nokogiri-backport 1.11.0
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +7 -0
- data/Gemfile +3 -0
- data/LICENSE-DEPENDENCIES.md +1682 -0
- data/LICENSE.md +9 -0
- data/README.md +272 -0
- data/bin/nokogiri +118 -0
- data/dependencies.yml +74 -0
- data/ext/java/nokogiri/EncodingHandler.java +124 -0
- data/ext/java/nokogiri/HtmlDocument.java +178 -0
- data/ext/java/nokogiri/HtmlElementDescription.java +148 -0
- data/ext/java/nokogiri/HtmlEntityLookup.java +79 -0
- data/ext/java/nokogiri/HtmlSaxParserContext.java +282 -0
- data/ext/java/nokogiri/HtmlSaxPushParser.java +222 -0
- data/ext/java/nokogiri/NokogiriService.java +597 -0
- data/ext/java/nokogiri/XmlAttr.java +162 -0
- data/ext/java/nokogiri/XmlAttributeDecl.java +129 -0
- data/ext/java/nokogiri/XmlCdata.java +82 -0
- data/ext/java/nokogiri/XmlComment.java +97 -0
- data/ext/java/nokogiri/XmlDocument.java +633 -0
- data/ext/java/nokogiri/XmlDocumentFragment.java +185 -0
- data/ext/java/nokogiri/XmlDtd.java +481 -0
- data/ext/java/nokogiri/XmlElement.java +68 -0
- data/ext/java/nokogiri/XmlElementContent.java +382 -0
- data/ext/java/nokogiri/XmlElementDecl.java +147 -0
- data/ext/java/nokogiri/XmlEntityDecl.java +157 -0
- data/ext/java/nokogiri/XmlEntityReference.java +101 -0
- data/ext/java/nokogiri/XmlNamespace.java +199 -0
- data/ext/java/nokogiri/XmlNode.java +1684 -0
- data/ext/java/nokogiri/XmlNodeSet.java +434 -0
- data/ext/java/nokogiri/XmlProcessingInstruction.java +100 -0
- data/ext/java/nokogiri/XmlReader.java +531 -0
- data/ext/java/nokogiri/XmlRelaxng.java +151 -0
- data/ext/java/nokogiri/XmlSaxParserContext.java +374 -0
- data/ext/java/nokogiri/XmlSaxPushParser.java +286 -0
- data/ext/java/nokogiri/XmlSchema.java +388 -0
- data/ext/java/nokogiri/XmlSyntaxError.java +138 -0
- data/ext/java/nokogiri/XmlText.java +110 -0
- data/ext/java/nokogiri/XmlXpathContext.java +301 -0
- data/ext/java/nokogiri/XsltStylesheet.java +347 -0
- data/ext/java/nokogiri/internals/ClosedStreamException.java +10 -0
- data/ext/java/nokogiri/internals/HtmlDomParserContext.java +252 -0
- data/ext/java/nokogiri/internals/IgnoreSchemaErrorsErrorHandler.java +20 -0
- data/ext/java/nokogiri/internals/NokogiriBlockingQueueInputStream.java +151 -0
- data/ext/java/nokogiri/internals/NokogiriDomParser.java +116 -0
- data/ext/java/nokogiri/internals/NokogiriEntityResolver.java +121 -0
- data/ext/java/nokogiri/internals/NokogiriErrorHandler.java +69 -0
- data/ext/java/nokogiri/internals/NokogiriHandler.java +327 -0
- data/ext/java/nokogiri/internals/NokogiriHelpers.java +734 -0
- data/ext/java/nokogiri/internals/NokogiriNamespaceCache.java +217 -0
- data/ext/java/nokogiri/internals/NokogiriNamespaceContext.java +127 -0
- data/ext/java/nokogiri/internals/NokogiriNonStrictErrorHandler.java +100 -0
- data/ext/java/nokogiri/internals/NokogiriNonStrictErrorHandler4NekoHtml.java +121 -0
- data/ext/java/nokogiri/internals/NokogiriStrictErrorHandler.java +78 -0
- data/ext/java/nokogiri/internals/NokogiriXPathFunction.java +180 -0
- data/ext/java/nokogiri/internals/NokogiriXPathFunctionResolver.java +72 -0
- data/ext/java/nokogiri/internals/NokogiriXPathVariableResolver.java +60 -0
- data/ext/java/nokogiri/internals/NokogiriXsltErrorListener.java +87 -0
- data/ext/java/nokogiri/internals/ParserContext.java +259 -0
- data/ext/java/nokogiri/internals/ReaderNode.java +488 -0
- data/ext/java/nokogiri/internals/SaveContextVisitor.java +778 -0
- data/ext/java/nokogiri/internals/SchemaErrorHandler.java +73 -0
- data/ext/java/nokogiri/internals/XalanDTMManagerPatch.java +168 -0
- data/ext/java/nokogiri/internals/XmlDeclHandler.java +42 -0
- data/ext/java/nokogiri/internals/XmlDomParserContext.java +274 -0
- data/ext/java/nokogiri/internals/XmlSaxParser.java +65 -0
- data/ext/java/nokogiri/internals/c14n/AttrCompare.java +119 -0
- data/ext/java/nokogiri/internals/c14n/C14nHelper.java +159 -0
- data/ext/java/nokogiri/internals/c14n/CanonicalFilter.java +37 -0
