libxml-ruby 2.8.0 → 3.2.3

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.
Files changed (205) hide show
  1. checksums.yaml +5 -5
  2. data/HISTORY +859 -775
  3. data/LICENSE +20 -20
  4. data/MANIFEST +166 -166
  5. data/README.rdoc +217 -184
  6. data/Rakefile +98 -78
  7. data/ext/libxml/extconf.rb +61 -116
  8. data/ext/libxml/libxml.c +80 -76
  9. data/ext/libxml/ruby_libxml.h +67 -75
  10. data/ext/libxml/ruby_xml.c +937 -893
  11. data/ext/libxml/ruby_xml.h +10 -10
  12. data/ext/libxml/ruby_xml_attr.c +333 -333
  13. data/ext/libxml/ruby_xml_attr.h +12 -12
  14. data/ext/libxml/ruby_xml_attr_decl.c +153 -153
  15. data/ext/libxml/ruby_xml_attr_decl.h +11 -11
  16. data/ext/libxml/ruby_xml_attributes.c +275 -275
  17. data/ext/libxml/ruby_xml_attributes.h +15 -15
  18. data/ext/libxml/ruby_xml_cbg.c +85 -85
  19. data/ext/libxml/ruby_xml_document.c +1123 -1147
  20. data/ext/libxml/ruby_xml_document.h +11 -11
  21. data/ext/libxml/ruby_xml_dtd.c +248 -268
  22. data/ext/libxml/ruby_xml_dtd.h +9 -9
  23. data/ext/libxml/ruby_xml_encoding.c +250 -260
  24. data/ext/libxml/ruby_xml_encoding.h +16 -19
  25. data/ext/libxml/ruby_xml_error.c +996 -996
  26. data/ext/libxml/ruby_xml_error.h +12 -12
  27. data/ext/libxml/ruby_xml_html_parser.c +89 -92
  28. data/ext/libxml/ruby_xml_html_parser.h +10 -10
  29. data/ext/libxml/ruby_xml_html_parser_context.c +337 -338
  30. data/ext/libxml/ruby_xml_html_parser_context.h +10 -10
  31. data/ext/libxml/ruby_xml_html_parser_options.c +46 -46
  32. data/ext/libxml/ruby_xml_html_parser_options.h +10 -10
  33. data/ext/libxml/ruby_xml_input_cbg.c +191 -191
  34. data/ext/libxml/ruby_xml_input_cbg.h +20 -20
  35. data/ext/libxml/ruby_xml_io.c +47 -50
  36. data/ext/libxml/ruby_xml_io.h +10 -10
  37. data/ext/libxml/ruby_xml_namespace.c +154 -153
  38. data/ext/libxml/ruby_xml_namespace.h +10 -10
  39. data/ext/libxml/ruby_xml_namespaces.c +293 -293
  40. data/ext/libxml/ruby_xml_namespaces.h +9 -9
  41. data/ext/libxml/ruby_xml_node.c +1406 -1452
  42. data/ext/libxml/ruby_xml_node.h +13 -11
  43. data/ext/libxml/ruby_xml_parser.c +91 -94
  44. data/ext/libxml/ruby_xml_parser.h +12 -12
  45. data/ext/libxml/ruby_xml_parser_context.c +999 -1001
  46. data/ext/libxml/ruby_xml_parser_context.h +10 -10
  47. data/ext/libxml/ruby_xml_parser_options.c +66 -66
  48. data/ext/libxml/ruby_xml_parser_options.h +12 -12
  49. data/ext/libxml/ruby_xml_reader.c +1239 -1228
  50. data/ext/libxml/ruby_xml_reader.h +17 -17
  51. data/ext/libxml/ruby_xml_relaxng.c +110 -111
  52. data/ext/libxml/ruby_xml_relaxng.h +10 -10
  53. data/ext/libxml/ruby_xml_sax2_handler.c +326 -328
  54. data/ext/libxml/ruby_xml_sax2_handler.h +10 -10
  55. data/ext/libxml/ruby_xml_sax_parser.c +116 -120
  56. data/ext/libxml/ruby_xml_sax_parser.h +10 -10
  57. data/ext/libxml/ruby_xml_schema.c +350 -301
  58. data/ext/libxml/ruby_xml_schema.h +806 -809
  59. data/ext/libxml/ruby_xml_schema_attribute.c +61 -109
  60. data/ext/libxml/ruby_xml_schema_attribute.h +15 -15
  61. data/ext/libxml/ruby_xml_schema_element.c +69 -94
  62. data/ext/libxml/ruby_xml_schema_element.h +14 -14
  63. data/ext/libxml/ruby_xml_schema_facet.c +46 -52
  64. data/ext/libxml/ruby_xml_schema_facet.h +13 -13
  65. data/ext/libxml/ruby_xml_schema_type.c +214 -259
  66. data/ext/libxml/ruby_xml_schema_type.h +9 -9
  67. data/ext/libxml/ruby_xml_version.h +9 -9
  68. data/ext/libxml/ruby_xml_writer.c +1133 -1137
  69. data/ext/libxml/ruby_xml_writer.h +10 -10
  70. data/ext/libxml/ruby_xml_xinclude.c +16 -16
  71. data/ext/libxml/ruby_xml_xinclude.h +11 -11
  72. data/ext/libxml/ruby_xml_xpath.c +194 -188
  73. data/ext/libxml/ruby_xml_xpath.h +13 -13
  74. data/ext/libxml/ruby_xml_xpath_context.c +360 -361
  75. data/ext/libxml/ruby_xml_xpath_context.h +9 -9
  76. data/ext/libxml/ruby_xml_xpath_expression.c +81 -81
