nokogiri 1.8.5 → 1.13.9

Sign up to get free protection for your applications and to get access to all the features.

Potentially problematic release.

This version of nokogiri might be problematic. Click here for more details.

Files changed (353) hide show
  1. checksums.yaml +4 -4
  2. data/Gemfile +3 -21
  3. data/LICENSE-DEPENDENCIES.md +1159 -868
  4. data/LICENSE.md +5 -28
  5. data/README.md +196 -90
  6. data/bin/nokogiri +63 -50
  7. data/dependencies.yml +13 -59
  8. data/ext/nokogiri/depend +38 -358
  9. data/ext/nokogiri/extconf.rb +765 -420
  10. data/ext/nokogiri/gumbo.c +584 -0
  11. data/ext/nokogiri/html4_document.c +166 -0
  12. data/ext/nokogiri/html4_element_description.c +294 -0
  13. data/ext/nokogiri/html4_entity_lookup.c +37 -0
  14. data/ext/nokogiri/html4_sax_parser_context.c +119 -0
  15. data/ext/nokogiri/html4_sax_push_parser.c +95 -0
  16. data/ext/nokogiri/libxml2_backwards_compat.c +121 -0
  17. data/ext/nokogiri/nokogiri.c +228 -91
  18. data/ext/nokogiri/nokogiri.h +199 -88
  19. data/ext/nokogiri/test_global_handlers.c +40 -0
  20. data/ext/nokogiri/xml_attr.c +42 -37
  21. data/ext/nokogiri/xml_attribute_decl.c +21 -21
  22. data/ext/nokogiri/xml_cdata.c +14 -19
  23. data/ext/nokogiri/xml_comment.c +19 -26
  24. data/ext/nokogiri/xml_document.c +296 -217
  25. data/ext/nokogiri/xml_document_fragment.c +12 -16
  26. data/ext/nokogiri/xml_dtd.c +64 -58
  27. data/ext/nokogiri/xml_element_content.c +31 -26
  28. data/ext/nokogiri/xml_element_decl.c +25 -25
  29. data/ext/nokogiri/xml_encoding_handler.c +43 -18
  30. data/ext/nokogiri/xml_entity_decl.c +37 -35
  31. data/ext/nokogiri/xml_entity_reference.c +16 -18
  32. data/ext/nokogiri/xml_namespace.c +99 -54
  33. data/ext/nokogiri/xml_node.c +1107 -658
  34. data/ext/nokogiri/xml_node_set.c +178 -166
  35. data/ext/nokogiri/xml_processing_instruction.c +17 -19
  36. data/ext/nokogiri/xml_reader.c +277 -175
  37. data/ext/nokogiri/xml_relax_ng.c +52 -28
  38. data/ext/nokogiri/xml_sax_parser.c +112 -112
  39. data/ext/nokogiri/xml_sax_parser_context.c +112 -86
  40. data/ext/nokogiri/xml_sax_push_parser.c +36 -27
  41. data/ext/nokogiri/xml_schema.c +114 -35
  42. data/ext/nokogiri/xml_syntax_error.c +42 -21
  43. data/ext/nokogiri/xml_text.c +14 -18
  44. data/ext/nokogiri/xml_xpath_context.c +226 -115
  45. data/ext/nokogiri/xslt_stylesheet.c +265 -173
  46. data/gumbo-parser/CHANGES.md +63 -0
  47. data/gumbo-parser/Makefile +101 -0
  48. data/gumbo-parser/THANKS +27 -0
  49. data/gumbo-parser/src/Makefile +34 -0
  50. data/gumbo-parser/src/README.md +41 -0
  51. data/gumbo-parser/src/ascii.c +75 -0
  52. data/gumbo-parser/src/ascii.h +115 -0
  53. data/gumbo-parser/src/attribute.c +42 -0
  54. data/gumbo-parser/src/attribute.h +17 -0
  55. data/gumbo-parser/src/char_ref.c +22225 -0
  56. data/gumbo-parser/src/char_ref.h +29 -0
  57. data/gumbo-parser/src/char_ref.rl +2154 -0
  58. data/gumbo-parser/src/error.c +626 -0
  59. data/gumbo-parser/src/error.h +148 -0
  60. data/gumbo-parser/src/foreign_attrs.c +104 -0
  61. data/gumbo-parser/src/foreign_attrs.gperf +27 -0
  62. data/gumbo-parser/src/gumbo.h +943 -0
  63. data/gumbo-parser/src/insertion_mode.h +33 -0
  64. data/gumbo-parser/src/macros.h +91 -0
  65. data/gumbo-parser/src/parser.c +4875 -0
  66. data/gumbo-parser/src/parser.h +41 -0