- data/ext/java/nokogiri/internals/c14n/CanonicalizationException.java +93 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer.java +252 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer11.java +639 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer11_OmitComments.java +38 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer11_WithComments.java +38 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer20010315.java +367 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer20010315Excl.java +295 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer20010315ExclOmitComments.java +40 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer20010315ExclWithComments.java +44 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer20010315OmitComments.java +44 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer20010315WithComments.java +43 -0
- data/ext/java/nokogiri/internals/c14n/CanonicalizerBase.java +630 -0
- data/ext/java/nokogiri/internals/c14n/CanonicalizerPhysical.java +173 -0
- data/ext/java/nokogiri/internals/c14n/CanonicalizerSpi.java +76 -0
- data/ext/java/nokogiri/internals/c14n/Constants.java +42 -0
- data/ext/java/nokogiri/internals/c14n/ElementProxy.java +293 -0
- data/ext/java/nokogiri/internals/c14n/HelperNodeList.java +93 -0
- data/ext/java/nokogiri/internals/c14n/IgnoreAllErrorHandler.java +79 -0
- data/ext/java/nokogiri/internals/c14n/InclusiveNamespaces.java +166 -0
- data/ext/java/nokogiri/internals/c14n/InvalidCanonicalizerException.java +76 -0
- data/ext/java/nokogiri/internals/c14n/NameSpaceSymbTable.java +402 -0
- data/ext/java/nokogiri/internals/c14n/NodeFilter.java +51 -0
- data/ext/java/nokogiri/internals/c14n/UtfHelpper.java +179 -0
- data/ext/java/nokogiri/internals/c14n/XMLUtils.java +507 -0
- data/ext/java/nokogiri/internals/dom2dtm/DOM2DTM.java +1745 -0
- data/ext/java/nokogiri/internals/dom2dtm/DOM2DTMdefaultNamespaceDeclarationNode.java +685 -0
- data/ext/nokogiri/depend +477 -0
- data/ext/nokogiri/extconf.rb +836 -0
- data/ext/nokogiri/html_document.c +171 -0
- data/ext/nokogiri/html_document.h +10 -0
- data/ext/nokogiri/html_element_description.c +279 -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 +116 -0
- data/ext/nokogiri/html_sax_parser_context.h +11 -0
- data/ext/nokogiri/html_sax_push_parser.c +87 -0
- data/ext/nokogiri/html_sax_push_parser.h +9 -0
- data/ext/nokogiri/nokogiri.c +135 -0
- data/ext/nokogiri/nokogiri.h +130 -0
- data/ext/nokogiri/xml_attr.c +103 -0
- data/ext/nokogiri/xml_attr.h +9 -0
- data/ext/nokogiri/xml_attribute_decl.c +70 -0
- data/ext/nokogiri/xml_attribute_decl.h +9 -0
- data/ext/nokogiri/xml_cdata.c +62 -0
- data/ext/nokogiri/xml_cdata.h +9 -0
- data/ext/nokogiri/xml_comment.c +69 -0
- data/ext/nokogiri/xml_comment.h +9 -0
- data/ext/nokogiri/xml_document.c +622 -0
- data/ext/nokogiri/xml_document.h +23 -0
- data/ext/nokogiri/xml_document_fragment.c +48 -0
- data/ext/nokogiri/xml_document_fragment.h +10 -0
- data/ext/nokogiri/xml_dtd.c +202 -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 +110 -0
- data/ext/nokogiri/xml_entity_decl.h +10 -0
- data/ext/nokogiri/xml_entity_reference.c +52 -0
- data/ext/nokogiri/xml_entity_reference.h +9 -0
- data/ext/nokogiri/xml_io.c +63 -0
- data/ext/nokogiri/xml_io.h +11 -0
- data/ext/nokogiri/xml_libxml2_hacks.c +112 -0
- data/ext/nokogiri/xml_libxml2_hacks.h +12 -0
- data/ext/nokogiri/xml_namespace.c +111 -0
- data/ext/nokogiri/xml_namespace.h +14 -0
- data/ext/nokogiri/xml_node.c +1773 -0
- data/ext/nokogiri/xml_node.h +13 -0
- data/ext/nokogiri/xml_node_set.c +486 -0
- data/ext/nokogiri/xml_node_set.h +12 -0
- data/ext/nokogiri/xml_processing_instruction.c +56 -0
- data/ext/nokogiri/xml_processing_instruction.h +9 -0
- data/ext/nokogiri/xml_reader.c +657 -0
- data/ext/nokogiri/xml_reader.h +10 -0
- data/ext/nokogiri/xml_relax_ng.c +179 -0
- data/ext/nokogiri/xml_relax_ng.h +9 -0
- data/ext/nokogiri/xml_sax_parser.c +305 -0
- data/ext/nokogiri/xml_sax_parser.h +39 -0
- data/ext/nokogiri/xml_sax_parser_context.c +262 -0
- data/ext/nokogiri/xml_sax_parser_context.h +10 -0
- data/ext/nokogiri/xml_sax_push_parser.c +159 -0
- data/ext/nokogiri/xml_sax_push_parser.h +9 -0
