libxml-ruby 0.9.8-x86-mswin32-60 → 0.9.9-x86-mswin32-60

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 (98) hide show
  1. data/CHANGES +41 -1
  2. data/LICENSE +3 -4
  3. data/README +37 -24
  4. data/Rakefile +2 -2
  5. data/ext/libxml/extconf.rb +31 -12
  6. data/ext/libxml/libxml.c +56 -858
  7. data/ext/libxml/ruby_libxml.h +93 -96
  8. data/ext/libxml/ruby_xml.c +855 -0
  9. data/ext/libxml/ruby_xml.h +9 -0
  10. data/ext/libxml/ruby_xml_attr.c +3 -9
  11. data/ext/libxml/ruby_xml_attr.h +2 -2
  12. data/ext/libxml/ruby_xml_attr_decl.c +2 -8
  13. data/ext/libxml/ruby_xml_attr_decl.h +1 -1
  14. data/ext/libxml/ruby_xml_attributes.c +6 -8
  15. data/ext/libxml/ruby_xml_attributes.h +1 -1
  16. data/ext/libxml/ruby_xml_document.c +915 -895
  17. data/ext/libxml/ruby_xml_document.h +2 -2
  18. data/ext/libxml/ruby_xml_dtd.c +257 -136
  19. data/ext/libxml/ruby_xml_dtd.h +1 -1
  20. data/ext/libxml/ruby_xml_encoding.c +55 -37
  21. data/ext/libxml/ruby_xml_encoding.h +1 -1
  22. data/ext/libxml/ruby_xml_error.c +526 -1058
  23. data/ext/libxml/ruby_xml_error.h +1 -1
  24. data/ext/libxml/ruby_xml_html_parser.c +2 -8
  25. data/ext/libxml/ruby_xml_html_parser.h +2 -2
  26. data/ext/libxml/ruby_xml_html_parser_context.c +175 -145
  27. data/ext/libxml/ruby_xml_html_parser_context.h +1 -1
  28. data/ext/libxml/ruby_xml_html_parser_options.c +12 -20
  29. data/ext/libxml/ruby_xml_html_parser_options.h +1 -1
  30. data/ext/libxml/ruby_xml_input_cbg.c +2 -8
  31. data/ext/libxml/ruby_xml_input_cbg.h +1 -1
  32. data/ext/libxml/ruby_xml_namespace.c +2 -8
  33. data/ext/libxml/ruby_xml_namespace.h +2 -2
  34. data/ext/libxml/ruby_xml_namespaces.c +1 -9
  35. data/ext/libxml/ruby_xml_namespaces.h +1 -1
  36. data/ext/libxml/ruby_xml_node.c +182 -121
  37. data/ext/libxml/ruby_xml_node.h +2 -2
  38. data/ext/libxml/ruby_xml_parser.c +2 -8
  39. data/ext/libxml/ruby_xml_parser.h +2 -2
  40. data/ext/libxml/ruby_xml_parser_context.c +952 -901
  41. data/ext/libxml/ruby_xml_parser_context.h +2 -2
  42. data/ext/libxml/ruby_xml_parser_options.c +2 -9
  43. data/ext/libxml/ruby_xml_parser_options.h +1 -1
  44. data/ext/libxml/ruby_xml_reader.c +1002 -993
  45. data/ext/libxml/ruby_xml_reader.h +1 -1
  46. data/ext/libxml/ruby_xml_relaxng.c +1 -7
  47. data/ext/libxml/ruby_xml_relaxng.h +1 -1
  48. data/ext/libxml/ruby_xml_sax2_handler.c +2 -2
  49. data/ext/libxml/ruby_xml_sax2_handler.h +1 -1
  50. data/ext/libxml/ruby_xml_sax_parser.c +2 -8
  51. data/ext/libxml/ruby_xml_sax_parser.h +2 -2
  52. data/ext/libxml/ruby_xml_schema.c +1 -7
  53. data/ext/libxml/ruby_xml_schema.h +1 -1
  54. data/ext/libxml/{version.h → ruby_xml_version.h} +2 -2
  55. data/ext/libxml/ruby_xml_xinclude.c +2 -8
  56. data/ext/libxml/ruby_xml_xinclude.h +2 -2
  57. data/ext/libxml/ruby_xml_xpath.c +17 -18
  58. data/ext/libxml/ruby_xml_xpath.h +2 -2
  59. data/ext/libxml/ruby_xml_xpath_context.c +387 -389
  60. data/ext/libxml/ruby_xml_xpath_context.h +2 -2
  61. data/ext/libxml/ruby_xml_xpath_expression.c +18 -8
  62. data/ext/libxml/ruby_xml_xpath_expression.h +1 -1
  63. data/ext/libxml/ruby_xml_xpath_object.c +19 -8
  64. data/ext/libxml/ruby_xml_xpath_object.h +1 -1
  65. data/ext/libxml/ruby_xml_xpointer.c +2 -8
  66. data/ext/libxml/ruby_xml_xpointer.h +2 -2
  67. data/ext/mingw/libxml_ruby.dll.a +0 -0
  68. data/ext/mingw/libxml_ruby.so +0 -0
  69. data/ext/vc/libxml_ruby.sln +7 -1
  70. data/lib/libxml.rb +1 -12
  71. data/lib/libxml/attr.rb +0 -3
  72. data/lib/libxml/attr_decl.rb +0 -3
  73. data/lib/libxml/attributes.rb +0 -3
  74. data/lib/libxml/document.rb +31 -5
  75. data/lib/libxml/error.rb +8 -4
  76. data/lib/libxml/properties.rb +0 -5
  77. data/lib/libxml/sax_callbacks.rb +30 -19
  78. data/lib/libxml/tree.rb +0 -1
  79. data/lib/libxml/xpath_object.rb +0 -13
  80. data/test/model/definition.dtd +8 -0
  81. data/test/tc_attributes.rb +4 -1
  82. data/test/tc_document.rb +16 -0
  83. data/test/tc_dtd.rb +30 -2
  84. data/test/tc_html_parser.rb +55 -10
  85. data/test/tc_node.rb +67 -1
  86. data/test/tc_node_edit.rb +26 -6
  87. data/test/tc_node_text.rb +41 -23
  88. data/test/tc_parser.rb +50 -0
  89. data/test/tc_reader.rb +15 -0
  90. data/test/tc_relaxng.rb +1 -1
  91. data/test/tc_sax_parser.rb +37 -5
  92. data/test/tc_schema.rb +1 -1
  93. data/test/tc_xpath.rb +1 -0
  94. data/test/tc_xpath_expression.rb +4 -2
  95. metadata +6 -6
  96. data/ext/libxml/ruby_xml_state.c +0 -51
  97. data/ext/libxml/ruby_xml_state.h +0 -11
  98. data/ext/vc/libxml_ruby.vcproj +0 -460
@@ -1,4 +1,4 @@
1
- /* $Id: ruby_xml_parser_context.h 711 2009-01-20 07:17:15Z cfis $ */
1
+ /* $Id: ruby_xml_parser_context.h 758 2009-01-25 20:36:03Z cfis $ */
2
2
 
3
3
  /* Please see the LICENSE file for copyright and distribution information */
4
4
 
@@ -7,6 +7,6 @@
7
7
 
8
8
  extern VALUE cXMLParserContext;
9
9
 
10
- void ruby_init_xml_parser_context(void);
10
+ void rxml_init_parser_context(void);
11
11
 
12
12
  #endif
@@ -5,7 +5,7 @@
5
5
  #include <stdarg.h>
6
6
  #include "ruby_libxml.h"
7
7
 
8
- /* Document-class: LibXML::XML::ParserOptions
8
+ /* Document-class: LibXML::XML::Parser::Options
9
9
  *
10
10
  * Options that control the operation of the HTMLParser. The easiest
11
11
  * way to set a parser's options is to use the methods
@@ -15,14 +15,7 @@
15
15
 
16
16
  VALUE mXMLParserOptions;
17
17
 
18
-
19
- // Rdoc needs to know
20
- #ifdef RDOC_NEVER_DEFINED
21
- mLibXML = rb_define_module("LibXML");
22
- mXML = rb_define_module_under(mLibXML, "XML");
23
- #endif
24
-
25
- void ruby_init_parser_options(void)
18
+ void rxml_init_parser_options(void)
26
19
  {
27
20
  mXMLParserOptions = rb_define_module_under(cXMLParser, "Options");
28
21
 
@@ -9,6 +9,6 @@
9
9
 
10
10
  extern VALUE mXMLParserOptions;
11
11
 
12
- void ruby_init_parser_options();
12
+ void rxml_init_parser_options();
13
13
 
14
14
  #endif
@@ -1,993 +1,1002 @@
1
- /* Copyright (c) 2006-2007 Apple Inc.
2
- * Please see the LICENSE file for copyright and distribution information. */
3
-
4
- #include "ruby_libxml.h"
5
- #include "ruby_xml_reader.h"
6
-
7
- /*
8
- * Document-class: LibXML::XML::Reader
9
- *
10
- * The XML::Reader class provides a simpler, alternative way of parsing an XML
11
- * document in contrast to XML::Parser or XML::SaxParser. A XML::Reader instance
12
- * acts like a cursor going forward in a document stream, stopping at each node
13
- * it encounters. To advance to the next node, simply cadd XML::Reader#read.
14
- *
15
- * The XML::Reader API closely matches the DOM Core specification and supports
16
- * namespaces, xml:base, entity handling and DTDs.
17
- *
18
- * To summarize, XML::Reader provides a far simpler API to use versus XML::SaxParser
19
- * and is more memory efficient than using XML::Parser to create a DOM tree.
20
- *
21
- * Example:
22
- *
23
- * parser = XML::Reader.string("<foo><bar>1</bar><bar>2</bar><bar>3</bar></foo>")
24
- * reader.read
25
- * assert_equal('foo', reader.name)
26
- * assert_equal(nil, reader.value)
27
- *
28
- * 3.times do |i|
29
- * reader.read
30
- * assert_equal(XML::Reader::TYPE_ELEMENT, reader.node_type)
31
- * assert_equal('bar', reader.name)
32
- * reader.read
33
- * assert_equal(XML::Reader::TYPE_TEXT, reader.node_type)
34
- * assert_equal((i + 1).to_s, reader.value)
35
- * reader.read
36
- * assert_equal(XML::Reader::TYPE_END_ELEMENT, reader.node_type)
37
- * end
38
- *
39
- * You can also parse documents (see XML::Reader.document),
40
- * strings (see XML::Parser.string) and io objects (see
41
- * XML::Parser.io).
42
- *
43
- * For a more in depth tutorial, albeit in C, see http://xmlsoft.org/xmlreader.html.*/
44
-
45
- VALUE cXMLReader;
46
-
47
- ID base_uri_SYMBOL;
48
- ID ENCODING_SYMBOL;
49
- ID OPTIONS_SYMBOL;
50
-
51
- static VALUE rxml_reader_wrap(xmlTextReaderPtr reader)
52
- {
53
- return Data_Wrap_Struct(cXMLReader, NULL, xmlFreeTextReader, reader);
54
- }
55
-
56
- static xmlTextReaderPtr rxml_text_reader_get(VALUE obj)
57
- {
58
- xmlTextReaderPtr xreader;
59
- Data_Get_Struct(obj, xmlTextReader, xreader);
60
- return xreader;
61
- }
62
-
63
- /*
64
- * call-seq:
65
- * XML::Reader.document(doc) -> XML::Reader
66
- *
67
- * Create an new reader for the specified document.
68
- */
69
- VALUE rxml_reader_document(VALUE klass, VALUE doc)
70
- {
71
- xmlDocPtr xdoc;
72
- xmlTextReaderPtr xreader;
73
-
74
- Data_Get_Struct(doc, xmlDoc, xdoc);
75
-
76
- xreader = xmlReaderWalker(xdoc);
77
-
78
- if (xreader == NULL)
79
- rxml_raise(&xmlLastError);
80
-
81
- return rxml_reader_wrap(xreader);
82
- }
83
-
84
- /* call-seq:
85
- * XML::Reader.file(path) -> XML::Reader
86
- * XML::Reader.file(path, :encoding => XML::Encoding::UTF_8,
87
- * :options => XML::Parser::Options::NOENT) -> XML::Parser
88
- *
89
- * Creates a new reader by parsing the specified file or uri.