  67. data/gumbo-parser/src/replacement.h +33 -0
  68. data/gumbo-parser/src/string_buffer.c +103 -0
  69. data/gumbo-parser/src/string_buffer.h +68 -0
  70. data/gumbo-parser/src/string_piece.c +48 -0
  71. data/gumbo-parser/src/svg_attrs.c +174 -0
  72. data/gumbo-parser/src/svg_attrs.gperf +77 -0
  73. data/gumbo-parser/src/svg_tags.c +137 -0
  74. data/gumbo-parser/src/svg_tags.gperf +55 -0
  75. data/gumbo-parser/src/tag.c +222 -0
  76. data/gumbo-parser/src/tag_lookup.c +382 -0
  77. data/gumbo-parser/src/tag_lookup.gperf +169 -0
  78. data/gumbo-parser/src/tag_lookup.h +13 -0
  79. data/gumbo-parser/src/token_buffer.c +79 -0
  80. data/gumbo-parser/src/token_buffer.h +71 -0
  81. data/gumbo-parser/src/token_type.h +17 -0
  82. data/gumbo-parser/src/tokenizer.c +3463 -0
  83. data/gumbo-parser/src/tokenizer.h +112 -0
  84. data/gumbo-parser/src/tokenizer_states.h +339 -0
  85. data/gumbo-parser/src/utf8.c +245 -0
  86. data/gumbo-parser/src/utf8.h +164 -0
  87. data/gumbo-parser/src/util.c +68 -0
  88. data/gumbo-parser/src/util.h +30 -0
  89. data/gumbo-parser/src/vector.c +111 -0
  90. data/gumbo-parser/src/vector.h +45 -0
  91. data/lib/nokogiri/class_resolver.rb +67 -0
  92. data/lib/nokogiri/css/node.rb +10 -8
  93. data/lib/nokogiri/css/parser.rb +397 -377
  94. data/lib/nokogiri/css/parser.y +250 -245
  95. data/lib/nokogiri/css/parser_extras.rb +54 -49
  96. data/lib/nokogiri/css/syntax_error.rb +3 -1
  97. data/lib/nokogiri/css/tokenizer.rb +107 -104
  98. data/lib/nokogiri/css/tokenizer.rex +3 -2
  99. data/lib/nokogiri/css/xpath_visitor.rb +218 -91
  100. data/lib/nokogiri/css.rb +50 -17
  101. data/lib/nokogiri/decorators/slop.rb +9 -7
  102. data/lib/nokogiri/extension.rb +31 -0
  103. data/lib/nokogiri/gumbo.rb +15 -0
  104. data/lib/nokogiri/html.rb +38 -27
  105. data/lib/nokogiri/{html → html4}/builder.rb +4 -2
  106. data/lib/nokogiri/{html → html4}/document.rb +103 -105
  107. data/lib/nokogiri/html4/document_fragment.rb +54 -0
  108. data/lib/nokogiri/{html → html4}/element_description.rb +3 -1
  109. data/lib/nokogiri/html4/element_description_defaults.rb +578 -0
  110. data/lib/nokogiri/{html → html4}/entity_lookup.rb +4 -2
  111. data/lib/nokogiri/{html → html4}/sax/parser.rb +17 -16
  112. data/lib/nokogiri/html4/sax/parser_context.rb +20 -0
  113. data/lib/nokogiri/{html → html4}/sax/push_parser.rb +12 -11
  114. data/lib/nokogiri/html4.rb +46 -0
  115. data/lib/nokogiri/html5/document.rb +91 -0
  116. data/lib/nokogiri/html5/document_fragment.rb +83 -0
  117. data/lib/nokogiri/html5/node.rb +100 -0
  118. data/lib/nokogiri/html5.rb +478 -0
  119. data/lib/nokogiri/jruby/dependencies.rb +21 -0
  120. data/lib/nokogiri/syntax_error.rb +2 -0
  121. data/lib/nokogiri/version/constant.rb +6 -0
  122. data/lib/nokogiri/version/info.rb +222 -0
  123. data/lib/nokogiri/version.rb +3 -108
  124. data/lib/nokogiri/xml/attr.rb +6 -3
  125. data/lib/nokogiri/xml/attribute_decl.rb +3 -1
  126. data/lib/nokogiri/xml/builder.rb +97 -53
  127. data/lib/nokogiri/xml/cdata.rb +3 -1
  128. data/lib/nokogiri/xml/character_data.rb +2 -0
  129. data/lib/nokogiri/xml/document.rb +224 -86
  130. data/lib/nokogiri/xml/document_fragment.rb +57 -44
  131. data/lib/nokogiri/xml/dtd.rb +4 -2
  132. data/lib/nokogiri/xml/element_content.rb +2 -0
  133. data/lib/nokogiri/xml/element_decl.rb +3 -1
