nokogiri 1.0.0 → 1.6.8.1

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 (309) hide show
  1. checksums.yaml +7 -0
  2. data/.autotest +26 -0
  3. data/.cross_rubies +9 -0
  4. data/.editorconfig +17 -0
  5. data/.gemtest +0 -0
  6. data/.travis.yml +51 -0
  7. data/CHANGELOG.rdoc +1160 -0
  8. data/CONTRIBUTING.md +42 -0
  9. data/C_CODING_STYLE.rdoc +33 -0
  10. data/Gemfile +22 -0
  11. data/LICENSE.txt +31 -0
  12. data/Manifest.txt +284 -40
  13. data/README.md +166 -0
  14. data/ROADMAP.md +111 -0
  15. data/Rakefile +310 -199
  16. data/STANDARD_RESPONSES.md +47 -0
  17. data/Y_U_NO_GEMSPEC.md +155 -0
  18. data/appveyor.yml +22 -0
  19. data/bin/nokogiri +118 -0
  20. data/build_all +45 -0
  21. data/dependencies.yml +29 -0
  22. data/ext/nokogiri/depend +358 -0
  23. data/ext/nokogiri/extconf.rb +664 -34
  24. data/ext/nokogiri/html_document.c +120 -33
  25. data/ext/nokogiri/html_document.h +1 -1
  26. data/ext/nokogiri/html_element_description.c +279 -0
  27. data/ext/nokogiri/html_element_description.h +10 -0
  28. data/ext/nokogiri/html_entity_lookup.c +32 -0
  29. data/ext/nokogiri/html_entity_lookup.h +8 -0
  30. data/ext/nokogiri/html_sax_parser_context.c +116 -0
  31. data/ext/nokogiri/html_sax_parser_context.h +11 -0
  32. data/ext/nokogiri/html_sax_push_parser.c +87 -0
  33. data/ext/nokogiri/html_sax_push_parser.h +9 -0
  34. data/ext/nokogiri/nokogiri.c +145 -0
  35. data/ext/nokogiri/nokogiri.h +131 -0
  36. data/ext/nokogiri/xml_attr.c +94 -0
  37. data/ext/nokogiri/xml_attr.h +9 -0
  38. data/ext/nokogiri/xml_attribute_decl.c +70 -0
  39. data/ext/nokogiri/xml_attribute_decl.h +9 -0
  40. data/ext/nokogiri/xml_cdata.c +23 -19
  41. data/ext/nokogiri/xml_cdata.h +1 -1
  42. data/ext/nokogiri/xml_comment.c +69 -0
  43. data/ext/nokogiri/xml_comment.h +9 -0
  44. data/ext/nokogiri/xml_document.c +501 -54
  45. data/ext/nokogiri/xml_document.h +14 -1
  46. data/ext/nokogiri/xml_document_fragment.c +48 -0
  47. data/ext/nokogiri/xml_document_fragment.h +10 -0
  48. data/ext/nokogiri/xml_dtd.c +109 -24
  49. data/ext/nokogiri/xml_dtd.h +3 -1
  50. data/ext/nokogiri/xml_element_content.c +123 -0
  51. data/ext/nokogiri/xml_element_content.h +10 -0
  52. data/ext/nokogiri/xml_element_decl.c +69 -0
  53. data/ext/nokogiri/xml_element_decl.h +9 -0
  54. data/ext/nokogiri/xml_encoding_handler.c +79 -0
  55. data/ext/nokogiri/xml_encoding_handler.h +8 -0
  56. data/ext/nokogiri/xml_entity_decl.c +110 -0
  57. data/ext/nokogiri/xml_entity_decl.h +10 -0
  58. data/ext/nokogiri/xml_entity_reference.c +52 -0
  59. data/ext/nokogiri/xml_entity_reference.h +9 -0
  60. data/ext/nokogiri/xml_io.c +60 -0
  61. data/ext/nokogiri/xml_io.h +11 -0
  62. data/ext/nokogiri/xml_libxml2_hacks.c +112 -0
  63. data/ext/nokogiri/xml_libxml2_hacks.h +12 -0
  64. data/ext/nokogiri/xml_namespace.c +117 -0
  65. data/ext/nokogiri/xml_namespace.h +13 -0
  66. data/ext/nokogiri/xml_node.c +1285 -315
  67. data/ext/nokogiri/xml_node.h +4 -6
  68. data/ext/nokogiri/xml_node_set.c +415 -54
  69. data/ext/nokogiri/xml_node_set.h +6 -2
  70. data/ext/nokogiri/xml_processing_instruction.c +56 -0
  71. data/ext/nokogiri/xml_processing_instruction.h +9 -0
  72. data/ext/nokogiri/xml_reader.c +316 -77
  73. data/ext/nokogiri/xml_reader.h +1 -1
  74. data/ext/nokogiri/xml_relax_ng.c +161 -0
  75. data/ext/nokogiri/xml_relax_ng.h +9 -0
  76. data/ext/nokogiri/xml_sax_parser.c +215 -80
  77. data/ext/nokogiri/xml_sax_parser.h +30 -1
  78. data/ext/nokogiri/xml_sax_parser_context.c +262 -0
  79. data/ext/nokogiri/xml_sax_parser_context.h +10 -0
  80. data/ext/nokogiri/xml_sax_push_parser.c +115 -0
  81. data/ext/nokogiri/xml_sax_push_parser.h +9 -0
  82. data/ext/nokogiri/xml_schema.c +205 -0
  83. data/ext/nokogiri/xml_schema.h +9 -0
  84. data/ext/nokogiri/xml_syntax_error.c +45 -175
  85. data/ext/nokogiri/xml_syntax_error.h +4 -2
  86. data/ext/nokogiri/xml_text.c +37 -14
  87. data/ext/nokogiri/xml_text.h +1 -1
  88. data/ext/nokogiri/xml_xpath_context.c +230 -13
  89. data/ext/nokogiri/xml_xpath_context.h +2 -1
  90. data/ext/nokogiri/xslt_stylesheet.c +196 -34
  91. data/ext/nokogiri/xslt_stylesheet.h +6 -1
  92. data/lib/nokogiri/css/node.rb +18 -61
  93. data/lib/nokogiri/css/parser.rb +725 -17
  94. data/lib/nokogiri/css/parser.y +126 -63
  95. data/lib/nokogiri/css/parser_extras.rb +91 -0
  96. data/lib/nokogiri/css/syntax_error.rb +7 -0
  97. data/lib/nokogiri/css/tokenizer.rb +148 -5
  98. data/lib/nokogiri/css/tokenizer.rex +31 -39
  99. data/lib/nokogiri/css/xpath_visitor.rb +109 -51
  100. data/lib/nokogiri/css.rb +24 -3
  101. data/lib/nokogiri/decorators/slop.rb +42 -0
  102. data/lib/nokogiri/html/builder.rb +27 -1
  103. data/lib/nokogiri/html/document.rb +329 -3
  104. data/lib/nokogiri/html/document_fragment.rb +39 -0
  105. data/lib/nokogiri/html/element_description.rb +23 -0
  106. data/lib/nokogiri/html/element_description_defaults.rb +671 -0
  107. data/lib/nokogiri/html/entity_lookup.rb +13 -0