- data/ext/nokogiri/xml_schema.c +276 -0
- data/ext/nokogiri/xml_schema.h +9 -0
- data/ext/nokogiri/xml_syntax_error.c +64 -0
- data/ext/nokogiri/xml_syntax_error.h +13 -0
- data/ext/nokogiri/xml_text.c +52 -0
- data/ext/nokogiri/xml_text.h +9 -0
- data/ext/nokogiri/xml_xpath_context.c +374 -0
- data/ext/nokogiri/xml_xpath_context.h +10 -0
- data/ext/nokogiri/xslt_stylesheet.c +263 -0
- data/ext/nokogiri/xslt_stylesheet.h +14 -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/css/node.rb +53 -0
- data/lib/nokogiri/css/parser.rb +751 -0
- data/lib/nokogiri/css/parser.y +272 -0
- data/lib/nokogiri/css/parser_extras.rb +94 -0
- data/lib/nokogiri/css/syntax_error.rb +8 -0
- data/lib/nokogiri/css/tokenizer.rb +154 -0
- data/lib/nokogiri/css/tokenizer.rex +55 -0
- data/lib/nokogiri/css/xpath_visitor.rb +260 -0
- data/lib/nokogiri/css.rb +28 -0
- data/lib/nokogiri/decorators/slop.rb +43 -0
- data/lib/nokogiri/html/builder.rb +36 -0
- data/lib/nokogiri/html/document.rb +322 -0
- data/lib/nokogiri/html/document_fragment.rb +50 -0
- data/lib/nokogiri/html/element_description.rb +24 -0
- data/lib/nokogiri/html/element_description_defaults.rb +672 -0
- data/lib/nokogiri/html/entity_lookup.rb +14 -0
- data/lib/nokogiri/html/sax/parser.rb +63 -0
- data/lib/nokogiri/html/sax/parser_context.rb +17 -0
- data/lib/nokogiri/html/sax/push_parser.rb +37 -0
- data/lib/nokogiri/html.rb +38 -0
- data/lib/nokogiri/jruby/dependencies.rb +20 -0
- data/lib/nokogiri/syntax_error.rb +5 -0
- data/lib/nokogiri/version/constant.rb +5 -0
- data/lib/nokogiri/version/info.rb +182 -0
- data/lib/nokogiri/version.rb +3 -0
- data/lib/nokogiri/xml/attr.rb +15 -0
- data/lib/nokogiri/xml/attribute_decl.rb +19 -0
- data/lib/nokogiri/xml/builder.rb +447 -0
- data/lib/nokogiri/xml/cdata.rb +12 -0
- data/lib/nokogiri/xml/character_data.rb +8 -0
- data/lib/nokogiri/xml/document.rb +290 -0
- data/lib/nokogiri/xml/document_fragment.rb +159 -0
- data/lib/nokogiri/xml/dtd.rb +33 -0
- data/lib/nokogiri/xml/element_content.rb +37 -0
- data/lib/nokogiri/xml/element_decl.rb +14 -0
- data/lib/nokogiri/xml/entity_decl.rb +20 -0
- data/lib/nokogiri/xml/entity_reference.rb +19 -0
- data/lib/nokogiri/xml/namespace.rb +14 -0
- data/lib/nokogiri/xml/node/save_options.rb +62 -0
- data/lib/nokogiri/xml/node.rb +1240 -0
- data/lib/nokogiri/xml/node_set.rb +372 -0
- data/lib/nokogiri/xml/notation.rb +7 -0
- data/lib/nokogiri/xml/parse_options.rb +127 -0
- data/lib/nokogiri/xml/pp/character_data.rb +19 -0
- data/lib/nokogiri/xml/pp/node.rb +57 -0
- data/lib/nokogiri/xml/pp.rb +3 -0
- data/lib/nokogiri/xml/processing_instruction.rb +9 -0
- data/lib/nokogiri/xml/reader.rb +116 -0
- data/lib/nokogiri/xml/relax_ng.rb +37 -0
- data/lib/nokogiri/xml/sax/document.rb +172 -0
- data/lib/nokogiri/xml/sax/parser.rb +123 -0
- data/lib/nokogiri/xml/sax/parser_context.rb +17 -0
- data/lib/nokogiri/xml/sax/push_parser.rb +61 -0
- data/lib/nokogiri/xml/sax.rb +5 -0
- data/lib/nokogiri/xml/schema.rb +72 -0
- data/lib/nokogiri/xml/searchable.rb +239 -0
- data/lib/nokogiri/xml/syntax_error.rb +71 -0
- data/lib/nokogiri/xml/text.rb +10 -0
- data/lib/nokogiri/xml/xpath/syntax_error.rb +12 -0
- data/lib/nokogiri/xml/xpath.rb +11 -0
- data/lib/nokogiri/xml/xpath_context.rb +17 -0
- data/lib/nokogiri/xml.rb +76 -0
- data/lib/nokogiri/xslt/stylesheet.rb +26 -0
- data/lib/nokogiri/xslt.rb +57 -0
- data/lib/nokogiri.rb +144 -0
- data/lib/serializer.jar +0 -0
- data/lib/xalan.jar +0 -0
- data/lib/xercesImpl.jar +0 -0
- data/lib/xml-apis.jar +0 -0
- data/lib/xsd/xmlparser/nokogiri.rb +103 -0
- metadata +531 -0
@@ -0,0 +1,657 @@
|
|
1
|
+
#include <xml_reader.h>
|
2
|
+
|
3
|
+
static void dealloc(xmlTextReaderPtr reader)
|
4
|
+
{
|
5
|
+
NOKOGIRI_DEBUG_START(reader);
|
6
|
+
xmlFreeTextReader(reader);
|
7
|
+
NOKOGIRI_DEBUG_END(reader);
|
8
|
+
}
|
9
|
+
|
10
|
+
static int has_attributes(xmlTextReaderPtr reader)
|
11
|
+
{
|
12
|
+
/*
|
13
|
+
* this implementation of xmlTextReaderHasAttributes explicitly includes
|
14
|
+
* namespaces and properties, because some earlier versions ignore
|
15
|
+
* namespaces.