  134. data/lib/nokogiri/xml/entity_decl.rb +4 -2
  135. data/lib/nokogiri/xml/entity_reference.rb +2 -0
  136. data/lib/nokogiri/xml/namespace.rb +3 -0
  137. data/lib/nokogiri/xml/node/save_options.rb +10 -5
  138. data/lib/nokogiri/xml/node.rb +895 -377
  139. data/lib/nokogiri/xml/node_set.rb +92 -65
  140. data/lib/nokogiri/xml/notation.rb +13 -0
  141. data/lib/nokogiri/xml/parse_options.rb +22 -8
  142. data/lib/nokogiri/xml/pp/character_data.rb +9 -6
  143. data/lib/nokogiri/xml/pp/node.rb +25 -26
  144. data/lib/nokogiri/xml/pp.rb +4 -2
  145. data/lib/nokogiri/xml/processing_instruction.rb +3 -1
  146. data/lib/nokogiri/xml/reader.rb +21 -28
  147. data/lib/nokogiri/xml/relax_ng.rb +8 -2
  148. data/lib/nokogiri/xml/sax/document.rb +45 -49
  149. data/lib/nokogiri/xml/sax/parser.rb +38 -34
  150. data/lib/nokogiri/xml/sax/parser_context.rb +8 -3
  151. data/lib/nokogiri/xml/sax/push_parser.rb +6 -5
  152. data/lib/nokogiri/xml/sax.rb +6 -4
  153. data/lib/nokogiri/xml/schema.rb +19 -9
  154. data/lib/nokogiri/xml/searchable.rb +112 -72
  155. data/lib/nokogiri/xml/syntax_error.rb +6 -4
  156. data/lib/nokogiri/xml/text.rb +2 -0
  157. data/lib/nokogiri/xml/xpath/syntax_error.rb +4 -2
  158. data/lib/nokogiri/xml/xpath.rb +15 -4
  159. data/lib/nokogiri/xml/xpath_context.rb +3 -3
  160. data/lib/nokogiri/xml.rb +38 -37
  161. data/lib/nokogiri/xslt/stylesheet.rb +3 -1
  162. data/lib/nokogiri/xslt.rb +29 -20
  163. data/lib/nokogiri.rb +49 -65
  164. data/lib/xsd/xmlparser/nokogiri.rb +26 -24
  165. data/patches/libxml2/0001-Remove-script-macro-support.patch +40 -0
  166. data/patches/libxml2/0002-Update-entities-to-remove-handling-of-ssi.patch +44 -0
  167. data/patches/libxml2/0003-libxml2.la-is-in-top_builddir.patch +25 -0
  168. data/patches/libxml2/0005-avoid-isnan-isinf.patch +81 -0
  169. data/patches/libxml2/0009-allow-wildcard-namespaces.patch +77 -0
  170. data/patches/libxslt/0001-update-automake-files-for-arm64.patch +3037 -0
  171. data/ports/archives/libxml2-2.10.3.tar.xz +0 -0
  172. data/ports/archives/libxslt-1.1.37.tar.xz +0 -0
  173. metadata +211 -266
  174. data/.autotest +0 -22
  175. data/.cross_rubies +0 -8
  176. data/.editorconfig +0 -17
  177. data/.gemtest +0 -0
  178. data/.travis.yml +0 -63
  179. data/CHANGELOG.md +0 -1368
  180. data/CONTRIBUTING.md +0 -42
  181. data/C_CODING_STYLE.rdoc +0 -33
  182. data/Gemfile-libxml-ruby +0 -3
  183. data/Manifest.txt +0 -370
  184. data/ROADMAP.md +0 -111
  185. data/Rakefile +0 -348
  186. data/SECURITY.md +0 -19
  187. data/STANDARD_RESPONSES.md +0 -47
  188. data/Y_U_NO_GEMSPEC.md +0 -155
  189. data/appveyor.yml +0 -29
  190. data/build_all +0 -44
  191. data/ext/nokogiri/html_document.c +0 -170
  192. data/ext/nokogiri/html_document.h +0 -10
  193. data/ext/nokogiri/html_element_description.c +0 -279
  194. data/ext/nokogiri/html_element_description.h +0 -10
  195. data/ext/nokogiri/html_entity_lookup.c +0 -32
  196. data/ext/nokogiri/html_entity_lookup.h +0 -8
  197. data/ext/nokogiri/html_sax_parser_context.c +0 -116
  198. data/ext/nokogiri/html_sax_parser_context.h +0 -11
  199. data/ext/nokogiri/html_sax_push_parser.c +0 -87
  200. data/ext/nokogiri/html_sax_push_parser.h +0 -9
  201. data/ext/nokogiri/xml_attr.h +0 -9
  202. data/ext/nokogiri/xml_attribute_decl.h +0 -9
  203. data/ext/nokogiri/xml_cdata.h +0 -9
