libxml-ruby 2.7.0 → 2.9.0
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +6 -14
- data/HISTORY +33 -11
- data/README.rdoc +7 -7
- data/Rakefile +80 -77
- data/ext/libxml/extconf.h +4 -5
- data/ext/libxml/extconf.rb +57 -118
- data/ext/libxml/libxml.c +4 -0
- data/ext/libxml/ruby_xml.c +977 -893
- data/ext/libxml/ruby_xml.h +20 -10
- data/ext/libxml/ruby_xml_attr.c +333 -333
- data/ext/libxml/ruby_xml_attr_decl.c +2 -2
- data/ext/libxml/ruby_xml_cbg.c +85 -85
- data/ext/libxml/ruby_xml_document.c +1133 -1147
- data/ext/libxml/ruby_xml_dtd.c +261 -268
- data/ext/libxml/ruby_xml_encoding.c +262 -260
- data/ext/libxml/ruby_xml_encoding.h +19 -19
- data/ext/libxml/ruby_xml_html_parser_context.c +337 -338
- data/ext/libxml/ruby_xml_input_cbg.c +191 -191
- data/ext/libxml/ruby_xml_io.c +52 -50
- data/ext/libxml/ruby_xml_namespace.c +2 -2
- data/ext/libxml/ruby_xml_node.c +1446 -1452
- data/ext/libxml/ruby_xml_parser_context.c +999 -1001
- data/ext/libxml/ruby_xml_reader.c +1226 -1228
- data/ext/libxml/ruby_xml_relaxng.c +110 -111
- data/ext/libxml/ruby_xml_sax2_handler.c +326 -328
- data/ext/libxml/ruby_xml_schema.c +300 -301
- data/ext/libxml/ruby_xml_version.h +3 -3
- data/ext/libxml/ruby_xml_writer.c +34 -16
- data/ext/libxml/ruby_xml_xpath.c +188 -187
- data/ext/libxml/ruby_xml_xpath_context.c +360 -361
- data/ext/libxml/ruby_xml_xpath_object.c +335 -335
- data/libxml-ruby.gemspec +47 -44
- data/test/tc_attr.rb +5 -7
- data/test/tc_attr_decl.rb +5 -6
- data/test/tc_attributes.rb +1 -2
- data/test/tc_canonicalize.rb +1 -2
- data/test/tc_deprecated_require.rb +1 -2
- data/test/tc_document.rb +4 -5
- data/test/tc_document_write.rb +2 -3
- data/test/tc_dtd.rb +4 -5
- data/test/tc_encoding.rb +126 -126
- data/test/tc_encoding_sax.rb +4 -3
- data/test/tc_error.rb +14 -15
- data/test/tc_html_parser.rb +15 -7
- data/test/tc_html_parser_context.rb +1 -2
- data/test/tc_namespace.rb +2 -3
- data/test/tc_namespaces.rb +5 -6
- data/test/tc_node.rb +2 -3
- data/test/tc_node_cdata.rb +2 -3
- data/test/tc_node_comment.rb +1 -2
- data/test/tc_node_copy.rb +1 -2
- data/test/tc_node_edit.rb +5 -7
- data/test/tc_node_pi.rb +1 -2
- data/test/tc_node_text.rb +2 -3
- data/test/tc_node_write.rb +2 -3
- data/test/tc_node_xlink.rb +1 -2
- data/test/tc_parser.rb +18 -24
- data/test/tc_parser_context.rb +6 -7
- data/test/tc_properties.rb +1 -2
- data/test/tc_reader.rb +9 -10
- data/test/tc_relaxng.rb +4 -5
- data/test/tc_sax_parser.rb +9 -10
- data/test/tc_schema.rb +4 -5
- data/test/tc_traversal.rb +1 -2
