nokogiri-backupify 1.5.0.beta.4

Sign up to get free protection for your applications and to get access to all the features.
Files changed (275) hide show
  1. data/.autotest +26 -0
  2. data/CHANGELOG.ja.rdoc +509 -0
  3. data/CHANGELOG.rdoc +490 -0
  4. data/Manifest.txt +274 -0
  5. data/README.ja.rdoc +106 -0
  6. data/README.rdoc +150 -0
  7. data/Rakefile +217 -0
  8. data/bin/nokogiri +54 -0
  9. data/deps.rip +5 -0
  10. data/ext/java/nokogiri/EncodingHandler.java +124 -0
  11. data/ext/java/nokogiri/HtmlDocument.java +146 -0
  12. data/ext/java/nokogiri/HtmlElementDescription.java +145 -0
  13. data/ext/java/nokogiri/HtmlEntityLookup.java +79 -0
  14. data/ext/java/nokogiri/HtmlSaxParserContext.java +256 -0
  15. data/ext/java/nokogiri/NokogiriService.java +466 -0
  16. data/ext/java/nokogiri/XmlAttr.java +183 -0
  17. data/ext/java/nokogiri/XmlAttributeDecl.java +130 -0
  18. data/ext/java/nokogiri/XmlCdata.java +89 -0
  19. data/ext/java/nokogiri/XmlComment.java +84 -0
  20. data/ext/java/nokogiri/XmlDocument.java +514 -0
  21. data/ext/java/nokogiri/XmlDocumentFragment.java +216 -0
  22. data/ext/java/nokogiri/XmlDtd.java +464 -0
  23. data/ext/java/nokogiri/XmlElement.java +221 -0
  24. data/ext/java/nokogiri/XmlElementContent.java +382 -0
  25. data/ext/java/nokogiri/XmlElementDecl.java +147 -0
  26. data/ext/java/nokogiri/XmlEntityDecl.java +161 -0
  27. data/ext/java/nokogiri/XmlEntityReference.java +75 -0
  28. data/ext/java/nokogiri/XmlNamespace.java +127 -0
  29. data/ext/java/nokogiri/XmlNode.java +1392 -0
  30. data/ext/java/nokogiri/XmlNodeSet.java +284 -0
  31. data/ext/java/nokogiri/XmlProcessingInstruction.java +103 -0
  32. data/ext/java/nokogiri/XmlReader.java +409 -0
  33. data/ext/java/nokogiri/XmlRelaxng.java +199 -0
  34. data/ext/java/nokogiri/XmlSaxParserContext.java +353 -0
  35. data/ext/java/nokogiri/XmlSaxPushParser.java +182 -0
  36. data/ext/java/nokogiri/XmlSchema.java +175 -0
  37. data/ext/java/nokogiri/XmlSyntaxError.java +114 -0
  38. data/ext/java/nokogiri/XmlText.java +135 -0
  39. data/ext/java/nokogiri/XmlXpathContext.java +175 -0
  40. data/ext/java/nokogiri/XsltStylesheet.java +181 -0
  41. data/ext/java/nokogiri/internals/HtmlDomParserContext.java +205 -0
  42. data/ext/java/nokogiri/internals/NokogiriDocumentCache.java +73 -0
  43. data/ext/java/nokogiri/internals/NokogiriErrorHandler.java +80 -0
  44. data/ext/java/nokogiri/internals/NokogiriHandler.java +326 -0
  45. data/ext/java/nokogiri/internals/NokogiriHelpers.java +583 -0
  46. data/ext/java/nokogiri/internals/NokogiriNamespaceCache.java +170 -0
  47. data/ext/java/nokogiri/internals/NokogiriNamespaceContext.java +118 -0
  48. data/ext/java/nokogiri/internals/NokogiriNonStrictErrorHandler.java +73 -0
  49. data/ext/java/nokogiri/internals/NokogiriNonStrictErrorHandler4NekoHtml.java +121 -0
  50. data/ext/java/nokogiri/internals/NokogiriStrictErrorHandler.java +78 -0
  51. data/ext/java/nokogiri/internals/NokogiriXPathFunction.java +120 -0
  52. data/ext/java/nokogiri/internals/NokogiriXPathFunctionResolver.java +56 -0
  53. data/ext/java/nokogiri/internals/ParserContext.java +278 -0
  54. data/ext/java/nokogiri/internals/PushInputStream.java +411 -0
  55. data/ext/java/nokogiri/internals/ReaderNode.java +473 -0
  56. data/ext/java/nokogiri/internals/SaveContext.java +282 -0
  57. data/ext/java/nokogiri/internals/SchemaErrorHandler.java +68 -0
  58. data/ext/java/nokogiri/internals/XmlDeclHandler.java +42 -0
  59. data/ext/java/nokogiri/internals/XmlDomParser.java +77 -0
  60. data/ext/java/nokogiri/internals/XmlDomParserContext.java +233 -0
  61. data/ext/java/nokogiri/internals/XmlSaxParser.java +65 -0
  62. data/ext/java/nokogiri/internals/XsltExtensionFunction.java +72 -0
