test_parser.py 17 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594
  1. # -*- coding: utf-8 -*-
  2. from __future__ import unicode_literals
  3. from django.contrib.auth import get_user_model
  4. from django.test import TestCase
  5. from misago.markup.parser import parse
  6. UserModel = get_user_model()
  7. class MockRequest(object):
  8. scheme = 'http'
  9. def __init__(self, user=None):
  10. self.user = user
  11. def get_host(self):
  12. return 'test.com'
  13. class MockPoster(object):
  14. username = 'LoremIpsum'
  15. slug = 'loremipsum'
  16. class BBCodeTests(TestCase):
  17. def test_inline_text(self):
  18. """inline elements are correctly parsed"""
  19. test_text = """
  20. Lorem **ipsum**, dolor met.
  21. Lorem [b]ipsum[/b], [i]dolor[/i] [u]met[/u].
  22. Lorem [b]**ipsum**[/b] [i]dolor[/i] [u]met[/u].
  23. Lorem [b]**ipsum[/b]** [i]dolor[/i] [u]met[/u].
  24. Lorem [b]__ipsum[/b]__ [i]dolor[/i] [u]met[/u].
  25. Lorem [b][i]ipsum[/i][/b].
  26. Lorem [b][i]ipsum[/b][/i].
  27. Lorem [b]ipsum[/B].
  28. """.strip()
  29. expected_result = """
  30. <p>Lorem <strong>ipsum</strong>, dolor met.</p>
  31. <p>Lorem <b>ipsum</b>, <i>dolor</i> <u>met</u>.</p>
  32. <p>Lorem <b><strong>ipsum</strong></b> <i>dolor</i> <u>met</u>.</p>
  33. <p>Lorem <b>**ipsum</b>** <i>dolor</i> <u>met</u>.</p>
  34. <p>Lorem <b>__ipsum</b>__ <i>dolor</i> <u>met</u>.</p>
  35. <p>Lorem <b><i>ipsum</i></b>.</p>
  36. <p>Lorem <b>[i]ipsum</b>[/i].</p>
  37. <p>Lorem <b>ipsum</b>.</p>
  38. """.strip()
  39. result = parse(test_text, MockRequest(), MockPoster(), minify=False)
  40. self.assertEqual(expected_result, result['parsed_text'])
  41. def test_hr(self):
  42. """hr bbcode is correctly parsed"""
  43. test_text = """
  44. Lorem ipsum.
  45. [hr]
  46. Dolor met.
  47. """.strip()
  48. expected_result = """
  49. <p>Lorem ipsum.</p>
  50. <hr/>
  51. <p>Dolor met.</p>
  52. """.strip()
  53. result = parse(test_text, MockRequest(), MockPoster(), minify=False)
  54. self.assertEqual(expected_result, result['parsed_text'])
  55. def test_img(self):
  56. """img bbcode is correctly parsed"""
  57. test_text = """
  58. Lorem ipsum [img]https://placekitten.com/g/1200/500[/img]
  59. Lorem ipsum [iMg]https://placekitten.com/g/1200/500[/ImG]
  60. Lorem ipsum !(https://placekitten.com/g/1200/500)
  61. """.strip()
  62. expected_result = """
  63. <p>Lorem ipsum <img alt="placekitten.com/g/1200/500" src="https://placekitten.com/g/1200/500"/></p>
  64. <p>Lorem ipsum <img alt="placekitten.com/g/1200/500" src="https://placekitten.com/g/1200/500"/></p>
  65. <p>Lorem ipsum <img alt="placekitten.com/g/1200/500" src="https://placekitten.com/g/1200/500"/></p>
  66. """.strip()
  67. result = parse(test_text, MockRequest(), MockPoster(), minify=False)
  68. self.assertEqual(expected_result, result['parsed_text'])
  69. def test_url(self):
  70. """url bbcode is correctly parsed"""
  71. test_text = """
  72. Lorem ipsum [url]https://placekitten.com/g/600/600[/url]
  73. Lorem ipsum [uRL=https://placekitten.com/g/400/400"]Label text![/UrL]
  74. Lorem ipsum [Lorem ipsum](https://placekitten.com/g/1200/500)
  75. """.strip()
  76. expected_result = """
  77. <p>Lorem ipsum <a href="" rel="nofollow">placekitten.com/g/600/600</a></p>
  78. <p>Lorem ipsum <a href="https://placekitten.com/g/400/400" rel="nofollow">Label text!</a></p>
  79. <p>Lorem ipsum <a href="https://placekitten.com/g/1200/500" rel="nofollow">Lorem ipsum</a></p>
  80. """.strip()
  81. result = parse(test_text, MockRequest(), MockPoster(), minify=False)
  82. self.assertEqual(expected_result, result['parsed_text'])
  83. class MinifyTests(TestCase):
  84. def test_minified_text(self):
  85. """parser minifies text successfully"""
  86. test_text = """
