superfeedr-nokogiri 1.4.0.20091116183308

Sign up to get free protection for your applications and to get access to all the features.
Files changed (270) hide show
  1. data/.autotest +27 -0
  2. data/CHANGELOG.ja.rdoc +330 -0
  3. data/CHANGELOG.rdoc +314 -0
  4. data/Manifest.txt +269 -0
  5. data/README.ja.rdoc +105 -0
  6. data/README.rdoc +118 -0
  7. data/Rakefile +244 -0
  8. data/bin/nokogiri +49 -0
  9. data/ext/nokogiri/extconf.rb +145 -0
  10. data/ext/nokogiri/html_document.c +145 -0
  11. data/ext/nokogiri/html_document.h +10 -0
  12. data/ext/nokogiri/html_element_description.c +272 -0
  13. data/ext/nokogiri/html_element_description.h +10 -0
  14. data/ext/nokogiri/html_entity_lookup.c +32 -0
  15. data/ext/nokogiri/html_entity_lookup.h +8 -0
  16. data/ext/nokogiri/html_sax_parser_context.c +92 -0
  17. data/ext/nokogiri/html_sax_parser_context.h +11 -0
  18. data/ext/nokogiri/nokogiri.c +89 -0
  19. data/ext/nokogiri/nokogiri.h +145 -0
  20. data/ext/nokogiri/xml_attr.c +92 -0
  21. data/ext/nokogiri/xml_attr.h +9 -0
  22. data/ext/nokogiri/xml_attribute_decl.c +67 -0
  23. data/ext/nokogiri/xml_attribute_decl.h +9 -0
  24. data/ext/nokogiri/xml_cdata.c +54 -0
  25. data/ext/nokogiri/xml_cdata.h +9 -0
  26. data/ext/nokogiri/xml_comment.c +52 -0
  27. data/ext/nokogiri/xml_comment.h +9 -0
  28. data/ext/nokogiri/xml_document.c +388 -0
  29. data/ext/nokogiri/xml_document.h +24 -0
  30. data/ext/nokogiri/xml_document_fragment.c +46 -0
  31. data/ext/nokogiri/xml_document_fragment.h +10 -0
  32. data/ext/nokogiri/xml_dtd.c +192 -0
  33. data/ext/nokogiri/xml_dtd.h +10 -0
  34. data/ext/nokogiri/xml_element_content.c +123 -0
  35. data/ext/nokogiri/xml_element_content.h +10 -0
  36. data/ext/nokogiri/xml_element_decl.c +69 -0
  37. data/ext/nokogiri/xml_element_decl.h +9 -0
  38. data/ext/nokogiri/xml_entity_decl.c +97 -0
  39. data/ext/nokogiri/xml_entity_decl.h +10 -0
  40. data/ext/nokogiri/xml_entity_reference.c +50 -0
  41. data/ext/nokogiri/xml_entity_reference.h +9 -0
  42. data/ext/nokogiri/xml_io.c +31 -0
  43. data/ext/nokogiri/xml_io.h +11 -0
  44. data/ext/nokogiri/xml_namespace.c +74 -0
  45. data/ext/nokogiri/xml_namespace.h +12 -0
  46. data/ext/nokogiri/xml_node.c +1060 -0
  47. data/ext/nokogiri/xml_node.h +13 -0
  48. data/ext/nokogiri/xml_node_set.c +397 -0
  49. data/ext/nokogiri/xml_node_set.h +9 -0
  50. data/ext/nokogiri/xml_processing_instruction.c +54 -0
  51. data/ext/nokogiri/xml_processing_instruction.h +9 -0
  52. data/ext/nokogiri/xml_reader.c +593 -0
  53. data/ext/nokogiri/xml_reader.h +10 -0
  54. data/ext/nokogiri/xml_relax_ng.c +159 -0
  55. data/ext/nokogiri/xml_relax_ng.h +9 -0
  56. data/ext/nokogiri/xml_sax_parser.c +286 -0
  57. data/ext/nokogiri/xml_sax_parser.h +43 -0
  58. data/ext/nokogiri/xml_sax_parser_context.c +155 -0
  59. data/ext/nokogiri/xml_sax_parser_context.h +10 -0
