glebm-nokogiri 1.4.2.1
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 +411 -0
- data/CHANGELOG.rdoc +397 -0
- data/Manifest.txt +276 -0
- data/README.ja.rdoc +106 -0
- data/README.rdoc +132 -0
- data/Rakefile +183 -0
- data/bin/nokogiri +49 -0
- data/deps.rip +5 -0
- data/ext/nokogiri/extconf.rb +97 -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 +95 -0
- data/ext/nokogiri/nokogiri.h +153 -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 +464 -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_namespace.c +84 -0
- data/ext/nokogiri/xml_namespace.h +13 -0
- data/ext/nokogiri/xml_node.c +1347 -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 +665 -0
- data/ext/nokogiri/xml_reader.h +10 -0
- data/ext/nokogiri/xml_relax_ng.c +168 -0
- data/ext/nokogiri/xml_relax_ng.h +9 -0
- data/ext/nokogiri/xml_sax_parser.c +286 -0
- data/ext/nokogiri/xml_sax_parser.h +39 -0
- data/ext/nokogiri/xml_sax_parser_context.c +159 -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 +276 -0
- data/ext/nokogiri/xml_xpath_context.h +9 -0
- data/ext/nokogiri/xslt_stylesheet.c +142 -0
- data/ext/nokogiri/xslt_stylesheet.h +9 -0
- data/lib/nokogiri.rb +133 -0
- data/lib/nokogiri/css.rb +25 -0
- data/lib/nokogiri/css/generated_parser.rb +669 -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 +232 -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 +169 -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 +386 -0
- data/lib/nokogiri/ffi/structs/common_node.rb +38 -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 +117 -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 +38 -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/weak_bucket.rb +40 -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 +162 -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 +36 -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 +556 -0
- data/lib/nokogiri/ffi/xml/node_set.rb +149 -0
- data/lib/nokogiri/ffi/xml/processing_instruction.rb +20 -0
- data/lib/nokogiri/ffi/xml/reader.rb +232 -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 +51 -0
- data/lib/nokogiri/ffi/xml/schema.rb +109 -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 +9 -0
- data/lib/nokogiri/ffi/xml/xpath_context.rb +148 -0
- data/lib/nokogiri/ffi/xslt/stylesheet.rb +53 -0
- data/lib/nokogiri/html.rb +35 -0
- data/lib/nokogiri/html/builder.rb +35 -0
- data/lib/nokogiri/html/document.rb +90 -0
- data/lib/nokogiri/html/document_fragment.rb +36 -0
- data/lib/nokogiri/html/element_description.rb +23 -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/syntax_error.rb +4 -0
- data/lib/nokogiri/version.rb +37 -0
- data/lib/nokogiri/version_warning.rb +14 -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 +194 -0
- data/lib/nokogiri/xml/document_fragment.rb +77 -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 +19 -0
- data/lib/nokogiri/xml/namespace.rb +13 -0
- data/lib/nokogiri/xml/node.rb +793 -0
- data/lib/nokogiri/xml/node/save_options.rb +42 -0
- data/lib/nokogiri/xml/node_set.rb +325 -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 +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 +48 -0
