nokogiri 1.10.9 → 1.18.3

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.

Potentially problematic release.


This version of nokogiri might be problematic. Click here for more details.

Files changed (230) hide show
  1. checksums.yaml +4 -4
  2. data/Gemfile +38 -0
  3. data/LICENSE-DEPENDENCIES.md +1632 -1022
  4. data/LICENSE.md +1 -1
  5. data/README.md +190 -95
  6. data/bin/nokogiri +63 -50
  7. data/dependencies.yml +34 -66
  8. data/ext/nokogiri/depend +38 -358
  9. data/ext/nokogiri/extconf.rb +909 -422
  10. data/ext/nokogiri/gumbo.c +610 -0
  11. data/ext/nokogiri/html4_document.c +171 -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.c +40 -0
  15. data/ext/nokogiri/html4_sax_parser_context.c +98 -0
  16. data/ext/nokogiri/html4_sax_push_parser.c +96 -0
  17. data/ext/nokogiri/libxml2_polyfill.c +114 -0
  18. data/ext/nokogiri/nokogiri.c +258 -105
  19. data/ext/nokogiri/nokogiri.h +207 -90
  20. data/ext/nokogiri/test_global_handlers.c +40 -0
  21. data/ext/nokogiri/xml_attr.c +18 -18
  22. data/ext/nokogiri/xml_attribute_decl.c +22 -22
  23. data/ext/nokogiri/xml_cdata.c +33 -33
  24. data/ext/nokogiri/xml_comment.c +19 -31
  25. data/ext/nokogiri/xml_document.c +499 -323
  26. data/ext/nokogiri/xml_document_fragment.c +17 -36
  27. data/ext/nokogiri/xml_dtd.c +65 -59
  28. data/ext/nokogiri/xml_element_content.c +63 -55
  29. data/ext/nokogiri/xml_element_decl.c +31 -31
  30. data/ext/nokogiri/xml_encoding_handler.c +54 -21
  31. data/ext/nokogiri/xml_entity_decl.c +37 -35
  32. data/ext/nokogiri/xml_entity_reference.c +17 -19
  33. data/ext/nokogiri/xml_namespace.c +131 -61
  34. data/ext/nokogiri/xml_node.c +1429 -723
  35. data/ext/nokogiri/xml_node_set.c +257 -225
  36. data/ext/nokogiri/xml_processing_instruction.c +18 -20
  37. data/ext/nokogiri/xml_reader.c +340 -231
  38. data/ext/nokogiri/xml_relax_ng.c +87 -99
  39. data/ext/nokogiri/xml_sax_parser.c +269 -176
  40. data/ext/nokogiri/xml_sax_parser_context.c +286 -152
  41. data/ext/nokogiri/xml_sax_push_parser.c +111 -64
  42. data/ext/nokogiri/xml_schema.c +132 -140
  43. data/ext/nokogiri/xml_syntax_error.c +52 -23
  44. data/ext/nokogiri/xml_text.c +37 -30
  45. data/ext/nokogiri/xml_xpath_context.c +373 -185
  46. data/ext/nokogiri/xslt_stylesheet.c +342 -191
  47. data/gumbo-parser/CHANGES.md +63 -0
  48. data/gumbo-parser/Makefile +129 -0
  49. data/gumbo-parser/THANKS +27 -0
  50. data/gumbo-parser/src/Makefile +34 -0
  51. data/gumbo-parser/src/README.md +41 -0
  52. data/gumbo-parser/src/ascii.c +75 -0
  53. data/gumbo-parser/src/ascii.h +115 -0
  54. data/gumbo-parser/src/attribute.c +42 -0
  55. data/gumbo-parser/src/attribute.h +17 -0
  56. data/gumbo-parser/src/char_ref.c +22225 -0
  57. data/gumbo-parser/src/char_ref.h +29 -0
  58. data/gumbo-parser/src/char_ref.rl +2154 -0
  59. data/gumbo-parser/src/error.c +658 -0
  60. data/gumbo-parser/src/error.h +152 -0
  61. data/gumbo-parser/src/foreign_attrs.c +103 -0
  62. data/gumbo-parser/src/foreign_attrs.gperf +27 -0
  63. data/gumbo-parser/src/insertion_mode.h +33 -0
  64. data/gumbo-parser/src/macros.h +91 -0
  65. data/gumbo-parser/src/nokogiri_gumbo.h +953 -0
  66. data/gumbo-parser/src/parser.c +4932 -0
  67. data/gumbo-parser/src/parser.h +41 -0
  68. data/gumbo-parser/src/replacement.h +33 -0
  69. data/gumbo-parser/src/string_buffer.c +103 -0
  70. data/gumbo-parser/src/string_buffer.h +68 -0
  71. data/gumbo-parser/src/string_piece.c +48 -0
  72. data/gumbo-parser/src/svg_attrs.c +174 -0
  73. data/gumbo-parser/src/svg_attrs.gperf +77 -0
  74. data/gumbo-parser/src/svg_tags.c +137 -0
  75. data/gumbo-parser/src/svg_tags.gperf +55 -0
  76. data/gumbo-parser/src/tag.c +223 -0
  77. data/gumbo-parser/src/tag_lookup.c +382 -0
  78. data/gumbo-parser/src/tag_lookup.gperf +170 -0
  79. data/gumbo-parser/src/tag_lookup.h +13 -0
  80. data/gumbo-parser/src/token_buffer.c +79 -0
  81. data/gumbo-parser/src/token_buffer.h +71 -0
  82. data/gumbo-parser/src/token_type.h +17 -0
  83. data/gumbo-parser/src/tokenizer.c +3464 -0
  84. data/gumbo-parser/src/tokenizer.h +112 -0
  85. data/gumbo-parser/src/tokenizer_states.h +339 -0
  86. data/gumbo-parser/src/utf8.c +245 -0
  87. data/gumbo-parser/src/utf8.h +164 -0
  88. data/gumbo-parser/src/util.c +66 -0
  89. data/gumbo-parser/src/util.h +34 -0
  90. data/gumbo-parser/src/vector.c +111 -0
  91. data/gumbo-parser/src/vector.h +45 -0
  92. data/lib/nokogiri/class_resolver.rb +67 -0
  93. data/lib/nokogiri/css/node.rb +14 -8
  94. data/lib/nokogiri/css/parser.rb +399 -377
  95. data/lib/nokogiri/css/parser.y +250 -245
  96. data/lib/nokogiri/css/parser_extras.rb +16 -71
  97. data/lib/nokogiri/css/selector_cache.rb +38 -0
  98. data/lib/nokogiri/css/syntax_error.rb +3 -1
  99. data/lib/nokogiri/css/tokenizer.rb +7 -5
  100. data/lib/nokogiri/css/tokenizer.rex +11 -9
  101. data/lib/nokogiri/css/xpath_visitor.rb +242 -96
  102. data/lib/nokogiri/css.rb +122 -17
  103. data/lib/nokogiri/decorators/slop.rb +11 -11
  104. data/lib/nokogiri/encoding_handler.rb +57 -0
  105. data/lib/nokogiri/extension.rb +32 -0
  106. data/lib/nokogiri/gumbo.rb +15 -0
  107. data/lib/nokogiri/html.rb +38 -27
  108. data/lib/nokogiri/{html → html4}/builder.rb +4 -2
  109. data/lib/nokogiri/html4/document.rb +235 -0
  110. data/lib/nokogiri/html4/document_fragment.rb +166 -0
  111. data/lib/nokogiri/{html → html4}/element_description.rb +3 -1
  112. data/lib/nokogiri/html4/element_description_defaults.rb +2040 -0
  113. data/lib/nokogiri/html4/encoding_reader.rb +121 -0
  114. data/lib/nokogiri/{html → html4}/entity_lookup.rb +4 -2
  115. data/lib/nokogiri/html4/sax/parser.rb +48 -0
  116. data/lib/nokogiri/html4/sax/parser_context.rb +15 -0
  117. data/lib/nokogiri/{html → html4}/sax/push_parser.rb +12 -11
  118. data/lib/nokogiri/html4.rb +42 -0
  119. data/lib/nokogiri/html5/builder.rb +40 -0
  120. data/lib/nokogiri/html5/document.rb +199 -0