|
16
|
+
*/
|
17
|
+
xmlNodePtr node ;
|
18
|
+
node = xmlTextReaderCurrentNode(reader);
|
19
|
+
if (node == NULL)
|
20
|
+
return(0);
|
21
|
+
|
22
|
+
if ((node->type == XML_ELEMENT_NODE) &&
|
23
|
+
((node->properties != NULL) || (node->nsDef != NULL)))
|
24
|
+
return(1);
|
25
|
+
return(0);
|
26
|
+
}
|
27
|
+
|
28
|
+
static void Nokogiri_xml_node_namespaces(xmlNodePtr node, VALUE attr_hash)
|
29
|
+
{
|
30
|
+
xmlNsPtr ns;
|
31
|
+
VALUE key;
|
32
|
+
|
33
|
+
if (node->type != XML_ELEMENT_NODE) return ;
|
34
|
+
|
35
|
+
ns = node->nsDef;
|
36
|
+
while (ns != NULL) {
|
37
|
+
|
38
|
+
key = rb_enc_str_new_cstr(XMLNS_PREFIX, rb_utf8_encoding());
|
39
|
+
if (ns->prefix) {
|
40
|
+
rb_str_cat_cstr(key, ":");
|
41
|
+
rb_str_cat_cstr(key, (const char*)ns->prefix);
|
42
|
+
}
|
43
|
+
|
44
|
+
key = rb_str_conv_enc(key, rb_utf8_encoding(), rb_default_internal_encoding());
|
45
|
+
rb_hash_aset(attr_hash,
|
46
|
+
key,
|
47
|
+
(ns->href ? NOKOGIRI_STR_NEW2(ns->href) : Qnil)
|
48
|
+
);
|
49
|
+
ns = ns->next ;
|
50
|
+
}
|
51
|
+
}
|
52
|
+
|
53
|
+
|
54
|
+
/*
|
55
|
+
* call-seq:
|
56
|
+
* default?
|
57
|
+
*
|
58
|
+
* Was an attribute generated from the default value in the DTD or schema?
|
59
|
+
*/
|
60
|
+
static VALUE default_eh(VALUE self)
|
61
|
+
{
|
62
|
+
xmlTextReaderPtr reader;
|
63
|
+
int eh;
|
64
|
+
|
65
|
+
Data_Get_Struct(self, xmlTextReader, reader);
|
66
|
+
eh = xmlTextReaderIsDefault(reader);
|
67
|
+
if(eh == 0) return Qfalse;
|
68
|
+
if(eh == 1) return Qtrue;
|
69
|
+
|
70
|
+
return Qnil;
|
71
|
+
}
|
72
|
+
|
73
|
+
/*
|
74
|
+
* call-seq:
|
75
|
+
* value?
|
76
|
+
*
|
77
|
+
* Does this node have a text value?
|
78
|
+
*/
|
79
|
+
static VALUE value_eh(VALUE self)
|
80
|
+
{
|
81
|
+
xmlTextReaderPtr reader;
|
82
|
+
int eh;
|
83
|
+
|
84
|
+
Data_Get_Struct(self, xmlTextReader, reader);
|
85
|
+
eh = xmlTextReaderHasValue(reader);
|
86
|
+
if(eh == 0) return Qfalse;
|
87
|
+
if(eh == 1) return Qtrue;
|
88
|
+
|
89
|
+
return Qnil;
|
90
|
+
}
|
91
|
+
|
92
|
+
/*
|
93
|
+
* call-seq:
|
94
|
+
* attributes?
|
95
|
+
*
|
96
|
+
* Does this node have attributes?
|
97
|
+
*/
|
98
|
+
static VALUE attributes_eh(VALUE self)
|
99
|
+
{
|
100
|
+
xmlTextReaderPtr reader;
|
101
|
+
int eh;
|
102
|
+
|
103
|
+
Data_Get_Struct(self, xmlTextReader, reader);
|
104
|
+
eh = has_attributes(reader);
|
105
|
+
if(eh == 0) return Qfalse;
|
106
|
+
if(eh == 1) return Qtrue;
|
107
|
+
|
108
|
+
return Qnil;
|
109
|
+
}
|
110
|
+
|
111
|
+
/*
|
112
|
+
* call-seq:
|
113
|
+
* namespaces
|
114
|
+
*
|
115
|
+
* Get a hash of namespaces for this Node
|
116
|
+
*/
|
117
|
+
static VALUE namespaces(VALUE self)
|
118
|
+
{
|
119
|
+
xmlTextReaderPtr reader;
|
120
|
+
xmlNodePtr ptr;
|
121
|
+
VALUE attr ;
|
122
|
+
|
123
|
+
Data_Get_Struct(self, xmlTextReader, reader);
|
124
|
+
|
125
|
+
attr = rb_hash_new() ;
|
126
|
+
|
127
|
+
if (! has_attributes(reader))
|
128
|
+
return attr ;
|
129
|
+
|
130
|
+
ptr = xmlTextReaderExpand(reader);
|
131
|
+
if(ptr == NULL) return Qnil;
|
132
|
+
|
133
|
+
Nokogiri_xml_node_namespaces(ptr, attr);
|
134
|
+
|
135
|
+
return attr ;
|
136
|
+
}
|
137
|
+
|
138
|
+
/*
|
139
|
+
* call-seq:
|
140
|
+
* attribute_nodes
|
141
|
+
*
|
142
|
+
* Get a list of attributes for this Node
|
143
|
+
*/
|
144
|
+
static VALUE attribute_nodes(VALUE self)
|
145
|
+
{
|
146
|
+
xmlTextReaderPtr reader;
|
147
|
+
xmlNodePtr ptr;
|
148
|
+
VALUE attr ;
|
149
|
+
|
150
|
+
Data_Get_Struct(self, xmlTextReader, reader);
|
151
|
+
|
152
|
+
attr = rb_ary_new() ;
|
153
|
+
|
154
|
+
if (! has_attributes(reader))
|
155
|
+
return attr ;
|
156
|
+
|
157
|
+
ptr = xmlTextReaderExpand(reader);
|
158
|
+
if(ptr == NULL) return Qnil;