  108. data/lib/nokogiri/html/sax/parser.rb +35 -4
  109. data/lib/nokogiri/html/sax/parser_context.rb +16 -0
  110. data/lib/nokogiri/html/sax/push_parser.rb +36 -0
  111. data/lib/nokogiri/html.rb +18 -76
  112. data/lib/nokogiri/syntax_error.rb +4 -0
  113. data/lib/nokogiri/version.rb +106 -1
  114. data/lib/nokogiri/xml/attr.rb +14 -0
  115. data/lib/nokogiri/xml/attribute_decl.rb +18 -0
  116. data/lib/nokogiri/xml/builder.rb +395 -31
  117. data/lib/nokogiri/xml/cdata.rb +4 -2
  118. data/lib/nokogiri/xml/character_data.rb +7 -0
  119. data/lib/nokogiri/xml/document.rb +267 -12
  120. data/lib/nokogiri/xml/document_fragment.rb +149 -0
  121. data/lib/nokogiri/xml/dtd.rb +27 -1
  122. data/lib/nokogiri/xml/element_content.rb +36 -0
  123. data/lib/nokogiri/xml/element_decl.rb +13 -0
  124. data/lib/nokogiri/xml/entity_decl.rb +19 -0
  125. data/lib/nokogiri/xml/namespace.rb +13 -0
  126. data/lib/nokogiri/xml/node/save_options.rb +61 -0
  127. data/lib/nokogiri/xml/node.rb +748 -109
  128. data/lib/nokogiri/xml/node_set.rb +200 -72
  129. data/lib/nokogiri/xml/parse_options.rb +120 -0
  130. data/lib/nokogiri/xml/pp/character_data.rb +18 -0
  131. data/lib/nokogiri/xml/pp/node.rb +56 -0
  132. data/lib/nokogiri/xml/pp.rb +2 -0
  133. data/lib/nokogiri/xml/processing_instruction.rb +8 -0
  134. data/lib/nokogiri/xml/reader.rb +102 -4
  135. data/lib/nokogiri/xml/relax_ng.rb +32 -0
  136. data/lib/nokogiri/xml/sax/document.rb +114 -2
  137. data/lib/nokogiri/xml/sax/parser.rb +97 -7
  138. data/lib/nokogiri/xml/sax/parser_context.rb +16 -0
  139. data/lib/nokogiri/xml/sax/push_parser.rb +60 -0
  140. data/lib/nokogiri/xml/sax.rb +2 -7
  141. data/lib/nokogiri/xml/schema.rb +63 -0
  142. data/lib/nokogiri/xml/searchable.rb +221 -0
  143. data/lib/nokogiri/xml/syntax_error.rb +27 -1
  144. data/lib/nokogiri/xml/text.rb +4 -1
  145. data/lib/nokogiri/xml/xpath/syntax_error.rb +11 -0
  146. data/lib/nokogiri/xml/xpath.rb +4 -0
  147. data/lib/nokogiri/xml/xpath_context.rb +3 -1
  148. data/lib/nokogiri/xml.rb +45 -38
  149. data/lib/nokogiri/xslt/stylesheet.rb +19 -0
  150. data/lib/nokogiri/xslt.rb +47 -2
  151. data/lib/nokogiri.rb +117 -24
  152. data/lib/xsd/xmlparser/nokogiri.rb +102 -0
  153. data/patches/sort-patches-by-date +25 -0
  154. data/ports/archives/libxml2-2.9.4.tar.gz +0 -0
  155. data/ports/archives/libxslt-1.1.29.tar.gz +0 -0
  156. data/suppressions/README.txt +1 -0
  157. data/suppressions/nokogiri_ree-1.8.7.358.supp +61 -0
  158. data/suppressions/nokogiri_ruby-1.8.7.370.supp +0 -0
  159. data/suppressions/nokogiri_ruby-1.9.2.320.supp +28 -0
  160. data/suppressions/nokogiri_ruby-1.9.3.327.supp +28 -0
  161. data/tasks/test.rb +100 -0
  162. data/test/css/test_nthiness.rb +73 -6
  163. data/test/css/test_parser.rb +184 -39
  164. data/test/css/test_tokenizer.rb +72 -19
  165. data/test/css/test_xpath_visitor.rb +44 -2
  166. data/test/decorators/test_slop.rb +20 -0
  167. data/test/files/2ch.html +108 -0
  168. data/test/files/GH_1042.html +18 -0
  169. data/test/files/address_book.rlx +12 -0
  170. data/test/files/address_book.xml +10 -0
  171. data/test/files/atom.xml +344 -0
  172. data/test/files/bar/bar.xsd +4 -0
  173. data/test/files/bogus.xml +0 -0
  174. data/test/files/dont_hurt_em_why.xml +422 -0
  175. data/test/files/encoding.html +82 -0
  176. data/test/files/encoding.xhtml +84 -0
  177. data/test/files/exslt.xml +8 -0
  178. data/test/files/exslt.xslt +35 -0
  179. data/test/files/foo/foo.xsd +4 -0
  180. data/test/files/metacharset.html +10 -0
  181. data/test/files/namespace_pressure_test.xml +1684 -0
  182. data/test/files/noencoding.html +47 -0
  183. data/test/files/po.xml +32 -0
  184. data/test/files/po.xsd +66 -0
  185. data/test/files/saml/saml20assertion_schema.xsd +283 -0
  186. data/test/files/saml/saml20protocol_schema.xsd +302 -0
  187. data/test/files/saml/xenc_schema.xsd +146 -0
  188. data/test/files/saml/xmldsig_schema.xsd +318 -0
  189. data/test/files/shift_jis.html +10 -0
  190. data/test/files/shift_jis.xml +5 -0
  191. data/test/files/shift_jis_no_charset.html +9 -0
  192. data/test/files/slow-xpath.xml +25509 -0
  193. data/test/files/snuggles.xml +3 -0
  194. data/test/files/staff.dtd +10 -0
  195. data/test/files/test_document_url/bar.xml +2 -0
  196. data/test/files/test_document_url/document.dtd +4 -0
  197. data/test/files/test_document_url/document.xml +6 -0
  198. data/test/files/tlm.html +2 -1
  199. data/test/files/to_be_xincluded.xml +2 -0
  200. data/test/files/valid_bar.xml +2 -0
  201. data/test/files/xinclude.xml +4 -0
  202. data/test/helper.rb +124 -13
  203. data/test/html/sax/test_parser.rb +118 -4
  204. data/test/html/sax/test_parser_context.rb +46 -0
  205. data/test/html/sax/test_push_parser.rb +87 -0
  206. data/test/html/test_builder.rb +94 -8
  207. data/test/html/test_document.rb +626 -11
  208. data/test/html/test_document_encoding.rb +145 -0
  209. data/test/html/test_document_fragment.rb +301 -0
  210. data/test/html/test_element_description.rb +105 -0
