nokogiri 1.10.10 → 1.13.9

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 (220) hide show
  1. checksums.yaml +4 -4
  2. data/Gemfile +5 -0
  3. data/LICENSE-DEPENDENCIES.md +1173 -884
  4. data/LICENSE.md +1 -1
  5. data/README.md +178 -96
  6. data/bin/nokogiri +63 -50
  7. data/dependencies.yml +13 -64
  8. data/ext/nokogiri/depend +38 -358
  9. data/ext/nokogiri/extconf.rb +761 -424
  10. data/ext/nokogiri/gumbo.c +584 -0
  11. data/ext/nokogiri/html4_document.c +166 -0
  12. data/ext/nokogiri/html4_element_description.c +294 -0
  13. data/ext/nokogiri/html4_entity_lookup.c +37 -0
  14. data/ext/nokogiri/html4_sax_parser_context.c +119 -0
  15. data/ext/nokogiri/html4_sax_push_parser.c +95 -0
  16. data/ext/nokogiri/libxml2_backwards_compat.c +121 -0
  17. data/ext/nokogiri/nokogiri.c +228 -91
  18. data/ext/nokogiri/nokogiri.h +199 -88
  19. data/ext/nokogiri/test_global_handlers.c +40 -0
  20. data/ext/nokogiri/xml_attr.c +17 -17
  21. data/ext/nokogiri/xml_attribute_decl.c +21 -21
  22. data/ext/nokogiri/xml_cdata.c +14 -19
  23. data/ext/nokogiri/xml_comment.c +19 -26
  24. data/ext/nokogiri/xml_document.c +296 -220
  25. data/ext/nokogiri/xml_document_fragment.c +12 -16
  26. data/ext/nokogiri/xml_dtd.c +64 -58
  27. data/ext/nokogiri/xml_element_content.c +31 -26
  28. data/ext/nokogiri/xml_element_decl.c +25 -25
  29. data/ext/nokogiri/xml_encoding_handler.c +43 -18
  30. data/ext/nokogiri/xml_entity_decl.c +37 -35
  31. data/ext/nokogiri/xml_entity_reference.c +16 -18
  32. data/ext/nokogiri/xml_namespace.c +98 -53
  33. data/ext/nokogiri/xml_node.c +1065 -653
  34. data/ext/nokogiri/xml_node_set.c +178 -166
  35. data/ext/nokogiri/xml_processing_instruction.c +17 -19
  36. data/ext/nokogiri/xml_reader.c +277 -175
  37. data/ext/nokogiri/xml_relax_ng.c +52 -28
  38. data/ext/nokogiri/xml_sax_parser.c +112 -112
  39. data/ext/nokogiri/xml_sax_parser_context.c +112 -86
  40. data/ext/nokogiri/xml_sax_push_parser.c +36 -27
  41. data/ext/nokogiri/xml_schema.c +98 -48
  42. data/ext/nokogiri/xml_syntax_error.c +42 -21
  43. data/ext/nokogiri/xml_text.c +14 -18
  44. data/ext/nokogiri/xml_xpath_context.c +226 -115
  45. data/ext/nokogiri/xslt_stylesheet.c +265 -173
  46. data/gumbo-parser/CHANGES.md +63 -0
  47. data/gumbo-parser/Makefile +101 -0
  48. data/gumbo-parser/THANKS +27 -0
  49. data/gumbo-parser/src/Makefile +34 -0
  50. data/gumbo-parser/src/README.md +41 -0
  51. data/gumbo-parser/src/ascii.c +75 -0
  52. data/gumbo-parser/src/ascii.h +115 -0
  53. data/gumbo-parser/src/attribute.c +42 -0
  54. data/gumbo-parser/src/attribute.h +17 -0
  55. data/gumbo-parser/src/char_ref.c +22225 -0
  56. data/gumbo-parser/src/char_ref.h +29 -0
  57. data/gumbo-parser/src/char_ref.rl +2154 -0
  58. data/gumbo-parser/src/error.c +626 -0
  59. data/gumbo-parser/src/error.h +148 -0
  60. data/gumbo-parser/src/foreign_attrs.c +104 -0
  61. data/gumbo-parser/src/foreign_attrs.gperf +27 -0
  62. data/gumbo-parser/src/gumbo.h +943 -0
  63. data/gumbo-parser/src/insertion_mode.h +33 -0
  64. data/gumbo-parser/src/macros.h +91 -0
  65. data/gumbo-parser/src/parser.c +4875 -0
  66. data/gumbo-parser/src/parser.h +41 -0
  67. data/gumbo-parser/src/replacement.h +33 -0
  68. data/gumbo-parser/src/string_buffer.c +103 -0
  69. data/gumbo-parser/src/string_buffer.h +68 -0
  70. data/gumbo-parser/src/string_piece.c +48 -0
  71. data/gumbo-parser/src/svg_attrs.c +174 -0
  72. data/gumbo-parser/src/svg_attrs.gperf +77 -0
  73. data/gumbo-parser/src/svg_tags.c +137 -0
  74. data/gumbo-parser/src/svg_tags.gperf +55 -0
  75. data/gumbo-parser/src/tag.c +222 -0
  76. data/gumbo-parser/src/tag_lookup.c +382 -0
  77. data/gumbo-parser/src/tag_lookup.gperf +169 -0
  78. data/gumbo-parser/src/tag_lookup.h +13 -0
  79. data/gumbo-parser/src/token_buffer.c +79 -0