  60. data/ext/nokogiri/xml_sax_push_parser.c +114 -0
  61. data/ext/nokogiri/xml_sax_push_parser.h +9 -0
  62. data/ext/nokogiri/xml_schema.c +156 -0
  63. data/ext/nokogiri/xml_schema.h +9 -0
  64. data/ext/nokogiri/xml_syntax_error.c +261 -0
  65. data/ext/nokogiri/xml_syntax_error.h +13 -0
  66. data/ext/nokogiri/xml_text.c +48 -0
  67. data/ext/nokogiri/xml_text.h +9 -0
  68. data/ext/nokogiri/xml_xpath.c +53 -0
  69. data/ext/nokogiri/xml_xpath.h +11 -0
  70. data/ext/nokogiri/xml_xpath_context.c +239 -0
  71. data/ext/nokogiri/xml_xpath_context.h +9 -0
  72. data/ext/nokogiri/xslt_stylesheet.c +131 -0
  73. data/ext/nokogiri/xslt_stylesheet.h +9 -0
  74. data/lib/nokogiri.rb +116 -0
  75. data/lib/nokogiri/css.rb +25 -0
  76. data/lib/nokogiri/css/generated_parser.rb +646 -0
  77. data/lib/nokogiri/css/generated_tokenizer.rb +142 -0
  78. data/lib/nokogiri/css/node.rb +99 -0
  79. data/lib/nokogiri/css/parser.rb +82 -0
  80. data/lib/nokogiri/css/parser.y +227 -0
  81. data/lib/nokogiri/css/syntax_error.rb +7 -0
  82. data/lib/nokogiri/css/tokenizer.rb +7 -0
  83. data/lib/nokogiri/css/tokenizer.rex +54 -0
  84. data/lib/nokogiri/css/xpath_visitor.rb +162 -0
  85. data/lib/nokogiri/decorators/slop.rb +33 -0
  86. data/lib/nokogiri/ffi/html/document.rb +28 -0
  87. data/lib/nokogiri/ffi/html/element_description.rb +85 -0
  88. data/lib/nokogiri/ffi/html/entity_lookup.rb +16 -0
  89. data/lib/nokogiri/ffi/html/sax/parser_context.rb +38 -0
  90. data/lib/nokogiri/ffi/io_callbacks.rb +42 -0
  91. data/lib/nokogiri/ffi/libxml.rb +356 -0
  92. data/lib/nokogiri/ffi/structs/common_node.rb +26 -0
  93. data/lib/nokogiri/ffi/structs/html_elem_desc.rb +24 -0
  94. data/lib/nokogiri/ffi/structs/html_entity_desc.rb +13 -0
  95. data/lib/nokogiri/ffi/structs/xml_alloc.rb +16 -0
  96. data/lib/nokogiri/ffi/structs/xml_attr.rb +19 -0
  97. data/lib/nokogiri/ffi/structs/xml_attribute.rb +27 -0
  98. data/lib/nokogiri/ffi/structs/xml_buffer.rb +16 -0
  99. data/lib/nokogiri/ffi/structs/xml_document.rb +108 -0
  100. data/lib/nokogiri/ffi/structs/xml_dtd.rb +28 -0
  101. data/lib/nokogiri/ffi/structs/xml_element.rb +26 -0
  102. data/lib/nokogiri/ffi/structs/xml_element_content.rb +17 -0
  103. data/lib/nokogiri/ffi/structs/xml_entity.rb +32 -0
  104. data/lib/nokogiri/ffi/structs/xml_enumeration.rb +12 -0
  105. data/lib/nokogiri/ffi/structs/xml_node.rb +28 -0
  106. data/lib/nokogiri/ffi/structs/xml_node_set.rb +53 -0
  107. data/lib/nokogiri/ffi/structs/xml_notation.rb +11 -0
  108. data/lib/nokogiri/ffi/structs/xml_ns.rb +15 -0
  109. data/lib/nokogiri/ffi/structs/xml_parser_context.rb +19 -0
  110. data/lib/nokogiri/ffi/structs/xml_relax_ng.rb +14 -0
  111. data/lib/nokogiri/ffi/structs/xml_sax_handler.rb +51 -0
  112. data/lib/nokogiri/ffi/structs/xml_sax_push_parser_context.rb +15 -0
  113. data/lib/nokogiri/ffi/structs/xml_schema.rb +13 -0
  114. data/lib/nokogiri/ffi/structs/xml_syntax_error.rb +31 -0