  211. data/test/html/test_named_characters.rb +14 -0
  212. data/test/html/test_node.rb +212 -0
  213. data/test/html/test_node_encoding.rb +85 -0
  214. data/test/namespaces/test_additional_namespaces_in_builder_doc.rb +14 -0
  215. data/test/namespaces/test_namespaces_aliased_default.rb +24 -0
  216. data/test/namespaces/test_namespaces_in_builder_doc.rb +75 -0
  217. data/test/namespaces/test_namespaces_in_cloned_doc.rb +31 -0
  218. data/test/namespaces/test_namespaces_in_created_doc.rb +75 -0
  219. data/test/namespaces/test_namespaces_in_parsed_doc.rb +80 -0
  220. data/test/namespaces/test_namespaces_preservation.rb +31 -0
  221. data/test/test_convert_xpath.rb +2 -47
  222. data/test/test_css_cache.rb +45 -0
  223. data/test/test_encoding_handler.rb +48 -0
  224. data/test/test_memory_leak.rb +156 -0
  225. data/test/test_nokogiri.rb +103 -1
  226. data/test/test_soap4r_sax.rb +52 -0
  227. data/test/test_xslt_transforms.rb +293 -8
  228. data/test/xml/node/test_save_options.rb +28 -0
  229. data/test/xml/node/test_subclass.rb +44 -0
  230. data/test/xml/sax/test_parser.rb +309 -8
  231. data/test/xml/sax/test_parser_context.rb +115 -0
  232. data/test/xml/sax/test_push_parser.rb +157 -0
  233. data/test/xml/test_attr.rb +67 -0
  234. data/test/xml/test_attribute_decl.rb +86 -0
  235. data/test/xml/test_builder.rb +327 -2
  236. data/test/xml/test_c14n.rb +180 -0
  237. data/test/xml/test_cdata.rb +32 -2
  238. data/test/xml/test_comment.rb +40 -0
  239. data/test/xml/test_document.rb +846 -35
  240. data/test/xml/test_document_encoding.rb +31 -0
  241. data/test/xml/test_document_fragment.rb +271 -0
  242. data/test/xml/test_dtd.rb +153 -9
  243. data/test/xml/test_dtd_encoding.rb +31 -0
  244. data/test/xml/test_element_content.rb +56 -0
  245. data/test/xml/test_element_decl.rb +73 -0
  246. data/test/xml/test_entity_decl.rb +122 -0
  247. data/test/xml/test_entity_reference.rb +251 -0
  248. data/test/xml/test_namespace.rb +96 -0
  249. data/test/xml/test_node.rb +1126 -105
  250. data/test/xml/test_node_attributes.rb +115 -0
  251. data/test/xml/test_node_encoding.rb +69 -0
  252. data/test/xml/test_node_inheritance.rb +32 -0
  253. data/test/xml/test_node_reparenting.rb +549 -0
  254. data/test/xml/test_node_set.rb +668 -9
  255. data/test/xml/test_parse_options.rb +64 -0
  256. data/test/xml/test_processing_instruction.rb +30 -0
  257. data/test/xml/test_reader.rb +589 -0
  258. data/test/xml/test_reader_encoding.rb +134 -0
  259. data/test/xml/test_relax_ng.rb +60 -0
  260. data/test/xml/test_schema.rb +142 -0
  261. data/test/xml/test_syntax_error.rb +30 -0
  262. data/test/xml/test_text.rb +49 -2
  263. data/test/xml/test_unparented_node.rb +440 -0
  264. data/test/xml/test_xinclude.rb +83 -0
  265. data/test/xml/test_xpath.rb +445 -0
  266. data/test/xslt/test_custom_functions.rb +133 -0
  267. data/test/xslt/test_exception_handling.rb +37 -0
  268. data/test_all +107 -0
  269. metadata +459 -115
  270. data/History.txt +0 -6
  271. data/README.ja.txt +0 -86
  272. data/README.txt +0 -87
  273. data/ext/nokogiri/html_sax_parser.c +0 -32
  274. data/ext/nokogiri/html_sax_parser.h +0 -11
  275. data/ext/nokogiri/native.c +0 -40
  276. data/ext/nokogiri/native.h +0 -51
  277. data/ext/nokogiri/xml_xpath.c +0 -46
  278. data/ext/nokogiri/xml_xpath.h +0 -11
  279. data/lib/nokogiri/css/generated_parser.rb +0 -653
  280. data/lib/nokogiri/css/generated_tokenizer.rb +0 -159
  281. data/lib/nokogiri/decorators/hpricot/node.rb +0 -58
  282. data/lib/nokogiri/decorators/hpricot/node_set.rb +0 -14
  283. data/lib/nokogiri/decorators/hpricot/xpath_visitor.rb +0 -17
  284. data/lib/nokogiri/decorators/hpricot.rb +0 -3
  285. data/lib/nokogiri/decorators.rb +0 -1
  286. data/lib/nokogiri/hpricot.rb +0 -47
  287. data/lib/nokogiri/xml/after_handler.rb +0 -18
  288. data/lib/nokogiri/xml/before_handler.rb +0 -32
  289. data/lib/nokogiri/xml/element.rb +0 -6
  290. data/lib/nokogiri/xml/entity_declaration.rb +0 -9
  291. data/nokogiri.gemspec +0 -34
  292. data/test/hpricot/files/basic.xhtml +0 -17
  293. data/test/hpricot/files/boingboing.html +0 -2266
  294. data/test/hpricot/files/cy0.html +0 -3653
  295. data/test/hpricot/files/immob.html +0 -400
  296. data/test/hpricot/files/pace_application.html +0 -1320
  297. data/test/hpricot/files/tenderlove.html +0 -16
  298. data/test/hpricot/files/uswebgen.html +0 -220
  299. data/test/hpricot/files/utf8.html +0 -1054
  300. data/test/hpricot/files/week9.html +0 -1723
  301. data/test/hpricot/files/why.xml +0 -19
  302. data/test/hpricot/load_files.rb +0 -7
  303. data/test/hpricot/test_alter.rb +0 -67
  304. data/test/hpricot/test_builder.rb +0 -27
  305. data/test/hpricot/test_parser.rb +0 -423
  306. data/test/hpricot/test_paths.rb +0 -15
  307. data/test/hpricot/test_preserved.rb +0 -78
  308. data/test/hpricot/test_xml.rb +0 -30
  309. data/test/test_reader.rb +0 -222
@@ -1,22 +1,92 @@
1
1
  #include <html_document.h>
2
2
 
