nokogiri-backport 1.11.0

Sign up to get free protection for your applications and to get access to all the features.
Files changed (239) hide show
  1. checksums.yaml +7 -0
  2. data/Gemfile +3 -0
  3. data/LICENSE-DEPENDENCIES.md +1682 -0
  4. data/LICENSE.md +9 -0
  5. data/README.md +272 -0
  6. data/bin/nokogiri +118 -0
  7. data/dependencies.yml +74 -0
  8. data/ext/java/nokogiri/EncodingHandler.java +124 -0
  9. data/ext/java/nokogiri/HtmlDocument.java +178 -0
  10. data/ext/java/nokogiri/HtmlElementDescription.java +148 -0
  11. data/ext/java/nokogiri/HtmlEntityLookup.java +79 -0
  12. data/ext/java/nokogiri/HtmlSaxParserContext.java +282 -0
  13. data/ext/java/nokogiri/HtmlSaxPushParser.java +222 -0
  14. data/ext/java/nokogiri/NokogiriService.java +597 -0
  15. data/ext/java/nokogiri/XmlAttr.java +162 -0
  16. data/ext/java/nokogiri/XmlAttributeDecl.java +129 -0
  17. data/ext/java/nokogiri/XmlCdata.java +82 -0
  18. data/ext/java/nokogiri/XmlComment.java +97 -0
  19. data/ext/java/nokogiri/XmlDocument.java +633 -0
  20. data/ext/java/nokogiri/XmlDocumentFragment.java +185 -0
  21. data/ext/java/nokogiri/XmlDtd.java +481 -0
  22. data/ext/java/nokogiri/XmlElement.java +68 -0
  23. data/ext/java/nokogiri/XmlElementContent.java +382 -0
  24. data/ext/java/nokogiri/XmlElementDecl.java +147 -0
  25. data/ext/java/nokogiri/XmlEntityDecl.java +157 -0
  26. data/ext/java/nokogiri/XmlEntityReference.java +101 -0
  27. data/ext/java/nokogiri/XmlNamespace.java +199 -0
  28. data/ext/java/nokogiri/XmlNode.java +1684 -0
  29. data/ext/java/nokogiri/XmlNodeSet.java +434 -0
  30. data/ext/java/nokogiri/XmlProcessingInstruction.java +100 -0
  31. data/ext/java/nokogiri/XmlReader.java +531 -0
  32. data/ext/java/nokogiri/XmlRelaxng.java +151 -0
  33. data/ext/java/nokogiri/XmlSaxParserContext.java +374 -0
  34. data/ext/java/nokogiri/XmlSaxPushParser.java +286 -0
  35. data/ext/java/nokogiri/XmlSchema.java +388 -0
  36. data/ext/java/nokogiri/XmlSyntaxError.java +138 -0
  37. data/ext/java/nokogiri/XmlText.java +110 -0
  38. data/ext/java/nokogiri/XmlXpathContext.java +301 -0
  39. data/ext/java/nokogiri/XsltStylesheet.java +347 -0
  40. data/ext/java/nokogiri/internals/ClosedStreamException.java +10 -0
  41. data/ext/java/nokogiri/internals/HtmlDomParserContext.java +252 -0
  42. data/ext/java/nokogiri/internals/IgnoreSchemaErrorsErrorHandler.java +20 -0
  43. data/ext/java/nokogiri/internals/NokogiriBlockingQueueInputStream.java +151 -0
  44. data/ext/java/nokogiri/internals/NokogiriDomParser.java +116 -0
  45. data/ext/java/nokogiri/internals/NokogiriEntityResolver.java +121 -0
  46. data/ext/java/nokogiri/internals/NokogiriErrorHandler.java +69 -0
  47. data/ext/java/nokogiri/internals/NokogiriHandler.java +327 -0
  48. data/ext/java/nokogiri/internals/NokogiriHelpers.java +734 -0
  49. data/ext/java/nokogiri/internals/NokogiriNamespaceCache.java +217 -0
  50. data/ext/java/nokogiri/internals/NokogiriNamespaceContext.java +127 -0
  51. data/ext/java/nokogiri/internals/NokogiriNonStrictErrorHandler.java +100 -0
  52. data/ext/java/nokogiri/internals/NokogiriNonStrictErrorHandler4NekoHtml.java +121 -0
  53. data/ext/java/nokogiri/internals/NokogiriStrictErrorHandler.java +78 -0
  54. data/ext/java/nokogiri/internals/NokogiriXPathFunction.java +180 -0
  55. data/ext/java/nokogiri/internals/NokogiriXPathFunctionResolver.java +72 -0
  56. data/ext/java/nokogiri/internals/NokogiriXPathVariableResolver.java +60 -0
  57. data/ext/java/nokogiri/internals/NokogiriXsltErrorListener.java +87 -0
  58. data/ext/java/nokogiri/internals/ParserContext.java +259 -0
  59. data/ext/java/nokogiri/internals/ReaderNode.java +488 -0
  60. data/ext/java/nokogiri/internals/SaveContextVisitor.java +778 -0
  61. data/ext/java/nokogiri/internals/SchemaErrorHandler.java +73 -0
  62. data/ext/java/nokogiri/internals/XalanDTMManagerPatch.java +168 -0
  63. data/ext/java/nokogiri/internals/XmlDeclHandler.java +42 -0
  64. data/ext/java/nokogiri/internals/XmlDomParserContext.java +274 -0
  65. data/ext/java/nokogiri/internals/XmlSaxParser.java +65 -0
  66. data/ext/java/nokogiri/internals/c14n/AttrCompare.java +119 -0