  63. data/ext/nokogiri/depend +358 -0
  64. data/ext/nokogiri/extconf.rb +124 -0
  65. data/ext/nokogiri/html_document.c +154 -0
  66. data/ext/nokogiri/html_document.h +10 -0
  67. data/ext/nokogiri/html_element_description.c +276 -0
  68. data/ext/nokogiri/html_element_description.h +10 -0
  69. data/ext/nokogiri/html_entity_lookup.c +32 -0
  70. data/ext/nokogiri/html_entity_lookup.h +8 -0
  71. data/ext/nokogiri/html_sax_parser_context.c +94 -0
  72. data/ext/nokogiri/html_sax_parser_context.h +11 -0
  73. data/ext/nokogiri/nokogiri.c +92 -0
  74. data/ext/nokogiri/nokogiri.h +160 -0
  75. data/ext/nokogiri/xml_attr.c +94 -0
  76. data/ext/nokogiri/xml_attr.h +9 -0
  77. data/ext/nokogiri/xml_attribute_decl.c +70 -0
  78. data/ext/nokogiri/xml_attribute_decl.h +9 -0
  79. data/ext/nokogiri/xml_cdata.c +56 -0
  80. data/ext/nokogiri/xml_cdata.h +9 -0
  81. data/ext/nokogiri/xml_comment.c +54 -0
  82. data/ext/nokogiri/xml_comment.h +9 -0
  83. data/ext/nokogiri/xml_document.c +478 -0
  84. data/ext/nokogiri/xml_document.h +23 -0
  85. data/ext/nokogiri/xml_document_fragment.c +48 -0
  86. data/ext/nokogiri/xml_document_fragment.h +10 -0
  87. data/ext/nokogiri/xml_dtd.c +202 -0
  88. data/ext/nokogiri/xml_dtd.h +10 -0
  89. data/ext/nokogiri/xml_element_content.c +123 -0
  90. data/ext/nokogiri/xml_element_content.h +10 -0
  91. data/ext/nokogiri/xml_element_decl.c +69 -0
  92. data/ext/nokogiri/xml_element_decl.h +9 -0
  93. data/ext/nokogiri/xml_encoding_handler.c +79 -0
  94. data/ext/nokogiri/xml_encoding_handler.h +8 -0
  95. data/ext/nokogiri/xml_entity_decl.c +110 -0
  96. data/ext/nokogiri/xml_entity_decl.h +10 -0
  97. data/ext/nokogiri/xml_entity_reference.c +52 -0
  98. data/ext/nokogiri/xml_entity_reference.h +9 -0
  99. data/ext/nokogiri/xml_io.c +31 -0
  100. data/ext/nokogiri/xml_io.h +11 -0
  101. data/ext/nokogiri/xml_libxml2_hacks.c +112 -0
  102. data/ext/nokogiri/xml_libxml2_hacks.h +12 -0
  103. data/ext/nokogiri/xml_namespace.c +84 -0
  104. data/ext/nokogiri/xml_namespace.h +13 -0
  105. data/ext/nokogiri/xml_node.c +1384 -0
  106. data/ext/nokogiri/xml_node.h +13 -0
  107. data/ext/nokogiri/xml_node_set.c +418 -0
  108. data/ext/nokogiri/xml_node_set.h +9 -0
  109. data/ext/nokogiri/xml_processing_instruction.c +56 -0
  110. data/ext/nokogiri/xml_processing_instruction.h +9 -0
  111. data/ext/nokogiri/xml_reader.c +684 -0
  112. data/ext/nokogiri/xml_reader.h +10 -0
  113. data/ext/nokogiri/xml_relax_ng.c +161 -0
  114. data/ext/nokogiri/xml_relax_ng.h +9 -0
  115. data/ext/nokogiri/xml_sax_parser.c +288 -0
  116. data/ext/nokogiri/xml_sax_parser.h +39 -0
  117. data/ext/nokogiri/xml_sax_parser_context.c +199 -0
  118. data/ext/nokogiri/xml_sax_parser_context.h +10 -0
  119. data/ext/nokogiri/xml_sax_push_parser.c +115 -0
  120. data/ext/nokogiri/xml_sax_push_parser.h +9 -0
  121. data/ext/nokogiri/xml_schema.c +205 -0
  122. data/ext/nokogiri/xml_schema.h +9 -0
  123. data/ext/nokogiri/xml_syntax_error.c +58 -0
  124. data/ext/nokogiri/xml_syntax_error.h +13 -0
  125. data/ext/nokogiri/xml_text.c +50 -0
  126. data/ext/nokogiri/xml_text.h +9 -0
  127. data/ext/nokogiri/xml_xpath_context.c +309 -0
  128. data/ext/nokogiri/xml_xpath_context.h +9 -0
  129. data/ext/nokogiri/xslt_stylesheet.c +258 -0
  130. data/ext/nokogiri/xslt_stylesheet.h +9 -0
  131. data/lib/isorelax.jar +0 -0
  132. data/lib/jing.jar +0 -0
  133. data/lib/nekodtd.jar +0 -0
  134. data/lib/nekohtml.jar +0 -0
  135. data/lib/nokogiri.rb +143 -0
  136. data/lib/nokogiri/css.rb +23 -0
  137. data/lib/nokogiri/css/node.rb +99 -0
  138. data/lib/nokogiri/css/parser.rb +677 -0
  139. data/lib/nokogiri/css/parser.y +237 -0
