nokogiri 1.8.5 → 1.15.3

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 (358) hide show
  1. checksums.yaml +4 -4
  2. data/Gemfile +40 -18
  3. data/LICENSE-DEPENDENCIES.md +1636 -1024
  4. data/LICENSE.md +5 -28
  5. data/README.md +203 -90
  6. data/bin/nokogiri +63 -50
  7. data/dependencies.yml +33 -61
  8. data/ext/nokogiri/depend +38 -358
  9. data/ext/nokogiri/extconf.rb +867 -417
  10. data/ext/nokogiri/gumbo.c +594 -0
  11. data/ext/nokogiri/html4_document.c +165 -0
  12. data/ext/nokogiri/html4_element_description.c +299 -0
  13. data/ext/nokogiri/html4_entity_lookup.c +37 -0
  14. data/ext/nokogiri/html4_sax_parser_context.c +108 -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 +251 -105
  18. data/ext/nokogiri/nokogiri.h +215 -90
  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 +22 -22
  22. data/ext/nokogiri/xml_cdata.c +40 -31
  23. data/ext/nokogiri/xml_comment.c +20 -27
  24. data/ext/nokogiri/xml_document.c +401 -237
  25. data/ext/nokogiri/xml_document_fragment.c +13 -17
  26. data/ext/nokogiri/xml_dtd.c +64 -58
  27. data/ext/nokogiri/xml_element_content.c +63 -55
  28. data/ext/nokogiri/xml_element_decl.c +31 -31
  29. data/ext/nokogiri/xml_encoding_handler.c +54 -21
  30. data/ext/nokogiri/xml_entity_decl.c +37 -35
  31. data/ext/nokogiri/xml_entity_reference.c +17 -19
  32. data/ext/nokogiri/xml_namespace.c +136 -62
  33. data/ext/nokogiri/xml_node.c +1387 -678
  34. data/ext/nokogiri/xml_node_set.c +246 -216
  35. data/ext/nokogiri/xml_processing_instruction.c +18 -20
  36. data/ext/nokogiri/xml_reader.c +347 -212
  37. data/ext/nokogiri/xml_relax_ng.c +86 -77
  38. data/ext/nokogiri/xml_sax_parser.c +149 -124
  39. data/ext/nokogiri/xml_sax_parser_context.c +145 -103
  40. data/ext/nokogiri/xml_sax_push_parser.c +64 -36
  41. data/ext/nokogiri/xml_schema.c +138 -81
  42. data/ext/nokogiri/xml_syntax_error.c +42 -21
  43. data/ext/nokogiri/xml_text.c +36 -26
  44. data/ext/nokogiri/xml_xpath_context.c +366 -178
  45. data/ext/nokogiri/xslt_stylesheet.c +335 -189
  46. data/gumbo-parser/CHANGES.md +63 -0
  47. data/gumbo-parser/Makefile +111 -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 +630 -0
  59. data/gumbo-parser/src/error.h +148 -0
  60. data/gumbo-parser/src/foreign_attrs.c +103 -0
  61. data/gumbo-parser/src/foreign_attrs.gperf +27 -0
  62. data/gumbo-parser/src/insertion_mode.h +33 -0
  63. data/gumbo-parser/src/macros.h +91 -0
  64. data/gumbo-parser/src/nokogiri_gumbo.h +944 -0
  65. data/gumbo-parser/src/parser.c +4891 -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 +223 -0
  76. data/gumbo-parser/src/tag_lookup.c +382 -0
  77. data/gumbo-parser/src/tag_lookup.gperf +170 -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 +66 -0
  88. data/gumbo-parser/src/util.h +34 -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 +224 -95
  100. data/lib/nokogiri/css.rb +56 -17
  101. data/lib/nokogiri/decorators/slop.rb +9 -7
  102. data/lib/nokogiri/encoding_handler.rb +57 -0
  103. data/lib/nokogiri/extension.rb +32 -0
  104. data/lib/nokogiri/gumbo.rb +15 -0
  105. data/lib/nokogiri/html.rb +38 -27
  106. data/lib/nokogiri/{html → html4}/builder.rb +4 -2
  107. data/lib/nokogiri/html4/document.rb +214 -0