  77. data/ext/libxml/ruby_xml_xpath_expression.h +10 -10
  78. data/ext/libxml/ruby_xml_xpath_object.c +338 -335
  79. data/ext/libxml/ruby_xml_xpath_object.h +17 -17
  80. data/ext/libxml/ruby_xml_xpointer.c +99 -99
  81. data/ext/libxml/ruby_xml_xpointer.h +11 -11
  82. data/ext/vc/libxml_ruby.sln +17 -15
  83. data/lib/libxml/node.rb +2 -78
  84. data/lib/libxml/parser.rb +0 -266
  85. data/lib/libxml/sax_parser.rb +0 -17
  86. data/lib/libxml/schema/attribute.rb +19 -19
  87. data/lib/libxml/schema/element.rb +19 -27
  88. data/lib/libxml/schema/type.rb +21 -29
  89. data/lib/libxml/schema.rb +47 -66
  90. data/lib/libxml-ruby.rb +30 -0
  91. data/lib/libxml.rb +3 -33
  92. data/libxml-ruby.gemspec +48 -44
  93. data/script/benchmark/depixelate +634 -634
  94. data/script/benchmark/hamlet.xml +9054 -9054
  95. data/script/benchmark/parsecount +170 -170
  96. data/script/benchmark/throughput +41 -41
  97. data/script/test +6 -6
  98. data/setup.rb +0 -1
  99. data/test/c14n/given/example-1.xml +14 -14
  100. data/test/c14n/given/example-2.xml +11 -11
  101. data/test/c14n/given/example-3.xml +18 -18
  102. data/test/c14n/given/example-4.xml +9 -9
  103. data/test/c14n/given/example-5.xml +12 -12
  104. data/test/c14n/given/example-6.xml +2 -2
  105. data/test/c14n/given/example-7.xml +11 -11
  106. data/test/c14n/given/example-8.xml +11 -11
  107. data/test/c14n/given/example-8.xpath +9 -9
  108. data/test/c14n/result/1-1-without-comments/example-1 +3 -3
  109. data/test/c14n/result/1-1-without-comments/example-2 +10 -10
  110. data/test/c14n/result/1-1-without-comments/example-3 +13 -13
  111. data/test/c14n/result/1-1-without-comments/example-4 +8 -8
  112. data/test/c14n/result/1-1-without-comments/example-5 +2 -2
  113. data/test/c14n/result/with-comments/example-1 +5 -5
  114. data/test/c14n/result/with-comments/example-2 +10 -10
  115. data/test/c14n/result/with-comments/example-3 +13 -13
  116. data/test/c14n/result/with-comments/example-4 +8 -8
  117. data/test/c14n/result/with-comments/example-5 +3 -3
  118. data/test/c14n/result/without-comments/example-1 +3 -3
  119. data/test/c14n/result/without-comments/example-2 +10 -10
  120. data/test/c14n/result/without-comments/example-3 +13 -13
  121. data/test/c14n/result/without-comments/example-4 +8 -8
  122. data/test/c14n/result/without-comments/example-5 +2 -2
  123. data/test/model/atom.xml +12 -12
  124. data/test/model/bands.iso-8859-1.xml +4 -4
  125. data/test/model/bands.utf-8.xml +4 -4
  126. data/test/model/bands.xml +4 -4
  127. data/test/model/books.xml +153 -153
  128. data/test/model/cwm_1_0.xml +11336 -0
  129. data/test/model/merge_bug_data.xml +58 -58
  130. data/test/model/ruby-lang.html +238 -238
  131. data/test/model/rubynet.xml +79 -79
  132. data/test/model/shiporder.rnc +28 -28
  133. data/test/model/shiporder.rng +86 -86
  134. data/test/model/shiporder.xml +22 -22
  135. data/test/model/shiporder.xsd +44 -40
  136. data/test/model/shiporder_bad.xsd +40 -0
  137. data/test/model/shiporder_import.xsd +45 -0
  138. data/test/model/soap.xml +27 -27
  139. data/test/model/xinclude.xml +4 -4
  140. data/test/{tc_attr.rb → test_attr.rb} +23 -25
  141. data/test/{tc_attr_decl.rb → test_attr_decl.rb} +13 -14
  142. data/test/{tc_attributes.rb → test_attributes.rb} +11 -18
  143. data/test/{tc_canonicalize.rb → test_canonicalize.rb} +36 -41
  144. data/test/test_deprecated_require.rb +12 -0
  145. data/test/{tc_document.rb → test_document.rb} +33 -27
  146. data/test/test_document_write.rb +146 -0
  147. data/test/{tc_dtd.rb → test_dtd.rb} +29 -29
  148. data/test/{tc_encoding.rb → test_encoding.rb} +129 -126
  149. data/test/{tc_encoding_sax.rb → test_encoding_sax.rb} +7 -6
  150. data/test/test_error.rb +178 -0
  151. data/test/test_helper.rb +4 -9
  152. data/test/test_html_parser.rb +162 -0
  153. data/test/test_html_parser_context.rb +23 -0