|
159
|
+
|
160
|
+
Nokogiri_xml_node_properties(ptr, attr);
|
161
|
+
|
162
|
+
return attr ;
|
163
|
+
}
|
164
|
+
|
165
|
+
/*
|
166
|
+
* call-seq:
|
167
|
+
* attribute_at(index)
|
168
|
+
*
|
169
|
+
* Get the value of attribute at +index+
|
170
|
+
*/
|
171
|
+
static VALUE attribute_at(VALUE self, VALUE index)
|
172
|
+
{
|
173
|
+
xmlTextReaderPtr reader;
|
174
|
+
xmlChar *value;
|
175
|
+
VALUE rb_value;
|
176
|
+
|
177
|
+
Data_Get_Struct(self, xmlTextReader, reader);
|
178
|
+
|
179
|
+
if(NIL_P(index)) return Qnil;
|
180
|
+
index = rb_Integer(index);
|
181
|
+
|
182
|
+
value = xmlTextReaderGetAttributeNo(
|
183
|
+
reader,
|
184
|
+
(int)NUM2INT(index)
|
185
|
+
);
|
186
|
+
if(value == NULL) return Qnil;
|
187
|
+
|
188
|
+
rb_value = NOKOGIRI_STR_NEW2(value);
|
189
|
+
xmlFree(value);
|
190
|
+
return rb_value;
|
191
|
+
}
|
192
|
+
|
193
|
+
/*
|
194
|
+
* call-seq:
|
195
|
+
* attribute(name)
|
196
|
+
*
|
197
|
+
* Get the value of attribute named +name+
|
198
|
+
*/
|
199
|
+
static VALUE reader_attribute(VALUE self, VALUE name)
|
200
|
+
{
|
201
|
+
xmlTextReaderPtr reader;
|
202
|
+
xmlChar *value ;
|
203
|
+
VALUE rb_value;
|
204
|
+
|
205
|
+
Data_Get_Struct(self, xmlTextReader, reader);
|
206
|
+
|
207
|
+
if(NIL_P(name)) return Qnil;
|
208
|
+
name = StringValue(name) ;
|
209
|
+
|
210
|
+
value = xmlTextReaderGetAttribute(reader, (xmlChar*)StringValueCStr(name));
|
211
|
+
if(value == NULL) return Qnil;
|
212
|
+
|
213
|
+
rb_value = NOKOGIRI_STR_NEW2(value);
|
214
|
+
xmlFree(value);
|
215
|
+
return rb_value;
|
216
|
+
}
|
217
|
+
|
218
|
+
/*
|
219
|
+
* call-seq:
|
220
|
+
* attribute_count
|
221
|
+
*
|
222
|
+
* Get the number of attributes for the current node
|
223
|
+
*/
|
224
|
+
static VALUE attribute_count(VALUE self)
|
225
|
+
{
|
226
|
+
xmlTextReaderPtr reader;
|
227
|
+
int count;
|
228
|
+
|
229
|
+
Data_Get_Struct(self, xmlTextReader, reader);
|
230
|
+
count = xmlTextReaderAttributeCount(reader);
|
231
|
+
if(count == -1) return Qnil;
|
232
|
+
|
233
|
+
return INT2NUM((long)count);
|
234
|
+
}
|
235
|
+
|
236
|
+
/*
|
237
|
+
* call-seq:
|
238
|
+
* depth
|
239
|
+
*
|
240
|
+
* Get the depth of the node
|
241
|
+
*/
|
242
|
+
static VALUE depth(VALUE self)
|
243
|
+
{
|
244
|
+
xmlTextReaderPtr reader;
|
245
|
+
int depth;
|
246
|
+
|
247
|
+
Data_Get_Struct(self, xmlTextReader, reader);
|
248
|
+
depth = xmlTextReaderDepth(reader);
|
249
|
+
if(depth == -1) return Qnil;
|
250
|
+
|
251
|
+
return INT2NUM((long)depth);
|
252
|
+
}
|
253
|
+
|
254
|
+
/*
|
255
|
+
* call-seq:
|
256
|
+
* xml_version
|
257
|
+
*
|
258
|
+
* Get the XML version of the document being read
|
259
|
+
*/
|
260
|
+
static VALUE xml_version(VALUE self)
|
261
|
+
{
|
262
|
+
xmlTextReaderPtr reader;
|
263
|
+
const char *version;
|
264
|
+
|
265
|
+
Data_Get_Struct(self, xmlTextReader, reader);
|
266
|
+
version = (const char *)xmlTextReaderConstXmlVersion(reader);
|
267
|
+
if(version == NULL) return Qnil;
|
268
|
+
|
269
|
+
return NOKOGIRI_STR_NEW2(version);
|
270
|
+
}
|
271
|
+
|
272
|
+
/*
|
273
|
+
* call-seq:
|
274
|
+
* lang
|
275
|
+
*
|
276
|
+
* Get the xml:lang scope within which the node resides.
|
277
|
+
*/
|
278
|
+
static VALUE lang(VALUE self)
|
279
|
+
{
|
280
|
+
xmlTextReaderPtr reader;
|
281
|
+
const char *lang;
|
282
|
+
|
283
|
+
Data_Get_Struct(self, xmlTextReader, reader);
|
284
|
+
lang = (const char *)xmlTextReaderConstXmlLang(reader);
|
285
|
+
if(lang == NULL) return Qnil;
|
286
|
+
|
287
|
+
return NOKOGIRI_STR_NEW2(lang);
|
288
|
+
}
|
289
|
+
|
290
|
+
/*
|
291
|
+
* call-seq:
|
292
|
+
* value
|
293
|
+
*
|
294
|
+
* Get the text value of the node if present. Returns a utf-8 encoded string.