  108. data/lib/nokogiri/html4/document_fragment.rb +54 -0
  109. data/lib/nokogiri/{html → html4}/element_description.rb +3 -1
  110. data/lib/nokogiri/html4/element_description_defaults.rb +2040 -0
  111. data/lib/nokogiri/html4/encoding_reader.rb +121 -0
  112. data/lib/nokogiri/{html → html4}/entity_lookup.rb +4 -2
  113. data/lib/nokogiri/{html → html4}/sax/parser.rb +17 -16
  114. data/lib/nokogiri/html4/sax/parser_context.rb +20 -0
  115. data/lib/nokogiri/{html → html4}/sax/push_parser.rb +12 -11
  116. data/lib/nokogiri/html4.rb +47 -0
  117. data/lib/nokogiri/html5/document.rb +168 -0
  118. data/lib/nokogiri/html5/document_fragment.rb +90 -0
  119. data/lib/nokogiri/html5/node.rb +103 -0
  120. data/lib/nokogiri/html5.rb +392 -0
  121. data/lib/nokogiri/jruby/dependencies.rb +3 -0
  122. data/lib/nokogiri/jruby/nokogiri_jars.rb +43 -0
  123. data/lib/nokogiri/syntax_error.rb +2 -0
  124. data/lib/nokogiri/version/constant.rb +6 -0
  125. data/lib/nokogiri/version/info.rb +223 -0
  126. data/lib/nokogiri/version.rb +3 -108
  127. data/lib/nokogiri/xml/attr.rb +55 -3
  128. data/lib/nokogiri/xml/attribute_decl.rb +6 -2
  129. data/lib/nokogiri/xml/builder.rb +98 -54
  130. data/lib/nokogiri/xml/cdata.rb +3 -1
  131. data/lib/nokogiri/xml/character_data.rb +2 -0
  132. data/lib/nokogiri/xml/document.rb +312 -126
  133. data/lib/nokogiri/xml/document_fragment.rb +104 -48
  134. data/lib/nokogiri/xml/dtd.rb +4 -2
  135. data/lib/nokogiri/xml/element_content.rb +12 -2
  136. data/lib/nokogiri/xml/element_decl.rb +6 -2
  137. data/lib/nokogiri/xml/entity_decl.rb +7 -3
  138. data/lib/nokogiri/xml/entity_reference.rb +2 -0
  139. data/lib/nokogiri/xml/namespace.rb +45 -0
  140. data/lib/nokogiri/xml/node/save_options.rb +23 -8
  141. data/lib/nokogiri/xml/node.rb +1093 -411
  142. data/lib/nokogiri/xml/node_set.rb +173 -67
  143. data/lib/nokogiri/xml/notation.rb +13 -0
  144. data/lib/nokogiri/xml/parse_options.rb +145 -52
  145. data/lib/nokogiri/xml/pp/character_data.rb +9 -6
  146. data/lib/nokogiri/xml/pp/node.rb +42 -30
  147. data/lib/nokogiri/xml/pp.rb +4 -2
  148. data/lib/nokogiri/xml/processing_instruction.rb +4 -1
  149. data/lib/nokogiri/xml/reader.rb +21 -28
  150. data/lib/nokogiri/xml/relax_ng.rb +8 -2
  151. data/lib/nokogiri/xml/sax/document.rb +45 -49
  152. data/lib/nokogiri/xml/sax/parser.rb +39 -36
  153. data/lib/nokogiri/xml/sax/parser_context.rb +8 -3
  154. data/lib/nokogiri/xml/sax/push_parser.rb +6 -5
  155. data/lib/nokogiri/xml/sax.rb +6 -4
  156. data/lib/nokogiri/xml/schema.rb +19 -9
  157. data/lib/nokogiri/xml/searchable.rb +120 -72
  158. data/lib/nokogiri/xml/syntax_error.rb +6 -4
  159. data/lib/nokogiri/xml/text.rb +2 -0
  160. data/lib/nokogiri/xml/xpath/syntax_error.rb +4 -2
  161. data/lib/nokogiri/xml/xpath.rb +15 -4
  162. data/lib/nokogiri/xml/xpath_context.rb +3 -3
  163. data/lib/nokogiri/xml.rb +38 -37
  164. data/lib/nokogiri/xslt/stylesheet.rb +3 -1
  165. data/lib/nokogiri/xslt.rb +101 -22
  166. data/lib/nokogiri.rb +59 -75
  167. data/lib/xsd/xmlparser/nokogiri.rb +29 -25
  168. data/patches/libxml2/0001-Remove-script-macro-support.patch +40 -0
  169. data/patches/libxml2/0002-Update-entities-to-remove-handling-of-ssi.patch +44 -0