  67. data/ext/java/nokogiri/internals/c14n/C14nHelper.java +159 -0
  68. data/ext/java/nokogiri/internals/c14n/CanonicalFilter.java +37 -0
  69. data/ext/java/nokogiri/internals/c14n/CanonicalizationException.java +93 -0
  70. data/ext/java/nokogiri/internals/c14n/Canonicalizer.java +252 -0
  71. data/ext/java/nokogiri/internals/c14n/Canonicalizer11.java +639 -0
  72. data/ext/java/nokogiri/internals/c14n/Canonicalizer11_OmitComments.java +38 -0
  73. data/ext/java/nokogiri/internals/c14n/Canonicalizer11_WithComments.java +38 -0
  74. data/ext/java/nokogiri/internals/c14n/Canonicalizer20010315.java +367 -0
  75. data/ext/java/nokogiri/internals/c14n/Canonicalizer20010315Excl.java +295 -0
  76. data/ext/java/nokogiri/internals/c14n/Canonicalizer20010315ExclOmitComments.java +40 -0
  77. data/ext/java/nokogiri/internals/c14n/Canonicalizer20010315ExclWithComments.java +44 -0
  78. data/ext/java/nokogiri/internals/c14n/Canonicalizer20010315OmitComments.java +44 -0
  79. data/ext/java/nokogiri/internals/c14n/Canonicalizer20010315WithComments.java +43 -0
  80. data/ext/java/nokogiri/internals/c14n/CanonicalizerBase.java +630 -0
  81. data/ext/java/nokogiri/internals/c14n/CanonicalizerPhysical.java +173 -0
  82. data/ext/java/nokogiri/internals/c14n/CanonicalizerSpi.java +76 -0
  83. data/ext/java/nokogiri/internals/c14n/Constants.java +42 -0
  84. data/ext/java/nokogiri/internals/c14n/ElementProxy.java +293 -0
  85. data/ext/java/nokogiri/internals/c14n/HelperNodeList.java +93 -0
  86. data/ext/java/nokogiri/internals/c14n/IgnoreAllErrorHandler.java +79 -0
  87. data/ext/java/nokogiri/internals/c14n/InclusiveNamespaces.java +166 -0
  88. data/ext/java/nokogiri/internals/c14n/InvalidCanonicalizerException.java +76 -0
  89. data/ext/java/nokogiri/internals/c14n/NameSpaceSymbTable.java +402 -0
  90. data/ext/java/nokogiri/internals/c14n/NodeFilter.java +51 -0
  91. data/ext/java/nokogiri/internals/c14n/UtfHelpper.java +179 -0
  92. data/ext/java/nokogiri/internals/c14n/XMLUtils.java +507 -0
  93. data/ext/java/nokogiri/internals/dom2dtm/DOM2DTM.java +1745 -0
  94. data/ext/java/nokogiri/internals/dom2dtm/DOM2DTMdefaultNamespaceDeclarationNode.java +685 -0
  95. data/ext/nokogiri/depend +477 -0
  96. data/ext/nokogiri/extconf.rb +836 -0
  97. data/ext/nokogiri/html_document.c +171 -0
  98. data/ext/nokogiri/html_document.h +10 -0
  99. data/ext/nokogiri/html_element_description.c +279 -0
  100. data/ext/nokogiri/html_element_description.h +10 -0
  101. data/ext/nokogiri/html_entity_lookup.c +32 -0
  102. data/ext/nokogiri/html_entity_lookup.h +8 -0
  103. data/ext/nokogiri/html_sax_parser_context.c +116 -0
  104. data/ext/nokogiri/html_sax_parser_context.h +11 -0
  105. data/ext/nokogiri/html_sax_push_parser.c +87 -0
  106. data/ext/nokogiri/html_sax_push_parser.h +9 -0
  107. data/ext/nokogiri/nokogiri.c +135 -0
  108. data/ext/nokogiri/nokogiri.h +130 -0
  109. data/ext/nokogiri/xml_attr.c +103 -0
  110. data/ext/nokogiri/xml_attr.h +9 -0
  111. data/ext/nokogiri/xml_attribute_decl.c +70 -0
  112. data/ext/nokogiri/xml_attribute_decl.h +9 -0
  113. data/ext/nokogiri/xml_cdata.c +62 -0
  114. data/ext/nokogiri/xml_cdata.h +9 -0
  115. data/ext/nokogiri/xml_comment.c +69 -0
  116. data/ext/nokogiri/xml_comment.h +9 -0
  117. data/ext/nokogiri/xml_document.c +622 -0
  118. data/ext/nokogiri/xml_document.h +23 -0
  119. data/ext/nokogiri/xml_document_fragment.c +48 -0
  120. data/ext/nokogiri/xml_document_fragment.h +10 -0
  121. data/ext/nokogiri/xml_dtd.c +202 -0
  122. data/ext/nokogiri/xml_dtd.h +10 -0
  123. data/ext/nokogiri/xml_element_content.c +123 -0
  124. data/ext/nokogiri/xml_element_content.h +10 -0
  125. data/ext/nokogiri/xml_element_decl.c +69 -0
  126. data/ext/nokogiri/xml_element_decl.h +9 -0
  127. data/ext/nokogiri/xml_encoding_handler.c +79 -0
  128. data/ext/nokogiri/xml_encoding_handler.h +8 -0
  129. data/ext/nokogiri/xml_entity_decl.c +110 -0
  130. data/ext/nokogiri/xml_entity_decl.h +10 -0
  131. data/ext/nokogiri/xml_entity_reference.c +52 -0
  132. data/ext/nokogiri/xml_entity_reference.h +9 -0
  133. data/ext/nokogiri/xml_io.c +63 -0
  134. data/ext/nokogiri/xml_io.h +11 -0