  204. data/ext/nokogiri/xml_comment.h +0 -9
  205. data/ext/nokogiri/xml_document.h +0 -23
  206. data/ext/nokogiri/xml_document_fragment.h +0 -10
  207. data/ext/nokogiri/xml_dtd.h +0 -10
  208. data/ext/nokogiri/xml_element_content.h +0 -10
  209. data/ext/nokogiri/xml_element_decl.h +0 -9
  210. data/ext/nokogiri/xml_encoding_handler.h +0 -8
  211. data/ext/nokogiri/xml_entity_decl.h +0 -10
  212. data/ext/nokogiri/xml_entity_reference.h +0 -9
  213. data/ext/nokogiri/xml_io.c +0 -61
  214. data/ext/nokogiri/xml_io.h +0 -11
  215. data/ext/nokogiri/xml_libxml2_hacks.c +0 -112
  216. data/ext/nokogiri/xml_libxml2_hacks.h +0 -12
  217. data/ext/nokogiri/xml_namespace.h +0 -15
  218. data/ext/nokogiri/xml_node.h +0 -13
  219. data/ext/nokogiri/xml_node_set.h +0 -12
  220. data/ext/nokogiri/xml_processing_instruction.h +0 -9
  221. data/ext/nokogiri/xml_reader.h +0 -10
  222. data/ext/nokogiri/xml_relax_ng.h +0 -9
  223. data/ext/nokogiri/xml_sax_parser.h +0 -39
  224. data/ext/nokogiri/xml_sax_parser_context.h +0 -10
  225. data/ext/nokogiri/xml_sax_push_parser.h +0 -9
  226. data/ext/nokogiri/xml_schema.h +0 -9
  227. data/ext/nokogiri/xml_syntax_error.h +0 -13
  228. data/ext/nokogiri/xml_text.h +0 -9
  229. data/ext/nokogiri/xml_xpath_context.h +0 -10
  230. data/ext/nokogiri/xslt_stylesheet.h +0 -14
  231. data/lib/nokogiri/html/document_fragment.rb +0 -49
  232. data/lib/nokogiri/html/element_description_defaults.rb +0 -671
  233. data/lib/nokogiri/html/sax/parser_context.rb +0 -16
  234. data/patches/libxml2/0001-Revert-Do-not-URI-escape-in-server-side-includes.patch +0 -78
  235. data/patches/libxml2/0002-Fix-nullptr-deref-with-XPath-logic-ops.patch +0 -54
  236. data/patches/libxml2/0003-Fix-infinite-loop-in-LZMA-decompression.patch +0 -50
  237. data/patches/sort-patches-by-date +0 -25
  238. data/ports/archives/libxml2-2.9.8.tar.gz +0 -0
  239. data/ports/archives/libxslt-1.1.32.tar.gz +0 -0
  240. data/suppressions/README.txt +0 -1
  241. data/suppressions/nokogiri_ruby-2.supp +0 -10
  242. data/tasks/test.rb +0 -100
  243. data/test/css/test_nthiness.rb +0 -226
  244. data/test/css/test_parser.rb +0 -386
  245. data/test/css/test_tokenizer.rb +0 -215
  246. data/test/css/test_xpath_visitor.rb +0 -96
  247. data/test/decorators/test_slop.rb +0 -23
  248. data/test/files/2ch.html +0 -108
  249. data/test/files/GH_1042.html +0 -18
  250. data/test/files/address_book.rlx +0 -12
  251. data/test/files/address_book.xml +0 -10
  252. data/test/files/atom.xml +0 -344
  253. data/test/files/bar/bar.xsd +0 -4
  254. data/test/files/bogus.xml +0 -0
  255. data/test/files/dont_hurt_em_why.xml +0 -422
  256. data/test/files/encoding.html +0 -82
  257. data/test/files/encoding.xhtml +0 -84
  258. data/test/files/exslt.xml +0 -8
  259. data/test/files/exslt.xslt +0 -35
  260. data/test/files/foo/foo.xsd +0 -4
  261. data/test/files/metacharset.html +0 -10
  262. data/test/files/namespace_pressure_test.xml +0 -1684
  263. data/test/files/noencoding.html +0 -47
  264. data/test/files/po.xml +0 -32
  265. data/test/files/po.xsd +0 -66
  266. data/test/files/saml/saml20assertion_schema.xsd +0 -283
  267. data/test/files/saml/saml20protocol_schema.xsd +0 -302
  268. data/test/files/saml/xenc_schema.xsd +0 -146
  269. data/test/files/saml/xmldsig_schema.xsd +0 -318
  270. data/test/files/shift_jis.html +0 -10