|
295
|
+
*/
|
296
|
+
static VALUE value(VALUE self)
|
297
|
+
{
|
298
|
+
xmlTextReaderPtr reader;
|
299
|
+
const char *value;
|
300
|
+
|
301
|
+
Data_Get_Struct(self, xmlTextReader, reader);
|
302
|
+
value = (const char *)xmlTextReaderConstValue(reader);
|
303
|
+
if(value == NULL) return Qnil;
|
304
|
+
|
305
|
+
return NOKOGIRI_STR_NEW2(value);
|
306
|
+
}
|
307
|
+
|
308
|
+
/*
|
309
|
+
* call-seq:
|
310
|
+
* prefix
|
311
|
+
*
|
312
|
+
* Get the shorthand reference to the namespace associated with the node.
|
313
|
+
*/
|
314
|
+
static VALUE prefix(VALUE self)
|
315
|
+
{
|
316
|
+
xmlTextReaderPtr reader;
|
317
|
+
const char *prefix;
|
318
|
+
|
319
|
+
Data_Get_Struct(self, xmlTextReader, reader);
|
320
|
+
prefix = (const char *)xmlTextReaderConstPrefix(reader);
|
321
|
+
if(prefix == NULL) return Qnil;
|
322
|
+
|
323
|
+
return NOKOGIRI_STR_NEW2(prefix);
|
324
|
+
}
|
325
|
+
|
326
|
+
/*
|
327
|
+
* call-seq:
|
328
|
+
* namespace_uri
|
329
|
+
*
|
330
|
+
* Get the URI defining the namespace associated with the node
|
331
|
+
*/
|
332
|
+
static VALUE namespace_uri(VALUE self)
|
333
|
+
{
|
334
|
+
xmlTextReaderPtr reader;
|
335
|
+
const char *uri;
|
336
|
+
|
337
|
+
Data_Get_Struct(self, xmlTextReader, reader);
|
338
|
+
uri = (const char *)xmlTextReaderConstNamespaceUri(reader);
|
339
|
+
if(uri == NULL) return Qnil;
|
340
|
+
|
341
|
+
return NOKOGIRI_STR_NEW2(uri);
|
342
|
+
}
|
343
|
+
|
344
|
+
/*
|
345
|
+
* call-seq:
|
346
|
+
* local_name
|
347
|
+
*
|
348
|
+
* Get the local name of the node
|
349
|
+
*/
|
350
|
+
static VALUE local_name(VALUE self)
|
351
|
+
{
|
352
|
+
xmlTextReaderPtr reader;
|
353
|
+
const char *name;
|
354
|
+
|
355
|
+
Data_Get_Struct(self, xmlTextReader, reader);
|
356
|
+
name = (const char *)xmlTextReaderConstLocalName(reader);
|
357
|
+
if(name == NULL) return Qnil;
|
358
|
+
|
359
|
+
return NOKOGIRI_STR_NEW2(name);
|
360
|
+
}
|
361
|
+
|
362
|
+
/*
|
363
|
+
* call-seq:
|
364
|
+
* name
|
365
|
+
*
|
366
|
+
* Get the name of the node. Returns a utf-8 encoded string.
|
367
|
+
*/
|
368
|
+
static VALUE name(VALUE self)
|
369
|
+
{
|
370
|
+
xmlTextReaderPtr reader;
|
371
|
+
const char *name;
|
372
|
+
|
373
|
+
Data_Get_Struct(self, xmlTextReader, reader);
|
374
|
+
name = (const char *)xmlTextReaderConstName(reader);
|
375
|
+
if(name == NULL) return Qnil;
|
376
|
+
|
377
|
+
return NOKOGIRI_STR_NEW2(name);
|
378
|
+
}
|
379
|
+
|
380
|
+
/*
|
381
|
+
* call-seq:
|
382
|
+
* base_uri
|
383
|
+
*
|
384
|
+
* Get the xml:base of the node
|
385
|
+
*/
|
386
|
+
static VALUE base_uri(VALUE self)
|
387
|
+
{
|
388
|
+
xmlTextReaderPtr reader;
|
389
|
+
const char * base_uri;
|
390
|
+
|
391
|
+
Data_Get_Struct(self, xmlTextReader, reader);
|
392
|
+
base_uri = (const char *)xmlTextReaderBaseUri(reader);
|
393
|
+
if (base_uri == NULL) return Qnil;
|
394
|
+
|
395
|
+
return NOKOGIRI_STR_NEW2(base_uri);
|
396
|
+
}
|
397
|
+
|
398
|
+
/*
|
399
|
+
* call-seq:
|
400
|
+
* state
|
401
|
+
*
|
402
|
+
* Get the state of the reader
|
403
|
+
*/
|
404
|
+
static VALUE state(VALUE self)
|
405
|
+
{
|
406
|
+
xmlTextReaderPtr reader;
|
407
|
+
Data_Get_Struct(self, xmlTextReader, reader);
|
408
|
+
return INT2NUM((long)xmlTextReaderReadState(reader));
|
409
|
+
}
|
410
|
+
|
411
|
+
/*
|
412
|
+
* call-seq:
|
413
|
+
* node_type
|
414
|
+
*
|
415
|
+
* Get the type of readers current node
|
416
|
+
*/
|
417
|
+
static VALUE node_type(VALUE self)
|
418
|
+
{
|
419
|
+
xmlTextReaderPtr reader;
|
420
|
+
Data_Get_Struct(self, xmlTextReader, reader);
|
421
|
+
return INT2NUM((long)xmlTextReaderNodeType(reader));
|
422
|
+
}
|
423
|
+
|
424
|
+
/*
|
425
|
+
* call-seq:
|
426
|
+
* read
|
427
|
+
*
|
428
|
+
* Move the Reader forward through the XML document.