  115. data/lib/nokogiri/ffi/structs/xml_text_reader.rb +12 -0
  116. data/lib/nokogiri/ffi/structs/xml_xpath_context.rb +37 -0
  117. data/lib/nokogiri/ffi/structs/xml_xpath_object.rb +35 -0
  118. data/lib/nokogiri/ffi/structs/xml_xpath_parser_context.rb +20 -0
  119. data/lib/nokogiri/ffi/structs/xslt_stylesheet.rb +13 -0
  120. data/lib/nokogiri/ffi/xml/attr.rb +41 -0
  121. data/lib/nokogiri/ffi/xml/attribute_decl.rb +27 -0
  122. data/lib/nokogiri/ffi/xml/cdata.rb +19 -0
  123. data/lib/nokogiri/ffi/xml/comment.rb +18 -0
  124. data/lib/nokogiri/ffi/xml/document.rb +135 -0
  125. data/lib/nokogiri/ffi/xml/document_fragment.rb +21 -0
  126. data/lib/nokogiri/ffi/xml/dtd.rb +69 -0
  127. data/lib/nokogiri/ffi/xml/element_content.rb +43 -0
  128. data/lib/nokogiri/ffi/xml/element_decl.rb +19 -0
  129. data/lib/nokogiri/ffi/xml/entity_decl.rb +27 -0
  130. data/lib/nokogiri/ffi/xml/entity_reference.rb +19 -0
  131. data/lib/nokogiri/ffi/xml/namespace.rb +44 -0
  132. data/lib/nokogiri/ffi/xml/node.rb +444 -0
  133. data/lib/nokogiri/ffi/xml/node_set.rb +133 -0
  134. data/lib/nokogiri/ffi/xml/processing_instruction.rb +20 -0
  135. data/lib/nokogiri/ffi/xml/reader.rb +227 -0
  136. data/lib/nokogiri/ffi/xml/relax_ng.rb +85 -0
  137. data/lib/nokogiri/ffi/xml/sax/parser.rb +142 -0
  138. data/lib/nokogiri/ffi/xml/sax/parser_context.rb +67 -0
  139. data/lib/nokogiri/ffi/xml/sax/push_parser.rb +39 -0
  140. data/lib/nokogiri/ffi/xml/schema.rb +92 -0
  141. data/lib/nokogiri/ffi/xml/syntax_error.rb +91 -0
  142. data/lib/nokogiri/ffi/xml/text.rb +18 -0
  143. data/lib/nokogiri/ffi/xml/xpath.rb +19 -0
  144. data/lib/nokogiri/ffi/xml/xpath_context.rb +135 -0
  145. data/lib/nokogiri/ffi/xslt/stylesheet.rb +47 -0
  146. data/lib/nokogiri/html.rb +35 -0
  147. data/lib/nokogiri/html/builder.rb +35 -0
  148. data/lib/nokogiri/html/document.rb +88 -0
  149. data/lib/nokogiri/html/document_fragment.rb +15 -0
  150. data/lib/nokogiri/html/element_description.rb +23 -0
  151. data/lib/nokogiri/html/entity_lookup.rb +13 -0
  152. data/lib/nokogiri/html/sax/parser.rb +48 -0
  153. data/lib/nokogiri/html/sax/parser_context.rb +16 -0
  154. data/lib/nokogiri/syntax_error.rb +4 -0
  155. data/lib/nokogiri/version.rb +33 -0
  156. data/lib/nokogiri/version_warning.rb +11 -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 +405 -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 +131 -0
  164. data/lib/nokogiri/xml/document_fragment.rb +69 -0
  165. data/lib/nokogiri/xml/dtd.rb +11 -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 +15 -0
  169. data/lib/nokogiri/xml/fragment_handler.rb +71 -0
  170. data/lib/nokogiri/xml/namespace.rb +13 -0
  171. data/lib/nokogiri/xml/node.rb +665 -0
  172. data/lib/nokogiri/xml/node/save_options.rb +42 -0
  173. data/lib/nokogiri/xml/node_set.rb +307 -0
  174. data/lib/nokogiri/xml/notation.rb +6 -0
