nokogiri-backupify 1.5.0.beta.4
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 +509 -0
- data/CHANGELOG.rdoc +490 -0
- data/Manifest.txt +274 -0
- data/README.ja.rdoc +106 -0
- data/README.rdoc +150 -0
- data/Rakefile +217 -0
- data/bin/nokogiri +54 -0
- data/deps.rip +5 -0
- data/ext/java/nokogiri/EncodingHandler.java +124 -0
- data/ext/java/nokogiri/HtmlDocument.java +146 -0
- data/ext/java/nokogiri/HtmlElementDescription.java +145 -0
- data/ext/java/nokogiri/HtmlEntityLookup.java +79 -0
- data/ext/java/nokogiri/HtmlSaxParserContext.java +256 -0
- data/ext/java/nokogiri/NokogiriService.java +466 -0
- data/ext/java/nokogiri/XmlAttr.java +183 -0
- data/ext/java/nokogiri/XmlAttributeDecl.java +130 -0
- data/ext/java/nokogiri/XmlCdata.java +89 -0
- data/ext/java/nokogiri/XmlComment.java +84 -0
- data/ext/java/nokogiri/XmlDocument.java +514 -0
- data/ext/java/nokogiri/XmlDocumentFragment.java +216 -0
- data/ext/java/nokogiri/XmlDtd.java +464 -0
- data/ext/java/nokogiri/XmlElement.java +221 -0
- data/ext/java/nokogiri/XmlElementContent.java +382 -0
- data/ext/java/nokogiri/XmlElementDecl.java +147 -0
- data/ext/java/nokogiri/XmlEntityDecl.java +161 -0
- data/ext/java/nokogiri/XmlEntityReference.java +75 -0
- data/ext/java/nokogiri/XmlNamespace.java +127 -0
- data/ext/java/nokogiri/XmlNode.java +1392 -0
- data/ext/java/nokogiri/XmlNodeSet.java +284 -0
- data/ext/java/nokogiri/XmlProcessingInstruction.java +103 -0
- data/ext/java/nokogiri/XmlReader.java +409 -0
- data/ext/java/nokogiri/XmlRelaxng.java +199 -0
- data/ext/java/nokogiri/XmlSaxParserContext.java +353 -0
- data/ext/java/nokogiri/XmlSaxPushParser.java +182 -0
- data/ext/java/nokogiri/XmlSchema.java +175 -0
- data/ext/java/nokogiri/XmlSyntaxError.java +114 -0
- data/ext/java/nokogiri/XmlText.java +135 -0
- data/ext/java/nokogiri/XmlXpathContext.java +175 -0
- data/ext/java/nokogiri/XsltStylesheet.java +181 -0
- data/ext/java/nokogiri/internals/HtmlDomParserContext.java +205 -0
- data/ext/java/nokogiri/internals/NokogiriDocumentCache.java +73 -0
- data/ext/java/nokogiri/internals/NokogiriErrorHandler.java +80 -0
- data/ext/java/nokogiri/internals/NokogiriHandler.java +326 -0
- data/ext/java/nokogiri/internals/NokogiriHelpers.java +583 -0
- data/ext/java/nokogiri/internals/NokogiriNamespaceCache.java +170 -0
- data/ext/java/nokogiri/internals/NokogiriNamespaceContext.java +118 -0
- data/ext/java/nokogiri/internals/NokogiriNonStrictErrorHandler.java +73 -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 +120 -0
- data/ext/java/nokogiri/internals/NokogiriXPathFunctionResolver.java +56 -0
- data/ext/java/nokogiri/internals/ParserContext.java +278 -0
- data/ext/java/nokogiri/internals/PushInputStream.java +411 -0
- data/ext/java/nokogiri/internals/ReaderNode.java +473 -0
- data/ext/java/nokogiri/internals/SaveContext.java +282 -0
- data/ext/java/nokogiri/internals/SchemaErrorHandler.java +68 -0
- data/ext/java/nokogiri/internals/XmlDeclHandler.java +42 -0
- data/ext/java/nokogiri/internals/XmlDomParser.java +77 -0
- data/ext/java/nokogiri/internals/XmlDomParserContext.java +233 -0
- data/ext/java/nokogiri/internals/XmlSaxParser.java +65 -0
- data/ext/java/nokogiri/internals/XsltExtensionFunction.java +72 -0
- data/ext/nokogiri/depend +358 -0