  121. data/lib/nokogiri/html5/document_fragment.rb +200 -0
  122. data/lib/nokogiri/html5/node.rb +103 -0
  123. data/lib/nokogiri/html5.rb +368 -0
  124. data/lib/nokogiri/jruby/dependencies.rb +3 -0
  125. data/lib/nokogiri/jruby/nokogiri_jars.rb +43 -0
  126. data/lib/nokogiri/syntax_error.rb +2 -0
  127. data/lib/nokogiri/version/constant.rb +6 -0
  128. data/lib/nokogiri/version/info.rb +224 -0
  129. data/lib/nokogiri/version.rb +3 -108
  130. data/lib/nokogiri/xml/attr.rb +55 -3
  131. data/lib/nokogiri/xml/attribute_decl.rb +6 -2
  132. data/lib/nokogiri/xml/builder.rb +83 -35
  133. data/lib/nokogiri/xml/cdata.rb +3 -1
  134. data/lib/nokogiri/xml/character_data.rb +2 -0
  135. data/lib/nokogiri/xml/document.rb +359 -130
  136. data/lib/nokogiri/xml/document_fragment.rb +170 -54
  137. data/lib/nokogiri/xml/dtd.rb +4 -2
  138. data/lib/nokogiri/xml/element_content.rb +12 -2
  139. data/lib/nokogiri/xml/element_decl.rb +6 -2
  140. data/lib/nokogiri/xml/entity_decl.rb +7 -3
  141. data/lib/nokogiri/xml/entity_reference.rb +2 -0
  142. data/lib/nokogiri/xml/namespace.rb +44 -0
  143. data/lib/nokogiri/xml/node/save_options.rb +23 -8
  144. data/lib/nokogiri/xml/node.rb +1168 -420
  145. data/lib/nokogiri/xml/node_set.rb +145 -67
  146. data/lib/nokogiri/xml/notation.rb +13 -0
  147. data/lib/nokogiri/xml/parse_options.rb +145 -52
  148. data/lib/nokogiri/xml/pp/character_data.rb +9 -6
  149. data/lib/nokogiri/xml/pp/node.rb +47 -30
  150. data/lib/nokogiri/xml/pp.rb +4 -2
  151. data/lib/nokogiri/xml/processing_instruction.rb +4 -1
  152. data/lib/nokogiri/xml/reader.rb +68 -41
  153. data/lib/nokogiri/xml/relax_ng.rb +60 -17
  154. data/lib/nokogiri/xml/sax/document.rb +198 -111
  155. data/lib/nokogiri/xml/sax/parser.rb +144 -67
  156. data/lib/nokogiri/xml/sax/parser_context.rb +119 -6
  157. data/lib/nokogiri/xml/sax/push_parser.rb +9 -5
  158. data/lib/nokogiri/xml/sax.rb +54 -4
  159. data/lib/nokogiri/xml/schema.rb +116 -39
  160. data/lib/nokogiri/xml/searchable.rb +139 -95
  161. data/lib/nokogiri/xml/syntax_error.rb +29 -5
  162. data/lib/nokogiri/xml/text.rb +2 -0
  163. data/lib/nokogiri/xml/xpath/syntax_error.rb +4 -2
  164. data/lib/nokogiri/xml/xpath.rb +15 -4
  165. data/lib/nokogiri/xml/xpath_context.rb +15 -4
  166. data/lib/nokogiri/xml.rb +45 -55
  167. data/lib/nokogiri/xslt/stylesheet.rb +32 -8
  168. data/lib/nokogiri/xslt.rb +103 -30
  169. data/lib/nokogiri.rb +59 -75
  170. data/lib/xsd/xmlparser/nokogiri.rb +32 -29
  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/libxml2/0019-xpath-Use-separate-static-hash-table-for-standard-fu.patch +244 -0
  175. data/patches/libxslt/0001-update-config.guess-and-config.sub-for-libxslt.patch +224 -0
  176. data/ports/archives/libxml2-2.13.6.tar.xz +0 -0
  177. data/ports/archives/libxslt-1.1.42.tar.xz +0 -0
  178. metadata +123 -295
  179. data/ext/nokogiri/html_document.c +0 -170
  180. data/ext/nokogiri/html_document.h +0 -10
  181. data/ext/nokogiri/html_element_description.c +0 -279
  182. data/ext/nokogiri/html_element_description.h +0 -10
  183. data/ext/nokogiri/html_entity_lookup.c +0 -32
  184. data/ext/nokogiri/html_entity_lookup.h +0 -8
  185. data/ext/nokogiri/html_sax_parser_context.c +0 -116
  186. data/ext/nokogiri/html_sax_parser_context.h +0 -11
  187. data/ext/nokogiri/html_sax_push_parser.c +0 -87
  188. data/ext/nokogiri/html_sax_push_parser.h +0 -9
  189. data/ext/nokogiri/xml_attr.h +0 -9
  190. data/ext/nokogiri/xml_attribute_decl.h +0 -9
  191. data/ext/nokogiri/xml_cdata.h +0 -9
  192. data/ext/nokogiri/xml_comment.h +0 -9
  193. data/ext/nokogiri/xml_document.h +0 -23
  194. data/ext/nokogiri/xml_document_fragment.h +0 -10
  195. data/ext/nokogiri/xml_dtd.h +0 -10
  196. data/ext/nokogiri/xml_element_content.h +0 -10
  197. data/ext/nokogiri/xml_element_decl.h +0 -9
  198. data/ext/nokogiri/xml_encoding_handler.h +0 -8
  199. data/ext/nokogiri/xml_entity_decl.h +0 -10
  200. data/ext/nokogiri/xml_entity_reference.h +0 -9
  201. data/ext/nokogiri/xml_io.c +0 -61
  202. data/ext/nokogiri/xml_io.h +0 -11
  203. data/ext/nokogiri/xml_libxml2_hacks.c +0 -112
  204. data/ext/nokogiri/xml_libxml2_hacks.h +0 -12
  205. data/ext/nokogiri/xml_namespace.h +0 -14
  206. data/ext/nokogiri/xml_node.h +0 -13
  207. data/ext/nokogiri/xml_node_set.h +0 -12
  208. data/ext/nokogiri/xml_processing_instruction.h +0 -9
  209. data/ext/nokogiri/xml_reader.h +0 -10
  210. data/ext/nokogiri/xml_relax_ng.h +0 -9
  211. data/ext/nokogiri/xml_sax_parser.h +0 -39
  212. data/ext/nokogiri/xml_sax_parser_context.h +0 -10
  213. data/ext/nokogiri/xml_sax_push_parser.h +0 -9
  214. data/ext/nokogiri/xml_schema.h +0 -9
  215. data/ext/nokogiri/xml_syntax_error.h +0 -13
  216. data/ext/nokogiri/xml_text.h +0 -9
  217. data/ext/nokogiri/xml_xpath_context.h +0 -10
  218. data/ext/nokogiri/xslt_stylesheet.h +0 -14
  219. data/lib/nokogiri/html/document.rb +0 -335
  220. data/lib/nokogiri/html/document_fragment.rb +0 -49
  221. data/lib/nokogiri/html/element_description_defaults.rb +0 -671
  222. data/lib/nokogiri/html/sax/parser.rb +0 -62
  223. data/lib/nokogiri/html/sax/parser_context.rb +0 -16
  224. data/patches/libxml2/0001-Revert-Do-not-URI-escape-in-server-side-includes.patch +0 -78
  225. data/patches/libxml2/0004-libxml2.la-is-in-top_builddir.patch +0 -25
  226. data/patches/libxml2/0005-Fix-infinite-loop-in-xmlStringLenDecodeEntities.patch +0 -32
  227. data/ports/archives/libxml2-2.9.10.tar.gz +0 -0
  228. data/ports/archives/libxslt-1.1.34.tar.gz +0 -0
  229. /data/patches/libxml2/{0002-Remove-script-macro-support.patch → 0001-Remove-script-macro-support.patch} +0 -0
  230. /data/patches/libxml2/{0003-Update-entities-to-remove-handling-of-ssi.patch → 0002-Update-entities-to-remove-handling-of-ssi.patch} +0 -0
@@ -1,26 +1,41 @@
1
1
  #ifndef NOKOGIRI_NATIVE
2
2
  #define NOKOGIRI_NATIVE
3
3
 