  175. data/lib/nokogiri/xml/parse_options.rb +85 -0
  176. data/lib/nokogiri/xml/pp.rb +2 -0
  177. data/lib/nokogiri/xml/pp/character_data.rb +18 -0
  178. data/lib/nokogiri/xml/pp/node.rb +56 -0
  179. data/lib/nokogiri/xml/processing_instruction.rb +8 -0
  180. data/lib/nokogiri/xml/reader.rb +74 -0
  181. data/lib/nokogiri/xml/relax_ng.rb +32 -0
  182. data/lib/nokogiri/xml/sax.rb +4 -0
  183. data/lib/nokogiri/xml/sax/document.rb +160 -0
  184. data/lib/nokogiri/xml/sax/parser.rb +115 -0
  185. data/lib/nokogiri/xml/sax/parser_context.rb +16 -0
  186. data/lib/nokogiri/xml/sax/push_parser.rb +60 -0
  187. data/lib/nokogiri/xml/schema.rb +61 -0
  188. data/lib/nokogiri/xml/syntax_error.rb +38 -0
  189. data/lib/nokogiri/xml/xpath.rb +10 -0
  190. data/lib/nokogiri/xml/xpath/syntax_error.rb +8 -0
  191. data/lib/nokogiri/xml/xpath_context.rb +16 -0
  192. data/lib/nokogiri/xslt.rb +48 -0
  193. data/lib/nokogiri/xslt/stylesheet.rb +25 -0
  194. data/lib/xsd/xmlparser/nokogiri.rb +71 -0
  195. data/tasks/test.rb +100 -0
  196. data/test/css/test_nthiness.rb +159 -0
  197. data/test/css/test_parser.rb +277 -0
  198. data/test/css/test_tokenizer.rb +183 -0
  199. data/test/css/test_xpath_visitor.rb +76 -0
  200. data/test/ffi/test_document.rb +35 -0
  201. data/test/files/2ch.html +108 -0
  202. data/test/files/address_book.rlx +12 -0
  203. data/test/files/address_book.xml +10 -0
  204. data/test/files/bar/bar.xsd +4 -0
  205. data/test/files/dont_hurt_em_why.xml +422 -0
  206. data/test/files/exslt.xml +8 -0
  207. data/test/files/exslt.xslt +35 -0
  208. data/test/files/foo/foo.xsd +4 -0
  209. data/test/files/po.xml +32 -0
  210. data/test/files/po.xsd +66 -0
  211. data/test/files/shift_jis.html +10 -0
  212. data/test/files/shift_jis.xml +5 -0
  213. data/test/files/snuggles.xml +3 -0
  214. data/test/files/staff.dtd +10 -0
  215. data/test/files/staff.xml +59 -0
  216. data/test/files/staff.xslt +32 -0
  217. data/test/files/tlm.html +850 -0
  218. data/test/files/valid_bar.xml +2 -0
  219. data/test/helper.rb +136 -0
  220. data/test/html/sax/test_parser.rb +64 -0
  221. data/test/html/sax/test_parser_context.rb +48 -0
  222. data/test/html/test_builder.rb +164 -0
  223. data/test/html/test_document.rb +390 -0
  224. data/test/html/test_document_encoding.rb +77 -0
  225. data/test/html/test_document_fragment.rb +132 -0
  226. data/test/html/test_element_description.rb +94 -0
  227. data/test/html/test_named_characters.rb +14 -0
  228. data/test/html/test_node.rb +228 -0
  229. data/test/html/test_node_encoding.rb +27 -0
  230. data/test/test_convert_xpath.rb +135 -0
  231. data/test/test_css_cache.rb +45 -0
  232. data/test/test_gc.rb +15 -0
  233. data/test/test_memory_leak.rb +77 -0
  234. data/test/test_nokogiri.rb +134 -0
  235. data/test/test_reader.rb +358 -0
  236. data/test/test_xslt_transforms.rb +131 -0
  237. data/test/xml/node/test_save_options.rb +20 -0