  154. data/test/test_namespace.rb +60 -0
  155. data/test/{tc_namespaces.rb → test_namespaces.rb} +34 -44
  156. data/test/{tc_node.rb → test_node.rb} +68 -47
  157. data/test/{tc_node_cdata.rb → test_node_cdata.rb} +12 -13
  158. data/test/{tc_node_comment.rb → test_node_comment.rb} +7 -8
  159. data/test/{tc_node_copy.rb → test_node_copy.rb} +4 -6
  160. data/test/{tc_node_edit.rb → test_node_edit.rb} +23 -41
  161. data/test/{tc_node_pi.rb → test_node_pi.rb} +37 -40
  162. data/test/{tc_node_text.rb → test_node_text.rb} +10 -12
  163. data/test/{tc_node_write.rb → test_node_write.rb} +18 -29
  164. data/test/test_node_xlink.rb +28 -0
  165. data/test/test_parser.rb +324 -0
  166. data/test/{tc_parser_context.rb → test_parser_context.rb} +42 -49
  167. data/test/{tc_properties.rb → test_properties.rb} +6 -7
  168. data/test/test_reader.rb +364 -0
  169. data/test/test_relaxng.rb +53 -0
  170. data/test/{tc_sax_parser.rb → test_sax_parser.rb} +44 -38
  171. data/test/test_schema.rb +231 -0
  172. data/test/test_suite.rb +38 -40
  173. data/test/{tc_traversal.rb → test_traversal.rb} +5 -6
  174. data/test/{tc_writer.rb → test_writer.rb} +468 -448
  175. data/test/{tc_xinclude.rb → test_xinclude.rb} +4 -5
  176. data/test/test_xml.rb +263 -0
  177. data/test/{tc_xpath.rb → test_xpath.rb} +31 -32
  178. data/test/{tc_xpath_context.rb → test_xpath_context.rb} +8 -9
  179. data/test/test_xpath_expression.rb +37 -0
  180. data/test/{tc_xpointer.rb → test_xpointer.rb} +16 -18
  181. metadata +122 -100
  182. data/lib/libxml/ns.rb +0 -22
  183. data/lib/libxml/properties.rb +0 -23
  184. data/lib/libxml/reader.rb +0 -29
  185. data/lib/libxml/xpath_object.rb +0 -16
  186. data/test/etc_doc_to_s.rb +0 -21
  187. data/test/ets_doc_file.rb +0 -17
  188. data/test/ets_doc_to_s.rb +0 -23
  189. data/test/ets_gpx.rb +0 -28
  190. data/test/ets_node_gc.rb +0 -23
  191. data/test/ets_test.xml +0 -2
  192. data/test/ets_tsr.rb +0 -11
  193. data/test/tc_deprecated_require.rb +0 -13
  194. data/test/tc_document_write.rb +0 -196
  195. data/test/tc_error.rb +0 -180
  196. data/test/tc_html_parser.rb +0 -153
  197. data/test/tc_html_parser_context.rb +0 -24
  198. data/test/tc_namespace.rb +0 -62
  199. data/test/tc_node_xlink.rb +0 -29
  200. data/test/tc_parser.rb +0 -381
  201. data/test/tc_reader.rb +0 -400
  202. data/test/tc_relaxng.rb +0 -54
  203. data/test/tc_schema.rb +0 -162
  204. data/test/tc_xml.rb +0 -226
  205. data/test/tc_xpath_expression.rb +0 -38
@@ -1,260 +1,250 @@
1
- /* Please see the LICENSE file for copyright and distribution information */
2
-
3
- #include <stdarg.h>
4
- #include "ruby_libxml.h"
5
-
6
- /*
7
- * Document-class: LibXML::XML::Encoding
8
- *
9
- * The encoding class exposes the encodings that libxml
10
- * supports via constants.
11
- *
12
- * LibXML converts all data sources to UTF8
13
- * internally before processing them. By default,
14
- * LibXML determines a data source's encoding
15
- * using the algorithm described on its
16
- * website[http://xmlsoft.org/encoding.html].
17
- *
18
- * However, you may override a data source's encoding
19
- * by using the encoding constants defined in this
20
- * module.
21
- *
22
- * Example 1:
23
- *
24
- * io = File.open('some_file', 'rb')
25
- * parser = XML::Parser.io(io, :encoding => XML::Encoding::ISO_8859_1)
26
- * doc = parser.parse
27
- *
28
- * Example 2:
29
- *
30
- * parser = XML::HTMLParser.file("some_file", :encoding => XML::Encoding::ISO_8859_1)
31
- * doc = parser.parse
32
- *
33
- * Example 3:
34
- *
35
- * document = XML::Document.new
36
- * document.encoding = XML::Encoding::ISO_8859_1
37
- * doc << XML::Node.new
38
- */
39
-
40
- VALUE mXMLEncoding;
41
-
42
- /*
43
- * call-seq:
44
- * Encoding.from_s("UTF_8") -> XML::Encoding::UTF_8
45
- *
46
- * Converts an encoding string to an encoding constant
47
- * defined on the XML::Encoding class.