  135. data/ext/nokogiri/xml_libxml2_hacks.c +112 -0
  136. data/ext/nokogiri/xml_libxml2_hacks.h +12 -0
  137. data/ext/nokogiri/xml_namespace.c +111 -0
  138. data/ext/nokogiri/xml_namespace.h +14 -0
  139. data/ext/nokogiri/xml_node.c +1773 -0
  140. data/ext/nokogiri/xml_node.h +13 -0
  141. data/ext/nokogiri/xml_node_set.c +486 -0
  142. data/ext/nokogiri/xml_node_set.h +12 -0
  143. data/ext/nokogiri/xml_processing_instruction.c +56 -0
  144. data/ext/nokogiri/xml_processing_instruction.h +9 -0
  145. data/ext/nokogiri/xml_reader.c +657 -0
  146. data/ext/nokogiri/xml_reader.h +10 -0
  147. data/ext/nokogiri/xml_relax_ng.c +179 -0
  148. data/ext/nokogiri/xml_relax_ng.h +9 -0
  149. data/ext/nokogiri/xml_sax_parser.c +305 -0
  150. data/ext/nokogiri/xml_sax_parser.h +39 -0
  151. data/ext/nokogiri/xml_sax_parser_context.c +262 -0
  152. data/ext/nokogiri/xml_sax_parser_context.h +10 -0
  153. data/ext/nokogiri/xml_sax_push_parser.c +159 -0
  154. data/ext/nokogiri/xml_sax_push_parser.h +9 -0
  155. data/ext/nokogiri/xml_schema.c +276 -0
  156. data/ext/nokogiri/xml_schema.h +9 -0
  157. data/ext/nokogiri/xml_syntax_error.c +64 -0
  158. data/ext/nokogiri/xml_syntax_error.h +13 -0
  159. data/ext/nokogiri/xml_text.c +52 -0
  160. data/ext/nokogiri/xml_text.h +9 -0
  161. data/ext/nokogiri/xml_xpath_context.c +374 -0
  162. data/ext/nokogiri/xml_xpath_context.h +10 -0
  163. data/ext/nokogiri/xslt_stylesheet.c +263 -0
  164. data/ext/nokogiri/xslt_stylesheet.h +14 -0
  165. data/lib/isorelax.jar +0 -0
  166. data/lib/jing.jar +0 -0
  167. data/lib/nekodtd.jar +0 -0
  168. data/lib/nekohtml.jar +0 -0
  169. data/lib/nokogiri/css/node.rb +53 -0
  170. data/lib/nokogiri/css/parser.rb +751 -0
  171. data/lib/nokogiri/css/parser.y +272 -0
  172. data/lib/nokogiri/css/parser_extras.rb +94 -0
  173. data/lib/nokogiri/css/syntax_error.rb +8 -0
  174. data/lib/nokogiri/css/tokenizer.rb +154 -0
  175. data/lib/nokogiri/css/tokenizer.rex +55 -0
  176. data/lib/nokogiri/css/xpath_visitor.rb +260 -0
  177. data/lib/nokogiri/css.rb +28 -0
  178. data/lib/nokogiri/decorators/slop.rb +43 -0
  179. data/lib/nokogiri/html/builder.rb +36 -0
  180. data/lib/nokogiri/html/document.rb +322 -0
  181. data/lib/nokogiri/html/document_fragment.rb +50 -0
  182. data/lib/nokogiri/html/element_description.rb +24 -0
  183. data/lib/nokogiri/html/element_description_defaults.rb +672 -0
  184. data/lib/nokogiri/html/entity_lookup.rb +14 -0
  185. data/lib/nokogiri/html/sax/parser.rb +63 -0
  186. data/lib/nokogiri/html/sax/parser_context.rb +17 -0
  187. data/lib/nokogiri/html/sax/push_parser.rb +37 -0
  188. data/lib/nokogiri/html.rb +38 -0
  189. data/lib/nokogiri/jruby/dependencies.rb +20 -0
  190. data/lib/nokogiri/syntax_error.rb +5 -0
  191. data/lib/nokogiri/version/constant.rb +5 -0
  192. data/lib/nokogiri/version/info.rb +182 -0
  193. data/lib/nokogiri/version.rb +3 -0
  194. data/lib/nokogiri/xml/attr.rb +15 -0
  195. data/lib/nokogiri/xml/attribute_decl.rb +19 -0
  196. data/lib/nokogiri/xml/builder.rb +447 -0
  197. data/lib/nokogiri/xml/cdata.rb +12 -0
  198. data/lib/nokogiri/xml/character_data.rb +8 -0
  199. data/lib/nokogiri/xml/document.rb +290 -0
  200. data/lib/nokogiri/xml/document_fragment.rb +159 -0
  201. data/lib/nokogiri/xml/dtd.rb +33 -0
  202. data/lib/nokogiri/xml/element_content.rb +37 -0
  203. data/lib/nokogiri/xml/element_decl.rb +14 -0
  204. data/lib/nokogiri/xml/entity_decl.rb +20 -0
  205. data/lib/nokogiri/xml/entity_reference.rb +19 -0
  206. data/lib/nokogiri/xml/namespace.rb +14 -0
  207. data/lib/nokogiri/xml/node/save_options.rb +62 -0
  208. data/lib/nokogiri/xml/node.rb +1240 -0
  209. data/lib/nokogiri/xml/node_set.rb +372 -0
  210. data/lib/nokogiri/xml/notation.rb +7 -0
  211. data/lib/nokogiri/xml/parse_options.rb +127 -0
  212. data/lib/nokogiri/xml/pp/character_data.rb +19 -0
  213. data/lib/nokogiri/xml/pp/node.rb +57 -0
  214. data/lib/nokogiri/xml/pp.rb +3 -0
  215. data/lib/nokogiri/xml/processing_instruction.rb +9 -0
  216. data/lib/nokogiri/xml/reader.rb +116 -0
  217. data/lib/nokogiri/xml/relax_ng.rb +37 -0
  218. data/lib/nokogiri/xml/sax/document.rb +172 -0
