rubyjedi-nokogiri_java 1.4.0.20100513161003-java
Sign up to get free protection for your applications and to get access to all the features.
- data/.autotest +26 -0
- data/CHANGELOG.ja.rdoc +330 -0
- data/CHANGELOG.rdoc +341 -0
- data/Manifest.txt +277 -0
- data/README.ja.rdoc +105 -0
- data/README.rdoc +125 -0
- data/Rakefile +307 -0
- data/bin/nokogiri +49 -0
- data/deps.rip +5 -0
- data/ext/nokogiri/extconf.rb +149 -0
- data/ext/nokogiri/html_document.c +145 -0
- data/ext/nokogiri/html_document.h +10 -0
- data/ext/nokogiri/html_element_description.c +272 -0
- data/ext/nokogiri/html_element_description.h +10 -0
- data/ext/nokogiri/html_entity_lookup.c +32 -0
- data/ext/nokogiri/html_entity_lookup.h +8 -0
- data/ext/nokogiri/html_sax_parser_context.c +92 -0
- data/ext/nokogiri/html_sax_parser_context.h +11 -0
- data/ext/nokogiri/nokogiri.c +96 -0
- data/ext/nokogiri/nokogiri.h +148 -0
- data/ext/nokogiri/xml_attr.c +92 -0
- data/ext/nokogiri/xml_attr.h +9 -0
- data/ext/nokogiri/xml_attribute_decl.c +67 -0
- data/ext/nokogiri/xml_attribute_decl.h +9 -0
- data/ext/nokogiri/xml_cdata.c +54 -0
- data/ext/nokogiri/xml_cdata.h +9 -0
- data/ext/nokogiri/xml_comment.c +52 -0
- data/ext/nokogiri/xml_comment.h +9 -0
- data/ext/nokogiri/xml_document.c +386 -0
- data/ext/nokogiri/xml_document.h +24 -0
- data/ext/nokogiri/xml_document_fragment.c +46 -0
- data/ext/nokogiri/xml_document_fragment.h +10 -0
- data/ext/nokogiri/xml_dtd.c +192 -0
- data/ext/nokogiri/xml_dtd.h +10 -0
- data/ext/nokogiri/xml_element_content.c +123 -0
- data/ext/nokogiri/xml_element_content.h +10 -0
- data/ext/nokogiri/xml_element_decl.c +69 -0
- data/ext/nokogiri/xml_element_decl.h +9 -0
- data/ext/nokogiri/xml_encoding_handler.c +79 -0
- data/ext/nokogiri/xml_encoding_handler.h +8 -0
- data/ext/nokogiri/xml_entity_decl.c +97 -0
- data/ext/nokogiri/xml_entity_decl.h +10 -0
- data/ext/nokogiri/xml_entity_reference.c +50 -0
- data/ext/nokogiri/xml_entity_reference.h +9 -0
- data/ext/nokogiri/xml_io.c +31 -0
- data/ext/nokogiri/xml_io.h +11 -0
- data/ext/nokogiri/xml_namespace.c +82 -0
- data/ext/nokogiri/xml_namespace.h +13 -0
- data/ext/nokogiri/xml_node.c +1080 -0
- data/ext/nokogiri/xml_node.h +13 -0
- data/ext/nokogiri/xml_node_set.c +405 -0
- data/ext/nokogiri/xml_node_set.h +9 -0
- data/ext/nokogiri/xml_processing_instruction.c +54 -0
- data/ext/nokogiri/xml_processing_instruction.h +9 -0
- data/ext/nokogiri/xml_reader.c +593 -0
- data/ext/nokogiri/xml_reader.h +10 -0
- data/ext/nokogiri/xml_relax_ng.c +159 -0
- data/ext/nokogiri/xml_relax_ng.h +9 -0
- data/ext/nokogiri/xml_sax_parser.c +283 -0
- data/ext/nokogiri/xml_sax_parser.h +43 -0
- data/ext/nokogiri/xml_sax_parser_context.c +157 -0
- data/ext/nokogiri/xml_sax_parser_context.h +10 -0
- data/ext/nokogiri/xml_sax_push_parser.c +114 -0
- data/ext/nokogiri/xml_sax_push_parser.h +9 -0
- data/ext/nokogiri/xml_schema.c +156 -0
- data/ext/nokogiri/xml_schema.h +9 -0
- data/ext/nokogiri/xml_syntax_error.c +52 -0
- data/ext/nokogiri/xml_syntax_error.h +13 -0
- data/ext/nokogiri/xml_text.c +48 -0
- data/ext/nokogiri/xml_text.h +9 -0
- data/ext/nokogiri/xml_xpath.c +53 -0
- data/ext/nokogiri/xml_xpath.h +11 -0
- data/ext/nokogiri/xml_xpath_context.c +239 -0
- data/ext/nokogiri/xml_xpath_context.h +9 -0
- data/ext/nokogiri/xslt_stylesheet.c +131 -0
- data/ext/nokogiri/xslt_stylesheet.h +9 -0
- data/lib/isorelax.jar +0 -0
- data/lib/jing.jar +0 -0
- data/lib/nekodtd.jar +0 -0
- data/lib/nekohtml.jar +0 -0
- data/lib/nokogiri.rb +123 -0
- data/lib/nokogiri/css.rb +25 -0
- data/lib/nokogiri/css/generated_parser.rb +659 -0
- data/lib/nokogiri/css/generated_tokenizer.rb +145 -0
- data/lib/nokogiri/css/node.rb +99 -0
- data/lib/nokogiri/css/parser.rb +82 -0
- data/lib/nokogiri/css/parser.y +230 -0
- data/lib/nokogiri/css/syntax_error.rb +7 -0
- data/lib/nokogiri/css/tokenizer.rb +7 -0
- data/lib/nokogiri/css/tokenizer.rex +55 -0
- data/lib/nokogiri/css/xpath_visitor.rb +164 -0
- data/lib/nokogiri/decorators/slop.rb +33 -0
- data/lib/nokogiri/ffi/encoding_handler.rb +42 -0
- data/lib/nokogiri/ffi/html/document.rb +28 -0