  87. Lorem ipsum.
  88. Lorem ipsum.
  89. """.strip()
  90. expected_result = """
  91. <p>Lorem ipsum.</p><p>Lorem ipsum.</p>
  92. """.strip()
  93. result = parse(test_text, MockRequest(), MockPoster(), minify=True)
  94. self.assertEqual(expected_result, result['parsed_text'])
  95. def test_minified_unicode_text(self):
  96. """parser minifies unicode text successfully"""
  97. test_text = """
  98. Bżęczyszczykiewłicz ipsum.
  99. Lorem ipsum.
  100. """.strip()
  101. expected_result = """
  102. <p>Bżęczyszczykiewłicz ipsum.</p><p>Lorem ipsum.</p>
  103. """.strip()
  104. result = parse(test_text, MockRequest(), MockPoster(), minify=True)
  105. self.assertEqual(expected_result, result['parsed_text'])
  106. def test_complex_paragraph(self):
  107. """parser minifies complex paragraph"""
  108. user = UserModel.objects.create_user('Bob', 'bob@test.com', 'Pass123')
  109. test_text = """
  110. Hey there @{}, how's going?
  111. """.strip().format(user)
  112. expected_result = """
  113. <p>Hey there <a href="{}">@{}</a>, how's going?</p>
  114. """.strip().format(user.get_absolute_url(), user)
  115. result = parse(test_text, MockRequest(user), user, minify=True)
  116. self.assertEqual(expected_result, result['parsed_text'])
  117. class CleanLinksTests(TestCase):
  118. def test_clean_current_link(self):
  119. """clean_links step cleans http://test.com"""
  120. test_text = """
  121. Lorem ipsum: http://test.com
  122. """.strip()
  123. expected_result = """
  124. <p>Lorem ipsum: <a href="/" rel="nofollow">test.com</a></p>
  125. """.strip()
  126. result = parse(test_text, MockRequest(), MockPoster(), minify=True)
  127. self.assertEqual(expected_result, result['parsed_text'])
  128. self.assertEqual(result['internal_links'], ['/'])
  129. self.assertEqual(result['images'], [])
  130. self.assertEqual(result['outgoing_links'], [])
  131. def test_clean_schemaless_link(self):
  132. """clean_links step cleans test.com"""
  133. test_text = """
  134. Lorem ipsum: test.com
  135. """.strip()
  136. expected_result = """
  137. <p>Lorem ipsum: <a href="/" rel="nofollow">test.com</a></p>
  138. """.strip()
  139. result = parse(test_text, MockRequest(), MockPoster(), minify=True)
  140. self.assertEqual(expected_result, result['parsed_text'])
  141. self.assertEqual(result['internal_links'], ['/'])
  142. self.assertEqual(result['images'], [])
  143. self.assertEqual(result['outgoing_links'], [])
  144. def test_trim_current_path(self):
  145. """clean_links step leaves http://test.com path"""
  146. test_text = """
  147. Lorem ipsum: http://test.com/somewhere-something/
  148. """.strip()
  149. expected_result = """
  150. <p>Lorem ipsum: <a href="/somewhere-something/" rel="nofollow">test.com/somewhere-something/</a></p>
  151. """.strip()
  152. result = parse(test_text, MockRequest(), MockPoster(), minify=True)
  153. self.assertEqual(expected_result, result['parsed_text'])
  154. self.assertEqual(result['internal_links'], ['/somewhere-something/'])
  155. self.assertEqual(result['images'], [])
  156. self.assertEqual(result['outgoing_links'], [])
  157. def test_clean_outgoing_link_domain(self):
  158. """clean_links step leaves outgoing domain link"""
  159. test_text = """
  160. Lorem ipsum: http://somewhere.com
  161. """.strip()
  162. expected_result = """
  163. <p>Lorem ipsum: <a href="http://somewhere.com" rel="nofollow">somewhere.com</a></p>
  164. """.strip()
  165. result = parse(test_text, MockRequest(), MockPoster(), minify=True)
  166. self.assertEqual(expected_result, result['parsed_text'])
  167. self.assertEqual(result['outgoing_links'], ['http://somewhere.com'])
  168. self.assertEqual(result['images'], [])
  169. self.assertEqual(result['internal_links'], [])
  170. def test_trim_outgoing_path(self):
  171. """clean_links step leaves outgoing link domain and path"""
  172. test_text = """
  173. Lorem ipsum: http://somewhere.com/somewhere-something/
  174. """.strip()
  175. expected_result = """
  176. <p>Lorem ipsum: <a href="http://somewhere.com/somewhere-something/" rel="nofollow">somewhere.com/somewhere-something/</a></p>
  177. """.strip()
  178. result = parse(test_text, MockRequest(), MockPoster(), minify=True)