  219. data/lib/nokogiri/xml/sax/parser.rb +123 -0
  220. data/lib/nokogiri/xml/sax/parser_context.rb +17 -0
  221. data/lib/nokogiri/xml/sax/push_parser.rb +61 -0
  222. data/lib/nokogiri/xml/sax.rb +5 -0
  223. data/lib/nokogiri/xml/schema.rb +72 -0
  224. data/lib/nokogiri/xml/searchable.rb +239 -0
  225. data/lib/nokogiri/xml/syntax_error.rb +71 -0
  226. data/lib/nokogiri/xml/text.rb +10 -0
  227. data/lib/nokogiri/xml/xpath/syntax_error.rb +12 -0
  228. data/lib/nokogiri/xml/xpath.rb +11 -0
  229. data/lib/nokogiri/xml/xpath_context.rb +17 -0
  230. data/lib/nokogiri/xml.rb +76 -0
  231. data/lib/nokogiri/xslt/stylesheet.rb +26 -0
  232. data/lib/nokogiri/xslt.rb +57 -0
  233. data/lib/nokogiri.rb +144 -0
  234. data/lib/serializer.jar +0 -0
  235. data/lib/xalan.jar +0 -0
  236. data/lib/xercesImpl.jar +0 -0
  237. data/lib/xml-apis.jar +0 -0
  238. data/lib/xsd/xmlparser/nokogiri.rb +103 -0
  239. metadata +531 -0
@@ -0,0 +1,87 @@
1
+ #include <html_sax_push_parser.h>
2
+
3
+ /*
4
+ * call-seq:
5
+ * native_write(chunk, last_chunk)
6
+ *
7
+ * Write +chunk+ to PushParser. +last_chunk+ triggers the end_document handle
8
+ */
9
+ static VALUE native_write(VALUE self, VALUE _chunk, VALUE _last_chunk)
10
+ {
11
+ xmlParserCtxtPtr ctx;
12
+ const char * chunk = NULL;
13
+ int size = 0;
14
+
15
+
16
+ Data_Get_Struct(self, xmlParserCtxt, ctx);
17
+
18
+ if(Qnil != _chunk) {
19
+ chunk = StringValuePtr(_chunk);
20
+ size = (int)RSTRING_LEN(_chunk);
21
+ }
22
+
23
+ if(htmlParseChunk(ctx, chunk, size, Qtrue == _last_chunk ? 1 : 0)) {
24
+ if (!(ctx->options & XML_PARSE_RECOVER)) {
25
+ xmlErrorPtr e = xmlCtxtGetLastError(ctx);
26
+ Nokogiri_error_raise(NULL, e);
27
+ }
28
+ }
29
+
30
+ return self;
31
+ }
32
+
33
+ /*
34
+ * call-seq:
35
+ * initialize_native(xml_sax, filename)
36
+ *
37
+ * Initialize the push parser with +xml_sax+ using +filename+
38
+ */
39
+ static VALUE initialize_native(VALUE self, VALUE _xml_sax, VALUE _filename,
40
+ VALUE encoding)
41
+ {
42
+ htmlSAXHandlerPtr sax;
43
+ const char * filename = NULL;
44
+ htmlParserCtxtPtr ctx;
45
+ xmlCharEncoding enc = XML_CHAR_ENCODING_NONE;
46
+
47
+ Data_Get_Struct(_xml_sax, xmlSAXHandler, sax);
48
+
49
+ if(_filename != Qnil) filename = StringValueCStr(_filename);
50
+
51
+ if (!NIL_P(encoding)) {
52
+ enc = xmlParseCharEncoding(StringValueCStr(encoding));
53
+ if (enc == XML_CHAR_ENCODING_ERROR)
54
+ rb_raise(rb_eArgError, "Unsupported Encoding");
55
+ }
56
+
57
+ ctx = htmlCreatePushParserCtxt(
58
+ sax,
59
+ NULL,
60
+ NULL,
61
+ 0,
62
+ filename,
63
+ enc
64
+ );
65
+ if(ctx == NULL)
66
+ rb_raise(rb_eRuntimeError, "Could not create a parser context");
67
+
68
+ ctx->userData = NOKOGIRI_SAX_TUPLE_NEW(ctx, self);
69
+
70
+ ctx->sax2 = 1;
71
+ DATA_PTR(self) = ctx;
72
+ return self;
73
+ }
74
+
75
+ VALUE cNokogiriHtmlSaxPushParser;
76
+ void init_html_sax_push_parser()
77
+ {
78
+ VALUE nokogiri = rb_define_module("Nokogiri");
79
+ VALUE html = rb_define_module_under(nokogiri, "HTML");
80
+ VALUE sax = rb_define_module_under(html, "SAX");
81
+ VALUE klass = rb_define_class_under(sax, "PushParser", cNokogiriXmlSaxPushParser);
82
+
83
+ cNokogiriHtmlSaxPushParser = klass;
84
+
85
+ rb_define_private_method(klass, "initialize_native", initialize_native, 3);
86
+ rb_define_private_method(klass, "native_write", native_write, 2);
87
+ }
@@ -0,0 +1,9 @@
1
+ #ifndef NOKOGIRI_HTML_SAX_PUSH_PARSER
2
+ #define NOKOGIRI_HTML_SAX_PUSH_PARSER
3
+
4
+ #include <nokogiri.h>
5
+
6
+ void init_html_sax_push_parser();
7
+
8
+ extern VALUE cNokogiriHtmlSaxPushParser ;
9
+ #endif
@@ -0,0 +1,135 @@
1
+ #include <nokogiri.h>
2
+
3
+ VALUE mNokogiri ;
4
+ VALUE mNokogiriXml ;
5
+ VALUE mNokogiriHtml ;
6
+ VALUE mNokogiriXslt ;
7
+ VALUE mNokogiriXmlSax ;
8
+ VALUE mNokogiriHtmlSax ;
9
+
10
+ #ifndef HAVE_VASPRINTF
11
+ /*
12
+ * Thank you Geoffroy Couprie for this implementation of vasprintf!
13
+ */
14
+ int vasprintf(char **strp, const char *fmt, va_list ap)
15
+ {
16
+ /* Mingw32/64 have a broken vsnprintf implementation that fails when
17
+ * using a zero-byte limit in order to retrieve the required size for malloc.
18
+ * So we use a one byte buffer instead.