  80. data/gumbo-parser/src/token_buffer.h +71 -0
  81. data/gumbo-parser/src/token_type.h +17 -0
  82. data/gumbo-parser/src/tokenizer.c +3463 -0
  83. data/gumbo-parser/src/tokenizer.h +112 -0
  84. data/gumbo-parser/src/tokenizer_states.h +339 -0
  85. data/gumbo-parser/src/utf8.c +245 -0
  86. data/gumbo-parser/src/utf8.h +164 -0
  87. data/gumbo-parser/src/util.c +68 -0
  88. data/gumbo-parser/src/util.h +30 -0
  89. data/gumbo-parser/src/vector.c +111 -0
  90. data/gumbo-parser/src/vector.h +45 -0
  91. data/lib/nokogiri/class_resolver.rb +67 -0
  92. data/lib/nokogiri/css/node.rb +10 -8
  93. data/lib/nokogiri/css/parser.rb +397 -377
  94. data/lib/nokogiri/css/parser.y +250 -245
  95. data/lib/nokogiri/css/parser_extras.rb +54 -49
  96. data/lib/nokogiri/css/syntax_error.rb +3 -1
  97. data/lib/nokogiri/css/tokenizer.rb +5 -3
  98. data/lib/nokogiri/css/tokenizer.rex +3 -2
  99. data/lib/nokogiri/css/xpath_visitor.rb +218 -91
  100. data/lib/nokogiri/css.rb +50 -17
  101. data/lib/nokogiri/decorators/slop.rb +9 -7
  102. data/lib/nokogiri/extension.rb +31 -0
  103. data/lib/nokogiri/gumbo.rb +15 -0
  104. data/lib/nokogiri/html.rb +38 -27
  105. data/lib/nokogiri/{html → html4}/builder.rb +4 -2
  106. data/lib/nokogiri/{html → html4}/document.rb +103 -105
  107. data/lib/nokogiri/html4/document_fragment.rb +54 -0
  108. data/lib/nokogiri/{html → html4}/element_description.rb +3 -1
  109. data/lib/nokogiri/html4/element_description_defaults.rb +578 -0
  110. data/lib/nokogiri/{html → html4}/entity_lookup.rb +4 -2
  111. data/lib/nokogiri/{html → html4}/sax/parser.rb +17 -16
  112. data/lib/nokogiri/html4/sax/parser_context.rb +20 -0
  113. data/lib/nokogiri/{html → html4}/sax/push_parser.rb +12 -11
  114. data/lib/nokogiri/html4.rb +46 -0
  115. data/lib/nokogiri/html5/document.rb +91 -0
  116. data/lib/nokogiri/html5/document_fragment.rb +83 -0
  117. data/lib/nokogiri/html5/node.rb +100 -0
  118. data/lib/nokogiri/html5.rb +478 -0
  119. data/lib/nokogiri/jruby/dependencies.rb +21 -0
  120. data/lib/nokogiri/syntax_error.rb +2 -0
  121. data/lib/nokogiri/version/constant.rb +6 -0
  122. data/lib/nokogiri/version/info.rb +222 -0
  123. data/lib/nokogiri/version.rb +3 -108
  124. data/lib/nokogiri/xml/attr.rb +6 -3
  125. data/lib/nokogiri/xml/attribute_decl.rb +3 -1
  126. data/lib/nokogiri/xml/builder.rb +74 -33
  127. data/lib/nokogiri/xml/cdata.rb +3 -1
  128. data/lib/nokogiri/xml/character_data.rb +2 -0
  129. data/lib/nokogiri/xml/document.rb +224 -86
  130. data/lib/nokogiri/xml/document_fragment.rb +46 -44
  131. data/lib/nokogiri/xml/dtd.rb +4 -2
  132. data/lib/nokogiri/xml/element_content.rb +2 -0
  133. data/lib/nokogiri/xml/element_decl.rb +3 -1
  134. data/lib/nokogiri/xml/entity_decl.rb +4 -2
  135. data/lib/nokogiri/xml/entity_reference.rb +2 -0
  136. data/lib/nokogiri/xml/namespace.rb +3 -0
  137. data/lib/nokogiri/xml/node/save_options.rb +10 -5
  138. data/lib/nokogiri/xml/node.rb +884 -378
  139. data/lib/nokogiri/xml/node_set.rb +51 -54
  140. data/lib/nokogiri/xml/notation.rb +13 -0
  141. data/lib/nokogiri/xml/parse_options.rb +22 -8
  142. data/lib/nokogiri/xml/pp/character_data.rb +9 -6
  143. data/lib/nokogiri/xml/pp/node.rb +25 -26
  144. data/lib/nokogiri/xml/pp.rb +4 -2
  145. data/lib/nokogiri/xml/processing_instruction.rb +3 -1
  146. data/lib/nokogiri/xml/reader.rb +21 -28
  147. data/lib/nokogiri/xml/relax_ng.rb +8 -2
  148. data/lib/nokogiri/xml/sax/document.rb +45 -49
  149. data/lib/nokogiri/xml/sax/parser.rb +38 -34
  150. data/lib/nokogiri/xml/sax/parser_context.rb +8 -3
  151. data/lib/nokogiri/xml/sax/push_parser.rb +6 -5
  152. data/lib/nokogiri/xml/sax.rb +6 -4
  153. data/lib/nokogiri/xml/schema.rb +19 -9
  154. data/lib/nokogiri/xml/searchable.rb +112 -72