- data/ext/nokogiri/extconf.rb +124 -0
- data/ext/nokogiri/html_document.c +154 -0
- data/ext/nokogiri/html_document.h +10 -0
- data/ext/nokogiri/html_element_description.c +276 -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 +94 -0
- data/ext/nokogiri/html_sax_parser_context.h +11 -0
- data/ext/nokogiri/nokogiri.c +92 -0
- data/ext/nokogiri/nokogiri.h +160 -0
- data/ext/nokogiri/xml_attr.c +94 -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 +56 -0
- data/ext/nokogiri/xml_cdata.h +9 -0
- data/ext/nokogiri/xml_comment.c +54 -0
- data/ext/nokogiri/xml_comment.h +9 -0
- data/ext/nokogiri/xml_document.c +478 -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 +31 -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 +84 -0
- data/ext/nokogiri/xml_namespace.h +13 -0
- data/ext/nokogiri/xml_node.c +1384 -0
- data/ext/nokogiri/xml_node.h +13 -0
- data/ext/nokogiri/xml_node_set.c +418 -0
- data/ext/nokogiri/xml_node_set.h +9 -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 +684 -0
- data/ext/nokogiri/xml_reader.h +10 -0
- data/ext/nokogiri/xml_relax_ng.c +161 -0
- data/ext/nokogiri/xml_relax_ng.h +9 -0
- data/ext/nokogiri/xml_sax_parser.c +288 -0
- data/ext/nokogiri/xml_sax_parser.h +39 -0
- data/ext/nokogiri/xml_sax_parser_context.c +199 -0
- data/ext/nokogiri/xml_sax_parser_context.h +10 -0
- data/ext/nokogiri/xml_sax_push_parser.c +115 -0
- data/ext/nokogiri/xml_sax_push_parser.h +9 -0
- data/ext/nokogiri/xml_schema.c +205 -0
- data/ext/nokogiri/xml_schema.h +9 -0
- data/ext/nokogiri/xml_syntax_error.c +58 -0
- data/ext/nokogiri/xml_syntax_error.h +13 -0
- data/ext/nokogiri/xml_text.c +50 -0
- data/ext/nokogiri/xml_text.h +9 -0
- data/ext/nokogiri/xml_xpath_context.c +309 -0
- data/ext/nokogiri/xml_xpath_context.h +9 -0
- data/ext/nokogiri/xslt_stylesheet.c +258 -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 +143 -0
- data/lib/nokogiri/css.rb +23 -0
- data/lib/nokogiri/css/node.rb +99 -0
- data/lib/nokogiri/css/parser.rb +677 -0
- data/lib/nokogiri/css/parser.y +237 -0
- data/lib/nokogiri/css/parser_extras.rb +91 -0
- data/lib/nokogiri/css/syntax_error.rb +7 -0
- data/lib/nokogiri/css/tokenizer.rb +152 -0
- data/lib/nokogiri/css/tokenizer.rex +55 -0
- data/lib/nokogiri/css/xpath_visitor.rb +171 -0
- data/lib/nokogiri/decorators/slop.rb +35 -0
- data/lib/nokogiri/html.rb +36 -0
- data/lib/nokogiri/html/builder.rb +35 -0
- data/lib/nokogiri/html/document.rb +221 -0
- data/lib/nokogiri/html/document_fragment.rb +41 -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 +52 -0
- data/lib/nokogiri/html/sax/parser_context.rb +16 -0
- data/lib/nokogiri/syntax_error.rb +4 -0
- data/lib/nokogiri/version.rb +35 -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 +418 -0
- data/lib/nokogiri/xml/cdata.rb +11 -0
- data/lib/nokogiri/xml/character_data.rb +7 -0
- data/lib/nokogiri/xml/document.rb +218 -0
- data/lib/nokogiri/xml/document_fragment.rb +84 -0
- data/lib/nokogiri/xml/dtd.rb +22 -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 +19 -0
- data/lib/nokogiri/xml/namespace.rb +13 -0
- data/lib/nokogiri/xml/node.rb +907 -0
- data/lib/nokogiri/xml/node/save_options.rb +45 -0