|
429
|
+
*/
|
430
|
+
static VALUE read_more(VALUE self)
|
431
|
+
{
|
432
|
+
xmlTextReaderPtr reader;
|
433
|
+
xmlErrorPtr error;
|
434
|
+
VALUE error_list;
|
435
|
+
int ret;
|
436
|
+
|
437
|
+
Data_Get_Struct(self, xmlTextReader, reader);
|
438
|
+
|
439
|
+
error_list = rb_funcall(self, rb_intern("errors"), 0);
|
440
|
+
|
441
|
+
xmlSetStructuredErrorFunc((void *)error_list, Nokogiri_error_array_pusher);
|
442
|
+
ret = xmlTextReaderRead(reader);
|
443
|
+
xmlSetStructuredErrorFunc(NULL, NULL);
|
444
|
+
|
445
|
+
if(ret == 1) return self;
|
446
|
+
if(ret == 0) return Qnil;
|
447
|
+
|
448
|
+
error = xmlGetLastError();
|
449
|
+
if(error)
|
450
|
+
rb_exc_raise(Nokogiri_wrap_xml_syntax_error(error));
|
451
|
+
else
|
452
|
+
rb_raise(rb_eRuntimeError, "Error pulling: %d", ret);
|
453
|
+
|
454
|
+
return Qnil;
|
455
|
+
}
|
456
|
+
|
457
|
+
/*
|
458
|
+
* call-seq:
|
459
|
+
* inner_xml
|
460
|
+
*
|
461
|
+
* Read the contents of the current node, including child nodes and markup.
|
462
|
+
* Returns a utf-8 encoded string.
|
463
|
+
*/
|
464
|
+
static VALUE inner_xml(VALUE self)
|
465
|
+
{
|
466
|
+
xmlTextReaderPtr reader;
|
467
|
+
xmlChar* value;
|
468
|
+
VALUE str;
|
469
|
+
|
470
|
+
Data_Get_Struct(self, xmlTextReader, reader);
|
471
|
+
|
472
|
+
value = xmlTextReaderReadInnerXml(reader);
|
473
|
+
|
474
|
+
str = Qnil;
|
475
|
+
if(value) {
|
476
|
+
str = NOKOGIRI_STR_NEW2((char*)value);
|
477
|
+
xmlFree(value);
|
478
|
+
}
|
479
|
+
|
480
|
+
return str;
|
481
|
+
}
|
482
|
+
|
483
|
+
/*
|
484
|
+
* call-seq:
|
485
|
+
* outer_xml
|
486
|
+
*
|
487
|
+
* Read the current node and its contents, including child nodes and markup.
|
488
|
+
* Returns a utf-8 encoded string.
|
489
|
+
*/
|
490
|
+
static VALUE outer_xml(VALUE self)
|
491
|
+
{
|
492
|
+
xmlTextReaderPtr reader;
|
493
|
+
xmlChar *value;
|
494
|
+
VALUE str = Qnil;
|
495
|
+
|
496
|
+
Data_Get_Struct(self, xmlTextReader, reader);
|
497
|
+
|
498
|
+
value = xmlTextReaderReadOuterXml(reader);
|
499
|
+
|
500
|
+
if(value) {
|
501
|
+
str = NOKOGIRI_STR_NEW2((char*)value);
|
502
|
+
xmlFree(value);
|
503
|
+
}
|
504
|
+
return str;
|
505
|
+
}
|
506
|
+
|
507
|
+
/*
|
508
|
+
* call-seq:
|
509
|
+
* from_memory(string, url = nil, encoding = nil, options = 0)
|
510
|
+
*
|
511
|
+
* Create a new reader that parses +string+
|
512
|
+
*/
|
513
|
+
static VALUE from_memory(int argc, VALUE *argv, VALUE klass)
|
514
|
+
{
|
515
|
+
VALUE rb_buffer, rb_url, encoding, rb_options;
|
516
|
+
xmlTextReaderPtr reader;
|
517
|
+
const char * c_url = NULL;
|
518
|
+
const char * c_encoding = NULL;
|
519
|
+
int c_options = 0;
|
520
|
+
VALUE rb_reader, args[3];
|
521
|
+
|
522
|
+
rb_scan_args(argc, argv, "13", &rb_buffer, &rb_url, &encoding, &rb_options);
|
523
|
+
|
524
|
+
if (!RTEST(rb_buffer)) rb_raise(rb_eArgError, "string cannot be nil");
|
525
|
+
if (RTEST(rb_url)) c_url = StringValueCStr(rb_url);
|
526
|
+
if (RTEST(encoding)) c_encoding = StringValueCStr(encoding);
|
527
|
+
if (RTEST(rb_options)) c_options = (int)NUM2INT(rb_options);
|
528
|
+
|
529
|
+
reader = xmlReaderForMemory(
|
530
|
+
StringValuePtr(rb_buffer),
|
531
|
+
(int)RSTRING_LEN(rb_buffer),
|
532
|
+
c_url,
|
533
|
+
c_encoding,
|
534
|
+
c_options
|
535
|
+
);
|
536
|
+
|
537
|
+
if(reader == NULL) {
|
538
|
+
xmlFreeTextReader(reader);
|
539
|
+
rb_raise(rb_eRuntimeError, "couldn't create a parser");
|
540
|
+
}
|
541
|
+
|
542
|
+
rb_reader = Data_Wrap_Struct(klass, NULL, dealloc, reader);
|
543
|
+
args[0] = rb_buffer;
|
544
|
+
args[1] = rb_url;
|
545
|
+
args[2] = encoding;
|
546
|
+
rb_obj_call_init(rb_reader, 3, args);
|
547
|
+
|
548
|
+
return rb_reader;
|
549
|
+
}
|
550
|
+
|
551
|
+
/*
|
552
|
+
* call-seq:
|
553
|
+
* from_io(io, url = nil, encoding = nil, options = 0)
|
554
|
+
*
|
555
|
+
* Create a new reader that parses +io+
|
556
|
+
*/
|
557
|
+
static VALUE from_io(int argc, VALUE *argv, VALUE klass)
|
558
|
+
{
|
559
|
+
