nokogiri 1.2.1-x86-mswin32-60 → 1.2.2-x86-mswin32-60
Sign up to get free protection for your applications and to get access to all the features.
Potentially problematic release.
This version of nokogiri might be problematic. Click here for more details.
- data/.autotest +15 -0
- data/{History.ja.txt → CHANGELOG.ja.rdoc} +30 -2
- data/{History.txt → CHANGELOG.rdoc} +28 -2
- data/Manifest.txt +13 -7
- data/{README.ja.txt → README.ja.rdoc} +3 -1
- data/{README.txt → README.rdoc} +7 -1
- data/Rakefile +8 -25
- data/ext/nokogiri/extconf.rb +4 -4
- data/ext/nokogiri/html_entity_lookup.c +30 -0
- data/ext/nokogiri/html_entity_lookup.h +8 -0
- data/ext/nokogiri/native.c +22 -0
- data/ext/nokogiri/native.h +27 -4
- data/ext/nokogiri/native.so +0 -0
- data/ext/nokogiri/xml_document.c +31 -4
- data/ext/nokogiri/xml_document.h +11 -0
- data/ext/nokogiri/xml_document_fragment.c +1 -1
- data/ext/nokogiri/xml_node.c +71 -58
- data/ext/nokogiri/xml_node_set.c +26 -0
- data/ext/nokogiri/xml_reader.c +4 -2
- data/ext/nokogiri/xml_sax_parser.c +0 -37
- data/ext/nokogiri/xml_sax_push_parser.c +2 -2
- data/ext/nokogiri/xml_xpath_context.c +34 -7
- data/lib/nokogiri.rb +25 -0
- data/lib/nokogiri/css/generated_tokenizer.rb +2 -2
- data/lib/nokogiri/css/node.rb +2 -0
- data/lib/nokogiri/css/parser.rb +3 -2
- data/lib/nokogiri/html.rb +9 -52
- data/lib/nokogiri/html/document.rb +2 -0
- data/lib/nokogiri/html/entity_lookup.rb +11 -0
- data/lib/nokogiri/version.rb +1 -1
- data/lib/nokogiri/xml.rb +1 -2
- data/lib/nokogiri/xml/builder.rb +18 -5
- data/lib/nokogiri/xml/document.rb +15 -1
- data/lib/nokogiri/xml/fragment_handler.rb +34 -0
- data/lib/nokogiri/xml/node.rb +104 -29
- data/lib/nokogiri/xml/node_set.rb +12 -10
- data/lib/nokogiri/xml/sax/parser.rb +3 -3
- data/lib/xsd/xmlparser/nokogiri.rb +53 -0
- data/tasks/test.rb +7 -5
- data/test/css/test_nthiness.rb +1 -0
- data/test/css/test_parser.rb +1 -0
- data/test/css/test_tokenizer.rb +1 -0
- data/test/css/test_xpath_visitor.rb +1 -0
- data/test/helper.rb +4 -0
- data/test/hpricot/test_alter.rb +1 -0
- data/test/html/sax/test_parser.rb +13 -0
- data/test/html/test_builder.rb +21 -0
- data/test/html/test_document.rb +36 -0
- data/test/html/test_document_encoding.rb +46 -0
- data/test/html/test_named_characters.rb +14 -0
- data/test/html/test_node.rb +80 -0
- data/test/test_convert_xpath.rb +1 -0
- data/test/test_css_cache.rb +1 -0
- data/test/test_nokogiri.rb +8 -0
- data/test/xml/sax/test_parser.rb +6 -0
- data/test/xml/sax/test_push_parser.rb +1 -0
- data/test/xml/test_builder.rb +9 -0
- data/test/xml/test_cdata.rb +1 -0
- data/test/xml/test_comment.rb +1 -0
- data/test/xml/test_document.rb +58 -0
- data/test/xml/test_document_encoding.rb +15 -14
- data/test/xml/test_document_fragment.rb +6 -0
- data/test/xml/test_dtd.rb +1 -0
- data/test/xml/test_dtd_encoding.rb +1 -0
- data/test/xml/test_entity_reference.rb +1 -0
- data/test/xml/test_node.rb +52 -4
- data/test/xml/test_node_encoding.rb +1 -0
- data/test/xml/test_node_set.rb +21 -1
- data/test/xml/test_processing_instruction.rb +1 -0
- data/test/xml/test_reader_encoding.rb +1 -0
- data/test/xml/test_unparented_node.rb +381 -0
- data/test/xml/test_xpath.rb +1 -0
- metadata +34 -16
- data/lib/nokogiri/xml/after_handler.rb +0 -18
- data/lib/nokogiri/xml/before_handler.rb +0 -33
- data/vendor/hoe.rb +0 -1020
data/.autotest
ADDED
@@ -1,11 +1,39 @@
|
|
1
|
-
=== 1.2.
