nokogiri 1.9.1 → 1.15.3

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 (226) hide show
  1. checksums.yaml +4 -4
  2. data/Gemfile +45 -0
  3. data/LICENSE-DEPENDENCIES.md +1636 -1024
  4. data/LICENSE.md +5 -28
  5. data/README.md +203 -89
  6. data/bin/nokogiri +63 -50
  7. data/dependencies.yml +33 -61
  8. data/ext/nokogiri/depend +38 -358
  9. data/ext/nokogiri/extconf.rb +864 -418
  10. data/ext/nokogiri/gumbo.c +594 -0
  11. data/ext/nokogiri/html4_document.c +165 -0
  12. data/ext/nokogiri/html4_element_description.c +299 -0
  13. data/ext/nokogiri/html4_entity_lookup.c +37 -0
  14. data/ext/nokogiri/html4_sax_parser_context.c +108 -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 +251 -105
  18. data/ext/nokogiri/nokogiri.h +215 -90
  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 +22 -22
  22. data/ext/nokogiri/xml_cdata.c +40 -31
  23. data/ext/nokogiri/xml_comment.c +20 -27
  24. data/ext/nokogiri/xml_document.c +401 -240
  25. data/ext/nokogiri/xml_document_fragment.c +13 -17
  26. data/ext/nokogiri/xml_dtd.c +64 -58
  27. data/ext/nokogiri/xml_element_content.c +63 -55
  28. data/ext/nokogiri/xml_element_decl.c +31 -31
  29. data/ext/nokogiri/xml_encoding_handler.c +54 -21
  30. data/ext/nokogiri/xml_entity_decl.c +37 -35
  31. data/ext/nokogiri/xml_entity_reference.c +17 -19
  32. data/ext/nokogiri/xml_namespace.c +135 -61
  33. data/ext/nokogiri/xml_node.c +1346 -677
  34. data/ext/nokogiri/xml_node_set.c +246 -216
  35. data/ext/nokogiri/xml_processing_instruction.c +18 -20
  36. data/ext/nokogiri/xml_reader.c +347 -212
  37. data/ext/nokogiri/xml_relax_ng.c +86 -77
  38. data/ext/nokogiri/xml_sax_parser.c +149 -124
  39. data/ext/nokogiri/xml_sax_parser_context.c +145 -103
  40. data/ext/nokogiri/xml_sax_push_parser.c +64 -36
  41. data/ext/nokogiri/xml_schema.c +138 -81
  42. data/ext/nokogiri/xml_syntax_error.c +42 -21
  43. data/ext/nokogiri/xml_text.c +36 -26
  44. data/ext/nokogiri/xml_xpath_context.c +366 -178
  45. data/ext/nokogiri/xslt_stylesheet.c +335 -189
  46. data/gumbo-parser/CHANGES.md +63 -0
  47. data/gumbo-parser/Makefile +111 -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 +630 -0
  59. data/gumbo-parser/src/error.h +148 -0
  60. data/gumbo-parser/src/foreign_attrs.c +103 -0
  61. data/gumbo-parser/src/foreign_attrs.gperf +27 -0
  62. data/gumbo-parser/src/insertion_mode.h +33 -0
  63. data/gumbo-parser/src/macros.h +91 -0
  64. data/gumbo-parser/src/nokogiri_gumbo.h +944 -0
  65. data/gumbo-parser/src/parser.c +4891 -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 +223 -0
  76. data/gumbo-parser/src/tag_lookup.c +382 -0
  77. data/gumbo-parser/src/tag_lookup.gperf +170 -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 +66 -0
  88. data/gumbo-parser/src/util.h +34 -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 +107 -104
  98. data/lib/nokogiri/css/tokenizer.rex +3 -2
  99. data/lib/nokogiri/css/xpath_visitor.rb +224 -95
  100. data/lib/nokogiri/css.rb +56 -17
  101. data/lib/nokogiri/decorators/slop.rb +9 -7
  102. data/lib/nokogiri/encoding_handler.rb +57 -0
  103. data/lib/nokogiri/extension.rb +32 -0
  104. data/lib/nokogiri/gumbo.rb +15 -0
  105. data/lib/nokogiri/html.rb +38 -27
  106. data/lib/nokogiri/{html → html4}/builder.rb +4 -2
  107. data/lib/nokogiri/html4/document.rb +214 -0
  108. data/lib/nokogiri/html4/document_fragment.rb +54 -0
  109. data/lib/nokogiri/{html → html4}/element_description.rb +3 -1
  110. data/lib/nokogiri/html4/element_description_defaults.rb +2040 -0
  111. data/lib/nokogiri/html4/encoding_reader.rb +121 -0
  112. data/lib/nokogiri/{html → html4}/entity_lookup.rb +4 -2
  113. data/lib/nokogiri/{html → html4}/sax/parser.rb +17 -16
  114. data/lib/nokogiri/html4/sax/parser_context.rb +20 -0
  115. data/lib/nokogiri/{html → html4}/sax/push_parser.rb +12 -11
  116. data/lib/nokogiri/html4.rb +47 -0
  117. data/lib/nokogiri/html5/document.rb +168 -0
  118. data/lib/nokogiri/html5/document_fragment.rb +90 -0
  119. data/lib/nokogiri/html5/node.rb +103 -0
  120. data/lib/nokogiri/html5.rb +392 -0
  121. data/lib/nokogiri/jruby/dependencies.rb +3 -0
  122. data/lib/nokogiri/jruby/nokogiri_jars.rb +43 -0
  123. data/lib/nokogiri/syntax_error.rb +2 -0
  124. data/lib/nokogiri/version/constant.rb +6 -0
  125. data/lib/nokogiri/version/info.rb +223 -0
  126. data/lib/nokogiri/version.rb +3 -108
  127. data/lib/nokogiri/xml/attr.rb +55 -3
  128. data/lib/nokogiri/xml/attribute_decl.rb +6 -2
  129. data/lib/nokogiri/xml/builder.rb +98 -54
  130. data/lib/nokogiri/xml/cdata.rb +3 -1
  131. data/lib/nokogiri/xml/character_data.rb +2 -0
  132. data/lib/nokogiri/xml/document.rb +312 -126
  133. data/lib/nokogiri/xml/document_fragment.rb +93 -48
  134. data/lib/nokogiri/xml/dtd.rb +4 -2
  135. data/lib/nokogiri/xml/element_content.rb +12 -2
  136. data/lib/nokogiri/xml/element_decl.rb +6 -2
  137. data/lib/nokogiri/xml/entity_decl.rb +7 -3
  138. data/lib/nokogiri/xml/entity_reference.rb +2 -0
  139. data/lib/nokogiri/xml/namespace.rb +45 -0
  140. data/lib/nokogiri/xml/node/save_options.rb +23 -8
  141. data/lib/nokogiri/xml/node.rb +1088 -418
  142. data/lib/nokogiri/xml/node_set.rb +173 -63
  143. data/lib/nokogiri/xml/notation.rb +13 -0
  144. data/lib/nokogiri/xml/parse_options.rb +145 -52
  145. data/lib/nokogiri/xml/pp/character_data.rb +9 -6
  146. data/lib/nokogiri/xml/pp/node.rb +42 -30
  147. data/lib/nokogiri/xml/pp.rb +4 -2
  148. data/lib/nokogiri/xml/processing_instruction.rb +4 -1
  149. data/lib/nokogiri/xml/reader.rb +21 -28
  150. data/lib/nokogiri/xml/relax_ng.rb +8 -2
  151. data/lib/nokogiri/xml/sax/document.rb +45 -49
  152. data/lib/nokogiri/xml/sax/parser.rb +39 -36
  153. data/lib/nokogiri/xml/sax/parser_context.rb +8 -3
  154. data/lib/nokogiri/xml/sax/push_parser.rb +6 -5
  155. data/lib/nokogiri/xml/sax.rb +6 -4
  156. data/lib/nokogiri/xml/schema.rb +19 -9
  157. data/lib/nokogiri/xml/searchable.rb +120 -72
  158. data/lib/nokogiri/xml/syntax_error.rb +6 -4
  159. data/lib/nokogiri/xml/text.rb +2 -0
  160. data/lib/nokogiri/xml/xpath/syntax_error.rb +4 -2
  161. data/lib/nokogiri/xml/xpath.rb +15 -4
  162. data/lib/nokogiri/xml/xpath_context.rb +3 -3
  163. data/lib/nokogiri/xml.rb +38 -37
  164. data/lib/nokogiri/xslt/stylesheet.rb +3 -1
  165. data/lib/nokogiri/xslt.rb +101 -22
  166. data/lib/nokogiri.rb +59 -75
  167. data/lib/xsd/xmlparser/nokogiri.rb +29 -25
  168. data/patches/libxml2/0001-Remove-script-macro-support.patch +40 -0
  169. data/patches/libxml2/0002-Update-entities-to-remove-handling-of-ssi.patch +44 -0
  170. data/patches/libxml2/0003-libxml2.la-is-in-top_builddir.patch +25 -0
  171. data/patches/libxml2/0009-allow-wildcard-namespaces.patch +77 -0
  172. data/patches/libxml2/0010-update-config.guess-and-config.sub-for-libxml2.patch +224 -0
  173. data/patches/libxml2/0011-rip-out-libxml2-s-libc_single_threaded-support.patch +30 -0
  174. data/patches/libxslt/0001-update-config.guess-and-config.sub-for-libxslt.patch +224 -0
  175. data/ports/archives/libxml2-2.11.4.tar.xz +0 -0
  176. data/ports/archives/libxslt-1.1.38.tar.xz +0 -0
  177. metadata +128 -265
  178. data/ext/nokogiri/html_document.c +0 -170
  179. data/ext/nokogiri/html_document.h +0 -10
  180. data/ext/nokogiri/html_element_description.c +0 -279
  181. data/ext/nokogiri/html_element_description.h +0 -10
  182. data/ext/nokogiri/html_entity_lookup.c +0 -32
  183. data/ext/nokogiri/html_entity_lookup.h +0 -8
  184. data/ext/nokogiri/html_sax_parser_context.c +0 -116
  185. data/ext/nokogiri/html_sax_parser_context.h +0 -11
  186. data/ext/nokogiri/html_sax_push_parser.c +0 -87
  187. data/ext/nokogiri/html_sax_push_parser.h +0 -9
  188. data/ext/nokogiri/xml_attr.h +0 -9
  189. data/ext/nokogiri/xml_attribute_decl.h +0 -9
  190. data/ext/nokogiri/xml_cdata.h +0 -9
  191. data/ext/nokogiri/xml_comment.h +0 -9
  192. data/ext/nokogiri/xml_document.h +0 -23
  193. data/ext/nokogiri/xml_document_fragment.h +0 -10
  194. data/ext/nokogiri/xml_dtd.h +0 -10
  195. data/ext/nokogiri/xml_element_content.h +0 -10
  196. data/ext/nokogiri/xml_element_decl.h +0 -9
  197. data/ext/nokogiri/xml_encoding_handler.h +0 -8
  198. data/ext/nokogiri/xml_entity_decl.h +0 -10
  199. data/ext/nokogiri/xml_entity_reference.h +0 -9
  200. data/ext/nokogiri/xml_io.c +0 -61
  201. data/ext/nokogiri/xml_io.h +0 -11
  202. data/ext/nokogiri/xml_libxml2_hacks.c +0 -112
  203. data/ext/nokogiri/xml_libxml2_hacks.h +0 -12
  204. data/ext/nokogiri/xml_namespace.h +0 -14
  205. data/ext/nokogiri/xml_node.h +0 -13
  206. data/ext/nokogiri/xml_node_set.h +0 -12
  207. data/ext/nokogiri/xml_processing_instruction.h +0 -9
  208. data/ext/nokogiri/xml_reader.h +0 -10
  209. data/ext/nokogiri/xml_relax_ng.h +0 -9
  210. data/ext/nokogiri/xml_sax_parser.h +0 -39
  211. data/ext/nokogiri/xml_sax_parser_context.h +0 -10
  212. data/ext/nokogiri/xml_sax_push_parser.h +0 -9
  213. data/ext/nokogiri/xml_schema.h +0 -9
  214. data/ext/nokogiri/xml_syntax_error.h +0 -13
  215. data/ext/nokogiri/xml_text.h +0 -9
  216. data/ext/nokogiri/xml_xpath_context.h +0 -10
  217. data/ext/nokogiri/xslt_stylesheet.h +0 -14
  218. data/lib/nokogiri/html/document.rb +0 -335
  219. data/lib/nokogiri/html/document_fragment.rb +0 -49
  220. data/lib/nokogiri/html/element_description_defaults.rb +0 -671
  221. data/lib/nokogiri/html/sax/parser_context.rb +0 -16
  222. data/patches/libxml2/0001-Revert-Do-not-URI-escape-in-server-side-includes.patch +0 -78
  223. data/patches/libxml2/0002-Fix-nullptr-deref-with-XPath-logic-ops.patch +0 -54
  224. data/patches/libxml2/0003-Fix-infinite-loop-in-LZMA-decompression.patch +0 -50
  225. data/ports/archives/libxml2-2.9.8.tar.gz +0 -0
  226. data/ports/archives/libxslt-1.1.32.tar.gz +0 -0
@@ -1,36 +1,61 @@
1
- #include <xml_node.h>
1
+ #include <nokogiri.h>
2
2
 
3
- static ID decorate, decorate_bang;
3
+ #include <stdbool.h>
4
4
 
5
- #ifdef DEBUG
6
- static void debug_node_dealloc(xmlNodePtr x)
7
- {
8
- NOKOGIRI_DEBUG_START(x)
9
- NOKOGIRI_DEBUG_END(x)
10
- }
11
- #else
12
- # define debug_node_dealloc 0
13
- #endif
5
+ // :stopdoc:
14
6
 
15
- static void mark(xmlNodePtr node)
7
+ VALUE cNokogiriXmlNode ;
8
+ static ID id_decorate, id_decorate_bang;
9
+
10
+ typedef xmlNodePtr(*pivot_reparentee_func)(xmlNodePtr, xmlNodePtr);
11
+
12
+ static void
13
+ _xml_node_mark(void *ptr)
16
14
  {
15
+ xmlNodePtr node = ptr;
16
+
17
+ if (!DOC_RUBY_OBJECT_TEST(node->doc)) {
18
+ return;
19
+ }
20
+
17
21
  xmlDocPtr doc = node->doc;
18
- if(doc->type == XML_DOCUMENT_NODE || doc->type == XML_HTML_DOCUMENT_NODE) {
19
- if(DOC_RUBY_OBJECT_TEST(doc)) {
22
+ if (doc->type == XML_DOCUMENT_NODE || doc->type == XML_HTML_DOCUMENT_NODE) {
23
+ if (DOC_RUBY_OBJECT_TEST(doc)) {
20
24
  rb_gc_mark(DOC_RUBY_OBJECT(doc));
21
25
  }
22
- } else if(node->doc->_private) {
26
+ } else if (node->doc->_private) {
23
27
  rb_gc_mark((VALUE)doc->_private);
24
28
  }
25
29
  }
26
30
 
27
- /* :nodoc: */
28
- typedef xmlNodePtr (*pivot_reparentee_func)(xmlNodePtr, xmlNodePtr);
31
+ #ifdef HAVE_RB_GC_LOCATION
32
+ static void
33
+ _xml_node_update_references(void *ptr)
34
+ {
35
+ xmlNodePtr node = ptr;
29
36
 
30
- /* :nodoc: */
31
- static void relink_namespace(xmlNodePtr reparented)
37
+ if (node->_private) {
38
+ node->_private = (void *)rb_gc_location((VALUE)node->_private);
39
+ }
40
+ }
41
+ #else
42
+ # define _xml_node_update_references 0
43
+ #endif
44
+
45
+ static const rb_data_type_t nokogiri_node_type = {
46
+ .wrap_struct_name = "Nokogiri::XML::Node",
47
+ .function = {
48
+ .dmark = _xml_node_mark,
49
+ .dcompact = _xml_node_update_references,
50
+ },
51
+ .flags = RUBY_TYPED_FREE_IMMEDIATELY,
52
+ };
53
+
54
+ static void
55
+ relink_namespace(xmlNodePtr reparented)
32
56
  {
33
57
  xmlNodePtr child;
58
+ xmlAttrPtr attr;
34
59
 
35
60
  if (reparented->type != XML_ATTRIBUTE_NODE &&
36
61
  reparented->type != XML_ELEMENT_NODE) { return; }
@@ -42,7 +67,7 @@ static void relink_namespace(xmlNodePtr reparented)
42
67
  name = xmlSplitQName2(reparented->name, &prefix);
43
68
 
44
69
  if (reparented->type == XML_ATTRIBUTE_NODE) {
45
- if (prefix == NULL || strcmp((char*)prefix, XMLNS_PREFIX) == 0) {
70
+ if (prefix == NULL || strcmp((char *)prefix, XMLNS_PREFIX) == 0) {
46
71
  xmlFree(name);
47
72
  xmlFree(prefix);
48
73
  return;
@@ -51,10 +76,6 @@ static void relink_namespace(xmlNodePtr reparented)
51
76
 
52
77
  ns = xmlSearchNs(reparented->doc, reparented, prefix);
53
78
 
54
- if (ns == NULL && reparented->parent) {
55
- ns = xmlSearchNs(reparented->doc, reparented->parent, prefix);
56
- }
57
-
58
79
  if (ns != NULL) {
59
80
  xmlNodeSetName(reparented, name);
60
81
  xmlSetNs(reparented, ns);
@@ -68,7 +89,9 @@ static void relink_namespace(xmlNodePtr reparented)
68
89
  if (reparented->type != XML_ELEMENT_NODE || !reparented->parent) { return; }
69
90
 