  271. data/test/files/shift_jis.xml +0 -5
  272. data/test/files/shift_jis_no_charset.html +0 -9
  273. data/test/files/slow-xpath.xml +0 -25509
  274. data/test/files/snuggles.xml +0 -3
  275. data/test/files/staff.dtd +0 -10
  276. data/test/files/staff.xml +0 -59
  277. data/test/files/staff.xslt +0 -32
  278. data/test/files/test_document_url/bar.xml +0 -2
  279. data/test/files/test_document_url/document.dtd +0 -4
  280. data/test/files/test_document_url/document.xml +0 -6
  281. data/test/files/tlm.html +0 -851
  282. data/test/files/to_be_xincluded.xml +0 -2
  283. data/test/files/valid_bar.xml +0 -2
  284. data/test/files/xinclude.xml +0 -4
  285. data/test/helper.rb +0 -271
  286. data/test/html/sax/test_parser.rb +0 -168
  287. data/test/html/sax/test_parser_context.rb +0 -46
  288. data/test/html/sax/test_parser_text.rb +0 -163
  289. data/test/html/sax/test_push_parser.rb +0 -87
  290. data/test/html/test_attributes.rb +0 -85
  291. data/test/html/test_builder.rb +0 -164
  292. data/test/html/test_document.rb +0 -712
  293. data/test/html/test_document_encoding.rb +0 -143
  294. data/test/html/test_document_fragment.rb +0 -310
  295. data/test/html/test_element_description.rb +0 -105
  296. data/test/html/test_named_characters.rb +0 -14
  297. data/test/html/test_node.rb +0 -212
  298. data/test/html/test_node_encoding.rb +0 -91
  299. data/test/namespaces/test_additional_namespaces_in_builder_doc.rb +0 -14
  300. data/test/namespaces/test_namespaces_aliased_default.rb +0 -24
  301. data/test/namespaces/test_namespaces_in_builder_doc.rb +0 -75
  302. data/test/namespaces/test_namespaces_in_cloned_doc.rb +0 -31
  303. data/test/namespaces/test_namespaces_in_created_doc.rb +0 -75
  304. data/test/namespaces/test_namespaces_in_parsed_doc.rb +0 -80
  305. data/test/namespaces/test_namespaces_preservation.rb +0 -31
  306. data/test/test_convert_xpath.rb +0 -135
  307. data/test/test_css_cache.rb +0 -47
  308. data/test/test_encoding_handler.rb +0 -48
  309. data/test/test_memory_leak.rb +0 -156
  310. data/test/test_nokogiri.rb +0 -138
  311. data/test/test_soap4r_sax.rb +0 -52
  312. data/test/test_xslt_transforms.rb +0 -314
  313. data/test/xml/node/test_save_options.rb +0 -28
  314. data/test/xml/node/test_subclass.rb +0 -44
  315. data/test/xml/sax/test_parser.rb +0 -402
  316. data/test/xml/sax/test_parser_context.rb +0 -115
  317. data/test/xml/sax/test_parser_text.rb +0 -202
  318. data/test/xml/sax/test_push_parser.rb +0 -265
  319. data/test/xml/test_attr.rb +0 -74
  320. data/test/xml/test_attribute_decl.rb +0 -86
  321. data/test/xml/test_builder.rb +0 -341
  322. data/test/xml/test_c14n.rb +0 -180
  323. data/test/xml/test_cdata.rb +0 -54
  324. data/test/xml/test_comment.rb +0 -40
  325. data/test/xml/test_document.rb +0 -982
  326. data/test/xml/test_document_encoding.rb +0 -31
  327. data/test/xml/test_document_fragment.rb +0 -298
  328. data/test/xml/test_dtd.rb +0 -187
  329. data/test/xml/test_dtd_encoding.rb +0 -31
  330. data/test/xml/test_element_content.rb +0 -56
  331. data/test/xml/test_element_decl.rb +0 -73
  332. data/test/xml/test_entity_decl.rb +0 -122
  333. data/test/xml/test_entity_reference.rb +0 -262
  334. data/test/xml/test_namespace.rb +0 -96
  335. data/test/xml/test_node.rb +0 -1325
  336. data/test/xml/test_node_attributes.rb +0 -115
  337. data/test/xml/test_node_encoding.rb +0 -75