90
- *
91
- * You may provide an optional hash table to control how the
92
- * parsing is performed. Valid options are:
93
- *
94
- * encoding - The document encoding, defaults to nil. Valid values
95
- * are the encoding constants defined on XML::Encoding.
96
- * options - Controls the execution of the parser, defaults to 0.
97
- * Valid values are the constants defined on
98
- * XML::Parser::Options. Mutliple options can be combined
99
- * by using Bitwise OR (|).
100
- */
101
- static VALUE rxml_reader_file(int argc, VALUE *argv, VALUE klass)
102
- {
103
- xmlTextReaderPtr xreader;
104
- VALUE path;
105
- VALUE options;
106
-
107
- const char *xencoding = NULL;
108
- int xoptions = 0;
109
-
110
- rb_scan_args(argc, argv, "11", &path, &options);
111
- Check_Type(path, T_STRING);
112
-
113
- if (!NIL_P(options))
114
- {
115
- VALUE encoding = Qnil;
116
- VALUE parserOptions = Qnil;
117
-
118
- Check_Type(options, T_HASH);
119
-
120
- encoding = rb_hash_aref(options, base_uri_SYMBOL);
121
- xencoding = NIL_P(encoding) ? NULL : xmlGetCharEncodingName(NUM2INT(encoding));
122
-
123
- parserOptions = rb_hash_aref(options, OPTIONS_SYMBOL);
124
- xoptions = NIL_P(parserOptions) ? 0 : NUM2INT(parserOptions);
125
- }
126
-
127
- xreader = xmlReaderForFile(StringValueCStr(path), xencoding, xoptions);
128
-
129
- if (xreader == NULL)
130
- rxml_raise(&xmlLastError);
131
-
132
- return rxml_reader_wrap(xreader);
133
- }
134
-
135
- /* call-seq:
136
- * XML::Reader.io(io) -> XML::Reader
137
- * XML::Reader.io(io, :encoding => XML::Encoding::UTF_8,
138
- * :options => XML::Parser::Options::NOENT) -> XML::Parser
139
- *
140
- * Creates a new reader by parsing the specified io object.
141
- *
142
- * You may provide an optional hash table to control how the
143
- * parsing is performed. Valid options are:
144
- *
145
- * base_uri - The base url for the parsed document.
146
- * encoding - The document encoding, defaults to nil. Valid values
147
- * are the encoding constants defined on XML::Encoding.
148
- * options - Controls the execution of the parser, defaults to 0.
149
- * Valid values are the constants defined on
150
- * XML::Parser::Options. Mutliple options can be combined
151
- * by using Bitwise OR (|).
152
- */
153
- static VALUE rxml_reader_io(int argc, VALUE *argv, VALUE klass)
154
- {
155
- xmlTextReaderPtr xreader;
156
- VALUE io;
157
- VALUE options;
158
- char *xbaseurl = NULL;
159
- const char *xencoding = NULL;
160
- int xoptions = 0;
161
-
162
- rb_scan_args(argc, argv, "11", &io, &options);
163
-
164
- if (!NIL_P(options))
165
- {
166
- VALUE baseurl = Qnil;
167
- VALUE encoding = Qnil;
168
- VALUE parserOptions = Qnil;
169
-
170
- Check_Type(options, T_HASH);
171
-
172
- baseurl = rb_hash_aref(options, base_uri_SYMBOL);
173
- xbaseurl = NIL_P(baseurl) ? NULL : StringValueCStr(baseurl);
174
-
175
- encoding = rb_hash_aref(options, ENCODING_SYMBOL);
176
- xencoding = NIL_P(encoding) ? NULL : xmlGetCharEncodingName(NUM2INT(encoding));
177
-
178
- parserOptions = rb_hash_aref(options, OPTIONS_SYMBOL);
179
- xoptions = NIL_P(parserOptions) ? 0 : NUM2INT(parserOptions);
180
- }
181
-
182
- xreader = xmlReaderForIO((xmlInputReadCallback) rxml_read_callback, NULL,
183
- (void *) io,
184
- xbaseurl, xencoding, xoptions);
185
-
186
- if (xreader == NULL)
187
- rxml_raise(&xmlLastError);
188
-
189
- return rxml_reader_wrap(xreader);
190
- }
191
-
192
- /* call-seq:
193
- * XML::Reader.string(io) -> XML::Reader
194
- * XML::Reader.string(io, :encoding => XML::Encoding::UTF_8,
195
- * :options => XML::Parser::Options::NOENT) -> XML::Parser
196
- *
197
- * Creates a new reader by parsing the specified string.
198
- *
199
- * You may provide an optional hash table to control how the
200
- * parsing is performed. Valid options are:
201
- *
202
- * base_uri - The base url for the parsed document.
203
- * encoding - The document encoding, defaults to nil. Valid values
204
- * are the encoding constants defined on XML::Encoding.
205
- * options - Controls the execution of the parser, defaults to 0.
206
- * Valid values are the constants defined on
207
- * XML::Parser::Options. Mutliple options can be combined
208
- * by using Bitwise OR (|).
209
- */
210
- static VALUE rxml_reader_string(int argc, VALUE *argv, VALUE klass)
211
- {
212
- xmlTextReaderPtr xreader;
213
- VALUE string;
214
- VALUE options;
215
- char *xbaseurl = NULL;
216
- const char *xencoding = NULL;
217
- int xoptions = 0;
218
-
219
- rb_scan_args(argc, argv, "11", &string, &options);
220
- Check_Type(string, T_STRING);
221
-
222
- if (!NIL_P(options))
223
- {
224
- VALUE baseurl = Qnil;
225
- VALUE encoding = Qnil;
226
- VALUE parserOptions = Qnil;
227
-
228
- Check_Type(options, T_HASH);
229
-
230
- baseurl = rb_hash_aref(options, base_uri_SYMBOL);
231
- xbaseurl = NIL_P(baseurl) ? NULL : StringValueCStr(baseurl);
232
-
233
- encoding = rb_hash_aref(options, ENCODING_SYMBOL);
234
- xencoding = NIL_P(encoding) ? NULL : xmlGetCharEncodingName(NUM2INT(encoding));
235
-
236
- parserOptions = rb_hash_aref(options, OPTIONS_SYMBOL);
237
- xoptions = NIL_P(parserOptions) ? 0 : NUM2INT(parserOptions);
238
- }
239
-
240
- xreader = xmlReaderForMemory(StringValueCStr(string), RSTRING_LEN(string),
241
- xbaseurl, xencoding, xoptions);
242
-
243
- if (xreader == NULL)
244
- rxml_raise(&xmlLastError);
245
-
246
- return rxml_reader_wrap(xreader);
247
- }
248
-
249
- /*
250
- * call-seq:
251
- * reader.close -> code
252
- *
253
- * This method releases any resources allocated by the current instance
254
- * changes the state to Closed and close any underlying input.
255
- */
256
- static VALUE rxml_reader_close(VALUE self)
257
- {
258
- return INT2FIX(xmlTextReaderClose(rxml_text_reader_get(self)));
259
- }
260
-
261
- /*
262
- * call-seq:
263
- * reader.move_to_attribute(val) -> code
264
- *
265
- * Move the position of the current instance to the attribute with the
266
- * specified index (if +val+ is an integer) or name (if +val+ is a string)
267
- * relative to the containing element.
268
- */
269
- static VALUE rxml_reader_move_to_attr(VALUE self, VALUE val)
270
- {
271
- xmlTextReaderPtr xreader;
272
- int ret;
273
-
274
- xreader = rxml_text_reader_get(self);
275
-
276
- if (TYPE(val) == T_FIXNUM)
277
- {
278
- ret = xmlTextReaderMoveToAttributeNo(xreader, FIX2INT(val));
279
- }
280
- else
281
- {
282
- ret = xmlTextReaderMoveToAttribute(xreader,
283
- (const xmlChar *) StringValueCStr(val));
284
- }
285
-
286
- return INT2FIX(ret);
287
- }
288
-
289
- /*
290
- * call-seq:
291
- * reader.move_to_first_attribute -> code
292
- *
293
- * Move the position of the current instance to the first attribute associated
294
- * with the current node.
295
- */
296
- static VALUE rxml_reader_move_to_first_attr(VALUE self)
297
- {
298
- return INT2FIX(xmlTextReaderMoveToFirstAttribute(rxml_text_reader_get(self)));
299
- }
300
-
301
- /*
302
- * call-seq:
303
- * reader.move_to_next_attribute -> code
304
- *
305
- * Move the position of the current instance to the next attribute associated
306
- * with the current node.
307
- */
308
- static VALUE rxml_reader_move_to_next_attr(VALUE self)
309
- {
310
- return INT2FIX(xmlTextReaderMoveToNextAttribute(rxml_text_reader_get(self)));
311
- }
312
-
313
- /*
314
- * call-seq:
315
- * reader.move_to_element -> code
316
- *
317
- * Move the position of the current instance to the node that contains the
318
- * current attribute node.
319
- */
320
- static VALUE rxml_reader_move_to_element(VALUE self)
321
- {
322
- return INT2FIX(xmlTextReaderMoveToElement(rxml_text_reader_get(self)));
323
- }
324
-
325
- /*
326
- * call-seq:
327
- * reader.next -> code
328
- *
329
- * Skip to the node following the current one in document order while avoiding
330
- * the subtree if any.
331
- */
332
- static VALUE rxml_reader_next(VALUE self)
333
- {
334
- return INT2FIX(xmlTextReaderNext(rxml_text_reader_get(self)));
335
- }
336
-
337
- /*
338
- * call-seq:
339
- * reader.next_sibling -> code
340
- *
341
- * Skip to the node following the current one in document order while avoiding
342
- * the subtree if any. Currently implemented only for Readers built on a
343
- * document.
344
- */
345
- static VALUE rxml_reader_next_sibling(VALUE self)
346
- {
347
- return INT2FIX(xmlTextReaderNextSibling(rxml_text_reader_get(self)));
348
- }
349
-
350
- /*
351
- * call-seq:
352
- * reader.node -> XML::Node
353
- *
354
- * Returns the reader's current node.
355
- * WARNING - Using this method is dangerous because the
356
- * the node may be destroyed on the next #read.
357
- */
358
- static VALUE rxml_reader_node(VALUE self)
359
- {
360
- xmlTextReaderPtr xreader = rxml_text_reader_get(self);
361
- xmlNodePtr xnode = xmlTextReaderCurrentNode(xreader);
362
- return rxml_node_wrap(xnode);
363
- }
364
-
365
- /*
366
- * call-seq:
367
- * reader.node_type -> type
368
- *
369
- * Get the node type of the current node. Reference:
370
- * http://dotgnu.org/pnetlib-doc/System/Xml/XmlNodeType.html
371
- */
372
- static VALUE rxml_reader_node_type(VALUE self)
373
- {
374
- return INT2FIX(xmlTextReaderNodeType(rxml_text_reader_get(self)));
375
- }
376
-
377
- /*
378
- * call-seq:
379
- * reader.normalization -> value
380
- *
381
- * The value indicating whether to normalize white space and attribute values.
382
- * Since attribute value and end of line normalizations are a MUST in the XML
383
- * specification only the value true is accepted. The broken bahaviour of
384
- * accepting out of range character entities like &#0; is of course not
385
- * supported either.
386
- *
387
- * Return 1 or -1 in case of error.
388
- */
389
- static VALUE rxml_reader_normalization(VALUE self)
390
- {
391
- return INT2FIX(xmlTextReaderNormalization(rxml_text_reader_get(self)));
392
- }
393
-
394
- /*
395
- * call-seq:
396
- * reader.read -> code
397
- *
398
- * Causes the reader to move to the next node in the stream, exposing its properties.