70
91
  /* Make sure that our reparented node has the correct namespaces */
71
- if (!reparented->ns && reparented->doc != (xmlDocPtr)reparented->parent) {
92
+ if (!reparented->ns &&
93
+ (reparented->doc != (xmlDocPtr)reparented->parent) &&
94
+ (rb_iv_get(DOC_RUBY_OBJECT(reparented->doc), "@namespace_inheritance") == Qtrue)) {
72
95
  xmlSetNs(reparented, reparented->parent->ns);
73
96
  }
74
97
 
@@ -91,7 +114,7 @@ static void relink_namespace(xmlNodePtr reparented)
91
114
  } else {
92
115
  reparented->nsDef = curr->next;
93
116
  }
94
- nokogiri_root_nsdef(curr, reparented->doc);
117
+ noko_xml_document_pin_namespace(curr, reparented->doc);
95
118
  } else {
96
119
  prev = curr;
97
120
  }
@@ -131,16 +154,19 @@ static void relink_namespace(xmlNodePtr reparented)
131
154
  }
132
155
 
133
156
  if (reparented->type == XML_ELEMENT_NODE) {
134
- child = (xmlNodePtr)((xmlElementPtr)reparented)->attributes;
135
- while(NULL != child) {
136
- relink_namespace(child);
137
- child = child->next;
157
+ attr = reparented->properties;
158
+ while (NULL != attr) {
159
+ relink_namespace((xmlNodePtr)attr);
160
+ attr = attr->next;
138
161
  }
139
162
  }
140
163
  }
141
164
 
142
- /* :nodoc: */
143
- static xmlNodePtr xmlReplaceNodeWrapper(xmlNodePtr pivot, xmlNodePtr new_node)
165
+
166
+ /* internal function meant to wrap xmlReplaceNode
167
+ and fix some issues we have with libxml2 merging nodes */
168
+ static xmlNodePtr
169
+ xmlReplaceNodeWrapper(xmlNodePtr pivot, xmlNodePtr new_node)
144
170
  {
145
171
  xmlNodePtr retval ;
146
172
 
@@ -163,22 +189,34 @@ static xmlNodePtr xmlReplaceNodeWrapper(xmlNodePtr pivot, xmlNodePtr new_node)
163
189
  return retval ;
164
190
  }
165
191
 
166
- /* :nodoc: */
167
- static VALUE reparent_node_with(VALUE pivot_obj, VALUE reparentee_obj, pivot_reparentee_func prf)
192
+
193
+ static void
194
+ raise_if_ancestor_of_self(xmlNodePtr self)
195
+ {
196
+ for (xmlNodePtr ancestor = self->parent ; ancestor ; ancestor = ancestor->parent) {
197
+ if (self == ancestor) {
198
+ rb_raise(rb_eRuntimeError, "cycle detected: node '%s' is an ancestor of itself", self->name);
199
+ }
200
+ }
201
+ }
202
+
203
+
204
+ static VALUE
205
+ reparent_node_with(VALUE pivot_obj, VALUE reparentee_obj, pivot_reparentee_func prf)
168
206
  {
169
207
  VALUE reparented_obj ;
170
- xmlNodePtr reparentee, pivot, reparented, next_text, new_next_text, parent ;
208
+ xmlNodePtr reparentee, original_reparentee, pivot, reparented, next_text, new_next_text, parent ;
171
209
  int original_ns_prefix_is_default = 0 ;
172
210
 
173
- if(!rb_obj_is_kind_of(reparentee_obj, cNokogiriXmlNode)) {
211
+ if (!rb_obj_is_kind_of(reparentee_obj, cNokogiriXmlNode)) {
174
212
  rb_raise(rb_eArgError, "node must be a Nokogiri::XML::Node");
175
213
  }
176
- if(rb_obj_is_kind_of(reparentee_obj, cNokogiriXmlDocument)) {
214
+ if (rb_obj_is_kind_of(reparentee_obj, cNokogiriXmlDocument)) {
177
215
  rb_raise(rb_eArgError, "node must be a Nokogiri::XML::Node");
178
216
  }
179
217
 
180
- Data_Get_Struct(reparentee_obj, xmlNode, reparentee);
181
- Data_Get_Struct(pivot_obj, xmlNode, pivot);
218
+ Noko_Node_Get_Struct(reparentee_obj, xmlNode, reparentee);
219
+ Noko_Node_Get_Struct(pivot_obj, xmlNode, pivot);
182
220
 
183
221
  /*
184
222
  * Check if nodes given are appropriate to have a parent-child
@@ -194,66 +232,66 @@ static VALUE reparent_node_with(VALUE pivot_obj, VALUE reparentee_obj, pivot_rep
194
232
 
195
233
  if (parent) {
196
234
  switch (parent->type) {
197
- case XML_DOCUMENT_NODE:
198
- case XML_HTML_DOCUMENT_NODE:
199
- switch (reparentee->type) {
200
- case XML_ELEMENT_NODE:
201
- case XML_PI_NODE:
202
- case XML_COMMENT_NODE:
203
- case XML_DOCUMENT_TYPE_NODE:
204
- /*
205
- * The DOM specification says no to adding text-like nodes
206
- * directly to a document, but we allow it for compatibility.
207
- */
208
- case XML_TEXT_NODE:
209
- case XML_CDATA_SECTION_NODE:
210
- case XML_ENTITY_REF_NODE:
211
- goto ok;
212
- default:
235
+ case XML_DOCUMENT_NODE:
236
+ case XML_HTML_DOCUMENT_NODE:
237
+ switch (reparentee->type) {
238
+ case XML_ELEMENT_NODE:
239
+ case XML_PI_NODE:
240
+ case XML_COMMENT_NODE:
241
+ case XML_DOCUMENT_TYPE_NODE:
242
+ /*
243
+ * The DOM specification says no to adding text-like nodes
244
+ * directly to a document, but we allow it for compatibility.
245
+ */
246
+ case XML_TEXT_NODE:
247
+ case XML_CDATA_SECTION_NODE:
248
+ case XML_ENTITY_REF_NODE:
249
+ goto ok;
250
+ default:
251
+ break;
252
+ }
213
253
  break;
214
- }
215
- break;
216
- case XML_DOCUMENT_FRAG_NODE:
217
- case XML_ENTITY_REF_NODE:
218
- case XML_ELEMENT_NODE:
219
- switch (reparentee->type) {
220
- case XML_ELEMENT_NODE:
221
- case XML_PI_NODE:
222
- case XML_COMMENT_NODE:
223
- case XML_TEXT_NODE:
224
- case XML_CDATA_SECTION_NODE:
254
+ case XML_DOCUMENT_FRAG_NODE:
225
255
  case XML_ENTITY_REF_NODE:
226
- goto ok;
227
- default:
256
+ case XML_ELEMENT_NODE:
257
+ switch (reparentee->type) {
258
+ case XML_ELEMENT_NODE:
259
+ case XML_PI_NODE:
260
+ case XML_COMMENT_NODE:
261
+ case XML_TEXT_NODE:
262
+ case XML_CDATA_SECTION_NODE:
263
+ case XML_ENTITY_REF_NODE:
264
+ goto ok;
265
+ default:
266
+ break;
267
+ }
268
+ break;
269
+ case XML_ATTRIBUTE_NODE:
270
+ switch (reparentee->type) {
271
+ case XML_TEXT_NODE:
272
+ case XML_ENTITY_REF_NODE:
273
+ goto ok;
274
+ default:
275
+ break;
276
+ }
228
277
  break;
229
- }
230
- break;
231
- case XML_ATTRIBUTE_NODE:
232
- switch (reparentee->type) {
233
278
  case XML_TEXT_NODE:
234
- case XML_ENTITY_REF_NODE:
235
- goto ok;
279
+ /*
280
+ * xmlAddChild() breaks the DOM specification in that it allows
281
+ * adding a text node to another, in which case text nodes are
282
+ * coalesced, but since our JRuby version does not support such
283
+ * operation, we should inhibit it.
284
+ */
285
+ break;
236
286
  default:
237
287
  break;
238
- }
239
- break;
240
- case XML_TEXT_NODE:
241
- /*
242
- * xmlAddChild() breaks the DOM specification in that it allows
243
- * adding a text node to another, in which case text nodes are
244
- * coalesced, but since our JRuby version does not support such
245
- * operation, we should inhibit it.
246
- */
247
- break;
248
- default:
249
- break;
250
288
  }
251
289
 
252
290
  rb_raise(rb_eArgError, "cannot reparent %s there", rb_obj_classname(reparentee_obj));
253
291
  }
254
292
 
255
293
  ok:
256
- xmlUnlinkNode(reparentee);
294
+ original_reparentee = reparentee;
257
295
 
258
296
  if (reparentee->doc != pivot->doc || reparentee->type == XML_TEXT_NODE) {
259
297
  /*
@@ -294,7 +332,7 @@ ok:
294
332
  original_ns_prefix_is_default = 1;
295
333
  }
296
334
 
297
- nokogiri_root_node(reparentee);
335
+ noko_xml_document_pin_node(reparentee);
298
336
 
299
337
  if (!(reparentee = xmlDocCopyNode(reparentee, pivot->doc, 1))) {
300
338
  rb_raise(rb_eRuntimeError, "Could not reparent node (xmlDocCopyNode)");
@@ -305,12 +343,14 @@ ok:
305
343
  * issue #391, where new node's prefix may become the string "default"
306
344
  * see libxml2 tree.c xmlNewReconciliedNs which implements this behavior.
307
345
  */
308
- xmlFree(reparentee->ns->prefix);
346
+ xmlFree(DISCARD_CONST_QUAL_XMLCHAR(reparentee->ns->prefix));
309
347
  reparentee->ns->prefix = NULL;
310
348
  }
311
349
  }
312
350
 
313
- if (prf != xmlAddPrevSibling && prf != xmlAddNextSibling
351
+ xmlUnlinkNode(original_reparentee);
352
+
353
+ if (prf != xmlAddPrevSibling && prf != xmlAddNextSibling && prf != xmlAddChild
314
354
  && reparentee->type == XML_TEXT_NODE && pivot->next && pivot->next->type == XML_TEXT_NODE) {
315
355
  /*
316
356
  * libxml merges text nodes in a right-to-left fashion, meaning that if
@@ -334,12 +374,12 @@ ok:
334
374
  new_next_text = xmlDocCopyNode(next_text, pivot->doc, 1) ;
335
375
 
336
376
  xmlUnlinkNode(next_text);
337
- nokogiri_root_node(next_text);
377
+ noko_xml_document_pin_node(next_text);
338
378
 
339
379
  xmlAddNextSibling(pivot, new_next_text);
340
380
  }
341
381
 
342
- if(!(reparented = (*prf)(pivot, reparentee))) {
382
+ if (!(reparented = (*prf)(pivot, reparentee))) {
343
383
  rb_raise(rb_eRuntimeError, "Could not reparent node");
344
384
  }
345
385
 
@@ -349,57 +389,432 @@ ok:
349
389
  * adjacent text nodes.
350
390
  */
351
391
  DATA_PTR(reparentee_obj) = reparented ;
392
+ reparented_obj = noko_xml_node_wrap(Qnil, reparented);
352
393
 
353
- relink_namespace(reparented);
394
+ rb_funcall(reparented_obj, id_decorate_bang, 0);
354
395
 
355
- reparented_obj = Nokogiri_wrap_xml_node(Qnil, reparented);
396
+ /* if we've created a cycle, raise an exception */
397
+ raise_if_ancestor_of_self(reparented);
356
398
 
357
- rb_funcall(reparented_obj, decorate_bang, 0);
399
+ relink_namespace(reparented);
358
400
 