- data/lib/nokogiri/xml/node_set.rb +350 -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 +112 -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 +164 -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 +57 -0
- data/lib/nokogiri/xml/syntax_error.rb +47 -0
- data/lib/nokogiri/xml/text.rb +9 -0
- data/lib/nokogiri/xml/xpath.rb +10 -0
- data/lib/nokogiri/xml/xpath/syntax_error.rb +11 -0
- data/lib/nokogiri/xml/xpath_context.rb +16 -0
- data/lib/nokogiri/xslt.rb +52 -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/cross_compile.rb +177 -0
- data/tasks/test.rb +94 -0
- data/test/css/test_nthiness.rb +159 -0
- data/test/css/test_parser.rb +303 -0
- data/test/css/test_tokenizer.rb +198 -0
- data/test/css/test_xpath_visitor.rb +85 -0
- data/test/decorators/test_slop.rb +16 -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 +171 -0
- data/test/html/sax/test_parser.rb +136 -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 +457 -0
- data/test/html/test_document_encoding.rb +123 -0
- data/test/html/test_document_fragment.rb +255 -0
- data/test/html/test_element_description.rb +100 -0
- data/test/html/test_named_characters.rb +14 -0
- data/test/html/test_node.rb +190 -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_memory_leak.rb +52 -0
- data/test/test_nokogiri.rb +132 -0
- data/test/test_reader.rb +403 -0
- data/test/test_soap4r_sax.rb +52 -0
- data/test/test_xslt_transforms.rb +189 -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 +338 -0
- data/test/xml/sax/test_parser_context.rb +113 -0
- data/test/xml/sax/test_push_parser.rb +156 -0
- data/test/xml/test_attr.rb +65 -0
- data/test/xml/test_attribute_decl.rb +86 -0
- data/test/xml/test_builder.rb +210 -0
- data/test/xml/test_cdata.rb +50 -0
- data/test/xml/test_comment.rb +29 -0
- data/test/xml/test_document.rb +675 -0
- data/test/xml/test_document_encoding.rb +26 -0
- data/test/xml/test_document_fragment.rb +192 -0
- data/test/xml/test_dtd.rb +107 -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 +122 -0
- data/test/xml/test_entity_reference.rb +21 -0
- data/test/xml/test_namespace.rb +70 -0
- data/test/xml/test_node.rb +899 -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 +321 -0
- data/test/xml/test_node_set.rb +708 -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 +47 -0
- data/test/xml/test_unparented_node.rb +381 -0
- data/test/xml/test_xpath.rb +237 -0
- data/test/xslt/test_custom_functions.rb +94 -0
- metadata +525 -0
@@ -0,0 +1,457 @@
|
|
1
|
+
require "helper"
|
2
|
+
|
3
|
+
module Nokogiri
|
4
|
+
module HTML
|
5
|
+
class TestDocument < Nokogiri::TestCase
|
6
|
+
def setup
|
7
|
+
super
|
8
|
+
@html = Nokogiri::HTML.parse(File.read(HTML_FILE))
|
9
|
+
end
|
10
|
+
|
11
|
+
def test_exceptions_remove_newlines
|
12
|
+
errors = @html.errors
|
13
|
+
assert errors.length > 0, 'has errors'
|
14
|
+
errors.each do |error|
|
15
|
+
assert_equal(error.to_s.chomp, error.to_s)
|
16
|
+
end
|
17
|
+
end
|
18
|
+
|
19
|
+
def test_fragment
|
20
|
+
fragment = @html.fragment
|
21
|
+
assert_equal 0, fragment.children.length
|
22
|
+
end
|
23
|
+
|
24
|
+
def test_document_takes_config_block
|
25
|
+
options = nil
|
26
|
+
Nokogiri::HTML(File.read(HTML_FILE), HTML_FILE) do |cfg|
|
27
|
+
options = cfg
|
28
|
+
options.nonet.nowarning.dtdattr
|
29
|
+
end
|
30
|
+
assert options.nonet?