  140. data/lib/nokogiri/css/parser_extras.rb +91 -0
  141. data/lib/nokogiri/css/syntax_error.rb +7 -0
  142. data/lib/nokogiri/css/tokenizer.rb +152 -0
  143. data/lib/nokogiri/css/tokenizer.rex +55 -0
  144. data/lib/nokogiri/css/xpath_visitor.rb +171 -0
  145. data/lib/nokogiri/decorators/slop.rb +35 -0
  146. data/lib/nokogiri/html.rb +36 -0
  147. data/lib/nokogiri/html/builder.rb +35 -0
  148. data/lib/nokogiri/html/document.rb +221 -0
  149. data/lib/nokogiri/html/document_fragment.rb +41 -0
  150. data/lib/nokogiri/html/element_description.rb +23 -0
  151. data/lib/nokogiri/html/element_description_defaults.rb +671 -0
  152. data/lib/nokogiri/html/entity_lookup.rb +13 -0
  153. data/lib/nokogiri/html/sax/parser.rb +52 -0
  154. data/lib/nokogiri/html/sax/parser_context.rb +16 -0
  155. data/lib/nokogiri/syntax_error.rb +4 -0
  156. data/lib/nokogiri/version.rb +35 -0
  157. data/lib/nokogiri/xml.rb +67 -0
  158. data/lib/nokogiri/xml/attr.rb +14 -0
  159. data/lib/nokogiri/xml/attribute_decl.rb +18 -0
  160. data/lib/nokogiri/xml/builder.rb +418 -0
  161. data/lib/nokogiri/xml/cdata.rb +11 -0
  162. data/lib/nokogiri/xml/character_data.rb +7 -0
  163. data/lib/nokogiri/xml/document.rb +218 -0
  164. data/lib/nokogiri/xml/document_fragment.rb +84 -0
  165. data/lib/nokogiri/xml/dtd.rb +22 -0
  166. data/lib/nokogiri/xml/element_content.rb +36 -0
  167. data/lib/nokogiri/xml/element_decl.rb +13 -0
  168. data/lib/nokogiri/xml/entity_decl.rb +19 -0
  169. data/lib/nokogiri/xml/namespace.rb +13 -0
  170. data/lib/nokogiri/xml/node.rb +907 -0
  171. data/lib/nokogiri/xml/node/save_options.rb +45 -0
  172. data/lib/nokogiri/xml/node_set.rb +350 -0
  173. data/lib/nokogiri/xml/notation.rb +6 -0
  174. data/lib/nokogiri/xml/parse_options.rb +85 -0
  175. data/lib/nokogiri/xml/pp.rb +2 -0
  176. data/lib/nokogiri/xml/pp/character_data.rb +18 -0
  177. data/lib/nokogiri/xml/pp/node.rb +56 -0
  178. data/lib/nokogiri/xml/processing_instruction.rb +8 -0
  179. data/lib/nokogiri/xml/reader.rb +112 -0
  180. data/lib/nokogiri/xml/relax_ng.rb +32 -0
  181. data/lib/nokogiri/xml/sax.rb +4 -0
  182. data/lib/nokogiri/xml/sax/document.rb +164 -0
  183. data/lib/nokogiri/xml/sax/parser.rb +115 -0
  184. data/lib/nokogiri/xml/sax/parser_context.rb +16 -0
  185. data/lib/nokogiri/xml/sax/push_parser.rb +60 -0
  186. data/lib/nokogiri/xml/schema.rb +57 -0
  187. data/lib/nokogiri/xml/syntax_error.rb +47 -0
  188. data/lib/nokogiri/xml/text.rb +9 -0
  189. data/lib/nokogiri/xml/xpath.rb +10 -0
  190. data/lib/nokogiri/xml/xpath/syntax_error.rb +11 -0
  191. data/lib/nokogiri/xml/xpath_context.rb +16 -0
  192. data/lib/nokogiri/xslt.rb +52 -0
  193. data/lib/nokogiri/xslt/stylesheet.rb +25 -0
  194. data/lib/xercesImpl.jar +0 -0
  195. data/lib/xsd/xmlparser/nokogiri.rb +90 -0
  196. data/tasks/cross_compile.rb +177 -0
  197. data/tasks/test.rb +94 -0
  198. data/test/css/test_nthiness.rb +159 -0
  199. data/test/css/test_parser.rb +303 -0
  200. data/test/css/test_tokenizer.rb +198 -0
  201. data/test/css/test_xpath_visitor.rb +85 -0
  202. data/test/decorators/test_slop.rb +16 -0
  203. data/test/files/2ch.html +108 -0
  204. data/test/files/address_book.rlx +12 -0
  205. data/test/files/address_book.xml +10 -0
  206. data/test/files/bar/bar.xsd +4 -0
  207. data/test/files/dont_hurt_em_why.xml +422 -0
  208. data/test/files/exslt.xml +8 -0
  209. data/test/files/exslt.xslt +35 -0
  210. data/test/files/foo/foo.xsd +4 -0
  211. data/test/files/po.xml +32 -0
  212. data/test/files/po.xsd +66 -0
  213. data/test/files/shift_jis.html +10 -0
  214. data/test/files/shift_jis.xml +5 -0
  215. data/test/files/snuggles.xml +3 -0
  216. data/test/files/staff.dtd +10 -0
  217. data/test/files/staff.xml +59 -0