359
401
  return reparented_obj ;
360
402
  }
361
403
 
404
+ // :startdoc:
362
405
 
363
406
  /*
364
- * call-seq:
365
- * document
407
+ * :call-seq:
408
+ * add_namespace_definition(prefix, href) → Nokogiri::XML::Namespace
409
+ * add_namespace(prefix, href) → Nokogiri::XML::Namespace
410
+ *
411
+ * :category: Manipulating Document Structure
412
+ *
413
+ * Adds a namespace definition to this node with +prefix+ using +href+ value, as if this node had
414
+ * included an attribute "xmlns:prefix=href".
415
+ *
416
+ * A default namespace definition for this node can be added by passing +nil+ for +prefix+.
417
+ *
418
+ * [Parameters]
419
+ * - +prefix+ (String, +nil+) An {XML Name}[https://www.w3.org/TR/xml-names/#ns-decl]
420
+ * - +href+ (String) The {URI reference}[https://www.w3.org/TR/xml-names/#sec-namespaces]
421
+ *
422
+ * [Returns] The new Nokogiri::XML::Namespace
423
+ *
424
+ * *Example:* adding a non-default namespace definition
425
+ *
426
+ * doc = Nokogiri::XML("<store><inventory></inventory></store>")
427
+ * inventory = doc.at_css("inventory")
428
+ * inventory.add_namespace_definition("automobile", "http://alices-autos.com/")
429
+ * inventory.add_namespace_definition("bicycle", "http://bobs-bikes.com/")
430
+ * inventory.add_child("<automobile:tire>Michelin model XGV, size 75R</automobile:tire>")
431
+ * doc.to_xml
432
+ * # => "<?xml version=\"1.0\"?>\n" +
433
+ * # "<store>\n" +
434
+ * # " <inventory xmlns:automobile=\"http://alices-autos.com/\" xmlns:bicycle=\"http://bobs-bikes.com/\">\n" +
435
+ * # " <automobile:tire>Michelin model XGV, size 75R</automobile:tire>\n" +
436
+ * # " </inventory>\n" +
437
+ * # "</store>\n"
438
+ *
439
+ * *Example:* adding a default namespace definition
440
+ *
441
+ * doc = Nokogiri::XML("<store><inventory><tire>Michelin model XGV, size 75R</tire></inventory></store>")
442
+ * doc.at_css("tire").add_namespace_definition(nil, "http://bobs-bikes.com/")
443
+ * doc.to_xml
444
+ * # => "<?xml version=\"1.0\"?>\n" +
445
+ * # "<store>\n" +
446
+ * # " <inventory>\n" +
447
+ * # " <tire xmlns=\"http://bobs-bikes.com/\">Michelin model XGV, size 75R</tire>\n" +
448
+ * # " </inventory>\n" +
449
+ * # "</store>\n"
450
+ *
451
+ */
452
+ static VALUE
453
+ rb_xml_node_add_namespace_definition(VALUE rb_node, VALUE rb_prefix, VALUE rb_href)
454
+ {
455
+ xmlNodePtr c_node, element;
456
+ xmlNsPtr c_namespace;
457
+ const xmlChar *c_prefix = (const xmlChar *)(NIL_P(rb_prefix) ? NULL : StringValueCStr(rb_prefix));
458
+
459
+ Noko_Node_Get_Struct(rb_node, xmlNode, c_node);
460
+ element = c_node ;
461
+
462
+ c_namespace = xmlSearchNs(c_node->doc, c_node, c_prefix);
463
+
464
+ if (!c_namespace) {
465
+ if (c_node->type != XML_ELEMENT_NODE) {
466
+ element = c_node->parent;
467
+ }
468
+ c_namespace = xmlNewNs(element, (const xmlChar *)StringValueCStr(rb_href), c_prefix);
469
+ }
470
+
471
+ if (!c_namespace) {
472
+ return Qnil ;
473
+ }
474
+
475
+ if (NIL_P(rb_prefix) || c_node != element) {
476
+ xmlSetNs(c_node, c_namespace);
477
+ }
478
+
479
+ return noko_xml_namespace_wrap(c_namespace, c_node->doc);
480
+ }
481
+
482
+
483
+ /*
484
+ * :call-seq: attribute(name) → Nokogiri::XML::Attr
485
+ *
486
+ * :category: Working With Node Attributes
487
+ *
488
+ * [Returns] Attribute (Nokogiri::XML::Attr) belonging to this node with name +name+.
489
+ *
490
+ * ⚠ Note that attribute namespaces are ignored and only the simple (non-namespace-prefixed) name is
491
+ * used to find a matching attribute. In case of a simple name collision, only one of the matching
492
+ * attributes will be returned. In this case, you will need to use #attribute_with_ns.
493
+ *
494
+ * *Example:*
495
+ *
496
+ * doc = Nokogiri::XML("<root><child size='large' class='big wide tall'/></root>")
497
+ * child = doc.at_css("child")
498
+ * child.attribute("size") # => #<Nokogiri::XML::Attr:0x550 name="size" value="large">
499
+ * child.attribute("class") # => #<Nokogiri::XML::Attr:0x564 name="class" value="big wide tall">
500
+ *
501
+ * *Example* showing that namespaced attributes will not be returned:
502
+ *
503
+ * ⚠ Note that only one of the two matching attributes is returned.
504
+ *
505
+ * doc = Nokogiri::XML(<<~EOF)
506
+ * <root xmlns:width='http://example.com/widths'
507
+ * xmlns:height='http://example.com/heights'>
508
+ * <child width:size='broad' height:size='tall'/>
509
+ * </root>
510
+ * EOF
511
+ * doc.at_css("child").attribute("size")
512
+ * # => #(Attr:0x550 {
513
+ * # name = "size",
514
+ * # namespace = #(Namespace:0x564 {
515
+ * # prefix = "width",
516
+ * # href = "http://example.com/widths"
517
+ * # }),
518
+ * # value = "broad"
519
+ * # })
520
+ */
521
+ static VALUE
522
+ rb_xml_node_attribute(VALUE self, VALUE name)
523
+ {
524
+ xmlNodePtr node;
525
+ xmlAttrPtr prop;
526
+ Noko_Node_Get_Struct(self, xmlNode, node);
527
+ prop = xmlHasProp(node, (xmlChar *)StringValueCStr(name));
528
+
529
+ if (! prop) { return Qnil; }
530
+ return noko_xml_node_wrap(Qnil, (xmlNodePtr)prop);
531
+ }
532
+
533
+
534
+ /*
535
+ * :call-seq: attribute_nodes() → Array<Nokogiri::XML::Attr>
536
+ *
537
+ * :category: Working With Node Attributes
538
+ *
539
+ * [Returns] Attributes (an Array of Nokogiri::XML::Attr) belonging to this node.
540
+ *
541
+ * Note that this is the preferred alternative to #attributes when the simple
542
+ * (non-namespace-prefixed) attribute names may collide.
543
+ *
544
+ * *Example:*
545
+ *
546
+ * Contrast this with the colliding-name example from #attributes.
547
+ *
548
+ * doc = Nokogiri::XML(<<~EOF)
549
+ * <root xmlns:width='http://example.com/widths'
550
+ * xmlns:height='http://example.com/heights'>
551
+ * <child width:size='broad' height:size='tall'/>
552
+ * </root>
553
+ * EOF
554
+ * doc.at_css("child").attribute_nodes
555
+ * # => [#(Attr:0x550 {
556
+ * # name = "size",
557
+ * # namespace = #(Namespace:0x564 {
558
+ * # prefix = "width",
559
+ * # href = "http://example.com/widths"
560
+ * # }),
561
+ * # value = "broad"
562
+ * # }),
563
+ * # #(Attr:0x578 {
564
+ * # name = "size",
565
+ * # namespace = #(Namespace:0x58c {
566
+ * # prefix = "height",
567
+ * # href = "http://example.com/heights"
568
+ * # }),
569
+ * # value = "tall"
570
+ * # })]
571
+ */
572
+ static VALUE
573
+ rb_xml_node_attribute_nodes(VALUE rb_node)
574
+ {
575
+ xmlNodePtr c_node;
576
+
577
+ Noko_Node_Get_Struct(rb_node, xmlNode, c_node);
578
+
579
+ return noko_xml_node_attrs(c_node);
580
+ }
581
+
582
+
583
+ /*
584
+ * :call-seq: attribute_with_ns(name, namespace) → Nokogiri::XML::Attr
585
+ *
586
+ * :category: Working With Node Attributes
587
+ *
588
+ * [Returns]
589
+ * Attribute (Nokogiri::XML::Attr) belonging to this node with matching +name+ and +namespace+.
590
+ *
591
+ * [Parameters]
592
+ * - +name+ (String): the simple (non-namespace-prefixed) name of the attribute
593
+ * - +namespace+ (String): the URI of the attribute's namespace
594
+ *
595
+ * See related: #attribute
596
+ *
597
+ * *Example:*
598
+ *
599
+ * doc = Nokogiri::XML(<<~EOF)
600
+ * <root xmlns:width='http://example.com/widths'
601
+ * xmlns:height='http://example.com/heights'>
602
+ * <child width:size='broad' height:size='tall'/>
603
+ * </root>
604
+ * EOF
605
+ * doc.at_css("child").attribute_with_ns("size", "http://example.com/widths")
606
+ * # => #(Attr:0x550 {
607
+ * # name = "size",
608
+ * # namespace = #(Namespace:0x564 {
609
+ * # prefix = "width",
610
+ * # href = "http://example.com/widths"
611
+ * # }),
612
+ * # value = "broad"
613
+ * # })
614
+ * doc.at_css("child").attribute_with_ns("size", "http://example.com/heights")
615
+ * # => #(Attr:0x578 {
616
+ * # name = "size",
617
+ * # namespace = #(Namespace:0x58c {
618
+ * # prefix = "height",
619
+ * # href = "http://example.com/heights"
620
+ * # }),
621
+ * # value = "tall"
622
+ * # })
623
+ */
624
+ static VALUE
625
+ rb_xml_node_attribute_with_ns(VALUE self, VALUE name, VALUE namespace)
626
+ {
627
+ xmlNodePtr node;
628
+ xmlAttrPtr prop;
629
+ Noko_Node_Get_Struct(self, xmlNode, node);
630
+ prop = xmlHasNsProp(node, (xmlChar *)StringValueCStr(name),
631
+ NIL_P(namespace) ? NULL : (xmlChar *)StringValueCStr(namespace));
632
+
633
+ if (! prop) { return Qnil; }
634
+ return noko_xml_node_wrap(Qnil, (xmlNodePtr)prop);
635
+ }
636
+
637
+
638
+
639
+ /*
640
+ * call-seq: blank? → Boolean
641
+ *
642
+ * [Returns] +true+ if the node is an empty or whitespace-only text or cdata node, else +false+.
643
+ *
644
+ * *Example:*
645
+ *
646
+ * Nokogiri("<root><child/></root>").root.child.blank? # => false
647
+ * Nokogiri("<root>\t \n</root>").root.child.blank? # => true
648
+ * Nokogiri("<root><![CDATA[\t \n]]></root>").root.child.blank? # => true
649
+ * Nokogiri("<root>not-blank</root>").root.child
650
+ * .tap { |n| n.content = "" }.blank # => true
651
+ */
652
+ static VALUE
653
+ rb_xml_node_blank_eh(VALUE self)
654
+ {
655
+ xmlNodePtr node;
656
+ Noko_Node_Get_Struct(self, xmlNode, node);
657
+ return (1 == xmlIsBlankNode(node)) ? Qtrue : Qfalse ;
658
+ }
659
+
660
+
661
+ /*
662
+ * :call-seq: child() → Nokogiri::XML::Node
663
+ *
664
+ * :category: Traversing Document Structure
665
+ *
666
+ * [Returns] First of this node's children, or +nil+ if there are no children
667
+ *
668
+ * This is a convenience method and is equivalent to:
669
+ *
670
+ * node.children.first
671
+ *
672
+ * See related: #children
673
+ */
674
+ static VALUE
675
+ rb_xml_node_child(VALUE self)
676
+ {
677
+ xmlNodePtr node, child;
678
+ Noko_Node_Get_Struct(self, xmlNode, node);
679
+
680
+ child = node->children;
681
+ if (!child) { return Qnil; }
682
+
683
+ return noko_xml_node_wrap(Qnil, child);
684
+ }
685
+
686
+
687
+ /*
688
+ * :call-seq: children() → Nokogiri::XML::NodeSet
689
+ *
690
+ * :category: Traversing Document Structure
691
+ *
692
+ * [Returns] Nokogiri::XML::NodeSet containing this node's children.
693
+ */
694
+ static VALUE
695
+ rb_xml_node_children(VALUE self)
696
+ {
697
+ xmlNodePtr node;
698
+ xmlNodePtr child;
699
+ xmlNodeSetPtr set;
700
+ VALUE document;
701
+ VALUE node_set;
702
+
703
+ Noko_Node_Get_Struct(self, xmlNode, node);
704
+
705
+ child = node->children;
706
+ set = xmlXPathNodeSetCreate(child);
707
+
708
+ document = DOC_RUBY_OBJECT(node->doc);
709
+
710
+ if (!child) { return noko_xml_node_set_wrap(set, document); }
711
+
712
+ child = child->next;
713
+ while (NULL != child) {
714
+ xmlXPathNodeSetAddUnique(set, child);
715
+ child = child->next;
716
+ }
717
+
718
+ node_set = noko_xml_node_set_wrap(set, document);
719
+
720
+ return node_set;
721
+ }
722
+
723
+
724
+ /*
725
+ * :call-seq:
726
+ * content() → String
727
+ * inner_text() → String
728
+ * text() → String
729
+ * to_str() → String
730
+ *
731
+ * [Returns]
732
+ * Contents of all the text nodes in this node's subtree, concatenated together into a single
733
+ * String.
734
+ *
735
+ * ⚠ Note that entities will _always_ be expanded in the returned String.
736
+ *
737
+ * See related: #inner_html
738
+ *
739
+ * *Example* of how entities are handled:
740
+ *
741
+ * Note that <tt>&lt;</tt> becomes <tt><</tt> in the returned String.
742
+ *
743
+ * doc = Nokogiri::XML.fragment("<child>a &lt; b</child>")
744
+ * doc.at_css("child").content
745
+ * # => "a < b"
746
+ *
747
+ * *Example* of how a subtree is handled:
748
+ *
749
+ * Note that the <tt><span></tt> tags are omitted and only the text node contents are returned,
750
+ * concatenated into a single string.
751
+ *
752
+ * doc = Nokogiri::XML.fragment("<child><span>first</span> <span>second</span></child>")
753
+ * doc.at_css("child").content
754
+ * # => "first second"
755
+ */
756
+ static VALUE
757
+ rb_xml_node_content(VALUE self)
758
+ {
759
+ xmlNodePtr node;
760
+ xmlChar *content;
761
+
762
+ Noko_Node_Get_Struct(self, xmlNode, node);
763
+
764
+ content = xmlNodeGetContent(node);
765
+ if (content) {
766
+ VALUE rval = NOKOGIRI_STR_NEW2(content);
767
+ xmlFree(content);
768
+ return rval;
769
+ }
770
+ return Qnil;
771
+ }
772
+
773
+
774
+ /*
775
+ * :call-seq: document() → Nokogiri::XML::Document
776
+ *
777
+ * :category: Traversing Document Structure
366
778
  *
367
- * Get the document for this Node
779
+ * [Returns] Parent Nokogiri::XML::Document for this node
368
780
  */
369
- static VALUE document(VALUE self)
781
+ static VALUE
782
+ rb_xml_node_document(VALUE self)
370
783
  {
371
784
  xmlNodePtr node;
372
- Data_Get_Struct(self, xmlNode, node);
785
+ Noko_Node_Get_Struct(self, xmlNode, node);
373
786
  return DOC_RUBY_OBJECT(node->doc);
374
787
  }
375
788
 
376
789
  /*
377
- * call-seq:
378
- * pointer_id
790
+ * :call-seq: pointer_id() → Integer
379
791
  *
380
- * Get the internal pointer number
792
+ * [Returns]
793
+ * A unique id for this node based on the internal memory structures. This method is used by #==
794
+ * to determine node identity.
381
795
  */
382
- static VALUE pointer_id(VALUE self)
796
+ static VALUE
797
+ rb_xml_node_pointer_id(VALUE self)
383
798
  {
384
799
  xmlNodePtr node;
385
- Data_Get_Struct(self, xmlNode, node);
800
+ Noko_Node_Get_Struct(self, xmlNode, node);
386
801
 
387
- return INT2NUM((long)(node));
802
+ return rb_uint2inum((uintptr_t)(node));
388
803
  }
389
804
 
390
805
  /*
391
- * call-seq:
392
- * encode_special_chars(string)
806
+ * :call-seq: encode_special_chars(string) → String
393
807
  *
394
808
  * Encode any special characters in +string+
395
809
  */
396
- static VALUE encode_special_chars(VALUE self, VALUE string)
810
+ static VALUE
811
+ encode_special_chars(VALUE self, VALUE string)
397
812
  {
398
813
  xmlNodePtr node;
399
814
  xmlChar *encoded;
400
815
  VALUE encoded_str;
401
816
 
402
- Data_Get_Struct(self, xmlNode, node);
817
+ Noko_Node_Get_Struct(self, xmlNode, node);
403
818
  encoded = xmlEncodeSpecialChars(
404
819
  node->doc,
405
820
  (const xmlChar *)StringValueCStr(string)
@@ -412,8 +827,8 @@ static VALUE encode_special_chars(VALUE self, VALUE string)
412
827
  }
413
828
 
414
829
  /*
415
- * call-seq:
416
- * create_internal_subset(name, external_id, system_id)
830
+ * :call-seq:
831
+ * create_internal_subset(name, external_id, system_id)
417
832
  *
418
833
  * Create the internal subset of a document.
419
834
  *
@@ -423,17 +838,18 @@ static VALUE encode_special_chars(VALUE self, VALUE string)
423
838
  * doc.create_internal_subset("chapter", nil, "chapter.dtd")
424
839
  * # => <!DOCTYPE chapter SYSTEM "chapter.dtd">
425
840
  */
426
- static VALUE create_internal_subset(VALUE self, VALUE name, VALUE external_id, VALUE system_id)
841
+ static VALUE
842
+ create_internal_subset(VALUE self, VALUE name, VALUE external_id, VALUE system_id)
427
843
  {
428
844
  xmlNodePtr node;
429
845
  xmlDocPtr doc;
430
846
  xmlDtdPtr dtd;
431
847
 
432
- Data_Get_Struct(self, xmlNode, node);
848
+ Noko_Node_Get_Struct(self, xmlNode, node);
433
849
 
434
850
  doc = node->doc;
435
851
 
436
- if(xmlGetIntSubset(doc)) {
852
+ if (xmlGetIntSubset(doc)) {
437
853
  rb_raise(rb_eRuntimeError, "Document already has an internal subset");
438
854
  }
439
855
 
@@ -444,28 +860,29 @@ static VALUE create_internal_subset(VALUE self, VALUE name, VALUE external_id, V
444
860
  NIL_P(system_id) ? NULL : (const xmlChar *)StringValueCStr(system_id)
445
861
  );
446
862
 
447
- if(!dtd) { return Qnil; }
863
+ if (!dtd) { return Qnil; }
448
864
 
449
- return Nokogiri_wrap_xml_node(Qnil, (xmlNodePtr)dtd);
865
+ return noko_xml_node_wrap(Qnil, (xmlNodePtr)dtd);
450
866
  }
451
867
 
452
868
  /*
453
- * call-seq:
454
- * create_external_subset(name, external_id, system_id)
869
+ * :call-seq:
870
+ * create_external_subset(name, external_id, system_id)
455
871
  *
456
872
  * Create an external subset
457
873
  */
458
- static VALUE create_external_subset(VALUE self, VALUE name, VALUE external_id, VALUE system_id)
874
+ static VALUE
875
+ create_external_subset(VALUE self, VALUE name, VALUE external_id, VALUE system_id)
459
876
  {
460
877
  xmlNodePtr node;
461
878
  xmlDocPtr doc;
462
879
  xmlDtdPtr dtd;
463
880
 
464
- Data_Get_Struct(self, xmlNode, node);
881
+ Noko_Node_Get_Struct(self, xmlNode, node);
465
882
 
466
883
  doc = node->doc;
467
884
 
468
- if(doc->extSubset) {
885
+ if (doc->extSubset) {
469
886
  rb_raise(rb_eRuntimeError, "Document already has an external subset");
470
887
  }
471
888
 
@@ -476,72 +893,78 @@ static VALUE create_external_subset(VALUE self, VALUE name, VALUE external_id, V
476
893
  NIL_P(system_id) ? NULL : (const xmlChar *)StringValueCStr(system_id)
477
894
  );
478
895
 
479
- if(!dtd) { return Qnil; }
896
+ if (!dtd) { return Qnil; }
480
897
 
481
- return Nokogiri_wrap_xml_node(Qnil, (xmlNodePtr)dtd);
898
+ return noko_xml_node_wrap(Qnil, (xmlNodePtr)dtd);
482
899
  }
483
900
 
484
901
  /*
485
- * call-seq:
486
- * external_subset
902
+ * :call-seq:
903
+ * external_subset()
487
904
  *
488
905
  * Get the external subset
489
906
  */
490
- static VALUE external_subset(VALUE self)
907
+ static VALUE
908
+ external_subset(VALUE self)
491
909
  {
492
910
  xmlNodePtr node;
493
911
  xmlDocPtr doc;
494
912
  xmlDtdPtr dtd;
495
913
 
496
- Data_Get_Struct(self, xmlNode, node);
914
+ Noko_Node_Get_Struct(self, xmlNode, node);
497
915
 
498
- if(!node->doc) { return Qnil; }
916
+ if (!node->doc) { return Qnil; }
499
917
 
500
918
  doc = node->doc;
501
919
  dtd = doc->extSubset;
502
920
 
503
- if(!dtd) { return Qnil; }
921
+ if (!dtd) { return Qnil; }
504
922
 
505
- return Nokogiri_wrap_xml_node(Qnil, (xmlNodePtr)dtd);
923
+ return noko_xml_node_wrap(Qnil, (xmlNodePtr)dtd);
506
924
  }
507
925
 
508
926
  /*
509
- * call-seq:
510
- * internal_subset
927
+ * :call-seq:
928
+ * internal_subset()
511
929
  *
512
930
  * Get the internal subset
513
931
  */
514
- static VALUE internal_subset(VALUE self)
932
+ static VALUE
933
+ internal_subset(VALUE self)
515
934
  {
516
935
  xmlNodePtr node;
517
936
  xmlDocPtr doc;
518
937
  xmlDtdPtr dtd;
519
938
 
520
- Data_Get_Struct(self, xmlNode, node);
939
+ Noko_Node_Get_Struct(self, xmlNode, node);
521
940
 
522
- if(!node->doc) { return Qnil; }
941
+ if (!node->doc) { return Qnil; }
523
942
 
524
943
  doc = node->doc;
525
944
  dtd = xmlGetIntSubset(doc);
526
945
 
527
- if(!dtd) { return Qnil; }
946
+ if (!dtd) { return Qnil; }
528
947
 
529
- return Nokogiri_wrap_xml_node(Qnil, (xmlNodePtr)dtd);
948
+ return noko_xml_node_wrap(Qnil, (xmlNodePtr)dtd);
530
949
  }
531
950
 