|
31
|
+
assert options.nowarning?
|
32
|
+
assert options.dtdattr?
|
33
|
+
end
|
34
|
+
|
35
|
+
def test_parse_takes_config_block
|
36
|
+
options = nil
|
37
|
+
Nokogiri::HTML.parse(File.read(HTML_FILE), HTML_FILE) do |cfg|
|
38
|
+
options = cfg
|
39
|
+
options.nonet.nowarning.dtdattr
|
40
|
+
end
|
41
|
+
assert options.nonet?
|
42
|
+
assert options.nowarning?
|
43
|
+
assert options.dtdattr?
|
44
|
+
end
|
45
|
+
|
46
|
+
def test_subclass
|
47
|
+
klass = Class.new(Nokogiri::HTML::Document)
|
48
|
+
doc = klass.new
|
49
|
+
assert_instance_of klass, doc
|
50
|
+
end
|
51
|
+
|
52
|
+
def test_subclass_initialize
|
53
|
+
klass = Class.new(Nokogiri::HTML::Document) do
|
54
|
+
attr_accessor :initialized_with
|
55
|
+
|
56
|
+
def initialize(*args)
|
57
|
+
@initialized_with = args
|
58
|
+
end
|
59
|
+
end
|
60
|
+
doc = klass.new("uri", "external_id", 1)
|
61
|
+
assert_equal ["uri", "external_id", 1], doc.initialized_with
|
62
|
+
end
|
63
|
+
|
64
|
+
def test_subclass_dup
|
65
|
+
klass = Class.new(Nokogiri::HTML::Document)
|
66
|
+
doc = klass.new.dup
|
67
|
+
assert_instance_of klass, doc
|
68
|
+
end
|
69
|
+
|
70
|
+
def test_subclass_parse
|
71
|
+
klass = Class.new(Nokogiri::HTML::Document)
|
72
|
+
doc = klass.parse(File.read(HTML_FILE))
|
73
|
+
assert_equal @html.to_s, doc.to_s
|
74
|
+
assert_instance_of klass, doc
|
75
|
+
end
|
76
|
+
|
77
|
+
def test_document_parse_method
|
78
|
+
html = Nokogiri::HTML::Document.parse(File.read(HTML_FILE))
|
79
|
+
assert_equal @html.to_s, html.to_s
|
80
|
+
end
|
81
|
+
|
82
|
+
###
|
83
|
+
# Nokogiri::HTML returns an empty Document when given a blank string GH#11
|
84
|
+
def test_empty_string_returns_empty_doc
|
85
|
+
doc = Nokogiri::HTML('')
|
86
|
+
assert_instance_of Nokogiri::HTML::Document, doc
|
87
|
+
assert_nil doc.root
|
88
|
+
end
|
89
|
+
|
90
|
+
unless Nokogiri.uses_libxml? && %w[2 6] === LIBXML_VERSION.split('.')[0..1]
|
91
|
+
# FIXME: this is a hack around broken libxml versions
|
92
|
+
def test_to_xhtml_with_indent
|
93
|
+
doc = Nokogiri::HTML('<html><body><a>foo</a></body></html>')
|
94
|
+
doc = Nokogiri::HTML(doc.to_xhtml(:indent => 2))
|
95
|
+
assert_indent 2, doc