  179. self.assertEqual(expected_result, result['parsed_text'])
  180. self.assertEqual(result['outgoing_links'], ['http://somewhere.com/somewhere-something/'])
  181. self.assertEqual(result['images'], [])
  182. self.assertEqual(result['internal_links'], [])
  183. def test_clean_local_image_src(self):
  184. """clean_links step cleans local image src"""
  185. test_text = """
  186. !(http://test.com/image.jpg)
  187. """.strip()
  188. expected_result = """
  189. <p><img alt="test.com/image.jpg" src="/image.jpg"/></p>
  190. """.strip()
  191. result = parse(test_text, MockRequest(), MockPoster(), minify=True)
  192. self.assertEqual(expected_result, result['parsed_text'])
  193. self.assertEqual(result['images'], ['/image.jpg'])
  194. self.assertEqual(result['internal_links'], [])
  195. self.assertEqual(result['outgoing_links'], [])
  196. def test_clean_remote_image_src(self):
  197. """clean_links step cleans remote image src"""
  198. test_text = """
  199. !(http://somewhere.com/image.jpg)
  200. """.strip()
  201. expected_result = """
  202. <p><img alt="somewhere.com/image.jpg" src="http://somewhere.com/image.jpg"/></p>
  203. """.strip()
  204. result = parse(test_text, MockRequest(), MockPoster(), minify=True)
  205. self.assertEqual(expected_result, result['parsed_text'])
  206. self.assertEqual(result['images'], ['somewhere.com/image.jpg'])
  207. self.assertEqual(result['internal_links'], [])
  208. self.assertEqual(result['outgoing_links'], [])
  209. def test_clean_linked_image(self):
  210. """parser handles image element nested in link"""
  211. test_text = """
  212. [![3.png](http://test.com/a/thumb/test/43/)](http://test.com/a/test/43/)
  213. """.strip()
  214. expected_result = """
  215. <p><a href="/a/test/43/" rel="nofollow"><img alt="3.png" src="/a/thumb/test/43/"/></a></p>
  216. """.strip()
  217. result = parse(test_text, MockRequest(), MockPoster(), minify=True)
  218. self.assertEqual(expected_result, result['parsed_text'])
  219. self.assertEqual(result['images'], ['/a/thumb/test/43/'])
  220. self.assertEqual(result['internal_links'], ['/a/test/43/'])
  221. self.assertEqual(result['outgoing_links'], [])
  222. def test_force_shva(self):
  223. """parser appends ?shva=1 bit to attachment links if flag is present"""
  224. test_text = """
  225. ![3.png](http://test.com/a/thumb/test/43/)
  226. """.strip()
  227. expected_result = """
  228. <p><img alt="3.png" src="/a/thumb/test/43/?shva=1"/></p>
  229. """.strip()
  230. result = parse(test_text, MockRequest(), MockPoster(), minify=True, force_shva=True)
  231. self.assertEqual(expected_result, result['parsed_text'])
  232. self.assertEqual(result['images'], ['/a/thumb/test/43/'])
  233. self.assertEqual(result['internal_links'], [])
  234. self.assertEqual(result['outgoing_links'], [])
  235. def test_remove_shva(self):
  236. """parser removes ?shva=1 bit from attachment links if flag is absent"""
  237. test_text = """
  238. ![3.png](http://test.com/a/thumb/test/43/?shva=1)
  239. """.strip()
  240. expected_result = """
  241. <p><img alt="3.png" src="/a/thumb/test/43/"/></p>
  242. """.strip()
  243. result = parse(test_text, MockRequest(), MockPoster(), minify=True)
  244. self.assertEqual(expected_result, result['parsed_text'])
  245. self.assertEqual(result['images'], ['/a/thumb/test/43/?shva=1'])
  246. self.assertEqual(result['internal_links'], [])
  247. self.assertEqual(result['outgoing_links'], [])
  248. class LinkifyTests(TestCase):
  249. def test_clean_current_link(self):
  250. """clean_links step cleans http://test.com"""
  251. test_text = """
  252. Lorem ipsum: `<http://test.com>`
  253. """.strip()
  254. expected_result = """
  255. <p>Lorem ipsum: <code>&lt;http://test.com&gt;</code></p>
  256. """.strip()
  257. result = parse(test_text, MockRequest(), MockPoster(), minify=True)
  258. self.assertEqual(expected_result, result['parsed_text'])
  259. self.assertEqual(result['internal_links'], [])
  260. self.assertEqual(result['images'], [])
  261. self.assertEqual(result['outgoing_links'], [])
  262. class StriketroughTests(TestCase):
  263. def test_striketrough(self):
  264. """striketrough markdown deletes test"""
  265. test_text = """