  170. data/patches/libxml2/0003-libxml2.la-is-in-top_builddir.patch +25 -0
  171. data/patches/libxml2/0009-allow-wildcard-namespaces.patch +77 -0
  172. data/patches/libxml2/0010-update-config.guess-and-config.sub-for-libxml2.patch +224 -0
  173. data/patches/libxml2/0011-rip-out-libxml2-s-libc_single_threaded-support.patch +30 -0
  174. data/patches/libxslt/0001-update-config.guess-and-config.sub-for-libxslt.patch +224 -0
  175. data/ports/archives/libxml2-2.11.4.tar.xz +0 -0
  176. data/ports/archives/libxslt-1.1.38.tar.xz +0 -0
  177. metadata +126 -399
  178. data/.autotest +0 -22
  179. data/.cross_rubies +0 -8
  180. data/.editorconfig +0 -17
  181. data/.gemtest +0 -0
  182. data/.travis.yml +0 -63
  183. data/CHANGELOG.md +0 -1368
  184. data/CONTRIBUTING.md +0 -42
  185. data/C_CODING_STYLE.rdoc +0 -33
  186. data/Gemfile-libxml-ruby +0 -3
  187. data/Manifest.txt +0 -370
  188. data/ROADMAP.md +0 -111
  189. data/Rakefile +0 -348
  190. data/SECURITY.md +0 -19
  191. data/STANDARD_RESPONSES.md +0 -47
  192. data/Y_U_NO_GEMSPEC.md +0 -155
  193. data/appveyor.yml +0 -29
  194. data/build_all +0 -44
  195. data/ext/nokogiri/html_document.c +0 -170
  196. data/ext/nokogiri/html_document.h +0 -10
  197. data/ext/nokogiri/html_element_description.c +0 -279
  198. data/ext/nokogiri/html_element_description.h +0 -10
  199. data/ext/nokogiri/html_entity_lookup.c +0 -32
  200. data/ext/nokogiri/html_entity_lookup.h +0 -8
  201. data/ext/nokogiri/html_sax_parser_context.c +0 -116
  202. data/ext/nokogiri/html_sax_parser_context.h +0 -11
  203. data/ext/nokogiri/html_sax_push_parser.c +0 -87
  204. data/ext/nokogiri/html_sax_push_parser.h +0 -9
  205. data/ext/nokogiri/xml_attr.h +0 -9
  206. data/ext/nokogiri/xml_attribute_decl.h +0 -9
  207. data/ext/nokogiri/xml_cdata.h +0 -9
  208. data/ext/nokogiri/xml_comment.h +0 -9
  209. data/ext/nokogiri/xml_document.h +0 -23
  210. data/ext/nokogiri/xml_document_fragment.h +0 -10
  211. data/ext/nokogiri/xml_dtd.h +0 -10
  212. data/ext/nokogiri/xml_element_content.h +0 -10
  213. data/ext/nokogiri/xml_element_decl.h +0 -9
  214. data/ext/nokogiri/xml_encoding_handler.h +0 -8
  215. data/ext/nokogiri/xml_entity_decl.h +0 -10
  216. data/ext/nokogiri/xml_entity_reference.h +0 -9
  217. data/ext/nokogiri/xml_io.c +0 -61
  218. data/ext/nokogiri/xml_io.h +0 -11
  219. data/ext/nokogiri/xml_libxml2_hacks.c +0 -112
  220. data/ext/nokogiri/xml_libxml2_hacks.h +0 -12
  221. data/ext/nokogiri/xml_namespace.h +0 -15
  222. data/ext/nokogiri/xml_node.h +0 -13
  223. data/ext/nokogiri/xml_node_set.h +0 -12
  224. data/ext/nokogiri/xml_processing_instruction.h +0 -9
  225. data/ext/nokogiri/xml_reader.h +0 -10
  226. data/ext/nokogiri/xml_relax_ng.h +0 -9
  227. data/ext/nokogiri/xml_sax_parser.h +0 -39
  228. data/ext/nokogiri/xml_sax_parser_context.h +0 -10
  229. data/ext/nokogiri/xml_sax_push_parser.h +0 -9
  230. data/ext/nokogiri/xml_schema.h +0 -9
  231. data/ext/nokogiri/xml_syntax_error.h +0 -13
  232. data/ext/nokogiri/xml_text.h +0 -9
  233. data/ext/nokogiri/xml_xpath_context.h +0 -10
  234. data/ext/nokogiri/xslt_stylesheet.h +0 -14
  235. data/lib/nokogiri/html/document.rb +0 -335
  236. data/lib/nokogiri/html/document_fragment.rb +0 -49