- data/test/tc_writer.rb +1 -2
- data/test/tc_xinclude.rb +1 -2
- data/test/tc_xml.rb +1 -2
- data/test/tc_xpath.rb +8 -9
- data/test/tc_xpath_context.rb +3 -4
- data/test/tc_xpath_expression.rb +3 -4
- data/test/tc_xpointer.rb +1 -3
- data/test/test_helper.rb +3 -1
- data/test/test_suite.rb +0 -1
- metadata +90 -72
- data/test/etc_doc_to_s.rb +0 -21
- data/test/ets_doc_file.rb +0 -17
- data/test/ets_doc_to_s.rb +0 -23
- data/test/ets_gpx.rb +0 -28
- data/test/ets_node_gc.rb +0 -23
- data/test/ets_test.xml +0 -2
- data/test/ets_tsr.rb +0 -11
- data/test/model/kml_sample.xml +0 -915
- data/test/remove_test.rb +0 -9
- data/test/tc_gc.rb +0 -86
- data/test/tc_parser.rb.orig +0 -384
@@ -1,260 +1,262 @@
|
|
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
|
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
|
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
|
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
|
-
#
|
190
|
-
return rb_str_new2(xstr);
|
191
|
-
|
192
|
-
|
193
|
-
|
194
|
-
|
195
|
-
|
196
|
-
|
197
|
-
|
198
|
-
|
199
|
-
|
200
|
-
|
201
|
-
|
202
|
-
|
203
|
-
|
204
|
-
|
205
|
-
|
206
|
-
|
207
|
-
|
208
|
-
|
209
|
-
|
210
|
-
#
|
211
|
-
|
212
|
-
|
213
|
-
|
214
|
-
/*
|
215
|
-
rb_define_const(mXMLEncoding, "
|
216
|
-
/*
|
217
|
-
rb_define_const(mXMLEncoding, "
|
218
|
-
/*
|
219
|
-
rb_define_const(mXMLEncoding, "
|
220
|
-
/*
|
221
|
-
rb_define_const(mXMLEncoding, "
|
222
|
-
/*
|
223
|
-
rb_define_const(mXMLEncoding, "
|
224
|
-
/*
|
225
|
-
rb_define_const(mXMLEncoding, "
|
226
|
-
/*
|
227
|
-
rb_define_const(mXMLEncoding, "
|
228
|
-
/*
|
229
|
-
rb_define_const(mXMLEncoding, "
|
230
|
-
/*
|
231
|
-
rb_define_const(mXMLEncoding, "
|
232
|
-
/*
|
233
|
-
rb_define_const(mXMLEncoding, "
|
234
|
-
/*
|
235
|
-
rb_define_const(mXMLEncoding, "
|
236
|
-
/*
|
237
|
-
rb_define_const(mXMLEncoding, "
|
238
|
-
/*
|
239
|
-
rb_define_const(mXMLEncoding, "
|
240
|
-
/*
|
241
|
-
rb_define_const(mXMLEncoding, "
|
242
|
-
/*
|
243
|
-
rb_define_const(mXMLEncoding, "
|
244
|
-
/*
|
245
|
-
rb_define_const(mXMLEncoding, "
|
246
|
-
/*
|
247
|
-
rb_define_const(mXMLEncoding, "
|
248
|
-
/*
|
249
|
-
rb_define_const(mXMLEncoding, "
|
250
|
-
/*
|
251
|
-
rb_define_const(mXMLEncoding, "
|
252
|
-
/*
|
253
|
-
rb_define_const(mXMLEncoding, "
|
254
|
-
/*
|
255
|
-
rb_define_const(mXMLEncoding, "
|
256
|
-
/*
|
257
|
-
rb_define_const(mXMLEncoding, "
|
258
|
-
/*
|
259
|
-
rb_define_const(mXMLEncoding, "
|
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
|
+