  155. data/lib/nokogiri/xml/syntax_error.rb +6 -4
  156. data/lib/nokogiri/xml/text.rb +2 -0
  157. data/lib/nokogiri/xml/xpath/syntax_error.rb +4 -2
  158. data/lib/nokogiri/xml/xpath.rb +15 -4
  159. data/lib/nokogiri/xml/xpath_context.rb +3 -3
  160. data/lib/nokogiri/xml.rb +38 -37
  161. data/lib/nokogiri/xslt/stylesheet.rb +3 -1
  162. data/lib/nokogiri/xslt.rb +29 -20
  163. data/lib/nokogiri.rb +49 -65
  164. data/lib/xsd/xmlparser/nokogiri.rb +26 -24
  165. data/patches/libxml2/{0002-Remove-script-macro-support.patch → 0001-Remove-script-macro-support.patch} +0 -0
  166. data/patches/libxml2/{0003-Update-entities-to-remove-handling-of-ssi.patch → 0002-Update-entities-to-remove-handling-of-ssi.patch} +0 -0
  167. data/patches/libxml2/{0004-libxml2.la-is-in-top_builddir.patch → 0003-libxml2.la-is-in-top_builddir.patch} +1 -1
  168. data/patches/libxml2/0005-avoid-isnan-isinf.patch +81 -0
  169. data/patches/libxml2/0009-allow-wildcard-namespaces.patch +77 -0
  170. data/patches/libxslt/0001-update-automake-files-for-arm64.patch +3037 -0
  171. data/ports/archives/libxml2-2.10.3.tar.xz +0 -0
  172. data/ports/archives/libxslt-1.1.37.tar.xz +0 -0
  173. metadata +189 -142
  174. data/ext/nokogiri/html_document.c +0 -170
  175. data/ext/nokogiri/html_document.h +0 -10
  176. data/ext/nokogiri/html_element_description.c +0 -279
  177. data/ext/nokogiri/html_element_description.h +0 -10
  178. data/ext/nokogiri/html_entity_lookup.c +0 -32
  179. data/ext/nokogiri/html_entity_lookup.h +0 -8
  180. data/ext/nokogiri/html_sax_parser_context.c +0 -116
  181. data/ext/nokogiri/html_sax_parser_context.h +0 -11
  182. data/ext/nokogiri/html_sax_push_parser.c +0 -87
  183. data/ext/nokogiri/html_sax_push_parser.h +0 -9
  184. data/ext/nokogiri/xml_attr.h +0 -9
  185. data/ext/nokogiri/xml_attribute_decl.h +0 -9
  186. data/ext/nokogiri/xml_cdata.h +0 -9
  187. data/ext/nokogiri/xml_comment.h +0 -9
  188. data/ext/nokogiri/xml_document.h +0 -23
  189. data/ext/nokogiri/xml_document_fragment.h +0 -10
  190. data/ext/nokogiri/xml_dtd.h +0 -10
  191. data/ext/nokogiri/xml_element_content.h +0 -10
  192. data/ext/nokogiri/xml_element_decl.h +0 -9
  193. data/ext/nokogiri/xml_encoding_handler.h +0 -8
  194. data/ext/nokogiri/xml_entity_decl.h +0 -10
  195. data/ext/nokogiri/xml_entity_reference.h +0 -9
  196. data/ext/nokogiri/xml_io.c +0 -61
  197. data/ext/nokogiri/xml_io.h +0 -11
  198. data/ext/nokogiri/xml_libxml2_hacks.c +0 -112
  199. data/ext/nokogiri/xml_libxml2_hacks.h +0 -12
  200. data/ext/nokogiri/xml_namespace.h +0 -14
  201. data/ext/nokogiri/xml_node.h +0 -13
  202. data/ext/nokogiri/xml_node_set.h +0 -12
  203. data/ext/nokogiri/xml_processing_instruction.h +0 -9
  204. data/ext/nokogiri/xml_reader.h +0 -10
  205. data/ext/nokogiri/xml_relax_ng.h +0 -9
  206. data/ext/nokogiri/xml_sax_parser.h +0 -39
  207. data/ext/nokogiri/xml_sax_parser_context.h +0 -10
  208. data/ext/nokogiri/xml_sax_push_parser.h +0 -9
  209. data/ext/nokogiri/xml_schema.h +0 -9
  210. data/ext/nokogiri/xml_syntax_error.h +0 -13
  211. data/ext/nokogiri/xml_text.h +0 -9
  212. data/ext/nokogiri/xml_xpath_context.h +0 -10
  213. data/ext/nokogiri/xslt_stylesheet.h +0 -14
  214. data/lib/nokogiri/html/document_fragment.rb +0 -49
  215. data/lib/nokogiri/html/element_description_defaults.rb +0 -671
  216. data/lib/nokogiri/html/sax/parser_context.rb +0 -16
  217. data/patches/libxml2/0001-Revert-Do-not-URI-escape-in-server-side-includes.patch +0 -78
  218. data/patches/libxml2/0005-Fix-infinite-loop-in-xmlStringLenDecodeEntities.patch +0 -32
  219. data/ports/archives/libxml2-2.9.10.tar.gz +0 -0
  220. data/ports/archives/libxslt-1.1.34.tar.gz +0 -0
@@ -1,4 +1,6 @@
1
- #include <xml_entity_decl.h>
1
+ #include <nokogiri.h>
2
+
3
+ VALUE cNokogiriXmlEntityDecl;
2
4
 