3
+ static ID id_encoding_found;
4
+
3
5
  /*
4
6
  * call-seq:
5
- * serialize
7
+ * new
6
8
  *
7
- * Serialize this document
9
+ * Create a new document
8
10
  */
9
- static VALUE serialize(VALUE self)
11
+ static VALUE new(int argc, VALUE *argv, VALUE klass)
10
12
  {
11
- xmlDocPtr doc;
12
- xmlChar *buf;
13
- int size;
14
- Data_Get_Struct(self, xmlDoc, doc);
13
+ VALUE uri, external_id, rest, rb_doc;
14
+ htmlDocPtr doc;
15
+
16
+ rb_scan_args(argc, argv, "0*", &rest);
17
+ uri = rb_ary_entry(rest, (long)0);
18
+ external_id = rb_ary_entry(rest, (long)1);
19
+
20
+ doc = htmlNewDoc(
21
+ RTEST(uri) ? (const xmlChar *)StringValueCStr(uri) : NULL,
22
+ RTEST(external_id) ? (const xmlChar *)StringValueCStr(external_id) : NULL
23
+ );
24
+ rb_doc = Nokogiri_wrap_xml_document(klass, doc);
25
+ rb_obj_call_init(rb_doc, argc, argv);
26
+ return rb_doc ;
27
+ }
28
+
29
+ /*
30
+ * call-seq:
31
+ * read_io(io, url, encoding, options)
32
+ *
33
+ * Read the HTML document from +io+ with given +url+, +encoding+,
34
+ * and +options+. See Nokogiri::HTML.parse
35
+ */
36
+ static VALUE read_io( VALUE klass,
37
+ VALUE io,
38
+ VALUE url,
39
+ VALUE encoding,
40
+ VALUE options )
41
+ {
42
+ const char * c_url = NIL_P(url) ? NULL : StringValueCStr(url);
43
+ const char * c_enc = NIL_P(encoding) ? NULL : StringValueCStr(encoding);
44
+ VALUE error_list = rb_ary_new();
45
+ VALUE document;
46
+ htmlDocPtr doc;
47
+
48
+ xmlResetLastError();
49
+ xmlSetStructuredErrorFunc((void *)error_list, Nokogiri_error_array_pusher);
50
+
51
+ doc = htmlReadIO(
52
+ io_read_callback,
53
+ io_close_callback,
54
+ (void *)io,
55
+ c_url,
56
+ c_enc,
57
+ (int)NUM2INT(options)
58
+ );
59
+ xmlSetStructuredErrorFunc(NULL, NULL);
15
60
 