|
1
|
+
=== 1.2.2 / 2009年3月14日
|
2
|
+
|
3
|
+
* 新しい機能
|
4
|
+
|
5
|
+
* Nokogiri は soap4r と一緒に使う事が可能。(XSD::XMLParser::Nokogiri 参照)
|
6
|
+
* Nokogiri::XML::Node#inner_html= はノードの中のHTMLをセット出来る
|
7
|
+
* Nokogiriのビルダーのインタフェースの改良
|
8
|
+
* Nokogiri::XML::Node#swap は、現在のノードに新しいhtmlを交換する事が出来る
|
9
|
+
|
10
|
+
|
11
|
+
* バグの修正
|
12
|
+
|
13
|
+
* ビルダーAPIのタグのネスティングを修正 (LH #41)
|
14
|
+
* Nokogiri::HTML.fragment はテキストだけのノードを適切に扱う事が出来る(LH #43)
|
15
|
+
* Nokogiri::XML::Node#before はテキストノードのはじめに挿入する事が出来る (LH #44)
|
16
|
+
* Nokogiri::XML::Node#after はテキストノードの文末に挿入する事が出来る
|
17
|
+
* Nokogiri::XML::Node#search 名前空間が自動的に登録されるようになった(LH#42)
|
18
|
+
* Nokogiri::XML::NodeSet#search 名前空間が自動的に登録されるようになった
|
19
|
+
* Nokogiri::HTML::NamedCharacters はlibxml2に委任
|
20
|
+
* Nokogiri::XML::Node#[] はSymbolを使う (LH #48)
|
21
|
+
* vasprintf にwindowsを修正 (Geffroy Couprie ありがとう!)
|
22
|
+
* Nokogiri::XML::Node#[]= はentityを符号化しない (LH #55)
|
23
|
+
* 名前空間はreparentedのノードに模写する (LH #56)
|
24
|
+
* StringのエンコードはRuby 1.9での初期設定を使用する
|
25
|
+
* Document#dup は新しいドキュメントに同じタイプを作る (LH #59)
|
26
|
+
* Document#parent は存在しない (LH #64)
|
27
|
+
|
28
|
+
|
29
|
+
=== 1.2.1 / 2009年2月23日
|
2
30
|
|
3
31
|
* 修正
|
4
32
|
|
5
33
|
* CSS のセレクターのスペースを修正
|
6
34
|
* Ruby 1.9 のStringのエンコードを修正 (角谷さんに感謝!)
|
7
35
|
|
8
|
-
=== 1.2.0 /
|
36
|
+
=== 1.2.0 / 2009年2月22日
|
9
37
|
|
10
38
|
* 新しい機能
|
11
39
|
* CSSサーチが CSS3 名前空間クエリをサポートするようになった
|
@@ -1,11 +1,37 @@
|
|
1
|
-
=== 1.2.