3
5
  /*
4
6
  * call-seq:
@@ -6,12 +8,13 @@
6
8
  *
7
9
  * Get the original_content before ref substitution
8
10
  */
9
- static VALUE original_content(VALUE self)
11
+ static VALUE
12
+ original_content(VALUE self)
10
13
  {
11
14
  xmlEntityPtr node;
12
- Data_Get_Struct(self, xmlEntity, node);
15
+ Noko_Node_Get_Struct(self, xmlEntity, node);
13
16
 
14
- if(!node->orig) return Qnil;
17
+ if (!node->orig) { return Qnil; }
15
18
 
16
19
  return NOKOGIRI_STR_NEW2(node->orig);
17
20
  }
@@ -22,12 +25,13 @@ static VALUE original_content(VALUE self)
22
25
  *
23
26
  * Get the content
24
27
  */
25
- static VALUE get_content(VALUE self)
28
+ static VALUE
29
+ get_content(VALUE self)
26
30
  {
27
31
  xmlEntityPtr node;
28
- Data_Get_Struct(self, xmlEntity, node);
32
+ Noko_Node_Get_Struct(self, xmlEntity, node);
29
33
 
30
- if(!node->content) return Qnil;
34
+ if (!node->content) { return Qnil; }
31
35
 
32
36
  return NOKOGIRI_STR_NEW(node->content, node->length);
33
37
  }
@@ -38,10 +42,11 @@ static VALUE get_content(VALUE self)
38
42
  *
39
43
  * Get the entity type
40
44
  */
41
- static VALUE entity_type(VALUE self)
45
+ static VALUE
46
+ entity_type(VALUE self)
42
47
  {
43
48
  xmlEntityPtr node;
44
- Data_Get_Struct(self, xmlEntity, node);
49
+ Noko_Node_Get_Struct(self, xmlEntity, node);
45
50
 