#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 xmlChar* xencoding)
|
169
|
+
{
|
170
|
+
rb_encoding* result;
|
171
|
+
if (xencoding)
|
172
|
+
{
|
173
|
+
xmlCharEncoding xmlEncoding = xmlParseCharEncoding((const char*)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 xmlChar* xstr, const xmlChar* xencoding)
|
185
|
+
{
|
186
|
+
#ifdef HAVE_RUBY_ENCODING_H
|
187
|
+
rb_encoding *rbencoding = rxml_figure_encoding(xencoding);
|
188
|
+
return rb_external_str_new_with_enc((const char*)xstr, strlen((const char*)xstr), rbencoding);
|
189
|
+
#else
|
190
|
+
return rb_str_new2((const char*)xstr);
|
191
|
+
#endif
|
192
|
+
}
|
193
|
+
|
194
|
+
VALUE rxml_new_cstr_len(const xmlChar* xstr, const long length, const xmlChar* xencoding)
|
195
|
+
{
|
196
|
+
#ifdef HAVE_RUBY_ENCODING_H
|
197
|
+
rb_encoding *rbencoding = rxml_figure_encoding(xencoding);
|
198
|
+
return rb_external_str_new_with_enc((const char*)xstr, length, rbencoding);
|
199
|
+
#else
|
200
|
+
return rb_str_new((const char*)xstr, length);
|
201
|
+
#endif
|
202
|
+
}
|
203
|
+
|
204
|
+
void rxml_init_encoding(void)
|
205
|
+
{
|
206
|
+
mXMLEncoding = rb_define_module_under(mXML, "Encoding");
|
207
|
+
rb_define_module_function(mXMLEncoding, "from_s", rxml_encoding_from_s, 1);
|
208
|
+
rb_define_module_function(mXMLEncoding, "to_s", rxml_encoding_to_s, 1);
|
209
|
+
|
210
|
+
#ifdef HAVE_RUBY_ENCODING_H
|
211
|
+
// rb_define_module_function(mXMLEncoding, "to_rb_encoding", rxml_encoding_to_rb_encoding, 2);
|
212
|
+
#endif
|
213
|
+
|
214
|
+
/* -1: No char encoding detected. */
|
215
|
+
rb_define_const(mXMLEncoding, "ERROR", INT2NUM(XML_CHAR_ENCODING_ERROR));
|
216
|
+
/* 0: No char encoding detected. */
|
217
|
+
rb_define_const(mXMLEncoding, "NONE", INT2NUM(XML_CHAR_ENCODING_NONE));
|
218
|
+
/* 1: UTF-8 */
|
219
|
+
rb_define_const(mXMLEncoding, "UTF_8", INT2NUM(XML_CHAR_ENCODING_UTF8));
|
220
|
+
/* 2: UTF-16 little endian. */
|
221
|
+
rb_define_const(mXMLEncoding, "UTF_16LE", INT2NUM(XML_CHAR_ENCODING_UTF16LE));
|
222
|
+
/* 3: UTF-16 big endian. */
|
223
|
+
rb_define_const(mXMLEncoding, "UTF_16BE", INT2NUM(XML_CHAR_ENCODING_UTF16BE));
|
224
|
+
/* 4: UCS-4 little endian. */
|
225
|
+
rb_define_const(mXMLEncoding, "UCS_4LE", INT2NUM(XML_CHAR_ENCODING_UCS4LE));
|
226
|
+
/* 5: UCS-4 big endian. */
|
227
|
+
rb_define_const(mXMLEncoding, "UCS_4BE", INT2NUM(XML_CHAR_ENCODING_UCS4BE));
|
228
|
+
/* 6: EBCDIC uh! */
|
229
|
+
rb_define_const(mXMLEncoding, "EBCDIC", INT2NUM(XML_CHAR_ENCODING_EBCDIC));
|
230
|
+