16
- htmlDocDumpMemory(doc, &buf, &size);
17
- VALUE rb_str = rb_str_new((char *)buf, (long)size);
18
- free(buf);
19
- return rb_str;
61
+ /*
62
+ * If EncodingFound has occurred in EncodingReader, make sure to do
63
+ * a cleanup and propagate the error.
64
+ */
65
+ if (rb_respond_to(io, id_encoding_found)) {
66
+ VALUE encoding_found = rb_funcall(io, id_encoding_found, 0);
67
+ if (!NIL_P(encoding_found)) {
68
+ xmlFreeDoc(doc);
69
+ rb_exc_raise(encoding_found);
70
+ }
71
+ }
72
+
73
+ if(doc == NULL) {
74
+ xmlErrorPtr error;
75
+
76
+ xmlFreeDoc(doc);
77
+
78
+ error = xmlGetLastError();
79
+ if(error)
80
+ rb_exc_raise(Nokogiri_wrap_xml_syntax_error(error));
81
+ else
82
+ rb_raise(rb_eRuntimeError, "Could not parse document");
83
+
84
+ return Qnil;
85
+ }
86
+
87
+ document = Nokogiri_wrap_xml_document(klass, doc);
88
+ rb_iv_set(document, "@errors", error_list);
89
+ return document;
20
90
  }
21
91
 