|
96
|
+
end
|
97
|
+
|
98
|
+
def test_write_to_xhtml_with_indent
|
99
|
+
io = StringIO.new
|
100
|
+
doc = Nokogiri::HTML('<html><body><a>foo</a></body></html>')
|
101
|
+
doc.write_xhtml_to io, :indent => 5
|
102
|
+
io.rewind
|
103
|
+
doc = Nokogiri::HTML(io.read)
|
104
|
+
assert_indent 5, doc
|
105
|
+
end
|
106
|
+
end
|
107
|
+
|
108
|
+
def test_swap_should_not_exist
|
109
|
+
assert_raises(NoMethodError) {
|
110
|
+
@html.swap
|
111
|
+
}
|
112
|
+
end
|
113
|
+
|
114
|
+
def test_namespace_should_not_exist
|
115
|
+
assert_raises(NoMethodError) {
|
116
|
+
@html.namespace
|
117
|
+
}
|
118
|
+
end
|
119
|
+
|
120
|
+
def test_meta_encoding
|
121
|
+
assert_equal 'UTF-8', @html.meta_encoding
|
122
|
+
|
123
|
+
html = Nokogiri::HTML(<<-eohtml)
|
124
|
+
<html>
|
125
|
+
<head>
|
126
|
+
<meta http-equiv="X-Content-Type" content="text/html; charset=Shift_JIS">
|
127
|
+
</head>
|
128
|
+
<body>
|
129
|
+
foo
|
130
|
+
</body>
|
131
|
+
</html>
|
132
|
+
eohtml
|
133
|
+
assert_nil html.meta_encoding
|
134
|
+
end
|
135
|
+
|
136
|
+
def test_meta_encoding=
|
137
|
+
@html.meta_encoding = 'EUC-JP'
|
138
|
+
assert_equal 'EUC-JP', @html.meta_encoding
|
139
|
+
end
|
140
|
+
|
141
|
+
def test_title
|
142
|
+
assert_equal 'Tender Lovemaking ', @html.title
|
143
|
+
doc = Nokogiri::HTML('<html><body>foo</body></html>')
|
144
|
+
assert_nil doc.title
|
145
|
+
end
|
146
|
+
|
147
|
+
def test_title=()
|
148
|
+
doc = Nokogiri::HTML(<<eohtml)
|
149
|
+
<html>
|
150
|
+
<head>
|
151
|
+
<title>old</title>
|
152
|
+
</head>
|
153
|
+
<body>
|
154
|
+
foo
|
155
|
+
</body>
|
156
|
+
</html>
|
157
|
+
eohtml
|
158
|
+
doc.title = 'new'
|
159
|
+
assert_equal 'new', doc.title
|
160
|
+
|
161
|
+
doc = Nokogiri::HTML(<<eohtml)
|
162
|
+
<html>
|
163
|
+
<head>
|
164
|
+
</head>
|
165
|
+
<body>
|
166
|
+
foo
|
167
|
+
</body>
|
168
|
+
</html>
|
169
|
+
eohtml
|
170
|
+
doc.title = 'new'
|
171
|
+
assert_equal 'new', doc.title
|
172
|
+
|
173
|
+
doc = Nokogiri::HTML(<<eohtml)
|
174
|
+
<html>
|
175
|
+
<body>
|
176
|
+
foo
|
177
|
+
</body>
|
178
|
+
</html>
|
179
|
+
eohtml
|
180
|
+
doc.title = 'new'
|
181
|
+
if Nokogiri.uses_libxml?