- data/lib/nokogiri/ffi/html/element_description.rb +81 -0
- data/lib/nokogiri/ffi/html/entity_lookup.rb +16 -0
- data/lib/nokogiri/ffi/html/sax/parser_context.rb +38 -0
- data/lib/nokogiri/ffi/io_callbacks.rb +42 -0
- data/lib/nokogiri/ffi/libxml.rb +372 -0
- data/lib/nokogiri/ffi/structs/common_node.rb +26 -0
- data/lib/nokogiri/ffi/structs/html_elem_desc.rb +24 -0
- data/lib/nokogiri/ffi/structs/html_entity_desc.rb +13 -0
- data/lib/nokogiri/ffi/structs/xml_alloc.rb +16 -0
- data/lib/nokogiri/ffi/structs/xml_attr.rb +19 -0
- data/lib/nokogiri/ffi/structs/xml_attribute.rb +27 -0
- data/lib/nokogiri/ffi/structs/xml_buffer.rb +16 -0
- data/lib/nokogiri/ffi/structs/xml_char_encoding_handler.rb +11 -0
- data/lib/nokogiri/ffi/structs/xml_document.rb +108 -0
- data/lib/nokogiri/ffi/structs/xml_dtd.rb +28 -0
- data/lib/nokogiri/ffi/structs/xml_element.rb +26 -0
- data/lib/nokogiri/ffi/structs/xml_element_content.rb +17 -0
- data/lib/nokogiri/ffi/structs/xml_entity.rb +32 -0
- data/lib/nokogiri/ffi/structs/xml_enumeration.rb +12 -0
- data/lib/nokogiri/ffi/structs/xml_node.rb +28 -0
- data/lib/nokogiri/ffi/structs/xml_node_set.rb +53 -0
- data/lib/nokogiri/ffi/structs/xml_notation.rb +11 -0
- data/lib/nokogiri/ffi/structs/xml_ns.rb +15 -0
- data/lib/nokogiri/ffi/structs/xml_parser_context.rb +19 -0
- data/lib/nokogiri/ffi/structs/xml_relax_ng.rb +14 -0
- data/lib/nokogiri/ffi/structs/xml_sax_handler.rb +51 -0
- data/lib/nokogiri/ffi/structs/xml_sax_push_parser_context.rb +124 -0
- data/lib/nokogiri/ffi/structs/xml_schema.rb +13 -0
- data/lib/nokogiri/ffi/structs/xml_syntax_error.rb +31 -0
- data/lib/nokogiri/ffi/structs/xml_text_reader.rb +12 -0
- data/lib/nokogiri/ffi/structs/xml_xpath_context.rb +37 -0
- data/lib/nokogiri/ffi/structs/xml_xpath_object.rb +35 -0
- data/lib/nokogiri/ffi/structs/xml_xpath_parser_context.rb +20 -0
- data/lib/nokogiri/ffi/structs/xslt_stylesheet.rb +13 -0
- data/lib/nokogiri/ffi/xml/attr.rb +41 -0
- data/lib/nokogiri/ffi/xml/attribute_decl.rb +27 -0
- data/lib/nokogiri/ffi/xml/cdata.rb +19 -0
- data/lib/nokogiri/ffi/xml/comment.rb +18 -0
- data/lib/nokogiri/ffi/xml/document.rb +135 -0
- data/lib/nokogiri/ffi/xml/document_fragment.rb +21 -0
- data/lib/nokogiri/ffi/xml/dtd.rb +67 -0
- data/lib/nokogiri/ffi/xml/element_content.rb +43 -0
- data/lib/nokogiri/ffi/xml/element_decl.rb +19 -0
- data/lib/nokogiri/ffi/xml/entity_decl.rb +27 -0
- data/lib/nokogiri/ffi/xml/entity_reference.rb +19 -0
- data/lib/nokogiri/ffi/xml/namespace.rb +44 -0
- data/lib/nokogiri/ffi/xml/node.rb +465 -0
- data/lib/nokogiri/ffi/xml/node_set.rb +146 -0
- data/lib/nokogiri/ffi/xml/processing_instruction.rb +20 -0
- data/lib/nokogiri/ffi/xml/reader.rb +227 -0
- data/lib/nokogiri/ffi/xml/relax_ng.rb +85 -0
- data/lib/nokogiri/ffi/xml/sax/parser.rb +135 -0
- data/lib/nokogiri/ffi/xml/sax/parser_context.rb +67 -0
- data/lib/nokogiri/ffi/xml/sax/push_parser.rb +55 -0
- data/lib/nokogiri/ffi/xml/schema.rb +92 -0
- data/lib/nokogiri/ffi/xml/syntax_error.rb +98 -0
- data/lib/nokogiri/ffi/xml/text.rb +18 -0
- data/lib/nokogiri/ffi/xml/xpath.rb +19 -0
- data/lib/nokogiri/ffi/xml/xpath_context.rb +135 -0
- data/lib/nokogiri/ffi/xslt/stylesheet.rb +50 -0
- data/lib/nokogiri/html.rb +36 -0
- data/lib/nokogiri/html/builder.rb +35 -0
- data/lib/nokogiri/html/document.rb +88 -0
- data/lib/nokogiri/html/document_fragment.rb +15 -0
- data/lib/nokogiri/html/element_description.rb +23 -0
- data/lib/nokogiri/html/element_description_defaults.rb +671 -0
- data/lib/nokogiri/html/entity_lookup.rb +13 -0
- data/lib/nokogiri/html/sax/parser.rb +48 -0
- data/lib/nokogiri/html/sax/parser_context.rb +16 -0
- data/lib/nokogiri/nokogiri.jar +0 -0
- data/lib/nokogiri/syntax_error.rb +4 -0
- data/lib/nokogiri/version.rb +33 -0
- data/lib/nokogiri/version_warning.rb +11 -0
- data/lib/nokogiri/xml.rb +67 -0
- data/lib/nokogiri/xml/attr.rb +14 -0
- data/lib/nokogiri/xml/attribute_decl.rb +18 -0
- data/lib/nokogiri/xml/builder.rb +405 -0
- data/lib/nokogiri/xml/cdata.rb +11 -0
- data/lib/nokogiri/xml/character_data.rb +7 -0