4
+ #include <ruby/defines.h> // https://github.com/sparklemotion/nokogiri/issues/2696
5
+
6
+ #ifdef _MSC_VER
7
+ # ifndef WIN32_LEAN_AND_MEAN
8
+ # define WIN32_LEAN_AND_MEAN
9
+ # endif /* WIN32_LEAN_AND_MEAN */
10
+
11
+ # ifndef WIN32
12
+ # define WIN32
13
+ # endif /* WIN32 */
14
+
15
+ # include <winsock2.h>
16
+ # include <ws2tcpip.h>
17
+ # include <windows.h>
18
+ #endif
19
+
20
+ #ifdef _WIN32
21
+ # define NOKOPUBFUN __declspec(dllexport)
22
+ # define NOKOPUBVAR __declspec(dllexport) extern
23
+ #else
24
+ # define NOKOPUBFUN
25
+ # define NOKOPUBVAR extern
26
+ #endif
27
+
4
28
  #include <stdlib.h>
5
29
  #include <string.h>
6
30
  #include <assert.h>
7
31
  #include <stdarg.h>
32
+ #include <stdio.h>
8
33
 
9
- #ifdef USE_INCLUDED_VASPRINTF
10
- int vasprintf (char **strp, const char *fmt, va_list ap);
11
- #else
12
-
13
- #define _GNU_SOURCE
14
- # include <stdio.h>
15
- #undef _GNU_SOURCE
16
-
17
- #endif
18
34
 