48
- */
49
- static VALUE rxml_encoding_from_s(VALUE klass, VALUE encoding)
50
- {
51
- xmlCharEncoding xencoding;
52
-
53
- if (encoding == Qnil)
54
- return Qnil;
55
-
56
- xencoding = xmlParseCharEncoding(StringValuePtr(encoding));
57
- return INT2NUM(xencoding);
58
- }
59
-
60
- /*
61
- * call-seq:
62
- * Encoding.to_s(XML::Encoding::UTF_8) -> "UTF-8"
63
- *
64
- * Converts an encoding constant defined on the XML::Encoding
65
- * class to its text representation.
66
- */
67
- static VALUE rxml_encoding_to_s(VALUE klass, VALUE encoding)
68
- {
69
- const char* xencoding = xmlGetCharEncodingName(NUM2INT(encoding));
70
-
71
- if (!xencoding)
72
- return Qnil;
73
- else
74
- return rxml_new_cstr(xencoding, xencoding);
75
- }
76
-
77
- #ifdef HAVE_RUBY_ENCODING_H
78
- /*
79
- * Converts an xmlCharEncoding enum value into a Ruby Encoding object (available
80
- * on Ruby 1.9.* and higher).
81
- */
82
- rb_encoding* rxml_xml_encoding_to_rb_encoding(VALUE klass, xmlCharEncoding xmlEncoding)
83
- {
84
- const char* encodingName;
85
-
86
- switch (xmlEncoding)
87
- {
88
- case XML_CHAR_ENCODING_UTF8:
89
- encodingName = "UTF-8";
90
- break;
91
- case XML_CHAR_ENCODING_UTF16LE:
92
- encodingName = "UTF-16LE";
93
- break;
94
- case XML_CHAR_ENCODING_UTF16BE:
95
- encodingName = "UTF-16BE";
96
- break;
97
- case XML_CHAR_ENCODING_UCS4LE:
98
- encodingName = "UCS-4LE";
99
- break;
100
- case XML_CHAR_ENCODING_UCS4BE:
101
- encodingName = "UCS-4BE";
102
- break;
103
- case XML_CHAR_ENCODING_UCS2:
104
- encodingName = "UCS-2";
105
- break;
106
- case XML_CHAR_ENCODING_8859_1:
107
- encodingName = "ISO8859-1";
108
- break;
109
- case XML_CHAR_ENCODING_8859_2:
110
- encodingName = "ISO8859-2";
111
- break;
112
- case XML_CHAR_ENCODING_8859_3:
113
- encodingName = "ISO8859-3";
114
- break;
115
- case XML_CHAR_ENCODING_8859_4:
116
- encodingName = "ISO8859-4";
117
- break;
118
- case XML_CHAR_ENCODING_8859_5:
119
- encodingName = "ISO8859-5";
120
- break;
121
- case XML_CHAR_ENCODING_8859_6:
122
- encodingName = "ISO8859-6";
123
- break;
124
- case XML_CHAR_ENCODING_8859_7:
125
- encodingName = "ISO8859-7";
126
- break;
127
- case XML_CHAR_ENCODING_8859_8:
128
- encodingName = "ISO8859-8";
129
- break;
130
- case XML_CHAR_ENCODING_8859_9:
131
- encodingName = "ISO8859-9";
132
- break;
133
- case XML_CHAR_ENCODING_2022_JP:
134
- encodingName = "ISO-2022-JP";
135
- break;
136
- case XML_CHAR_ENCODING_SHIFT_JIS:
137
- encodingName = "SHIFT-JIS";
138
- break;
139
- case XML_CHAR_ENCODING_EUC_JP:
140
- encodingName = "EUC-JP";
141
- break;
142
- case XML_CHAR_ENCODING_ASCII:
143
- encodingName = "US-ASCII";
144
- break;
145
- default:
146
- /* Covers XML_CHAR_ENCODING_ERROR, XML_CHAR_ENCODING_NONE, XML_CHAR_ENCODING_EBCDIC */
147
- encodingName = "ASCII-8BIT";
148
- break;
149
- }
150
-
151
- return rb_enc_find(encodingName);
152
- }
153
-
154
- /*
155
- * call-seq:
156
- * Input.encoding_to_rb_encoding(Input::ENCODING) -> Encoding
157
- *
158
- * Converts an encoding constant defined on the XML::Encoding
159
- * class to a Ruby encoding object (available on Ruby 1.9.* and higher).