22
92
  /*
@@ -33,16 +103,36 @@ static VALUE read_memory( VALUE klass,
33
103
  VALUE options )
34
104
  {
35
105
  const char * c_buffer = StringValuePtr(string);
36
- const char * c_url = (url == Qnil) ? NULL : StringValuePtr(url);
37
- const char * c_enc = (encoding == Qnil) ? NULL : StringValuePtr(encoding);
38
- int len = NUM2INT(rb_funcall(string, rb_intern("length"), 0));
106
+ const char * c_url = NIL_P(url) ? NULL : StringValueCStr(url);
107
+ const char * c_enc = NIL_P(encoding) ? NULL : StringValueCStr(encoding);
108
+ int len = (int)RSTRING_LEN(string);
109
+ VALUE error_list = rb_ary_new();
110
+ VALUE document;
111
+ htmlDocPtr doc;
39
112
 
40
- htmlDocPtr doc = htmlReadMemory(c_buffer, len, c_url, c_enc, NUM2INT(options));
113
+ xmlResetLastError();
114
+ xmlSetStructuredErrorFunc((void *)error_list, Nokogiri_error_array_pusher);
41
115
 
42
- if(doc == NULL)
43
- doc = htmlNewDoc((const xmlChar *)c_url, NULL);
116
+ doc = htmlReadMemory(c_buffer, len, c_url, c_enc, (int)NUM2INT(options));
117
+ xmlSetStructuredErrorFunc(NULL, NULL);
44
118
 
45
- return Nokogiri_wrap_xml_document(klass, doc);
119
+ if(doc == NULL) {
120
+ xmlErrorPtr error;
121
+
122
+ xmlFreeDoc(doc);
123
+
124
+ error = xmlGetLastError();
125
+ if(error)
126
+ rb_exc_raise(Nokogiri_wrap_xml_syntax_error(error));
127
+ else
128
+ rb_raise(rb_eRuntimeError, "Could not parse document");
129
+
130
+ return Qnil;
131
+ }
132
+
133
+ document = Nokogiri_wrap_xml_document(klass, doc);
134
+ rb_iv_set(document, "@errors", error_list);
135
+ return document;
46
136
  }
47
137
 
48
138
  /*
@@ -55,29 +145,26 @@ static VALUE type(VALUE self)
55
145
  {
56
146
  htmlDocPtr doc;
57
147
  Data_Get_Struct(self, xmlDoc, doc);
58
- return INT2NUM((int)doc->type);
148
+ return INT2NUM((long)doc->type);
59
149
  }
60
150
 
61
151
  VALUE cNokogiriHtmlDocument ;
62
152
  void init_html_document()
63
153
  {
64
- /*
65
- * HACK. This is so that rdoc will work with this C file.
66
- */
67
- /*
68
- VALUE nokogiri = rb_define_module("Nokogiri");
69
- VALUE html = rb_define_module_under(nokogiri, "HTML");
70
- VALUE xml = rb_define_module_under(nokogiri, "XML");
71
- VALUE node = rb_define_class_under(xml, "Node", rb_cObject);
72
- VALUE xml_doc = rb_define_class_under(xml, "Document", node);
73
- VALUE klass = rb_define_class_under(html, "Document", xml_doc);
74
- */
154
+ VALUE nokogiri = rb_define_module("Nokogiri");
155
+ VALUE html = rb_define_module_under(nokogiri, "HTML");
156
+ VALUE xml = rb_define_module_under(nokogiri, "XML");
157
+ VALUE node = rb_define_class_under(xml, "Node", rb_cObject);
158
+ VALUE xml_doc = rb_define_class_under(xml, "Document", node);
159
+ VALUE klass = rb_define_class_under(html, "Document", xml_doc);
75
160
 
76
- VALUE klass ;
77
- klass = cNokogiriHtmlDocument = rb_const_get(mNokogiriHtml, rb_intern("Document"));
161
+ cNokogiriHtmlDocument = klass;
78
162
 
79
163
  rb_define_singleton_method(klass, "read_memory", read_memory, 4);
164
+ rb_define_singleton_method(klass, "read_io", read_io, 4);
165
+ rb_define_singleton_method(klass, "new", new, -1);
80
166
 
81
167
  rb_define_method(klass, "type", type, 0);
82
- rb_define_method(klass, "serialize", serialize, 0);
168
+
169
+ id_encoding_found = rb_intern("encoding_found");
83
170
  }
@@ -1,7 +1,7 @@
1
1
  #ifndef NOKOGIRI_HTML_DOCUMENT
2
2
  #define NOKOGIRI_HTML_DOCUMENT
3
3
 