399
- *
400
- * Returns true if a node was successfully read or false if there are no more
401
- * nodes to read. On errors, an exception is raised.*/
402
- static VALUE rxml_reader_read(VALUE self)
403
- {
404
- int result = xmlTextReaderRead(rxml_text_reader_get(self));
405
- switch(result)
406
- {
407
- case -1:
408
- rxml_raise(&xmlLastError);
409
- return Qnil;
410
- break;
411
- case 0:
412
- return Qfalse;
413
- case 1:
414
- return Qtrue;
415
- default:
416
- rb_raise(rb_eRuntimeError,
417
- "xmlTextReaderRead did not return -1, 0 or 1. Return value was: %d", result);
418
- }
419
- }
420
-
421
- /*
422
- * call-seq:
423
- * reader.read_attribute_value -> code
424
- *
425
- * Parse an attribute value into one or more Text and EntityReference nodes.
426
- *
427
- * Return 1 in case of success, 0 if the reader was not positionned on an
428
- * attribute node or all the attribute values have been read, or -1 in case of
429
- * error.
430
- */
431
- static VALUE rxml_reader_read_attr_value(VALUE self)
432
- {
433
- return INT2FIX(xmlTextReaderReadAttributeValue(rxml_text_reader_get(self)));
434
- }
435
-
436
- /*
437
- * call-seq:
438
- * reader.read_inner_xml -> data
439
- *
440
- * Read the contents of the current node, including child nodes and markup.
441
- *
442
- * Return a string containing the XML content, or nil if the current node is
443
- * neither an element nor attribute, or has no child nodes.
444
- */
445
- static VALUE rxml_reader_read_inner_xml(VALUE self)
446
- {
447
- const xmlChar *result = xmlTextReaderReadInnerXml(rxml_text_reader_get(self));
448
- return (result == NULL ? Qnil : rb_str_new2((const char*)result));
449
- }
450
-
451
- /*
452
- * call-seq:
453
- * reader.read_outer_xml -> data
454
- *
455
- * Read the contents of the current node, including child nodes and markup.
456
- *
457
- * Return a string containing the XML content, or nil if the current node is
458
- * neither an element nor attribute, or has no child nodes.
459
- */
460
- static VALUE rxml_reader_read_outer_xml(VALUE self)
461
- {
462
- const xmlChar *result = xmlTextReaderReadOuterXml(rxml_text_reader_get(self));
463
- return (result == NULL ? Qnil : rb_str_new2((const char*)result));
464
- }
465
-
466
- /*
467
- * call-seq:
468
- * reader.read_state -> state
469
- *
470
- * Get the read state of the reader.
471
- */
472
- static VALUE rxml_reader_read_state(VALUE self)
473
- {
474
- return INT2FIX(xmlTextReaderReadState(rxml_text_reader_get(self)));
475
- }
476
-
477
- /*
478
- * call-seq:
479
- * reader.read_string -> string
480
- *
481
- * Read the contents of an element or a text node as a string.
482
- *
483
- * Return a string containing the contents of the Element or Text node, or nil
484
- * if the reader is positioned on any other type of node.
485
- */
486
- static VALUE rxml_reader_read_string(VALUE self)
487
- {
488
- const xmlChar *result = xmlTextReaderReadString(rxml_text_reader_get(self));
489
- return (result == NULL ? Qnil : rb_str_new2((const char*)result));
490
- }
491
-
492
- /*
493
- * call-seq:
494
- * reader.relax_ng_validate(rng) -> code
495
- *
496
- * Use RelaxNG to validate the document as it is processed. Activation is only
497
- * possible before the first read. If +rng+ is nil, the RelaxNG validation is
498
- * desactivated.
499
- *
500
- * Return 0 in case the RelaxNG validation could be (des)activated and -1 in
501
- * case of error.
502
- */
503
- static VALUE rxml_reader_relax_ng_validate(VALUE self, VALUE rng)
504
- {
505
- char *xrng = NIL_P(rng) ? NULL : StringValueCStr(rng);
506
- return INT2FIX(xmlTextReaderRelaxNGValidate(rxml_text_reader_get(self), xrng));
507
- }
508
-
509
- #if LIBXML_VERSION >= 20620
510
- /*
511
- * call-seq:
512
- * reader.schema_validate(schema) -> code
513
- *
514
- * Use W3C XSD schema to validate the document as it is processed. Activation
515
- * is only possible before the first read. If +schema+ is nil, then XML Schema
516
- * validation is desactivated.
517
- *
518
- * Return 0 in case the schemas validation could be (de)activated and -1 in
519
- * case of error.
520
- */
521
- static VALUE
522
- rxml_reader_schema_validate(VALUE self, VALUE xsd)
523
- {
524
- char *xxsd = NIL_P(xsd) ? NULL : StringValueCStr(xsd);
525
- int status = xmlTextReaderSchemaValidate(rxml_text_reader_get(self), xxsd);
526
- return INT2FIX(status);
527
- }
528
- #endif
529
-
530
- /*
531
- * call-seq:
532
- * reader.name -> name
533
- *
534
- * Return the qualified name of the node.
535
- */
536
- static VALUE rxml_reader_name(VALUE self)
537
- {
538
- const xmlChar *result = xmlTextReaderConstName(rxml_text_reader_get(self));
539
- return (result == NULL ? Qnil : rb_str_new2((const char*)result));
540
- }
541
-
542
- /*
543
- * call-seq:
544
- * reader.local_name -> name
545
- *
546
- * Return the local name of the node.
547
- */
548
- static VALUE rxml_reader_local_name(VALUE self)
549
- {
550
- const xmlChar *result = xmlTextReaderConstLocalName(rxml_text_reader_get(self));
551
- return (result == NULL ? Qnil : rb_str_new2((const char*)result));
552
- }
553
-
554
- /*
555
- * call-seq:
556
- * reader.attribute_count -> count
557
- *
558
- * Provide the number of attributes of the current node.
559
- */
560
- static VALUE rxml_reader_attr_count(VALUE self)
561
- {
562
- return INT2FIX(xmlTextReaderAttributeCount(rxml_text_reader_get(self)));
563
- }
564
-
565
- /*
566
- * call-seq:
567
- * reader.encoding -> XML::Encoding::UTF_8
568
- *
569
- * Returns the encoding of the document being read. Note you
570
- * first have to read data from the reader for encoding
571
- * to return a value
572
- *
573
- * reader = XML::Reader.file(XML_FILE)
574
- * assert_nil(reader.encoding)
575
- * reader.read
576
- * assert_equal(XML::Encoding::UTF_8, reader.encoding)
577
- *
578
- * In addition, libxml always appears to return nil for the encoding
579
- * when parsing strings.
580
- */
581
- static VALUE rxml_reader_encoding(VALUE self)
582
- {
583
- xmlTextReaderPtr xreader = rxml_text_reader_get(self);
584
- const xmlChar *xencoding = xmlTextReaderConstEncoding(xreader);
585
- if (xencoding)
586
- return INT2NUM(xmlParseCharEncoding(xencoding));
587
- else
588
- return INT2NUM(XML_CHAR_ENCODING_NONE);
589
- }
590
-
591
- /*
592
- * call-seq:
593
- * reader.base_uri -> URI
594
- *
595
- * Determine the base URI of the node.
596
- */
597
- static VALUE rxml_reader_base_uri(VALUE self)
598
- {
599
- const xmlChar *result = xmlTextReaderConstBaseUri(rxml_text_reader_get(self));
600
- return (result == NULL ? Qnil : rb_str_new2((const char*)result));
601
- }
602
-
603
- /*
604
- * call-seq:
605
- * reader.namespace_uri -> URI
606
- *
607
- * Determine the namespace URI of the node.
608
- */
609
- static VALUE rxml_reader_namespace_uri(VALUE self)
610
- {
611
- const xmlChar *result = xmlTextReaderConstNamespaceUri(rxml_text_reader_get(self));
612
- return (result == NULL ? Qnil : rb_str_new2((const char*)result));
613
- }
614
-
615
- /*
616
- * call-seq:
617
- * reader.value -> text
618
- *
619
- * Provide the text value of the node if present.
620
- */
621
- static VALUE rxml_reader_value(VALUE self)
622
- {
623
- const xmlChar *result = xmlTextReaderConstValue(rxml_text_reader_get(self));
624
- return (result == NULL ? Qnil : rb_str_new2((const char*)result));
625
- }
626
-
627
- /*
628
- * call-seq:
629
- * reader.prefix -> prefix
630
- *
631
- * Get a shorthand reference to the namespace associated with the node.
632
- */
633
- static VALUE rxml_reader_prefix(VALUE self)
634
- {
635
- const xmlChar *result = xmlTextReaderConstPrefix(rxml_text_reader_get(self));
636
- return (result == NULL ? Qnil : rb_str_new2((const char*)result));
637
- }
638
-
639
- /*
640
- * call-seq:
641
- * reader.depth -> depth
642
- *
643
- * Get the depth of the node in the tree.
644
- */
645
- static VALUE rxml_reader_depth(VALUE self)
646
- {
647
- return INT2FIX(xmlTextReaderDepth(rxml_text_reader_get(self)));
648
- }
649
-
650
- /*
651
- * call-seq:
652
- * reader.quote_char -> char
653
- *
654
- * Get the quotation mark character used to enclose the value of an attribute,
655
- * as an integer value (and -1 in case of error).
656
- */
657
- static VALUE rxml_reader_quote_char(VALUE self)
658
- {
659
- return INT2FIX(xmlTextReaderQuoteChar(rxml_text_reader_get(self)));
660
- }
661
-
662
- /*
663
- * call-seq:
664
- * reader.standalone -> code
665
- *
666
- * Determine the standalone status of the document being read.
667
- *
668
- * Return 1 if the document was declared to be standalone, 0 if it was
669
- * declared to be not standalone, or -1 if the document did not specify its
670
- * standalone status or in case of error.
671
- */
672
- static VALUE rxml_reader_standalone(VALUE self)
673
- {
674
- return INT2FIX(xmlTextReaderStandalone(rxml_text_reader_get(self)));
675
- }
676
-
677
- /*
678
- * call-seq:
679
- * reader.xml_lang -> value
680
- *
681
- * Get the xml:lang scope within which the node resides.
682
- */
683
- static VALUE rxml_reader_xml_lang(VALUE self)
684
- {
685
- const xmlChar *result = xmlTextReaderConstXmlLang(rxml_text_reader_get(self));
686
- return (result == NULL ? Qnil : rb_str_new2((const char*)result));
687
- }
688
-
689
- /*
690
- * call-seq:
691
- * reader.xml_version -> version
692
- *
693
- * Determine the XML version of the document being read.
694
- */
695
- static VALUE rxml_reader_xml_version(VALUE self)
696
- {
697
- const xmlChar *result = xmlTextReaderConstXmlVersion(rxml_text_reader_get(self));
698
- return (result == NULL ? Qnil : rb_str_new2((const char*)result));
699
- }
700
-
701
- /*
702
- * call-seq:
703
- * reader.has_attributes? -> bool
704
- *
705
- * Get whether the node has attributes.
706
- */
707
- static VALUE rxml_reader_has_attributes(VALUE self)
708
- {
709
- return xmlTextReaderHasAttributes(rxml_text_reader_get(self)) ? Qtrue
710
- : Qfalse;
711
- }
712
-
713
- /*
714
- * call-seq:
715
- * reader.has_value? -> bool
716
- *
717
- * Get whether the node can have a text value.
718
- */
719
- static VALUE rxml_reader_has_value(VALUE self)
720
- {
721
- return xmlTextReaderHasValue(rxml_text_reader_get(self)) ? Qtrue : Qfalse;
722
- }
723
-
724
- /*
725
- * call-seq:
726
- * reader[key] -> value
727
- *
728
- * Provide the value of the attribute with the specified index (if +key+ is an
729
- * integer) or with the specified name (if +key+ is a string) relative to the
730
- * containing element, as a string.