- data/lib/nokogiri/xslt/stylesheet.rb +25 -0
- data/lib/xsd/xmlparser/nokogiri.rb +90 -0
- data/tasks/cross_compile.rb +158 -0
- data/tasks/test.rb +94 -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 +85 -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 +169 -0
- data/test/html/sax/test_parser.rb +74 -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 +398 -0
- data/test/html/test_document_encoding.rb +77 -0
- data/test/html/test_document_fragment.rb +182 -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 +181 -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 +87 -0
- data/test/test_nokogiri.rb +138 -0
- data/test/test_reader.rb +386 -0
- data/test/test_soap4r_sax.rb +52 -0
- data/test/test_xslt_transforms.rb +188 -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 +307 -0
- data/test/xml/sax/test_parser_context.rb +63 -0
- data/test/xml/sax/test_push_parser.rb +139 -0
- data/test/xml/test_attr.rb +38 -0
- data/test/xml/test_attribute_decl.rb +82 -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 +668 -0
- data/test/xml/test_document_encoding.rb +26 -0
- data/test/xml/test_document_fragment.rb +180 -0
- data/test/xml/test_dtd.rb +82 -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 +120 -0
- data/test/xml/test_entity_reference.rb +21 -0
- data/test/xml/test_namespace.rb +68 -0
- data/test/xml/test_node.rb +865 -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 +293 -0
- data/test/xml/test_node_set.rb +649 -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 +38 -0
- data/test/xml/test_unparented_node.rb +381 -0
- data/test/xml/test_xpath.rb +138 -0
- metadata +533 -0
|
@@ -0,0 +1,182 @@
|
|
|
1
|
+
# -*- coding: utf-8 -*-
|
|
2
|
+
require "helper"
|
|
3
|
+
|
|
4
|
+
module Nokogiri
|
|
5
|
+
module HTML
|
|
6
|
+
class TestDocumentFragment < Nokogiri::TestCase
|
|
7
|
+
def setup
|
|
8
|
+
super
|
|
9
|
+
@html = Nokogiri::HTML.parse(File.read(HTML_FILE), HTML_FILE)
|
|
10
|
+
end
|
|
11
|
+
|
|
12
|
+
def test_no_contextual_parsing_on_unlinked_nodes
|
|
13
|
+
node = @html.css('body').first
|
|
14
|
+
node.unlink
|
|
15
|
+
assert_raises(RuntimeError) do
|
|
16
|
+
node.parse('<br />')
|
|
17
|
+
end
|
|
18
|
+
end
|
|
19
|
+
|
|
20
|
+
def test_parse_in_context
|
|
21
|
+
assert_equal('<br>', @html.root.parse('<br />').to_s)
|
|
22
|
+
end
|
|
23
|
+
|
|
24
|
+
def test_ancestors_search
|
|
25
|
+
html = %q{
|
|
26
|
+
<div>
|
|
27
|
+
<ul>
|
|
28
|
+
<li>foo</li>
|
|
29
|
+
</ul>
|
|
30
|
+
</div>
|
|
31
|
+
}
|
|
32
|
+
fragment = Nokogiri::HTML.fragment html
|
|
33
|
+
li = fragment.at('li')
|
|
34
|
+
assert li.matches?('li')
|
|
35
|
+
end
|
|
36
|
+
|
|
37
|
+
def test_fun_encoding
|
|
38
|
+
string = %Q(<body>こんにちは</body>)
|
|
39
|
+
html = Nokogiri::HTML::DocumentFragment.parse(
|
|
40
|
+
string
|
|
41
|
+
).to_html(:encoding => 'UTF-8')
|
|
42
|
+
assert_equal string, html
|
|
43
|
+
end
|
|
44
|
+
|
|
45
|
+
def test_new
|
|
46
|
+
fragment = Nokogiri::HTML::DocumentFragment.new(@html)
|
|
47
|
+
end
|
|
48
|
+