19
+ */
20
+ char tmp[1];
21
+ int len = vsnprintf (tmp, 1, fmt, ap) + 1;
22
+ char *res = (char *)malloc((unsigned int)len);
23
+ if (res == NULL) {
24
+ return -1;
25
+ }
26
+ *strp = res;
27
+ return vsnprintf(res, (unsigned int)len, fmt, ap);
28
+ }
29
+ #endif
30
+
31
+ #include "ruby/util.h"
32
+
33
+ void nokogiri_root_node(xmlNodePtr node)
34
+ {
35
+ xmlDocPtr doc;
36
+ nokogiriTuplePtr tuple;
37
+
38
+ doc = node->doc;
39
+ if (doc->type == XML_DOCUMENT_FRAG_NODE) { doc = doc->doc; }
40
+ tuple = (nokogiriTuplePtr)doc->_private;
41
+ st_insert(tuple->unlinkedNodes, (st_data_t)node, (st_data_t)node);
42
+ }
43
+
44
+ void nokogiri_root_nsdef(xmlNsPtr ns, xmlDocPtr doc)
45
+ {
46
+ nokogiriTuplePtr tuple;
47
+
48
+ if (doc->type == XML_DOCUMENT_FRAG_NODE) { doc = doc->doc; }
49
+ tuple = (nokogiriTuplePtr)doc->_private;
50
+ st_insert(tuple->unlinkedNodes, (st_data_t)ns, (st_data_t)ns);
51
+ }
52
+
53
+ void Init_nokogiri()
54
+ {
55
+ xmlMemSetup(
56
+ (xmlFreeFunc)ruby_xfree,
57
+ (xmlMallocFunc)ruby_xmalloc,
58
+ (xmlReallocFunc)ruby_xrealloc,
59
+ ruby_strdup
60
+ );
61
+
62
+ mNokogiri = rb_define_module("Nokogiri");
63
+ mNokogiriXml = rb_define_module_under(mNokogiri, "XML");
64
+ mNokogiriHtml = rb_define_module_under(mNokogiri, "HTML");
65
+ mNokogiriXslt = rb_define_module_under(mNokogiri, "XSLT");
66
+ mNokogiriXmlSax = rb_define_module_under(mNokogiriXml, "SAX");
67
+ mNokogiriHtmlSax = rb_define_module_under(mNokogiriHtml, "SAX");
68
+
69
+ rb_const_set(mNokogiri, rb_intern("LIBXML_COMPILED_VERSION"), NOKOGIRI_STR_NEW2(LIBXML_DOTTED_VERSION));
70
+ rb_const_set(mNokogiri, rb_intern("LIBXML_LOADED_VERSION"), NOKOGIRI_STR_NEW2(xmlParserVersion));
71
+
72
+ rb_const_set(mNokogiri, rb_intern("LIBXSLT_COMPILED_VERSION"), NOKOGIRI_STR_NEW2(LIBXSLT_DOTTED_VERSION));
73
+ rb_const_set(mNokogiri, rb_intern("LIBXSLT_LOADED_VERSION"), NOKOGIRI_STR_NEW2(xsltEngineVersion));
74
+
75
+ #ifdef NOKOGIRI_PACKAGED_LIBRARIES
76
+ rb_const_set(mNokogiri, rb_intern("PACKAGED_LIBRARIES"), Qtrue);
77
+ #ifdef NOKOGIRI_PRECOMPILED_LIBRARIES
78
+ rb_const_set(mNokogiri, rb_intern("PRECOMPILED_LIBRARIES"), Qtrue);
79
+ #else
80
+ rb_const_set(mNokogiri, rb_intern("PRECOMPILED_LIBRARIES"), Qfalse);
81
+ #endif
82
+ rb_const_set(mNokogiri, rb_intern("LIBXML2_PATCHES"), rb_str_split(NOKOGIRI_STR_NEW2(NOKOGIRI_LIBXML2_PATCHES), " "));
83
+ rb_const_set(mNokogiri, rb_intern("LIBXSLT_PATCHES"), rb_str_split(NOKOGIRI_STR_NEW2(NOKOGIRI_LIBXSLT_PATCHES), " "));
84
+ #else
85
+ rb_const_set(mNokogiri, rb_intern("PACKAGED_LIBRARIES"), Qfalse);
86
+ rb_const_set(mNokogiri, rb_intern("PRECOMPILED_LIBRARIES"), Qfalse);
87
+ rb_const_set(mNokogiri, rb_intern("LIBXML2_PATCHES"), Qnil);
88
+ rb_const_set(mNokogiri, rb_intern("LIBXSLT_PATCHES"), Qnil);
89
+ #endif
90
+
91
+ #ifdef LIBXML_ICONV_ENABLED
92
+ rb_const_set(mNokogiri, rb_intern("LIBXML_ICONV_ENABLED"), Qtrue);
93
+ #else
94
+ rb_const_set(mNokogiri, rb_intern("LIBXML_ICONV_ENABLED"), Qfalse);
95
+ #endif
96
+
97
+ #ifdef NOKOGIRI_OTHER_LIBRARY_VERSIONS
98
+ rb_const_set(mNokogiri, rb_intern("OTHER_LIBRARY_VERSIONS"), NOKOGIRI_STR_NEW2(NOKOGIRI_OTHER_LIBRARY_VERSIONS));
99
+ #endif
100
+
101
+ xmlInitParser();
102
+
103
+ init_xml_document();
104
+ init_html_document();
105
+ init_xml_node();
106
+ init_xml_document_fragment();
107
+ init_xml_text();
108
+ init_xml_cdata();
109
+ init_xml_processing_instruction();
110
+ init_xml_attr();
111
+ init_xml_entity_reference();
112
+ init_xml_comment();
113
+ init_xml_node_set();
114
+ init_xml_xpath_context();
115
+ init_xml_sax_parser_context();
116
+ init_xml_sax_parser();
117
+ init_xml_sax_push_parser();
118
+ init_xml_reader();
119
+ init_xml_dtd();
120
+ init_xml_element_content();
121
+ init_xml_attribute_decl();
122
+ init_xml_element_decl();
123
+ init_xml_entity_decl();
124
+ init_xml_namespace();
125
+ init_html_sax_parser_context();
126
+ init_html_sax_push_parser();
127
+ init_xslt_stylesheet();
128
+ init_xml_syntax_error();
129
+ init_html_entity_lookup();
130
+ init_html_element_description();
131
+ init_xml_schema();
132
+ init_xml_relax_ng();
133
+ init_nokogiri_io();
134
+ init_xml_encoding_handler();