  266. Lorem ~~ipsum, dolor~~ met.
  267. """.strip()
  268. expected_result = """
  269. <p>Lorem <del>ipsum, dolor</del> met.</p>
  270. """.strip()
  271. result = parse(test_text, MockRequest(), MockPoster(), minify=False)
  272. self.assertEqual(expected_result, result['parsed_text'])
  273. class QuoteTests(TestCase):
  274. def test_quotes(self):
  275. """bbcode for quote is supported"""
  276. test_text = """
  277. Lorem ipsum.
  278. [quote]Dolor met[/quote]
  279. [quote]Dolor <b>met</b>[/quote]
  280. [quote]Dolor **met**[quote]Dolor met[/quote][/quote]
  281. """.strip()
  282. expected_result = """
  283. <p>Lorem ipsum.</p>
  284. <aside class="quote-block">
  285. <div class="quote-heading"></div>
  286. <blockquote class="quote-body">
  287. <p>Dolor met</p>
  288. </blockquote>
  289. </aside>
  290. <aside class="quote-block">
  291. <div class="quote-heading"></div>
  292. <blockquote class="quote-body">
  293. <p>Dolor &lt;b&gt;met&lt;/b&gt;</p>
  294. </blockquote>
  295. </aside>
  296. <aside class="quote-block">
  297. <div class="quote-heading"></div>
  298. <blockquote class="quote-body">
  299. <p>Dolor <strong>met</strong></p>
  300. <aside class="quote-block">
  301. <div class="quote-heading"></div>
  302. <blockquote class="quote-body">
  303. <p>Dolor met</p>
  304. </blockquote>
  305. </aside>
  306. </blockquote>
  307. </aside>
  308. """.strip()
  309. result = parse(test_text, MockRequest(), MockPoster(), minify=False)
  310. self.assertEqual(expected_result, result['parsed_text'])
  311. def test_authored_quotes(self):
  312. """bbcode for authored quote is supported and handles mentions as well"""
  313. test_text = """
  314. Lorem ipsum.
  315. [quote]Dolor met[/quote]
  316. [quote=\"@Bob\"]Dolor <b>met</b>[/quote]
  317. [quote]Dolor **met**[quote=@Bob]Dolor met[/quote][/quote]
  318. """.strip()
  319. expected_result = """
  320. <p>Lorem ipsum.</p>
  321. <aside class="quote-block">
  322. <div class="quote-heading"></div>
  323. <blockquote class="quote-body">
  324. <p>Dolor met</p>
  325. </blockquote>
  326. </aside>
  327. <aside class="quote-block">
  328. <div class="quote-heading">@Bob</div>
  329. <blockquote class="quote-body">
  330. <p>Dolor &lt;b&gt;met&lt;/b&gt;</p>
  331. </blockquote>
  332. </aside>
  333. <aside class="quote-block">
  334. <div class="quote-heading"></div>
  335. <blockquote class="quote-body">
  336. <p>Dolor <strong>met</strong></p>
  337. <aside class="quote-block">
  338. <div class="quote-heading">@Bob</div>
  339. <blockquote class="quote-body">
  340. <p>Dolor met</p>
  341. </blockquote>
  342. </aside>
  343. </blockquote>
  344. </aside>
  345. """.strip()
  346. request = MockRequest(user=MockPoster())
  347. result = parse(test_text, request, MockPoster(), minify=False)
  348. self.assertEqual(expected_result, result['parsed_text'])
  349. def test_custom_quote_title(self):
  350. """parser handles custom quotetitle"""
  351. test_text = """