19
35
  #include <libxml/parser.h>
36
+ #include <libxml/tree.h>
20
37
  #include <libxml/entities.h>
21
- #include <libxml/parserInternals.h>
22
38
  #include <libxml/xpath.h>
23
- #include <libxml/xpathInternals.h>
24
39
  #include <libxml/xmlreader.h>
25
40
  #include <libxml/xmlsave.h>
26
41
  #include <libxml/xmlschemas.h>
@@ -28,94 +43,196 @@ int vasprintf (char **strp, const char *fmt, va_list ap);
28
43
  #include <libxml/HTMLtree.h>
29
44
  #include <libxml/relaxng.h>
30
45
  #include <libxml/xinclude.h>
31
- #include <libxslt/extensions.h>
32
46
  #include <libxml/c14n.h>
33
- #include <ruby.h>
34
- #include <ruby/st.h>
35
- #include <ruby/encoding.h>
47
+ #include <libxml/parserInternals.h>
48
+ #include <libxml/xpathInternals.h>
36
49
 
37
- #ifndef NORETURN
38
- # if defined(__GNUC__)
39
- # define NORETURN(name) __attribute__((noreturn)) name
40
- # else
41
- # define NORETURN(name) name
42
- # endif
43
- #endif
50
+ #include <libxslt/extensions.h>
51
+ #include <libxslt/xsltconfig.h>
52
+ #include <libxslt/xsltutils.h>
53
+ #include <libxslt/transform.h>
54
+ #include <libxslt/imports.h>
55
+ #include <libxslt/xsltInternals.h>
44
56
 
45
- #define NOKOGIRI_STR_NEW2(str) \
46
- NOKOGIRI_STR_NEW(str, strlen((const char *)(str)))
47
-
48
- #define NOKOGIRI_STR_NEW(str, len) \
49
- rb_external_str_new_with_enc((const char *)(str), (long)(len), rb_utf8_encoding())
50
-
51
- #define RBSTR_OR_QNIL(_str) \
52
- (_str ? NOKOGIRI_STR_NEW2(_str) : Qnil)
53
-
54
- #include <xml_libxml2_hacks.h>
55
-
56
- #include <xml_io.h>
57
- #include <xml_document.h>
58
- #include <html_entity_lookup.h>
59
- #include <html_document.h>
60
- #include <xml_node.h>
61
- #include <xml_text.h>
62
- #include <xml_cdata.h>
63
- #include <xml_attr.h>
64
- #include <xml_processing_instruction.h>
65
- #include <xml_entity_reference.h>
66
- #include <xml_document_fragment.h>
67
- #include <xml_comment.h>
68
- #include <xml_node_set.h>
69
- #include <xml_dtd.h>
70
- #include <xml_attribute_decl.h>
71
- #include <xml_element_decl.h>
72
- #include <xml_entity_decl.h>
73
- #include <xml_xpath_context.h>
74
- #include <xml_element_content.h>
75
- #include <xml_sax_parser_context.h>
76
- #include <xml_sax_parser.h>
77
- #include <xml_sax_push_parser.h>
78
- #include <xml_reader.h>
79
- #include <html_sax_parser_context.h>
80
- #include <html_sax_push_parser.h>
81
- #include <xslt_stylesheet.h>
82
- #include <xml_syntax_error.h>
83
- #include <xml_schema.h>
84
- #include <xml_relax_ng.h>
85
- #include <html_element_description.h>
86
- #include <xml_namespace.h>
87
- #include <xml_encoding_handler.h>
88
-
89
- extern VALUE mNokogiri ;
90
- extern VALUE mNokogiriXml ;
91
- extern VALUE mNokogiriXmlSax ;
92
- extern VALUE mNokogiriHtml ;
93
- extern VALUE mNokogiriHtmlSax ;
94
- extern VALUE mNokogiriXslt ;
95
-
96
- void nokogiri_root_node(xmlNodePtr);
97
- void nokogiri_root_nsdef(xmlNsPtr, xmlDocPtr);
98
-
99
- #ifdef DEBUG
100
-
101
- #define NOKOGIRI_DEBUG_START(p) if (getenv("NOKOGIRI_NO_FREE")) return ; if (getenv("NOKOGIRI_DEBUG")) fprintf(stderr,"nokogiri: %s:%d %p start\n", __FILE__, __LINE__, p);
102
- #define NOKOGIRI_DEBUG_END(p) if (getenv("NOKOGIRI_DEBUG")) fprintf(stderr,"nokogiri: %s:%d %p end\n", __FILE__, __LINE__, p);
57
+ #include <libexslt/exslt.h>
103
58
 
104
- #else
59
+ /* libxml2_polyfill.c */
60
+ #ifndef HAVE_XMLCTXTSETOPTIONS
61
+ int xmlCtxtSetOptions(xmlParserCtxtPtr ctxt, int options);
62
+ #endif
63
+ #ifndef HAVE_XMLCTXTGETOPTIONS
64
+ int xmlCtxtGetOptions(xmlParserCtxtPtr ctxt);
65
+ #endif
66
+ #ifndef HAVE_XMLSWITCHENCODINGNAME
67
+ int xmlSwitchEncodingName(xmlParserCtxtPtr ctxt, const char *encoding);
68
+ #endif
105
69
 
106
- #define NOKOGIRI_DEBUG_START(p)
107
- #define NOKOGIRI_DEBUG_END(p)
70
+ #define XMLNS_PREFIX "xmlns"
71
+ #define XMLNS_PREFIX_LEN 6 /* including either colon or \0 */
108
72
 
73
+ #ifndef xmlErrorConstPtr
74
+ # if LIBXML_VERSION >= 21200
75
+ # define xmlErrorConstPtr const xmlError *
76
+ # else
77
+ # define xmlErrorConstPtr xmlError *
78
+ # endif
109
79
  #endif
110
80
 
111
- #ifndef __builtin_expect
112
- # if defined(__GNUC__)
113
- # define __builtin_expect(expr, c) __builtin_expect((long)(expr), (long)(c))
114
- # endif
81
+ #include <ruby.h>
82
+ #include <ruby/st.h>
83
+ #include <ruby/encoding.h>
84
+ #include <ruby/util.h>
85
+ #include <ruby/version.h>
86
+
87
+ #define NOKOGIRI_STR_NEW2(str) NOKOGIRI_STR_NEW(str, strlen((const char *)(str)))
88
+ #define NOKOGIRI_STR_NEW(str, len) rb_external_str_new_with_enc((const char *)(str), (long)(len), rb_utf8_encoding())
89
+ #define RBSTR_OR_QNIL(_str) (_str ? NOKOGIRI_STR_NEW2(_str) : Qnil)
90
+
91
+ #ifndef NORETURN_DECL
92
+ # if defined(__GNUC__)
93
+ # define NORETURN_DECL __attribute__ ((noreturn))
94
+ # else
95
+ # define NORETURN_DECL
96
+ # endif
115
97
  #endif
116
98
 