- data/lib/nokogiri/xml/document.rb +163 -0
- data/lib/nokogiri/xml/document_fragment.rb +73 -0
- data/lib/nokogiri/xml/dtd.rb +11 -0
- data/lib/nokogiri/xml/element_content.rb +36 -0
- data/lib/nokogiri/xml/element_decl.rb +13 -0
- data/lib/nokogiri/xml/entity_decl.rb +15 -0
- data/lib/nokogiri/xml/fragment_handler.rb +73 -0
- data/lib/nokogiri/xml/namespace.rb +13 -0
- data/lib/nokogiri/xml/node.rb +730 -0
- data/lib/nokogiri/xml/node/save_options.rb +42 -0
- data/lib/nokogiri/xml/node_set.rb +318 -0
- data/lib/nokogiri/xml/notation.rb +6 -0
- data/lib/nokogiri/xml/parse_options.rb +85 -0
- data/lib/nokogiri/xml/pp.rb +2 -0
- data/lib/nokogiri/xml/pp/character_data.rb +18 -0
- data/lib/nokogiri/xml/pp/node.rb +56 -0
- data/lib/nokogiri/xml/processing_instruction.rb +8 -0
- data/lib/nokogiri/xml/reader.rb +74 -0
- data/lib/nokogiri/xml/relax_ng.rb +32 -0
- data/lib/nokogiri/xml/sax.rb +4 -0
- data/lib/nokogiri/xml/sax/document.rb +160 -0
- data/lib/nokogiri/xml/sax/parser.rb +115 -0
- data/lib/nokogiri/xml/sax/parser_context.rb +16 -0
- data/lib/nokogiri/xml/sax/push_parser.rb +60 -0
- data/lib/nokogiri/xml/schema.rb +61 -0
- data/lib/nokogiri/xml/syntax_error.rb +43 -0
- data/lib/nokogiri/xml/xpath.rb +10 -0
- data/lib/nokogiri/xml/xpath/syntax_error.rb +8 -0
- data/lib/nokogiri/xml/xpath_context.rb +16 -0
- data/lib/nokogiri/xslt.rb +48 -0
- data/lib/nokogiri/xslt/stylesheet.rb +25 -0
- data/lib/xercesImpl.jar +0 -0
- data/lib/xsd/xmlparser/nokogiri.rb +90 -0
- data/tasks/test.rb +100 -0
- data/test/css/test_nthiness.rb +159 -0
- data/test/css/test_parser.rb +282 -0
- data/test/css/test_tokenizer.rb +190 -0
- data/test/css/test_xpath_visitor.rb +76 -0
- data/test/ffi/test_document.rb +35 -0
- data/test/files/2ch.html +108 -0
- data/test/files/address_book.rlx +12 -0
- data/test/files/address_book.xml +10 -0
- data/test/files/bar/bar.xsd +4 -0
- data/test/files/dont_hurt_em_why.xml +422 -0
- data/test/files/exslt.xml +8 -0
- data/test/files/exslt.xslt +35 -0
- data/test/files/foo/foo.xsd +4 -0
- data/test/files/po.xml +32 -0
- data/test/files/po.xsd +66 -0
- data/test/files/shift_jis.html +10 -0
- data/test/files/shift_jis.xml +5 -0
- data/test/files/snuggles.xml +3 -0
- data/test/files/staff.dtd +10 -0
- data/test/files/staff.xml +59 -0
- data/test/files/staff.xslt +32 -0
- data/test/files/tlm.html +850 -0
- data/test/files/valid_bar.xml +2 -0
- data/test/helper.rb +137 -0
- data/test/html/sax/test_parser.rb +83 -0
- data/test/html/sax/test_parser_context.rb +48 -0
- data/test/html/test_builder.rb +164 -0
- data/test/html/test_document.rb +385 -0
- data/test/html/test_document_encoding.rb +77 -0
- data/test/html/test_document_fragment.rb +157 -0
- data/test/html/test_element_description.rb +98 -0
- data/test/html/test_named_characters.rb +14 -0
- data/test/html/test_node.rb +242 -0
- data/test/html/test_node_encoding.rb +27 -0
- data/test/test_convert_xpath.rb +135 -0
- data/test/test_css_cache.rb +45 -0
- data/test/test_encoding_handler.rb +46 -0
- data/test/test_jruby.rb +40 -0
- data/test/test_memory_leak.rb +87 -0
- data/test/test_nokogiri.rb +140 -0
- data/test/test_reader.rb +358 -0
- data/test/test_soap4r_sax.rb +52 -0
- data/test/test_xslt_transforms.rb +150 -0
- data/test/xml/node/test_save_options.rb +20 -0
- data/test/xml/node/test_subclass.rb +44 -0
- data/test/xml/sax/test_parser.rb +314 -0
- data/test/xml/sax/test_parser_context.rb +63 -0
- data/test/xml/sax/test_push_parser.rb +135 -0
- data/test/xml/test_attr.rb +38 -0
- data/test/xml/test_attribute_decl.rb +90 -0
- data/test/xml/test_builder.rb +167 -0
- data/test/xml/test_cdata.rb +38 -0
- data/test/xml/test_comment.rb +29 -0
- data/test/xml/test_document.rb +638 -0
- data/test/xml/test_document_encoding.rb +26 -0
- data/test/xml/test_document_fragment.rb +149 -0
- data/test/xml/test_dtd.rb +92 -0
- data/test/xml/test_dtd_encoding.rb +33 -0
- data/test/xml/test_element_content.rb +56 -0
- data/test/xml/test_element_decl.rb +73 -0
- data/test/xml/test_entity_decl.rb +83 -0
- data/test/xml/test_entity_reference.rb +21 -0
- data/test/xml/test_namespace.rb +70 -0