  352. Lorem ipsum.
  353. [quote=\"Lorem ipsum very test\"]Dolor <b>met</b>[/quote]
  354. """.strip()
  355. expected_result = """
  356. <p>Lorem ipsum.</p>
  357. <aside class="quote-block">
  358. <div class="quote-heading">Lorem ipsum very test</div>
  359. <blockquote class="quote-body">
  360. <p>Dolor &lt;b&gt;met&lt;/b&gt;</p>
  361. </blockquote>
  362. </aside>
  363. """.strip()
  364. result = parse(test_text, MockRequest(), MockPoster(), minify=False)
  365. self.assertEqual(expected_result, result['parsed_text'])
  366. def test_hr_edge_case(self):
  367. """test for weird edge case in which hr gets moved outside of quote"""
  368. test_text = """
  369. Lorem ipsum.
  370. [quote]
  371. Dolor met
  372. - - - - -
  373. Amet elit
  374. [/quote]
  375. """.strip()
  376. expected_result = """
  377. <p>Lorem ipsum.</p>
  378. <aside class="quote-block">
  379. <div class="quote-heading"></div>
  380. <blockquote class="quote-body">
  381. <p>Dolor met</p>
  382. <hr/>
  383. <p>Amet elit</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. class CodeTests(TestCase):
  390. def test_code(self):
  391. """code bbcode is correctly parsed"""
  392. test_text = """
  393. Lorem ipsum.
  394. [code]
  395. Dolor [b]met.[/b]
  396. [/code]
  397. """.strip()
  398. expected_result = """
  399. <p>Lorem ipsum.</p>
  400. <pre><code>Dolor [b]met.[/b]</code></pre>
  401. """.strip()
  402. result = parse(test_text, MockRequest(), MockPoster(), minify=False)
  403. self.assertEqual(expected_result, result['parsed_text'])
  404. def test_inline_code(self):
  405. """inline code bbcode is correctly parsed"""
  406. test_text = """
  407. Lorem ipsum.
  408. [code]Dolor [b]met.[/b][/code]
  409. """.strip()
  410. expected_result = """
  411. <p>Lorem ipsum.</p>
  412. <pre><code>Dolor [b]met.[/b]</code></pre>
  413. """.strip()
  414. result = parse(test_text, MockRequest(), MockPoster(), minify=False)
  415. self.assertEqual(expected_result, result['parsed_text'])
  416. def test_code_strip(self):
  417. """code bbcode trims its content"""
  418. test_text = """
  419. Lorem ipsum.
  420. [code]
  421. Dolor [b]met.[/b]
  422. [/code]
  423. """.strip()
  424. expected_result = """
  425. <p>Lorem ipsum.</p>
  426. <pre><code> Dolor [b]met.[/b]</code></pre>
  427. """.strip()
  428. result = parse(test_text, MockRequest(), MockPoster(), minify=False)
  429. self.assertEqual(expected_result, result['parsed_text'])
  430. def test_code_language(self):
  431. """code bbcode with language is correctly parsed"""
  432. test_text = """
  433. Lorem ipsum.
  434. [code="python"]
  435. Dolor [b]met.[/b]
  436. [/code]
  437. """.strip()
  438. expected_result = """
  439. <p>Lorem ipsum.</p>
  440. <pre><code class="python">Dolor [b]met.[/b]</code></pre>
  441. """.strip()
  442. result = parse(test_text, MockRequest(), MockPoster(), minify=False)
  443. self.assertEqual(expected_result, result['parsed_text'])
  444. def test_code_language_optional_quotes(self):
  445. """code quotes around language name are optional"""
  446. test_text = """
  447. Lorem ipsum.
  448. [code=python"]
  449. Dolor [b]met.[/b]
  450. [/code]
  451. """.strip()
  452. expected_result = """
  453. <p>Lorem ipsum.</p>
  454. <pre><code class="python">Dolor [b]met.[/b]</code></pre>
  455. """.strip()
  456. result = parse(test_text, MockRequest(), MockPoster(), minify=False)
  457. self.assertEqual(expected_result, result['parsed_text'])
  458. test_text = """
  459. Lorem ipsum.
  460. [code="python]
  461. Dolor [b]met.[/b]
  462. [/code]
  463. """.strip()
  464. expected_result = """
  465. <p>Lorem ipsum.</p>
  466. <pre><code class="python">Dolor [b]met.[/b]</code></pre>
  467. """.strip()
  468. result = parse(test_text, MockRequest(), MockPoster(), minify=False)
  469. self.assertEqual(expected_result, result['parsed_text'])