  218. data/test/files/staff.xslt +32 -0
  219. data/test/files/tlm.html +850 -0
  220. data/test/files/valid_bar.xml +2 -0
  221. data/test/helper.rb +171 -0
  222. data/test/html/sax/test_parser.rb +136 -0
  223. data/test/html/sax/test_parser_context.rb +48 -0
  224. data/test/html/test_builder.rb +164 -0
  225. data/test/html/test_document.rb +457 -0
  226. data/test/html/test_document_encoding.rb +123 -0
  227. data/test/html/test_document_fragment.rb +255 -0
  228. data/test/html/test_element_description.rb +100 -0
  229. data/test/html/test_named_characters.rb +14 -0
  230. data/test/html/test_node.rb +190 -0
  231. data/test/html/test_node_encoding.rb +27 -0
  232. data/test/test_convert_xpath.rb +135 -0
  233. data/test/test_css_cache.rb +45 -0
  234. data/test/test_encoding_handler.rb +46 -0
  235. data/test/test_memory_leak.rb +52 -0
  236. data/test/test_nokogiri.rb +132 -0
  237. data/test/test_reader.rb +403 -0
  238. data/test/test_soap4r_sax.rb +52 -0
  239. data/test/test_xslt_transforms.rb +189 -0
  240. data/test/xml/node/test_save_options.rb +20 -0
  241. data/test/xml/node/test_subclass.rb +44 -0
  242. data/test/xml/sax/test_parser.rb +338 -0
  243. data/test/xml/sax/test_parser_context.rb +113 -0
  244. data/test/xml/sax/test_push_parser.rb +156 -0
  245. data/test/xml/test_attr.rb +65 -0
  246. data/test/xml/test_attribute_decl.rb +86 -0
  247. data/test/xml/test_builder.rb +210 -0
  248. data/test/xml/test_cdata.rb +50 -0
  249. data/test/xml/test_comment.rb +29 -0
  250. data/test/xml/test_document.rb +675 -0
  251. data/test/xml/test_document_encoding.rb +26 -0
  252. data/test/xml/test_document_fragment.rb +192 -0
  253. data/test/xml/test_dtd.rb +107 -0
  254. data/test/xml/test_dtd_encoding.rb +33 -0
  255. data/test/xml/test_element_content.rb +56 -0
  256. data/test/xml/test_element_decl.rb +73 -0
  257. data/test/xml/test_entity_decl.rb +122 -0
  258. data/test/xml/test_entity_reference.rb +21 -0
  259. data/test/xml/test_namespace.rb +70 -0
  260. data/test/xml/test_node.rb +899 -0
  261. data/test/xml/test_node_attributes.rb +34 -0
  262. data/test/xml/test_node_encoding.rb +107 -0
  263. data/test/xml/test_node_reparenting.rb +321 -0
  264. data/test/xml/test_node_set.rb +708 -0
  265. data/test/xml/test_parse_options.rb +52 -0
  266. data/test/xml/test_processing_instruction.rb +30 -0
  267. data/test/xml/test_reader_encoding.rb +126 -0
  268. data/test/xml/test_relax_ng.rb +60 -0
  269. data/test/xml/test_schema.rb +89 -0
  270. data/test/xml/test_syntax_error.rb +12 -0
  271. data/test/xml/test_text.rb +47 -0
  272. data/test/xml/test_unparented_node.rb +381 -0
  273. data/test/xml/test_xpath.rb +237 -0
  274. data/test/xslt/test_custom_functions.rb +94 -0
  275. metadata +525 -0
@@ -0,0 +1,39 @@
1
+ #ifndef NOKOGIRI_XML_SAX_PARSER
2
+ #define NOKOGIRI_XML_SAX_PARSER
3
+
4
+ #include <nokogiri.h>
5
+
6
+ void init_xml_sax_parser();
7
+
8
+ extern VALUE cNokogiriXmlSaxParser ;
9
+
10
+ typedef struct _nokogiriSAXTuple {
11
+ xmlParserCtxtPtr ctxt;
12
+ VALUE self;
13
+ } nokogiriSAXTuple;
14
+
15
+ typedef nokogiriSAXTuple * nokogiriSAXTuplePtr;
16
+
17
+ #define NOKOGIRI_SAX_SELF(_ctxt) \
18
+ ((nokogiriSAXTuplePtr)(_ctxt))->self
19
+
20
+ #define NOKOGIRI_SAX_CTXT(_ctxt) \
21
+ ((nokogiriSAXTuplePtr)(_ctxt))->ctxt
22
+
23
+ #define NOKOGIRI_SAX_TUPLE_NEW(_ctxt, _self) \
24
+ nokogiri_sax_tuple_new(_ctxt, _self)
25
+
26
+ static inline nokogiriSAXTuplePtr
27
+ nokogiri_sax_tuple_new(xmlParserCtxtPtr ctxt, VALUE self)
28
+ {
29
+ nokogiriSAXTuplePtr tuple = malloc(sizeof(nokogiriSAXTuple));
30
+ tuple->self = self;
31
+ tuple->ctxt = ctxt;
32
+ return tuple;
33
+ }
34
+
35
+ #define NOKOGIRI_SAX_TUPLE_DESTROY(_tuple) \