/* 7: UCS-4 unusual ordering. */
|
231
|
+
rb_define_const(mXMLEncoding, "UCS_4_2143", INT2NUM(XML_CHAR_ENCODING_UCS4_2143));
|
232
|
+
/* 8: UCS-4 unusual ordering. */
|
233
|
+
rb_define_const(mXMLEncoding, "UCS_4_3412", INT2NUM(XML_CHAR_ENCODING_UCS4_3412));
|
234
|
+
/* 9: UCS-2. */
|
235
|
+
rb_define_const(mXMLEncoding, "UCS_2", INT2NUM(XML_CHAR_ENCODING_UCS2));
|
236
|
+
/* 10: ISO-8859-1 ISO Latin 1. */
|
237
|
+
rb_define_const(mXMLEncoding, "ISO_8859_1", INT2NUM(XML_CHAR_ENCODING_8859_1));
|
238
|
+
/* 11: ISO-8859-2 ISO Latin 2. */
|
239
|
+
rb_define_const(mXMLEncoding, "ISO_8859_2", INT2NUM(XML_CHAR_ENCODING_8859_2));
|
240
|
+
/* 12: ISO-8859-3. */
|
241
|
+
rb_define_const(mXMLEncoding, "ISO_8859_3", INT2NUM(XML_CHAR_ENCODING_8859_3));
|
242
|
+
/* 13: ISO-8859-4. */
|
243
|
+
rb_define_const(mXMLEncoding, "ISO_8859_4", INT2NUM(XML_CHAR_ENCODING_8859_4));
|
244
|
+
/* 14: ISO-8859-5. */
|
245
|
+
rb_define_const(mXMLEncoding, "ISO_8859_5", INT2NUM(XML_CHAR_ENCODING_8859_5));
|
246
|
+
/* 15: ISO-8859-6. */
|
247
|
+
rb_define_const(mXMLEncoding, "ISO_8859_6", INT2NUM(XML_CHAR_ENCODING_8859_6));
|
248
|
+
/* 16: ISO-8859-7. */
|
249
|
+
rb_define_const(mXMLEncoding, "ISO_8859_7", INT2NUM(XML_CHAR_ENCODING_8859_7));
|
250
|
+
/* 17: ISO-8859-8. */
|
251
|
+
rb_define_const(mXMLEncoding, "ISO_8859_8", INT2NUM(XML_CHAR_ENCODING_8859_8));
|
252
|
+
/* 18: ISO-8859-9. */
|
253
|
+
rb_define_const(mXMLEncoding, "ISO_8859_9", INT2NUM(XML_CHAR_ENCODING_8859_9));
|
254
|
+
/* 19: ISO-2022-JP. */
|
255
|
+
rb_define_const(mXMLEncoding, "ISO_2022_JP", INT2NUM(XML_CHAR_ENCODING_2022_JP));
|
256
|
+
/* 20: Shift_JIS. */
|
257
|
+
rb_define_const(mXMLEncoding, "SHIFT_JIS", INT2NUM(XML_CHAR_ENCODING_SHIFT_JIS));
|
258
|
+
/* 21: EUC-JP. */
|
259
|
+
rb_define_const(mXMLEncoding, "EUC_JP", INT2NUM(XML_CHAR_ENCODING_EUC_JP));
|
260
|
+
/* 22: pure ASCII. */
|
261
|
+
rb_define_const(mXMLEncoding, "ASCII", INT2NUM(XML_CHAR_ENCODING_ASCII));
|
262
|
+
}
|
@@ -1,19 +1,19 @@
|
|
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
|
12
|
-
VALUE rxml_new_cstr_len(const
|
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
|
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
|
+
// Ruby 1.8/1.9 encoding compatibility
|
11
|
+
VALUE rxml_new_cstr(const xmlChar* xstr, const xmlChar* xencoding);
|
12
|
+
VALUE rxml_new_cstr_len(const xmlChar* xstr, const long length, const xmlChar* 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 xmlChar* xencoding);
|
17
|
+
#endif
|
18
|
+
|
19
|
+
#endif
|