160
- */
161
- VALUE rxml_encoding_to_rb_encoding(VALUE klass, VALUE encoding)
162
- {
163
- xmlCharEncoding xmlEncoding = (xmlCharEncoding)NUM2INT(encoding);
164
- rb_encoding* rbencoding = rxml_xml_encoding_to_rb_encoding(klass, xmlEncoding);
165
- return rb_enc_from_encoding(rbencoding);
166
- }
167
-
168
- rb_encoding* rxml_figure_encoding(const char* xencoding)
169
- {
170
- rb_encoding* result;
171
- if (xencoding)
172
- {
173
- xmlCharEncoding xmlEncoding = xmlParseCharEncoding(xencoding);
174
- result = rxml_xml_encoding_to_rb_encoding(mXMLEncoding, xmlEncoding);
175
- }
176
- else
177
- {
178
- result = rb_utf8_encoding();
179
- }
180
- return result;
181
- }
182
- #endif
183
-
184
- VALUE rxml_new_cstr(const char* xstr, const char* xencoding)
185
- {
186
- #ifdef HAVE_RUBY_ENCODING_H
187
- rb_encoding *rbencoding = rxml_figure_encoding(xencoding);
188
- return rb_external_str_new_with_enc(xstr, strlen(xstr), rbencoding);
189
- #endif
190
- return rb_str_new2(xstr);
191
- }
192
-
193
- VALUE rxml_new_cstr_len(const char* xstr, const int length, const char* xencoding)
194
- {
195
- #ifdef HAVE_RUBY_ENCODING_H
196
- rb_encoding *rbencoding = rxml_figure_encoding(xencoding);
197
- return rb_external_str_new_with_enc(xstr, length, rbencoding);
198
- #endif
199
- return rb_str_new(xstr, length);
200
- }
201
-
202
- void rxml_init_encoding(void)
203
- {
204
- mXMLEncoding = rb_define_module_under(mXML, "Encoding");
205
- rb_define_module_function(mXMLEncoding, "from_s", rxml_encoding_from_s, 1);
206
- rb_define_module_function(mXMLEncoding, "to_s", rxml_encoding_to_s, 1);
207
-
208
- #ifdef HAVE_RUBY_ENCODING_H
209
- // rb_define_module_function(mXMLEncoding, "to_rb_encoding", rxml_encoding_to_rb_encoding, 2);
210
- #endif
211
-
212
- /* -1: No char encoding detected. */
213
- rb_define_const(mXMLEncoding, "ERROR", INT2NUM(XML_CHAR_ENCODING_ERROR));
214
- /* 0: No char encoding detected. */
215
- rb_define_const(mXMLEncoding, "NONE", INT2NUM(XML_CHAR_ENCODING_NONE));
216
- /* 1: UTF-8 */
217
- rb_define_const(mXMLEncoding, "UTF_8", INT2NUM(XML_CHAR_ENCODING_UTF8));
218
- /* 2: UTF-16 little endian. */
219
- rb_define_const(mXMLEncoding, "UTF_16LE", INT2NUM(XML_CHAR_ENCODING_UTF16LE));
220
- /* 3: UTF-16 big endian. */
221
- rb_define_const(mXMLEncoding, "UTF_16BE", INT2NUM(XML_CHAR_ENCODING_UTF16BE));
222
- /* 4: UCS-4 little endian. */
223
- rb_define_const(mXMLEncoding, "UCS_4LE", INT2NUM(XML_CHAR_ENCODING_UCS4LE));
224
- /* 5: UCS-4 big endian. */
225
- rb_define_const(mXMLEncoding, "UCS_4BE", INT2NUM(XML_CHAR_ENCODING_UCS4BE));
226
- /* 6: EBCDIC uh! */
227
- rb_define_const(mXMLEncoding, "EBCDIC", INT2NUM(XML_CHAR_ENCODING_EBCDIC));
228
- /* 7: UCS-4 unusual ordering. */
229
- rb_define_const(mXMLEncoding, "UCS_4_2143", INT2NUM(XML_CHAR_ENCODING_UCS4_2143));
230
- /* 8: UCS-4 unusual ordering. */
231
- rb_define_const(mXMLEncoding, "UCS_4_3412", INT2NUM(XML_CHAR_ENCODING_UCS4_3412));
232
- /* 9: UCS-2. */
233
- rb_define_const(mXMLEncoding, "UCS_2", INT2NUM(XML_CHAR_ENCODING_UCS2));
234
- /* 10: ISO-8859-1 ISO Latin 1. */
235
- rb_define_const(mXMLEncoding, "ISO_8859_1", INT2NUM(XML_CHAR_ENCODING_8859_1));
236
- /* 11: ISO-8859-2 ISO Latin 2. */
237
- rb_define_const(mXMLEncoding, "ISO_8859_2", INT2NUM(XML_CHAR_ENCODING_8859_2));
238
- /* 12: ISO-8859-3. */
239
- rb_define_const(mXMLEncoding, "ISO_8859_3", INT2NUM(XML_CHAR_ENCODING_8859_3));
240
- /* 13: ISO-8859-4. */
241
- rb_define_const(mXMLEncoding, "ISO_8859_4", INT2NUM(XML_CHAR_ENCODING_8859_4));
242
- /* 14: ISO-8859-5. */
243
- rb_define_const(mXMLEncoding, "ISO_8859_5", INT2NUM(XML_CHAR_ENCODING_8859_5));
244
- /* 15: ISO-8859-6. */
245
- rb_define_const(mXMLEncoding, "ISO_8859_6", INT2NUM(XML_CHAR_ENCODING_8859_6));
246
- /* 16: ISO-8859-7. */
247
- rb_define_const(mXMLEncoding, "ISO_8859_7", INT2NUM(XML_CHAR_ENCODING_8859_7));
248
- /* 17: ISO-8859-8. */
249
- rb_define_const(mXMLEncoding, "ISO_8859_8", INT2NUM(XML_CHAR_ENCODING_8859_8));
250
- /* 18: ISO-8859-9. */
251
- rb_define_const(mXMLEncoding, "ISO_8859_9", INT2NUM(XML_CHAR_ENCODING_8859_9));
252
- /* 19: ISO-2022-JP. */
253
- rb_define_const(mXMLEncoding, "ISO_2022_JP", INT2NUM(XML_CHAR_ENCODING_2022_JP));
254
- /* 20: Shift_JIS. */
255
- rb_define_const(mXMLEncoding, "SHIFT_JIS", INT2NUM(XML_CHAR_ENCODING_SHIFT_JIS));
256
- /* 21: EUC-JP. */
257
- rb_define_const(mXMLEncoding, "EUC_JP", INT2NUM(XML_CHAR_ENCODING_EUC_JP));
258
- /* 22: pure ASCII. */
259
- rb_define_const(mXMLEncoding, "ASCII", INT2NUM(XML_CHAR_ENCODING_ASCII));
260
- }
1
+ /* Please see the LICENSE file for copyright and distribution information */
2
+
3
+ #include <stdarg.h>
4
+ #include "ruby_libxml.h"
5
+
6
+ /*
7
+ * Document-class: LibXML::XML::Encoding
8
+ *
9
+ * The encoding class exposes the encodings that libxml
10
+ * supports via constants.