  338. data/test/xml/test_node_inheritance.rb +0 -32
  339. data/test/xml/test_node_reparenting.rb +0 -592
  340. data/test/xml/test_node_set.rb +0 -809
  341. data/test/xml/test_parse_options.rb +0 -64
  342. data/test/xml/test_processing_instruction.rb +0 -30
  343. data/test/xml/test_reader.rb +0 -620
  344. data/test/xml/test_reader_encoding.rb +0 -134
  345. data/test/xml/test_relax_ng.rb +0 -60
  346. data/test/xml/test_schema.rb +0 -142
  347. data/test/xml/test_syntax_error.rb +0 -36
  348. data/test/xml/test_text.rb +0 -60
  349. data/test/xml/test_unparented_node.rb +0 -483
  350. data/test/xml/test_xinclude.rb +0 -83
  351. data/test/xml/test_xpath.rb +0 -470
  352. data/test/xslt/test_custom_functions.rb +0 -133
  353. data/test/xslt/test_exception_handling.rb +0 -37
data/test/html/sax/test_push_parser.rb DELETED
@@ -1,87 +0,0 @@
1
- # -*- coding: utf-8 -*-
2
-
3
- require "helper"
4
-
5
- module Nokogiri
6
- module HTML
7
- module SAX
8
- class TestPushParser < Nokogiri::SAX::TestCase
9
- def setup
10
- super
11
- @parser = HTML::SAX::PushParser.new(Doc.new)
12
- end
13
-
14
- def test_end_document_called
15
- @parser.<<(<<-eoxml)
16
- <p id="asdfasdf">
17
- <!-- This is a comment -->
18
- Paragraph 1
19
- </p>
20
- eoxml
21
- assert ! @parser.document.end_document_called
22
- @parser.finish
23
- assert @parser.document.end_document_called
24
- end
25
-
26
- def test_start_element
27
- @parser.<<(<<-eoxml)
28
- <!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN" "http://www.w3.org/TR/REC-html40/loose.dtd">
29
- <html><head><body><p id="asdfasdf">
30
- eoxml
31
-
32
- assert_equal [["html", []], ["head", []], ["body", []], ["p", [["id", "asdfasdf"]]]],
33
- @parser.document.start_elements
34
-
35
- @parser.<<(<<-eoxml)
36
- <!-- This is a comment -->
37
- Paragraph 1
38
- </p></body></html>
39
- eoxml
40
- assert_equal [' This is a comment '], @parser.document.comments
41
- @parser.finish
42
- end
43
-
44
-
45
- def test_chevron_partial_html
46
- @parser.<<(<<-eoxml)
47
- <p id="asdfasdf">
48
- eoxml
49
-
50
- @parser.<<(<<-eoxml)
51
- <!-- This is a comment -->
52
- Paragraph 1
53
- </p>
54
- eoxml
55
- assert_equal [' This is a comment '], @parser.document.comments
56
- @parser.finish
57
- end
58
-
59
- def test_chevron
60
- @parser.<<(<<-eoxml)
61
- <p id="asdfasdf">
62
- <!-- This is a comment -->
63
- Paragraph 1
64
- </p>
65
- eoxml
66
- @parser.finish
67
- assert_equal [' This is a comment '], @parser.document.comments
68
- end
69
-
70
- def test_default_options
71
- assert_equal 0, @parser.options
72
- end
73
-
74
- def test_broken_encoding
75
- skip("ultra hard to fix for pure Java version") if Nokogiri.jruby?
76
- @parser.options |= XML::ParseOptions::RECOVER
77
- # This is ISO_8859-1:
78
- @parser.<< "<?xml version='1.0' encoding='UTF-8'?><r>Gau\337</r>"
79
- @parser.finish
80
- assert(@parser.document.errors.size >= 1)
81
- assert_equal "Gau\337", @parser.document.data.join
82
- assert_equal [["r"], ["body"], ["html"]], @parser.document.end_elements
83
- end
84
- end
85
- end
86
- end
87
- end
data/test/html/test_attributes.rb DELETED
@@ -1,85 +0,0 @@
1
- require "helper"
2
-
3
- module Nokogiri
4
- module HTML
5
- class TestAttr < Nokogiri::TestCase
6
- unless Nokogiri::VersionInfo.instance.libxml2? && Nokogiri::VersionInfo.instance.libxml2_using_system?