|
1
|
+
=== 1.2.2 / 2009-03-14
|
2
|
+
|
3
|
+
* New features
|
4
|
+
|
5
|
+
* Nokogiri may be used with soap4r. See XSD::XMLParser::Nokogiri
|
6
|
+
* Nokogiri::XML::Node#inner_html= to set the inner html for a node
|
7
|
+
* Nokogiri builder interface improvements
|
8
|
+
* Nokogiri::XML::Node#swap swaps html for current node (LH #50)
|
9
|
+
|
10
|
+
* Bugfixes
|
11
|
+
|
12
|
+
* Fixed a tag nesting problem in the Builder API (LH #41)
|
13
|
+
* Nokogiri::HTML.fragment will properly handle text only nodes (LH #43)
|
14
|
+
* Nokogiri::XML::Node#before will prepend text nodes (LH #44)
|
15
|
+
* Nokogiri::XML::Node#after will append text nodes
|
16
|
+
* Nokogiri::XML::Node#search automatically registers root namepsaces (LH #42)
|
17
|
+
* Nokogiri::XML::NodeSet#search automatically registers namespaces
|
18
|
+
* Nokogiri::HTML::NamedCharacters delegates to libxml2
|
19
|
+
* Nokogiri::XML::Node#[] can take a symbol (LH #48)
|
20
|
+
* vasprintf for windows updated. Thanks Geoffroy Couprie!
|
21
|
+
* Nokogiri::XML::Node#[]= should not encode entities (LH #55)
|
22
|
+
* Namespaces should be copied to reparented nodes (LH #56)
|
23
|
+
* Nokogiri uses encoding set on the string for default in Ruby 1.9
|
24
|
+
* Document#dup should create a new document of the same type (LH #59)
|
25
|
+
* Document should not have a parent method (LH #64)
|
26
|
+
|
27
|
+
=== 1.2.1 / 2009-02-23
|
2
28
|
|
3
29
|
* Bugfixes
|
4
30
|
|
5
31
|
* Fixed a CSS selector space bug
|
6
32
|
* Fixed Ruby 1.9 String Encoding (Thanks 角谷さん!)
|
7
33
|
|
8
|
-
=== 1.2.0 /
|
34
|
+
=== 1.2.0 / 2009-02-22
|
9
35
|
|
10
36
|
* New features
|
11
37
|
|
data/Manifest.txt
CHANGED
@@ -1,12 +1,15 @@
|
|
1
|
-
|
2
|
-
|
1
|
+
.autotest
|
2
|
+
CHANGELOG.ja.rdoc
|
3
|
+
CHANGELOG.rdoc
|
3
4
|
Manifest.txt
|
4
|
-
README.ja.
|
5
|
-
README.
|
5
|
+
README.ja.rdoc
|
6
|
+
README.rdoc
|
6
7
|
Rakefile
|
7
8
|
ext/nokogiri/extconf.rb
|
8
9
|
ext/nokogiri/html_document.c
|
9
10
|
ext/nokogiri/html_document.h
|
11
|
+
ext/nokogiri/html_entity_lookup.c
|
12
|
+
ext/nokogiri/html_entity_lookup.h
|
10
13
|
ext/nokogiri/html_sax_parser.c
|
11
14
|
ext/nokogiri/html_sax_parser.h
|
12
15
|
ext/nokogiri/native.c
|
@@ -71,13 +74,12 @@ lib/nokogiri/hpricot.rb
|
|
71
74
|
lib/nokogiri/html.rb
|
72
75
|
lib/nokogiri/html/builder.rb
|
73
76
|
lib/nokogiri/html/document.rb
|
77
|
+
lib/nokogiri/html/entity_lookup.rb
|
74
78
|
lib/nokogiri/html/sax/parser.rb
|
75
79
|
lib/nokogiri/syntax_error.rb
|
76
80
|
lib/nokogiri/version.rb
|
77
81
|
lib/nokogiri/xml.rb
|
78
|
-
lib/nokogiri/xml/after_handler.rb
|
79
82
|
lib/nokogiri/xml/attr.rb
|
80
|
-
lib/nokogiri/xml/before_handler.rb
|
81
83
|
lib/nokogiri/xml/builder.rb
|
82
84
|
lib/nokogiri/xml/cdata.rb
|
83
85
|
lib/nokogiri/xml/comment.rb
|
@@ -86,6 +88,7 @@ lib/nokogiri/xml/document_fragment.rb
|
|
86
88
|
lib/nokogiri/xml/dtd.rb
|
87
89