532
951
  /*
533
- * call-seq:
534
- * dup
535
- * dup(depth)
536
- * dup(depth, new_parent_doc)
952
+ * :call-seq:
953
+ * dup → Nokogiri::XML::Node
954
+ * dup(depth) → Nokogiri::XML::Node
955
+ * dup(depth, new_parent_doc) → Nokogiri::XML::Node
537
956
  *
538
957
  * Copy this node.
539
- * An optional depth may be passed in. 0 is a shallow copy, 1 (the default) is a deep copy.
540
- * An optional new_parent_doc may also be passed in, which will be the new
541
- * node's parent document. Defaults to the current node's document.
542
- * current document.
958
+ *
959
+ * [Parameters]
960
+ * - +depth+ 0 is a shallow copy, 1 (the default) is a deep copy.
961
+ * - +new_parent_doc+
962
+ * The new node's parent Document. Defaults to the this node's document.
963
+ *
964
+ * [Returns] The new Nokgiri::XML::Node
543
965
  */
544
- static VALUE duplicate_node(int argc, VALUE *argv, VALUE self)
966
+ static VALUE
967
+ duplicate_node(int argc, VALUE *argv, VALUE self)
545
968
  {
546
969
  VALUE r_level, r_new_parent_doc;
547
970
  int level;
@@ -549,7 +972,7 @@ static VALUE duplicate_node(int argc, VALUE *argv, VALUE self)
549
972
  xmlDocPtr new_parent_doc;
550
973
  xmlNodePtr node, dup;
551
974
 
552
- Data_Get_Struct(self, xmlNode, node);
975
+ Noko_Node_Get_Struct(self, xmlNode, node);
553
976
 
554
977
  n_args = rb_scan_args(argc, argv, "02", &r_level, &r_new_parent_doc);
555
978
 
@@ -561,44 +984,33 @@ static VALUE duplicate_node(int argc, VALUE *argv, VALUE self)
561
984
  if (n_args < 2) {
562
985
  new_parent_doc = node->doc;
563
986
  } else {
564
- Data_Get_Struct(r_new_parent_doc, xmlDoc, new_parent_doc);
987
+ new_parent_doc = noko_xml_document_unwrap(r_new_parent_doc);
565
988
  }
566
989
 
567
990
  dup = xmlDocCopyNode(node, new_parent_doc, level);
568
- if(dup == NULL) { return Qnil; }
991
+ if (dup == NULL) { return Qnil; }
569
992
 
570
- nokogiri_root_node(dup);
993
+ noko_xml_document_pin_node(dup);
571
994
 
572
- return Nokogiri_wrap_xml_node(rb_obj_class(self), dup);
995
+ return noko_xml_node_wrap(rb_obj_class(self), dup);
573
996
  }
574
997
 
575
998
  /*
576
- * call-seq:
577
- * unlink
999
+ * :call-seq:
1000
+ * unlink() → self
578
1001
  *
579
1002
  * Unlink this node from its current context.
580
1003
  */
581
- static VALUE unlink_node(VALUE self)
1004
+ static VALUE
1005
+ unlink_node(VALUE self)
582
1006
  {
583
1007
  xmlNodePtr node;
584
- Data_Get_Struct(self, xmlNode, node);
1008
+ Noko_Node_Get_Struct(self, xmlNode, node);
585
1009
  xmlUnlinkNode(node);
586
- nokogiri_root_node(node);
1010
+ noko_xml_document_pin_node(node);
587
1011
  return self;
588
1012
  }
589
1013
 
590
- /*
591
- * call-seq:
592
- * blank?
593
- *
594
- * Is this node blank?
595
- */
596
- static VALUE blank_eh(VALUE self)
597
- {
598
- xmlNodePtr node;
599
- Data_Get_Struct(self, xmlNode, node);
600
- return (1 == xmlIsBlankNode(node)) ? Qtrue : Qfalse ;
601
- }
602
1014
 
603
1015
  /*
604
1016
  * call-seq:
@@ -606,15 +1018,16 @@ static VALUE blank_eh(VALUE self)
606
1018
  *
607
1019
  * Returns the next sibling node
608
1020
  */
609
- static VALUE next_sibling(VALUE self)
1021
+ static VALUE
1022
+ next_sibling(VALUE self)
610
1023
  {
611
1024
  xmlNodePtr node, sibling;
612
- Data_Get_Struct(self, xmlNode, node);
1025
+ Noko_Node_Get_Struct(self, xmlNode, node);
613
1026
 
614
1027
  sibling = node->next;
615
- if(!sibling) { return Qnil; }
1028
+ if (!sibling) { return Qnil; }
616
1029
 
617
- return Nokogiri_wrap_xml_node(Qnil, sibling) ;
1030
+ return noko_xml_node_wrap(Qnil, sibling) ;
618
1031
  }
619
1032
 
620
1033
  /*
@@ -623,15 +1036,16 @@ static VALUE next_sibling(VALUE self)
623
1036
  *
624
1037
  * Returns the previous sibling node
625
1038
  */
626
- static VALUE previous_sibling(VALUE self)
1039
+ static VALUE
1040
+ previous_sibling(VALUE self)
627
1041
  {
628
1042
  xmlNodePtr node, sibling;
629
- Data_Get_Struct(self, xmlNode, node);
1043
+ Noko_Node_Get_Struct(self, xmlNode, node);
630
1044
 
631
1045
  sibling = node->prev;
632
- if(!sibling) { return Qnil; }
1046
+ if (!sibling) { return Qnil; }
633
1047
 
634
- return Nokogiri_wrap_xml_node(Qnil, sibling);
1048
+ return noko_xml_node_wrap(Qnil, sibling);
635
1049
  }
636
1050
 
637
1051
  /*
@@ -640,15 +1054,16 @@ static VALUE previous_sibling(VALUE self)
640
1054
  *
641
1055
  * Returns the next Nokogiri::XML::Element type sibling node.
642
1056
  */
643
- static VALUE next_element(VALUE self)
1057
+ static VALUE
1058
+ next_element(VALUE self)
644
1059
  {
645
1060
  xmlNodePtr node, sibling;
646
- Data_Get_Struct(self, xmlNode, node);
1061
+ Noko_Node_Get_Struct(self, xmlNode, node);
647
1062
 
648
1063
  sibling = xmlNextElementSibling(node);
649
- if(!sibling) { return Qnil; }
1064
+ if (!sibling) { return Qnil; }
650
1065
 
651
- return Nokogiri_wrap_xml_node(Qnil, sibling);
1066
+ return noko_xml_node_wrap(Qnil, sibling);
652
1067
  }
653
1068
 
654
1069
  /*
@@ -657,82 +1072,60 @@ static VALUE next_element(VALUE self)
657
1072
  *
658
1073
  * Returns the previous Nokogiri::XML::Element type sibling node.
659
1074
  */
660
- static VALUE previous_element(VALUE self)
1075
+ static VALUE
1076
+ previous_element(VALUE self)
661
1077
  {
662
1078
  xmlNodePtr node, sibling;
663
- Data_Get_Struct(self, xmlNode, node);
1079
+ Noko_Node_Get_Struct(self, xmlNode, node);
664
1080
 
665
1081
  /*
666
1082
  * note that we don't use xmlPreviousElementSibling here because it's buggy pre-2.7.7.
667
1083
  */
668
1084
  sibling = node->prev;
669
- if(!sibling) { return Qnil; }
1085
+ if (!sibling) { return Qnil; }
670
1086
 
671
- while(sibling && sibling->type != XML_ELEMENT_NODE) {
1087
+ while (sibling && sibling->type != XML_ELEMENT_NODE) {
672
1088
  sibling = sibling->prev;
673
1089
  }
674
1090
 
675
- return sibling ? Nokogiri_wrap_xml_node(Qnil, sibling) : Qnil ;
1091
+ return sibling ? noko_xml_node_wrap(Qnil, sibling) : Qnil ;
676
1092
  }
677
1093
 
678
1094
  /* :nodoc: */
679
- static VALUE replace(VALUE self, VALUE new_node)
1095
+ static VALUE
1096
+ replace(VALUE self, VALUE new_node)
680
1097
  {
681
1098
  VALUE reparent = reparent_node_with(self, new_node, xmlReplaceNodeWrapper);
682
1099
 
683
1100
  xmlNodePtr pivot;
684
- Data_Get_Struct(self, xmlNode, pivot);
685
- nokogiri_root_node(pivot);
1101
+ Noko_Node_Get_Struct(self, xmlNode, pivot);
1102
+ noko_xml_document_pin_node(pivot);
686
1103
 
687
1104
  return reparent;
688
1105
  }
689
1106
 
690
1107
  /*
691
- * call-seq:
692
- * children
1108
+ * :call-seq:
1109
+ * element_children() → NodeSet
1110
+ * elements() → NodeSet
693
1111
  *
694
- * Get the list of children for this node as a NodeSet
695
- */
696
- static VALUE children(VALUE self)
697
- {
698
- xmlNodePtr node;
699
- xmlNodePtr child;
700
- xmlNodeSetPtr set;
701
- VALUE document;
702
- VALUE node_set;
703
-
704
- Data_Get_Struct(self, xmlNode, node);
705
-
706
- child = node->children;
707
- set = xmlXPathNodeSetCreate(child);
708
-
709
- document = DOC_RUBY_OBJECT(node->doc);
710
-
711
- if(!child) { return Nokogiri_wrap_xml_node_set(set, document); }
712
-
713
- child = child->next;
714
- while(NULL != child) {
715
- xmlXPathNodeSetAddUnique(set, child);
716
- child = child->next;
717
- }
718
-
719
- node_set = Nokogiri_wrap_xml_node_set(set, document);
720
-
721
- return node_set;
722
- }
723
-
724
- /*
725
- * call-seq:
726
- * element_children
1112
+ * [Returns]
1113
+ * The node's child elements as a NodeSet. Only children that are elements will be returned, which
1114
+ * notably excludes Text nodes.
727
1115
  *
728
- * Get the list of children for this node as a NodeSet. All nodes will be
729
- * element nodes.
1116
+ * *Example:*
730
1117
  *
731
- * Example:
1118
+ * Note that #children returns the Text node "hello" while #element_children does not.
732
1119
  *
733
- * @doc.root.element_children.all? { |x| x.element? } # => true
1120
+ * div = Nokogiri::HTML5("<div>hello<span>world</span>").at_css("div")
1121
+ * div.element_children
1122
+ * # => [#<Nokogiri::XML::Element:0x50 name="span" children=[#<Nokogiri::XML::Text:0x3c "world">]>]
1123
+ * div.children
1124
+ * # => [#<Nokogiri::XML::Text:0x64 "hello">,
1125
+ * # #<Nokogiri::XML::Element:0x50 name="span" children=[#<Nokogiri::XML::Text:0x3c "world">]>]
734
1126
  */
735
- static VALUE element_children(VALUE self)
1127
+ static VALUE
1128
+ rb_xml_node_element_children(VALUE self)
736
1129
  {
737
1130
  xmlNodePtr node;
738
1131
  xmlNodePtr child;
@@ -740,83 +1133,78 @@ static VALUE element_children(VALUE self)
740
1133
  VALUE document;
741
1134
  VALUE node_set;
742
1135
 
743
- Data_Get_Struct(self, xmlNode, node);
1136
+ Noko_Node_Get_Struct(self, xmlNode, node);
744
1137
 
745
1138
  child = xmlFirstElementChild(node);
746
1139
  set = xmlXPathNodeSetCreate(child);
747
1140
 
748
1141
  document = DOC_RUBY_OBJECT(node->doc);
749
1142
 
750
- if(!child) { return Nokogiri_wrap_xml_node_set(set, document); }
751
-
752
- child = xmlNextElementSibling(child);
753
- while(NULL != child) {
754
- xmlXPathNodeSetAddUnique(set, child);
755
- child = xmlNextElementSibling(child);
756
- }
757
-
758
- node_set = Nokogiri_wrap_xml_node_set(set, document);
1143
+ if (!child) { return noko_xml_node_set_wrap(set, document); }
759
1144
 
760
- return node_set;
761
- }
762
-
763
- /*
764
- * call-seq:
765
- * child
766
- *
767
- * Returns the child node
768
- */
769
- static VALUE child(VALUE self)
770
- {
771
- xmlNodePtr node, child;
772
- Data_Get_Struct(self, xmlNode, node);
1145
+ child = xmlNextElementSibling(child);
1146
+ while (NULL != child) {
1147
+ xmlXPathNodeSetAddUnique(set, child);
1148
+ child = xmlNextElementSibling(child);
1149
+ }
773
1150
 
774
- child = node->children;
775
- if(!child) { return Qnil; }
1151
+ node_set = noko_xml_node_set_wrap(set, document);
776
1152
 
777
- return Nokogiri_wrap_xml_node(Qnil, child);
1153
+ return node_set;
778
1154
  }
779
1155
 
780
1156
  /*
781
- * call-seq:
782
- * first_element_child
1157
+ * :call-seq:
1158
+ * first_element_child() → Node
783
1159
  *
784
- * Returns the first child node of this node that is an element.
1160
+ * [Returns] The first child Node that is an element.
785
1161
  *
786
- * Example:
1162
+ * *Example:*
787
1163
  *
788
- * @doc.root.first_element_child.element? # => true
1164
+ * Note that the "hello" child, which is a Text node, is skipped and the <tt><span></tt> element is
1165
+ * returned.
1166
+ *
1167
+ * div = Nokogiri::HTML5("<div>hello<span>world</span>").at_css("div")
1168
+ * div.first_element_child
1169
+ * # => #(Element:0x3c { name = "span", children = [ #(Text "world")] })
789
1170
  */
790
- static VALUE first_element_child(VALUE self)
1171
+ static VALUE
1172
+ rb_xml_node_first_element_child(VALUE self)
791
1173
  {
792
1174
  xmlNodePtr node, child;
793
- Data_Get_Struct(self, xmlNode, node);
1175
+ Noko_Node_Get_Struct(self, xmlNode, node);
794
1176
 
795
1177
  child = xmlFirstElementChild(node);
796
- if(!child) { return Qnil; }
1178
+ if (!child) { return Qnil; }
797
1179
 
798
- return Nokogiri_wrap_xml_node(Qnil, child);
1180
+ return noko_xml_node_wrap(Qnil, child);
799
1181
  }
800
1182
 
801
1183
  /*
802
- * call-seq:
803
- * last_element_child
1184
+ * :call-seq:
1185
+ * last_element_child() → Node
804
1186
  *
805
- * Returns the last child node of this node that is an element.
1187
+ * [Returns] The last child Node that is an element.
806
1188
  *
807
- * Example:
1189
+ * *Example:*
808
1190
  *
809
- * @doc.root.last_element_child.element? # => true
1191
+ * Note that the "hello" child, which is a Text node, is skipped and the <tt><span>yes</span></tt>
1192
+ * element is returned.
1193
+ *
1194
+ * div = Nokogiri::HTML5("<div><span>no</span><span>yes</span>skip</div>").at_css("div")
1195
+ * div.last_element_child
1196
+ * # => #(Element:0x3c { name = "span", children = [ #(Text "yes")] })
810
1197
  */
811
- static VALUE last_element_child(VALUE self)
1198
+ static VALUE
1199
+ rb_xml_node_last_element_child(VALUE self)
812
1200
  {
813
1201
  xmlNodePtr node, child;
814
- Data_Get_Struct(self, xmlNode, node);
1202
+ Noko_Node_Get_Struct(self, xmlNode, node);
815
1203
 
816
1204
  child = xmlLastElementChild(node);
817
- if(!child) { return Qnil; }
1205
+ if (!child) { return Qnil; }
818
1206
 
819
- return Nokogiri_wrap_xml_node(Qnil, child);
1207
+ return noko_xml_node_wrap(Qnil, child);
820
1208
  }
821
1209
 
822
1210
  /*
@@ -825,11 +1213,12 @@ static VALUE last_element_child(VALUE self)
825
1213
  *
826
1214
  * Returns true if +attribute+ is set
827
1215
  */
828
- static VALUE key_eh(VALUE self, VALUE attribute)
1216
+ static VALUE
1217
+ key_eh(VALUE self, VALUE attribute)
829
1218
  {
830
1219
  xmlNodePtr node;
831
- Data_Get_Struct(self, xmlNode, node);
832
- if(xmlHasProp(node, (xmlChar *)StringValueCStr(attribute))) {
1220
+ Noko_Node_Get_Struct(self, xmlNode, node);
1221
+ if (xmlHasProp(node, (xmlChar *)StringValueCStr(attribute))) {
833
1222
  return Qtrue;
834
1223
  }
835
1224
  return Qfalse;
@@ -841,12 +1230,13 @@ static VALUE key_eh(VALUE self, VALUE attribute)
841
1230
  *
842
1231
  * Returns true if +attribute+ is set with +namespace+
843
1232
  */
844
- static VALUE namespaced_key_eh(VALUE self, VALUE attribute, VALUE namespace)
1233
+ static VALUE
1234
+ namespaced_key_eh(VALUE self, VALUE attribute, VALUE namespace)
845
1235
  {
846
1236
  xmlNodePtr node;
847
- Data_Get_Struct(self, xmlNode, node);
848
- if(xmlHasNsProp(node, (xmlChar *)StringValueCStr(attribute),
849
- NIL_P(namespace) ? NULL : (xmlChar *)StringValueCStr(namespace))) {
1237
+ Noko_Node_Get_Struct(self, xmlNode, node);
1238
+ if (xmlHasNsProp(node, (xmlChar *)StringValueCStr(attribute),
1239
+ NIL_P(namespace) ? NULL : (xmlChar *)StringValueCStr(namespace))) {
850
1240
  return Qtrue;
851
1241
  }
852
1242
  return Qfalse;
@@ -858,11 +1248,12 @@ static VALUE namespaced_key_eh(VALUE self, VALUE attribute, VALUE namespace)
858
1248
  *
859
1249
  * Set the +property+ to +value+
860
1250
  */
861
- static VALUE set(VALUE self, VALUE property, VALUE value)
1251
+ static VALUE
1252
+ set(VALUE self, VALUE property, VALUE value)
862
1253
  {
863
1254
  xmlNodePtr node, cur;
864
1255
  xmlAttrPtr prop;
865
- Data_Get_Struct(self, xmlNode, node);
1256
+ Noko_Node_Get_Struct(self, xmlNode, node);
866
1257
 