117
- #define XMLNS_PREFIX "xmlns"
118
- #define XMLNS_PREFIX_LEN 6 /* including either colon or \0 */
119
- #define XMLNS_BUFFER_LEN 128
99
+ #ifndef PRINTFLIKE_DECL
100
+ # if defined(__GNUC__)
101
+ # define PRINTFLIKE_DECL(stringidx, argidx) __attribute__ ((format(printf,stringidx,argidx)))
102
+ # else
103
+ # define PRINTFLIKE_DECL(stringidx, argidx)
104
+ # endif
105
+ #endif
120
106
 
107
+ #if defined(TRUFFLERUBY) && !defined(NOKOGIRI_PACKAGED_LIBRARIES)
108
+ # define TRUFFLERUBY_NOKOGIRI_SYSTEM_LIBRARIES
121
109
  #endif
110
+
111
+ NOKOPUBVAR VALUE mNokogiri ;
112
+ NOKOPUBVAR VALUE mNokogiriGumbo ;
113
+ NOKOPUBVAR VALUE mNokogiriHtml4 ;
114
+ NOKOPUBVAR VALUE mNokogiriHtml4Sax ;
115
+ NOKOPUBVAR VALUE mNokogiriHtml5 ;
116
+ NOKOPUBVAR VALUE mNokogiriXml ;
117
+ NOKOPUBVAR VALUE mNokogiriXmlSax ;
118
+ NOKOPUBVAR VALUE mNokogiriXmlXpath ;
119
+ NOKOPUBVAR VALUE mNokogiriXslt ;
120
+
121
+ NOKOPUBVAR VALUE cNokogiriEncodingHandler;
122
+ NOKOPUBVAR VALUE cNokogiriSyntaxError;
123
+ NOKOPUBVAR VALUE cNokogiriXmlAttr;
124
+ NOKOPUBVAR VALUE cNokogiriXmlAttributeDecl;
125
+ NOKOPUBVAR VALUE cNokogiriXmlCData;
126
+ NOKOPUBVAR VALUE cNokogiriXmlCharacterData;
127
+ NOKOPUBVAR VALUE cNokogiriXmlComment;
128
+ NOKOPUBVAR VALUE cNokogiriXmlDocument ;
129
+ NOKOPUBVAR VALUE cNokogiriXmlDocumentFragment;
130
+ NOKOPUBVAR VALUE cNokogiriXmlDtd;
131
+ NOKOPUBVAR VALUE cNokogiriXmlElement ;
132
+ NOKOPUBVAR VALUE cNokogiriXmlElementContent;
133
+ NOKOPUBVAR VALUE cNokogiriXmlElementDecl;
134
+ NOKOPUBVAR VALUE cNokogiriXmlEntityDecl;
135
+ NOKOPUBVAR VALUE cNokogiriXmlEntityReference;
136
+ NOKOPUBVAR VALUE cNokogiriXmlNamespace ;
137
+ NOKOPUBVAR VALUE cNokogiriXmlNode ;
138
+ NOKOPUBVAR VALUE cNokogiriXmlNodeSet ;
139
+ NOKOPUBVAR VALUE cNokogiriXmlProcessingInstruction;
140
+ NOKOPUBVAR VALUE cNokogiriXmlReader;
141
+ NOKOPUBVAR VALUE cNokogiriXmlRelaxNG;
142
+ NOKOPUBVAR VALUE cNokogiriXmlSaxParser ;
143
+ NOKOPUBVAR VALUE cNokogiriXmlSaxParserContext;
144
+ NOKOPUBVAR VALUE cNokogiriXmlSaxPushParser ;
145
+ NOKOPUBVAR VALUE cNokogiriXmlSchema;
146
+ NOKOPUBVAR VALUE cNokogiriXmlSyntaxError;
147
+ NOKOPUBVAR VALUE cNokogiriXmlText ;
148
+ NOKOPUBVAR VALUE cNokogiriXmlXpathContext;
149
+ NOKOPUBVAR VALUE cNokogiriXmlXpathSyntaxError;
150
+ NOKOPUBVAR VALUE cNokogiriXsltStylesheet ;
151
+
152
+ NOKOPUBVAR VALUE cNokogiriHtml4Document ;
153
+ NOKOPUBVAR VALUE cNokogiriHtml4SaxPushParser ;
154
+ NOKOPUBVAR VALUE cNokogiriHtml4ElementDescription ;
155
+ NOKOPUBVAR VALUE cNokogiriHtml4SaxParser;
156
+ NOKOPUBVAR VALUE cNokogiriHtml4SaxParserContext;
157
+ NOKOPUBVAR VALUE cNokogiriHtml5Document ;
158
+
159
+ typedef struct _nokogiriTuple {
160
+ VALUE doc;
161
+ st_table *unlinkedNodes;
162
+ VALUE node_cache;
163
+ } nokogiriTuple;
164
+ typedef nokogiriTuple *nokogiriTuplePtr;
165
+
166
+ typedef struct _libxmlStructuredErrorHandlerState {
167
+ void *user_data;
168
+ xmlStructuredErrorFunc handler;
169
+ } libxmlStructuredErrorHandlerState ;
170
+
171
+ typedef struct _nokogiriXsltStylesheetTuple {
172
+ xsltStylesheetPtr ss;
173
+ VALUE func_instances;
174
+ } nokogiriXsltStylesheetTuple;
175
+
176
+ void noko_xml_document_pin_node(xmlNodePtr);
177
+ void noko_xml_document_pin_namespace(xmlNsPtr, xmlDocPtr);
178
+ int noko_xml_document_has_wrapped_blank_nodes_p(xmlDocPtr c_document);
179
+
180
+ int noko_io_read(void *ctx, char *buffer, int len);
181
+ int noko_io_write(void *ctx, char *buffer, int len);
182
+ int noko_io_close(void *ctx);
183
+
184
+ #define Noko_Node_Get_Struct(obj,type,sval) ((sval) = (type*)DATA_PTR(obj))
185
+ #define Noko_Namespace_Get_Struct(obj,type,sval) ((sval) = (type*)DATA_PTR(obj))
186
+
187
+ VALUE noko_xml_node_wrap(VALUE klass, xmlNodePtr node) ;
188
+ VALUE noko_xml_node_wrap_node_set_result(xmlNodePtr node, VALUE node_set) ;
189
+ VALUE noko_xml_node_attrs(xmlNodePtr node) ;
190
+
191
+ VALUE noko_xml_namespace_wrap(xmlNsPtr node, xmlDocPtr doc);
192
+ VALUE noko_xml_namespace_wrap_xpath_copy(xmlNsPtr node);