7
- #
8
- # libxml2 >= 2.9.2 fails to escape comments within some attributes. It
9
- # wants to ensure these comments can be treated as "server-side includes",
10
- # but as a result fails to ensure that serialization is well-formed,
11
- # resulting in an opportunity for XSS injection of code into a final
12
- # re-parsed document (presumably in a browser).
13
- #
14
- # the offending commit is:
15
- #
16
- # https://github.com/GNOME/libxml2/commit/960f0e2
17
- #
18
- # we'll test this by parsing the HTML, serializing it, then
19
- # re-parsing it to ensure there isn't any ambiguity in the output
20
- # that might allow code injection into a browser consuming
21
- # "sanitized" output.
22
- #
23
- # complaints have been made upstream about this behavior, notably at
24
- #
25
- # https://bugzilla.gnome.org/show_bug.cgi?id=769760
26
- #
27
- # and multiple CVEs have been declared and fixed in downstream
28
- # libraries as a result, a list is being kept up to date here:
29
- #
30
- # https://github.com/flavorjones/loofah/issues/144
31
- #
32
- [
33
- #
34
- # these tags and attributes are determined by the code at:
35
- #
36
- # https://git.gnome.org/browse/libxml2/tree/HTMLtree.c?h=v2.9.2#n714
37
- #
38
- {tag: "a", attr: "href"},
39
- {tag: "div", attr: "href"},
40
- {tag: "a", attr: "action"},
41
- {tag: "div", attr: "action"},
42
- {tag: "a", attr: "src"},
43
- {tag: "div", attr: "src"},
44
- {tag: "a", attr: "name"},
45
- #
46
- # note that div+name is _not_ affected by the libxml2 issue.
47
- # but we test it anyway to ensure our logic isn't modifying
48
- # attributes that don't need modifying.
49
- #
50
- {tag: "div", attr: "name", unescaped: true},
51
- ].each do |config|
52
-
53
- define_method "test_uri_escaping_of_#{config[:attr]}_attr_in_#{config[:tag]}_tag" do
54
- html = %{<#{config[:tag]} #{config[:attr]}='examp<!--" unsafeattr=unsafevalue()>-->le.com'>test</#{config[:tag]}>}
55
-
56
- reparsed = HTML.fragment(HTML.fragment(html).to_html)
57
- attributes = reparsed.at_css(config[:tag]).attribute_nodes
58
-
59
- assert_equal [config[:attr]], attributes.collect(&:name)
60
- if Nokogiri::VersionInfo.instance.libxml2?
61
- if config[:unescaped]
62
- #
63
- # this attribute was emitted wrapped in single-quotes, so a double quote is A-OK.
64
- # assert that this attribute's serialization is unaffected.
65
- #
66
- assert_equal %{examp<!--" unsafeattr=unsafevalue()>-->le.com}, attributes.first.value
67
- else
68
- #
69
- # let's match the behavior in libxml < 2.9.2.
70
- # test that this attribute's serialization is well-formed and sanitized.
71
- #
72
- assert_equal %{examp<!--%22%20unsafeattr=unsafevalue()>-->le.com}, attributes.first.value
73
- end
74
- else
75
- #
76
- # yay for consistency in javaland. move along, nothing to see here.