36
+ free(_tuple) \
37
+
38
+ #endif
39
+
@@ -0,0 +1,199 @@
1
+ #include <xml_sax_parser_context.h>
2
+
3
+ VALUE cNokogiriXmlSaxParserContext ;
4
+
5
+ static void deallocate(xmlParserCtxtPtr ctxt)
6
+ {
7
+ NOKOGIRI_DEBUG_START(handler);
8
+
9
+ ctxt->sax = NULL;
10
+
11
+ xmlFreeParserCtxt(ctxt);
12
+
13
+ NOKOGIRI_DEBUG_END(handler);
14
+ }
15
+
16
+ /*
17
+ * call-seq:
18
+ * parse_io(io, encoding)
19
+ *
20
+ * Parse +io+ object with +encoding+
21
+ */
22
+ static VALUE parse_io(VALUE klass, VALUE io, VALUE encoding)
23
+ {
24
+ xmlCharEncoding enc = (xmlCharEncoding)NUM2INT(encoding);
25
+
26
+ xmlParserCtxtPtr ctxt = xmlCreateIOParserCtxt(
27
+ NULL,
28
+ NULL,
29
+ (xmlInputReadCallback)io_read_callback,
30
+ (xmlInputCloseCallback)io_close_callback,
31
+ (void *)io,
32
+ enc
33
+ );
34
+
35
+ return Data_Wrap_Struct(klass, NULL, deallocate, ctxt);
36
+ }
37
+
38
+ /*
39
+ * call-seq:
40
+ * parse_file(filename)
41
+ *
42
+ * Parse file given +filename+
43
+ */
44
+ static VALUE parse_file(VALUE klass, VALUE filename)
45
+ {
46
+ xmlParserCtxtPtr ctxt = xmlCreateFileParserCtxt(StringValuePtr(filename));
47
+ return Data_Wrap_Struct(klass, NULL, deallocate, ctxt);
48
+ }
49
+
50
+ /*
51
+ * call-seq:
52
+ * parse_memory(data)
53
+ *
54
+ * Parse the XML stored in memory in +data+
55
+ */
56
+ static VALUE parse_memory(VALUE klass, VALUE data)
57
+ {
58
+ xmlParserCtxtPtr ctxt;
59
+
60
+ if(NIL_P(data)) rb_raise(rb_eArgError, "data cannot be nil");
61
+ if(!(int)RSTRING_LEN(data))
62
+ rb_raise(rb_eRuntimeError, "data cannot be empty");
63
+
64
+ ctxt = xmlCreateMemoryParserCtxt(
65
+ StringValuePtr(data),
66
+ (int)RSTRING_LEN(data)
67
+ );
68
+
69
+ return Data_Wrap_Struct(klass, NULL, deallocate, ctxt);
70
+ }
71
+
72
+ /*
73
+ * call-seq:
74
+ * parse_with(sax_handler)
75
+ *
76
+ * Use +sax_handler+ and parse the current document
77
+ */
78
+ static VALUE parse_with(VALUE self, VALUE sax_handler)
79
+ {
80
+ xmlParserCtxtPtr ctxt;
81
+ xmlSAXHandlerPtr sax;
82
+
83
+ if(!rb_obj_is_kind_of(sax_handler, cNokogiriXmlSaxParser))
84
+ rb_raise(rb_eArgError, "argument must be a Nokogiri::XML::SAX::Parser");
85
+
86
+ Data_Get_Struct(self, xmlParserCtxt, ctxt);
87
+ Data_Get_Struct(sax_handler, xmlSAXHandler, sax);
88
+
89
+ /* Free the sax handler since we'll assign our own */
90
+ if(ctxt->sax && ctxt->sax != (xmlSAXHandlerPtr)&xmlDefaultSAXHandler)
91
+ xmlFree(ctxt->sax);
92
+
93
+ ctxt->sax = sax;
94
+ ctxt->userData = (void *)NOKOGIRI_SAX_TUPLE_NEW(ctxt, sax_handler);
95
+
96
+ xmlParseDocument(ctxt);
97
+
98
+ if(NULL != ctxt->myDoc) xmlFreeDoc(ctxt->myDoc);
99
+
100
+ NOKOGIRI_SAX_TUPLE_DESTROY(ctxt->userData);
101
+
102
+ return Qnil ;
103
+ }
104
+
105
+ /*
106
+ * call-seq:
107
+ * replace_entities=(boolean)
108
+ *
109
+ * Should this parser replace entities? &amp; will get converted to '&' if
110
+ * set to true
111
+ */
112
+ static VALUE set_replace_entities(VALUE self, VALUE value)
113
+ {
114
+ xmlParserCtxtPtr ctxt;
115
+ Data_Get_Struct(self, xmlParserCtxt, ctxt);
116
+
117
+ if(Qfalse == value)
118
+ ctxt->replaceEntities = 0;
119
+ else
120
+ ctxt->replaceEntities = 1;
121
+
122
+ return value;
123
+ }
124
+
125
+ /*
126
+ * call-seq:
127
+ * replace_entities
128
+ *
129
+ * Should this parser replace entities? &amp; will get converted to '&' if
130
+ * set to true
131
+ */
132
+ static VALUE get_replace_entities(VALUE self)
133
+ {
134
+ xmlParserCtxtPtr ctxt;
135
+ Data_Get_Struct(self, xmlParserCtxt, ctxt);
136
+
137
+ if(0 == ctxt->replaceEntities)
138
+ return Qfalse;
139
+ else
140
+ return Qtrue;
141
+ }
142
+
143
+ /*
144
+ * call-seq: line
145
+ *
146
+ * Get the current line the parser context is processing.