|
|
49
|
+
def test_fragment_should_have_document
|
|
50
|
+
fragment = Nokogiri::HTML::DocumentFragment.new(@html)
|
|
51
|
+
assert_equal @html, fragment.document
|
|
52
|
+
end
|
|
53
|
+
|
|
54
|
+
def test_empty_fragment_should_be_searchable_by_css
|
|
55
|
+
fragment = Nokogiri::HTML.fragment("")
|
|
56
|
+
assert_equal 0, fragment.css("a").size
|
|
57
|
+
end
|
|
58
|
+
|
|
59
|
+
def test_empty_fragment_should_be_searchable
|
|
60
|
+
fragment = Nokogiri::HTML.fragment("")
|
|
61
|
+
assert_equal 0, fragment.search("//a").size
|
|
62
|
+
end
|
|
63
|
+
|
|
64
|
+
def test_name
|
|
65
|
+
fragment = Nokogiri::HTML::DocumentFragment.new(@html)
|
|
66
|
+
assert_equal '#document-fragment', fragment.name
|
|
67
|
+
end
|
|
68
|
+
|
|
69
|
+
def test_static_method
|
|
70
|
+
fragment = Nokogiri::HTML::DocumentFragment.parse("<div>a</div>")
|
|
71
|
+
assert_instance_of Nokogiri::HTML::DocumentFragment, fragment
|
|
72
|
+
end
|
|
73
|
+
|
|
74
|
+
def test_many_fragments
|
|
75
|
+
100.times { Nokogiri::HTML::DocumentFragment.new(@html) }
|
|
76
|
+
end
|
|
77
|
+
|
|
78
|
+
def test_subclass
|
|
79
|
+
klass = Class.new(Nokogiri::HTML::DocumentFragment)
|
|
80
|
+
fragment = klass.new(@html, "<div>a</div>")
|
|
81
|
+
assert_instance_of klass, fragment
|
|
82
|
+
end
|
|
83
|
+
|
|
84
|
+
def test_subclass_parse
|
|
85
|
+
klass = Class.new(Nokogiri::HTML::DocumentFragment)
|
|
86
|
+
doc = klass.parse("<div>a</div>")
|
|
87
|
+
assert_instance_of klass, doc
|
|
88
|
+
end
|
|
89
|
+
|
|
90
|
+
def test_html_fragment
|
|
91
|
+
fragment = Nokogiri::HTML.fragment("<div>a</div>")
|
|
92
|
+
assert_equal "<div>a</div>", fragment.to_s
|
|
93
|
+
end
|
|
94
|
+
|
|
95
|
+
def test_html_fragment_has_outer_text
|
|
96
|
+
doc = "a<div>b</div>c"
|
|
97
|
+
fragment = Nokogiri::HTML::Document.new.fragment(doc)
|
|
98
|
+
if Nokogiri::VERSION_INFO['libxml']['loaded'] <= "2.6.16"
|
|
99
|
+
assert_equal "a<div>b</div><p>c</p>", fragment.to_s
|
|
100
|
+
else
|
|
101
|
+
assert_equal "a<div>b</div>c", fragment.to_s
|
|
102
|
+
end
|
|
103
|
+
end
|
|
104
|
+
|
|
105
|
+
def test_html_fragment_case_insensitivity
|
|
106
|
+
doc = "<Div>b</Div>"
|
|
107
|
+
fragment = Nokogiri::HTML::Document.new.fragment(doc)
|
|
108
|
+
assert_equal "<div>b</div>", fragment.to_s
|
|
109
|
+
end
|
|
110
|
+
|
|
111
|
+
def test_html_fragment_with_leading_whitespace
|
|
112
|
+
doc = " <div>b</div> "
|
|
113
|
+
fragment = Nokogiri::HTML::Document.new.fragment(doc)
|
|
114
|
+
assert_equal "<div>b</div>", fragment.to_s
|
|
115
|
+
end
|
|
116
|
+
|
|
117
|
+
def test_html_fragment_with_leading_whitespace_and_newline
|
|
118
|
+
doc = " \n<div>b</div> "
|
|
119
|
+
fragment = Nokogiri::HTML::Document.new.fragment(doc)
|
|
120
|
+
assert_equal "<div>b</div>", fragment.to_s
|
|
121
|
+
end
|
|
122
|
+
|
|
123
|
+
def test_html_fragment_with_leading_text_and_newline
|
|
124
|
+
fragment = HTML::Document.new.fragment("First line\nSecond line<br>Broken line")
|
|
125
|
+
assert_equal fragment.to_s, "First line\nSecond line<br>Broken line"
|
|
126
|
+
end
|
|