  238. data/test/xml/node/test_subclass.rb +44 -0
  239. data/test/xml/sax/test_parser.rb +307 -0
  240. data/test/xml/sax/test_parser_context.rb +56 -0
  241. data/test/xml/sax/test_push_parser.rb +131 -0
  242. data/test/xml/test_attr.rb +38 -0
  243. data/test/xml/test_attribute_decl.rb +82 -0
  244. data/test/xml/test_builder.rb +167 -0
  245. data/test/xml/test_cdata.rb +38 -0
  246. data/test/xml/test_comment.rb +29 -0
  247. data/test/xml/test_document.rb +607 -0
  248. data/test/xml/test_document_encoding.rb +26 -0
  249. data/test/xml/test_document_fragment.rb +138 -0
  250. data/test/xml/test_dtd.rb +82 -0
  251. data/test/xml/test_dtd_encoding.rb +33 -0
  252. data/test/xml/test_element_content.rb +56 -0
  253. data/test/xml/test_element_decl.rb +73 -0
  254. data/test/xml/test_entity_decl.rb +83 -0
  255. data/test/xml/test_entity_reference.rb +21 -0
  256. data/test/xml/test_namespace.rb +68 -0
  257. data/test/xml/test_node.rb +889 -0
  258. data/test/xml/test_node_attributes.rb +34 -0
  259. data/test/xml/test_node_encoding.rb +107 -0
  260. data/test/xml/test_node_set.rb +531 -0
  261. data/test/xml/test_parse_options.rb +52 -0
  262. data/test/xml/test_processing_instruction.rb +30 -0
  263. data/test/xml/test_reader_encoding.rb +126 -0
  264. data/test/xml/test_relax_ng.rb +60 -0
  265. data/test/xml/test_schema.rb +89 -0
  266. data/test/xml/test_syntax_error.rb +27 -0
  267. data/test/xml/test_text.rb +30 -0
  268. data/test/xml/test_unparented_node.rb +381 -0
  269. data/test/xml/test_xpath.rb +106 -0
  270. metadata +430 -0
@@ -0,0 +1,43 @@
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
+ ({ \
19
+ nokogiriSAXTuplePtr _tuple = (nokogiriSAXTuplePtr)(_ctxt); \
20
+ _tuple->self; \
21
+ })
22
+
23
+ #define NOKOGIRI_SAX_CTXT(_ctxt) \
24
+ ({ \
25
+ nokogiriSAXTuplePtr _tuple = (nokogiriSAXTuplePtr)(_ctxt); \
26
+ _tuple->ctxt; \
27
+ })
28
+
29
+ #define NOKOGIRI_SAX_TUPLE_NEW(_ctxt, _self) \
30
+ ({ \
31
+ nokogiriSAXTuplePtr _tuple = malloc(sizeof(nokogiriSAXTuple)); \
32
+ _tuple->self = _self; \
33
+ _tuple->ctxt = _ctxt; \
34
+ _tuple; \
35
+ })
36
+
37
+ #define NOKOGIRI_SAX_TUPLE_DESTROY(_tuple) \
38
+ ({ \
39
+ free(_tuple); \
40
+ })
41
+
42
+ #endif
43
+
@@ -0,0 +1,155 @@
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
+ if(NIL_P(data)) rb_raise(rb_eArgError, "data cannot be nil");
59
+ if(!(int)RSTRING_LEN(data))
60
+ rb_raise(rb_eRuntimeError, "data cannot be empty");
61
+
62
+ xmlParserCtxtPtr ctxt = xmlCreateMemoryParserCtxt(
63
+ StringValuePtr(data),
64
+ (int)RSTRING_LEN(data)
65
+ );
66
+
67
+ return Data_Wrap_Struct(klass, NULL, deallocate, ctxt);
68
+ }
69
+
70
+ /*
71
+ * call-seq:
72
+ * parse_with(sax_handler)
73
+ *
74
+ * Use +sax_handler+ and parse the current document
75
+ */
76
+ static VALUE parse_with(VALUE self, VALUE sax_handler)
77
+ {
78