731
- */
732
- static VALUE rxml_reader_attribute(VALUE self, VALUE key)
733
- {
734
- xmlTextReaderPtr reader;
735
- xmlChar *attr;
736
-
737
- reader = rxml_text_reader_get(self);
738
-
739
- if (TYPE(key) == T_FIXNUM)
740
- {
741
- attr = xmlTextReaderGetAttributeNo(reader, FIX2INT(key));
742
- }
743
- else
744
- {
745
- attr = xmlTextReaderGetAttribute(reader, (const xmlChar *) StringValueCStr(key));
746
- }
747
- return (attr == NULL ? Qnil : rb_str_new2((const char*)attr));
748
- }
749
-
750
- /*
751
- * call-seq:
752
- * reader.lookup_namespace(prefix) -> value
753
- *
754
- * Resolve a namespace prefix in the scope of the current element.
755
- * To return the default namespace, specify nil as +prefix+.
756
- */
757
- static VALUE rxml_reader_lookup_namespace(VALUE self, VALUE prefix)
758
- {
759
- const xmlChar *result = xmlTextReaderLookupNamespace(rxml_text_reader_get(
760
- self), (const xmlChar *) StringValueCStr(prefix));
761
- return (result == NULL ? Qnil : rb_str_new2((const char*)result));
762
- }
763
-
764
- /*
765
- * call-seq:
766
- * reader.expand -> node
767
- *
768
- * Read the contents of the current node and the full subtree. It then makes
769
- * the subtree available until the next read call.
770
- *
771
- * Return an XML::Node object, or nil in case of error.
772
- */
773
- static VALUE rxml_reader_expand(VALUE self)
774
- {
775
- xmlNodePtr node;
776
- xmlDocPtr doc;
777
- xmlTextReaderPtr reader = rxml_text_reader_get(self);
778
- node = xmlTextReaderExpand(reader);
779
-
780
- if (!node)
781
- return Qnil;
782
-
783
- /* Okay this is tricky. By accessing the returned node, we
784
- take ownership of the reader's document. Thus we need to
785
- tell the reader to not free it. Otherwise it will be
786
- freed twice - once when the Ruby document wrapper goes
787
- out of scope and once when the reader goes out of scope. */
788
-
789
- xmlTextReaderPreserve(reader);
790
- doc = xmlTextReaderCurrentDoc(reader);
791
- rxml_document_wrap(doc);
792
-
793
- return rxml_node_wrap(node);
794
- }
795
-
796
- #if LIBXML_VERSION >= 20618
797
- /*
798
- * call-seq:
799
- * reader.byte_consumed -> value
800
- *
801
- * This method provides the current index of the parser used by the reader,
802
- * relative to the start of the current entity.
803
- */
804
- static VALUE
805
- rxml_reader_byte_consumed(VALUE self)
806
- {
807
- return INT2NUM(xmlTextReaderByteConsumed(rxml_text_reader_get(self)));
808
- }
809
- #endif
810
-
811
- #if LIBXML_VERSION >= 20617
812
- /*
813
- * call-seq:
814
- * reader.column_number -> number
815
- *
816
- * Provide the column number of the current parsing point.
817
- */
818
- static VALUE
819
- rxml_reader_column_number(VALUE self)
820
- {
821
- return INT2NUM(xmlTextReaderGetParserColumnNumber(rxml_text_reader_get(self)));
822
- }
823
-
824
- /*
825
- * call-seq:
826
- * reader.line_number -> number
827
- *
828
- * Provide the line number of the current parsing point.
829
- */
830
- static VALUE
831
- rxml_reader_line_number(VALUE self)
832
- {
833
- return INT2NUM(xmlTextReaderGetParserLineNumber(rxml_text_reader_get(self)));
834
- }
835
- #endif
836
-
837
- /*
838
- * call-seq:
839
- * reader.default? -> bool
840
- *
841
- * Return whether an Attribute node was generated from the default value
842
- * defined in the DTD or schema.
843
- */
844
- static VALUE rxml_reader_default(VALUE self)
845
- {
846
- return xmlTextReaderIsDefault(rxml_text_reader_get(self)) ? Qtrue : Qfalse;
847
- }
848
-
849
- /*
850
- * call-seq:
851
- * reader.namespace_declaration? -> bool
852
- *
853
- * Determine whether the current node is a namespace declaration rather than a
854
- * regular attribute.
855
- */
856
- static VALUE rxml_reader_namespace_declaration(VALUE self)
857
- {
858
- return xmlTextReaderIsNamespaceDecl(rxml_text_reader_get(self)) ? Qtrue
859
- : Qfalse;
860
- }
861
-
862
- /*
863
- * call-seq:
864
- * reader.empty_element? -> bool
865
- *
866
- * Check if the current node is empty.
867
- */
868
- static VALUE rxml_reader_empty_element(VALUE self)
869
- {
870
- return xmlTextReaderIsEmptyElement(rxml_text_reader_get(self)) ? Qtrue
871
- : Qfalse;
872
- }
873
-
874
- /*
875
- * call-seq:
876
- * reader.valid? -> bool
877
- *
878
- * Retrieve the validity status from the parser context.
879
- */
880
- static VALUE rxml_reader_valid(VALUE self)
881
- {
882
- return xmlTextReaderIsValid(rxml_text_reader_get(self)) ? Qtrue : Qfalse;
883
- }
884
-
885
- /* Rdoc needs to know. */
886
- #ifdef RDOC_NEVER_DEFINED
887
- mLibXML = rb_define_module("LibXML");
888
- mXML = rb_define_module_under(mLibXML, "XML");
889
- #endif
890
-
891
- void ruby_init_xml_reader(void)
892
- {
893
- base_uri_SYMBOL = ID2SYM(rb_intern("base_uri"));
894
- ENCODING_SYMBOL = ID2SYM(rb_intern("encoding"));
895
- OPTIONS_SYMBOL = ID2SYM(rb_intern("options"));
896
-
897
- cXMLReader = rb_define_class_under(mXML, "Reader", rb_cObject);
898
-
899
- rb_define_singleton_method(cXMLReader, "document", rxml_reader_document, 1);
900
- rb_define_singleton_method(cXMLReader, "file", rxml_reader_file, -1);
901
- rb_define_singleton_method(cXMLReader, "io", rxml_reader_io, -1);
902
- rb_define_singleton_method(cXMLReader, "string", rxml_reader_string, -1);
903
-
904
- rb_define_method(cXMLReader, "[]", rxml_reader_attribute, 1);
905
- rb_define_method(cXMLReader, "attribute_count", rxml_reader_attr_count, 0);
906
- rb_define_method(cXMLReader, "base_uri", rxml_reader_base_uri, 0);
907
- #if LIBXML_VERSION >= 20618
908
- rb_define_method(cXMLReader, "byte_consumed", rxml_reader_byte_consumed, 0);
909
- #endif
910
- rb_define_method(cXMLReader, "close", rxml_reader_close, 0);
911
- #if LIBXML_VERSION >= 20617
912
- rb_define_method(cXMLReader, "column_number", rxml_reader_column_number, 0);
913
- #endif
914
- rb_define_method(cXMLReader, "depth", rxml_reader_depth, 0);
915
- rb_define_method(cXMLReader, "encoding", rxml_reader_encoding, 0);
916
- rb_define_method(cXMLReader, "expand", rxml_reader_expand, 0);
917
- rb_define_method(cXMLReader, "has_attributes?", rxml_reader_has_attributes, 0);
918
- rb_define_method(cXMLReader, "has_value?", rxml_reader_has_value, 0);
919
- #if LIBXML_VERSION >= 20617
920
- rb_define_method(cXMLReader, "line_number", rxml_reader_line_number, 0);
921
- #endif
922
- rb_define_method(cXMLReader, "local_name", rxml_reader_local_name, 0);
923
- rb_define_method(cXMLReader, "lookup_namespace", rxml_reader_lookup_namespace, 1);
924
- rb_define_method(cXMLReader, "move_to_attribute", rxml_reader_move_to_attr, 1);
925
- rb_define_method(cXMLReader, "move_to_first_attribute", rxml_reader_move_to_first_attr, 0);
926
- rb_define_method(cXMLReader, "move_to_next_attribute", rxml_reader_move_to_next_attr, 0);
927
- rb_define_method(cXMLReader, "move_to_element", rxml_reader_move_to_element, 0);
928
- rb_define_method(cXMLReader, "name", rxml_reader_name, 0);
929
- rb_define_method(cXMLReader, "namespace_uri", rxml_reader_namespace_uri, 0);
930
- rb_define_method(cXMLReader, "next", rxml_reader_next, 0);
931
- rb_define_method(cXMLReader, "next_sibling", rxml_reader_next_sibling, 0);
932
- rb_define_method(cXMLReader, "node", rxml_reader_node, 0);
933
- rb_define_method(cXMLReader, "node_type", rxml_reader_node_type, 0);
934
- rb_define_method(cXMLReader, "normalization", rxml_reader_normalization, 0);
935
- rb_define_method(cXMLReader, "prefix", rxml_reader_prefix, 0);
936
- rb_define_method(cXMLReader, "quote_char", rxml_reader_quote_char, 0);
937
- rb_define_method(cXMLReader, "read", rxml_reader_read, 0);
938
- rb_define_method(cXMLReader, "read_attribute_value", rxml_reader_read_attr_value, 0);
939
- rb_define_method(cXMLReader, "read_inner_xml", rxml_reader_read_inner_xml, 0);
940
- rb_define_method(cXMLReader, "read_outer_xml", rxml_reader_read_outer_xml, 0);
941
- rb_define_method(cXMLReader, "read_state", rxml_reader_read_state, 0);
942
- rb_define_method(cXMLReader, "read_string", rxml_reader_read_string, 0);
943
- rb_define_method(cXMLReader, "relax_ng_validate", rxml_reader_relax_ng_validate, 1);
944
- rb_define_method(cXMLReader, "standalone", rxml_reader_standalone, 0);
945
- #if LIBXML_VERSION >= 20620
946
- rb_define_method(cXMLReader, "schema_validate", rxml_reader_schema_validate, 1);
947
- #endif
948
- rb_define_method(cXMLReader, "value", rxml_reader_value, 0);
949
- rb_define_method(cXMLReader, "xml_lang", rxml_reader_xml_lang, 0);
950
- rb_define_method(cXMLReader, "xml_version", rxml_reader_xml_version, 0);
951
- rb_define_method(cXMLReader, "default?", rxml_reader_default, 0);
952
- rb_define_method(cXMLReader, "empty_element?", rxml_reader_empty_element, 0);
953
- rb_define_method(cXMLReader, "namespace_declaration?", rxml_reader_namespace_declaration, 0);
954
- rb_define_method(cXMLReader, "valid?", rxml_reader_valid, 0);
955
-
956
- /* Constants */
957
- rb_define_const(cXMLReader, "LOADDTD", INT2FIX(XML_PARSER_LOADDTD));
958
- rb_define_const(cXMLReader, "DEFAULTATTRS", INT2FIX(XML_PARSER_DEFAULTATTRS));
959
- rb_define_const(cXMLReader, "VALIDATE", INT2FIX(XML_PARSER_VALIDATE));
960
- rb_define_const(cXMLReader, "SUBST_ENTITIES", INT2FIX(XML_PARSER_SUBST_ENTITIES));
961
-
962
- rb_define_const(cXMLReader, "SEVERITY_VALIDITY_WARNING", INT2FIX(XML_PARSER_SEVERITY_VALIDITY_WARNING));
963
- rb_define_const(cXMLReader, "SEVERITY_VALIDITY_ERROR", INT2FIX(XML_PARSER_SEVERITY_VALIDITY_ERROR));
964
- rb_define_const(cXMLReader, "SEVERITY_WARNING", INT2FIX(XML_PARSER_SEVERITY_WARNING));
965
- rb_define_const(cXMLReader, "SEVERITY_ERROR", INT2FIX(XML_PARSER_SEVERITY_ERROR));
966
-
967
- rb_define_const(cXMLReader, "TYPE_NONE", INT2FIX(XML_READER_TYPE_NONE));
968
- rb_define_const(cXMLReader, "TYPE_ELEMENT", INT2FIX(XML_READER_TYPE_ELEMENT));
969
- rb_define_const(cXMLReader, "TYPE_ATTRIBUTE", INT2FIX(XML_READER_TYPE_ATTRIBUTE));
970
- rb_define_const(cXMLReader, "TYPE_TEXT", INT2FIX(XML_READER_TYPE_TEXT));
971
- rb_define_const(cXMLReader, "TYPE_CDATA", INT2FIX(XML_READER_TYPE_CDATA));
972
- rb_define_const(cXMLReader, "TYPE_ENTITY_REFERENCE", INT2FIX(XML_READER_TYPE_ENTITY_REFERENCE));
973
- rb_define_const(cXMLReader, "TYPE_ENTITY", INT2FIX(XML_READER_TYPE_ENTITY));
974
- rb_define_const(cXMLReader, "TYPE_PROCESSING_INSTRUCTION", INT2FIX(XML_READER_TYPE_PROCESSING_INSTRUCTION));
975
- rb_define_const(cXMLReader, "TYPE_COMMENT", INT2FIX(XML_READER_TYPE_COMMENT));
976
- rb_define_const(cXMLReader, "TYPE_DOCUMENT", INT2FIX(XML_READER_TYPE_DOCUMENT));
977
- rb_define_const(cXMLReader, "TYPE_DOCUMENT_TYPE", INT2FIX(XML_READER_TYPE_DOCUMENT_TYPE));
978
- rb_define_const(cXMLReader, "TYPE_DOCUMENT_FRAGMENT", INT2FIX(XML_READER_TYPE_DOCUMENT_FRAGMENT));
979
- rb_define_const(cXMLReader, "TYPE_NOTATION", INT2FIX(XML_READER_TYPE_NOTATION));
980
- rb_define_const(cXMLReader, "TYPE_WHITESPACE", INT2FIX(XML_READER_TYPE_WHITESPACE));
981
- rb_define_const(cXMLReader, "TYPE_SIGNIFICANT_WHITESPACE", INT2FIX(XML_READER_TYPE_SIGNIFICANT_WHITESPACE));
982
- rb_define_const(cXMLReader, "TYPE_END_ELEMENT", INT2FIX(XML_READER_TYPE_END_ELEMENT));
983
- rb_define_const(cXMLReader, "TYPE_END_ENTITY", INT2FIX(XML_READER_TYPE_END_ENTITY));
984
- rb_define_const(cXMLReader, "TYPE_XML_DECLARATION", INT2FIX(XML_READER_TYPE_XML_DECLARATION));
985
-
986
- /* Read states */
987
- rb_define_const(cXMLReader, "MODE_INITIAL", INT2FIX(XML_TEXTREADER_MODE_INITIAL));
988
- rb_define_const(cXMLReader, "MODE_INTERACTIVE", INT2FIX(XML_TEXTREADER_MODE_INTERACTIVE));
989
- rb_define_const(cXMLReader, "MODE_ERROR", INT2FIX(XML_TEXTREADER_MODE_ERROR));
990
- rb_define_const(cXMLReader, "MODE_EOF", INT2FIX(XML_TEXTREADER_MODE_EOF));
991
- rb_define_const(cXMLReader, "MODE_CLOSED", INT2FIX(XML_TEXTREADER_MODE_CLOSED));
992
- rb_define_const(cXMLReader, "MODE_READING", INT2FIX(XML_TEXTREADER_MODE_READING));
993
- }
1
+ /* Copyright (c) 2006-2007 Apple Inc.