  237. data/lib/nokogiri/html/element_description_defaults.rb +0 -671
  238. data/lib/nokogiri/html/sax/parser_context.rb +0 -16
  239. data/patches/libxml2/0001-Revert-Do-not-URI-escape-in-server-side-includes.patch +0 -78
  240. data/patches/libxml2/0002-Fix-nullptr-deref-with-XPath-logic-ops.patch +0 -54
  241. data/patches/libxml2/0003-Fix-infinite-loop-in-LZMA-decompression.patch +0 -50
  242. data/patches/sort-patches-by-date +0 -25
  243. data/ports/archives/libxml2-2.9.8.tar.gz +0 -0
  244. data/ports/archives/libxslt-1.1.32.tar.gz +0 -0
  245. data/suppressions/README.txt +0 -1
  246. data/suppressions/nokogiri_ruby-2.supp +0 -10
  247. data/tasks/test.rb +0 -100
  248. data/test/css/test_nthiness.rb +0 -226
  249. data/test/css/test_parser.rb +0 -386
  250. data/test/css/test_tokenizer.rb +0 -215
  251. data/test/css/test_xpath_visitor.rb +0 -96
  252. data/test/decorators/test_slop.rb +0 -23
  253. data/test/files/2ch.html +0 -108
  254. data/test/files/GH_1042.html +0 -18
  255. data/test/files/address_book.rlx +0 -12
  256. data/test/files/address_book.xml +0 -10
  257. data/test/files/atom.xml +0 -344
  258. data/test/files/bar/bar.xsd +0 -4
  259. data/test/files/bogus.xml +0 -0
  260. data/test/files/dont_hurt_em_why.xml +0 -422
  261. data/test/files/encoding.html +0 -82
  262. data/test/files/encoding.xhtml +0 -84
  263. data/test/files/exslt.xml +0 -8
  264. data/test/files/exslt.xslt +0 -35
  265. data/test/files/foo/foo.xsd +0 -4
  266. data/test/files/metacharset.html +0 -10
  267. data/test/files/namespace_pressure_test.xml +0 -1684
  268. data/test/files/noencoding.html +0 -47
  269. data/test/files/po.xml +0 -32
  270. data/test/files/po.xsd +0 -66
  271. data/test/files/saml/saml20assertion_schema.xsd +0 -283
  272. data/test/files/saml/saml20protocol_schema.xsd +0 -302
  273. data/test/files/saml/xenc_schema.xsd +0 -146
  274. data/test/files/saml/xmldsig_schema.xsd +0 -318
  275. data/test/files/shift_jis.html +0 -10
  276. data/test/files/shift_jis.xml +0 -5
  277. data/test/files/shift_jis_no_charset.html +0 -9
  278. data/test/files/slow-xpath.xml +0 -25509
  279. data/test/files/snuggles.xml +0 -3
  280. data/test/files/staff.dtd +0 -10
  281. data/test/files/staff.xml +0 -59
  282. data/test/files/staff.xslt +0 -32
  283. data/test/files/test_document_url/bar.xml +0 -2
  284. data/test/files/test_document_url/document.dtd +0 -4
  285. data/test/files/test_document_url/document.xml +0 -6
  286. data/test/files/tlm.html +0 -851
  287. data/test/files/to_be_xincluded.xml +0 -2
  288. data/test/files/valid_bar.xml +0 -2
  289. data/test/files/xinclude.xml +0 -4
  290. data/test/helper.rb +0 -271
  291. data/test/html/sax/test_parser.rb +0 -168
  292. data/test/html/sax/test_parser_context.rb +0 -46
  293. data/test/html/sax/test_parser_text.rb +0 -163
  294. data/test/html/sax/test_push_parser.rb +0 -87
  295. data/test/html/test_attributes.rb +0 -85
  296. data/test/html/test_builder.rb +0 -164
  297. data/test/html/test_document.rb +0 -712
  298. data/test/html/test_document_encoding.rb +0 -143
  299. data/test/html/test_document_fragment.rb +0 -310
  300. data/test/html/test_element_description.rb +0 -105