+ if(!rb_obj_is_kind_of(sax_handler, cNokogiriXmlSaxParser))
79
+ rb_raise(rb_eArgError, "argument must be a Nokogiri::XML::SAX::Parser");
80
+
81
+ xmlParserCtxtPtr ctxt;
82
+ Data_Get_Struct(self, xmlParserCtxt, ctxt);
83
+
84
+ xmlSAXHandlerPtr sax;
85
+ Data_Get_Struct(sax_handler, xmlSAXHandler, sax);
86
+
87
+ // Free the sax handler since we'll assign our own
88
+ if(ctxt->sax && ctxt->sax != (xmlSAXHandlerPtr)&xmlDefaultSAXHandler)
89
+ xmlFree(ctxt->sax);
90
+
91
+ ctxt->sax = sax;
92
+ ctxt->userData = (void *)NOKOGIRI_SAX_TUPLE_NEW(ctxt, sax_handler);
93
+
94
+ xmlParseDocument(ctxt);
95
+
96
+ if(NULL != ctxt->myDoc) xmlFreeDoc(ctxt->myDoc);
97
+
98
+ NOKOGIRI_SAX_TUPLE_DESTROY(ctxt->userData);
99
+ }
100
+
101
+ /*
102
+ * call-seq:
103
+ * replace_entities=(boolean)
104
+ *
105
+ * Should this parser replace entities? &amp; will get converted to '&' if
106
+ * set to true
107
+ */
108
+ static VALUE set_replace_entities(VALUE self, VALUE value)
109
+ {
110
+ xmlParserCtxtPtr ctxt;
111
+ Data_Get_Struct(self, xmlParserCtxt, ctxt);
112
+
113
+ if(Qfalse == value)
114
+ ctxt->replaceEntities = 0;
115
+ else
116
+ ctxt->replaceEntities = 1;
117
+
118
+ return value;
119
+ }
120
+
121
+ /*
122
+ * call-seq:
123
+ * replace_entities
124
+ *
125
+ * Should this parser replace entities? &amp; will get converted to '&' if
126
+ * set to true
127
+ */
128
+ static VALUE get_replace_entities(VALUE self)
129
+ {
130
+ xmlParserCtxtPtr ctxt;
131
+ Data_Get_Struct(self, xmlParserCtxt, ctxt);
132
+
133
+ if(0 == ctxt->replaceEntities)
134
+ return Qfalse;
135
+ else
136
+ return Qtrue;
137
+ }
138
+
139
+ void init_xml_sax_parser_context()
140
+ {
141
+ VALUE nokogiri = rb_define_module("Nokogiri");
142
+ VALUE xml = rb_define_module_under(nokogiri, "XML");
143
+ VALUE sax = rb_define_module_under(xml, "SAX");
144
+ VALUE klass = rb_define_class_under(sax, "ParserContext", rb_cObject);
145
+
146
+ cNokogiriXmlSaxParserContext = klass;
147
+
148
+ rb_define_singleton_method(klass, "io", parse_io, 2);
149
+ rb_define_singleton_method(klass, "memory", parse_memory, 1);
150
+ rb_define_singleton_method(klass, "file", parse_file, 1);
151
+
152
+ rb_define_method(klass, "parse_with", parse_with, 1);
153
+ rb_define_method(klass, "replace_entities=", set_replace_entities, 1);
154
+ rb_define_method(klass, "replace_entities", get_replace_entities, 0);
155
+ }
@@ -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 +1,114 @@
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
+ Data_Get_Struct(self, xmlParserCtxt, ctx);
28
+
29
+ const char * chunk = NULL;
30
+ int size = 0;
31
+
32
+ if(Qnil != _chunk) {
33
+ chunk = StringValuePtr(_chunk);
34
+ size = RSTRING_LEN(_chunk);
35
+ }
36
+
37
+ if(xmlParseChunk(ctx, chunk, size, Qtrue == _last_chunk ? 1 : 0)) {
38
+ if (!(ctx->options & XML_PARSE_RECOVER)) {
39
+ xmlErrorPtr e = xmlCtxtGetLastError(ctx);
40
+ Nokogiri_error_raise(NULL, e);