|
182
|
+
assert_nil doc.title
|
183
|
+
else
|
184
|
+
assert_equal 'new', doc.title
|
185
|
+
end
|
186
|
+
end
|
187
|
+
|
188
|
+
def test_meta_encoding_without_head
|
189
|
+
html = Nokogiri::HTML('<html><body>foo</body></html>')
|
190
|
+
assert_nil html.meta_encoding
|
191
|
+
|
192
|
+
html.meta_encoding = 'EUC-JP'
|
193
|
+
assert_nil html.meta_encoding
|
194
|
+
end
|
195
|
+
|
196
|
+
def test_root_node_parent_is_document
|
197
|
+
parent = @html.root.parent
|
198
|
+
assert_equal @html, parent
|
199
|
+
assert_instance_of Nokogiri::HTML::Document, parent
|
200
|
+
end
|
201
|
+
|
202
|
+
def test_parse_handles_nil_gracefully
|
203
|
+
assert_nothing_raised do
|
204
|
+
@doc = Nokogiri::HTML::Document.parse(nil)
|
205
|
+
end
|
206
|
+
assert_instance_of Nokogiri::HTML::Document, @doc
|
207
|
+
end
|
208
|
+
|
209
|
+
def test_parse_empty_document
|
210
|
+
doc = Nokogiri::HTML("\n")
|
211
|
+
assert_equal 0, doc.css('a').length
|
212
|
+
assert_equal 0, doc.xpath('//a').length
|
213
|
+
assert_equal 0, doc.search('//a').length
|
214
|
+
end
|
215
|
+
|
216
|
+
def test_HTML_function
|
217
|
+
html = Nokogiri::HTML(File.read(HTML_FILE))
|
218
|
+
assert html.html?
|
219
|
+
end
|
220
|
+
|
221
|
+
def test_parse_io
|
222
|
+
assert File.open(HTML_FILE, 'rb') { |f|
|
223
|
+
Document.read_io(f, nil, 'UTF-8',
|
224
|
+
XML::ParseOptions::NOERROR | XML::ParseOptions::NOWARNING
|
225
|
+
)
|
226
|
+
}
|
227
|
+
end
|
228
|
+
|
229
|
+
def test_to_xhtml
|
230
|
+
assert_match 'XHTML', @html.to_xhtml
|
231
|
+
assert_match 'XHTML', @html.to_xhtml(:encoding => 'UTF-8')
|
232
|
+
assert_match 'UTF-8', @html.to_xhtml(:encoding => 'UTF-8')
|
233
|
+
end
|
234
|
+
|
235
|
+
def test_no_xml_header
|
236
|
+
html = Nokogiri::HTML(<<-eohtml)
|
237
|
+
<html>
|
238
|
+
</html>
|
239
|
+
eohtml
|
240
|
+
assert html.to_html.length > 0, 'html length is too short'
|
241
|
+
assert_no_match(/^<\?xml/, html.to_html)
|
242
|
+
end
|
243
|
+
|
244
|
+
def test_document_has_error
|
245
|
+
html = Nokogiri::HTML(<<-eohtml)
|
246
|
+
<html>
|
247
|
+
<body>
|
248
|
+
<div awesome="asdf>
|
249
|
+
<p>inside div tag</p>
|
250
|
+
</div>
|
251
|
+
<p>outside div tag</p>
|
252
|
+
</body>
|
253
|
+
</html>
|
254
|
+
eohtml
|
255
|
+
assert html.errors.length > 0
|
256
|
+
end
|
257
|
+
|
258
|
+
def test_relative_css
|
259
|
+
html = Nokogiri::HTML(<<-eohtml)
|
260
|
+
<html>
|
261
|
+
<body>
|
262
|
+
<div>
|
263
|
+
<p>inside div tag</p>
|
264
|
+
</div>
|
265
|
+
<p>outside div tag</p>
|
266
|
+
</body>
|
267
|
+
</html>
|
268
|
+
eohtml
|
269
|
+
set = html.search('div').search('p')
|
270
|
+
assert_equal(1, set.length)
|
271
|
+
assert_equal('inside div tag', set.first.inner_text)
|
272
|
+
end
|
273
|
+
|
274
|
+
def test_multi_css
|
275
|
+
html = Nokogiri::HTML(<<-eohtml)
|
276
|
+
<html>
|
277
|
+
<body>
|
278
|
+
<div>
|
279
|
+
<p>p tag</p>
|
280
|
+
<a>a tag</a>
|
281
|
+
</div>
|
282
|
+
</body>
|
283
|
+
</html>
|
284
|
+
eohtml
|
285
|
+
set = html.css('p, a')
|
286
|
+
assert_equal(2, set.length)
|
287
|
+
assert_equal ['a tag', 'p tag'].sort, set.map { |x| x.content }.sort
|
288
|
+
end
|
289
|
+
|
290
|
+
def test_inner_text
|
291
|
+
html = Nokogiri::HTML(<<-eohtml)
|
292
|
+
<html>
|
293
|
+
<body>
|
294
|
+
<div>
|
295
|
+
<p>
|
296
|
+
Hello world!