VALUE rb_io, rb_url, encoding, rb_options;
|
560
|
+
xmlTextReaderPtr reader;
|
561
|
+
const char * c_url = NULL;
|
562
|
+
const char * c_encoding = NULL;
|
563
|
+
int c_options = 0;
|
564
|
+
VALUE rb_reader, args[3];
|
565
|
+
|
566
|
+
rb_scan_args(argc, argv, "13", &rb_io, &rb_url, &encoding, &rb_options);
|
567
|
+
|
568
|
+
if (!RTEST(rb_io)) rb_raise(rb_eArgError, "io cannot be nil");
|
569
|
+
if (RTEST(rb_url)) c_url = StringValueCStr(rb_url);
|
570
|
+
if (RTEST(encoding)) c_encoding = StringValueCStr(encoding);
|
571
|
+
if (RTEST(rb_options)) c_options = (int)NUM2INT(rb_options);
|
572
|
+
|
573
|
+
reader = xmlReaderForIO(
|
574
|
+
(xmlInputReadCallback)io_read_callback,
|
575
|
+
(xmlInputCloseCallback)io_close_callback,
|
576
|
+
(void *)rb_io,
|
577
|
+
c_url,
|
578
|
+
c_encoding,
|
579
|
+
c_options
|
580
|
+
);
|
581
|
+
|
582
|
+
if(reader == NULL) {
|
583
|
+
xmlFreeTextReader(reader);
|
584
|
+
rb_raise(rb_eRuntimeError, "couldn't create a parser");
|
585
|
+
}
|
586
|
+
|
587
|
+
rb_reader = Data_Wrap_Struct(klass, NULL, dealloc, reader);
|
588
|
+
args[0] = rb_io;
|
589
|
+
args[1] = rb_url;
|
590
|
+
args[2] = encoding;
|
591
|
+
rb_obj_call_init(rb_reader, 3, args);
|
592
|
+
|
593
|
+
return rb_reader;
|
594
|
+
}
|
595
|
+
|
596
|
+
/*
|
597
|
+
* call-seq:
|
598
|
+
* reader.empty_element? # => true or false
|
599
|
+
*
|
600
|
+
* Returns true if the current node is empty, otherwise false.
|
601
|
+
*/
|
602
|
+
static VALUE empty_element_p(VALUE self)
|
603
|
+
{
|
604
|
+
xmlTextReaderPtr reader;
|
605
|
+
|
606
|
+
Data_Get_Struct(self, xmlTextReader, reader);
|
607
|
+
|
608
|
+
if(xmlTextReaderIsEmptyElement(reader))
|
609
|
+
return Qtrue;
|
610
|
+
|
611
|
+
return Qfalse;
|
612
|
+
}
|
613
|
+
|
614
|
+
VALUE cNokogiriXmlReader;
|
615
|
+
|
616
|
+
void init_xml_reader()
|
617
|
+
{
|
618
|
+
VALUE module = rb_define_module("Nokogiri");
|
619
|
+
VALUE xml = rb_define_module_under(module, "XML");
|
620
|
+
|
621
|
+
/*
|
622
|
+
* The Reader parser allows you to effectively pull parse an XML document.
|
623
|
+
* Once instantiated, call Nokogiri::XML::Reader#each to iterate over each
|
624
|
+
* node. Note that you may only iterate over the document once!
|
625
|
+
*/
|
626
|
+
VALUE klass = rb_define_class_under(xml, "Reader", rb_cObject);
|
627
|
+
|
628
|
+
cNokogiriXmlReader = klass;
|
629
|
+
|
630
|
+
rb_define_singleton_method(klass, "from_memory", from_memory, -1);
|
631
|
+
rb_define_singleton_method(klass, "from_io", from_io, -1);
|
632
|
+
|
633
|
+
rb_define_method(klass, "read", read_more, 0);
|
634
|
+
rb_define_method(klass, "inner_xml", inner_xml, 0);
|
635
|
+
rb_define_method(klass, "outer_xml", outer_xml, 0);
|
636
|
+
rb_define_method(klass, "state", state, 0);
|
637
|
+
rb_define_method(klass, "node_type", node_type, 0);
|
638
|
+
rb_define_method(klass, "name", name, 0);
|
639
|
+
rb_define_method(klass, "local_name", local_name, 0);
|
640
|
+
rb_define_method(klass, "namespace_uri", namespace_uri, 0);
|
641
|
+
rb_define_method(klass, "prefix", prefix, 0);
|
642
|
+
rb_define_method(klass, "value", value, 0);
|
643
|
+
rb_define_method(klass, "lang", lang, 0);
|
644
|
+
rb_define_method(klass, "xml_version", xml_version, 0);
|
645
|
+
rb_define_method(klass, "depth", depth, 0);
|
646
|
+
rb_define_method(klass, "attribute_count", attribute_count, 0);
|
647
|
+
rb_define_method(klass, "attribute", reader_attribute, 1);
|
648
|
+
rb_define_method(klass, "namespaces", namespaces, 0);
|
649
|
+
rb_define_method(klass, "attribute_at", attribute_at, 1);
|
650
|
+
rb_define_method(klass, "empty_element?", empty_element_p, 0);
|
651
|
+
rb_define_method(klass, "attributes?", attributes_eh, 0);
|
652
|
+
rb_define_method(klass, "value?", value_eh, 0);
|
653
|
+
rb_define_method(klass, "default?", default_eh, 0);
|
654
|
+
rb_define_method(klass, "base_uri", base_uri, 0);
|
655
|
+
|
656
|
+
rb_define_private_method(klass, "attr_nodes", attribute_nodes, 0);
|
657
|
+
}
|