|
lib/nokogiri/xml/element.rb
|
88
90
|
lib/nokogiri/xml/entity_declaration.rb
|
91
|
+
lib/nokogiri/xml/fragment_handler.rb
|
89
92
|
lib/nokogiri/xml/node.rb
|
90
93
|
lib/nokogiri/xml/node/save_options.rb
|
91
94
|
lib/nokogiri/xml/node_set.rb
|
@@ -103,6 +106,7 @@ lib/nokogiri/xml/xpath/syntax_error.rb
|
|
103
106
|
lib/nokogiri/xml/xpath_context.rb
|
104
107
|
lib/nokogiri/xslt.rb
|
105
108
|
lib/nokogiri/xslt/stylesheet.rb
|
109
|
+
lib/xsd/xmlparser/nokogiri.rb
|
106
110
|
tasks/test.rb
|
107
111
|
test/css/test_nthiness.rb
|
108
112
|
test/css/test_parser.rb
|
@@ -135,6 +139,8 @@ test/hpricot/test_xml.rb
|
|
135
139
|
test/html/sax/test_parser.rb
|
136
140
|
test/html/test_builder.rb
|
137
141
|
test/html/test_document.rb
|
142
|
+
test/html/test_document_encoding.rb
|
143
|
+
test/html/test_named_characters.rb
|
138
144
|
test/html/test_node.rb
|
139
145
|
test/test_convert_xpath.rb
|
140
146
|
test/test_css_cache.rb
|
@@ -162,5 +168,5 @@ test/xml/test_node_set.rb
|
|
162
168
|
test/xml/test_processing_instruction.rb
|
163
169
|
test/xml/test_reader_encoding.rb
|
164
170
|
test/xml/test_text.rb
|
171
|
+
test/xml/test_unparented_node.rb
|
165
172
|
test/xml/test_xpath.rb
|
166
|
-
vendor/hoe.rb
|
data/{README.txt → README.rdoc}
RENAMED
@@ -42,8 +42,12 @@ The bug tracker is available here:
|
|
42
42
|
require 'nokogiri'
|
43
43
|
require 'open-uri'
|
44
44
|
|
45
|
+
# Get a Nokogiri::HTML:Document for the page we’re interested in...
|
46
|
+
|
45
47
|
doc = Nokogiri::HTML(open('http://www.google.com/search?q=tenderlove'))
|
46
48
|
|
49
|
+
# Do funky things with it using Nokogiri::XML::Node methods...
|
50
|
+
|
47
51
|
####
|
48
52
|
# Search for nodes by css
|
49
53
|
doc.css('h3.r a.l').each do |link|
|
@@ -66,8 +70,10 @@ The bug tracker is available here:
|
|
66
70
|
== REQUIREMENTS:
|
67
71
|
|
68
72
|
* ruby 1.8 or 1.9
|
69
|
-
*
|
73
|
+
* libxml2
|
74
|
+
* libxml2-dev
|
70
75
|
* libxslt
|
76
|
+
* libxslt-dev
|
71
77
|
|
72
78
|
== INSTALL:
|
73
79
|
|
data/Rakefile
CHANGED
@@ -2,6 +2,7 @@
|
|
2
2
|
|
3
3
|
require 'rubygems'
|
4
4
|
require 'rake'
|
5
|
+
require 'hoe'
|
5
6
|
|
6
7
|
|
7
8
|
kind = Config::CONFIG['DLEXT']
|
@@ -10,8 +11,6 @@ windows = RUBY_PLATFORM =~ /mswin/i ? true : false
|
|
10
11
|
LIB_DIR = File.expand_path(File.join(File.dirname(__FILE__), 'lib'))
|
11
12
|
$LOAD_PATH << LIB_DIR
|
12
13
|
|
13
|
-
require 'vendor/hoe'
|
14
|
-
|
15
14
|
GENERATED_PARSER = "lib/nokogiri/css/generated_parser.rb"
|
16
15
|
GENERATED_TOKENIZER = "lib/nokogiri/css/generated_tokenizer.rb"
|
17
16
|
|
@@ -22,6 +21,9 @@ require 'nokogiri/version'
|
|
22
21
|
HOE = Hoe.new('nokogiri', Nokogiri::VERSION) do |p|
|
23
22
|
p.developer('Aaron Patterson', 'aaronp@rubyforge.org')
|
24
23
|
p.developer('Mike Dalessio', 'mike.dalessio@gmail.com')
|
24
|
+
p.readme_file = ['README', ENV['HLANG'], 'rdoc'].compact.join('.')