867
1258
  /* If a matching attribute node already exists, then xmlSetProp will destroy
868
1259
  * the existing node's children. However, if Nokogiri has a node object
@@ -871,13 +1262,13 @@ static VALUE set(VALUE self, VALUE property, VALUE value)
871
1262
  * We can avoid this by unlinking these nodes first.
872
1263
  */
873
1264
  if (node->type != XML_ELEMENT_NODE) {
874
- return(Qnil);
1265
+ return (Qnil);
875
1266
  }
876
1267
  prop = xmlHasProp(node, (xmlChar *)StringValueCStr(property));
877
1268
  if (prop && prop->children) {
878
1269
  for (cur = prop->children; cur; cur = cur->next) {
879
1270
  if (cur->_private) {
880
- nokogiri_root_node(cur);
1271
+ noko_xml_document_pin_node(cur);
881
1272
  xmlUnlinkNode(cur);
882
1273
  }
883
1274
  }
@@ -895,7 +1286,8 @@ static VALUE set(VALUE self, VALUE property, VALUE value)
895
1286
  *
896
1287
  * Get the value for +attribute+
897
1288
  */
898
- static VALUE get(VALUE self, VALUE rattribute)
1289
+ static VALUE
1290
+ get(VALUE self, VALUE rattribute)
899
1291
  {
900
1292
  xmlNodePtr node;
901
1293
  xmlChar *value = 0;
@@ -906,10 +1298,10 @@ static VALUE get(VALUE self, VALUE rattribute)
906
1298
 
907
1299
  if (NIL_P(rattribute)) { return Qnil; }
908
1300
 
909
- Data_Get_Struct(self, xmlNode, node);
1301
+ Noko_Node_Get_Struct(self, xmlNode, node);
910
1302
  attribute = xmlCharStrdup(StringValueCStr(rattribute));
911
1303
 
912
- colon = (xmlChar *)(uintptr_t)xmlStrchr(attribute, (const xmlChar)':');
1304
+ colon = DISCARD_CONST_QUAL_XMLCHAR(xmlStrchr(attribute, (const xmlChar)':'));
913
1305
  if (colon) {
914
1306
  /* split the attribute string into separate prefix and name by
915
1307
  * null-terminating the prefix at the colon */
@@ -921,7 +1313,7 @@ static VALUE get(VALUE self, VALUE rattribute)
921
1313
  if (ns) {
922
1314
  value = xmlGetNsProp(node, attr_name, ns->href);
923
1315
  } else {
924
- value = xmlGetProp(node, (xmlChar*)StringValueCStr(rattribute));
1316
+ value = xmlGetProp(node, (xmlChar *)StringValueCStr(rattribute));
925
1317
  }
926
1318
  } else {
927
1319
  value = xmlGetNoNsProp(node, attribute);
@@ -942,15 +1334,16 @@ static VALUE get(VALUE self, VALUE rattribute)
942
1334
  *
943
1335
  * Set the namespace to +namespace+
944
1336
  */
945
- static VALUE set_namespace(VALUE self, VALUE namespace)
1337
+ static VALUE
1338
+ set_namespace(VALUE self, VALUE namespace)
946
1339
  {
947
1340
  xmlNodePtr node;
948
1341
  xmlNsPtr ns = NULL;
949
1342
 
950
- Data_Get_Struct(self, xmlNode, node);
1343
+ Noko_Node_Get_Struct(self, xmlNode, node);
951
1344
 
952
- if(!NIL_P(namespace)) {
953
- Data_Get_Struct(namespace, xmlNs, ns);
1345
+ if (!NIL_P(namespace)) {
1346
+ Noko_Namespace_Get_Struct(namespace, xmlNs, ns);
954
1347
  }
955
1348
 
956
1349
  xmlSetNs(node, ns);
@@ -959,138 +1352,140 @@ static VALUE set_namespace(VALUE self, VALUE namespace)
959
1352
  }
960
1353
 
961
1354
  /*
962
- * call-seq:
963
- * attribute(name)
1355
+ * :call-seq:
1356
+ * namespace() → Namespace
964
1357
  *
965
- * Get the attribute node with +name+
966
- */
967
- static VALUE attr(VALUE self, VALUE name)
968
- {
969
- xmlNodePtr node;
970
- xmlAttrPtr prop;
971
- Data_Get_Struct(self, xmlNode, node);
972
- prop = xmlHasProp(node, (xmlChar *)StringValueCStr(name));
973
-
974
- if(! prop) { return Qnil; }
975
- return Nokogiri_wrap_xml_node(Qnil, (xmlNodePtr)prop);
976
- }
977
-
978
- /*
979
- * call-seq:
980
- * attribute_with_ns(name, namespace)
1358
+ * [Returns] The Namespace of the element or attribute node, or +nil+ if there is no namespace.
981
1359
  *
982
- * Get the attribute node with +name+ and +namespace+
983
- */
984
- static VALUE attribute_with_ns(VALUE self, VALUE name, VALUE namespace)
985
- {
986
- xmlNodePtr node;
987
- xmlAttrPtr prop;
988
- Data_Get_Struct(self, xmlNode, node);
989
- prop = xmlHasNsProp(node, (xmlChar *)StringValueCStr(name),
990
- NIL_P(namespace) ? NULL : (xmlChar *)StringValueCStr(namespace));
991
-
992
- if(! prop) { return Qnil; }
993
- return Nokogiri_wrap_xml_node(Qnil, (xmlNodePtr)prop);
994
- }
995
-
996
- /*
997
- * call-seq:
998
- * attribute_nodes()
1360
+ * *Example:*
999
1361
  *
1000
- * returns a list containing the Node attributes.
1362
+ * doc = Nokogiri::XML(<<~EOF)
1363
+ * <root>
1364
+ * <first/>
1365
+ * <second xmlns="http://example.com/child"/>
1366
+ * <foo:third xmlns:foo="http://example.com/foo"/>
1367
+ * </root>
1368
+ * EOF
1369
+ * doc.at_xpath("//first").namespace
1370
+ * # => nil
1371
+ * doc.at_xpath("//xmlns:second", "xmlns" => "http://example.com/child").namespace
1372
+ * # => #(Namespace:0x3c { href = "http://example.com/child" })
1373
+ * doc.at_xpath("//foo:third", "foo" => "http://example.com/foo").namespace
1374
+ * # => #(Namespace:0x50 { prefix = "foo", href = "http://example.com/foo" })
1001
1375
  */
1002
- static VALUE attribute_nodes(VALUE self)
1376
+ static VALUE
1377
+ rb_xml_node_namespace(VALUE rb_node)
1003
1378
  {
1004
- /* this code in the mode of xmlHasProp() */
1005
- xmlNodePtr node;
1006
- VALUE attr;
1379
+ xmlNodePtr c_node ;
1380
+ Noko_Node_Get_Struct(rb_node, xmlNode, c_node);
1007
1381
 
1008
- Data_Get_Struct(self, xmlNode, node);
1009
-
1010
- attr = rb_ary_new();
1011
- Nokogiri_xml_node_properties(node, attr);
1382
+ if (c_node->ns) {
1383
+ return noko_xml_namespace_wrap(c_node->ns, c_node->doc);
1384
+ }
1012
1385
 
1013
- return attr ;
1386
+ return Qnil ;
1014
1387
  }
1015
1388
 
1016
-
1017
1389
  /*
1018
- * call-seq:
1019
- * namespace()
1390
+ * :call-seq:
1391
+ * namespace_definitions() → Array<Nokogiri::XML::Namespace>
1020
1392
  *
1021
- * returns the namespace of the element or attribute node as a Namespace
1022
- * object, or nil if there is no namespace for the element or attribute.
1023
- */
1024
- static VALUE namespace(VALUE self)
1025
- {
1026
- xmlNodePtr node ;
1027
- Data_Get_Struct(self, xmlNode, node);
1028
-
1029
- if (node->ns) {
1030
- return Nokogiri_wrap_xml_namespace(node->doc, node->ns);
1031
- }
1032
-
1033
- return Qnil ;
1034
- }
1035
-
1036
- /*
1037
- * call-seq:
1038
- * namespace_definitions()
1393
+ * [Returns]
1394
+ * Namespaces that are defined directly on this node, as an Array of Namespace objects. The array
1395
+ * will be empty if no namespaces are defined on this node.
1039
1396
  *
1040
- * returns namespaces defined on self element directly, as an array of Namespace objects. Includes both a default namespace (as in"xmlns="), and prefixed namespaces (as in "xmlns:prefix=").
1397
+ * *Example:*
1398
+ *
1399
+ * doc = Nokogiri::XML(<<~EOF)
1400
+ * <root xmlns="http://example.com/root">
1401
+ * <first/>
1402
+ * <second xmlns="http://example.com/child" xmlns:unused="http://example.com/unused"/>
1403
+ * <foo:third xmlns:foo="http://example.com/foo"/>
1404
+ * </root>
1405
+ * EOF
1406
+ * doc.at_xpath("//root:first", "root" => "http://example.com/root").namespace_definitions
1407
+ * # => []
1408
+ * doc.at_xpath("//xmlns:second", "xmlns" => "http://example.com/child").namespace_definitions
1409
+ * # => [#(Namespace:0x3c { href = "http://example.com/child" }),
1410
+ * # #(Namespace:0x50 {
1411
+ * # prefix = "unused",
1412
+ * # href = "http://example.com/unused"
1413
+ * # })]
1414
+ * doc.at_xpath("//foo:third", "foo" => "http://example.com/foo").namespace_definitions
1415
+ * # => [#(Namespace:0x64 { prefix = "foo", href = "http://example.com/foo" })]
1041
1416
  */
1042
- static VALUE namespace_definitions(VALUE self)
1417
+ static VALUE
1418
+ namespace_definitions(VALUE rb_node)
1043
1419
  {
1044
1420
  /* this code in the mode of xmlHasProp() */
1045
- xmlNodePtr node ;
1046
- VALUE list;
1047
- xmlNsPtr ns;
1048
-
1049
- Data_Get_Struct(self, xmlNode, node);
1421
+ xmlNodePtr c_node ;
1422
+ xmlNsPtr c_namespace;
1423
+ VALUE definitions = rb_ary_new();
1050
1424
 
1051
- list = rb_ary_new();
1425
+ Noko_Node_Get_Struct(rb_node, xmlNode, c_node);
1052
1426
 
1053
- ns = node->nsDef;
1054
-
1055
- if(!ns) { return list; }
1427
+ c_namespace = c_node->nsDef;
1428
+ if (!c_namespace) {
1429
+ return definitions;
1430
+ }
1056
1431
 
1057
- while(NULL != ns) {
1058
- rb_ary_push(list, Nokogiri_wrap_xml_namespace(node->doc, ns));
1059
- ns = ns->next;
1432
+ while (c_namespace != NULL) {
1433
+ rb_ary_push(definitions, noko_xml_namespace_wrap(c_namespace, c_node->doc));
1434
+ c_namespace = c_namespace->next;
1060
1435
  }
1061
1436
 
1062
- return list;
1437
+ return definitions;
1063
1438
  }
1064
1439
 
1065
1440
  /*
1066
- * call-seq:
1067
- * namespace_scopes()
1441
+ * :call-seq:
1442
+ * namespace_scopes() → Array<Nokogiri::XML::Namespace>
1068
1443
  *
1069
- * returns namespaces in scope for self -- those defined on self element
1070
- * directly or any ancestor node -- as an array of Namespace objects. Default
1071
- * namespaces ("xmlns=" style) for self are included in this array; Default
1072
- * namespaces for ancestors, however, are not. See also #namespaces
1444
+ * [Returns] Array of all the Namespaces on this node and its ancestors.
1445
+ *
1446
+ * See also #namespaces
1447
+ *
1448
+ * *Example:*
1449
+ *
1450
+ * doc = Nokogiri::XML(<<~EOF)
1451
+ * <root xmlns="http://example.com/root" xmlns:bar="http://example.com/bar">
1452
+ * <first/>
1453
+ * <second xmlns="http://example.com/child"/>
1454
+ * <third xmlns:foo="http://example.com/foo"/>
1455
+ * </root>
1456
+ * EOF
1457
+ * doc.at_xpath("//root:first", "root" => "http://example.com/root").namespace_scopes
1458
+ * # => [#(Namespace:0x3c { href = "http://example.com/root" }),
1459
+ * # #(Namespace:0x50 { prefix = "bar", href = "http://example.com/bar" })]
1460
+ * doc.at_xpath("//child:second", "child" => "http://example.com/child").namespace_scopes
1461
+ * # => [#(Namespace:0x64 { href = "http://example.com/child" }),
1462
+ * # #(Namespace:0x50 { prefix = "bar", href = "http://example.com/bar" })]
1463
+ * doc.at_xpath("//root:third", "root" => "http://example.com/root").namespace_scopes
1464
+ * # => [#(Namespace:0x78 { prefix = "foo", href = "http://example.com/foo" }),
1465
+ * # #(Namespace:0x3c { href = "http://example.com/root" }),
1466
+ * # #(Namespace:0x50 { prefix = "bar", href = "http://example.com/bar" })]
1073
1467
  */
1074
- static VALUE namespace_scopes(VALUE self)
1468
+ static VALUE
1469
+ rb_xml_node_namespace_scopes(VALUE rb_node)
1075
1470
  {
1076
- xmlNodePtr node ;
1077
- VALUE list;
1078
- xmlNsPtr *ns_list;
1471
+ xmlNodePtr c_node ;
1472
+ xmlNsPtr *namespaces;
1473
+ VALUE scopes = rb_ary_new();
1079
1474
  int j;
1080
1475
 
1081
- Data_Get_Struct(self, xmlNode, node);
1476
+ Noko_Node_Get_Struct(rb_node, xmlNode, c_node);
1082
1477
 
1083
- list = rb_ary_new();
1084
- ns_list = xmlGetNsList(node->doc, node);
1085
-
1086
- if(!ns_list) { return list; }
1478
+ namespaces = xmlGetNsList(c_node->doc, c_node);
1479
+ if (!namespaces) {
1480
+ return scopes;
1481
+ }
1087
1482
 
1088
- for (j = 0 ; ns_list[j] != NULL ; ++j) {
1089
- rb_ary_push(list, Nokogiri_wrap_xml_namespace(node->doc, ns_list[j]));
1483
+ for (j = 0 ; namespaces[j] != NULL ; ++j) {
1484
+ rb_ary_push(scopes, noko_xml_namespace_wrap(namespaces[j], c_node->doc));
1090
1485
  }
1091
1486
 
1092
- xmlFree(ns_list);
1093
- return list;
1487
+ xmlFree(namespaces);
1488
+ return scopes;
1094
1489
  }
1095
1490
 
1096
1491
  /*
@@ -1099,11 +1494,12 @@ static VALUE namespace_scopes(VALUE self)
1099
1494
  *
1100
1495
  * Get the type for this Node
1101
1496
  */
1102
- static VALUE node_type(VALUE self)
1497
+ static VALUE
1498
+ node_type(VALUE self)
1103
1499
  {
1104
1500
  xmlNodePtr node;
1105
- Data_Get_Struct(self, xmlNode, node);
1106
- return INT2NUM((long)node->type);
1501
+ Noko_Node_Get_Struct(self, xmlNode, node);
1502
+ return INT2NUM(node->type);
1107
1503
  }
1108
1504
 
1109
1505
  /*
@@ -1112,16 +1508,17 @@ static VALUE node_type(VALUE self)
1112
1508
  *
1113
1509
  * Set the content for this Node
1114
1510
  */
1115
- static VALUE set_native_content(VALUE self, VALUE content)
1511
+ static VALUE
1512
+ set_native_content(VALUE self, VALUE content)
1116
1513
  {
1117
1514
  xmlNodePtr node, child, next ;
1118
- Data_Get_Struct(self, xmlNode, node);
1515
+ Noko_Node_Get_Struct(self, xmlNode, node);
1119
1516
 
1120
1517
  child = node->children;
1121
1518
  while (NULL != child) {
1122
1519
  next = child->next ;
1123
1520
  xmlUnlinkNode(child) ;
1124
- nokogiri_root_node(child);
1521
+ noko_xml_document_pin_node(child);
1125
1522
  child = next ;
1126
1523
  }
1127
1524
 
@@ -1129,41 +1526,20 @@ static VALUE set_native_content(VALUE self, VALUE content)
1129
1526
  return content;
1130
1527
  }
1131
1528
 