  301. data/test/html/test_named_characters.rb +0 -14
  302. data/test/html/test_node.rb +0 -212
  303. data/test/html/test_node_encoding.rb +0 -91
  304. data/test/namespaces/test_additional_namespaces_in_builder_doc.rb +0 -14
  305. data/test/namespaces/test_namespaces_aliased_default.rb +0 -24
  306. data/test/namespaces/test_namespaces_in_builder_doc.rb +0 -75
  307. data/test/namespaces/test_namespaces_in_cloned_doc.rb +0 -31
  308. data/test/namespaces/test_namespaces_in_created_doc.rb +0 -75
  309. data/test/namespaces/test_namespaces_in_parsed_doc.rb +0 -80
  310. data/test/namespaces/test_namespaces_preservation.rb +0 -31
  311. data/test/test_convert_xpath.rb +0 -135
  312. data/test/test_css_cache.rb +0 -47
  313. data/test/test_encoding_handler.rb +0 -48
  314. data/test/test_memory_leak.rb +0 -156
  315. data/test/test_nokogiri.rb +0 -138
  316. data/test/test_soap4r_sax.rb +0 -52
  317. data/test/test_xslt_transforms.rb +0 -314
  318. data/test/xml/node/test_save_options.rb +0 -28
  319. data/test/xml/node/test_subclass.rb +0 -44
  320. data/test/xml/sax/test_parser.rb +0 -402
  321. data/test/xml/sax/test_parser_context.rb +0 -115
  322. data/test/xml/sax/test_parser_text.rb +0 -202
  323. data/test/xml/sax/test_push_parser.rb +0 -265
  324. data/test/xml/test_attr.rb +0 -74
  325. data/test/xml/test_attribute_decl.rb +0 -86
  326. data/test/xml/test_builder.rb +0 -341
  327. data/test/xml/test_c14n.rb +0 -180
  328. data/test/xml/test_cdata.rb +0 -54
  329. data/test/xml/test_comment.rb +0 -40
  330. data/test/xml/test_document.rb +0 -982
  331. data/test/xml/test_document_encoding.rb +0 -31
  332. data/test/xml/test_document_fragment.rb +0 -298
  333. data/test/xml/test_dtd.rb +0 -187
  334. data/test/xml/test_dtd_encoding.rb +0 -31
  335. data/test/xml/test_element_content.rb +0 -56
  336. data/test/xml/test_element_decl.rb +0 -73
  337. data/test/xml/test_entity_decl.rb +0 -122
  338. data/test/xml/test_entity_reference.rb +0 -262
  339. data/test/xml/test_namespace.rb +0 -96
  340. data/test/xml/test_node.rb +0 -1325
  341. data/test/xml/test_node_attributes.rb +0 -115
  342. data/test/xml/test_node_encoding.rb +0 -75
  343. data/test/xml/test_node_inheritance.rb +0 -32
  344. data/test/xml/test_node_reparenting.rb +0 -592
  345. data/test/xml/test_node_set.rb +0 -809
  346. data/test/xml/test_parse_options.rb +0 -64
  347. data/test/xml/test_processing_instruction.rb +0 -30
  348. data/test/xml/test_reader.rb +0 -620
  349. data/test/xml/test_reader_encoding.rb +0 -134
  350. data/test/xml/test_relax_ng.rb +0 -60
  351. data/test/xml/test_schema.rb +0 -142
  352. data/test/xml/test_syntax_error.rb +0 -36
  353. data/test/xml/test_text.rb +0 -60
  354. data/test/xml/test_unparented_node.rb +0 -483
  355. data/test/xml/test_xinclude.rb +0 -83
  356. data/test/xml/test_xpath.rb +0 -470
  357. data/test/xslt/test_custom_functions.rb +0 -133
  358. data/test/xslt/test_exception_handling.rb +0 -37
@@ -1,151 +1,154 @@
1
+ # frozen_string_literal: true
1
2
  #--
2
3
  # DO NOT MODIFY!!!!
3
- # This file is automatically generated by rex 1.0.5
4
+ # This file is automatically generated by rex 1.0.7
4
5
  # from lexical definition file "lib/nokogiri/css/tokenizer.rex".
5
6
  #++
6
7
 