77
- #
78
- assert_equal %{examp<!--%22 unsafeattr=unsafevalue()>-->le.com}, attributes.first.value
79
- end
80
- end
81
- end
82
- end
83
- end
84
- end
85
- end
data/test/html/test_builder.rb DELETED
@@ -1,164 +0,0 @@
1
- require "helper"
2
-
3
- module Nokogiri
4
- module HTML
5
- class TestBuilder < Nokogiri::TestCase
6
- def test_top_level_function_builds
7
- foo = nil
8
- Nokogiri() { |xml| foo = xml }
9
- assert_instance_of Nokogiri::HTML::Builder, foo
10
- end
11
-
12
- def test_builder_with_explicit_tags
13
- html_doc = Nokogiri::HTML::Builder.new {
14
- div.slide(:class => 'another_class') {
15
- node = Nokogiri::XML::Node.new("id", doc)
16
- node.content = "hello"
17
- insert(node)
18
- }
19
- }.doc
20
- assert_equal 1, html_doc.css('div.slide > id').length
21
- assert_equal 'hello', html_doc.at('div.slide > id').content
22
- end
23
-
24
- def test_hash_as_attributes_for_attribute_method
25
- html = Nokogiri::HTML::Builder.new { ||
26
- div.slide(:class => 'another_class') {
27
- span 'Slide 1'
28
- }
29
- }.to_html
30
- assert_match 'class="slide another_class"', html
31
- end
32
-
33
- def test_hash_as_attributes
34
- builder = Nokogiri::HTML::Builder.new do
35
- div(:id => 'awesome') {
36
- h1 "america"
37
- }
38
- end
39
- assert_equal('<div id="awesome"><h1>america</h1></div>',
40
- builder.doc.root.to_html.gsub(/\n/, '').gsub(/>\s*</, '><'))
41
- end
42
-
43
- def test_href_with_attributes
44
- uri = 'http://tenderlovemaking.com/'
45
- built = Nokogiri::XML::Builder.new {
46
- div {
47
- a('King Khan & The Shrines', :href => uri)
48
- }
49
- }
50
- assert_equal 'http://tenderlovemaking.com/',
51
- built.doc.at('a')[:href]
52
- end
53
-
54
- def test_tag_nesting
55
- builder = Nokogiri::HTML::Builder.new do
56
- body {
57
- span.left ''
58
- span.middle {
59
- div.icon ''
60
- }
61
- span.right ''
62
- }
63
- end
64
- assert node = builder.doc.css('span.right').first
65
- assert_equal 'middle', node.previous_sibling['class']
66
- end
67
-
68
- def test_has_ampersand
69
- builder = Nokogiri::HTML::Builder.new do
70
- div.rad.thing! {
71
- text "<awe&some>"
72
- b "hello & world"
73
- }
74
- end
75
- assert_equal(
76
- '<div class="rad" id="thing">&lt;awe&amp;some&gt;<b>hello &amp; world</b></div>',
77
- builder.doc.root.to_html.gsub(/\n/, ''))
78
- end
79
-
80
- def test_multi_tags
81
- builder = Nokogiri::HTML::Builder.new do
82
- div.rad.thing! {
83
- text "<awesome>"
84
- b "hello"
85
- }
86
- end
87
- assert_equal(
88
- '<div class="rad" id="thing">&lt;awesome&gt;<b>hello</b></div>',
89
- builder.doc.root.to_html.gsub(/\n/, ''))
90
- end
91
-
92
- def test_attributes_plus_block
93
- builder = Nokogiri::HTML::Builder.new do
94
- div.rad.thing! {
95
- text "<awesome>"
96
- }
97
- end
98
- assert_equal('<div class="rad" id="thing">&lt;awesome&gt;</div>',
99
- builder.doc.root.to_html.chomp)
100
- end
101
-
102
- def test_builder_adds_attributes
103
- builder = Nokogiri::HTML::Builder.new do
104
- div.rad.thing! "tender div"
105
- end
106
- assert_equal('<div class="rad" id="thing">tender div</div>',
107
- builder.doc.root.to_html.chomp)
108
- end
109
-
110
- def test_bold_tag
111
- builder = Nokogiri::HTML::Builder.new do
112
- b "bold tag"
113
- end
114
- assert_equal('<b>bold tag</b>', builder.doc.root.to_html.chomp)
115
- end
116
-
117
- def test_html_then_body_tag
118
- builder = Nokogiri::HTML::Builder.new do
119
- html {
120
- body {
121
- b "bold tag"
122
- }
123
- }
124
- end
125
- assert_equal('<html><body><b>bold tag</b></body></html>',
126
- builder.doc.root.to_html.chomp.gsub(/>\s*</, '><'))
127
- end
128
-
129
- def test_instance_eval_with_delegation_to_block_context
130
- class << self
131
- def foo
132
- "foo!"
133
- end
134
- end
135
-
136
- builder = Nokogiri::HTML::Builder.new { text foo }
137
- assert builder.to_html.include?("foo!")
138
- end
139
-
140
- def test_builder_with_param
141
- doc = Nokogiri::HTML::Builder.new { |html|
142
- html.body {
143
- html.p "hello world"
144
- }
145
- }.doc
146
-
147
- assert node = doc.xpath('//body/p').first
148
- assert_equal 'hello world', node.content
149
- end
150
-
151
- def test_builder_with_id
152
- text = "hello world"
153
- doc = Nokogiri::HTML::Builder.new { |html|
154
- html.body {
155
- html.id_ text
156
- }
157
- }.doc
158
-
159
- assert node = doc.xpath('//body/id').first
160
- assert_equal text, node.content
161
- end
162
- end
163
- end
164
- end