(self):
  454. """code bbcode with language is correctly parsed"""
  455. test_text = """
  456. Lorem ipsum.
  457. [code="python"]
  458. Dolor [b]met.[/b]
  459. [/code]
  460. """.strip()
  461. expected_result = """
  462. <p>Lorem ipsum.</p>
  463. <pre><code class="python">Dolor [b]met.[/b]</code></pre>
  464. """.strip()
  465. result = parse(test_text, MockRequest(), MockPoster(), minify=False)
  466. self.assertEqual(expected_result, result["parsed_text"])
  467. def test_code_language_optional_quotes(self):
  468. """code quotes around language name are optional"""
  469. test_text = """
  470. Lorem ipsum.
  471. [code=python"]
  472. Dolor [b]met.[/b]
  473. [/code]
  474. """.strip()
  475. expected_result = """
  476. <p>Lorem ipsum.</p>
  477. <pre><code class="python">Dolor [b]met.[/b]</code></pre>
  478. """.strip()
  479. result = parse(test_text, MockRequest(), MockPoster(), minify=False)
  480. self.assertEqual(expected_result, result["parsed_text"])
  481. test_text = """
  482. Lorem ipsum.
  483. [code="python]
  484. Dolor [b]met.[/b]
  485. [/code]
  486. """.strip()
  487. expected_result = """
  488. <p>Lorem ipsum.</p>
  489. <pre><code class="python">Dolor [b]met.[/b]</code></pre>
  490. """.strip()
  491. result = parse(test_text, MockRequest(), MockPoster(), minify=False)
  492. self.assertEqual(expected_result, result["parsed_text"])