7
8
  module Nokogiri
8
9
  module CSS
9
- class Tokenizer # :nodoc:
10
- require 'strscan'
10
+ # :nodoc: all
11
+ class Tokenizer
12
+ require 'strscan'
11
13
 
12
- class ScanError < StandardError ; end
14
+ class ScanError < StandardError ; end
13
15
 
14
- attr_reader :lineno
15
- attr_reader :filename
16
- attr_accessor :state
16
+ attr_reader :lineno
17
+ attr_reader :filename
18
+ attr_accessor :state
17
19
 
18
- def scan_setup(str)
19
- @ss = StringScanner.new(str)
20
- @lineno = 1
21
- @state = nil
22
- end
20
+ def scan_setup(str)
21
+ @ss = StringScanner.new(str)
22
+ @lineno = 1
23
+ @state = nil
24
+ end
23
25
 
24
- def action
25
- yield
26
- end
26
+ def action
27
+ yield
28
+ end
27
29
 
28
- def scan_str(str)
29
- scan_setup(str)
30
- do_parse
31
- end
32
- alias :scan :scan_str
30
+ def scan_str(str)
31
+ scan_setup(str)
32
+ do_parse
33
+ end
34
+ alias :scan :scan_str
33
35
 
34
- def load_file( filename )
35
- @filename = filename
36
- open(filename, "r") do |f|
37
- scan_setup(f.read)
38
- end
39
- end
36
+ def load_file( filename )
37
+ @filename = filename
38
+ File.open(filename, "r") do |f|
39
+ scan_setup(f.read)
40
+ end
41
+ end
40
42
 
41
- def scan_file( filename )
42
- load_file(filename)
43
- do_parse
44
- end
43
+ def scan_file( filename )
44
+ load_file(filename)
45
+ do_parse
46
+ end
45
47
 
46
48
 
47
- def next_token
48
- return if @ss.eos?
49
-
50
- # skips empty actions
51
- until token = _next_token or @ss.eos?; end
52
- token
53
- end
49
+ def next_token
50
+ return if @ss.eos?
54
51
 
55
- def _next_token
56
- text = @ss.peek(1)
57
- @lineno += 1 if text == "\n"
58
- token = case @state
59
- when nil
60
- case
61
- when (text = @ss.scan(/has\([\s]*/))
62
- action { [:HAS, text] }
52
+ # skips empty actions
53
+ until token = _next_token or @ss.eos?; end
54
+ token
55
+ end
63
56
 
64
- when (text = @ss.scan(/[-@]?([_A-Za-z]|[^\0-\177]|\\[0-9A-Fa-f]{1,6}(\r\n|[\s])?|\\[^\n\r\f0-9A-Fa-f])([_A-Za-z0-9-]|[^\0-\177]|\\[0-9A-Fa-f]{1,6}(\r\n|[\s])?|\\[^\n\r\f0-9A-Fa-f])*\([\s]*/))
65
- action { [:FUNCTION, text] }
57
+ def _next_token
58
+ text = @ss.peek(1)
59
+ @lineno += 1 if text == "\n"
60
+ token = case @state
61
+ when nil
62
+ case
63
+ when (text = @ss.scan(/has\([\s]*/))
64
+ action { [:HAS, text] }
66
65
 
67
- when (text = @ss.scan(/[-@]?([_A-Za-z]|[^\0-\177]|\\[0-9A-Fa-f]{1,6}(\r\n|[\s])?|\\[^\n\r\f0-9A-Fa-f])([_A-Za-z0-9-]|[^\0-\177]|\\[0-9A-Fa-f]{1,6}(\r\n|[\s])?|\\[^\n\r\f0-9A-Fa-f])*/))
68
- action { [:IDENT, text] }
66
+ when (text = @ss.scan(/-?([_A-Za-z]|[^\0-\177]|\\[0-9A-Fa-f]{1,6}(\r\n|[\s])?|\\[^\n\r\f0-9A-Fa-f])([_A-Za-z0-9-]|[^\0-\177]|\\[0-9A-Fa-f]{1,6}(\r\n|[\s])?|\\[^\n\r\f0-9A-Fa-f])*\([\s]*/))
67
+ action { [:FUNCTION, text] }
69
68
 
70
- when (text = @ss.scan(/\#([_A-Za-z0-9-]|[^\0-\177]|\\[0-9A-Fa-f]{1,6}(\r\n|[\s])?|\\[^\n\r\f0-9A-Fa-f])+/))
71
- action { [:HASH, text] }
69
+ when (text = @ss.scan(/-?([_A-Za-z]|[^\0-\177]|\\[0-9A-Fa-f]{1,6}(\r\n|[\s])?|\\[^\n\r\f0-9A-Fa-f])([_A-Za-z0-9-]|[^\0-\177]|\\[0-9A-Fa-f]{1,6}(\r\n|[\s])?|\\[^\n\r\f0-9A-Fa-f])*/))
70
+ action { [:IDENT, text] }
72
71
 