1132
- /*
1133
- * call-seq:
1134
- * content
1135
- *
1136
- * Returns the content for this Node
1137
- */
1138
- static VALUE get_native_content(VALUE self)
1139
- {
1140
- xmlNodePtr node;
1141
- xmlChar * content;
1142
-
1143
- Data_Get_Struct(self, xmlNode, node);
1144
-
1145
- content = xmlNodeGetContent(node);
1146
- if(content) {
1147
- VALUE rval = NOKOGIRI_STR_NEW2(content);
1148
- xmlFree(content);
1149
- return rval;
1150
- }
1151
- return Qnil;
1152
- }
1153
-
1154
1529
  /*
1155
1530
  * call-seq:
1156
1531
  * lang=
1157
1532
  *
1158
1533
  * Set the language of a node, i.e. the values of the xml:lang attribute.
1159
1534
  */
1160
- static VALUE set_lang(VALUE self_rb, VALUE lang_rb)
1535
+ static VALUE
1536
+ set_lang(VALUE self_rb, VALUE lang_rb)
1161
1537
  {
1162
1538
  xmlNodePtr self ;
1163
- xmlChar* lang ;
1539
+ xmlChar *lang ;
1164
1540
 
1165
- Data_Get_Struct(self_rb, xmlNode, self);
1166
- lang = (xmlChar*)StringValueCStr(lang_rb);
1541
+ Noko_Node_Get_Struct(self_rb, xmlNode, self);
1542
+ lang = (xmlChar *)StringValueCStr(lang_rb);
1167
1543
 
1168
1544
  xmlNodeSetLang(self, lang);
1169
1545
 
@@ -1177,13 +1553,14 @@ static VALUE set_lang(VALUE self_rb, VALUE lang_rb)
1177
1553
  * Searches the language of a node, i.e. the values of the xml:lang attribute or
1178
1554
  * the one carried by the nearest ancestor.
1179
1555
  */
1180
- static VALUE get_lang(VALUE self_rb)
1556
+ static VALUE
1557
+ get_lang(VALUE self_rb)
1181
1558
  {
1182
1559
  xmlNodePtr self ;
1183
- xmlChar* lang ;
1560
+ xmlChar *lang ;
1184
1561
  VALUE lang_rb ;
1185
1562
 
1186
- Data_Get_Struct(self_rb, xmlNode, self);
1563
+ Noko_Node_Get_Struct(self_rb, xmlNode, self);
1187
1564
 
1188
1565
  lang = xmlNodeGetLang(self);
1189
1566
  if (lang) {
@@ -1196,7 +1573,8 @@ static VALUE get_lang(VALUE self_rb)
1196
1573
  }
1197
1574
 
1198
1575
  /* :nodoc: */
1199
- static VALUE add_child(VALUE self, VALUE new_child)
1576
+ static VALUE
1577
+ add_child(VALUE self, VALUE new_child)
1200
1578
  {
1201
1579
  return reparent_node_with(self, new_child, xmlAddChild);
1202
1580
  }
@@ -1207,15 +1585,16 @@ static VALUE add_child(VALUE self, VALUE new_child)
1207
1585
  *
1208
1586
  * Get the parent Node for this Node
1209
1587
  */
1210
- static VALUE get_parent(VALUE self)
1588
+ static VALUE
1589
+ get_parent(VALUE self)
1211
1590
  {
1212
1591
  xmlNodePtr node, parent;
1213
- Data_Get_Struct(self, xmlNode, node);
1592
+ Noko_Node_Get_Struct(self, xmlNode, node);
1214
1593
 
1215
1594
  parent = node->parent;
1216
- if(!parent) { return Qnil; }
1595
+ if (!parent) { return Qnil; }
1217
1596
 
1218
- return Nokogiri_wrap_xml_node(Qnil, parent) ;
1597
+ return noko_xml_node_wrap(Qnil, parent) ;
1219
1598
  }
1220
1599
 
1221
1600
  /*
@@ -1224,11 +1603,12 @@ static VALUE get_parent(VALUE self)
1224
1603
  *
1225
1604
  * Set the name for this Node
1226
1605
  */
1227
- static VALUE set_name(VALUE self, VALUE new_name)
1606
+ static VALUE
1607
+ set_name(VALUE self, VALUE new_name)
1228
1608
  {
1229
1609
  xmlNodePtr node;
1230
- Data_Get_Struct(self, xmlNode, node);
1231
- xmlNodeSetName(node, (xmlChar*)StringValueCStr(new_name));
1610
+ Noko_Node_Get_Struct(self, xmlNode, node);
1611
+ xmlNodeSetName(node, (xmlChar *)StringValueCStr(new_name));
1232
1612
  return new_name;
1233
1613
  }
1234
1614
 
@@ -1238,11 +1618,12 @@ static VALUE set_name(VALUE self, VALUE new_name)
1238
1618
  *
1239
1619
  * Returns the name for this Node
1240
1620
  */
1241
- static VALUE get_name(VALUE self)
1621
+ static VALUE
1622
+ get_name(VALUE self)
1242
1623
  {
1243
1624
  xmlNodePtr node;
1244
- Data_Get_Struct(self, xmlNode, node);
1245
- if(node->name) {
1625
+ Noko_Node_Get_Struct(self, xmlNode, node);
1626
+ if (node->name) {
1246
1627
  return NOKOGIRI_STR_NEW2(node->name);
1247
1628
  }
1248
1629
  return Qnil;
@@ -1254,28 +1635,39 @@ static VALUE get_name(VALUE self)
1254
1635
  *
1255
1636
  * Returns the path associated with this Node
1256
1637
  */
1257
- static VALUE path(VALUE self)
1638
+ static VALUE
1639
+ rb_xml_node_path(VALUE rb_node)
1258
1640
  {
1259
- xmlNodePtr node;
1260
- xmlChar *path ;
1641
+ xmlNodePtr c_node;
1642
+ xmlChar *c_path ;
1261
1643
  VALUE rval;
1262
1644
 
1263
- Data_Get_Struct(self, xmlNode, node);
1645
+ Noko_Node_Get_Struct(rb_node, xmlNode, c_node);
1646
+
1647
+ c_path = xmlGetNodePath(c_node);
1648
+ if (c_path == NULL) {
1649
+ // see https://github.com/sparklemotion/nokogiri/issues/2250
1650
+ // this behavior is clearly undesirable, but is what libxml <= 2.9.10 returned, and so we
1651
+ // do this for now to preserve the behavior across libxml2 versions.
1652
+ rval = NOKOGIRI_STR_NEW2("?");
1653
+ } else {
1654
+ rval = NOKOGIRI_STR_NEW2(c_path);
1655
+ xmlFree(c_path);
1656
+ }
1264
1657
 
1265
- path = xmlGetNodePath(node);
1266
- rval = NOKOGIRI_STR_NEW2(path);
1267
- xmlFree(path);
1268
1658
  return rval ;
1269
1659
  }
1270
1660
 
1271
1661
  /* :nodoc: */
1272
- static VALUE add_next_sibling(VALUE self, VALUE new_sibling)
1662
+ static VALUE
1663
+ add_next_sibling(VALUE self, VALUE new_sibling)
1273
1664
  {
1274
1665
  return reparent_node_with(self, new_sibling, xmlAddNextSibling) ;
1275
1666
  }
1276
1667
 
1277
1668
  /* :nodoc: */
1278
- static VALUE add_previous_sibling(VALUE self, VALUE new_sibling)
1669
+ static VALUE
1670
+ add_previous_sibling(VALUE self, VALUE new_sibling)
1279
1671
  {
1280
1672
  return reparent_node_with(self, new_sibling, xmlAddPrevSibling) ;
1281
1673
  }
@@ -1286,7 +1678,8 @@ static VALUE add_previous_sibling(VALUE self, VALUE new_sibling)
1286
1678
  *
1287
1679
  * Write this Node to +io+ with +encoding+ and +options+
1288
1680
  */
1289
- static VALUE native_write_to(
1681
+ static VALUE
1682
+ native_write_to(
1290
1683
  VALUE self,
1291
1684
  VALUE io,
1292
1685
  VALUE encoding,
@@ -1295,10 +1688,10 @@ static VALUE native_write_to(
1295
1688
  )
1296
1689
  {
1297
1690
  xmlNodePtr node;
1298
- const char * before_indent;
1691
+ const char *before_indent;
1299
1692
  xmlSaveCtxtPtr savectx;
1300
1693
 
1301
- Data_Get_Struct(self, xmlNode, node);
1694
+ Noko_Node_Get_Struct(self, xmlNode, node);
1302
1695
 
1303
1696
  xmlIndentTreeOutput = 1;
1304
1697
 
@@ -1307,8 +1700,8 @@ static VALUE native_write_to(
1307
1700
  xmlTreeIndentString = StringValueCStr(indent_string);
1308
1701
 
1309
1702
  savectx = xmlSaveToIO(
1310
- (xmlOutputWriteCallback)io_write_callback,
1311
- (xmlOutputCloseCallback)io_close_callback,
1703
+ (xmlOutputWriteCallback)noko_io_write,
1704
+ (xmlOutputCloseCallback)noko_io_close,
1312
1705
  (void *)io,
1313
1706
  RTEST(encoding) ? StringValueCStr(encoding) : NULL,
1314
1707
  (int)NUM2INT(options)
@@ -1321,93 +1714,366 @@ static VALUE native_write_to(
1321
1714
  return io;
1322
1715
  }
1323
1716
 
1324
- /*
1325
- * call-seq:
1326
- * line
1327
- *
1328
- * Returns the line for this Node
1329
- */
1330
- static VALUE line(VALUE self)
1717
+
1718
+ static inline void
1719
+ output_partial_string(VALUE out, char const *str, size_t length)
1331
1720
  {
1332
- xmlNodePtr node;
1333
- Data_Get_Struct(self, xmlNode, node);
1721
+ if (length) {
1722
+ rb_enc_str_buf_cat(out, str, (long)length, rb_utf8_encoding());
1723
+ }
1724
+ }
1334
1725
 
1335
- return INT2NUM(xmlGetLineNo(node));
1726
+ static inline void
1727
+ output_char(VALUE out, char ch)
1728
+ {
1729
+ output_partial_string(out, &ch, 1);
1336
1730
  }
1337
1731
 
1338
- /*
1339
- * call-seq:
1340
- * add_namespace_definition(prefix, href)
1341
- *
1342
- * Adds a namespace definition with +prefix+ using +href+ value. The result is
1343
- * as if parsed XML for this node had included an attribute
1344
- * 'xmlns:prefix=value'. A default namespace for this node ("xmlns=") can be
1345
- * added by passing 'nil' for prefix. Namespaces added this way will not
1346
- * show up in #attributes, but they will be included as an xmlns attribute
1347
- * when the node is serialized to XML.
1348
- */
1349
- static VALUE add_namespace_definition(VALUE self, VALUE prefix, VALUE href)
1732
+ static inline void
1733
+ output_string(VALUE out, char const *str)
1350
1734
  {
1351
- xmlNodePtr node, namespace;
1352
- xmlNsPtr ns;
1735
+ output_partial_string(out, str, strlen(str));
1736
+ }
1737
+
1738
+ static inline void
1739
+ output_tagname(VALUE out, xmlNodePtr elem)
1740
+ {
1741
+ // Elements in the HTML, MathML, and SVG namespaces do not use a namespace
1742
+ // prefix in the HTML syntax.
1743
+ char const *name = (char const *)elem->name;
1744
+ xmlNsPtr ns = elem->ns;
1745
+ if (ns && ns->href && ns->prefix
1746
+ && strcmp((char const *)ns->href, "http://www.w3.org/1999/xhtml")
1747
+ && strcmp((char const *)ns->href, "http://www.w3.org/1998/Math/MathML")
1748
+ && strcmp((char const *)ns->href, "http://www.w3.org/2000/svg")) {
1749
+ output_string(out, (char const *)elem->ns->prefix);
1750
+ output_char(out, ':');
1751
+ char const *colon = strchr(name, ':');
1752
+ if (colon) {
1753
+ name = colon + 1;
1754
+ }
1755
+ }
1756
+ output_string(out, name);
1757
+ }
1758
+
1759
+ static inline void
1760
+ output_attr_name(VALUE out, xmlAttrPtr attr)
1761
+ {
1762
+ xmlNsPtr ns = attr->ns;
1763
+ char const *name = (char const *)attr->name;
1764
+ if (ns && ns->href) {
1765
+ char const *uri = (char const *)ns->href;
1766
+ char const *localname = strchr(name, ':');
1767
+ if (localname) {
1768
+ ++localname;
1769
+ } else {
1770
+ localname = name;
1771
+ }
1772
+
1773
+ if (!strcmp(uri, "http://www.w3.org/XML/1998/namespace")) {
1774
+ output_string(out, "xml:");
1775
+ name = localname;
1776
+ } else if (!strcmp(uri, "http://www.w3.org/2000/xmlns/")) {
1777
+ // xmlns:xmlns -> xmlns
1778
+ // xmlns:foo -> xmlns:foo
1779
+ if (strcmp(localname, "xmlns")) {
1780
+ output_string(out, "xmlns:");
1781
+ }
1782
+ name = localname;
1783
+ } else if (!strcmp(uri, "http://www.w3.org/1999/xlink")) {
1784
+ output_string(out, "xlink:");
1785
+ name = localname;
1786
+ } else if (ns->prefix) {
1787
+ output_string(out, (char const *)ns->prefix);
1788
+ output_char(out, ':');
1789
+ name = localname;
1790
+ }
1791
+ }
1792
+ output_string(out, name);
1793
+ }
1794
+
1795
+ static void
1796
+ output_escaped_string(VALUE out, xmlChar const *start, bool attr)
1797
+ {
1798
+ xmlChar const *next = start;
1799
+ int ch;
1800
+
1801
+ while ((ch = *next) != 0) {
1802
+ char const *replacement = NULL;
1803
+ size_t replaced_bytes = 1;
1804
+ if (ch == '&') {
1805
+ replacement = "&amp;";
1806
+ } else if (ch == 0xC2 && next[1] == 0xA0) {
1807
+ // U+00A0 NO-BREAK SPACE has the UTF-8 encoding C2 A0.
1808
+ replacement = "&nbsp;";
1809
+ replaced_bytes = 2;
1810
+ } else if (attr && ch == '"') {
1811
+ replacement = "&quot;";
1812
+ } else if (!attr && ch == '<') {
1813
+ replacement = "&lt;";
1814
+ } else if (!attr && ch == '>') {
1815
+ replacement = "&gt;";
1816
+ } else {
1817
+ ++next;
1818
+ continue;
1819
+ }
1820
+ output_partial_string(out, (char const *)start, next - start);
1821
+ output_string(out, replacement);
1822
+ next += replaced_bytes;
1823
+ start = next;
1824
+ }
1825
+ output_partial_string(out, (char const *)start, next - start);
1826
+ }
1353
1827
 
1354
- Data_Get_Struct(self, xmlNode, node);
1355
- namespace = node ;
1828
+ static bool
1829
+ should_prepend_newline(xmlNodePtr node)
1830
+ {
1831
+ char const *name = (char const *)node->name;
1832
+ xmlNodePtr child = node->children;
1833
+
1834
+ if (!name || !child || (strcmp(name, "pre") && strcmp(name, "textarea") && strcmp(name, "listing"))) {
1835
+ return false;
1836
+ }
1837
+
1838
+ return child->type == XML_TEXT_NODE && child->content && child->content[0] == '\n';
1839
+ }
1356
1840
 
1357
- ns = xmlSearchNs(
1358
- node->doc,
1359
- node,
1360
- (const xmlChar *)(NIL_P(prefix) ? NULL : StringValueCStr(prefix))
1361
- );
1841
+ static VALUE
1842
+ rb_prepend_newline(VALUE self)
1843
+ {
1844
+ xmlNodePtr node;
1845
+ Noko_Node_Get_Struct(self, xmlNode, node);
1846
+ return should_prepend_newline(node) ? Qtrue : Qfalse;
1847
+ }
1362
1848
 