4
- #include <native.h>
4
+ #include <nokogiri.h>
5
5
 
6
6
  void init_html_document();
7
7
 
@@ -0,0 +1,279 @@
1
+ #include <html_element_description.h>
2
+
3
+ /*
4
+ * call-seq:
5
+ * required_attributes
6
+ *
7
+ * A list of required attributes for this element
8
+ */
9
+ static VALUE required_attributes(VALUE self)
10
+ {
11
+ htmlElemDesc * description;
12
+ VALUE list;
13
+ int i;
14
+
15
+ Data_Get_Struct(self, htmlElemDesc, description);
16
+
17
+ list = rb_ary_new();
18
+
19
+ if(NULL == description->attrs_req) return list;
20
+
21
+ for(i = 0; description->attrs_depr[i]; i++) {
22
+ rb_ary_push(list, NOKOGIRI_STR_NEW2(description->attrs_req[i]));
23
+ }
24
+
25
+ return list;
26
+ }
27
+
28
+ /*
29
+ * call-seq:
30
+ * deprecated_attributes
31
+ *
32
+ * A list of deprecated attributes for this element
33
+ */
34
+ static VALUE deprecated_attributes(VALUE self)
35
+ {
36
+ htmlElemDesc * description;
37
+ VALUE list;
38
+ int i;
39
+
40
+ Data_Get_Struct(self, htmlElemDesc, description);
41
+
42
+ list = rb_ary_new();
43
+
44
+ if(NULL == description->attrs_depr) return list;
45
+
46
+ for(i = 0; description->attrs_depr[i]; i++) {
47
+ rb_ary_push(list, NOKOGIRI_STR_NEW2(description->attrs_depr[i]));
48
+ }
49
+
50
+ return list;
51
+ }
52
+
53
+ /*
54
+ * call-seq:
55
+ * optional_attributes
56
+ *
57
+ * A list of optional attributes for this element
58
+ */
59
+ static VALUE optional_attributes(VALUE self)
60
+ {
61
+ htmlElemDesc * description;
62
+ VALUE list;
63
+ int i;
64
+
65
+ Data_Get_Struct(self, htmlElemDesc, description);
66
+
67
+ list = rb_ary_new();
68
+
69
+ if(NULL == description->attrs_opt) return list;
70
+
71
+ for(i = 0; description->attrs_opt[i]; i++) {
72
+ rb_ary_push(list, NOKOGIRI_STR_NEW2(description->attrs_opt[i]));
73
+ }
74
+
75
+ return list;
76
+ }
77
+
78
+ /*
79
+ * call-seq:
80
+ * default_sub_element
81
+ *
82
+ * The default sub element for this element
83
+ */
84
+ static VALUE default_sub_element(VALUE self)
85
+ {
86
+ htmlElemDesc * description;
87
+ Data_Get_Struct(self, htmlElemDesc, description);
88
+
89
+ if (description->defaultsubelt)
90
+ return NOKOGIRI_STR_NEW2(description->defaultsubelt);
91
+
92
+ return Qnil;
93
+ }
94
+
95
+ /*
96
+ * call-seq:
97
+ * sub_elements
98
+ *
99
+ * A list of allowed sub elements for this element.
100
+ */
101
+ static VALUE sub_elements(VALUE self)
102
+ {
103
+ htmlElemDesc * description;
104
+ VALUE list;
105
+ int i;
106
+
107
+ Data_Get_Struct(self, htmlElemDesc, description);
108
+
109
+ list = rb_ary_new();
110
+
111
+ if(NULL == description->subelts) return list;
112
+
113
+ for(i = 0; description->subelts[i]; i++) {
114
+ rb_ary_push(list, NOKOGIRI_STR_NEW2(description->subelts[i]));
115
+ }
116
+
117
+ return list;
118
+ }
119
+
120
+ /*
121
+ * call-seq:
122
+ * description
123
+ *
124
+ * The description for this element
125
+ */
126
+ static VALUE description(VALUE self)
127
+ {
128
+ htmlElemDesc * description;
129
+ Data_Get_Struct(self, htmlElemDesc, description);
130
+
131
+ return NOKOGIRI_STR_NEW2(description->desc);
132
+ }
133
+
134
+ /*
135
+ * call-seq:
136
+ * inline?
137
+ *
138
+ * Is this element an inline element?
139
+ */
140
+ static VALUE inline_eh(VALUE self)
141
+ {
142
+ htmlElemDesc * description;
143
+ Data_Get_Struct(self, htmlElemDesc, description);
144
+
145
+ if(description->isinline) return Qtrue;
146
+ return Qfalse;
147
+ }
148
+
149
+ /*
150
+ * call-seq:
151
+ * deprecated?
152
+ *
153
+ * Is this element deprecated?
154
+ */
155
+ static VALUE deprecated_eh(VALUE self)
156
+ {
157
+ htmlElemDesc * description;
158
+ Data_Get_Struct(self, htmlElemDesc, description);
159
+
160
+ if(description->depr) return Qtrue;
161
+ return Qfalse;
162
+ }
163
+
164
+ /*
165
+ * call-seq:
166
+ * empty?
167
+ *
168
+ * Is this an empty element?
169
+ */
170
+ static VALUE empty_eh(VALUE self)
171
+ {
172
+ htmlElemDesc * description;
173
+ Data_Get_Struct(self, htmlElemDesc, description);
174
+
175
+ if(description->empty) return Qtrue;
176
+ return Qfalse;
177
+ }
178
+
179
+ /*
180
+ * call-seq:
181
+ * save_end_tag?
182
+ *
183
+ * Should the end tag be saved?
184
+ */
185
+ static VALUE save_end_tag_eh(VALUE self)
186
+ {
187
+ htmlElemDesc * description;
188
+ Data_Get_Struct(self, htmlElemDesc, description);
189
+
190
+ if(description->saveEndTag) return Qtrue;
191
+ return Qfalse;
192
+ }
193
+
194
+ /*
195
+ * call-seq:
196
+ * implied_end_tag?
197
+ *
198
+ * Can the end tag be implied for this tag?
199
+ */
200
+ static VALUE implied_end_tag_eh(VALUE self)
201
+ {
202
+ htmlElemDesc * description;
203
+ Data_Get_Struct(self, htmlElemDesc, description);
204
+
205
+ if(description->endTag) return Qtrue;
206
+ return Qfalse;
207
+ }
208
+
209
+ /*
210
+ * call-seq:
211
+ * implied_start_tag?
212
+ *
213
+ * Can the start tag be implied for this tag?
214
+ */
215
+ static VALUE implied_start_tag_eh(VALUE self)
216
+ {
217
+ htmlElemDesc * description;
218
+ Data_Get_Struct(self, htmlElemDesc, description);
219
+
220
+ if(description->startTag) return Qtrue;
221
+ return Qfalse;
222
+ }
223
+
224
+ /*
225
+ * call-seq:
226
+ * name
227
+ *
228
+ * Get the tag name for this ElemementDescription
229
+ */
230
+ static VALUE name(VALUE self)
231
+ {
232
+ htmlElemDesc * description;
233
+ Data_Get_Struct(self, htmlElemDesc, description);
234
+
235
+ if(NULL == description->name) return Qnil;
236
+ return NOKOGIRI_STR_NEW2(description->name);
237
+ }
238
+
239
+ /*
240
+ * call-seq:
241
+ * [](tag_name)
242
+ *
243
+ * Get ElemementDescription for +tag_name+
244
+ */
245
+ static VALUE get_description(VALUE klass, VALUE tag_name)
246
+ {
247
+ const htmlElemDesc * description = htmlTagLookup(
248
+ (const xmlChar *)StringValueCStr(tag_name)
249
+ );
250
+
251
+ if(NULL == description) return Qnil;
252
+ return Data_Wrap_Struct(klass, 0, 0, (void *)description);
253
+ }
254
+
255
+ VALUE cNokogiriHtmlElementDescription ;
256
+ void init_html_element_description()
257
+ {
258
+ VALUE nokogiri = rb_define_module("Nokogiri");
259
+ VALUE html = rb_define_module_under(nokogiri, "HTML");
260