73
- when (text = @ss.scan(/[\s]*~=[\s]*/))
74
- action { [:INCLUDES, text] }
72
+ when (text = @ss.scan(/\#([_A-Za-z0-9-]|[^\0-\177]|\\[0-9A-Fa-f]{1,6}(\r\n|[\s])?|\\[^\n\r\f0-9A-Fa-f])+/))
73
+ action { [:HASH, text] }
75
74
 
76
- when (text = @ss.scan(/[\s]*\|=[\s]*/))
77
- action { [:DASHMATCH, text] }
75
+ when (text = @ss.scan(/[\s]*~=[\s]*/))
76
+ action { [:INCLUDES, text] }
78
77
 
79
- when (text = @ss.scan(/[\s]*\^=[\s]*/))
80
- action { [:PREFIXMATCH, text] }
78
+ when (text = @ss.scan(/[\s]*\|=[\s]*/))
79
+ action { [:DASHMATCH, text] }
81
80
 
82
- when (text = @ss.scan(/[\s]*\$=[\s]*/))
83
- action { [:SUFFIXMATCH, text] }
81
+ when (text = @ss.scan(/[\s]*\^=[\s]*/))
82
+ action { [:PREFIXMATCH, text] }
84
83
 
85
- when (text = @ss.scan(/[\s]*\*=[\s]*/))
86
- action { [:SUBSTRINGMATCH, text] }
84
+ when (text = @ss.scan(/[\s]*\$=[\s]*/))
85
+ action { [:SUFFIXMATCH, text] }
87
86
 
88
- when (text = @ss.scan(/[\s]*!=[\s]*/))
89
- action { [:NOT_EQUAL, text] }
87
+ when (text = @ss.scan(/[\s]*\*=[\s]*/))
88
+ action { [:SUBSTRINGMATCH, text] }
90
89
 
91
- when (text = @ss.scan(/[\s]*=[\s]*/))
92
- action { [:EQUAL, text] }
90
+ when (text = @ss.scan(/[\s]*!=[\s]*/))
91
+ action { [:NOT_EQUAL, text] }
93
92
 
94
- when (text = @ss.scan(/[\s]*\)/))
95
- action { [:RPAREN, text] }
93
+ when (text = @ss.scan(/[\s]*=[\s]*/))
94
+ action { [:EQUAL, text] }
96
95
 
97
- when (text = @ss.scan(/\[[\s]*/))
98
- action { [:LSQUARE, text] }
96
+ when (text = @ss.scan(/[\s]*\)/))
97
+ action { [:RPAREN, text] }
99
98
 
100
- when (text = @ss.scan(/[\s]*\]/))
101
- action { [:RSQUARE, text] }
99
+ when (text = @ss.scan(/\[[\s]*/))
100
+ action { [:LSQUARE, text] }
102
101
 
103
- when (text = @ss.scan(/[\s]*\+[\s]*/))
104
- action { [:PLUS, text] }
102
+ when (text = @ss.scan(/[\s]*\]/))
103
+ action { [:RSQUARE, text] }
105
104
 
106
- when (text = @ss.scan(/[\s]*>[\s]*/))
107
- action { [:GREATER, text] }
105
+ when (text = @ss.scan(/[\s]*\+[\s]*/))
106
+ action { [:PLUS, text] }
108
107
 
109
- when (text = @ss.scan(/[\s]*,[\s]*/))
110
- action { [:COMMA, text] }
108
+ when (text = @ss.scan(/[\s]*>[\s]*/))
109
+ action { [:GREATER, text] }
111
110
 
112
- when (text = @ss.scan(/[\s]*~[\s]*/))
113
- action { [:TILDE, text] }
111
+ when (text = @ss.scan(/[\s]*,[\s]*/))
112
+ action { [:COMMA, text] }
114
113
 
115
- when (text = @ss.scan(/\:not\([\s]*/))
116
- action { [:NOT, text] }
114
+ when (text = @ss.scan(/[\s]*~[\s]*/))
115
+ action { [:TILDE, text] }
117
116
 
118
- when (text = @ss.scan(/-?([0-9]+|[0-9]*\.[0-9]+)/))
119
- action { [:NUMBER, text] }
117
+ when (text = @ss.scan(/\:not\([\s]*/))
118
+ action { [:NOT, text] }
120
119
 