46
51
  return INT2NUM((int)node->etype);
47
52
  }
@@ -52,12 +57,13 @@ static VALUE entity_type(VALUE self)
52
57
  *
53
58
  * Get the external identifier for PUBLIC
54
59
  */
55
- static VALUE external_id(VALUE self)
60
+ static VALUE
61
+ external_id(VALUE self)
56
62
  {
57
63
  xmlEntityPtr node;
58
- Data_Get_Struct(self, xmlEntity, node);
64
+ Noko_Node_Get_Struct(self, xmlEntity, node);
59
65
 
60
- if(!node->ExternalID) return Qnil;
66
+ if (!node->ExternalID) { return Qnil; }
61
67
 
62
68
  return NOKOGIRI_STR_NEW2(node->ExternalID);
63
69
  }
@@ -68,43 +74,39 @@ static VALUE external_id(VALUE self)
68
74
  *
69
75
  * Get the URI for a SYSTEM or PUBLIC Entity
70
76
  */
71
- static VALUE system_id(VALUE self)
77
+ static VALUE
78
+ system_id(VALUE self)
72
79
  {
73
80
  xmlEntityPtr node;
74
- Data_Get_Struct(self, xmlEntity, node);
81
+ Noko_Node_Get_Struct(self, xmlEntity, node);
75
82
 
76
- if(!node->SystemID) return Qnil;
83
+ if (!node->SystemID) { return Qnil; }
77
84
 
78
85
  return NOKOGIRI_STR_NEW2(node->SystemID);
79
86
  }
80
87
 
81
- VALUE cNokogiriXmlEntityDecl;
82
-
83
- void init_xml_entity_decl()
88
+ void
89
+ noko_init_xml_entity_decl()
84
90
  {
85
- VALUE nokogiri = rb_define_module("Nokogiri");
86
- VALUE xml = rb_define_module_under(nokogiri, "XML");
87
- VALUE node = rb_define_class_under(xml, "Node", rb_cObject);
88
- VALUE klass = rb_define_class_under(xml, "EntityDecl", node);
89
-
90
- cNokogiriXmlEntityDecl = klass;
91
+ assert(cNokogiriXmlNode);
92
+ cNokogiriXmlEntityDecl = rb_define_class_under(mNokogiriXml, "EntityDecl", cNokogiriXmlNode);
91
93
 
92
- rb_define_method(klass, "original_content", original_content, 0);
93
- rb_define_method(klass, "content", get_content, 0);
94
- rb_define_method(klass, "entity_type", entity_type, 0);
95
- rb_define_method(klass, "external_id", external_id, 0);
96
- rb_define_method(klass, "system_id", system_id, 0);
94
+ rb_define_method(cNokogiriXmlEntityDecl, "original_content", original_content, 0);
95
+ rb_define_method(cNokogiriXmlEntityDecl, "content", get_content, 0);
96
+ rb_define_method(cNokogiriXmlEntityDecl, "entity_type", entity_type, 0);
97
+ rb_define_method(cNokogiriXmlEntityDecl, "external_id", external_id, 0);
98
+ rb_define_method(cNokogiriXmlEntityDecl, "system_id", system_id, 0);
97
99
 