11
+ *
12
+ * LibXML converts all data sources to UTF8
13
+ * internally before processing them. By default,
14
+ * LibXML determines a data source's encoding
15
+ * using the algorithm described on its
16
+ * website[http://xmlsoft.org/encoding.html].
17
+ *
18
+ * However, you may override a data source's encoding
19
+ * by using the encoding constants defined in this
20
+ * module.
21
+ *
22
+ * Example 1:
23
+ *
24
+ * io = File.open('some_file', 'rb')
25
+ * parser = XML::Parser.io(io, :encoding => XML::Encoding::ISO_8859_1)
26
+ * doc = parser.parse
27
+ *
28
+ * Example 2:
29
+ *
30
+ * parser = XML::HTMLParser.file("some_file", :encoding => XML::Encoding::ISO_8859_1)
31
+ * doc = parser.parse
32
+ *
33
+ * Example 3:
34
+ *
35
+ * document = XML::Document.new
36
+ * document.encoding = XML::Encoding::ISO_8859_1
37
+ * doc << XML::Node.new
38
+ */
39
+
40
+ VALUE mXMLEncoding;
41
+
42
+ /*
43
+ * call-seq:
44
+ * Encoding.from_s("UTF_8") -> XML::Encoding::UTF_8
45
+ *
46
+ * Converts an encoding string to an encoding constant
47
+ * defined on the XML::Encoding class.
48
+ */
49
+ static VALUE rxml_encoding_from_s(VALUE klass, VALUE encoding)
50
+ {
51
+ xmlCharEncoding xencoding;
52
+
53
+ if (encoding == Qnil)
54
+ return Qnil;
55
+
56
+ xencoding = xmlParseCharEncoding(StringValuePtr(encoding));
57
+ return INT2NUM(xencoding);
58
+ }
59
+
60
+ /*
61
+ * call-seq:
62
+ * Encoding.to_s(XML::Encoding::UTF_8) -> "UTF-8"
63
+ *
64
+ * Converts an encoding constant defined on the XML::Encoding
65
+ * class to its text representation.
66
+ */
67
+ static VALUE rxml_encoding_to_s(VALUE klass, VALUE encoding)
68
+ {
69
+ const xmlChar* xencoding = (const xmlChar*)xmlGetCharEncodingName(NUM2INT(encoding));
70
+
71
+ if (!xencoding)
72
+ return Qnil;
73
+ else
74
+ return rxml_new_cstr(xencoding, xencoding);
75
+ }
76
+
77
+ /*
78
+ * Converts an xmlCharEncoding enum value into a Ruby Encoding object (available
79
+ * on Ruby 1.9.* and higher).