127
|
+
|
|
128
|
+
def test_html_fragment_with_leading_whitespace_and_text_and_newline
|
|
129
|
+
fragment = HTML::Document.new.fragment(" First line\nSecond line<br>Broken line")
|
|
130
|
+
assert_equal "First line\nSecond line<br>Broken line", fragment.to_s
|
|
131
|
+
end
|
|
132
|
+
|
|
133
|
+
def test_html_fragment_with_leading_entity
|
|
134
|
+
failed = ""test<br/>test""
|
|
135
|
+
fragment = Nokogiri::HTML::DocumentFragment.parse(failed)
|
|
136
|
+
assert_equal '"test<br>test"', fragment.to_html
|
|
137
|
+
end
|
|
138
|
+
|
|
139
|
+
def test_to_s
|
|
140
|
+
doc = "<span>foo<br></span><span>bar</span>"
|
|
141
|
+
fragment = Nokogiri::HTML::Document.new.fragment(doc)
|
|
142
|
+
assert_equal "<span>foo<br></span><span>bar</span>", fragment.to_s
|
|
143
|
+
end
|
|
144
|
+
|
|
145
|
+
def test_to_html
|
|
146
|
+
doc = "<span>foo<br></span><span>bar</span>"
|
|
147
|
+
fragment = Nokogiri::HTML::Document.new.fragment(doc)
|
|
148
|
+
assert_equal "<span>foo<br></span><span>bar</span>", fragment.to_html
|
|
149
|
+
end
|
|
150
|
+
|
|
151
|
+
def test_to_xhtml
|
|
152
|
+
doc = "<span>foo<br></span><span>bar</span>"
|
|
153
|
+
fragment = Nokogiri::HTML::Document.new.fragment(doc)
|
|
154
|
+
if Nokogiri::VERSION_INFO['libxml']['loaded'] >= "2.7.0"
|
|
155
|
+
assert_equal "<span>foo<br /></span><span>bar</span>", fragment.to_xhtml
|
|
156
|
+
else
|
|
157
|
+
assert_equal "<span>foo<br></span><span>bar</span>", fragment.to_xhtml
|
|
158
|
+
end
|
|
159
|
+
end
|
|
160
|
+
|
|
161
|
+
def test_to_xml
|
|
162
|
+
doc = "<span>foo<br></span><span>bar</span>"
|
|
163
|
+
fragment = Nokogiri::HTML::Document.new.fragment(doc)
|
|
164
|
+
assert_equal "<span>foo<br/></span><span>bar</span>", fragment.to_xml
|
|
165
|
+
end
|
|
166
|
+
|
|
167
|
+
def test_fragment_script_tag_with_cdata
|
|
168
|
+
doc = HTML::Document.new
|
|
169
|
+
fragment = doc.fragment("<script>var foo = 'bar';</script>")
|
|
170
|
+
assert_equal("<script>var foo = 'bar';</script>",
|
|
171
|
+
fragment.to_s)
|
|
172
|
+
end
|
|
173
|
+
|
|
174
|
+
def test_fragment_with_comment
|
|
175
|
+
doc = HTML::Document.new
|
|
176
|
+
fragment = doc.fragment("<p>hello<!-- your ad here --></p>")
|
|
177
|
+
assert_equal("<p>hello<!-- your ad here --></p>",
|
|
178
|
+
fragment.to_s)
|
|
179
|
+
end
|
|
180
|
+
end
|
|
181
|
+
end
|
|
182
|
+
end
|
|
@@ -0,0 +1,98 @@
|
|
|
1
|
+
require "helper"
|
|
2
|
+
|
|
3
|
+
module Nokogiri
|
|
4
|
+
module HTML
|
|
5
|
+
class TestElementDescription < Nokogiri::TestCase
|
|
6
|
+
def test_fetch_nonexistent
|
|
7
|
+
assert_nil ElementDescription['foo']
|
|
8
|
+
end
|
|
9
|
+
|
|
10
|
+
def test_fetch_element_description
|
|
11
|
+
assert desc = ElementDescription['a']
|
|
12
|
+
assert_instance_of ElementDescription, desc
|
|
13
|
+
end
|
|
14
|
+
|
|
15
|
+
def test_name
|
|
16
|
+
assert_equal 'a', ElementDescription['a'].name
|
|
17
|
+
end
|
|
18
|
+
|
|
19
|
+
def test_implied_start_tag?
|
|
20
|
+
assert !ElementDescription['a'].implied_start_tag?
|
|
21
|
+
end
|
|
22
|
+
|
|
23
|
+
def test_implied_end_tag?