121
- when (text = @ss.scan(/[\s]*\/\/[\s]*/))
122
- action { [:DOUBLESLASH, text] }
120
+ when (text = @ss.scan(/-?([0-9]+|[0-9]*\.[0-9]+)/))
121
+ action { [:NUMBER, text] }
123
122
 
124
- when (text = @ss.scan(/[\s]*\/[\s]*/))
125
- action { [:SLASH, text] }
123
+ when (text = @ss.scan(/[\s]*\/\/[\s]*/))
124
+ action { [:DOUBLESLASH, text] }
126
125
 
127
- when (text = @ss.scan(/U\+[0-9a-f?]{1,6}(-[0-9a-f]{1,6})?/))
128
- action {[:UNICODE_RANGE, text] }
126
+ when (text = @ss.scan(/[\s]*\/[\s]*/))
127
+ action { [:SLASH, text] }
129
128
 
130
- when (text = @ss.scan(/[\s]+/))
131
- action { [:S, text] }
129
+ when (text = @ss.scan(/U\+[0-9a-f?]{1,6}(-[0-9a-f]{1,6})?/))
130
+ action {[:UNICODE_RANGE, text] }
132
131
 
133
- when (text = @ss.scan(/"([^\n\r\f"]|\n|\r\n|\r|\f|[^\0-\177]|\\[0-9A-Fa-f]{1,6}(\r\n|[\s])?|\\[^\n\r\f0-9A-Fa-f])*(?<!\\)(?:\\{2})*"|'([^\n\r\f']|\n|\r\n|\r|\f|[^\0-\177]|\\[0-9A-Fa-f]{1,6}(\r\n|[\s])?|\\[^\n\r\f0-9A-Fa-f])*(?<!\\)(?:\\{2})*'/))
134
- action { [:STRING, text] }
132
+ when (text = @ss.scan(/[\s]+/))
133
+ action { [:S, text] }
135
134
 
136
- when (text = @ss.scan(/./))
137
- action { [text, text] }
135
+ when (text = @ss.scan(/"([^\n\r\f"]|\n|\r\n|\r|\f|[^\0-\177]|\\[0-9A-Fa-f]{1,6}(\r\n|[\s])?|\\[^\n\r\f0-9A-Fa-f])*(?<!\\)(?:\\{2})*"|'([^\n\r\f']|\n|\r\n|\r|\f|[^\0-\177]|\\[0-9A-Fa-f]{1,6}(\r\n|[\s])?|\\[^\n\r\f0-9A-Fa-f])*(?<!\\)(?:\\{2})*'/))
136
+ action { [:STRING, text] }
138
137
 
139
- else
140
- text = @ss.string[@ss.pos .. -1]
141
- raise ScanError, "can not match: '" + text + "'"
142
- end # if
138
+ when (text = @ss.scan(/./))
139
+ action { [text, text] }
143
140
 
144
- else
145
- raise ScanError, "undefined state: '" + state.to_s + "'"
146
- end # case state
147
- token
148
- end # def _next_token
141
+
142
+ else
143
+ text = @ss.string[@ss.pos .. -1]
144
+ raise ScanError, "can not match: '" + text + "'"
145
+ end # if
146
+
147
+ else
148
+ raise ScanError, "undefined state: '" + state.to_s + "'"
149
+ end # case state
150
+ token
151
+ end # def _next_token
149
152
 
150
153
  end # class
151
154
  end
@@ -1,6 +1,7 @@
1
1
  module Nokogiri
2
2
  module CSS
3
- class Tokenizer # :nodoc:
3
+ # :nodoc: all
4
+ class Tokenizer
4
5
 
5
6
  macro
6
7
  nl \n|\r\n|\r|\f
@@ -12,7 +13,7 @@ macro
12
13
  escape {unicode}|\\[^\n\r\f0-9A-Fa-f]
13
14
  nmchar [_A-Za-z0-9-]|{nonascii}|{escape}
14
15
  nmstart [_A-Za-z]|{nonascii}|{escape}
15
- ident [-@]?({nmstart})({nmchar})*
16
+ ident -?({nmstart})({nmchar})*
16
17
  name ({nmchar})+
17
18
  string1 "([^\n\r\f"]|{nl}|{nonascii}|{escape})*(?<!\\)(?:\\{2})*"
18
19
  string2 '([^\n\r\f']|{nl}|{nonascii}|{escape})*(?<!\\)(?:\\{2})*'