- data/test/xml/test_node.rb +740 -0
- data/test/xml/test_node_attributes.rb +34 -0
- data/test/xml/test_node_encoding.rb +107 -0
- data/test/xml/test_node_reparenting.rb +279 -0
- data/test/xml/test_node_set.rb +577 -0
- data/test/xml/test_parse_options.rb +52 -0
- data/test/xml/test_processing_instruction.rb +30 -0
- data/test/xml/test_reader_encoding.rb +126 -0
- data/test/xml/test_relax_ng.rb +60 -0
- data/test/xml/test_schema.rb +89 -0
- data/test/xml/test_syntax_error.rb +12 -0
- data/test/xml/test_text.rb +30 -0
- data/test/xml/test_unparented_node.rb +381 -0
- data/test/xml/test_xpath.rb +169 -0
- metadata +477 -0
@@ -0,0 +1,67 @@
|
|
1
|
+
#include <xml_attribute_decl.h>
|
2
|
+
|
3
|
+
/*
|
4
|
+
* call-seq:
|
5
|
+
* attribute_type
|
6
|
+
*
|
7
|
+
* The attribute_type for this AttributeDecl
|
8
|
+
*/
|
9
|
+
static VALUE attribute_type(VALUE self)
|
10
|
+
{
|
11
|
+
xmlAttributePtr node;
|
12
|
+
Data_Get_Struct(self, xmlAttribute, node);
|
13
|
+
return INT2NUM((long)node->atype);
|
14
|
+
}
|
15
|
+
|
16
|
+
/*
|
17
|
+
* call-seq:
|
18
|
+
* default
|
19
|
+
*
|
20
|
+
* The default value
|
21
|
+
*/
|
22
|
+
static VALUE default_value(VALUE self)
|
23
|
+
{
|
24
|
+
xmlAttributePtr node;
|
25
|
+
Data_Get_Struct(self, xmlAttribute, node);
|
26
|
+
|
27
|
+
if(node->defaultValue) return NOKOGIRI_STR_NEW2(node->defaultValue);
|
28
|
+
return Qnil;
|
29
|
+
}
|
30
|
+
|
31
|
+
/*
|
32
|
+
* call-seq:
|
33
|
+
* enumeration
|
34
|
+
*
|
35
|
+
* An enumeration of possible values
|
36
|
+
*/
|
37
|
+
static VALUE enumeration(VALUE self)
|
38
|
+
{
|
39
|
+
xmlAttributePtr node;
|
40
|
+
Data_Get_Struct(self, xmlAttribute, node);
|
41
|
+
|
42
|
+
VALUE list = rb_ary_new();
|
43
|
+
xmlEnumerationPtr enm = node->tree;
|
44
|
+
|
45
|
+
while(enm) {
|
46
|
+
rb_ary_push(list, NOKOGIRI_STR_NEW2(enm->name));
|
47
|
+
enm = enm->next;
|
48
|
+
}
|
49
|
+
|
50
|
+
return list;
|
51
|
+
}
|
52
|
+
|
53
|
+
VALUE cNokogiriXmlAttributeDecl;
|
54
|
+
|
55
|
+
void init_xml_attribute_decl()
|
56
|
+
{
|
57
|
+
VALUE nokogiri = rb_define_module("Nokogiri");
|
58
|
+
VALUE xml = rb_define_module_under(nokogiri, "XML");
|
59
|
+
VALUE node = rb_define_class_under(xml, "Node", rb_cObject);
|
60
|
+
VALUE klass = rb_define_class_under(xml, "AttributeDecl", node);
|
61
|
+
|
62
|
+
cNokogiriXmlAttributeDecl = klass;
|
63
|
+
|
64
|
+
rb_define_method(klass, "attribute_type", attribute_type, 0);
|
65
|
+
rb_define_method(klass, "default", default_value, 0);
|
66
|
+
rb_define_method(klass, "enumeration", enumeration, 0);
|
67
|
+
}
|
@@ -0,0 +1,54 @@
|
|
1
|
+
#include <xml_cdata.h>
|
2
|
+
|
3
|
+
/*
|
4
|
+
* call-seq:
|
5
|
+
* new(document, content)
|
6
|
+
*
|
7
|
+
* Create a new CDATA element on the +document+ with +content+
|
8
|
+
*/
|
9
|
+
static VALUE new(int argc, VALUE *argv, VALUE klass)
|
10
|
+
{
|
11
|
+
xmlDocPtr xml_doc;
|
12
|
+
VALUE doc;
|
13
|
+
VALUE content;
|
14
|
+
VALUE rest;
|
15
|
+
|
16
|
+
rb_scan_args(argc, argv, "2*", &doc, &content, &rest);
|
17
|
+
|
18
|
+
Data_Get_Struct(doc, xmlDoc, xml_doc);
|
19
|
+
|
20
|
+
xmlNodePtr node = xmlNewCDataBlock(
|
21
|
+
xml_doc->doc,
|
22
|
+
NIL_P(content) ? NULL : (const xmlChar *)StringValuePtr(content),
|
23
|
+
NIL_P(content) ? 0 : (int)RSTRING_LEN(content)
|
24
|
+
);
|
25
|
+
|
26
|
+
NOKOGIRI_ROOT_NODE(node);
|
27
|
+
|
28
|
+
VALUE rb_node = Nokogiri_wrap_xml_node(klass, node);
|
29
|
+
rb_obj_call_init(rb_node, argc, argv);
|
30
|
+
|
31
|
+
if(rb_block_given_p()) rb_yield(rb_node);
|
32
|
+
|
33
|
+
return rb_node;
|
34
|
+
}
|
35
|
+
|
36
|
+
VALUE cNokogiriXmlCData;
|
37
|
+
void init_xml_cdata()
|
38
|
+
{
|
39
|
+
VALUE nokogiri = rb_define_module("Nokogiri");
|
40
|
+
VALUE xml = rb_define_module_under(nokogiri, "XML");
|
41
|
+
VALUE node = rb_define_class_under(xml, "Node", rb_cObject);
|
42
|
+
VALUE char_data = rb_define_class_under(xml, "CharacterData", node);
|
43
|
+
VALUE text = rb_define_class_under(xml, "Text", char_data);
|
44
|
+
|
45
|
+
/*
|
46
|
+
* CData represents a CData node in an xml document.