|
|
24
|
+
assert !ElementDescription['a'].implied_end_tag?
|
|
25
|
+
assert ElementDescription['p'].implied_end_tag?
|
|
26
|
+
end
|
|
27
|
+
|
|
28
|
+
def test_save_end_tag?
|
|
29
|
+
assert !ElementDescription['a'].save_end_tag?
|
|
30
|
+
assert ElementDescription['br'].save_end_tag?
|
|
31
|
+
end
|
|
32
|
+
|
|
33
|
+
def test_empty?
|
|
34
|
+
assert ElementDescription['br'].empty?
|
|
35
|
+
assert !ElementDescription['a'].empty?
|
|
36
|
+
end
|
|
37
|
+
|
|
38
|
+
def test_deprecated?
|
|
39
|
+
assert ElementDescription['applet'].deprecated?
|
|
40
|
+
assert !ElementDescription['br'].deprecated?
|
|
41
|
+
end
|
|
42
|
+
|
|
43
|
+
def test_inline?
|
|
44
|
+
assert ElementDescription['a'].inline?
|
|
45
|
+
assert !ElementDescription['div'].inline?
|
|
46
|
+
end
|
|
47
|
+
|
|
48
|
+
def test_block?
|
|
49
|
+
element = ElementDescription['a']
|
|
50
|
+
assert_equal(!element.inline?, element.block?)
|
|
51
|
+
end
|
|
52
|
+
|
|
53
|
+
def test_description
|
|
54
|
+
assert ElementDescription['a'].description
|
|
55
|
+
end
|
|
56
|
+
|
|
57
|
+
def test_subelements
|
|
58
|
+
sub_elements = ElementDescription['body'].sub_elements
|
|
59
|
+
if Nokogiri::LIBXML_VERSION == '2.7.7'
|
|
60
|
+
assert_equal 65, sub_elements.length
|
|
61
|
+
else
|
|
62
|
+
assert_equal 61, sub_elements.length
|
|
63
|
+
end
|
|
64
|
+
end
|
|
65
|
+
|
|
66
|
+
def test_default_sub_element
|
|
67
|
+
assert_equal 'div', ElementDescription['body'].default_sub_element
|
|
68
|
+
end
|
|
69
|
+
|
|
70
|
+
def test_optional_attributes
|
|
71
|
+
attrs = ElementDescription['table'].optional_attributes
|
|
72
|
+
assert attrs
|
|
73
|
+
end
|
|
74
|
+
|
|
75
|
+
def test_deprecated_attributes
|
|
76
|
+
attrs = ElementDescription['table'].deprecated_attributes
|
|
77
|
+
assert attrs
|
|
78
|
+
assert_equal 2, attrs.length
|
|
79
|
+
end
|
|
80
|
+
|
|
81
|
+
def test_required_attributes
|
|
82
|
+
attrs = ElementDescription['table'].required_attributes
|
|
83
|
+
assert attrs
|
|
84
|
+
assert_equal 0, attrs.length
|
|
85
|
+
end
|
|
86
|
+
|
|
87
|
+
def test_inspect
|
|
88
|
+
desc = ElementDescription['input']
|
|
89
|
+
assert_match desc.name, desc.inspect
|
|
90
|
+
end
|
|
91
|
+
|
|
92
|
+
def test_to_s
|
|
93
|
+
desc = ElementDescription['input']
|
|
94
|
+
assert_match desc.name, desc.to_s
|
|
95
|
+
end
|
|
96
|
+
end
|
|
97
|
+
end
|
|
98
|
+
end
|
|
@@ -0,0 +1,14 @@
|
|
|
1
|
+
require "helper"
|
|
2
|
+
|
|
3
|
+
module Nokogiri
|
|
4
|
+
module HTML
|
|
5
|
+
class TestNamedCharacters < Nokogiri::TestCase
|
|
6
|
+
def test_named_character
|
|
7
|
+
copy = NamedCharacters.get('copy')
|
|
8
|
+
assert_equal 169, NamedCharacters['copy']
|
|
9
|
+
assert_equal copy.value, NamedCharacters['copy']
|
|
10
|
+
assert copy.description
|
|
11
|
+
end
|
|
12
|
+
end
|