98
100
  rb_const_set(cNokogiriXmlEntityDecl, rb_intern("INTERNAL_GENERAL"),
99
- INT2NUM(XML_INTERNAL_GENERAL_ENTITY));
101
+ INT2NUM(XML_INTERNAL_GENERAL_ENTITY));
100
102
  rb_const_set(cNokogiriXmlEntityDecl, rb_intern("EXTERNAL_GENERAL_PARSED"),
101
- INT2NUM(XML_EXTERNAL_GENERAL_PARSED_ENTITY));
103
+ INT2NUM(XML_EXTERNAL_GENERAL_PARSED_ENTITY));
102
104
  rb_const_set(cNokogiriXmlEntityDecl, rb_intern("EXTERNAL_GENERAL_UNPARSED"),
103
- INT2NUM(XML_EXTERNAL_GENERAL_UNPARSED_ENTITY));
105
+ INT2NUM(XML_EXTERNAL_GENERAL_UNPARSED_ENTITY));
104
106
  rb_const_set(cNokogiriXmlEntityDecl, rb_intern("INTERNAL_PARAMETER"),
105
- INT2NUM(XML_INTERNAL_PARAMETER_ENTITY));
107
+ INT2NUM(XML_INTERNAL_PARAMETER_ENTITY));
106
108
  rb_const_set(cNokogiriXmlEntityDecl, rb_intern("EXTERNAL_PARAMETER"),
107
- INT2NUM(XML_EXTERNAL_PARAMETER_ENTITY));
109
+ INT2NUM(XML_EXTERNAL_PARAMETER_ENTITY));
108
110
  rb_const_set(cNokogiriXmlEntityDecl, rb_intern("INTERNAL_PREDEFINED"),
109
- INT2NUM(XML_INTERNAL_PREDEFINED_ENTITY));
111
+ INT2NUM(XML_INTERNAL_PREDEFINED_ENTITY));
110
112
  }
@@ -1,4 +1,6 @@
1
- #include <xml_entity_reference.h>
1
+ #include <nokogiri.h>
2
+
3
+ VALUE cNokogiriXmlEntityReference;
2
4
 
3
5
  /*
4
6
  * call-seq:
@@ -6,7 +8,8 @@
6
8
  *
7
9
  * Create a new EntityReference element on the +document+ with +name+
8
10
  */
9
- static VALUE new(int argc, VALUE *argv, VALUE klass)
11
+ static VALUE
12
+ new (int argc, VALUE *argv, VALUE klass)
10
13
  {
11
14
  xmlDocPtr xml_doc;
12
15
  xmlNodePtr node;
@@ -20,33 +23,28 @@ static VALUE new(int argc, VALUE *argv, VALUE klass)
20
23
  Data_Get_Struct(document, xmlDoc, xml_doc);
21
24
 
22
25
  node = xmlNewReference(
23
- xml_doc,
24
- (const xmlChar *)StringValueCStr(name)
25
- );
26
+ xml_doc,
27
+ (const xmlChar *)StringValueCStr(name)
28
+ );
26
29
 
27
- nokogiri_root_node(node);
30
+ noko_xml_document_pin_node(node);
28
31
 
29
- rb_node = Nokogiri_wrap_xml_node(klass, node);
32
+ rb_node = noko_xml_node_wrap(klass, node);
30
33
  rb_obj_call_init(rb_node, argc, argv);
31
34
 
32
- if(rb_block_given_p()) rb_yield(rb_node);
35
+ if (rb_block_given_p()) { rb_yield(rb_node); }
33
36
 
34
37
  return rb_node;
35
38
  }
36
39
 
37
- VALUE cNokogiriXmlEntityReference;
38
- void init_xml_entity_reference()
40
+ void
41
+ noko_init_xml_entity_reference()
39
42
  {
40
- VALUE nokogiri = rb_define_module("Nokogiri");
41
- VALUE xml = rb_define_module_under(nokogiri, "XML");
42
- VALUE node = rb_define_class_under(xml, "Node", rb_cObject);
43
-
43
+ assert(cNokogiriXmlNode);
44
44
  /*
45
45
  * EntityReference represents an EntityReference node in an xml document.
46
46
  */
47
- VALUE klass = rb_define_class_under(xml, "EntityReference", node);
48
-
49
- cNokogiriXmlEntityReference = klass;
47
+ cNokogiriXmlEntityReference = rb_define_class_under(mNokogiriXml, "EntityReference", cNokogiriXmlNode);
50
48
 
51
- rb_define_singleton_method(klass, "new", new, -1);
49
+ rb_define_singleton_method(cNokogiriXmlEntityReference, "new", new, -1);
52
50
  }
@@ -1,28 +1,79 @@
1
- #include <xml_namespace.h>
1
+ #include <nokogiri.h>
2
+
3
+ /*
4
+ * The lifecycle of a Namespace node is more complicated than other Nodes, for two reasons:
5
+ *
6
+ * 1. the underlying C structure has a different layout than all the other node structs, with the
7
+ * `_private` member where we store a pointer to Ruby object data not being in first position.
8
+ * 2. xmlNs structures returned in an xmlNodeset from an XPath query are copies of the document's
9
+ * namespaces, and so do not share the same memory lifecycle as everything else in a document.
10
+ *
11
+ * As a result of 1, you may see special handling of XML_NAMESPACE_DECL node types throughout the
12
+ * Nokogiri C code, though I intend to wrap up that logic in ruby_object_{get,set} functions
13
+ * shortly.
14
+ *
15
+ * As a result of 2, you will see we have special handling in this file and in xml_node_set.c to
16
+ * carefully manage the memory lifecycle of xmlNs structs to match the Ruby object's GC
17
+ * lifecycle. In xml_node_set.c we have local versions of xmlXPathNodeSetDel() and
18
+ * xmlXPathFreeNodeSet() that avoid freeing xmlNs structs in the node set. In this file, we decide
19
+ * whether or not to call dealloc_namespace() depending on whether the xmlNs struct appears to be
20
+ * in an xmlNodeSet (and thus the result of an XPath query) or not.
21
+ *
22
+ * Yes, this is madness.
23
+ */
2
24
 