|
47
|
+
*/
|
48
|
+
VALUE klass = rb_define_class_under(xml, "CDATA", text);
|
49
|
+
|
50
|
+
|
51
|
+
cNokogiriXmlCData = klass;
|
52
|
+
|
53
|
+
rb_define_singleton_method(klass, "new", new, -1);
|
54
|
+
}
|
@@ -0,0 +1,52 @@
|
|
1
|
+
#include <xml_comment.h>
|
2
|
+
|
3
|
+
/*
|
4
|
+
* call-seq:
|
5
|
+
* new(document, content)
|
6
|
+
*
|
7
|
+
* Create a new Comment element on the +document+ with +content+
|
8
|
+
*/
|
9
|
+
static VALUE new(int argc, VALUE *argv, VALUE klass)
|
10
|
+
{
|
11
|
+
xmlDocPtr xml_doc;
|
12
|
+
VALUE document;
|
13
|
+
VALUE content;
|
14
|
+
VALUE rest;
|
15
|
+
|
16
|
+
rb_scan_args(argc, argv, "2*", &document, &content, &rest);
|
17
|
+
|
18
|
+
Data_Get_Struct(document, xmlDoc, xml_doc);
|
19
|
+
|
20
|
+
xmlNodePtr node = xmlNewDocComment(
|
21
|
+
xml_doc,
|
22
|
+
(const xmlChar *)StringValuePtr(content)
|
23
|
+
);
|
24
|
+
|
25
|
+
VALUE rb_node = Nokogiri_wrap_xml_node(klass, node);
|
26
|
+
rb_obj_call_init(rb_node, argc, argv);
|
27
|
+
|
28
|
+
NOKOGIRI_ROOT_NODE(node);
|
29
|
+
|
30
|
+
if(rb_block_given_p()) rb_yield(rb_node);
|
31
|
+
|
32
|
+
return rb_node;
|
33
|
+
}
|
34
|
+
|
35
|
+
VALUE cNokogiriXmlComment;
|
36
|
+
void init_xml_comment()
|
37
|
+
{
|
38
|
+
VALUE nokogiri = rb_define_module("Nokogiri");
|
39
|
+
VALUE xml = rb_define_module_under(nokogiri, "XML");
|
40
|
+
VALUE node = rb_define_class_under(xml, "Node", rb_cObject);
|
41
|
+
VALUE char_data = rb_define_class_under(xml, "CharacterData", node);
|
42
|
+
|
43
|
+
/*
|
44
|
+
* Comment represents a comment node in an xml document.
|
45
|
+
*/
|
46
|
+
VALUE klass = rb_define_class_under(xml, "Comment", char_data);
|
47
|
+
|
48
|
+
|
49
|
+
cNokogiriXmlComment = klass;
|
50
|
+
|
51
|
+
rb_define_singleton_method(klass, "new", new, -1);
|
52
|
+
}
|
@@ -0,0 +1,386 @@
|
|
1
|
+
#include <xml_document.h>
|
2
|
+
|
3
|
+
static int dealloc_node_i(xmlNodePtr key, xmlNodePtr node, xmlDocPtr doc)
|
4
|
+
{
|
5
|
+
switch(node->type) {
|
6
|
+
case XML_ATTRIBUTE_NODE:
|
7
|
+
xmlFreePropList((xmlAttrPtr)node);
|
8
|
+
break;
|
9
|
+
default:
|
10
|
+
if(node->parent == NULL) {
|
11
|
+
xmlAddChild((xmlNodePtr)doc, node);
|
12
|
+
}
|
13
|
+
}
|
14
|
+
return ST_CONTINUE;
|
15
|
+
}
|
16
|
+
|
17
|
+
static void dealloc(xmlDocPtr doc)
|
18
|
+
{
|
19
|
+
NOKOGIRI_DEBUG_START(doc);
|
20
|
+
|
21
|
+
st_table *node_hash = DOC_UNLINKED_NODE_HASH(doc);
|
22
|
+
|
23
|
+
xmlDeregisterNodeFunc func = xmlDeregisterNodeDefault(NULL);
|
24
|
+
|
25
|
+
st_foreach(node_hash, dealloc_node_i, (st_data_t)doc);
|
26
|
+
st_free_table(node_hash);
|
27
|
+
|
28
|
+
free(doc->_private);
|
29
|
+
doc->_private = NULL;
|
30
|
+
xmlFreeDoc(doc);
|
31
|
+
|
32
|
+
xmlDeregisterNodeDefault(func);
|
33
|
+
|
34
|
+
NOKOGIRI_DEBUG_END(doc);
|
35
|
+
}
|
36
|
+
|
37
|
+
static void recursively_remove_namespaces_from_node(xmlNodePtr node)
|
38
|
+
{
|
39
|
+
xmlNodePtr child ;
|
40
|
+
|
41
|
+
xmlSetNs(node, NULL);
|
42
|
+
|
43
|
+
for (child = node->children ; child ; child = child->next)
|
44
|
+
recursively_remove_namespaces_from_node(child);
|
45
|
+
}
|
46
|
+
|
47
|
+
/*
|
48
|
+
* call-seq:
|
49
|
+
* url
|
50
|
+
*
|
51
|
+
* Get the url name for this document.