147
+ */
148
+ static VALUE line(VALUE self)
149
+ {
150
+ xmlParserCtxtPtr ctxt;
151
+ xmlParserInputPtr io;
152
+
153
+ Data_Get_Struct(self, xmlParserCtxt, ctxt);
154
+
155
+ io = ctxt->input;
156
+ if(io)
157
+ return INT2NUM(io->line);
158
+
159
+ return Qnil;
160
+ }
161
+
162
+ /*
163
+ * call-seq: column
164
+ *
165
+ * Get the current column the parser context is processing.
166
+ */
167
+ static VALUE column(VALUE self)
168
+ {
169
+ xmlParserCtxtPtr ctxt;
170
+ xmlParserInputPtr io;
171
+
172
+ Data_Get_Struct(self, xmlParserCtxt, ctxt);
173
+
174
+ io = ctxt->input;
175
+ if(io)
176
+ return INT2NUM(io->col);
177
+
178
+ return Qnil;
179
+ }
180
+
181
+ void init_xml_sax_parser_context()
182
+ {
183
+ VALUE nokogiri = rb_define_module("Nokogiri");
184
+ VALUE xml = rb_define_module_under(nokogiri, "XML");
185
+ VALUE sax = rb_define_module_under(xml, "SAX");
186
+ VALUE klass = rb_define_class_under(sax, "ParserContext", rb_cObject);
187
+
188
+ cNokogiriXmlSaxParserContext = klass;
189
+
190
+ rb_define_singleton_method(klass, "io", parse_io, 2);
191
+ rb_define_singleton_method(klass, "memory", parse_memory, 1);
192
+ rb_define_singleton_method(klass, "file", parse_file, 1);
193
+
194
+ rb_define_method(klass, "parse_with", parse_with, 1);
195
+ rb_define_method(klass, "replace_entities=", set_replace_entities, 1);
196
+ rb_define_method(klass, "replace_entities", get_replace_entities, 0);
197
+ rb_define_method(klass, "line", line, 0);
198
+ rb_define_method(klass, "column", column, 0);
199
+ }
@@ -0,0 +1,10 @@
1
+ #ifndef NOKOGIRI_XML_SAX_PARSER_CONTEXT
2
+ #define NOKOGIRI_XML_SAX_PARSER_CONTEXT
3
+
4
+ #include <nokogiri.h>
5
+
6
+ extern VALUE cNokogiriXmlSaxParserContext;
7
+
8
+ void init_xml_sax_parser_context();
9
+
10
+ #endif
@@ -0,0 +1,115 @@
1
+ #include <xml_sax_push_parser.h>
2
+
3
+ static void deallocate(xmlParserCtxtPtr ctx)
4
+ {
5
+ NOKOGIRI_DEBUG_START(ctx);
6
+ if(ctx != NULL) {
7
+ NOKOGIRI_SAX_TUPLE_DESTROY(ctx->userData);
8
+ xmlFreeParserCtxt(ctx);
9
+ }
10
+ NOKOGIRI_DEBUG_END(ctx);
11
+ }
12
+
13
+ static VALUE allocate(VALUE klass)
14
+ {
15
+ return Data_Wrap_Struct(klass, NULL, deallocate, NULL);
16
+ }
17
+
18
+ /*
19
+ * call-seq:
20
+ * native_write(chunk, last_chunk)
21
+ *
22
+ * Write +chunk+ to PushParser. +last_chunk+ triggers the end_document handle
23
+ */
24
+ static VALUE native_write(VALUE self, VALUE _chunk, VALUE _last_chunk)
25
+ {
26
+ xmlParserCtxtPtr ctx;
27
+ const char * chunk = NULL;
28
+ int size = 0;
29
+
30
+
31
+ Data_Get_Struct(self, xmlParserCtxt, ctx);
32
+
33
+ if(Qnil != _chunk) {
34
+ chunk = StringValuePtr(_chunk);
35
+ size = (int)RSTRING_LEN(_chunk);
36
+ }
37
+
38
+ if(xmlParseChunk(ctx, chunk, size, Qtrue == _last_chunk ? 1 : 0)) {
39
+ if (!(ctx->options & XML_PARSE_RECOVER)) {
40
+ xmlErrorPtr e = xmlCtxtGetLastError(ctx);
41
+ Nokogiri_error_raise(NULL, e);
42
+ }
43
+ }
44
+
45
+ return self;
46
+ }
47
+
48
+ /*
49
+ * call-seq:
50
+ * initialize_native(xml_sax, filename)
51
+ *
52
+ * Initialize the push parser with +xml_sax+ using +filename+
53
+ */
54
+ static VALUE initialize_native(VALUE self, VALUE _xml_sax, VALUE _filename)
55
+ {
56
+ xmlSAXHandlerPtr sax;
57