|
13
|
+
end
|
|
14
|
+
end
|
|
@@ -0,0 +1,181 @@
|
|
|
1
|
+
require "helper"
|
|
2
|
+
|
|
3
|
+
require 'nkf'
|
|
4
|
+
|
|
5
|
+
module Nokogiri
|
|
6
|
+
module HTML
|
|
7
|
+
class TestNode < Nokogiri::TestCase
|
|
8
|
+
def setup
|
|
9
|
+
super
|
|
10
|
+
@html = Nokogiri::HTML(<<-eohtml)
|
|
11
|
+
<html>
|
|
12
|
+
<head></head>
|
|
13
|
+
<body>
|
|
14
|
+
<div class='baz'><a href="foo" class="bar">first</a></div>
|
|
15
|
+
</body>
|
|
16
|
+
</html>
|
|
17
|
+
eohtml
|
|
18
|
+
end
|
|
19
|
+
|
|
20
|
+
def test_to_a
|
|
21
|
+
assert_equal [['class', 'bar'], ['href', 'foo']],@html.at('a').to_a.sort
|
|
22
|
+
end
|
|
23
|
+
|
|
24
|
+
def test_attr
|
|
25
|
+
node = @html.at('div.baz')
|
|
26
|
+
assert_equal node['class'], node.attr('class')
|
|
27
|
+
end
|
|
28
|
+
|
|
29
|
+
def test_get_attribute
|
|
30
|
+
element = @html.at('div')
|
|
31
|
+
assert_equal 'baz', element.get_attribute('class')
|
|
32
|
+
assert_equal 'baz', element['class']
|
|
33
|
+
element['href'] = "javascript:alert(\"AGGA-KA-BOO!\")"
|
|
34
|
+
assert_match(/%22AGGA-KA-BOO!%22/, element.to_html)
|
|
35
|
+
end
|
|
36
|
+
|
|
37
|
+
def test_css_path_round_trip
|
|
38
|
+
doc = Nokogiri::HTML(File.read(HTML_FILE))
|
|
39
|
+
%w{ #header small div[2] div.post body }.each do |css_sel|
|
|
40
|
+
ele = doc.at css_sel
|
|
41
|
+
assert_equal ele, doc.at(ele.css_path), ele.css_path
|
|
42
|
+
end
|
|
43
|
+
end
|
|
44
|
+
|
|
45
|
+
def test_path_round_trip
|
|
46
|
+
doc = Nokogiri::HTML(File.read(HTML_FILE))
|
|
47
|
+
%w{ #header small div[2] div.post body }.each do |css_sel|
|
|
48
|
+
ele = doc.at css_sel
|
|
49
|
+
assert_equal ele, doc.at(ele.path), ele.path
|
|
50
|
+
end
|
|
51
|
+
end
|
|
52
|
+
|
|
53
|
+
def test_append_with_document
|
|
54
|
+
assert_raises(ArgumentError) do
|
|
55
|
+
@html.root << Nokogiri::HTML::Document.new
|
|
56
|
+
end
|
|
57
|
+
end
|
|
58
|
+
|
|
59
|
+
###
|
|
60
|
+
# Make sure a document that doesn't declare a meta encoding returns
|
|
61
|
+
# nil.
|
|
62
|
+
def test_meta_encoding
|
|
63
|
+
assert_nil @html.meta_encoding
|
|
64
|
+
end
|
|
65
|
+
|
|
66
|
+
def test_description
|
|
67
|
+
assert desc = @html.at('a.bar').description
|
|
68
|
+
assert_equal 'a', desc.name
|
|
69
|
+
end
|
|
70
|
+
|
|
71
|
+
def test_ancestors_with_selector
|
|
72
|
+
assert node = @html.at('a.bar').child
|
|
73
|
+
assert list = node.ancestors('.baz')
|
|
74
|
+
assert_equal 1, list.length
|
|
75
|
+
assert_equal 'div', list.first.name
|
|
76
|
+
end
|
|
77
|
+
|
|
78
|
+
def test_matches_inside_fragment
|
|
79
|
+
fragment = DocumentFragment.new @html
|
|
80
|
+
fragment << XML::Node.new('a', @html)
|
|
81
|
+
|
|
82
|
+
a = fragment.children.last
|
|
83
|
+
assert a.matches?('a'), 'a should match'
|
|
84
|
+
end
|
|
85
|
+
|
|
86
|
+
def test_css_matches?