80
+ */
81
+ rb_encoding* rxml_xml_encoding_to_rb_encoding(VALUE klass, xmlCharEncoding xmlEncoding)
82
+ {
83
+ const char* encodingName;
84
+
85
+ switch (xmlEncoding)
86
+ {
87
+ case XML_CHAR_ENCODING_UTF8:
88
+ encodingName = "UTF-8";
89
+ break;
90
+ case XML_CHAR_ENCODING_UTF16LE:
91
+ encodingName = "UTF-16LE";
92
+ break;
93
+ case XML_CHAR_ENCODING_UTF16BE:
94
+ encodingName = "UTF-16BE";
95
+ break;
96
+ case XML_CHAR_ENCODING_UCS4LE:
97
+ encodingName = "UCS-4LE";
98
+ break;
99
+ case XML_CHAR_ENCODING_UCS4BE:
100
+ encodingName = "UCS-4BE";
101
+ break;
102
+ case XML_CHAR_ENCODING_UCS2:
103
+ encodingName = "UCS-2";
104
+ break;
105
+ case XML_CHAR_ENCODING_8859_1:
106
+ encodingName = "ISO8859-1";
107
+ break;
108
+ case XML_CHAR_ENCODING_8859_2:
109
+ encodingName = "ISO8859-2";
110
+ break;
111
+ case XML_CHAR_ENCODING_8859_3:
112
+ encodingName = "ISO8859-3";
113
+ break;
114
+ case XML_CHAR_ENCODING_8859_4:
115
+ encodingName = "ISO8859-4";
116
+ break;
117
+ case XML_CHAR_ENCODING_8859_5:
118
+ encodingName = "ISO8859-5";
119
+ break;
120
+ case XML_CHAR_ENCODING_8859_6:
121
+ encodingName = "ISO8859-6";
122
+ break;
123
+ case XML_CHAR_ENCODING_8859_7:
124
+ encodingName = "ISO8859-7";
125
+ break;
126
+ case XML_CHAR_ENCODING_8859_8:
127
+ encodingName = "ISO8859-8";
128
+ break;
129
+ case XML_CHAR_ENCODING_8859_9:
130
+ encodingName = "ISO8859-9";
131
+ break;
132
+ case XML_CHAR_ENCODING_2022_JP:
133
+ encodingName = "ISO-2022-JP";
134
+ break;
135
+ case XML_CHAR_ENCODING_SHIFT_JIS:
136
+ encodingName = "SHIFT-JIS";
137
+ break;
138
+ case XML_CHAR_ENCODING_EUC_JP:
139
+ encodingName = "EUC-JP";
140
+ break;
141
+ case XML_CHAR_ENCODING_ASCII:
142
+ encodingName = "US-ASCII";
143
+ break;
144
+ default:
145
+ /* Covers XML_CHAR_ENCODING_ERROR, XML_CHAR_ENCODING_NONE, XML_CHAR_ENCODING_EBCDIC */
146
+ encodingName = "ASCII-8BIT";
147
+ break;
148
+ }
149
+
150
+ return rb_enc_find(encodingName);
151
+ }
152
+
153
+ /*
154
+ * call-seq:
155
+ * Input.encoding_to_rb_encoding(Input::ENCODING) -> Encoding
156
+ *
157
+ * Converts an encoding constant defined on the XML::Encoding
158
+ * class to a Ruby encoding object (available on Ruby 1.9.* and higher).
159
+ */
160
+ VALUE rxml_encoding_to_rb_encoding(VALUE klass, VALUE encoding)
161
+ {
162
+ xmlCharEncoding xmlEncoding = (xmlCharEncoding)NUM2INT(encoding);
163
+ rb_encoding* rbencoding = rxml_xml_encoding_to_rb_encoding(klass, xmlEncoding);
164
+ return rb_enc_from_encoding(rbencoding);
165
+ }
166
+
167
+ rb_encoding* rxml_figure_encoding(const xmlChar* xencoding)
168
+ {
169
+ rb_encoding* result;
170
+ if (xencoding)
171
+ {
172
+ xmlCharEncoding xmlEncoding = xmlParseCharEncoding((const char*)xencoding);
173
+ result = rxml_xml_encoding_to_rb_encoding(mXMLEncoding, xmlEncoding);
174
+ }
175
+ else
176
+ {
177
+ result = rb_utf8_encoding();
178
+ }
179
+ return result;
180
+ }
181
+
182
+ VALUE rxml_new_cstr(const xmlChar* xstr, const xmlChar* xencoding)
183
+ {
184
+ rb_encoding *rbencoding = rxml_figure_encoding(xencoding);
185
+ return rb_external_str_new_with_enc((const char*)xstr, strlen((const char*)xstr), rbencoding);
186
+ }
187
+
188
+ VALUE rxml_new_cstr_len(const xmlChar* xstr, const long length, const xmlChar* xencoding)
189
+ {
190
+ rb_encoding *rbencoding = rxml_figure_encoding(xencoding);
191
+ return rb_external_str_new_with_enc((const char*)xstr, length, rbencoding);
192
+ }
193
+
194
+ void rxml_init_encoding(void)
195
+ {
196
+ mXMLEncoding = rb_define_module_under(mXML, "Encoding");
197
+ rb_define_module_function(mXMLEncoding, "from_s", rxml_encoding_from_s, 1);
198
+ rb_define_module_function(mXMLEncoding, "to_s", rxml_encoding_to_s, 1);
199
+
200
+ rb_define_module_function(mXMLEncoding, "to_rb_encoding", rxml_encoding_to_rb_encoding, 2);
201
+
202
+ /* -1: No char encoding detected. */
203
+ rb_define_const(mXMLEncoding, "ERROR", INT2NUM(XML_CHAR_ENCODING_ERROR));