193
+
194
+ VALUE noko_xml_element_content_wrap(VALUE doc, xmlElementContentPtr element);
195
+
196
+ VALUE noko_xml_node_set_wrap(xmlNodeSetPtr node_set, VALUE document) ;
197
+ xmlNodeSetPtr noko_xml_node_set_unwrap(VALUE rb_node_set) ;
198
+
199
+ VALUE noko_xml_document_wrap_with_init_args(VALUE klass, xmlDocPtr doc, int argc, VALUE *argv);
200
+ VALUE noko_xml_document_wrap(VALUE klass, xmlDocPtr doc);
201
+ xmlDocPtr noko_xml_document_unwrap(VALUE rb_document);
202
+ NOKOPUBFUN VALUE Nokogiri_wrap_xml_document(VALUE klass,
203
+ xmlDocPtr doc); /* deprecated. use noko_xml_document_wrap() instead. */
204
+
205
+ xmlSAXHandlerPtr noko_xml_sax_parser_unwrap(VALUE rb_sax_handler);
206
+
207
+ xmlParserCtxtPtr noko_xml_sax_push_parser_unwrap(VALUE rb_parser);
208
+
209
+ VALUE noko_xml_sax_parser_context_wrap(VALUE klass, xmlParserCtxtPtr c_context);
210
+ xmlParserCtxtPtr noko_xml_sax_parser_context_unwrap(VALUE rb_context);
211
+ void noko_xml_sax_parser_context_set_encoding(xmlParserCtxtPtr c_context, VALUE rb_encoding);
212
+
213
+ #define DOC_RUBY_OBJECT_TEST(x) ((nokogiriTuplePtr)(x->_private))
214
+ #define DOC_RUBY_OBJECT(x) (((nokogiriTuplePtr)(x->_private))->doc)
215
+ #define DOC_UNLINKED_NODE_HASH(x) (((nokogiriTuplePtr)(x->_private))->unlinkedNodes)
216
+ #define DOC_NODE_CACHE(x) (((nokogiriTuplePtr)(x->_private))->node_cache)
217
+ #define NOKOGIRI_NAMESPACE_EH(node) ((node)->type == XML_NAMESPACE_DECL)
218
+
219
+ #define DISCARD_CONST_QUAL(t, v) ((t)(uintptr_t)(v))
220
+ #define DISCARD_CONST_QUAL_XMLCHAR(v) DISCARD_CONST_QUAL(xmlChar *, v)
221
+
222
+ #if HAVE_RB_CATEGORY_WARNING
223
+ # define NOKO_WARN_DEPRECATION(message...) rb_category_warning(RB_WARN_CATEGORY_DEPRECATED, message)
224
+ #else
225
+ # define NOKO_WARN_DEPRECATION(message...) rb_warning(message)
226
+ #endif
227
+
228
+ void noko__structured_error_func_save(libxmlStructuredErrorHandlerState *handler_state);
229
+ void noko__structured_error_func_save_and_set(libxmlStructuredErrorHandlerState *handler_state, void *user_data,
230
+ xmlStructuredErrorFunc handler);
231
+ void noko__structured_error_func_restore(libxmlStructuredErrorHandlerState *handler_state);
232
+ VALUE noko_xml_syntax_error__wrap(xmlErrorConstPtr error);
233
+ void noko__error_array_pusher(void *ctx, xmlErrorConstPtr error);
234
+ NORETURN_DECL void noko__error_raise(void *ctx, xmlErrorConstPtr error);
235
+ void Nokogiri_marshal_xpath_funcall_and_return_values(xmlXPathParserContextPtr ctx, int nargs, VALUE handler,
236
+ const char *function_name) ;
237
+
238
+ #endif /* NOKOGIRI_NATIVE */
@@ -0,0 +1,40 @@
1
+ #include <nokogiri.h>
2
+
3
+ static VALUE foreign_error_handler_block = Qnil;
4
+
5
+ static void
6
+ foreign_error_handler(void *user_data, xmlErrorConstPtr c_error)
7
+ {
8
+ rb_funcall(foreign_error_handler_block, rb_intern("call"), 0);
9
+ }
10
+
11
+ /*
12
+ * call-seq:
13
+ * __foreign_error_handler { ... } -> nil
14
+ *
15
+ * Override libxml2's global error handlers to call the block. This method thus has very little
16
+ * value except to test that Nokogiri is properly setting error handlers elsewhere in the code. See
17
+ * test/helper.rb for how this is being used.
18
+ */
19
+ static VALUE
20
+ rb_foreign_error_handler(VALUE klass)
21
+ {
22
+ rb_need_block();
23
+ foreign_error_handler_block = rb_block_proc();
24
+ xmlSetStructuredErrorFunc(NULL, foreign_error_handler);
25
+ return Qnil;
26
+ }
27
+
28
+ /*
29
+ * Document-module: Nokogiri::Test
30
+ *
31
+ * The Nokogiri::Test module should only be used for testing Nokogiri.
32
+ * Do NOT use this outside of the Nokogiri test suite.
33
+ */
34
+ void
35
+ noko_init_test_global_handlers(void)
36
+ {
37
+ VALUE mNokogiriTest = rb_define_module_under(mNokogiri, "Test");
38
+
39
+ rb_define_singleton_method(mNokogiriTest, "__foreign_error_handler", rb_foreign_error_handler, 0);
40
+ }
@@ -1,19 +1,22 @@
1
- #include <xml_attr.h>
1
+ #include <nokogiri.h>
2
+
3
+ VALUE cNokogiriXmlAttr;
2
4
 