|
25
|
+
p.history_file = ['CHANGELOG', ENV['HLANG'], 'rdoc'].compact.join('.')
|
26
|
+
p.extra_rdoc_files = FileList['*.rdoc']
|
25
27
|
p.clean_globs = [
|
26
28
|
'ext/nokogiri/Makefile',
|
27
29
|
'ext/nokogiri/*.{o,so,bundle,a,log,dll}',
|
@@ -102,7 +104,7 @@ file GENERATED_TOKENIZER => "lib/nokogiri/css/tokenizer.rex" do |t|
|
|
102
104
|
begin
|
103
105
|
sh "frex --independent -o #{t.name} #{t.prerequisites.first}"
|
104
106
|
rescue
|
105
|
-
abort "need frex, sudo gem install
|
107
|
+
abort "need frex, sudo gem install tenderlove-frex -s http://gems.github.com"
|
106
108
|
end
|
107
109
|
end
|
108
110
|
|
@@ -254,31 +256,12 @@ namespace :install do
|
|
254
256
|
desc "Install frex and racc for development"
|
255
257
|
task :deps => %w(frex racc)
|
256
258
|
|
257
|
-
|
258
|
-
|
259
|
-
file "stash/racc-1.4.5-all.tar.gz" => "stash" do |t|
|
260
|
-
puts "Downloading racc to #{t.name}..."
|
261
|
-
|
262
|
-
Dir.chdir File.dirname(t.name) do
|
263
|
-
url = "http://i.loveruby.net/archive/racc/racc-1.4.5-all.tar.gz"
|
264
|
-
system "wget #{url} || curl -O #{url}"
|
265
|
-
end
|
266
|
-
end
|
267
|
-
|
268
|
-
task :racc => "stash/racc-1.4.5-all.tar.gz" do |t|
|
269
|
-
sh "tar xvf #{t.prerequisites.first} -C stash"
|
270
|
-
|
271
|
-
Dir.chdir "stash/#{File.basename(t.prerequisites.first, ".tar.gz")}" do
|
272
|
-
sh "ruby setup.rb config"
|
273
|
-
sh "ruby setup.rb setup"
|
274
|
-
sh "sudo ruby setup.rb install"
|
275
|
-
end
|
276
|
-
|
277
|
-
puts "The racc binary is likely in #{::Config::CONFIG["bindir"]}."