135
+ }
@@ -0,0 +1,130 @@
1
+ #ifndef NOKOGIRI_NATIVE
2
+ #define NOKOGIRI_NATIVE
3
+
4
+ #if _MSC_VER
5
+ # ifndef WIN32_LEAN_AND_MEAN
6
+ # define WIN32_LEAN_AND_MEAN
7
+ # endif /* WIN32_LEAN_AND_MEAN */
8
+
9
+ # ifndef WIN32
10
+ # define WIN32
11
+ # endif /* WIN32 */
12
+
13
+ # include <winsock2.h>
14
+ # include <ws2tcpip.h>
15
+ # include <windows.h>
16
+ #endif
17
+
18
+ #include <stdlib.h>
19
+ #include <string.h>
20
+ #include <assert.h>
21
+ #include <stdarg.h>
22
+ #include <stdio.h>
23
+
24
+ #include <libxml/parser.h>
25
+ #include <libxml/entities.h>
26
+ #include <libxml/parserInternals.h>
27
+ #include <libxml/xpath.h>
28
+ #include <libxml/xpathInternals.h>
29
+ #include <libxml/xmlreader.h>
30
+ #include <libxml/xmlsave.h>
31
+ #include <libxml/xmlschemas.h>
32
+ #include <libxml/HTMLparser.h>
33
+ #include <libxml/HTMLtree.h>
34
+ #include <libxml/relaxng.h>
35
+ #include <libxml/xinclude.h>
36
+ #include <libxslt/extensions.h>
37
+ #include <libxslt/xsltconfig.h>
38
+ #include <libxml/c14n.h>
39
+
40
+ #include <ruby.h>
41
+ #include <ruby/st.h>
42
+ #include <ruby/encoding.h>
43
+
44
+ #ifndef NORETURN
45
+ # if defined(__GNUC__)
46
+ # define NORETURN(name) __attribute__((noreturn)) name
47
+ # else
48
+ # define NORETURN(name) name
49
+ # endif
50
+ #endif
51
+
52
+ #define NOKOGIRI_STR_NEW2(str) \
53
+ NOKOGIRI_STR_NEW(str, strlen((const char *)(str)))
54
+
55
+ #define NOKOGIRI_STR_NEW(str, len) \
56
+ rb_external_str_new_with_enc((const char *)(str), (long)(len), rb_utf8_encoding())
57
+
58
+ #define RBSTR_OR_QNIL(_str) \
59
+ (_str ? NOKOGIRI_STR_NEW2(_str) : Qnil)
60
+
61
+ #include <xml_libxml2_hacks.h>
62
+
63
+ #include <xml_io.h>
64
+ #include <xml_document.h>
65
+ #include <html_entity_lookup.h>
66
+ #include <html_document.h>
67
+ #include <xml_node.h>
68
+ #include <xml_text.h>
69
+ #include <xml_cdata.h>
70
+ #include <xml_attr.h>
71
+ #include <xml_processing_instruction.h>
72
+ #include <xml_entity_reference.h>
73
+ #include <xml_document_fragment.h>
74
+ #include <xml_comment.h>
75
+ #include <xml_node_set.h>
76
+ #include <xml_dtd.h>
77
+ #include <xml_attribute_decl.h>
78
+ #include <xml_element_decl.h>
79
+ #include <xml_entity_decl.h>
80
+ #include <xml_xpath_context.h>
81
+ #include <xml_element_content.h>
82
+ #include <xml_sax_parser_context.h>
83
+ #include <xml_sax_parser.h>
84
+ #include <xml_sax_push_parser.h>
85
+ #include <xml_reader.h>
86
+ #include <html_sax_parser_context.h>
87
+ #include <html_sax_push_parser.h>
88
+ #include <xslt_stylesheet.h>
89
+ #include <xml_syntax_error.h>
90
+ #include <xml_schema.h>
91
+ #include <xml_relax_ng.h>
92
+ #include <html_element_description.h>
93
+ #include <xml_namespace.h>
94
+ #include <xml_encoding_handler.h>
95
+
96
+ extern VALUE mNokogiri ;
97
+ extern VALUE mNokogiriXml ;
98
+ extern VALUE mNokogiriXmlSax ;
99
+ extern VALUE mNokogiriHtml ;
100
+ extern VALUE mNokogiriHtmlSax ;
101
+ extern VALUE mNokogiriXslt ;
102
+
103
+ int vasprintf(char **strp, const char *fmt, va_list ap);
104
+
105
+ void nokogiri_root_node(xmlNodePtr);
106
+ void nokogiri_root_nsdef(xmlNsPtr, xmlDocPtr);
107
+
108
+ #ifdef DEBUG
109
+
110
+ # define NOKOGIRI_DEBUG_START(p) if (getenv("NOKOGIRI_NO_FREE")) return ; if (getenv("NOKOGIRI_DEBUG")) fprintf(stderr,"nokogiri: %s:%d %p start\n", __FILE__, __LINE__, p);
111
+ # define NOKOGIRI_DEBUG_END(p) if (getenv("NOKOGIRI_DEBUG")) fprintf(stderr,"nokogiri: %s:%d %p end\n", __FILE__, __LINE__, p);
112
+
113
+ #else
114
+
115
+ # define NOKOGIRI_DEBUG_START(p)
116
+ # define NOKOGIRI_DEBUG_END(p)
117
+
118
+ #endif
119
+
120
+ #ifndef __builtin_expect
121
+ # if defined(__GNUC__)
122
+ # define __builtin_expect(expr, c) __builtin_expect((long)(expr), (long)(c))
123
+ # endif
124
+ #endif
125
+
126
+ #define XMLNS_PREFIX "xmlns"
127
+ #define XMLNS_PREFIX_LEN 6 /* including either colon or \0 */
128
+ #define XMLNS_BUFFER_LEN 128
129
+
130
+ #endif
@@ -0,0 +1,103 @@
1
+ #include <xml_attr.h>
2
+
3
+ /*
4
+ * call-seq:
5
+ * value=(content)
6
+ *
7
+ * Set the value for this Attr to +content+. Use `nil` to remove the value
8
+ * (e.g., a HTML boolean attribute).