3
5
  /*
4
6
  * call-seq:
5
7
  * value=(content)
6
8
  *
7
- * Set the value for this Attr to +content+. Use `nil` to remove the value
9
+ * Set the value for this Attr to +content+. Use +nil+ to remove the value
8
10
  * (e.g., a HTML boolean attribute).
9
11
  */
10
- static VALUE set_value(VALUE self, VALUE content)
12
+ static VALUE
13
+ set_value(VALUE self, VALUE content)
11
14
  {
12
15
  xmlAttrPtr attr;
13
16
  xmlChar *value;
14
17
  xmlNode *cur;
15
18
 
16
- Data_Get_Struct(self, xmlAttr, attr);
19
+ Noko_Node_Get_Struct(self, xmlAttr, attr);
17
20
 
18
21
  if (attr->children) {
19
22
  xmlFreeNodeList(attr->children);
@@ -49,7 +52,8 @@ static VALUE set_value(VALUE self, VALUE content)
49
52
  *
50
53
  * Create a new Attr element on the +document+ with +name+
51
54
  */
52
- static VALUE new(int argc, VALUE *argv, VALUE klass)
55
+ static VALUE
56
+ new (int argc, VALUE *argv, VALUE klass)
53
57
  {
54
58
  xmlDocPtr xml_doc;
55
59
  VALUE document;
@@ -64,7 +68,7 @@ static VALUE new(int argc, VALUE *argv, VALUE klass)
64
68
  rb_raise(rb_eArgError, "parameter must be a Nokogiri::XML::Document");
65
69
  }
66
70
 
67
- Data_Get_Struct(document, xmlDoc, xml_doc);
71
+ xml_doc = noko_xml_document_unwrap(document);
68
72
 
69
73
  node = xmlNewDocProp(
70
74
  xml_doc,
@@ -72,9 +76,9 @@ static VALUE new(int argc, VALUE *argv, VALUE klass)
72
76
  NULL
73
77
  );
74
78
 
75
- nokogiri_root_node((xmlNodePtr)node);
79
+ noko_xml_document_pin_node((xmlNodePtr)node);
76
80
 
77
- rb_node = Nokogiri_wrap_xml_node(klass, (xmlNodePtr)node);
81
+ rb_node = noko_xml_node_wrap(klass, (xmlNodePtr)node);
78
82
  rb_obj_call_init(rb_node, argc, argv);
79
83
 
80
84
  if (rb_block_given_p()) {
@@ -84,20 +88,16 @@ static VALUE new(int argc, VALUE *argv, VALUE klass)
84
88
  return rb_node;
85
89
  }
86
90
 
87
- VALUE cNokogiriXmlAttr;
88
- void init_xml_attr()
91
+ void
92
+ noko_init_xml_attr(void)
89
93
  {
90
- VALUE nokogiri = rb_define_module("Nokogiri");
91
- VALUE xml = rb_define_module_under(nokogiri, "XML");
92
- VALUE node = rb_define_class_under(xml, "Node", rb_cObject);
93
-
94
+ assert(cNokogiriXmlNode);
94
95
  /*
95
96
  * Attr represents a Attr node in an xml document.
96
97
  */
97
- VALUE klass = rb_define_class_under(xml, "Attr", node);
98
+ cNokogiriXmlAttr = rb_define_class_under(mNokogiriXml, "Attr", cNokogiriXmlNode);
98
99
 
99
- cNokogiriXmlAttr = klass;
100
+ rb_define_singleton_method(cNokogiriXmlAttr, "new", new, -1);
100
101
 
101
- rb_define_singleton_method(klass, "new", new, -1);
102
- rb_define_method(klass, "value=", set_value, 1);
102
+ rb_define_method(cNokogiriXmlAttr, "value=", set_value, 1);
103
103
  }
@@ -1,4 +1,6 @@
1
- #include <xml_attribute_decl.h>
1
+ #include <nokogiri.h>
2
+
3
+ VALUE cNokogiriXmlAttributeDecl;
2
4
 
3
5
  /*
4
6
  * call-seq:
@@ -6,11 +8,12 @@
6
8
  *
7
9
  * The attribute_type for this AttributeDecl
8
10
  */
9
- static VALUE attribute_type(VALUE self)
11
+ static VALUE
12
+ attribute_type(VALUE self)
10
13
  {
11
14
  xmlAttributePtr node;
12
- Data_Get_Struct(self, xmlAttribute, node);
13
- return INT2NUM((long)node->atype);
15
+ Noko_Node_Get_Struct(self, xmlAttribute, node);
16
+ return INT2NUM(node->atype);
14
17
  }
15
18
 
16
19
  /*
@@ -19,12 +22,13 @@ static VALUE attribute_type(VALUE self)
19
22
  *
20
23
  * The default value
21
24
  */
22
- static VALUE default_value(VALUE self)
25
+ static VALUE
26
+ default_value(VALUE self)
23
27
  {
24
28
  xmlAttributePtr node;
25
- Data_Get_Struct(self, xmlAttribute, node);
29
+ Noko_Node_Get_Struct(self, xmlAttribute, node);
26
30
 
27
- if(node->defaultValue) return NOKOGIRI_STR_NEW2(node->defaultValue);
31
+ if (node->defaultValue) { return NOKOGIRI_STR_NEW2(node->defaultValue); }
28
32
  return Qnil;
29
33
  }
30
34
 
@@ -34,18 +38,19 @@ static VALUE default_value(VALUE self)
34
38
  *
35
39
  * An enumeration of possible values
36
40
  */
37
- static VALUE enumeration(VALUE self)
41
+ static VALUE
42
+ enumeration(VALUE self)
38
43
  {
39
44
  xmlAttributePtr node;
40
45
  xmlEnumerationPtr enm;
41
46
  VALUE list;
42
47
 
43
- Data_Get_Struct(self, xmlAttribute, node);
48
+ Noko_Node_Get_Struct(self, xmlAttribute, node);
44
49
 
45
50
  list = rb_ary_new();
46
51
  enm = node->tree;
47
52
 
48
- while(enm) {
53
+ while (enm) {
49
54
  rb_ary_push(list, NOKOGIRI_STR_NEW2(enm->name));
50
55
  enm = enm->next;
51
56
  }
@@ -53,18 +58,13 @@ static VALUE enumeration(VALUE self)
53
58
  return list;
54
59
  }
55
60
 