|
259
|
+
task :racc do |t|
|
260
|
+
sh "sudo gem install racc"
|
278
261
|
end
|
279
262
|
|
280
263
|
task :frex do
|
281
|
-
sh "sudo gem install
|
264
|
+
sh "sudo gem install tenderlove-frex -s http://gems.github.com"
|
282
265
|
end
|
283
266
|
end
|
284
267
|
|
data/ext/nokogiri/extconf.rb
CHANGED
@@ -57,10 +57,10 @@ unless find_header('libxml/parser.h', *HEADER_DIRS)
|
|
57
57
|
end
|
58
58
|
|
59
59
|
unless find_header('libxslt/xslt.h', *HEADER_DIRS)
|
60
|
-
abort "libxslt is missing. try 'port install libxslt' or 'yum install libxslt'"
|
60
|
+
abort "libxslt is missing. try 'port install libxslt' or 'yum install libxslt-devel'"
|
61
61
|
end
|
62
62
|
unless find_header('libexslt/exslt.h', *HEADER_DIRS)
|
63
|
-
abort "libxslt is missing. try 'port install libxslt' or 'yum install libxslt'"
|
63
|
+
abort "libxslt is missing. try 'port install libxslt' or 'yum install libxslt-devel'"
|
64
64
|
end
|
65
65
|
|
66
66
|
unless find_library('xml2', 'xmlParseDoc', *LIB_DIRS)
|
@@ -68,11 +68,11 @@ unless find_library('xml2', 'xmlParseDoc', *LIB_DIRS)
|
|
68
68
|
end
|
69
69
|
|
70
70
|
unless find_library('xslt', 'xsltParseStylesheetDoc', *LIB_DIRS)
|
71
|
-
abort "libxslt is missing. try 'port install libxslt' or 'yum install libxslt'"
|
71
|
+
abort "libxslt is missing. try 'port install libxslt' or 'yum install libxslt-devel'"
|
72
72
|
end
|
73
73
|
|
74
74
|
unless find_library('exslt', 'exsltFuncRegister', *LIB_DIRS)
|
75
|
-
abort "libxslt is missing. try 'port install libxslt' or 'yum install libxslt'"
|
75
|
+
abort "libxslt is missing. try 'port install libxslt' or 'yum install libxslt-devel'"
|
76
76
|
end
|
77
77
|
|
78
78
|
create_makefile('nokogiri/native')
|
@@ -0,0 +1,30 @@
|
|
1
|
+
#include <html_entity_lookup.h>
|
2
|
+
|
3
|
+
/*
|
4
|
+
* call-seq:
|
5
|
+
* get(key)
|
6
|
+
*
|
7
|
+
* Get the HTML::EntityDescription for +key+
|
8
|
+
*/
|
9
|
+
static VALUE get(VALUE self, VALUE key)
|
10
|
+
{
|
11
|
+
const htmlEntityDesc * desc =
|
12
|
+
htmlEntityLookup((const xmlChar *)StringValuePtr(key));
|
13
|
+
|
14
|
+
if(NULL == desc) return Qnil;
|
15
|
+
VALUE klass = rb_const_get(mNokogiriHtml, rb_intern("EntityDescription"));
|
16
|
+
|
17
|
+
return rb_funcall(klass, rb_intern("new"), 3,
|
18
|
+
INT2NUM((int)desc->value),
|
19
|
+
NOKOGIRI_STR_NEW2(desc->name, "UTF-8"),
|
20
|
+
NOKOGIRI_STR_NEW2(desc->desc, "UTF-8"));
|
21
|
+
}
|
22
|
+
|
23
|
+
void init_html_entity_lookup()
|
24
|
+
{
|
25
|
+
VALUE nokogiri = rb_define_module("Nokogiri");
|
26
|
+
VALUE html = rb_define_module_under(nokogiri, "HTML");
|
27
|
+
VALUE klass = rb_define_class_under(html, "EntityLookup", rb_cObject);
|
28
|
+
|
29
|
+
rb_define_method(klass, "get", get, 1);
|
30
|
+
}
|
data/ext/nokogiri/native.c
CHANGED
@@ -7,6 +7,23 @@ VALUE mNokogiriXslt ;
|
|
7
7
|
VALUE mNokogiriXmlSax ;
|
8
8
|
VALUE mNokogiriHtmlSax ;
|
9
9
|
|
10
|
+
#ifdef XP_WIN
|
11
|
+
/*
|
12
|
+
* I srsly hate windows. it doesn't have vasprintf.
|
13
|
+
* Thank you Geoffroy Couprie for this implementation of vasprintf!