9
+ */
10
+ static VALUE set_value(VALUE self, VALUE content)
11
+ {
12
+ xmlAttrPtr attr;
13
+ xmlChar *value;
14
+ xmlNode *cur;
15
+
16
+ Data_Get_Struct(self, xmlAttr, attr);
17
+
18
+ if (attr->children) {
19
+ xmlFreeNodeList(attr->children);
20
+ }
21
+ attr->children = attr->last = NULL;
22
+
23
+ if (content == Qnil) {
24
+ return content;
25
+ }
26
+
27
+ value = xmlEncodeEntitiesReentrant(attr->doc, (unsigned char *)StringValueCStr(content));
28
+ if (xmlStrlen(value) == 0) {
29
+ attr->children = xmlNewDocText(attr->doc, value);
30
+ } else {
31
+ attr->children = xmlStringGetNodeList(attr->doc, value);
32
+ }
33
+ xmlFree(value);
34
+
35
+ for (cur = attr->children; cur; cur = cur->next) {
36
+ cur->parent = (xmlNode *)attr;
37
+ cur->doc = attr->doc;
38
+ if (cur->next == NULL) {
39
+ attr->last = cur;
40
+ }
41
+ }
42
+
43
+ return content;
44
+ }
45
+
46
+ /*
47
+ * call-seq:
48
+ * new(document, name)
49
+ *
50
+ * Create a new Attr element on the +document+ with +name+
51
+ */
52
+ static VALUE new(int argc, VALUE *argv, VALUE klass)
53
+ {
54
+ xmlDocPtr xml_doc;
55
+ VALUE document;
56
+ VALUE name;
57
+ VALUE rest;
58
+ xmlAttrPtr node;
59
+ VALUE rb_node;
60
+
61
+ rb_scan_args(argc, argv, "2*", &document, &name, &rest);
62
+
63
+ if (! rb_obj_is_kind_of(document, cNokogiriXmlDocument)) {
64
+ rb_raise(rb_eArgError, "parameter must be a Nokogiri::XML::Document");
65
+ }
66
+
67
+ Data_Get_Struct(document, xmlDoc, xml_doc);
68
+
69
+ node = xmlNewDocProp(
70
+ xml_doc,
71
+ (const xmlChar *)StringValueCStr(name),
72
+ NULL
73
+ );
74
+
75
+ nokogiri_root_node((xmlNodePtr)node);
76
+
77
+ rb_node = Nokogiri_wrap_xml_node(klass, (xmlNodePtr)node);
78
+ rb_obj_call_init(rb_node, argc, argv);
79
+
80
+ if (rb_block_given_p()) {
81
+ rb_yield(rb_node);
82
+ }
83
+
84
+ return rb_node;
85
+ }
86
+
87
+ VALUE cNokogiriXmlAttr;
88
+ void init_xml_attr()
89
+ {
90
+ VALUE nokogiri = rb_define_module("Nokogiri");
91
+ VALUE xml = rb_define_module_under(nokogiri, "XML");
92
+ VALUE node = rb_define_class_under(xml, "Node", rb_cObject);
93
+
94
+ /*
95
+ * Attr represents a Attr node in an xml document.
96
+ */
97
+ VALUE klass = rb_define_class_under(xml, "Attr", node);
98
+
99
+ cNokogiriXmlAttr = klass;
100
+
101
+ rb_define_singleton_method(klass, "new", new, -1);
102
+ rb_define_method(klass, "value=", set_value, 1);
103
+ }
@@ -0,0 +1,9 @@
1
+ #ifndef NOKOGIRI_XML_ATTR
2
+ #define NOKOGIRI_XML_ATTR
3
+
4
+ #include <nokogiri.h>
5
+
6
+ void init_xml_attr();
7
+
8
+ extern VALUE cNokogiriXmlAttr;
9
+ #endif
@@ -0,0 +1,70 @@
1
+ #include <xml_attribute_decl.h>
2
+
3
+ /*
4
+ * call-seq:
5
+ * attribute_type
6
+ *
7
+ * The attribute_type for this AttributeDecl
8
+ */
9
+ static VALUE attribute_type(VALUE self)
10
+ {
11
+ xmlAttributePtr node;
12
+ Data_Get_Struct(self, xmlAttribute, node);
13
+ return INT2NUM((long)node->atype);
14
+ }
15
+
16
+ /*
17
+ * call-seq:
18
+ * default
19
+ *
20
+ * The default value
21
+ */
22
+ static VALUE default_value(VALUE self)
23
+ {
24
+ xmlAttributePtr node;
25
+ Data_Get_Struct(self, xmlAttribute, node);
26
+
27
+ if(node->defaultValue) return NOKOGIRI_STR_NEW2(node->defaultValue);
28
+ return Qnil;
29
+ }
30
+
31
+ /*
32
+ * call-seq:
33
+ * enumeration
34
+ *
35
+ * An enumeration of possible values
36
+ */
37
+ static VALUE enumeration(VALUE self)
38
+ {
39
+ xmlAttributePtr node;
40
+ xmlEnumerationPtr enm;
41
+ VALUE list;
42
+
43
+ Data_Get_Struct(self, xmlAttribute, node);
44
+
45
+ list = rb_ary_new();
46
+ enm = node->tree;
47
+
48
+ while(enm) {
49
+ rb_ary_push(list, NOKOGIRI_STR_NEW2(enm->name));
50
+ enm = enm->next;
51
+ }
52
+
53
+ return list;
54
+ }
55
+
56
+ VALUE cNokogiriXmlAttributeDecl;
57
+
58
+ void init_xml_attribute_decl()
59
+ {
60
+ VALUE nokogiri = rb_define_module("Nokogiri");
61
+ VALUE xml = rb_define_module_under(nokogiri, "XML");
62
+ VALUE node = rb_define_class_under(xml, "Node", rb_cObject);
63
+ VALUE klass = rb_define_class_under(xml, "AttributeDecl", node);
64
+
65
+ cNokogiriXmlAttributeDecl = klass;
66
+
67
+ rb_define_method(klass, "attribute_type", attribute_type, 0);
68
+ rb_define_method(klass, "default", default_value, 0);
69
+ rb_define_method(klass, "enumeration", enumeration, 0);
70
+ }
@@ -0,0 +1,9 @@
1
+ #ifndef NOKOGIRI_XML_ATTRIBUTE_DECL
2
+ #define NOKOGIRI_XML_ATTRIBUTE_DECL
3
+
4
+ #include <nokogiri.h>
5
+
6
+ void init_xml_attribute_decl();
7
+
8
+ extern VALUE cNokogiriXmlAttributeDecl;
9
+ #endif
@@ -0,0 +1,62 @@
1
+ #include <xml_cdata.h>
2
+
3
+ /*
4
+ * call-seq:
5
+ * new(document, content)
6
+ *
7
+ * Create a new CDATA element on the +document+ with +content+
8
+ *
9
+ * If +content+ cannot be implicitly converted to a string, this method will
10
+ * raise a TypeError exception.