56
- VALUE cNokogiriXmlAttributeDecl;
57
-
58
- void init_xml_attribute_decl()
61
+ void
62
+ noko_init_xml_attribute_decl(void)
59
63
  {
60
- VALUE nokogiri = rb_define_module("Nokogiri");
61
- VALUE xml = rb_define_module_under(nokogiri, "XML");
62
- VALUE node = rb_define_class_under(xml, "Node", rb_cObject);
63
- VALUE klass = rb_define_class_under(xml, "AttributeDecl", node);
64
-
65
- cNokogiriXmlAttributeDecl = klass;
64
+ assert(cNokogiriXmlNode);
65
+ cNokogiriXmlAttributeDecl = rb_define_class_under(mNokogiriXml, "AttributeDecl", cNokogiriXmlNode);
66
66
 
67
- rb_define_method(klass, "attribute_type", attribute_type, 0);
68
- rb_define_method(klass, "default", default_value, 0);
69
- rb_define_method(klass, "enumeration", enumeration, 0);
67
+ rb_define_method(cNokogiriXmlAttributeDecl, "attribute_type", attribute_type, 0);
68
+ rb_define_method(cNokogiriXmlAttributeDecl, "default", default_value, 0);
69
+ rb_define_method(cNokogiriXmlAttributeDecl, "enumeration", enumeration, 0);
70
70
  }
@@ -1,4 +1,6 @@
1
- #include <xml_cdata.h>
1
+ #include <nokogiri.h>
2
+
3
+ VALUE cNokogiriXmlCData;
2
4
 
3
5
  /*
4
6
  * call-seq:
@@ -9,54 +11,52 @@
9
11
  * If +content+ cannot be implicitly converted to a string, this method will
10
12
  * raise a TypeError exception.
11
13
  */
12
- static VALUE new(int argc, VALUE *argv, VALUE klass)
14
+ static VALUE
15
+ rb_xml_cdata_s_new(int argc, VALUE *argv, VALUE klass)
13
16
  {
14
- xmlDocPtr xml_doc;
15
- xmlNodePtr node;
16
- VALUE doc;
17
- VALUE content;
18
- VALUE rest;
17
+ xmlDocPtr c_document;
18
+ xmlNodePtr c_node;
19
+ VALUE rb_document;
20
+ VALUE rb_content;
21
+ VALUE rb_rest;
19
22
  VALUE rb_node;
20
- xmlChar *content_str = NULL;
21
- int content_str_len = 0;
22
-
23
- rb_scan_args(argc, argv, "2*", &doc, &content, &rest);
24
23
 
25
- Data_Get_Struct(doc, xmlDoc, xml_doc);
24
+ rb_scan_args(argc, argv, "2*", &rb_document, &rb_content, &rb_rest);
26
25
 
27
- if (!NIL_P(content)) {
28
- content_str = (xmlChar *)StringValuePtr(content);
29
- content_str_len = RSTRING_LEN(content);
26
+ Check_Type(rb_content, T_STRING);
27
+ if (!rb_obj_is_kind_of(rb_document, cNokogiriXmlNode)) {
28
+ rb_raise(rb_eTypeError,
29
+ "expected first parameter to be a Nokogiri::XML::Document, received %"PRIsVALUE,
30
+ rb_obj_class(rb_document));
30
31
  }
31
32
 
32
- node = xmlNewCDataBlock(xml_doc->doc, content_str, content_str_len);
33
-
34
- nokogiri_root_node(node);
33
+ if (!rb_obj_is_kind_of(rb_document, cNokogiriXmlDocument)) {
34
+ xmlNodePtr deprecated_node_type_arg;
35
+ NOKO_WARN_DEPRECATION("Passing a Node as the first parameter to CDATA.new is deprecated. Please pass a Document instead. This will become an error in Nokogiri v1.17.0."); // TODO: deprecated in v1.15.3, remove in v1.17.0
36
+ Noko_Node_Get_Struct(rb_document, xmlNode, deprecated_node_type_arg);
37
+ c_document = deprecated_node_type_arg->doc;
38
+ } else {
39
+ c_document = noko_xml_document_unwrap(rb_document);
40
+ }
35
41
 
36
- rb_node = Nokogiri_wrap_xml_node(klass, node);
42
+ c_node = xmlNewCDataBlock(c_document, (xmlChar *)StringValueCStr(rb_content), RSTRING_LENINT(rb_content));
43
+ noko_xml_document_pin_node(c_node);
44
+ rb_node = noko_xml_node_wrap(klass, c_node);
37
45
  rb_obj_call_init(rb_node, argc, argv);
38
46
 
39
- if(rb_block_given_p()) { rb_yield(rb_node); }
47
+ if (rb_block_given_p()) { rb_yield(rb_node); }
40
48
 
41
49
  return rb_node;
42
50
  }
43
51
 
44
- VALUE cNokogiriXmlCData;
45
- void init_xml_cdata()
52
+ void
53
+ noko_init_xml_cdata(void)
46
54
  {
47
- VALUE nokogiri = rb_define_module("Nokogiri");
48
- VALUE xml = rb_define_module_under(nokogiri, "XML");
49
- VALUE node = rb_define_class_under(xml, "Node", rb_cObject);
50
- VALUE char_data = rb_define_class_under(xml, "CharacterData", node);
51
- VALUE text = rb_define_class_under(xml, "Text", char_data);
52
-
55
+ assert(cNokogiriXmlText);
53
56
  /*
54
57
  * CData represents a CData node in an xml document.
55
58
  */
56
- VALUE klass = rb_define_class_under(xml, "CDATA", text);
57
-
58
-
59
- cNokogiriXmlCData = klass;
59
+ cNokogiriXmlCData = rb_define_class_under(mNokogiriXml, "CDATA", cNokogiriXmlText);
60
60
 
61
- rb_define_singleton_method(klass, "new", new, -1);
61
+ rb_define_singleton_method(cNokogiriXmlCData, "new", rb_xml_cdata_s_new, -1);
62
62
  }