|
14
|
+
*/
|
15
|
+
int vasprintf (char **strp, const char *fmt, va_list ap)
|
16
|
+
{
|
17
|
+
int len = vsnprintf (NULL, 0, fmt, ap) + 1;
|
18
|
+
char *res = (char *)malloc((unsigned int)len);
|
19
|
+
if (res == NULL)
|
20
|
+
return -1;
|
21
|
+
*strp = res;
|
22
|
+
return vsnprintf(res, (unsigned int)len, fmt, ap);
|
23
|
+
}
|
24
|
+
#endif
|
25
|
+
|
26
|
+
|
10
27
|
void Init_native()
|
11
28
|
{
|
12
29
|
xmlMemSetup(
|
@@ -27,6 +44,10 @@ void Init_native()
|
|
27
44
|
rb_intern("LIBXML_VERSION"),
|
28
45
|
NOKOGIRI_STR_NEW2(LIBXML_DOTTED_VERSION, "UTF-8")
|
29
46
|
);
|
47
|
+
rb_const_set( mNokogiri,
|
48
|
+
rb_intern("LIBXML_PARSER_VERSION"),
|
49
|
+
NOKOGIRI_STR_NEW2(xmlParserVersion, "UTF-8")
|
50
|
+
);
|
30
51
|
|
31
52
|
init_xml_document();
|
32
53
|
init_html_document();
|
@@ -48,4 +69,5 @@ void Init_native()
|
|
48
69
|
init_html_sax_parser();
|
49
70
|
init_xslt_stylesheet();
|
50
71
|
init_xml_syntax_error();
|
72
|
+
init_html_entity_lookup();
|
51
73
|
}
|
data/ext/nokogiri/native.h
CHANGED
@@ -30,16 +30,26 @@
|
|
30
30
|
#define NOKOGIRI_STR_NEW2(str, encoding) \
|
31
31
|
({ \
|
32
32
|
VALUE _string = rb_str_new2((const char *)str); \
|
33
|
-
if(NULL != encoding) \
|
34
|
-
|
33
|
+
if(NULL != encoding) { \
|
34
|
+
int _enc = rb_enc_find_index(encoding); \
|
35
|
+
if(_enc == -1) \
|
36
|
+
rb_enc_associate_index(_string, rb_enc_find_index("ASCII")); \
|
37
|
+
else \
|
38
|
+
rb_enc_associate_index(_string, _enc); \
|
39
|
+
} \
|
35
40
|
_string; \
|
36
41
|
})
|
37
42
|
|
38
43
|
#define NOKOGIRI_STR_NEW(str, len, encoding) \
|
39
44
|
({ \
|
40
45
|
VALUE _string = rb_str_new((const char *)str, (long)len); \
|
41
|
-
if(NULL != encoding) \
|
42
|
-
|
46
|
+
if(NULL != encoding) { \
|
47
|
+
int _enc = rb_enc_find_index(encoding); \
|
48
|
+
if(_enc == -1) \
|
49
|
+
rb_enc_associate_index(_string, rb_enc_find_index("ASCII")); \
|
50
|
+
else \
|
51
|
+
rb_enc_associate_index(_string, _enc); \
|
52
|
+
} \
|
43
53
|
_string; \
|
44
54
|
})
|
45
55
|
|
@@ -52,8 +62,13 @@
|
|
52
62
|
rb_str_new((const char *)str, (long)len)
|
53
63
|
#endif
|
54
64
|
|
65
|
+
#ifdef XP_WIN
|
66
|
+
int vasprintf (char **strp, const char *fmt, va_list ap);
|
67
|
+
#endif
|
68
|
+
|
55
69
|
#include <xml_io.h>
|
56
70
|
#include <xml_document.h>
|
71
|
+
#include <html_entity_lookup.h>
|
57
72
|
#include <html_document.h>
|
58
73
|
#include <xml_node.h>
|
59
74
|
#include <xml_text.h>
|
@@ -81,6 +96,14 @@ extern VALUE mNokogiriHtml ;
|
|
81
96
|
extern VALUE mNokogiriHtmlSax ;
|
82
97
|
extern VALUE mNokogiriXslt ;
|
83
98
|
|
99
|
+
#define NOKOGIRI_ROOT_NODE(_node) \
|
100
|
+
({ \
|
101
|
+