2
+ * Please see the LICENSE file for copyright and distribution information. */
3
+
4
+ #include "ruby_libxml.h"
5
+ #include "ruby_xml_reader.h"
6
+
7
+ /*
8
+ * Document-class: LibXML::XML::Reader
9
+ *
10
+ * The XML::Reader class provides a simpler, alternative way of parsing an XML
11
+ * document in contrast to XML::Parser or XML::SaxParser. A XML::Reader instance
12
+ * acts like a cursor going forward in a document stream, stopping at each node
13
+ * it encounters. To advance to the next node, simply cadd XML::Reader#read.
14
+ *
15
+ * The XML::Reader API closely matches the DOM Core specification and supports
16
+ * namespaces, xml:base, entity handling and DTDs.
17
+ *
18
+ * To summarize, XML::Reader provides a far simpler API to use versus XML::SaxParser
19
+ * and is more memory efficient than using XML::Parser to create a DOM tree.
20
+ *
21
+ * Example:
22
+ *
23
+ * parser = XML::Reader.string("<foo><bar>1</bar><bar>2</bar><bar>3</bar></foo>")
24
+ * reader.read
25
+ * assert_equal('foo', reader.name)
26
+ * assert_equal(nil, reader.value)
27
+ *
28
+ * 3.times do |i|
29
+ * reader.read
30
+ * assert_equal(XML::Reader::TYPE_ELEMENT, reader.node_type)
31
+ * assert_equal('bar', reader.name)
32
+ * reader.read
33
+ * assert_equal(XML::Reader::TYPE_TEXT, reader.node_type)
34
+ * assert_equal((i + 1).to_s, reader.value)
35
+ * reader.read
36
+ * assert_equal(XML::Reader::TYPE_END_ELEMENT, reader.node_type)
37
+ * end
38
+ *
39
+ * You can also parse documents (see XML::Reader.document),
40
+ * strings (see XML::Parser.string) and io objects (see
41
+ * XML::Parser.io).
42
+ *
43
+ * For a more in depth tutorial, albeit in C, see http://xmlsoft.org/xmlreader.html.*/
44
+
45
+ VALUE cXMLReader;
46
+
47
+ ID BASE_URI_SYMBOL;
48
+ ID ENCODING_SYMBOL;
49
+ ID IO_ATTR;
50
+ ID OPTIONS_SYMBOL;
51
+
52
+
53
+ static void rxml_reader_free(xmlTextReaderPtr reader)
54
+ {
55
+ xmlFreeTextReader(reader);
56
+ }
57
+
58
+ static VALUE rxml_reader_wrap(xmlTextReaderPtr reader)
59
+ {
60
+ return Data_Wrap_Struct(cXMLReader, NULL, rxml_reader_free, reader);
61
+ }
62
+
63
+ static xmlTextReaderPtr rxml_text_reader_get(VALUE obj)
64
+ {
65
+ xmlTextReaderPtr xreader;
66
+ Data_Get_Struct(obj, xmlTextReader, xreader);
67
+ return xreader;
68
+ }
69
+
70
+ /*
71
+ * call-seq:
72
+ * XML::Reader.document(doc) -> XML::Reader
73
+ *
74
+ * Create an new reader for the specified document.
75
+ */
76
+ VALUE rxml_reader_document(VALUE klass, VALUE doc)
77
+ {
78
+ xmlDocPtr xdoc;
79
+ xmlTextReaderPtr xreader;
80
+
81
+ Data_Get_Struct(doc, xmlDoc, xdoc);
82
+
83
+ xreader = xmlReaderWalker(xdoc);
84
+
85
+ if (xreader == NULL)
86
+ rxml_raise(&xmlLastError);
87
+
88
+ return rxml_reader_wrap(xreader);
89
+ }
90
+
91
+ /* call-seq:
92
+ * XML::Reader.file(path) -> XML::Reader
93
+ * XML::Reader.file(path, :encoding => XML::Encoding::UTF_8,
94
+ * :options => XML::Parser::Options::NOENT) -> XML::Parser
95
+ *
96
+ * Creates a new reader by parsing the specified file or uri.
97
+ *
98
+ * You may provide an optional hash table to control how the
99
+ * parsing is performed. Valid options are:
100
+ *
101
+ * encoding - The document encoding, defaults to nil. Valid values
102
+ * are the encoding constants defined on XML::Encoding.
103
+ * options - Controls the execution of the parser, defaults to 0.
104
+ * Valid values are the constants defined on
105
+ * XML::Parser::Options. Mutliple options can be combined
106
+ * by using Bitwise OR (|).
107
+ */
108
+ static VALUE rxml_reader_file(int argc, VALUE *argv, VALUE klass)
109
+ {
110
+ xmlTextReaderPtr xreader;
111
+ VALUE path;
112
+ VALUE options;
113
+
114
+ const char *xencoding = NULL;
115
+ int xoptions = 0;
116
+
117
+ rb_scan_args(argc, argv, "11", &path, &options);
118
+ Check_Type(path, T_STRING);
119
+
120
+ if (!NIL_P(options))
121
+ {
122
+ VALUE encoding = Qnil;
123
+ VALUE parserOptions = Qnil;
124
+
125
+ Check_Type(options, T_HASH);
126
+
127
+ encoding = rb_hash_aref(options, BASE_URI_SYMBOL);
128
+ xencoding = NIL_P(encoding) ? NULL : xmlGetCharEncodingName(NUM2INT(encoding));
129
+
130
+ parserOptions = rb_hash_aref(options, OPTIONS_SYMBOL);
131
+ xoptions = NIL_P(parserOptions) ? 0 : NUM2INT(parserOptions);
132
+ }
133
+
134
+ xreader = xmlReaderForFile(StringValueCStr(path), xencoding, xoptions);
135
+
136
+ if (xreader == NULL)
137
+ rxml_raise(&xmlLastError);
138
+
139
+ return rxml_reader_wrap(xreader);
140
+ }
141
+
142
+ /* call-seq:
143
+ * XML::Reader.io(io) -> XML::Reader
144
+ * XML::Reader.io(io, :encoding => XML::Encoding::UTF_8,
145
+ * :options => XML::Parser::Options::NOENT) -> XML::Parser
146
+ *
147
+ * Creates a new reader by parsing the specified io object.
148
+ *
149
+ * You may provide an optional hash table to control how the
150
+ * parsing is performed. Valid options are:
151
+ *
152
+ * base_uri - The base url for the parsed document.
153
+ * encoding - The document encoding, defaults to nil. Valid values
154
+ * are the encoding constants defined on XML::Encoding.
155
+ * options - Controls the execution of the parser, defaults to 0.
156
+ * Valid values are the constants defined on
157
+ * XML::Parser::Options. Mutliple options can be combined
158
+ * by using Bitwise OR (|).
159
+ */
160
+ static VALUE rxml_reader_io(int argc, VALUE *argv, VALUE klass)
161
+ {
162
+ xmlTextReaderPtr xreader;
163
+ VALUE result;
164
+ VALUE io;
165
+ VALUE options;
166
+ char *xbaseurl = NULL;
167
+ const char *xencoding = NULL;
168
+ int xoptions = 0;
169
+
170
+ rb_scan_args(argc, argv, "11", &io, &options);
171
+
172
+ if (!NIL_P(options))
173
+ {
174
+ VALUE baseurl = Qnil;
175
+ VALUE encoding = Qnil;
176
+ VALUE parserOptions = Qnil;
177
+
178
+ Check_Type(options, T_HASH);
179
+
180
+ baseurl = rb_hash_aref(options, BASE_URI_SYMBOL);
181
+ xbaseurl = NIL_P(baseurl) ? NULL : StringValueCStr(baseurl);
182
+
183
+ encoding = rb_hash_aref(options, ENCODING_SYMBOL);
184
+ xencoding = NIL_P(encoding) ? NULL : xmlGetCharEncodingName(NUM2INT(encoding));
185
+
186
+ parserOptions = rb_hash_aref(options, OPTIONS_SYMBOL);
187
+ xoptions = NIL_P(parserOptions) ? 0 : NUM2INT(parserOptions);
188
+ }
189
+
190
+ xreader = xmlReaderForIO((xmlInputReadCallback) rxml_read_callback, NULL,
191
+ (void *) io,
192
+ xbaseurl, xencoding, xoptions);
193
+
194
+ if (xreader == NULL)
195
+ rxml_raise(&xmlLastError);
196
+
197
+ result = rxml_reader_wrap(xreader);
198
+
199
+ /* Attach io object to parser so it won't get freed.*/
200
+ rb_ivar_set(result, IO_ATTR, io);
201
+
202
+ return result;
203
+ }
204
+
205
+ /* call-seq:
206
+ * XML::Reader.string(io) -> XML::Reader
207
+ * XML::Reader.string(io, :encoding => XML::Encoding::UTF_8,
208
+ * :options => XML::Parser::Options::NOENT) -> XML::Parser
209
+ *
210
+ * Creates a new reader by parsing the specified string.