1363
- if(!ns) {
1364
- if (node->type != XML_ELEMENT_NODE) {
1365
- namespace = node->parent;
1849
+ static bool
1850
+ is_one_of(xmlNodePtr node, char const *const *tagnames, size_t num_tagnames)
1851
+ {
1852
+ char const *name = (char const *)node->name;
1853
+ if (name == NULL) { // fragments don't have a name
1854
+ return false;
1855
+ }
1856
+ for (size_t idx = 0; idx < num_tagnames; ++idx) {
1857
+ if (!strcmp(name, tagnames[idx])) {
1858
+ return true;
1366
1859
  }
1367
- ns = xmlNewNs(
1368
- namespace,
1369
- (const xmlChar *)StringValueCStr(href),
1370
- (const xmlChar *)(NIL_P(prefix) ? NULL : StringValueCStr(prefix))
1371
- );
1372
1860
  }
1861
+ return false;
1862
+
1863
+ }
1864
+
1865
+ static void
1866
+ output_node(
1867
+ VALUE out,
1868
+ xmlNodePtr node,
1869
+ bool preserve_newline
1870
+ )
1871
+ {
1872
+ static char const *const VOID_ELEMENTS[] = {
1873
+ "area", "base", "basefont", "bgsound", "br", "col", "embed", "frame", "hr",
1874
+ "img", "input", "keygen", "link", "meta", "param", "source", "track", "wbr",
1875
+ };
1876
+
1877
+ static char const *const UNESCAPED_TEXT_ELEMENTS[] = {
1878
+ "style", "script", "xmp", "iframe", "noembed", "noframes", "plaintext", "noscript",
1879
+ };
1880
+
1881
+ switch (node->type) {
1882
+ case XML_ELEMENT_NODE:
1883
+ // Serialize the start tag.
1884
+ output_char(out, '<');
1885
+ output_tagname(out, node);
1886
+
1887
+ // Add attributes.
1888
+ for (xmlAttrPtr attr = node->properties; attr; attr = attr->next) {
1889
+ output_char(out, ' ');
1890
+ output_attr_name(out, attr);
1891
+ if (attr->children) {
1892
+ output_string(out, "=\"");
1893
+ xmlChar *value = xmlNodeListGetString(attr->doc, attr->children, 1);
1894
+ output_escaped_string(out, value, true);
1895
+ xmlFree(value);
1896
+ output_char(out, '"');
1897
+ } else {
1898
+ // Output name=""
1899
+ output_string(out, "=\"\"");
1900
+ }
1901
+ }
1902
+ output_char(out, '>');
1903
+
1904
+ // Add children and end tag if element is not void.
1905
+ if (!is_one_of(node, VOID_ELEMENTS, sizeof VOID_ELEMENTS / sizeof VOID_ELEMENTS[0])) {
1906
+ if (preserve_newline && should_prepend_newline(node)) {
1907
+ output_char(out, '\n');
1908
+ }
1909
+ for (xmlNodePtr child = node->children; child; child = child->next) {
1910
+ output_node(out, child, preserve_newline);
1911
+ }
1912
+ output_string(out, "</");
1913
+ output_tagname(out, node);
1914
+ output_char(out, '>');
1915
+ }
1916
+ break;
1917
+
1918
+ case XML_TEXT_NODE:
1919
+ if (node->parent
1920
+ && is_one_of(node->parent, UNESCAPED_TEXT_ELEMENTS,
1921
+ sizeof UNESCAPED_TEXT_ELEMENTS / sizeof UNESCAPED_TEXT_ELEMENTS[0])) {
1922
+ output_string(out, (char const *)node->content);
1923
+ } else {
1924
+ output_escaped_string(out, node->content, false);
1925
+ }
1926
+ break;
1927
+
1928
+ case XML_CDATA_SECTION_NODE:
1929
+ output_string(out, "<![CDATA[");
1930
+ output_string(out, (char const *)node->content);
1931
+ output_string(out, "]]>");
1932
+ break;
1933
+
1934
+ case XML_COMMENT_NODE:
1935
+ output_string(out, "<!--");
1936
+ output_string(out, (char const *)node->content);
1937
+ output_string(out, "-->");
1938
+ break;
1939
+
1940
+ case XML_PI_NODE:
1941
+ output_string(out, "<?");
1942
+ output_string(out, (char const *)node->content);
1943
+ output_char(out, '>');
1944
+ break;
1945
+
1946
+ case XML_DOCUMENT_TYPE_NODE:
1947
+ case XML_DTD_NODE:
1948
+ output_string(out, "<!DOCTYPE ");
1949
+ output_string(out, (char const *)node->name);
1950
+ output_string(out, ">");
1951
+ break;
1952
+
1953
+ case XML_DOCUMENT_NODE:
1954
+ case XML_DOCUMENT_FRAG_NODE:
1955
+ case XML_HTML_DOCUMENT_NODE:
1956
+ for (xmlNodePtr child = node->children; child; child = child->next) {
1957
+ output_node(out, child, preserve_newline);
1958
+ }
1959
+ break;
1960
+
1961
+ default:
1962
+ rb_raise(rb_eRuntimeError, "Unsupported document node (%d); this is a bug in Nokogiri", node->type);
1963
+ break;
1964
+ }
1965
+ }
1373
1966
 
1374
- if (!ns) { return Qnil ; }
1967
+ static VALUE
1968
+ html_standard_serialize(
1969
+ VALUE self,
1970
+ VALUE preserve_newline
1971
+ )
1972
+ {
1973
+ xmlNodePtr node;
1974
+ Noko_Node_Get_Struct(self, xmlNode, node);
1975
+ VALUE output = rb_str_buf_new(4096);
1976
+ output_node(output, node, RTEST(preserve_newline));
1977
+ return output;
1978
+ }
1375
1979
 
1376
- if(NIL_P(prefix) || node != namespace) { xmlSetNs(node, ns); }
1980
+ /*
1981
+ * :call-seq:
1982
+ * line() → Integer
1983
+ *
1984
+ * [Returns] The line number of this Node.
1985
+ *
1986
+ * ---
1987
+ *
1988
+ * <b> ⚠ The CRuby and JRuby implementations differ in important ways! </b>
1989
+ *
1990
+ * Semantic differences:
1991
+ * - The CRuby method reflects the node's line number <i>in the parsed string</i>
1992
+ * - The JRuby method reflects the node's line number <i>in the final DOM structure</i> after
1993
+ * corrections have been applied
1994
+ *
1995
+ * Performance differences:
1996
+ * - The CRuby method is {O(1)}[https://en.wikipedia.org/wiki/Time_complexity#Constant_time]
1997
+ * (constant time)
1998
+ * - The JRuby method is {O(n)}[https://en.wikipedia.org/wiki/Time_complexity#Linear_time] (linear
1999
+ * time, where n is the number of nodes before/above the element in the DOM)
2000
+ *
2001
+ * If you'd like to help improve the JRuby implementation, please review these issues and reach out
2002
+ * to the maintainers:
2003
+ * - https://github.com/sparklemotion/nokogiri/issues/1223
2004
+ * - https://github.com/sparklemotion/nokogiri/pull/2177
2005
+ * - https://github.com/sparklemotion/nokogiri/issues/2380
2006
+ */
2007
+ static VALUE
2008
+ rb_xml_node_line(VALUE rb_node)
2009
+ {
2010
+ xmlNodePtr c_node;
2011
+ Noko_Node_Get_Struct(rb_node, xmlNode, c_node);
1377
2012
 
1378
- return Nokogiri_wrap_xml_namespace(node->doc, ns);
2013
+ return LONG2NUM(xmlGetLineNo(c_node));
1379
2014
  }
1380
2015
 
1381
2016
  /*
1382
2017
  * call-seq:
1383
- * new(name, document)
2018
+ * line=(num)
1384
2019
  *
1385
- * Create a new node with +name+ sharing GC lifecycle with +document+
2020
+ * Sets the line for this Node. num must be less than 65535.
1386
2021
  */
1387
- static VALUE new(int argc, VALUE *argv, VALUE klass)
2022
+ static VALUE
2023
+ rb_xml_node_line_set(VALUE rb_node, VALUE rb_line_number)
1388
2024
  {
1389
- xmlDocPtr doc;
1390
- xmlNodePtr node;
1391
- VALUE name;
1392
- VALUE document;
2025
+ xmlNodePtr c_node;
2026
+ int line_number = NUM2INT(rb_line_number);
2027
+
2028
+ Noko_Node_Get_Struct(rb_node, xmlNode, c_node);
2029
+
2030
+ // libxml2 optionally uses xmlNode.psvi to store longer line numbers, but only for text nodes.
2031
+ // search for "psvi" in SAX2.c and tree.c to learn more.
2032
+ if (line_number < 65535) {
2033
+ c_node->line = (short) line_number;
2034
+ } else {
2035
+ c_node->line = 65535;
2036
+ if (c_node->type == XML_TEXT_NODE) {
2037
+ c_node->psvi = (void *)(ptrdiff_t) line_number;
2038
+ }
2039
+ }
2040
+
2041
+ return rb_line_number;
2042
+ }
2043
+
2044
+ /* :nodoc: documented in lib/nokogiri/xml/node.rb */
2045
+ static VALUE
2046
+ rb_xml_node_new(int argc, VALUE *argv, VALUE klass)
2047
+ {
2048
+ xmlNodePtr c_document_node;
2049
+ xmlNodePtr c_node;
2050
+ VALUE rb_name;
2051
+ VALUE rb_document_node;
1393
2052
  VALUE rest;
1394
2053
  VALUE rb_node;
1395
2054
 
1396
- rb_scan_args(argc, argv, "2*", &name, &document, &rest);
2055
+ rb_scan_args(argc, argv, "2*", &rb_name, &rb_document_node, &rest);
1397
2056
 
1398
- Data_Get_Struct(document, xmlDoc, doc);
2057
+ if (!rb_obj_is_kind_of(rb_document_node, cNokogiriXmlNode)) {
2058
+ rb_raise(rb_eArgError, "document must be a Nokogiri::XML::Node");
2059
+ }
2060
+ if (!rb_obj_is_kind_of(rb_document_node, cNokogiriXmlDocument)) {
2061
+ // TODO: deprecate allowing Node
2062
+ NOKO_WARN_DEPRECATION("Passing a Node as the second parameter to Node.new is deprecated. Please pass a Document instead, or prefer an alternative constructor like Node#add_child. This will become an error in a future release of Nokogiri.");
2063
+ }
2064
+ Noko_Node_Get_Struct(rb_document_node, xmlNode, c_document_node);
1399
2065
 
1400
- node = xmlNewNode(NULL, (xmlChar *)StringValueCStr(name));
1401
- node->doc = doc->doc;
1402
- nokogiri_root_node(node);
2066
+ c_node = xmlNewNode(NULL, (xmlChar *)StringValueCStr(rb_name));
2067
+ c_node->doc = c_document_node->doc;
2068
+ noko_xml_document_pin_node(c_node);
1403
2069
 
1404
- rb_node = Nokogiri_wrap_xml_node(
2070
+ rb_node = noko_xml_node_wrap(
1405
2071
  klass == cNokogiriXmlNode ? (VALUE)NULL : klass,
1406
- node
2072
+ c_node
1407
2073
  );
1408
2074
  rb_obj_call_init(rb_node, argc, argv);
1409
2075
 
1410
- if(rb_block_given_p()) { rb_yield(rb_node); }
2076
+ if (rb_block_given_p()) { rb_yield(rb_node); }
1411
2077
 
1412
2078
  return rb_node;
1413
2079
  }
@@ -1418,13 +2084,14 @@ static VALUE new(int argc, VALUE *argv, VALUE klass)
1418
2084
  *
1419
2085
  * Returns the Node as html.
1420
2086
  */
1421
- static VALUE dump_html(VALUE self)
2087
+ static VALUE
2088
+ dump_html(VALUE self)
1422
2089
  {
1423
2090
  xmlBufferPtr buf ;
1424
2091
  xmlNodePtr node ;
1425
2092
  VALUE html;
1426
2093
 
1427
- Data_Get_Struct(self, xmlNode, node);
2094
+ Noko_Node_Get_Struct(self, xmlNode, node);
1428
2095
 
1429
2096
  buf = xmlBufferCreate() ;
1430
2097
  htmlNodeDump(buf, node->doc, node);
@@ -1439,13 +2106,14 @@ static VALUE dump_html(VALUE self)
1439
2106
  *
1440
2107
  * Compare this Node to +other+ with respect to their Document
1441
2108
  */
1442
- static VALUE compare(VALUE self, VALUE _other)
2109
+ static VALUE
2110
+ compare(VALUE self, VALUE _other)
1443
2111
  {
1444
2112
  xmlNodePtr node, other;
1445
- Data_Get_Struct(self, xmlNode, node);
1446
- Data_Get_Struct(_other, xmlNode, other);
2113
+ Noko_Node_Get_Struct(self, xmlNode, node);
2114
+ Noko_Node_Get_Struct(_other, xmlNode, other);
1447
2115
 
1448
- return INT2NUM((long)xmlXPathCmpNodes(other, node));
2116
+ return INT2NUM(xmlXPathCmpNodes(other, node));
1449
2117
  }
1450
2118
 
1451
2119
 
@@ -1456,13 +2124,14 @@ static VALUE compare(VALUE self, VALUE _other)
1456
2124
  * Loads and substitutes all xinclude elements below the node. The
1457
2125
  * parser context will be initialized with +options+.
1458
2126
  */
1459
- static VALUE process_xincludes(VALUE self, VALUE options)
2127
+ static VALUE
2128
+ process_xincludes(VALUE self, VALUE options)
1460
2129
  {
1461
2130
  int rcode ;
1462
2131
  xmlNodePtr node;
1463
2132
  VALUE error_list = rb_ary_new();
1464
2133
 
1465
- Data_Get_Struct(self, xmlNode, node);
2134
+ Noko_Node_Get_Struct(self, xmlNode, node);
1466
2135
 
1467
2136
  xmlSetStructuredErrorFunc((void *)error_list, Nokogiri_error_array_pusher);
1468
2137
  rcode = xmlXIncludeProcessTreeFlags(node, (int)NUM2INT(options));
@@ -1472,7 +2141,7 @@ static VALUE process_xincludes(VALUE self, VALUE options)
1472
2141
  xmlErrorPtr error;
1473
2142
 
1474
2143
  error = xmlGetLastError();
1475
- if(error) {
2144
+ if (error) {
1476
2145
  rb_exc_raise(Nokogiri_wrap_xml_syntax_error(error));
1477
2146
  } else {
1478
2147
  rb_raise(rb_eRuntimeError, "Could not perform xinclude substitution");
@@ -1484,7 +2153,8 @@ static VALUE process_xincludes(VALUE self, VALUE options)
1484
2153
 
1485
2154
 
1486
2155
  /* TODO: DOCUMENT ME */
1487
- static VALUE in_context(VALUE self, VALUE _str, VALUE _options)
2156
+ static VALUE
2157
+ in_context(VALUE self, VALUE _str, VALUE _options)
1488
2158
  {
1489
2159
  xmlNodePtr node, list = 0, tmp, child_iter, node_children, doc_children;
1490
2160
  xmlNodeSetPtr set;
@@ -1492,7 +2162,7 @@ static VALUE in_context(VALUE self, VALUE _str, VALUE _options)
1492
2162
  VALUE doc, err;
1493
2163
  int doc_is_empty;
1494
2164
 
1495
- Data_Get_Struct(self, xmlNode, node);
2165
+ Noko_Node_Get_Struct(self, xmlNode, node);
1496
2166
 
1497
2167
  doc = DOC_RUBY_OBJECT(node->doc);
1498
2168
  err = rb_iv_get(doc, "@errors");
@@ -1533,9 +2203,7 @@ static VALUE in_context(VALUE self, VALUE _str, VALUE _options)
1533
2203
  */
1534
2204
  child_iter = node->doc->children ;
1535
2205
  while (child_iter) {
1536
- if (child_iter->parent != (xmlNodePtr)node->doc) {
1537
- child_iter->parent = (xmlNodePtr)node->doc;
1538
- }
2206
+ child_iter->parent = (xmlNodePtr)node->doc;
1539
2207
  child_iter = child_iter->next;
1540
2208
  }
1541
2209
 
@@ -1545,12 +2213,17 @@ static VALUE in_context(VALUE self, VALUE _str, VALUE _options)
1545
2213
 
1546
2214
  xmlSetStructuredErrorFunc(NULL, NULL);
1547
2215
 
1548
- /* Workaround for a libxml2 bug where a parsing error may leave a broken
2216
+ /*
2217
+ * Workaround for a libxml2 bug where a parsing error may leave a broken
1549
2218
  * node reference in node->doc->children.
2219
+ *
2220
+ * https://bugzilla.gnome.org/show_bug.cgi?id=668155
2221
+ *
1550
2222
  * This workaround is limited to when a parse error occurs, the document
1551
2223
  * went from having no children to having children, and the context node is
1552
2224
  * part of a document fragment.
1553
- * https://bugzilla.gnome.org/show_bug.cgi?id=668155
2225
+ *
2226
+ * TODO: This was fixed in libxml 2.8.0 by 71a243d
1554
2227
  */
1555
2228
  if (error != XML_ERR_OK && doc_is_empty && node->doc->children != NULL) {
1556
2229
  child_iter = node;
@@ -1565,12 +2238,12 @@ static VALUE in_context(VALUE self, VALUE _str, VALUE _options)
1565
2238
 
1566
2239
  /* FIXME: This probably needs to handle more constants... */
1567
2240
  switch (error) {
1568
- case XML_ERR_INTERNAL_ERROR:
1569
- case XML_ERR_NO_MEMORY:
1570
- rb_raise(rb_eRuntimeError, "error parsing fragment (%d)", error);
1571
- break;
1572
- default:
1573
- break;
2241
+ case XML_ERR_INTERNAL_ERROR:
2242
+ case XML_ERR_NO_MEMORY:
2243
+ rb_raise(rb_eRuntimeError, "error parsing fragment (%d)", error);
2244
+ break;
2245
+ default:
2246
+ break;
1574
2247
  }
1575
2248
 
1576
2249
  set = xmlXPathNodeSetCreate(NULL);
@@ -1579,178 +2252,174 @@ static VALUE in_context(VALUE self, VALUE _str, VALUE _options)
1579
2252
  tmp = list->next;
1580
2253
  list->next = NULL;
1581
2254
  xmlXPathNodeSetAddUnique(set, list);
1582
- nokogiri_root_node(list);
2255
+ noko_xml_document_pin_node(list);
1583
2256
  list = tmp;
1584
2257
  }
1585
2258
 
1586
- return Nokogiri_wrap_xml_node_set(set, doc);
2259
+ return noko_xml_node_set_wrap(set, doc);
1587
2260
  }
1588
2261
 
1589
-
1590
- VALUE Nokogiri_wrap_xml_node(VALUE klass, xmlNodePtr node)
2262
+ VALUE
2263
+ noko_xml_node_wrap(VALUE rb_class, xmlNodePtr c_node)
1591
2264
  {
1592
- VALUE document = Qnil ;
1593
- VALUE node_cache = Qnil ;
1594
- VALUE rb_node = Qnil ;
2265
+ VALUE rb_document, rb_node_cache, rb_node;
1595
2266
  nokogiriTuplePtr node_has_a_document;
1596
- xmlDocPtr doc;
1597
- void (*mark_method)(xmlNodePtr) = NULL ;
2267
+ xmlDocPtr c_doc;
1598
2268
 