41
+ }
42
+ }
43
+
44
+ return self;
45
+ }
46
+
47
+ /*
48
+ * call-seq:
49
+ * initialize_native(xml_sax, filename)
50
+ *
51
+ * Initialize the push parser with +xml_sax+ using +filename+
52
+ */
53
+ static VALUE initialize_native(VALUE self, VALUE _xml_sax, VALUE _filename)
54
+ {
55
+ xmlSAXHandlerPtr sax;
56
+
57
+ Data_Get_Struct(_xml_sax, xmlSAXHandler, sax);
58
+
59
+ const char * filename = NULL;
60
+
61
+ if(_filename != Qnil) filename = StringValuePtr(_filename);
62
+
63
+ xmlParserCtxtPtr ctx = xmlCreatePushParserCtxt(
64
+ sax,
65
+ NULL,
66
+ NULL,
67
+ 0,
68
+ filename
69
+ );
70
+ if(ctx == NULL)
71
+ rb_raise(rb_eRuntimeError, "Could not create a parser context");
72
+
73
+ ctx->userData = NOKOGIRI_SAX_TUPLE_NEW(ctx, self);
74
+
75
+ ctx->sax2 = 1;
76
+ DATA_PTR(self) = ctx;
77
+ return self;
78
+ }
79
+
80
+ static VALUE get_options(VALUE self)
81
+ {
82
+ xmlParserCtxtPtr ctx;
83
+ Data_Get_Struct(self, xmlParserCtxt, ctx);
84
+
85
+ return INT2NUM(ctx->options);
86
+ }
87
+
88
+ static VALUE set_options(VALUE self, VALUE options)
89
+ {
90
+ xmlParserCtxtPtr ctx;
91
+ Data_Get_Struct(self, xmlParserCtxt, ctx);
92
+
93
+ if (xmlCtxtUseOptions(ctx, (int)NUM2INT(options)) != 0)
94
+ rb_raise(rb_eRuntimeError, "Cannot set XML parser context options");
95
+
96
+ return Qnil;
97
+ }
98
+
99
+ VALUE cNokogiriXmlSaxPushParser ;
100
+ void init_xml_sax_push_parser()
101
+ {
102
+ VALUE nokogiri = rb_define_module("Nokogiri");
103
+ VALUE xml = rb_define_module_under(nokogiri, "XML");
104
+ VALUE sax = rb_define_module_under(xml, "SAX");
105
+ VALUE klass = rb_define_class_under(sax, "PushParser", rb_cObject);
106
+
107
+ cNokogiriXmlSaxPushParser = klass;
108
+
109
+ rb_define_alloc_func(klass, allocate);
110
+ rb_define_private_method(klass, "initialize_native", initialize_native, 2);
111
+ rb_define_private_method(klass, "native_write", native_write, 2);
112
+ rb_define_method(klass, "options", get_options, 0);
113
+ rb_define_method(klass, "options=", set_options, 1);
114
+ }
@@ -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 +1,156 @@
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
+
21
+ Data_Get_Struct(self, xmlSchema, schema);
22
+ Data_Get_Struct(document, xmlDoc, doc);
23
+
24
+ VALUE errors = rb_ary_new();
25
+
26
+ xmlSchemaValidCtxtPtr valid_ctxt = xmlSchemaNewValidCtxt(schema);
27
+
28
+ if(NULL == valid_ctxt) {
29
+ // we have a problem
30
+ rb_raise(rb_eRuntimeError, "Could not create a validation context");
31
+ }
32
+
33
+ #ifdef HAVE_XMLSCHEMASETVALIDSTRUCTUREDERRORS
34
+ xmlSchemaSetValidStructuredErrors(
35
+ valid_ctxt,
36
+ Nokogiri_error_array_pusher,
37
+ (void *)errors
38
+ );
39
+ #endif
40
+
41
+ xmlSchemaValidateDoc(valid_ctxt, doc);
42
+
43
+ xmlSchemaFreeValidCtxt(valid_ctxt);
44
+
45
+ return errors;
46
+ }
47
+
48
+ /*
49
+ * call-seq:
50
+ * read_memory(string)
51
+ *
52