nokogiriTuplePtr tuple = (nokogiriTuplePtr)(_node->doc->_private); \
|
102
|
+
xmlNodeSetPtr node_set = (xmlNodeSetPtr)(tuple->unlinkedNodes); \
|
103
|
+
xmlXPathNodeSetAdd(node_set, _node); \
|
104
|
+
_node; \
|
105
|
+
})
|
106
|
+
|
84
107
|
#ifdef DEBUG
|
85
108
|
|
86
109
|
#define NOKOGIRI_DEBUG_START(p) if (getenv("NOKOGIRI_NO_FREE")) return ; if (getenv("NOKOGIRI_DEBUG")) fprintf(stderr,"nokogiri: %s:%d %p start\n", __FILE__, __LINE__, p);
|
data/ext/nokogiri/native.so
CHANGED
Binary file
|
data/ext/nokogiri/xml_document.c
CHANGED
@@ -3,8 +3,23 @@
|
|
3
3
|
static void dealloc(xmlDocPtr doc)
|
4
4
|
{
|
5
5
|
NOKOGIRI_DEBUG_START(doc);
|
6
|
+
|
7
|
+
nokogiriTuplePtr tuple = doc->_private;
|
8
|
+
xmlNodeSetPtr node_set = tuple->unlinkedNodes;
|
9
|
+
|
10
|
+
int i;
|
11
|
+
for(i = 0; i < node_set->nodeNr; i++) {
|
12
|
+
xmlAddChild((xmlNodePtr)doc, node_set->nodeTab[i]);
|
13
|
+
}
|
14
|
+
|
15
|
+
if (node_set->nodeTab != NULL)
|
16
|
+
xmlFree(node_set->nodeTab);
|
17
|
+
xmlFree(node_set);
|
18
|
+
|
19
|
+
free(doc->_private);
|
6
20
|
doc->_private = NULL;
|
7
21
|
xmlFreeDoc(doc);
|
22
|
+
|
8
23
|
NOKOGIRI_DEBUG_END(doc);
|
9
24
|
}
|
10
25
|
|
@@ -187,7 +202,11 @@ static VALUE duplicate_node(int argc, VALUE *argv, VALUE self)
|
|
187
202
|
dup = xmlCopyDoc(doc, NUM2INT(level));
|
188
203
|
if(dup == NULL) return Qnil;
|
189
204
|
|
190
|
-
|
205
|
+
dup->type = doc->type;
|
206
|
+
if(dup->type == XML_DOCUMENT_NODE)
|
207
|
+
return Nokogiri_wrap_xml_document(cNokogiriXmlDocument, dup);
|
208
|
+
else
|
209
|
+
return Nokogiri_wrap_xml_document(cNokogiriHtmlDocument, dup);
|
191
210
|
}
|
192
211
|
|
193
212
|
/*
|
@@ -255,7 +274,6 @@ void init_xml_document()
|
|
255
274
|
rb_define_method(klass, "encoding", encoding, 0);
|
256
275
|
rb_define_method(klass, "dup", duplicate_node, -1);
|
257
276
|
rb_define_method(klass, "url", url, 0);
|
258
|
-
rb_undef_method(klass, "parent");
|
259
277
|
}
|
260
278
|
|
261
279
|
|
@@ -263,10 +281,19 @@ void init_xml_document()
|
|
263
281
|
VALUE Nokogiri_wrap_xml_document(VALUE klass, xmlDocPtr doc)
|
264
282
|
{
|
265
283
|
VALUE rb_doc = Qnil;
|
284
|
+
nokogiriTuplePtr tuple = (nokogiriTuplePtr)malloc(sizeof(nokogiriTuple));
|
266
285
|
|
267
|
-
rb_doc = Data_Wrap_Struct(
|
286
|
+
rb_doc = Data_Wrap_Struct(
|
287
|
+
klass ? klass : cNokogiriXmlDocument,
|
288
|
+
0,
|
289
|
+
dealloc,
|
290
|
+
doc
|
291
|
+
);
|
268
292
|
rb_iv_set(rb_doc, "@decorators", Qnil);
|
269
|
-
|
293
|
+
|
294
|
+
tuple->doc = (void *)rb_doc;
|
295
|
+
tuple->unlinkedNodes = xmlXPathNodeSetCreate(NULL);
|
296
|
+
doc->_private = tuple ;
|
270
297
|
|
271
298
|
return rb_doc ;
|
272
299
|
}
|