|
297
|
+
</p>
|
298
|
+
</div>
|
299
|
+
</body>
|
300
|
+
</html>
|
301
|
+
eohtml
|
302
|
+
node = html.xpath('//div').first
|
303
|
+
assert_equal('Hello world!', node.inner_text.strip)
|
304
|
+
end
|
305
|
+
|
306
|
+
def test_find_by_xpath
|
307
|
+
found = @html.xpath('//div/a')
|
308
|
+
assert_equal 3, found.length
|
309
|
+
end
|
310
|
+
|
311
|
+
def test_find_by_css
|
312
|
+
found = @html.css('div > a')
|
313
|
+
assert_equal 3, found.length
|
314
|
+
end
|
315
|
+
|
316
|
+
def test_find_by_css_with_square_brackets
|
317
|
+
found = @html.css("div[@id='header'] > h1")
|
318
|
+
found = @html.css("div[@id='header'] h1") # this blows up on commit 6fa0f6d329d9dbf1cc21c0ac72f7e627bb4c05fc
|
319
|
+
assert_equal 1, found.length
|
320
|
+
end
|
321
|
+
|
322
|
+
def test_find_with_function
|
323
|
+
assert @html.css("div:awesome() h1", Class.new {
|
324
|
+
def awesome divs
|
325
|
+
[divs.first]
|
326
|
+
end
|
327
|
+
}.new)
|
328
|
+
end
|
329
|
+
|
330
|
+
def test_dup_shallow
|
331
|
+
found = @html.search('//div/a').first
|
332
|
+
dup = found.dup(0)
|
333
|
+
assert dup
|
334
|
+
assert_equal '', dup.content
|
335
|
+
end
|
336
|
+
|
337
|
+
def test_search_can_handle_xpath_and_css
|
338
|
+
found = @html.search('//div/a', 'div > p')
|
339
|
+
length = @html.xpath('//div/a').length +
|
340
|
+
@html.css('div > p').length
|
341
|
+
assert_equal length, found.length
|
342
|
+
end
|
343
|
+
|
344
|
+
def test_dup_document
|
345
|
+
assert dup = @html.dup
|
346
|
+
assert_not_equal dup, @html
|
347
|
+
assert @html.html?
|
348
|
+
assert_instance_of Nokogiri::HTML::Document, dup
|
349
|
+
assert dup.html?, 'duplicate should be html'
|
350
|
+
assert_equal @html.to_s, dup.to_s
|
351
|
+
end
|
352
|
+
|
353
|
+
def test_dup_document_shallow
|
354
|
+
assert dup = @html.dup(0)
|
355
|
+
assert_not_equal dup, @html
|
356
|
+
end
|
357
|
+
|
358
|
+
def test_dup
|
359
|
+
found = @html.search('//div/a').first
|
360
|
+
dup = found.dup
|
361
|
+
assert dup
|
362
|
+
assert_equal found.content, dup.content
|
363
|
+
assert_equal found.document, dup.document
|
364
|
+
end
|
365
|
+
|
366
|
+
def test_inner_html
|
367
|
+
html = Nokogiri::HTML(<<-eohtml)
|
368
|
+
<html>
|
369
|
+
<body>
|
370
|
+
<div>
|
371
|
+
<p>
|
372
|
+
Hello world!