|
52
|
+
*/
|
53
|
+
static VALUE url(VALUE self)
|
54
|
+
{
|
55
|
+
xmlDocPtr doc;
|
56
|
+
Data_Get_Struct(self, xmlDoc, doc);
|
57
|
+
|
58
|
+
if(doc->URL) return NOKOGIRI_STR_NEW2(doc->URL);
|
59
|
+
|
60
|
+
return Qnil;
|
61
|
+
}
|
62
|
+
|
63
|
+
/*
|
64
|
+
* call-seq:
|
65
|
+
* root=
|
66
|
+
*
|
67
|
+
* Set the root element on this document
|
68
|
+
*/
|
69
|
+
static VALUE set_root(VALUE self, VALUE root)
|
70
|
+
{
|
71
|
+
xmlDocPtr doc;
|
72
|
+
xmlNodePtr new_root;
|
73
|
+
|
74
|
+
Data_Get_Struct(self, xmlDoc, doc);
|
75
|
+
Data_Get_Struct(root, xmlNode, new_root);
|
76
|
+
|
77
|
+
xmlNodePtr old_root = NULL;
|
78
|
+
|
79
|
+
/* If the new root's document is not the same as the current document,
|
80
|
+
* then we need to dup the node in to this document. */
|
81
|
+
if(new_root->doc != doc) {
|
82
|
+
old_root = xmlDocGetRootElement(doc);
|
83
|
+
if (!(new_root = xmlDocCopyNode(new_root, doc, 1))) {
|
84
|
+
rb_raise(rb_eRuntimeError, "Could not reparent node (xmlDocCopyNode)");
|
85
|
+
}
|
86
|
+
}
|
87
|
+
|
88
|
+
xmlDocSetRootElement(doc, new_root);
|
89
|
+
if(old_root) NOKOGIRI_ROOT_NODE(old_root);
|
90
|
+
return root;
|
91
|
+
}
|
92
|
+
|
93
|
+
/*
|
94
|
+
* call-seq:
|
95
|
+
* root
|
96
|
+
*
|
97
|
+
* Get the root node for this document.
|
98
|
+
*/
|
99
|
+
static VALUE root(VALUE self)
|
100
|
+
{
|
101
|
+
xmlDocPtr doc;
|
102
|
+
Data_Get_Struct(self, xmlDoc, doc);
|
103
|
+
|
104
|
+
xmlNodePtr root = xmlDocGetRootElement(doc);
|
105
|
+
|
106
|
+
if(!root) return Qnil;
|
107
|
+
return Nokogiri_wrap_xml_node(Qnil, root) ;
|
108
|
+
}
|
109
|
+
|
110
|
+
/*
|
111
|
+
* call-seq:
|
112
|
+
* encoding= encoding
|
113
|
+
*
|
114
|
+
* Set the encoding string for this Document
|
115
|
+
*/
|
116
|
+
static VALUE set_encoding(VALUE self, VALUE encoding)
|
117
|
+
{
|
118
|
+
xmlDocPtr doc;
|
119
|
+
Data_Get_Struct(self, xmlDoc, doc);
|
120
|
+
|
121
|
+
doc->encoding = xmlStrdup((xmlChar *)StringValuePtr(encoding));
|
122
|
+
|
123
|
+
return encoding;
|
124
|
+
}
|
125
|
+
|
126
|
+
/*
|
127
|
+
* call-seq:
|
128
|
+
* encoding
|
129
|
+
*
|
130
|
+
* Get the encoding for this Document
|
131
|
+
*/
|
132
|
+
static VALUE encoding(VALUE self)
|
133
|
+
{
|
134
|
+
xmlDocPtr doc;
|
135
|
+
Data_Get_Struct(self, xmlDoc, doc);
|
136
|
+
|
137
|
+
if(!doc->encoding) return Qnil;
|
138
|
+
return NOKOGIRI_STR_NEW2(doc->encoding);
|
139
|
+
}
|
140
|
+
|
141
|
+
/*
|
142
|
+
* call-seq:
|
143
|
+
* version
|
144
|
+
*
|
145
|
+
* Get the XML version for this Document
|
146
|
+
*/
|
147
|
+
static VALUE version(VALUE self)
|
148
|
+
{
|
149
|
+
xmlDocPtr doc;
|
150
|
+
Data_Get_Struct(self, xmlDoc, doc);
|
151
|
+
|
152
|
+
if(!doc->version) return Qnil;
|
153
|
+
return NOKOGIRI_STR_NEW2(doc->version);
|
154
|
+
}
|
155
|
+
|
156
|
+
/*
|
157
|
+
* call-seq:
|
158
|
+
* read_io(io, url, encoding, options)
|
159
|
+
*
|
160
|
+
* Create a new document from an IO object
|
161
|
+
*/
|
162
|
+
static VALUE read_io( VALUE klass,
|
163
|
+
VALUE io,
|
164
|
+
VALUE url,
|
165
|
+
VALUE encoding,
|
166
|
+
VALUE options )
|
167
|
+
{
|
168
|
+
const char * c_url = NIL_P(url) ? NULL : StringValuePtr(url);
|
169
|
+
const char * c_enc = NIL_P(encoding) ? NULL : StringValuePtr(encoding);
|
170
|
+
VALUE error_list = rb_ary_new();
|
171
|
+
|
172
|
+
xmlResetLastError();
|
173
|
+
xmlSetStructuredErrorFunc((void *)error_list, Nokogiri_error_array_pusher);
|
174
|
+
|
175
|
+
xmlDocPtr doc = xmlReadIO(
|
176
|
+
(xmlInputReadCallback)io_read_callback,
|
177
|
+
(xmlInputCloseCallback)io_close_callback,
|