3
25
  VALUE cNokogiriXmlNamespace ;
4
26
 
5
- static void dealloc_namespace(xmlNsPtr ns)
27
+ static void
28
+ _xml_namespace_dealloc(void *ptr)
6
29
  {
7
30
  /*
8
- *
9
31
  * this deallocator is only used for namespace nodes that are part of an xpath
10
- * node set.
11
- *
12
- * see Nokogiri_wrap_xml_namespace() for more details.
13
- *
32
+ * node set. see noko_xml_namespace_wrap().
14
33
  */
34
+ xmlNsPtr ns = ptr;
15
35
  NOKOGIRI_DEBUG_START(ns) ;
36
+
16
37
  if (ns->href) {
17
- xmlFree((xmlChar *)(uintptr_t)ns->href);
38
+ xmlFree(DISCARD_CONST_QUAL_XMLCHAR(ns->href));
18
39
  }
19
40
  if (ns->prefix) {
20
- xmlFree((xmlChar *)(uintptr_t)ns->prefix);
41
+ xmlFree(DISCARD_CONST_QUAL_XMLCHAR(ns->prefix));
21
42
  }
22
43
  xmlFree(ns);
23
44
  NOKOGIRI_DEBUG_END(ns) ;
24
45
  }
25
46
 
47
+ #ifdef HAVE_RB_GC_LOCATION
48
+ static void
49
+ _xml_namespace_update_references(void *ptr)
50
+ {
51
+ xmlNsPtr ns = ptr;
52
+ if (ns->_private) {
53
+ ns->_private = (void *)rb_gc_location((VALUE)ns->_private);
54
+ }
55
+ }
56
+ #else
57
+ # define _xml_namespace_update_references 0
58
+ #endif
59
+
60
+ static const rb_data_type_t nokogiri_xml_namespace_type_with_dealloc = {
61
+ "Nokogiri/XMLNamespace/WithDealloc",
62
+ {0, _xml_namespace_dealloc, 0, _xml_namespace_update_references},
63
+ 0, 0,
64
+ #ifdef RUBY_TYPED_FREE_IMMEDIATELY
65
+ RUBY_TYPED_FREE_IMMEDIATELY,
66
+ #endif
67
+ };
68
+
69
+ static const rb_data_type_t nokogiri_xml_namespace_type_without_dealloc = {
70
+ "Nokogiri/XMLNamespace/WithoutDealloc",
71
+ {0, 0, 0, _xml_namespace_update_references},
72
+ 0, 0,
73
+ #ifdef RUBY_TYPED_FREE_IMMEDIATELY
74
+ RUBY_TYPED_FREE_IMMEDIATELY,
75
+ #endif
76
+ };
26
77
 
27
78
  /*
28
79
  * call-seq:
@@ -30,12 +81,13 @@ static void dealloc_namespace(xmlNsPtr ns)
30
81
  *
31
82
  * Get the prefix for this namespace. Returns +nil+ if there is no prefix.
32
83
  */
33
- static VALUE prefix(VALUE self)
84
+ static VALUE
85
+ prefix(VALUE self)
34
86
  {
35
87
  xmlNsPtr ns;
36
88
 
37
- Data_Get_Struct(self, xmlNs, ns);
38
- if(!ns->prefix) return Qnil;
89
+ Noko_Namespace_Get_Struct(self, xmlNs, ns);
90
+ if (!ns->prefix) { return Qnil; }
39
91
 
40
92
  return NOKOGIRI_STR_NEW2(ns->prefix);
41
93
  }
@@ -46,66 +98,59 @@ static VALUE prefix(VALUE self)
46
98
  *
47
99
  * Get the href for this namespace
48
100
  */
49
- static VALUE href(VALUE self)
101
+ static VALUE
102
+ href(VALUE self)
50
103
  {
51
104
  xmlNsPtr ns;
52
105
 