204
+ /* 0: No char encoding detected. */
205
+ rb_define_const(mXMLEncoding, "NONE", INT2NUM(XML_CHAR_ENCODING_NONE));
206
+ /* 1: UTF-8 */
207
+ rb_define_const(mXMLEncoding, "UTF_8", INT2NUM(XML_CHAR_ENCODING_UTF8));
208
+ /* 2: UTF-16 little endian. */
209
+ rb_define_const(mXMLEncoding, "UTF_16LE", INT2NUM(XML_CHAR_ENCODING_UTF16LE));
210
+ /* 3: UTF-16 big endian. */
211
+ rb_define_const(mXMLEncoding, "UTF_16BE", INT2NUM(XML_CHAR_ENCODING_UTF16BE));
212
+ /* 4: UCS-4 little endian. */
213
+ rb_define_const(mXMLEncoding, "UCS_4LE", INT2NUM(XML_CHAR_ENCODING_UCS4LE));
214
+ /* 5: UCS-4 big endian. */
215
+ rb_define_const(mXMLEncoding, "UCS_4BE", INT2NUM(XML_CHAR_ENCODING_UCS4BE));
216
+ /* 6: EBCDIC uh! */
217
+ rb_define_const(mXMLEncoding, "EBCDIC", INT2NUM(XML_CHAR_ENCODING_EBCDIC));
218
+ /* 7: UCS-4 unusual ordering. */
219
+ rb_define_const(mXMLEncoding, "UCS_4_2143", INT2NUM(XML_CHAR_ENCODING_UCS4_2143));
220
+ /* 8: UCS-4 unusual ordering. */
221
+ rb_define_const(mXMLEncoding, "UCS_4_3412", INT2NUM(XML_CHAR_ENCODING_UCS4_3412));
222
+ /* 9: UCS-2. */
223
+ rb_define_const(mXMLEncoding, "UCS_2", INT2NUM(XML_CHAR_ENCODING_UCS2));
224
+ /* 10: ISO-8859-1 ISO Latin 1. */
225
+ rb_define_const(mXMLEncoding, "ISO_8859_1", INT2NUM(XML_CHAR_ENCODING_8859_1));
226
+ /* 11: ISO-8859-2 ISO Latin 2. */
227
+ rb_define_const(mXMLEncoding, "ISO_8859_2", INT2NUM(XML_CHAR_ENCODING_8859_2));
228
+ /* 12: ISO-8859-3. */
229
+ rb_define_const(mXMLEncoding, "ISO_8859_3", INT2NUM(XML_CHAR_ENCODING_8859_3));
230
+ /* 13: ISO-8859-4. */
231
+ rb_define_const(mXMLEncoding, "ISO_8859_4", INT2NUM(XML_CHAR_ENCODING_8859_4));
232
+ /* 14: ISO-8859-5. */
233
+ rb_define_const(mXMLEncoding, "ISO_8859_5", INT2NUM(XML_CHAR_ENCODING_8859_5));
234
+ /* 15: ISO-8859-6. */
235
+ rb_define_const(mXMLEncoding, "ISO_8859_6", INT2NUM(XML_CHAR_ENCODING_8859_6));
236
+ /* 16: ISO-8859-7. */
237
+ rb_define_const(mXMLEncoding, "ISO_8859_7", INT2NUM(XML_CHAR_ENCODING_8859_7));
238
+ /* 17: ISO-8859-8. */
239
+ rb_define_const(mXMLEncoding, "ISO_8859_8", INT2NUM(XML_CHAR_ENCODING_8859_8));
240
+ /* 18: ISO-8859-9. */
241
+ rb_define_const(mXMLEncoding, "ISO_8859_9", INT2NUM(XML_CHAR_ENCODING_8859_9));
242
+ /* 19: ISO-2022-JP. */
243
+ rb_define_const(mXMLEncoding, "ISO_2022_JP", INT2NUM(XML_CHAR_ENCODING_2022_JP));
244
+ /* 20: Shift_JIS. */
245
+ rb_define_const(mXMLEncoding, "SHIFT_JIS", INT2NUM(XML_CHAR_ENCODING_SHIFT_JIS));
246
+ /* 21: EUC-JP. */
247
+ rb_define_const(mXMLEncoding, "EUC_JP", INT2NUM(XML_CHAR_ENCODING_EUC_JP));
248
+ /* 22: pure ASCII. */
249
+ rb_define_const(mXMLEncoding, "ASCII", INT2NUM(XML_CHAR_ENCODING_ASCII));
250
+ }
@@ -1,19 +1,16 @@
1
- /* Please see the LICENSE file for copyright and distribution information */
2
-
3
- #ifndef __RXML_ENCODING__
4
- #define __RXML_ENCODING__
5
-
6
- extern VALUE mXMLEncoding;
7
-
8
- void rxml_init_encoding();
9
-
10
- // Ruby 1.8/1.9 encoding compatibility
11
- VALUE rxml_new_cstr(const char* xstr, const char* xencoding);
12
- VALUE rxml_new_cstr_len(const char* xstr, const int length, const char* xencoding);
13
-
14
- #ifdef HAVE_RUBY_ENCODING_H
15
- rb_encoding* rxml_xml_encoding_to_rb_encoding(VALUE klass, xmlCharEncoding xmlEncoding);
16
- rb_encoding* rxml_figure_encoding(const char* xencoding);
17
- #endif
18
-
19
- #endif
1
+ /* Please see the LICENSE file for copyright and distribution information */
2
+
3
+ #ifndef __RXML_ENCODING__
4
+ #define __RXML_ENCODING__
5
+
6
+ extern VALUE mXMLEncoding;
7
+
8
+ void rxml_init_encoding();
9
+
10
+ VALUE rxml_new_cstr(const xmlChar* xstr, const xmlChar* xencoding);
11
+ VALUE rxml_new_cstr_len(const xmlChar* xstr, const long length, const xmlChar* xencoding);
12
+
13
+ rb_encoding* rxml_xml_encoding_to_rb_encoding(VALUE klass, xmlCharEncoding xmlEncoding);
14
+ rb_encoding* rxml_figure_encoding(const xmlChar* xencoding);
15
+
16
+ #endif