+ const char * filename = NULL;
58
+ xmlParserCtxtPtr ctx;
59
+
60
+ Data_Get_Struct(_xml_sax, xmlSAXHandler, sax);
61
+
62
+ if(_filename != Qnil) filename = StringValuePtr(_filename);
63
+
64
+ ctx = xmlCreatePushParserCtxt(
65
+ sax,
66
+ NULL,
67
+ NULL,
68
+ 0,
69
+ filename
70
+ );
71
+ if(ctx == NULL)
72
+ rb_raise(rb_eRuntimeError, "Could not create a parser context");
73
+
74
+ ctx->userData = NOKOGIRI_SAX_TUPLE_NEW(ctx, self);
75
+
76
+ ctx->sax2 = 1;
77
+ DATA_PTR(self) = ctx;
78
+ return self;
79
+ }
80
+
81
+ static VALUE get_options(VALUE self)
82
+ {
83
+ xmlParserCtxtPtr ctx;
84
+ Data_Get_Struct(self, xmlParserCtxt, ctx);
85
+
86
+ return INT2NUM(ctx->options);
87
+ }
88
+
89
+ static VALUE set_options(VALUE self, VALUE options)
90
+ {
91
+ xmlParserCtxtPtr ctx;
92
+ Data_Get_Struct(self, xmlParserCtxt, ctx);
93
+
94
+ if (xmlCtxtUseOptions(ctx, (int)NUM2INT(options)) != 0)
95
+ rb_raise(rb_eRuntimeError, "Cannot set XML parser context options");
96
+
97
+ return Qnil;
98
+ }
99
+
100
+ VALUE cNokogiriXmlSaxPushParser ;
101
+ void init_xml_sax_push_parser()
102
+ {
103
+ VALUE nokogiri = rb_define_module("Nokogiri");
104
+ VALUE xml = rb_define_module_under(nokogiri, "XML");
105
+ VALUE sax = rb_define_module_under(xml, "SAX");
106
+ VALUE klass = rb_define_class_under(sax, "PushParser", rb_cObject);
107
+
108
+ cNokogiriXmlSaxPushParser = klass;
109
+
110
+ rb_define_alloc_func(klass, allocate);
111
+ rb_define_private_method(klass, "initialize_native", initialize_native, 2);
112
+ rb_define_private_method(klass, "native_write", native_write, 2);
113
+ rb_define_method(klass, "options", get_options, 0);
114
+ rb_define_method(klass, "options=", set_options, 1);
115
+ }
@@ -0,0 +1,9 @@
1
+ #ifndef NOKOGIRI_XML_SAX_PUSH_PARSER
2
+ #define NOKOGIRI_XML_SAX_PUSH_PARSER
3
+
4
+ #include <nokogiri.h>
5
+
6
+ void init_xml_sax_push_parser();
7
+
8
+ extern VALUE cNokogiriXmlSaxPushParser ;
9
+ #endif
@@ -0,0 +1,205 @@
1
+ #include <xml_schema.h>
2
+
3
+ static void dealloc(xmlSchemaPtr schema)
4
+ {
5
+ NOKOGIRI_DEBUG_START(schema);
6
+ xmlSchemaFree(schema);
7
+ NOKOGIRI_DEBUG_END(schema);
8
+ }
9
+
10
+ /*
11
+ * call-seq:
12
+ * validate_document(document)
13
+ *
14
+ * Validate a Nokogiri::XML::Document against this Schema.
15
+ */
16
+ static VALUE validate_document(VALUE self, VALUE document)
17
+ {
18
+ xmlDocPtr doc;
19
+ xmlSchemaPtr schema;
20
+ xmlSchemaValidCtxtPtr valid_ctxt;
21
+ VALUE errors;
22
+
23
+ Data_Get_Struct(self, xmlSchema, schema);
24
+ Data_Get_Struct(document, xmlDoc, doc);
25
+
26
+ errors = rb_ary_new();
27
+
28
+ valid_ctxt = xmlSchemaNewValidCtxt(schema);
29
+
30
+ if(NULL == valid_ctxt) {
31
+ /* we have a problem */
32
+ rb_raise(rb_eRuntimeError, "Could not create a validation context");
33
+ }
34
+
35
+ #ifdef HAVE_XMLSCHEMASETVALIDSTRUCTUREDERRORS
36
+ xmlSchemaSetValidStructuredErrors(
37
+ valid_ctxt,
38
+ Nokogiri_error_array_pusher,
39
+ (void *)errors
40
+ );
41
+ #endif
42
+
43
+ xmlSchemaValidateDoc(valid_ctxt, doc);
44
+
45
+ xmlSchemaFreeValidCtxt(valid_ctxt);
46
+
47
+ return errors;
48
+ }
49
+
50
+ /*
51
+ * call-seq:
52
+ * validate_file(filename)
53
+ *
54
+ * Validate a file against this Schema.