53
- Data_Get_Struct(self, xmlNs, ns);
54
- if(!ns->href) return Qnil;
106
+ Noko_Namespace_Get_Struct(self, xmlNs, ns);
107
+ if (!ns->href) { return Qnil; }
55
108
 
56
109
  return NOKOGIRI_STR_NEW2(ns->href);
57
110
  }
58
111
 
59
- static int part_of_an_xpath_node_set_eh(xmlNsPtr node)
112
+ VALUE
113
+ noko_xml_namespace_wrap(xmlNsPtr c_namespace, xmlDocPtr c_document)
60
114
  {
61
- return (node->next && ! NOKOGIRI_NAMESPACE_EH(node->next));
62
- }
63
-
64
- VALUE Nokogiri_wrap_xml_namespace(xmlDocPtr doc, xmlNsPtr node)
65
- {
66
- VALUE ns = 0, document, node_cache;
67
-
68
- assert(doc->type == XML_DOCUMENT_NODE || doc->type == XML_HTML_DOCUMENT_NODE);
69
-
70
- if (node->_private) return (VALUE)node->_private;
115
+ VALUE rb_namespace;
71
116
 
72
- if (doc->type == XML_DOCUMENT_FRAG_NODE) doc = doc->doc;
117
+ if (c_namespace->_private) {
118
+ return (VALUE)c_namespace->_private;
119
+ }
73
120
 
74
- if (DOC_RUBY_OBJECT_TEST(doc)) {
75
- document = DOC_RUBY_OBJECT(doc);
121
+ if (c_document) {
122
+ rb_namespace = TypedData_Wrap_Struct(cNokogiriXmlNamespace,
123
+ &nokogiri_xml_namespace_type_without_dealloc,
124
+ c_namespace);
76
125
 
77
- if (part_of_an_xpath_node_set_eh(node)) {
78
- /*
79
- * this is a duplicate returned as part of an xpath query node set, and so
80
- * we need to make sure we manage this memory.
81
- *
82
- * see comments in xml_node_set.c for more details.
83
- */
84
- ns = Data_Wrap_Struct(cNokogiriXmlNamespace, 0, dealloc_namespace, node);
85
- } else {
86
- ns = Data_Wrap_Struct(cNokogiriXmlNamespace, 0, 0, node);
87
- node_cache = rb_iv_get(document, "@node_cache");
88
- rb_ary_push(node_cache, ns);
126
+ if (DOC_RUBY_OBJECT_TEST(c_document)) {
127
+ rb_iv_set(rb_namespace, "@document", DOC_RUBY_OBJECT(c_document));
128
+ rb_ary_push(DOC_NODE_CACHE(c_document), rb_namespace);
89
129
  }
90
-
91
- rb_iv_set(ns, "@document", document);
92
130
  } else {
93
- ns = Data_Wrap_Struct(cNokogiriXmlNamespace, 0, 0, node);
131
+ rb_namespace = TypedData_Wrap_Struct(cNokogiriXmlNamespace,
132
+ &nokogiri_xml_namespace_type_with_dealloc,
133
+ c_namespace);
94
134
  }
95
135
 
96
- node->_private = (void *)ns;
136
+ c_namespace->_private = (void *)rb_namespace;
97
137
 
98
- return ns;
138
+ return rb_namespace;
139
+ }
140
+
141
+ VALUE
142
+ noko_xml_namespace_wrap_xpath_copy(xmlNsPtr c_namespace)
143
+ {
144
+ return noko_xml_namespace_wrap(c_namespace, NULL);
99
145
  }
100
146
 
101
- void init_xml_namespace()
147
+ void
148
+ noko_init_xml_namespace()
102
149
  {
103
- VALUE nokogiri = rb_define_module("Nokogiri");
104
- VALUE xml = rb_define_module_under(nokogiri, "XML");
105
- VALUE klass = rb_define_class_under(xml, "Namespace", rb_cObject);
150
+ cNokogiriXmlNamespace = rb_define_class_under(mNokogiriXml, "Namespace", rb_cObject);
106
151
 
107
- cNokogiriXmlNamespace = klass;
152
+ rb_undef_alloc_func(cNokogiriXmlNamespace);
108
153
 
109
- rb_define_method(klass, "prefix", prefix, 0);
110
- rb_define_method(klass, "href", href, 0);
154
+ rb_define_method(cNokogiriXmlNamespace, "prefix", prefix, 0);
155
+ rb_define_method(cNokogiriXmlNamespace, "href", href, 0);
111
156
  }