178
|
+
(void *)io,
|
179
|
+
c_url,
|
180
|
+
c_enc,
|
181
|
+
(int)NUM2INT(options)
|
182
|
+
);
|
183
|
+
xmlSetStructuredErrorFunc(NULL, NULL);
|
184
|
+
|
185
|
+
if(doc == NULL) {
|
186
|
+
xmlFreeDoc(doc);
|
187
|
+
|
188
|
+
xmlErrorPtr error = xmlGetLastError();
|
189
|
+
if(error)
|
190
|
+
rb_exc_raise(Nokogiri_wrap_xml_syntax_error((VALUE)NULL, error));
|
191
|
+
else
|
192
|
+
rb_raise(rb_eRuntimeError, "Could not parse document");
|
193
|
+
|
194
|
+
return Qnil;
|
195
|
+
}
|
196
|
+
|
197
|
+
VALUE document = Nokogiri_wrap_xml_document(klass, doc);
|
198
|
+
rb_iv_set(document, "@errors", error_list);
|
199
|
+
return document;
|
200
|
+
}
|
201
|
+
|
202
|
+
/*
|
203
|
+
* call-seq:
|
204
|
+
* read_memory(string, url, encoding, options)
|
205
|
+
*
|
206
|
+
* Create a new document from a String
|
207
|
+
*/
|
208
|
+
static VALUE read_memory( VALUE klass,
|
209
|
+
VALUE string,
|
210
|
+
VALUE url,
|
211
|
+
VALUE encoding,
|
212
|
+
VALUE options )
|
213
|
+
{
|
214
|
+
const char * c_buffer = StringValuePtr(string);
|
215
|
+
const char * c_url = NIL_P(url) ? NULL : StringValuePtr(url);
|
216
|
+
const char * c_enc = NIL_P(encoding) ? NULL : StringValuePtr(encoding);
|
217
|
+
int len = RSTRING_LEN(string);
|
218
|
+
VALUE error_list = rb_ary_new();
|
219
|
+
|
220
|
+
xmlResetLastError();
|
221
|
+
xmlSetStructuredErrorFunc((void *)error_list, Nokogiri_error_array_pusher);
|
222
|
+
xmlDocPtr doc = xmlReadMemory(c_buffer, len, c_url, c_enc, (int)NUM2INT(options));
|
223
|
+
xmlSetStructuredErrorFunc(NULL, NULL);
|
224
|
+
|
225
|
+
if(doc == NULL) {
|
226
|
+
xmlFreeDoc(doc);
|
227
|
+
|
228
|
+
xmlErrorPtr error = xmlGetLastError();
|
229
|
+
if(error)
|
230
|
+
rb_exc_raise(Nokogiri_wrap_xml_syntax_error((VALUE)NULL, error));
|
231
|
+
else
|
232
|
+
rb_raise(rb_eRuntimeError, "Could not parse document");
|
233
|
+
|
234
|
+
return Qnil;
|
235
|
+
}
|
236
|
+
|
237
|
+
VALUE document = Nokogiri_wrap_xml_document(klass, doc);
|
238
|
+
rb_iv_set(document, "@errors", error_list);
|
239
|
+
return document;
|
240
|
+
}
|
241
|
+
|
242
|
+
/*
|
243
|
+
* call-seq:
|
244
|
+
* dup
|
245
|
+
*
|
246
|
+
* Copy this Document. An optional depth may be passed in, but it defaults
|
247
|
+
* to a deep copy. 0 is a shallow copy, 1 is a deep copy.
|
248
|
+
*/
|
249
|
+
static VALUE duplicate_node(int argc, VALUE *argv, VALUE self)
|
250
|
+
{
|
251
|
+
VALUE level;
|
252
|
+
|
253
|
+
if(rb_scan_args(argc, argv, "01", &level) == 0)
|
254
|
+
level = INT2NUM((long)1);
|
255
|
+
|
256
|
+
xmlDocPtr doc, dup;
|
257
|
+
Data_Get_Struct(self, xmlDoc, doc);
|
258
|
+
|
259
|
+
dup = xmlCopyDoc(doc, (int)NUM2INT(level));
|
260
|
+
if(dup == NULL) return Qnil;
|
261
|
+
|
262
|
+
dup->type = doc->type;
|
263
|
+
return Nokogiri_wrap_xml_document(rb_obj_class(self), dup);
|
264
|
+
}
|
265
|
+
|
266
|
+
/*
|
267
|
+
* call-seq:
|
268
|
+
* new(version = default)
|
269
|
+
*
|
270
|
+
* Create a new document with +version+ (defaults to "1.0")
|
271
|
+
*/
|
272
|
+
static VALUE new(int argc, VALUE *argv, VALUE klass)
|
273
|
+
{
|
274
|
+
VALUE version, rest, rb_doc ;
|
275
|
+
|
276
|
+
rb_scan_args(argc, argv, "0*", &rest);
|
277
|
+
version = rb_ary_entry(rest, (long)0);
|
278
|
+
if (NIL_P(version)) version = rb_str_new2("1.0");
|
279
|
+
|
280
|
+
xmlDocPtr doc = xmlNewDoc((xmlChar *)StringValuePtr(version));
|
281
|
+
rb_doc = Nokogiri_wrap_xml_document(klass, doc);
|
282
|
+
rb_obj_call_init(rb_doc, argc, argv);
|
283
|
+
return rb_doc ;
|
284
|
+
}
|
285
|
+
|
286
|
+
/*
|
287
|
+
* call-seq:
|
288
|
+
* remove_namespaces!