11
+ */
12
+ static VALUE new(int argc, VALUE *argv, VALUE klass)
13
+ {
14
+ xmlDocPtr xml_doc;
15
+ xmlNodePtr node;
16
+ VALUE doc;
17
+ VALUE content;
18
+ VALUE rest;
19
+ VALUE rb_node;
20
+ xmlChar *content_str = NULL;
21
+ int content_str_len = 0;
22
+
23
+ rb_scan_args(argc, argv, "2*", &doc, &content, &rest);
24
+
25
+ Data_Get_Struct(doc, xmlDoc, xml_doc);
26
+
27
+ if (!NIL_P(content)) {
28
+ content_str = (xmlChar *)StringValuePtr(content);
29
+ content_str_len = RSTRING_LEN(content);
30
+ }
31
+
32
+ node = xmlNewCDataBlock(xml_doc->doc, content_str, content_str_len);
33
+
34
+ nokogiri_root_node(node);
35
+
36
+ rb_node = Nokogiri_wrap_xml_node(klass, node);
37
+ rb_obj_call_init(rb_node, argc, argv);
38
+
39
+ if(rb_block_given_p()) { rb_yield(rb_node); }
40
+
41
+ return rb_node;
42
+ }
43
+
44
+ VALUE cNokogiriXmlCData;
45
+ void init_xml_cdata()
46
+ {
47
+ VALUE nokogiri = rb_define_module("Nokogiri");
48
+ VALUE xml = rb_define_module_under(nokogiri, "XML");
49
+ VALUE node = rb_define_class_under(xml, "Node", rb_cObject);
50
+ VALUE char_data = rb_define_class_under(xml, "CharacterData", node);
51
+ VALUE text = rb_define_class_under(xml, "Text", char_data);
52
+
53
+ /*
54
+ * CData represents a CData node in an xml document.
55
+ */
56
+ VALUE klass = rb_define_class_under(xml, "CDATA", text);
57
+
58
+
59
+ cNokogiriXmlCData = klass;
60
+
61
+ rb_define_singleton_method(klass, "new", new, -1);
62
+ }
@@ -0,0 +1,9 @@
1
+ #ifndef NOKOGIRI_XML_CDATA
2
+ #define NOKOGIRI_XML_CDATA
3
+
4
+ #include <nokogiri.h>
5
+
6
+ void init_xml_cdata();
7
+
8
+ extern VALUE cNokogiriXmlCData;
9
+ #endif
@@ -0,0 +1,69 @@
1
+ #include <xml_comment.h>
2
+
3
+ static ID document_id ;
4
+
5
+ /*
6
+ * call-seq:
7
+ * new(document_or_node, content)
8
+ *
9
+ * Create a new Comment element on the +document+ with +content+.
10
+ * Alternatively, if a +node+ is passed, the +node+'s document is used.
11
+ */
12
+ static VALUE new(int argc, VALUE *argv, VALUE klass)
13
+ {
14
+ xmlDocPtr xml_doc;
15
+ xmlNodePtr node;
16
+ VALUE document;
17
+ VALUE content;
18
+ VALUE rest;
19
+ VALUE rb_node;
20
+
21
+ rb_scan_args(argc, argv, "2*", &document, &content, &rest);
22
+
23
+ if (rb_obj_is_kind_of(document, cNokogiriXmlNode))
24
+ {
25
+ document = rb_funcall(document, document_id, 0);
26
+ }
27
+ else if ( !rb_obj_is_kind_of(document, cNokogiriXmlDocument)
28
+ && !rb_obj_is_kind_of(document, cNokogiriXmlDocumentFragment))
29
+ {
30
+ rb_raise(rb_eArgError, "first argument must be a XML::Document or XML::Node");
31
+ }
32
+
33
+ Data_Get_Struct(document, xmlDoc, xml_doc);
34
+
35
+ node = xmlNewDocComment(
36
+ xml_doc,
37
+ (const xmlChar *)StringValueCStr(content)
38
+ );
39
+
40
+ rb_node = Nokogiri_wrap_xml_node(klass, node);
41
+ rb_obj_call_init(rb_node, argc, argv);
42
+
43
+ nokogiri_root_node(node);
44
+
45
+ if(rb_block_given_p()) rb_yield(rb_node);
46
+
47
+ return rb_node;
48
+ }
49
+
50
+ VALUE cNokogiriXmlComment;
51
+ void init_xml_comment()
52
+ {
53
+ VALUE nokogiri = rb_define_module("Nokogiri");
54
+ VALUE xml = rb_define_module_under(nokogiri, "XML");
55
+ VALUE node = rb_define_class_under(xml, "Node", rb_cObject);
56
+ VALUE char_data = rb_define_class_under(xml, "CharacterData", node);
57
+
58
+ /*
59
+ * Comment represents a comment node in an xml document.
60
+ */
61
+ VALUE klass = rb_define_class_under(xml, "Comment", char_data);
62
+
63
+
64
+ cNokogiriXmlComment = klass;
65
+
66
+ rb_define_singleton_method(klass, "new", new, -1);
67
+
68
+ document_id = rb_intern("document");
69
+ }
@@ -0,0 +1,9 @@
1
+ #ifndef NOKOGIRI_XML_COMMENT
2
+ #define NOKOGIRI_XML_COMMENT
3
+
4
+ #include <nokogiri.h>
5
+
6
+ void init_xml_comment();
7
+
8
+ extern VALUE cNokogiriXmlComment;
9
+ #endif