|
373
|
+
</p>
|
374
|
+
</div>
|
375
|
+
</body>
|
376
|
+
</html>
|
377
|
+
eohtml
|
378
|
+
node = html.xpath('//div').first
|
379
|
+
assert_equal('<p>Helloworld!</p>', node.inner_html.gsub(/\s/, ''))
|
380
|
+
end
|
381
|
+
|
382
|
+
def test_round_trip
|
383
|
+
doc = Nokogiri::HTML(@html.inner_html)
|
384
|
+
assert_equal @html.root.to_html, doc.root.to_html
|
385
|
+
end
|
386
|
+
|
387
|
+
def test_fragment_contains_text_node
|
388
|
+
fragment = Nokogiri::HTML.fragment('fooo')
|
389
|
+
assert_equal 1, fragment.children.length
|
390
|
+
assert_equal 'fooo', fragment.inner_text
|
391
|
+
end
|
392
|
+
|
393
|
+
def test_fragment_includes_two_tags
|
394
|
+
assert_equal 2, Nokogiri::HTML.fragment("<br/><hr/>").children.length
|
395
|
+
end
|
396
|
+
|
397
|
+
def test_relative_css_finder
|
398
|
+
doc = Nokogiri::HTML(<<-eohtml)
|
399
|
+
<html>
|
400
|
+
<body>
|
401
|
+
<div class="red">
|
402
|
+
<p>
|
403
|
+
inside red
|
404
|
+
</p>
|
405
|
+
</div>
|
406
|
+
<div class="green">
|
407
|
+
<p>
|
408
|
+
inside green
|
409
|
+
</p>
|
410
|
+
</div>
|
411
|
+
</body>
|
412
|
+
</html>
|
413
|
+
eohtml
|
414
|
+
red_divs = doc.css('div.red')
|
415
|
+
assert_equal 1, red_divs.length
|
416
|
+
p_tags = red_divs.first.css('p')
|
417
|
+
assert_equal 1, p_tags.length
|
418
|
+
assert_equal 'inside red', p_tags.first.text.strip
|
419
|
+
end
|
420
|
+
|
421
|
+
def test_find_classes
|
422
|
+
doc = Nokogiri::HTML(<<-eohtml)
|
423
|
+
<html>
|
424
|
+
<body>
|
425
|
+
<p class="red">RED</p>
|
426
|
+
<p class="awesome red">RED</p>
|
427
|
+
<p class="notred">GREEN</p>
|
428
|
+
<p class="green notred">GREEN</p>
|
429
|
+
</body>
|
430
|
+
</html>
|
431
|
+
eohtml
|
432
|
+
list = doc.css('.red')
|
433
|
+
assert_equal 2, list.length
|
434
|
+
assert_equal %w{ RED RED }, list.map { |x| x.text }
|
435
|
+
end
|
436
|
+
|
437
|
+
def test_parse_can_take_io
|
438
|
+
html = nil
|
439
|
+
File.open(HTML_FILE, 'rb') { |f|
|
440
|
+
html = Nokogiri::HTML(f)
|
441
|
+
}
|
442
|
+
assert html.html?
|
443
|
+
end
|
444
|
+
|
445
|
+
def test_html?
|
446
|
+
assert !@html.xml?
|
447
|
+
assert @html.html?
|
448
|
+
end
|
449
|
+
|
450
|
+
def test_serialize
|
451
|
+
assert @html.serialize
|
452
|
+
assert @html.to_html
|
453
|
+
end
|
454
|
+
end
|
455
|
+
end
|
456
|
+
end
|
457
|
+
|