|
289
|
+
*
|
290
|
+
* Remove all namespaces from all nodes in the document.
|
291
|
+
*
|
292
|
+
* This could be useful for developers who either don't understand namespaces
|
293
|
+
* or don't care about them.
|
294
|
+
*
|
295
|
+
* The following example shows a use case, and you can decide for yourself
|
296
|
+
* whether this is a good thing or not:
|
297
|
+
*
|
298
|
+
* doc = Nokogiri::XML <<-EOXML
|
299
|
+
* <root>
|
300
|
+
* <car xmlns:part="http://general-motors.com/">
|
301
|
+
* <part:tire>Michelin Model XGV</part:tire>
|
302
|
+
* </car>
|
303
|
+
* <bicycle xmlns:part="http://schwinn.com/">
|
304
|
+
* <part:tire>I'm a bicycle tire!</part:tire>
|
305
|
+
* </bicycle>
|
306
|
+
* </root>
|
307
|
+
* EOXML
|
308
|
+
*
|
309
|
+
* doc.xpath("//tire").to_s # => ""
|
310
|
+
* doc.xpath("//part:tire", "part" => "http://general-motors.com/").to_s # => "<part:tire>Michelin Model XGV</part:tire>"
|
311
|
+
* doc.xpath("//part:tire", "part" => "http://schwinn.com/").to_s # => "<part:tire>I'm a bicycle tire!</part:tire>"
|
312
|
+
*
|
313
|
+
* doc.remove_namespaces!
|
314
|
+
*
|
315
|
+
* doc.xpath("//tire").to_s # => "<tire>Michelin Model XGV</tire><tire>I'm a bicycle tire!</tire>"
|
316
|
+
* doc.xpath("//part:tire", "part" => "http://general-motors.com/").to_s # => ""
|
317
|
+
* doc.xpath("//part:tire", "part" => "http://schwinn.com/").to_s # => ""
|
318
|
+
*
|
319
|
+
* For more information on why this probably is *not* a good thing in general,
|
320
|
+
* please direct your browser to
|
321
|
+
* http://tenderlovemaking.com/2009/04/23/namespaces-in-xml/
|
322
|
+
*/
|
323
|
+
VALUE remove_namespaces_bang(VALUE self)
|
324
|
+
{
|
325
|
+
xmlDocPtr doc ;
|
326
|
+
Data_Get_Struct(self, xmlDoc, doc);
|
327
|
+
|
328
|
+
recursively_remove_namespaces_from_node((xmlNodePtr)doc);
|
329
|
+
return self;
|
330
|
+
}
|
331
|
+
|
332
|
+
|
333
|
+
VALUE cNokogiriXmlDocument ;
|
334
|
+
void init_xml_document()
|
335
|
+
{
|
336
|
+
VALUE nokogiri = rb_define_module("Nokogiri");
|
337
|
+
VALUE xml = rb_define_module_under(nokogiri, "XML");
|
338
|
+
VALUE node = rb_define_class_under(xml, "Node", rb_cObject);
|
339
|
+
|
340
|
+
/*
|
341
|
+
* Nokogiri::XML::Document wraps an xml document.
|
342
|
+
*/
|
343
|
+
VALUE klass = rb_define_class_under(xml, "Document", node);
|
344
|
+
|
345
|
+
cNokogiriXmlDocument = klass;
|
346
|
+
|
347
|
+
rb_define_singleton_method(klass, "read_memory", read_memory, 4);
|
348
|
+
rb_define_singleton_method(klass, "read_io", read_io, 4);
|
349
|
+
rb_define_singleton_method(klass, "new", new, -1);
|
350
|
+
|
351
|
+
rb_define_method(klass, "root", root, 0);
|
352
|
+
rb_define_method(klass, "root=", set_root, 1);
|
353
|
+
rb_define_method(klass, "encoding", encoding, 0);
|
354
|
+
rb_define_method(klass, "encoding=", set_encoding, 1);
|
355
|
+
rb_define_method(klass, "version", version, 0);
|
356
|
+
rb_define_method(klass, "dup", duplicate_node, -1);
|
357
|
+
rb_define_method(klass, "url", url, 0);
|
358
|
+
rb_define_method(klass, "remove_namespaces!", remove_namespaces_bang, 0);
|
359
|
+
}
|
360
|
+
|
361
|
+
|
362
|
+
/* this takes klass as a param because it's used for HtmlDocument, too. */
|
363
|
+
VALUE Nokogiri_wrap_xml_document(VALUE klass, xmlDocPtr doc)
|
364
|
+
{
|
365
|
+
nokogiriTuplePtr tuple = (nokogiriTuplePtr)malloc(sizeof(nokogiriTuple));
|
366
|
+
|
367
|
+
VALUE rb_doc = Data_Wrap_Struct(
|
368
|
+
klass ? klass : cNokogiriXmlDocument,
|
369
|
+
0,
|
370
|
+
dealloc,
|
371
|
+
doc
|
372
|
+
);
|
373
|
+
|
374
|
+
VALUE cache = rb_ary_new();
|
375
|
+
rb_iv_set(rb_doc, "@decorators", Qnil);
|
376
|
+
rb_iv_set(rb_doc, "@node_cache", cache);
|
377
|
+
|
378
|
+
tuple->doc = (void *)rb_doc;
|
379
|
+
tuple->unlinkedNodes = st_init_numtable_with_size(128);
|
380
|
+
tuple->node_cache = cache;
|
381
|
+
doc->_private = tuple ;
|
382
|
+
|
383
|
+
rb_obj_call_init(rb_doc, 0, NULL);
|
384
|
+
|
385
|
+
return rb_doc ;
|
386
|
+
}
|