211
+ *
212
+ * You may provide an optional hash table to control how the
213
+ * parsing is performed. Valid options are:
214
+ *
215
+ * base_uri - The base url for the parsed document.
216
+ * encoding - The document encoding, defaults to nil. Valid values
217
+ * are the encoding constants defined on XML::Encoding.
218
+ * options - Controls the execution of the parser, defaults to 0.
219
+ * Valid values are the constants defined on
220
+ * XML::Parser::Options. Mutliple options can be combined
221
+ * by using Bitwise OR (|).
222
+ */
223
+ static VALUE rxml_reader_string(int argc, VALUE *argv, VALUE klass)
224
+ {
225
+ xmlTextReaderPtr xreader;
226
+ VALUE string;
227
+ VALUE options;
228
+ char *xbaseurl = NULL;
229
+ const char *xencoding = NULL;
230
+ int xoptions = 0;
231
+
232
+ rb_scan_args(argc, argv, "11", &string, &options);
233
+ Check_Type(string, T_STRING);
234
+
235
+ if (!NIL_P(options))
236
+ {
237
+ VALUE baseurl = Qnil;
238
+ VALUE encoding = Qnil;
239
+ VALUE parserOptions = Qnil;
240
+
241
+ Check_Type(options, T_HASH);
242
+
243
+ baseurl = rb_hash_aref(options, BASE_URI_SYMBOL);
244
+ xbaseurl = NIL_P(baseurl) ? NULL : StringValueCStr(baseurl);
245
+
246
+ encoding = rb_hash_aref(options, ENCODING_SYMBOL);
247
+ xencoding = NIL_P(encoding) ? NULL : xmlGetCharEncodingName(NUM2INT(encoding));
248
+
249
+ parserOptions = rb_hash_aref(options, OPTIONS_SYMBOL);
250
+ xoptions = NIL_P(parserOptions) ? 0 : NUM2INT(parserOptions);
251
+ }
252
+
253
+ xreader = xmlReaderForMemory(StringValueCStr(string), RSTRING_LEN(string),
254
+ xbaseurl, xencoding, xoptions);
255
+
256
+ if (xreader == NULL)
257
+ rxml_raise(&xmlLastError);
258
+
259
+ return rxml_reader_wrap(xreader);
260
+ }
261
+
262
+ /*
263
+ * call-seq:
264
+ * reader.close -> code
265
+ *
266
+ * This method releases any resources allocated by the current instance
267
+ * changes the state to Closed and close any underlying input.
268
+ */
269
+ static VALUE rxml_reader_close(VALUE self)
270
+ {
271
+ return INT2FIX(xmlTextReaderClose(rxml_text_reader_get(self)));
272
+ }
273
+
274
+ /*
275
+ * call-seq:
276
+ * reader.move_to_attribute(val) -> code
277
+ *
278
+ * Move the position of the current instance to the attribute with the
279
+ * specified index (if +val+ is an integer) or name (if +val+ is a string)
280
+ * relative to the containing element.
281
+ */
282
+ static VALUE rxml_reader_move_to_attr(VALUE self, VALUE val)
283
+ {
284
+ xmlTextReaderPtr xreader;
285
+ int ret;
286
+
287
+ xreader = rxml_text_reader_get(self);
288
+
289
+ if (TYPE(val) == T_FIXNUM)
290
+ {
291
+ ret = xmlTextReaderMoveToAttributeNo(xreader, FIX2INT(val));
292
+ }
293
+ else
294
+ {
295
+ ret = xmlTextReaderMoveToAttribute(xreader,
296
+ (const xmlChar *) StringValueCStr(val));
297
+ }
298
+
299
+ return INT2FIX(ret);
300
+ }
301
+
302
+ /*
303
+ * call-seq:
304
+ * reader.move_to_first_attribute -> code
305
+ *
306
+ * Move the position of the current instance to the first attribute associated
307
+ * with the current node.
308
+ */
309
+ static VALUE rxml_reader_move_to_first_attr(VALUE self)
310
+ {
311
+ return INT2FIX(xmlTextReaderMoveToFirstAttribute(rxml_text_reader_get(self)));
312
+ }
313
+
314
+ /*
315
+ * call-seq:
316
+ * reader.move_to_next_attribute -> code
317
+ *
318
+ * Move the position of the current instance to the next attribute associated
319
+ * with the current node.
320
+ */
321
+ static VALUE rxml_reader_move_to_next_attr(VALUE self)
322
+ {
323
+ return INT2FIX(xmlTextReaderMoveToNextAttribute(rxml_text_reader_get(self)));
324
+ }
325
+
326
+ /*
327
+ * call-seq:
328
+ * reader.move_to_element -> code
329
+ *
330
+ * Move the position of the current instance to the node that contains the
331
+ * current attribute node.
332
+ */
333
+ static VALUE rxml_reader_move_to_element(VALUE self)
334
+ {
335
+ return INT2FIX(xmlTextReaderMoveToElement(rxml_text_reader_get(self)));
336
+ }
337
+
338
+ /*
339
+ * call-seq:
340
+ * reader.next -> code
341
+ *
342
+ * Skip to the node following the current one in document order while avoiding
343
+ * the subtree if any.
344
+ */
345
+ static VALUE rxml_reader_next(VALUE self)
346
+ {
347
+ return INT2FIX(xmlTextReaderNext(rxml_text_reader_get(self)));
348
+ }
349
+
350
+ /*
351
+ * call-seq:
352
+ * reader.next_sibling -> code
353
+ *
354
+ * Skip to the node following the current one in document order while avoiding
355
+ * the subtree if any. Currently implemented only for Readers built on a
356
+ * document.
357
+ */
358
+ static VALUE rxml_reader_next_sibling(VALUE self)
359
+ {
360
+ return INT2FIX(xmlTextReaderNextSibling(rxml_text_reader_get(self)));
361
+ }
362
+
363
+ /*
364
+ * call-seq:
365
+ * reader.node -> XML::Node
366
+ *
367
+ * Returns the reader's current node. It will return
368
+ * nil if Reader#read has not yet been called.
369
+ * WARNING - Using this method is dangerous because the
370
+ * the node may be destroyed on the next #read.
371
+ */
372
+ static VALUE rxml_reader_node(VALUE self)
373
+ {
374
+ xmlTextReaderPtr xreader = rxml_text_reader_get(self);
375
+ xmlNodePtr xnode = xmlTextReaderCurrentNode(xreader);
376
+ return xnode ? rxml_node_wrap(xnode) : Qnil;
377
+ }
378
+
379
+ /*
380
+ * call-seq:
381
+ * reader.node_type -> type
382
+ *
383
+ * Get the node type of the current node. Reference:
384
+ * http://dotgnu.org/pnetlib-doc/System/Xml/XmlNodeType.html
385
+ */
386
+ static VALUE rxml_reader_node_type(VALUE self)
387
+ {
388
+ return INT2FIX(xmlTextReaderNodeType(rxml_text_reader_get(self)));
389
+ }
390
+
391
+ /*
392
+ * call-seq:
393
+ * reader.normalization -> value
394
+ *
395
+ * The value indicating whether to normalize white space and attribute values.
396
+ * Since attribute value and end of line normalizations are a MUST in the XML
397
+ * specification only the value true is accepted. The broken bahaviour of
398
+ * accepting out of range character entities like &#0; is of course not
399
+ * supported either.
400
+ *
401
+ * Return 1 or -1 in case of error.
402
+ */
403
+ static VALUE rxml_reader_normalization(VALUE self)
404
+ {
405
+ return INT2FIX(xmlTextReaderNormalization(rxml_text_reader_get(self)));
406
+ }
407
+
408
+ /*
409
+ * call-seq:
410
+ * reader.read -> code
411
+ *
412
+ * Causes the reader to move to the next node in the stream, exposing its properties.
413
+ *
414
+ * Returns true if a node was successfully read or false if there are no more
415
+ * nodes to read. On errors, an exception is raised.*/
416
+ static VALUE rxml_reader_read(VALUE self)
417
+ {
418
+ int result = xmlTextReaderRead(rxml_text_reader_get(self));
419
+ switch(result)
420
+ {
421
+ case -1:
422
+ rxml_raise(&xmlLastError);
423
+ return Qnil;
424
+ break;
425
+ case 0:
426
+ return Qfalse;
427
+ case 1:
428
+ return Qtrue;
429
+ default:
430
+ rb_raise(rb_eRuntimeError,
431
+ "xmlTextReaderRead did not return -1, 0 or 1. Return value was: %d", result);
432
+ }
433
+ }
434
+
435
+ /*
436
+ * call-seq:
437
+ * reader.read_attribute_value -> code
438
+ *
439
+ * Parse an attribute value into one or more Text and EntityReference nodes.
440
+ *
441
+ * Return 1 in case of success, 0 if the reader was not positionned on an
442
+ * attribute node or all the attribute values have been read, or -1 in case of
443
+ * error.
444
+ */
445
+ static VALUE rxml_reader_read_attr_value(VALUE self)
446
+ {
447
+ return INT2FIX(xmlTextReaderReadAttributeValue(rxml_text_reader_get(self)));
448
+ }
449
+
450
+ /*
451
+ * call-seq:
452
+ * reader.read_inner_xml -> data
453
+ *
454
+ * Read the contents of the current node, including child nodes and markup.
455
+ *
456
+ * Return a string containing the XML content, or nil if the current node is
457
+ * neither an element nor attribute, or has no child nodes.
458
+ */
459
+ static VALUE rxml_reader_read_inner_xml(VALUE self)
460
+ {
461
+ const xmlChar *result = xmlTextReaderReadInnerXml(rxml_text_reader_get(self));
462
+ return (result == NULL ? Qnil : rb_str_new2((const char*)result));
463
+ }
464
+
465
+ /*
466
+ * call-seq:
467
+ * reader.read_outer_xml -> data
468
+ *
469
+ * Read the contents of the current node, including child nodes and markup.
470
+ *
471
+ * Return a string containing the XML content, or nil if the current node is
472
+ * neither an element nor attribute, or has no child nodes.
473
+ */
474
+ static VALUE rxml_reader_read_outer_xml(VALUE self)
475
+ {
476
+ const xmlChar *result = xmlTextReaderReadOuterXml(rxml_text_reader_get(self));
477
+ return (result == NULL ? Qnil : rb_str_new2((const char*)result));
478
+ }
479
+
480
+ /*
481
+ * call-seq:
482
+ * reader.read_state -> state
483
+ *
484
+ * Get the read state of the reader.
485
+ */
486
+ static VALUE rxml_reader_read_state(VALUE self)
487
+ {
488
+ return INT2FIX(xmlTextReaderReadState(rxml_text_reader_get(self)));
489
+ }
490
+
491
+ /*
492
+ * call-seq:
493
+ * reader.read_string -> string
494
+ *
495
+ * Read the contents of an element or a text node as a string.
496
+ *
497
+ * Return a string containing the contents of the Element or Text node, or nil
498
+ * if the reader is positioned on any other type of node.
499
+ */
500
+ static VALUE rxml_reader_read_string(VALUE self)
501
+ {
502
+ const xmlChar *result = xmlTextReaderReadString(rxml_text_reader_get(self));
503
+ return (result == NULL ? Qnil : rb_str_new2((const char*)result));
504
+ }
505
+
506
+ /*
507
+ * call-seq:
508
+ * reader.relax_ng_validate(rng) -> code
509
+ *
510
+ * Use RelaxNG to validate the document as it is processed. Activation is only
511
+ * possible before the first read. If +rng+ is nil, the RelaxNG validation is
512
+ * desactivated.
513
+ *
514
+ * Return 0 in case the RelaxNG validation could be (des)activated and -1 in
515
+ * case of error.
516
+ */
517
+ static VALUE rxml_reader_relax_ng_validate(VALUE self, VALUE rng)
518
+ {
519
+ char *xrng = NIL_P(rng) ? NULL : StringValueCStr(rng);
520
+ return INT2FIX(xmlTextReaderRelaxNGValidate(rxml_text_reader_get(self), xrng));
521
+ }
522
+
523
+ #if LIBXML_VERSION >= 20620
524
+ /*
525
+ * call-seq:
526
+ * reader.schema_validate(schema) -> code
527
+ *
528
+ * Use W3C XSD schema to validate the document as it is processed. Activation
529
+ * is only possible before the first read. If +schema+ is nil, then XML Schema
530
+ * validation is desactivated.