+ * Create a new Schema from the contents of +string+
53
+ */
54
+ static VALUE read_memory(VALUE klass, VALUE content)
55
+ {
56
+
57
+ xmlSchemaParserCtxtPtr ctx = xmlSchemaNewMemParserCtxt(
58
+ (const char *)StringValuePtr(content),
59
+ RSTRING_LEN(content)
60
+ );
61
+
62
+ VALUE errors = rb_ary_new();
63
+ xmlSetStructuredErrorFunc((void *)errors, Nokogiri_error_array_pusher);
64
+
65
+ #ifdef HAVE_XMLSCHEMASETPARSERSTRUCTUREDERRORS
66
+ xmlSchemaSetParserStructuredErrors(
67
+ ctx,
68
+ Nokogiri_error_array_pusher,
69
+ (void *)errors
70
+ );
71
+ #endif
72
+
73
+ xmlSchemaPtr schema = xmlSchemaParse(ctx);
74
+
75
+ xmlSetStructuredErrorFunc(NULL, NULL);
76
+ xmlSchemaFreeParserCtxt(ctx);
77
+
78
+ if(NULL == schema) {
79
+ xmlErrorPtr error = xmlGetLastError();
80
+ if(error)
81
+ Nokogiri_error_raise(NULL, error);
82
+ else
83
+ rb_raise(rb_eRuntimeError, "Could not parse document");
84
+
85
+ return Qnil;
86
+ }
87
+
88
+ VALUE rb_schema = Data_Wrap_Struct(klass, 0, dealloc, schema);
89
+ rb_iv_set(rb_schema, "@errors", errors);
90
+
91
+ return rb_schema;
92
+ }
93
+
94
+ /*
95
+ * call-seq:
96
+ * from_document(doc)
97
+ *
98
+ * Create a new Schema from the Nokogiri::XML::Document +doc+
99
+ */
100
+ static VALUE from_document(VALUE klass, VALUE document)
101
+ {
102
+ xmlDocPtr doc;
103
+ Data_Get_Struct(document, xmlDoc, doc);
104
+
105
+ // In case someone passes us a node. ugh.
106
+ doc = doc->doc;
107
+
108
+ xmlSchemaParserCtxtPtr ctx = xmlSchemaNewDocParserCtxt(doc);
109
+
110
+ VALUE errors = rb_ary_new();
111
+ xmlSetStructuredErrorFunc((void *)errors, Nokogiri_error_array_pusher);
112
+
113
+ #ifdef HAVE_XMLSCHEMASETPARSERSTRUCTUREDERRORS
114
+ xmlSchemaSetParserStructuredErrors(
115
+ ctx,
116
+ Nokogiri_error_array_pusher,
117
+ (void *)errors
118
+ );
119
+ #endif
120
+
121
+ xmlSchemaPtr schema = xmlSchemaParse(ctx);
122
+
123
+ xmlSetStructuredErrorFunc(NULL, NULL);
124
+ xmlSchemaFreeParserCtxt(ctx);
125
+
126
+ if(NULL == schema) {
127
+ xmlErrorPtr error = xmlGetLastError();
128
+ if(error)
129
+ Nokogiri_error_raise(NULL, error);
130
+ else
131
+ rb_raise(rb_eRuntimeError, "Could not parse document");
132
+
133
+ return Qnil;
134
+ }
135
+
136
+ VALUE rb_schema = Data_Wrap_Struct(klass, 0, dealloc, schema);
137
+ rb_iv_set(rb_schema, "@errors", errors);
138
+
139
+ return rb_schema;
140
+
141
+ return Qnil;
142
+ }
143
+
144
+ VALUE cNokogiriXmlSchema;
145
+ void init_xml_schema()
146
+ {
147
+ VALUE nokogiri = rb_define_module("Nokogiri");
148
+ VALUE xml = rb_define_module_under(nokogiri, "XML");
149
+ VALUE klass = rb_define_class_under(xml, "Schema", rb_cObject);
150
+
151
+ cNokogiriXmlSchema = klass;
152
+
153
+ rb_define_singleton_method(klass, "read_memory", read_memory, 1);
154
+ rb_define_singleton_method(klass, "from_document", from_document, 1);
155
+ rb_define_private_method(klass, "validate_document", validate_document, 1);
156
+ }