1599
- assert(node);
2269
+ assert(c_node);
1600
2270
 
1601
- if(node->type == XML_DOCUMENT_NODE || node->type == XML_HTML_DOCUMENT_NODE) {
1602
- return DOC_RUBY_OBJECT(node->doc);
2271
+ if (c_node->type == XML_DOCUMENT_NODE || c_node->type == XML_HTML_DOCUMENT_NODE) {
2272
+ return DOC_RUBY_OBJECT(c_node->doc);
1603
2273
  }
1604
2274
 
1605
- /* It's OK if the node doesn't have a fully-realized document (as in XML::Reader). */
1606
- /* see https://github.com/sparklemotion/nokogiri/issues/95 */
1607
- /* and https://github.com/sparklemotion/nokogiri/issues/439 */
1608
- doc = node->doc;
1609
- if (doc->type == XML_DOCUMENT_FRAG_NODE) { doc = doc->doc; }
1610
- node_has_a_document = DOC_RUBY_OBJECT_TEST(doc);
2275
+ c_doc = c_node->doc;
2276
+
2277
+ // Nodes yielded from XML::Reader don't have a fully-realized Document
2278
+ node_has_a_document = DOC_RUBY_OBJECT_TEST(c_doc);
1611
2279
 
1612
- if(node->_private && node_has_a_document) {
1613
- return (VALUE)node->_private;
2280
+ if (c_node->_private && node_has_a_document) {
2281
+ return (VALUE)c_node->_private;
1614
2282
  }
1615
2283
 
1616
- if(!RTEST(klass)) {
1617
- switch(node->type) {
1618
- case XML_ELEMENT_NODE:
1619
- klass = cNokogiriXmlElement;
1620
- break;
1621
- case XML_TEXT_NODE:
1622
- klass = cNokogiriXmlText;
1623
- break;
1624
- case XML_ATTRIBUTE_NODE:
1625
- klass = cNokogiriXmlAttr;
1626
- break;
1627
- case XML_ENTITY_REF_NODE:
1628
- klass = cNokogiriXmlEntityReference;
1629
- break;
1630
- case XML_COMMENT_NODE:
1631
- klass = cNokogiriXmlComment;
1632
- break;
1633
- case XML_DOCUMENT_FRAG_NODE:
1634
- klass = cNokogiriXmlDocumentFragment;
1635
- break;
1636
- case XML_PI_NODE:
1637
- klass = cNokogiriXmlProcessingInstruction;
1638
- break;
1639
- case XML_ENTITY_DECL:
1640
- klass = cNokogiriXmlEntityDecl;
1641
- break;
1642
- case XML_CDATA_SECTION_NODE:
1643
- klass = cNokogiriXmlCData;
1644
- break;
1645
- case XML_DTD_NODE:
1646
- klass = cNokogiriXmlDtd;
1647
- break;
1648
- case XML_ATTRIBUTE_DECL:
1649
- klass = cNokogiriXmlAttributeDecl;
1650
- break;
1651
- case XML_ELEMENT_DECL:
1652
- klass = cNokogiriXmlElementDecl;
1653
- break;
1654
- default:
1655
- klass = cNokogiriXmlNode;
2284
+ if (!RTEST(rb_class)) {
2285
+ switch (c_node->type) {
2286
+ case XML_ELEMENT_NODE:
2287
+ rb_class = cNokogiriXmlElement;
2288
+ break;
2289
+ case XML_TEXT_NODE:
2290
+ rb_class = cNokogiriXmlText;
2291
+ break;
2292
+ case XML_ATTRIBUTE_NODE:
2293
+ rb_class = cNokogiriXmlAttr;
2294
+ break;
2295
+ case XML_ENTITY_REF_NODE:
2296
+ rb_class = cNokogiriXmlEntityReference;
2297
+ break;
2298
+ case XML_COMMENT_NODE:
2299
+ rb_class = cNokogiriXmlComment;
2300
+ break;
2301
+ case XML_DOCUMENT_FRAG_NODE:
2302
+ rb_class = cNokogiriXmlDocumentFragment;
2303
+ break;
2304
+ case XML_PI_NODE:
2305
+ rb_class = cNokogiriXmlProcessingInstruction;
2306
+ break;
2307
+ case XML_ENTITY_DECL:
2308
+ rb_class = cNokogiriXmlEntityDecl;
2309
+ break;
2310
+ case XML_CDATA_SECTION_NODE:
2311
+ rb_class = cNokogiriXmlCData;
2312
+ break;
2313
+ case XML_DTD_NODE:
2314
+ rb_class = cNokogiriXmlDtd;
2315
+ break;
2316
+ case XML_ATTRIBUTE_DECL:
2317
+ rb_class = cNokogiriXmlAttributeDecl;
2318
+ break;
2319
+ case XML_ELEMENT_DECL:
2320
+ rb_class = cNokogiriXmlElementDecl;
2321
+ break;
2322
+ default:
2323
+ rb_class = cNokogiriXmlNode;
1656
2324
  }
1657
2325
  }
1658
2326
 
1659
- mark_method = node_has_a_document ? mark : NULL ;
1660
-
1661
- rb_node = Data_Wrap_Struct(klass, mark_method, debug_node_dealloc, node) ;
1662
- node->_private = (void *)rb_node;
2327
+ rb_node = TypedData_Wrap_Struct(rb_class, &nokogiri_node_type, c_node) ;
2328
+ c_node->_private = (void *)rb_node;
1663
2329
 
1664
2330
  if (node_has_a_document) {
1665
- document = DOC_RUBY_OBJECT(doc);
1666
- node_cache = DOC_NODE_CACHE(doc);
1667
- rb_ary_push(node_cache, rb_node);
1668
- rb_funcall(document, decorate, 1, rb_node);
2331
+ rb_document = DOC_RUBY_OBJECT(c_doc);
2332
+ rb_node_cache = DOC_NODE_CACHE(c_doc);
2333
+ rb_ary_push(rb_node_cache, rb_node);
2334
+ rb_funcall(rb_document, id_decorate, 1, rb_node);
1669
2335
  }
1670
2336
 
1671
2337
  return rb_node ;
1672
2338
  }
1673
2339
 
1674
2340
 
1675
- void Nokogiri_xml_node_properties(xmlNodePtr node, VALUE attr_list)
2341
+ /*
2342
+ * return Array<Nokogiri::XML::Attr> containing the node's attributes
2343
+ */
2344
+ VALUE
2345
+ noko_xml_node_attrs(xmlNodePtr c_node)
1676
2346
  {
1677
- xmlAttrPtr prop;
1678
- prop = node->properties ;
1679
- while (prop != NULL) {
1680
- rb_ary_push(attr_list, Nokogiri_wrap_xml_node(Qnil, (xmlNodePtr)prop));
1681
- prop = prop->next ;
2347
+ VALUE rb_properties = rb_ary_new();
2348
+ xmlAttrPtr c_property;
2349
+
2350
+ c_property = c_node->properties ;
2351
+ while (c_property != NULL) {
2352
+ rb_ary_push(rb_properties, noko_xml_node_wrap(Qnil, (xmlNodePtr)c_property));
2353
+ c_property = c_property->next ;
1682
2354
  }
2355
+
2356
+ return rb_properties;
1683
2357
  }
1684
2358
 
1685
- VALUE cNokogiriXmlNode ;
1686
- VALUE cNokogiriXmlElement ;
1687
-
1688
- void init_xml_node()
1689
- {
1690
- VALUE nokogiri = rb_define_module("Nokogiri");
1691
- VALUE xml = rb_define_module_under(nokogiri, "XML");
1692
- VALUE klass = rb_define_class_under(xml, "Node", rb_cObject);
1693
-
1694
- cNokogiriXmlNode = klass;
1695
-
1696
- cNokogiriXmlElement = rb_define_class_under(xml, "Element", klass);
1697
-
1698
- rb_define_singleton_method(klass, "new", new, -1);
1699
-
1700
- rb_define_method(klass, "add_namespace_definition", add_namespace_definition, 2);
1701
- rb_define_method(klass, "node_name", get_name, 0);
1702
- rb_define_method(klass, "document", document, 0);
1703
- rb_define_method(klass, "node_name=", set_name, 1);
1704
- rb_define_method(klass, "parent", get_parent, 0);
1705
- rb_define_method(klass, "child", child, 0);
1706
- rb_define_method(klass, "first_element_child", first_element_child, 0);
1707
- rb_define_method(klass, "last_element_child", last_element_child, 0);
1708
- rb_define_method(klass, "children", children, 0);
1709
- rb_define_method(klass, "element_children", element_children, 0);
1710
- rb_define_method(klass, "next_sibling", next_sibling, 0);
1711
- rb_define_method(klass, "previous_sibling", previous_sibling, 0);
1712
- rb_define_method(klass, "next_element", next_element, 0);
1713
- rb_define_method(klass, "previous_element", previous_element, 0);
1714
- rb_define_method(klass, "node_type", node_type, 0);
1715
- rb_define_method(klass, "path", path, 0);
1716
- rb_define_method(klass, "key?", key_eh, 1);
1717
- rb_define_method(klass, "namespaced_key?", namespaced_key_eh, 2);
1718
- rb_define_method(klass, "blank?", blank_eh, 0);
1719
- rb_define_method(klass, "attribute_nodes", attribute_nodes, 0);
1720
- rb_define_method(klass, "attribute", attr, 1);
1721
- rb_define_method(klass, "attribute_with_ns", attribute_with_ns, 2);
1722
- rb_define_method(klass, "namespace", namespace, 0);
1723
- rb_define_method(klass, "namespace_definitions", namespace_definitions, 0);
1724
- rb_define_method(klass, "namespace_scopes", namespace_scopes, 0);
1725
- rb_define_method(klass, "encode_special_chars", encode_special_chars, 1);
1726
- rb_define_method(klass, "dup", duplicate_node, -1);
1727
- rb_define_method(klass, "unlink", unlink_node, 0);
1728
- rb_define_method(klass, "internal_subset", internal_subset, 0);
1729
- rb_define_method(klass, "external_subset", external_subset, 0);
1730
- rb_define_method(klass, "create_internal_subset", create_internal_subset, 3);
1731
- rb_define_method(klass, "create_external_subset", create_external_subset, 3);
1732
- rb_define_method(klass, "pointer_id", pointer_id, 0);
1733
- rb_define_method(klass, "line", line, 0);
1734
- rb_define_method(klass, "content", get_native_content, 0);
1735
- rb_define_method(klass, "native_content=", set_native_content, 1);
1736
- rb_define_method(klass, "lang", get_lang, 0);
1737
- rb_define_method(klass, "lang=", set_lang, 1);
1738
-
1739
- rb_define_private_method(klass, "process_xincludes", process_xincludes, 1);
1740
- rb_define_private_method(klass, "in_context", in_context, 2);
1741
- rb_define_private_method(klass, "add_child_node", add_child, 1);
1742
- rb_define_private_method(klass, "add_previous_sibling_node", add_previous_sibling, 1);
1743
- rb_define_private_method(klass, "add_next_sibling_node", add_next_sibling, 1);
1744
- rb_define_private_method(klass, "replace_node", replace, 1);
1745
- rb_define_private_method(klass, "dump_html", dump_html, 0);
1746
- rb_define_private_method(klass, "native_write_to", native_write_to, 4);
1747
- rb_define_private_method(klass, "get", get, 1);
1748
- rb_define_private_method(klass, "set", set, 2);
1749
- rb_define_private_method(klass, "set_namespace", set_namespace, 1);
1750
- rb_define_private_method(klass, "compare", compare, 1);
1751
-
1752
- decorate = rb_intern("decorate");
1753
- decorate_bang = rb_intern("decorate!");
1754
- }
1755
-
1756
- /* vim: set noet sw=4 sws=4 */
2359
+ void
2360
+ noko_init_xml_node(void)
2361
+ {
2362
+ cNokogiriXmlNode = rb_define_class_under(mNokogiriXml, "Node", rb_cObject);
2363
+
2364
+ rb_undef_alloc_func(cNokogiriXmlNode);
2365
+
2366
+ rb_define_singleton_method(cNokogiriXmlNode, "new", rb_xml_node_new, -1);
2367
+
2368
+ rb_define_method(cNokogiriXmlNode, "add_namespace_definition", rb_xml_node_add_namespace_definition, 2);
2369
+ rb_define_method(cNokogiriXmlNode, "attribute", rb_xml_node_attribute, 1);
2370
+ rb_define_method(cNokogiriXmlNode, "attribute_nodes", rb_xml_node_attribute_nodes, 0);
2371
+ rb_define_method(cNokogiriXmlNode, "attribute_with_ns", rb_xml_node_attribute_with_ns, 2);
2372
+ rb_define_method(cNokogiriXmlNode, "blank?", rb_xml_node_blank_eh, 0);
2373
+ rb_define_method(cNokogiriXmlNode, "child", rb_xml_node_child, 0);
2374
+ rb_define_method(cNokogiriXmlNode, "children", rb_xml_node_children, 0);
2375
+ rb_define_method(cNokogiriXmlNode, "content", rb_xml_node_content, 0);
2376
+ rb_define_method(cNokogiriXmlNode, "create_external_subset", create_external_subset, 3);
2377
+ rb_define_method(cNokogiriXmlNode, "create_internal_subset", create_internal_subset, 3);
2378
+ rb_define_method(cNokogiriXmlNode, "document", rb_xml_node_document, 0);
2379
+ rb_define_method(cNokogiriXmlNode, "dup", duplicate_node, -1);
2380
+ rb_define_method(cNokogiriXmlNode, "element_children", rb_xml_node_element_children, 0);
2381
+ rb_define_method(cNokogiriXmlNode, "encode_special_chars", encode_special_chars, 1);
2382
+ rb_define_method(cNokogiriXmlNode, "external_subset", external_subset, 0);
2383
+ rb_define_method(cNokogiriXmlNode, "first_element_child", rb_xml_node_first_element_child, 0);
2384
+ rb_define_method(cNokogiriXmlNode, "internal_subset", internal_subset, 0);
2385
+ rb_define_method(cNokogiriXmlNode, "key?", key_eh, 1);
2386
+ rb_define_method(cNokogiriXmlNode, "lang", get_lang, 0);
2387
+ rb_define_method(cNokogiriXmlNode, "lang=", set_lang, 1);
2388
+ rb_define_method(cNokogiriXmlNode, "last_element_child", rb_xml_node_last_element_child, 0);
2389
+ rb_define_method(cNokogiriXmlNode, "line", rb_xml_node_line, 0);
2390
+ rb_define_method(cNokogiriXmlNode, "line=", rb_xml_node_line_set, 1);
2391
+ rb_define_method(cNokogiriXmlNode, "namespace", rb_xml_node_namespace, 0);
2392
+ rb_define_method(cNokogiriXmlNode, "namespace_definitions", namespace_definitions, 0);
2393
+ rb_define_method(cNokogiriXmlNode, "namespace_scopes", rb_xml_node_namespace_scopes, 0);
2394
+ rb_define_method(cNokogiriXmlNode, "namespaced_key?", namespaced_key_eh, 2);
2395
+ rb_define_method(cNokogiriXmlNode, "native_content=", set_native_content, 1);
2396
+ rb_define_method(cNokogiriXmlNode, "next_element", next_element, 0);
2397
+ rb_define_method(cNokogiriXmlNode, "next_sibling", next_sibling, 0);
2398
+ rb_define_method(cNokogiriXmlNode, "node_name", get_name, 0);
2399
+ rb_define_method(cNokogiriXmlNode, "node_name=", set_name, 1);
2400
+ rb_define_method(cNokogiriXmlNode, "node_type", node_type, 0);
2401
+ rb_define_method(cNokogiriXmlNode, "parent", get_parent, 0);
2402
+ rb_define_method(cNokogiriXmlNode, "path", rb_xml_node_path, 0);
2403
+ rb_define_method(cNokogiriXmlNode, "pointer_id", rb_xml_node_pointer_id, 0);
2404
+ rb_define_method(cNokogiriXmlNode, "previous_element", previous_element, 0);
2405
+ rb_define_method(cNokogiriXmlNode, "previous_sibling", previous_sibling, 0);
2406
+ rb_define_method(cNokogiriXmlNode, "unlink", unlink_node, 0);
2407
+
2408
+ rb_define_private_method(cNokogiriXmlNode, "add_child_node", add_child, 1);
2409
+ rb_define_private_method(cNokogiriXmlNode, "add_next_sibling_node", add_next_sibling, 1);
2410
+ rb_define_private_method(cNokogiriXmlNode, "add_previous_sibling_node", add_previous_sibling, 1);
2411
+ rb_define_private_method(cNokogiriXmlNode, "compare", compare, 1);
2412
+ rb_define_private_method(cNokogiriXmlNode, "dump_html", dump_html, 0);
2413
+ rb_define_private_method(cNokogiriXmlNode, "get", get, 1);
2414
+ rb_define_private_method(cNokogiriXmlNode, "in_context", in_context, 2);
2415
+ rb_define_private_method(cNokogiriXmlNode, "native_write_to", native_write_to, 4);
2416
+ rb_define_private_method(cNokogiriXmlNode, "prepend_newline?", rb_prepend_newline, 0);
2417
+ rb_define_private_method(cNokogiriXmlNode, "html_standard_serialize", html_standard_serialize, 1);
2418
+ rb_define_private_method(cNokogiriXmlNode, "process_xincludes", process_xincludes, 1);
2419
+ rb_define_private_method(cNokogiriXmlNode, "replace_node", replace, 1);
2420
+ rb_define_private_method(cNokogiriXmlNode, "set", set, 2);
2421
+ rb_define_private_method(cNokogiriXmlNode, "set_namespace", set_namespace, 1);
2422
+
2423
+ id_decorate = rb_intern("decorate");
2424
+ id_decorate_bang = rb_intern("decorate!");
2425
+ }