531
+ *
532
+ * Return 0 in case the schemas validation could be (de)activated and -1 in
533
+ * case of error.
534
+ */
535
+ static VALUE
536
+ rxml_reader_schema_validate(VALUE self, VALUE xsd)
537
+ {
538
+ char *xxsd = NIL_P(xsd) ? NULL : StringValueCStr(xsd);
539
+ int status = xmlTextReaderSchemaValidate(rxml_text_reader_get(self), xxsd);
540
+ return INT2FIX(status);
541
+ }
542
+ #endif
543
+
544
+ /*
545
+ * call-seq:
546
+ * reader.name -> name
547
+ *
548
+ * Return the qualified name of the node.
549
+ */
550
+ static VALUE rxml_reader_name(VALUE self)
551
+ {
552
+ const xmlChar *result = xmlTextReaderConstName(rxml_text_reader_get(self));
553
+ return (result == NULL ? Qnil : rb_str_new2((const char*)result));
554
+ }
555
+
556
+ /*
557
+ * call-seq:
558
+ * reader.local_name -> name
559
+ *
560
+ * Return the local name of the node.
561
+ */
562
+ static VALUE rxml_reader_local_name(VALUE self)
563
+ {
564
+ const xmlChar *result = xmlTextReaderConstLocalName(rxml_text_reader_get(self));
565
+ return (result == NULL ? Qnil : rb_str_new2((const char*)result));
566
+ }
567
+
568
+ /*
569
+ * call-seq:
570
+ * reader.attribute_count -> count
571
+ *
572
+ * Provide the number of attributes of the current node.
573
+ */
574
+ static VALUE rxml_reader_attr_count(VALUE self)
575
+ {
576
+ return INT2FIX(xmlTextReaderAttributeCount(rxml_text_reader_get(self)));
577
+ }
578
+
579
+ /*
580
+ * call-seq:
581
+ * reader.encoding -> XML::Encoding::UTF_8
582
+ *
583
+ * Returns the encoding of the document being read. Note you
584
+ * first have to read data from the reader for encoding
585
+ * to return a value
586
+ *
587
+ * reader = XML::Reader.file(XML_FILE)
588
+ * assert_nil(reader.encoding)
589
+ * reader.read
590
+ * assert_equal(XML::Encoding::UTF_8, reader.encoding)
591
+ *
592
+ * In addition, libxml always appears to return nil for the encoding
593
+ * when parsing strings.
594
+ */
595
+ static VALUE rxml_reader_encoding(VALUE self)
596
+ {
597
+ xmlTextReaderPtr xreader = rxml_text_reader_get(self);
598
+ const xmlChar *xencoding = xmlTextReaderConstEncoding(xreader);
599
+ if (xencoding)
600
+ return INT2NUM(xmlParseCharEncoding(xencoding));
601
+ else
602
+ return INT2NUM(XML_CHAR_ENCODING_NONE);
603
+ }
604
+
605
+ /*
606
+ * call-seq:
607
+ * reader.base_uri -> URI
608
+ *
609
+ * Determine the base URI of the node.
610
+ */
611
+ static VALUE rxml_reader_base_uri(VALUE self)
612
+ {
613
+ const xmlChar *result = xmlTextReaderConstBaseUri(rxml_text_reader_get(self));
614
+ return (result == NULL ? Qnil : rb_str_new2((const char*)result));
615
+ }
616
+
617
+ /*
618
+ * call-seq:
619
+ * reader.namespace_uri -> URI
620
+ *
621
+ * Determine the namespace URI of the node.
622
+ */
623
+ static VALUE rxml_reader_namespace_uri(VALUE self)
624
+ {
625
+ const xmlChar *result = xmlTextReaderConstNamespaceUri(rxml_text_reader_get(self));
626
+ return (result == NULL ? Qnil : rb_str_new2((const char*)result));
627
+ }
628
+
629
+ /*
630
+ * call-seq:
631
+ * reader.value -> text
632
+ *
633
+ * Provide the text value of the node if present.
634
+ */
635
+ static VALUE rxml_reader_value(VALUE self)
636
+ {
637
+ const xmlChar *result = xmlTextReaderConstValue(rxml_text_reader_get(self));
638
+ return (result == NULL ? Qnil : rb_str_new2((const char*)result));
639
+ }
640
+
641
+ /*
642
+ * call-seq:
643
+ * reader.prefix -> prefix
644
+ *
645
+ * Get a shorthand reference to the namespace associated with the node.
646
+ */
647
+ static VALUE rxml_reader_prefix(VALUE self)
648
+ {
649
+ const xmlChar *result = xmlTextReaderConstPrefix(rxml_text_reader_get(self));
650
+ return (result == NULL ? Qnil : rb_str_new2((const char*)result));
651
+ }
652
+
653
+ /*
654
+ * call-seq:
655
+ * reader.depth -> depth
656
+ *
657
+ * Get the depth of the node in the tree.
658
+ */
659
+ static VALUE rxml_reader_depth(VALUE self)
660
+ {
661
+ return INT2FIX(xmlTextReaderDepth(rxml_text_reader_get(self)));
662
+ }
663
+
664
+ /*
665
+ * call-seq:
666
+ * reader.quote_char -> char
667
+ *
668
+ * Get the quotation mark character used to enclose the value of an attribute,
669
+ * as an integer value (and -1 in case of error).
670
+ */
671
+ static VALUE rxml_reader_quote_char(VALUE self)
672
+ {
673
+ return INT2FIX(xmlTextReaderQuoteChar(rxml_text_reader_get(self)));
674
+ }
675
+
676
+ /*
677
+ * call-seq:
678
+ * reader.standalone -> code
679
+ *
680
+ * Determine the standalone status of the document being read.
681
+ *
682
+ * Return 1 if the document was declared to be standalone, 0 if it was
683
+ * declared to be not standalone, or -1 if the document did not specify its
684
+ * standalone status or in case of error.
685
+ */
686
+ static VALUE rxml_reader_standalone(VALUE self)
687
+ {
688
+ return INT2FIX(xmlTextReaderStandalone(rxml_text_reader_get(self)));
689
+ }
690
+
691
+ /*
692
+ * call-seq:
693
+ * reader.xml_lang -> value
694
+ *
695
+ * Get the xml:lang scope within which the node resides.
696
+ */
697
+ static VALUE rxml_reader_xml_lang(VALUE self)
698
+ {
699
+ const xmlChar *result = xmlTextReaderConstXmlLang(rxml_text_reader_get(self));
700
+ return (result == NULL ? Qnil : rb_str_new2((const char*)result));
701
+ }
702
+
703
+ /*
704
+ * call-seq:
705
+ * reader.xml_version -> version
706
+ *
707
+ * Determine the XML version of the document being read.
708
+ */
709
+ static VALUE rxml_reader_xml_version(VALUE self)
710
+ {
711
+ const xmlChar *result = xmlTextReaderConstXmlVersion(rxml_text_reader_get(self));
712
+ return (result == NULL ? Qnil : rb_str_new2((const char*)result));
713
+ }
714
+
715
+ /*
716
+ * call-seq:
717
+ * reader.has_attributes? -> bool
718
+ *
719
+ * Get whether the node has attributes.
720
+ */
721
+ static VALUE rxml_reader_has_attributes(VALUE self)
722
+ {
723
+ return xmlTextReaderHasAttributes(rxml_text_reader_get(self)) ? Qtrue
724
+ : Qfalse;
725
+ }
726
+
727
+ /*
728
+ * call-seq:
729
+ * reader.has_value? -> bool
730
+ *
731
+ * Get whether the node can have a text value.
732
+ */
733
+ static VALUE rxml_reader_has_value(VALUE self)
734
+ {
735
+ return xmlTextReaderHasValue(rxml_text_reader_get(self)) ? Qtrue : Qfalse;
736
+ }
737
+
738
+ /*
739
+ * call-seq:
740
+ * reader[key] -> value
741
+ *
742
+ * Provide the value of the attribute with the specified index (if +key+ is an
743
+ * integer) or with the specified name (if +key+ is a string) relative to the
744
+ * containing element, as a string.
745
+ */
746
+ static VALUE rxml_reader_attribute(VALUE self, VALUE key)
747
+ {
748
+ xmlTextReaderPtr reader;
749
+ xmlChar *attr;
750
+
751
+ reader = rxml_text_reader_get(self);
752
+
753
+ if (TYPE(key) == T_FIXNUM)
754
+ {
755
+ attr = xmlTextReaderGetAttributeNo(reader, FIX2INT(key));
756
+ }
757
+ else
758
+ {
759
+ attr = xmlTextReaderGetAttribute(reader, (const xmlChar *) StringValueCStr(key));
760
+ }
761
+ return (attr == NULL ? Qnil : rb_str_new2((const char*)attr));
762
+ }
763
+
764
+ /*
765
+ * call-seq:
766
+ * reader.lookup_namespace(prefix) -> value
767
+ *
768
+ * Resolve a namespace prefix in the scope of the current element.
769
+ * To return the default namespace, specify nil as +prefix+.
770
+ */
771
+ static VALUE rxml_reader_lookup_namespace(VALUE self, VALUE prefix)
772
+ {
773
+ const xmlChar *result = xmlTextReaderLookupNamespace(rxml_text_reader_get(
774
+ self), (const xmlChar *) StringValueCStr(prefix));
775
+ return (result == NULL ? Qnil : rb_str_new2((const char*)result));
776
+ }
777
+
778
+ /*
779
+ * call-seq:
780
+ * reader.expand -> node
781
+ *
782
+ * Read the contents of the current node and the full subtree. It then makes
783
+ * the subtree available until the next read call.
784
+ *
785
+ * Return an XML::Node object, or nil in case of error.
786
+ */
787
+ static VALUE rxml_reader_expand(VALUE self)
788
+ {
789
+ xmlNodePtr node;
790
+ xmlDocPtr doc;
791
+ xmlTextReaderPtr reader = rxml_text_reader_get(self);
792
+ node = xmlTextReaderExpand(reader);
793
+
794
+ if (!node)
795
+ return Qnil;
796
+
797
+ /* Okay this is tricky. By accessing the returned node, we
798
+ take ownership of the reader's document. Thus we need to
799
+ tell the reader to not free it. Otherwise it will be
800
+ freed twice - once when the Ruby document wrapper goes
801
+ out of scope and once when the reader goes out of scope. */
802
+
803
+ xmlTextReaderPreserve(reader);
804
+ doc = xmlTextReaderCurrentDoc(reader);
805
+ rxml_document_wrap(doc);
806
+
807
+ return rxml_node_wrap(node);
808
+ }
809
+
810
+ #if LIBXML_VERSION >= 20618
811
+ /*
812
+ * call-seq:
813
+ * reader.byte_consumed -> value
814
+ *
815
+ * This method provides the current index of the parser used by the reader,
816
+ * relative to the start of the current entity.
817
+ */
818
+ static VALUE
819
+ rxml_reader_byte_consumed(VALUE self)
820
+ {
821
+ return INT2NUM(xmlTextReaderByteConsumed(rxml_text_reader_get(self)));
822
+ }
823
+ #endif
824
+
825
+ #if LIBXML_VERSION >= 20617
826
+ /*
827
+ * call-seq:
828
+ * reader.column_number -> number
829
+ *
830
+ * Provide the column number of the current parsing point.
831
+ */
832
+ static VALUE
833
+ rxml_reader_column_number(VALUE self)
834
+ {
835
+ return INT2NUM(xmlTextReaderGetParserColumnNumber(rxml_text_reader_get(self)));
836
+ }
837
+
838
+ /*
839
+ * call-seq:
840
+ * reader.line_number -> number
841
+ *
842
+ * Provide the line number of the current parsing point.