55
+ */
56
+ static VALUE validate_file(VALUE self, VALUE rb_filename)
57
+ {
58
+ xmlSchemaPtr schema;
59
+ xmlSchemaValidCtxtPtr valid_ctxt;
60
+ const char *filename ;
61
+ VALUE errors;
62
+
63
+ Data_Get_Struct(self, xmlSchema, schema);
64
+ filename = (const char*)StringValuePtr(rb_filename) ;
65
+
66
+ errors = rb_ary_new();
67
+
68
+ valid_ctxt = xmlSchemaNewValidCtxt(schema);
69
+
70
+ if(NULL == valid_ctxt) {
71
+ /* we have a problem */
72
+ rb_raise(rb_eRuntimeError, "Could not create a validation context");
73
+ }
74
+
75
+ #ifdef HAVE_XMLSCHEMASETVALIDSTRUCTUREDERRORS
76
+ xmlSchemaSetValidStructuredErrors(
77
+ valid_ctxt,
78
+ Nokogiri_error_array_pusher,
79
+ (void *)errors
80
+ );
81
+ #endif
82
+
83
+ xmlSchemaValidateFile(valid_ctxt, filename, 0);
84
+
85
+ xmlSchemaFreeValidCtxt(valid_ctxt);
86
+
87
+ return errors;
88
+ }
89
+
90
+ /*
91
+ * call-seq:
92
+ * read_memory(string)
93
+ *
94
+ * Create a new Schema from the contents of +string+
95
+ */
96
+ static VALUE read_memory(VALUE klass, VALUE content)
97
+ {
98
+ xmlSchemaPtr schema;
99
+ xmlSchemaParserCtxtPtr ctx = xmlSchemaNewMemParserCtxt(
100
+ (const char *)StringValuePtr(content),
101
+ (int)RSTRING_LEN(content)
102
+ );
103
+ VALUE rb_schema;
104
+ VALUE errors = rb_ary_new();
105
+ xmlSetStructuredErrorFunc((void *)errors, Nokogiri_error_array_pusher);
106
+
107
+ #ifdef HAVE_XMLSCHEMASETPARSERSTRUCTUREDERRORS
108
+ xmlSchemaSetParserStructuredErrors(
109
+ ctx,
110
+ Nokogiri_error_array_pusher,
111
+ (void *)errors
112
+ );
113
+ #endif
114
+
115
+ schema = xmlSchemaParse(ctx);
116
+
117
+ xmlSetStructuredErrorFunc(NULL, NULL);
118
+ xmlSchemaFreeParserCtxt(ctx);
119
+
120
+ if(NULL == schema) {
121
+ xmlErrorPtr error = xmlGetLastError();
122
+ if(error)
123
+ Nokogiri_error_raise(NULL, error);
124
+ else
125
+ rb_raise(rb_eRuntimeError, "Could not parse document");
126
+
127
+ return Qnil;
128
+ }
129
+
130
+ rb_schema = Data_Wrap_Struct(klass, 0, dealloc, schema);
131
+ rb_iv_set(rb_schema, "@errors", errors);
132
+
133
+ return rb_schema;
134
+ }
135
+
136
+ /*
137
+ * call-seq:
138
+ * from_document(doc)
139
+ *
140
+ * Create a new Schema from the Nokogiri::XML::Document +doc+
141
+ */
142
+ static VALUE from_document(VALUE klass, VALUE document)
143
+ {
144
+ xmlDocPtr doc;
145
+ xmlSchemaParserCtxtPtr ctx;
146
+ xmlSchemaPtr schema;
147
+ VALUE errors;
148
+ VALUE rb_schema;
149
+
150
+ Data_Get_Struct(document, xmlDoc, doc);
151
+
152
+ /* In case someone passes us a node. ugh. */
153
+ doc = doc->doc;
154
+
155
+ ctx = xmlSchemaNewDocParserCtxt(doc);
156
+
157
+ errors = rb_ary_new();
158
+ xmlSetStructuredErrorFunc((void *)errors, Nokogiri_error_array_pusher);
159
+
160
+ #ifdef HAVE_XMLSCHEMASETPARSERSTRUCTUREDERRORS
161
+ xmlSchemaSetParserStructuredErrors(
162
+ ctx,
163
+ Nokogiri_error_array_pusher,
164
+ (void *)errors
165
+ );
166
+ #endif
167
+
168
+ schema = xmlSchemaParse(ctx);
169
+
170
+ xmlSetStructuredErrorFunc(NULL, NULL);
171
+ xmlSchemaFreeParserCtxt(ctx);
172
+
173
+ if(NULL == schema) {
174
+ xmlErrorPtr error = xmlGetLastError();
175
+ if(error)
176
+ Nokogiri_error_raise(NULL, error);
177
+ else
178
+ rb_raise(rb_eRuntimeError, "Could not parse document");
179
+
180
+ return Qnil;
181
+ }
182
+
183
+ rb_schema = Data_Wrap_Struct(klass, 0, dealloc, schema);
184
+ rb_iv_set(rb_schema, "@errors", errors);
185
+
186
+ return rb_schema;
187
+
188
+ return Qnil;
189
+ }
190
+
191
+ VALUE cNokogiriXmlSchema;
192
+ void init_xml_schema()
193
+ {
194
+ VALUE nokogiri = rb_define_module("Nokogiri");
195
+ VALUE xml = rb_define_module_under(nokogiri, "XML");
196
+ VALUE klass = rb_define_class_under(xml, "Schema", rb_cObject);
197
+
198
+ cNokogiriXmlSchema = klass;
199
+
200
+ rb_define_singleton_method(klass, "read_memory", read_memory, 1);
201
+ rb_define_singleton_method(klass, "from_document", from_document, 1);
202
+
203
+ rb_define_private_method(klass, "validate_document", validate_document, 1);
204
+ rb_define_private_method(klass, "validate_file", validate_file, 1);
205
+ }