+ VALUE klass = rb_define_class_under(html, "ElementDescription",rb_cObject);
261
+
262
+ cNokogiriHtmlElementDescription = klass;
263
+
264
+ rb_define_singleton_method(klass, "[]", get_description, 1);
265
+
266
+ rb_define_method(klass, "name", name, 0);
267
+ rb_define_method(klass, "implied_start_tag?", implied_start_tag_eh, 0);
268
+ rb_define_method(klass, "implied_end_tag?", implied_end_tag_eh, 0);
269
+ rb_define_method(klass, "save_end_tag?", save_end_tag_eh, 0);
270
+ rb_define_method(klass, "empty?", empty_eh, 0);
271
+ rb_define_method(klass, "deprecated?", deprecated_eh, 0);
272
+ rb_define_method(klass, "inline?", inline_eh, 0);
273
+ rb_define_method(klass, "description", description, 0);
274
+ rb_define_method(klass, "sub_elements", sub_elements, 0);
275
+ rb_define_method(klass, "default_sub_element", default_sub_element, 0);
276
+ rb_define_method(klass, "optional_attributes", optional_attributes, 0);
277
+ rb_define_method(klass, "deprecated_attributes", deprecated_attributes, 0);
278
+ rb_define_method(klass, "required_attributes", required_attributes, 0);
279
+ }
@@ -0,0 +1,10 @@
1
+ #ifndef NOKOGIRI_HTML_ELEMENT_DESCRIPTION
2
+ #define NOKOGIRI_HTML_ELEMENT_DESCRIPTION
3
+
4
+ #include <nokogiri.h>
5
+
6
+ void init_html_element_description();
7
+
8
+ extern VALUE cNokogiriHtmlElementDescription ;
9
+
10
+ #endif
@@ -0,0 +1,32 @@
1
+ #include <html_entity_lookup.h>
2
+
3
+ /*
4
+ * call-seq:
5
+ * get(key)
6
+ *
7
+ * Get the HTML::EntityDescription for +key+
8
+ */
9
+ static VALUE get(VALUE self, VALUE key)
10
+ {
11
+ const htmlEntityDesc * desc =
12
+ htmlEntityLookup((const xmlChar *)StringValueCStr(key));
13
+ VALUE klass, args[3];
14
+
15
+ if(NULL == desc) return Qnil;
16
+ klass = rb_const_get(mNokogiriHtml, rb_intern("EntityDescription"));
17
+
18
+ args[0] = INT2NUM((long)desc->value);
19
+ args[1] = NOKOGIRI_STR_NEW2(desc->name);
20
+ args[2] = NOKOGIRI_STR_NEW2(desc->desc);
21
+
22
+ return rb_class_new_instance(3, args, klass);
23
+ }
24
+
25
+ void init_html_entity_lookup()
26
+ {
27
+ VALUE nokogiri = rb_define_module("Nokogiri");
28
+ VALUE html = rb_define_module_under(nokogiri, "HTML");
29
+ VALUE klass = rb_define_class_under(html, "EntityLookup", rb_cObject);
30
+
31
+ rb_define_method(klass, "get", get, 1);
32
+ }
@@ -0,0 +1,8 @@
1
+ #ifndef NOKOGIRI_HTML_ENTITY_LOOKUP
2
+ #define NOKOGIRI_HTML_ENTITY_LOOKUP
3
+
4
+ #include <nokogiri.h>
5
+
6
+ void init_html_entity_lookup();
7
+
8
+ #endif
@@ -0,0 +1,116 @@
1
+ #include <html_sax_parser_context.h>
2
+
3
+ VALUE cNokogiriHtmlSaxParserContext ;
4
+
5
+ static void deallocate(xmlParserCtxtPtr ctxt)
6
+ {
7
+ NOKOGIRI_DEBUG_START(handler);
8
+
9
+ ctxt->sax = NULL;
10
+
11
+ htmlFreeParserCtxt(ctxt);
12
+
13
+ NOKOGIRI_DEBUG_END(handler);
14
+ }
15
+
16
+ static VALUE
17
+ parse_memory(VALUE klass, VALUE data, VALUE encoding)
18
+ {
19
+ htmlParserCtxtPtr ctxt;
20
+
21
+ if (NIL_P(data))
22
+ rb_raise(rb_eArgError, "data cannot be nil");
23
+ if (!(int)RSTRING_LEN(data))
24
+ rb_raise(rb_eRuntimeError, "data cannot be empty");
25
+
26
+ ctxt = htmlCreateMemoryParserCtxt(StringValuePtr(data),
27
+ (int)RSTRING_LEN(data));
28
+ if (ctxt->sax) {
29
+ xmlFree(ctxt->sax);
30
+ ctxt->sax = NULL;
31
+ }
32
+
33
+ if (RTEST(encoding)) {
34
+ xmlCharEncodingHandlerPtr enc = xmlFindCharEncodingHandler(StringValueCStr(encoding));
35
+ if (enc != NULL) {
36
+ xmlSwitchToEncoding(ctxt, enc);
37
+ if (ctxt->errNo == XML_ERR_UNSUPPORTED_ENCODING) {
38
+ rb_raise(rb_eRuntimeError, "Unsupported encoding %s",
39
+ StringValueCStr(encoding));
40
+ }
41
+ }
42
+ }
43
+
44
+ return Data_Wrap_Struct(klass, NULL, deallocate, ctxt);
45
+ }
46
+
47
+ static VALUE parse_file(VALUE klass, VALUE filename, VALUE encoding)
48
+ {
49
+ htmlParserCtxtPtr ctxt = htmlCreateFileParserCtxt(
50
+ StringValueCStr(filename),
51
+ StringValueCStr(encoding)
52
+ );
53
+ return Data_Wrap_Struct(klass, NULL, deallocate, ctxt);
54
+ }
55
+
56
+ static VALUE
57
+ parse_doc(VALUE ctxt_val)
58
+ {
59
+ htmlParserCtxtPtr ctxt = (htmlParserCtxtPtr)ctxt_val;
60
+ htmlParseDocument(ctxt);
61
+ return Qnil;
62
+ }
63
+
64
+ static VALUE
65
+ parse_doc_finalize(VALUE ctxt_val)
66
+ {
67
+ htmlParserCtxtPtr ctxt = (htmlParserCtxtPtr)ctxt_val;
68
+
69
+ if (ctxt->myDoc)
70
+ xmlFreeDoc(ctxt->myDoc);
71
+
72
+ NOKOGIRI_SAX_TUPLE_DESTROY(ctxt->userData);
73
+ return Qnil;
74
+ }
75
+
76
+ static VALUE
77
+ parse_with(VALUE self, VALUE sax_handler)
78
+ {
79
+ htmlParserCtxtPtr ctxt;
80
+ htmlSAXHandlerPtr sax;
81
+
82
+ if (!rb_obj_is_kind_of(sax_handler, cNokogiriXmlSaxParser))
83
+ rb_raise(rb_eArgError, "argument must be a Nokogiri::XML::SAX::Parser");
84
+
85
+ Data_Get_Struct(self, htmlParserCtxt, ctxt);
86
+ Data_Get_Struct(sax_handler, htmlSAXHandler, sax);
87
+
88
+ /* Free the sax handler since we'll assign our own */
89
+ if (ctxt->sax && ctxt->sax != (xmlSAXHandlerPtr)&xmlDefaultSAXHandler)
90
+ xmlFree(ctxt->sax);
91
+
92
+ ctxt->sax = sax;
93
+ ctxt->userData = (void *)NOKOGIRI_SAX_TUPLE_NEW(ctxt, sax_handler);
94
+
95
+ rb_ensure(parse_doc, (VALUE)ctxt, parse_doc_finalize, (VALUE)ctxt);
96
+
97
+ return self;
98
+ }
99
+
100
+ void init_html_sax_parser_context()
101
+ {
102
+ VALUE nokogiri = rb_define_module("Nokogiri");
103
+ VALUE xml = rb_define_module_under(nokogiri, "XML");
104
+ VALUE html = rb_define_module_under(nokogiri, "HTML");
105
+ VALUE sax = rb_define_module_under(xml, "SAX");
106
+ VALUE hsax = rb_define_module_under(html, "SAX");
107
+ VALUE pc = rb_define_class_under(sax, "ParserContext", rb_cObject);
108
+ VALUE klass = rb_define_class_under(hsax, "ParserContext", pc);
109
+
110
+ cNokogiriHtmlSaxParserContext = klass;
111
+
112
+ rb_define_singleton_method(klass, "memory", parse_memory, 2);
113
+ rb_define_singleton_method(klass, "file", parse_file, 2);
114
+
115
+ rb_define_method(klass, "parse_with", parse_with, 1);
116
+ }
@@ -0,0 +1,11 @@
1
+ #ifndef NOKOGIRI_HTML_SAX_PARSER_CONTEXT
2
+ #define NOKOGIRI_HTML_SAX_PARSER_CONTEXT
3
+
4
+ #include <nokogiri.h>
5
+
6
+ extern VALUE cNokogiriHtmlSaxParserContext;
7
+
8
+ void init_html_sax_parser_context();
9
+
10
+ #endif
11
+