843
+ */
844
+ static VALUE
845
+ rxml_reader_line_number(VALUE self)
846
+ {
847
+ return INT2NUM(xmlTextReaderGetParserLineNumber(rxml_text_reader_get(self)));
848
+ }
849
+ #endif
850
+
851
+ /*
852
+ * call-seq:
853
+ * reader.default? -> bool
854
+ *
855
+ * Return whether an Attribute node was generated from the default value
856
+ * defined in the DTD or schema.
857
+ */
858
+ static VALUE rxml_reader_default(VALUE self)
859
+ {
860
+ return xmlTextReaderIsDefault(rxml_text_reader_get(self)) ? Qtrue : Qfalse;
861
+ }
862
+
863
+ /*
864
+ * call-seq:
865
+ * reader.namespace_declaration? -> bool
866
+ *
867
+ * Determine whether the current node is a namespace declaration rather than a
868
+ * regular attribute.
869
+ */
870
+ static VALUE rxml_reader_namespace_declaration(VALUE self)
871
+ {
872
+ return xmlTextReaderIsNamespaceDecl(rxml_text_reader_get(self)) ? Qtrue
873
+ : Qfalse;
874
+ }
875
+
876
+ /*
877
+ * call-seq:
878
+ * reader.empty_element? -> bool
879
+ *
880
+ * Check if the current node is empty.
881
+ */
882
+ static VALUE rxml_reader_empty_element(VALUE self)
883
+ {
884
+ return xmlTextReaderIsEmptyElement(rxml_text_reader_get(self)) ? Qtrue
885
+ : Qfalse;
886
+ }
887
+
888
+ /*
889
+ * call-seq:
890
+ * reader.valid? -> bool
891
+ *
892
+ * Retrieve the validity status from the parser context.
893
+ */
894
+ static VALUE rxml_reader_valid(VALUE self)
895
+ {
896
+ return xmlTextReaderIsValid(rxml_text_reader_get(self)) ? Qtrue : Qfalse;
897
+ }
898
+
899
+ void rxml_init_reader(void)
900
+ {
901
+ BASE_URI_SYMBOL = ID2SYM(rb_intern("base_uri"));
902
+ ENCODING_SYMBOL = ID2SYM(rb_intern("encoding"));
903
+ IO_ATTR = ID2SYM(rb_intern("@io"));
904
+ OPTIONS_SYMBOL = ID2SYM(rb_intern("options"));
905
+
906
+ cXMLReader = rb_define_class_under(mXML, "Reader", rb_cObject);
907
+
908
+ rb_define_singleton_method(cXMLReader, "document", rxml_reader_document, 1);
909
+ rb_define_singleton_method(cXMLReader, "file", rxml_reader_file, -1);
910
+ rb_define_singleton_method(cXMLReader, "io", rxml_reader_io, -1);
911
+ rb_define_singleton_method(cXMLReader, "string", rxml_reader_string, -1);
912
+
913
+ rb_define_method(cXMLReader, "[]", rxml_reader_attribute, 1);
914
+ rb_define_method(cXMLReader, "attribute_count", rxml_reader_attr_count, 0);
915
+ rb_define_method(cXMLReader, "base_uri", rxml_reader_base_uri, 0);
916
+ #if LIBXML_VERSION >= 20618
917
+ rb_define_method(cXMLReader, "byte_consumed", rxml_reader_byte_consumed, 0);
918
+ #endif
919
+ rb_define_method(cXMLReader, "close", rxml_reader_close, 0);
920
+ #if LIBXML_VERSION >= 20617
921
+ rb_define_method(cXMLReader, "column_number", rxml_reader_column_number, 0);
922
+ #endif
923
+ rb_define_method(cXMLReader, "depth", rxml_reader_depth, 0);
924
+ rb_define_method(cXMLReader, "encoding", rxml_reader_encoding, 0);
925
+ rb_define_method(cXMLReader, "expand", rxml_reader_expand, 0);
926
+ rb_define_method(cXMLReader, "has_attributes?", rxml_reader_has_attributes, 0);
927
+ rb_define_method(cXMLReader, "has_value?", rxml_reader_has_value, 0);
928
+ #if LIBXML_VERSION >= 20617
929
+ rb_define_method(cXMLReader, "line_number", rxml_reader_line_number, 0);
930
+ #endif
931
+ rb_define_method(cXMLReader, "local_name", rxml_reader_local_name, 0);
932
+ rb_define_method(cXMLReader, "lookup_namespace", rxml_reader_lookup_namespace, 1);
933
+ rb_define_method(cXMLReader, "move_to_attribute", rxml_reader_move_to_attr, 1);
934
+ rb_define_method(cXMLReader, "move_to_first_attribute", rxml_reader_move_to_first_attr, 0);
935
+ rb_define_method(cXMLReader, "move_to_next_attribute", rxml_reader_move_to_next_attr, 0);
936
+ rb_define_method(cXMLReader, "move_to_element", rxml_reader_move_to_element, 0);
937
+ rb_define_method(cXMLReader, "name", rxml_reader_name, 0);
938
+ rb_define_method(cXMLReader, "namespace_uri", rxml_reader_namespace_uri, 0);
939
+ rb_define_method(cXMLReader, "next", rxml_reader_next, 0);
940
+ rb_define_method(cXMLReader, "next_sibling", rxml_reader_next_sibling, 0);
941
+ rb_define_method(cXMLReader, "node", rxml_reader_node, 0);
942
+ rb_define_method(cXMLReader, "node_type", rxml_reader_node_type, 0);
943
+ rb_define_method(cXMLReader, "normalization", rxml_reader_normalization, 0);
944
+ rb_define_method(cXMLReader, "prefix", rxml_reader_prefix, 0);
945
+ rb_define_method(cXMLReader, "quote_char", rxml_reader_quote_char, 0);
946
+ rb_define_method(cXMLReader, "read", rxml_reader_read, 0);
947
+ rb_define_method(cXMLReader, "read_attribute_value", rxml_reader_read_attr_value, 0);
948
+ rb_define_method(cXMLReader, "read_inner_xml", rxml_reader_read_inner_xml, 0);
949
+ rb_define_method(cXMLReader, "read_outer_xml", rxml_reader_read_outer_xml, 0);
950
+ rb_define_method(cXMLReader, "read_state", rxml_reader_read_state, 0);
951
+ rb_define_method(cXMLReader, "read_string", rxml_reader_read_string, 0);
952
+ rb_define_method(cXMLReader, "relax_ng_validate", rxml_reader_relax_ng_validate, 1);
953
+ rb_define_method(cXMLReader, "standalone", rxml_reader_standalone, 0);
954
+ #if LIBXML_VERSION >= 20620
955
+ rb_define_method(cXMLReader, "schema_validate", rxml_reader_schema_validate, 1);
956
+ #endif
957
+ rb_define_method(cXMLReader, "value", rxml_reader_value, 0);
958
+ rb_define_method(cXMLReader, "xml_lang", rxml_reader_xml_lang, 0);
959
+ rb_define_method(cXMLReader, "xml_version", rxml_reader_xml_version, 0);
960
+ rb_define_method(cXMLReader, "default?", rxml_reader_default, 0);
961
+ rb_define_method(cXMLReader, "empty_element?", rxml_reader_empty_element, 0);
962
+ rb_define_method(cXMLReader, "namespace_declaration?", rxml_reader_namespace_declaration, 0);
963
+ rb_define_method(cXMLReader, "valid?", rxml_reader_valid, 0);
964
+
965
+ /* Constants */
966
+ rb_define_const(cXMLReader, "LOADDTD", INT2FIX(XML_PARSER_LOADDTD));
967
+ rb_define_const(cXMLReader, "DEFAULTATTRS", INT2FIX(XML_PARSER_DEFAULTATTRS));
968
+ rb_define_const(cXMLReader, "VALIDATE", INT2FIX(XML_PARSER_VALIDATE));
969
+ rb_define_const(cXMLReader, "SUBST_ENTITIES", INT2FIX(XML_PARSER_SUBST_ENTITIES));
970
+
971
+ rb_define_const(cXMLReader, "SEVERITY_VALIDITY_WARNING", INT2FIX(XML_PARSER_SEVERITY_VALIDITY_WARNING));
972
+ rb_define_const(cXMLReader, "SEVERITY_VALIDITY_ERROR", INT2FIX(XML_PARSER_SEVERITY_VALIDITY_ERROR));
973
+ rb_define_const(cXMLReader, "SEVERITY_WARNING", INT2FIX(XML_PARSER_SEVERITY_WARNING));
974
+ rb_define_const(cXMLReader, "SEVERITY_ERROR", INT2FIX(XML_PARSER_SEVERITY_ERROR));
975
+
976
+ rb_define_const(cXMLReader, "TYPE_NONE", INT2FIX(XML_READER_TYPE_NONE));
977
+ rb_define_const(cXMLReader, "TYPE_ELEMENT", INT2FIX(XML_READER_TYPE_ELEMENT));
978
+ rb_define_const(cXMLReader, "TYPE_ATTRIBUTE", INT2FIX(XML_READER_TYPE_ATTRIBUTE));
979
+ rb_define_const(cXMLReader, "TYPE_TEXT", INT2FIX(XML_READER_TYPE_TEXT));
980
+ rb_define_const(cXMLReader, "TYPE_CDATA", INT2FIX(XML_READER_TYPE_CDATA));
981
+ rb_define_const(cXMLReader, "TYPE_ENTITY_REFERENCE", INT2FIX(XML_READER_TYPE_ENTITY_REFERENCE));
982
+ rb_define_const(cXMLReader, "TYPE_ENTITY", INT2FIX(XML_READER_TYPE_ENTITY));
983
+ rb_define_const(cXMLReader, "TYPE_PROCESSING_INSTRUCTION", INT2FIX(XML_READER_TYPE_PROCESSING_INSTRUCTION));
984
+ rb_define_const(cXMLReader, "TYPE_COMMENT", INT2FIX(XML_READER_TYPE_COMMENT));
985
+ rb_define_const(cXMLReader, "TYPE_DOCUMENT", INT2FIX(XML_READER_TYPE_DOCUMENT));
986
+ rb_define_const(cXMLReader, "TYPE_DOCUMENT_TYPE", INT2FIX(XML_READER_TYPE_DOCUMENT_TYPE));
987
+ rb_define_const(cXMLReader, "TYPE_DOCUMENT_FRAGMENT", INT2FIX(XML_READER_TYPE_DOCUMENT_FRAGMENT));
988
+ rb_define_const(cXMLReader, "TYPE_NOTATION", INT2FIX(XML_READER_TYPE_NOTATION));
989
+ rb_define_const(cXMLReader, "TYPE_WHITESPACE", INT2FIX(XML_READER_TYPE_WHITESPACE));
990
+ rb_define_const(cXMLReader, "TYPE_SIGNIFICANT_WHITESPACE", INT2FIX(XML_READER_TYPE_SIGNIFICANT_WHITESPACE));
991
+ rb_define_const(cXMLReader, "TYPE_END_ELEMENT", INT2FIX(XML_READER_TYPE_END_ELEMENT));
992
+ rb_define_const(cXMLReader, "TYPE_END_ENTITY", INT2FIX(XML_READER_TYPE_END_ENTITY));
993
+ rb_define_const(cXMLReader, "TYPE_XML_DECLARATION", INT2FIX(XML_READER_TYPE_XML_DECLARATION));
994
+
995
+ /* Read states */
996
+ rb_define_const(cXMLReader, "MODE_INITIAL", INT2FIX(XML_TEXTREADER_MODE_INITIAL));
997
+ rb_define_const(cXMLReader, "MODE_INTERACTIVE", INT2FIX(XML_TEXTREADER_MODE_INTERACTIVE));
998
+ rb_define_const(cXMLReader, "MODE_ERROR", INT2FIX(XML_TEXTREADER_MODE_ERROR));
999
+ rb_define_const(cXMLReader, "MODE_EOF", INT2FIX(XML_TEXTREADER_MODE_EOF));
1000
+ rb_define_const(cXMLReader, "MODE_CLOSED", INT2FIX(XML_TEXTREADER_MODE_CLOSED));
1001
+ rb_define_const(cXMLReader, "MODE_READING", INT2FIX(XML_TEXTREADER_MODE_READING));
1002
+ }