|
|
87
|
+
assert node = @html.at('a.bar')
|
|
88
|
+
assert node.matches?('a.bar')
|
|
89
|
+
end
|
|
90
|
+
|
|
91
|
+
def test_xpath_matches?
|
|
92
|
+
assert node = @html.at('//a')
|
|
93
|
+
assert node.matches?('//a')
|
|
94
|
+
end
|
|
95
|
+
|
|
96
|
+
def test_unlink_then_swap
|
|
97
|
+
node = @html.at('a')
|
|
98
|
+
node.unlink
|
|
99
|
+
|
|
100
|
+
another_node = @html.at('div')
|
|
101
|
+
assert another_node, 'should have a node'
|
|
102
|
+
|
|
103
|
+
# This used to segv
|
|
104
|
+
assert_nothing_raised do
|
|
105
|
+
node.add_previous_sibling another_node
|
|
106
|
+
end
|
|
107
|
+
end
|
|
108
|
+
|
|
109
|
+
def test_swap
|
|
110
|
+
@html.at('div').swap('<a href="foo">bar</a>')
|
|
111
|
+
a_tag = @html.css('a').first
|
|
112
|
+
assert_equal 'body', a_tag.parent.name
|
|
113
|
+
assert_equal 0, @html.css('div').length
|
|
114
|
+
end
|
|
115
|
+
|
|
116
|
+
def test_swap_with_regex_characters
|
|
117
|
+
@html.at('div').swap('<a href="foo">ba)r</a>')
|
|
118
|
+
a_tag = @html.css('a').first
|
|
119
|
+
assert_equal 'ba)r', a_tag.text
|
|
120
|
+
end
|
|
121
|
+
|
|
122
|
+
def test_attribute_decodes_entities
|
|
123
|
+
node = @html.at('div')
|
|
124
|
+
node['href'] = 'foo&bar'
|
|
125
|
+
assert_equal 'foo&bar', node['href']
|
|
126
|
+
node['href'] += '&baz'
|
|
127
|
+
assert_equal 'foo&bar&baz', node['href']
|
|
128
|
+
end
|
|
129
|
+
|
|
130
|
+
def test_fragment_handler_does_not_regurge_on_invalid_attributes
|
|
131
|
+
iframe = %Q{<iframe style="width: 0%; height: 0px" src="http://someurl" allowtransparency></iframe>}
|
|
132
|
+
assert_nothing_raised { @html.at('div').fragment(iframe) }
|
|
133
|
+
end
|
|
134
|
+
|
|
135
|
+
def test_fragment
|
|
136
|
+
fragment = @html.fragment(<<-eohtml)
|
|
137
|
+
hello
|
|
138
|
+
<div class="foo">
|
|
139
|
+
<p>bar</p>
|
|
140
|
+
</div>
|
|
141
|
+
world
|
|
142
|
+
eohtml
|
|
143
|
+
assert_match(/^hello/, fragment.inner_html.strip)
|
|
144
|
+
assert_equal 3, fragment.children.length
|
|
145
|
+
assert p_tag = fragment.css('p').first
|
|
146
|
+
assert_equal 'div', p_tag.parent.name
|
|
147
|
+
assert_equal 'foo', p_tag.parent['class']
|
|
148
|
+
end
|
|
149
|
+
|
|
150
|
+
def test_fragment_serialization
|
|
151
|
+
fragment = Nokogiri::HTML.fragment("<div>foo</div>")
|
|
152
|
+
assert_equal "<div>foo</div>", fragment.serialize.chomp
|
|
153
|
+
assert_equal "<div>foo</div>", fragment.to_xml.chomp
|
|
154
|
+
assert_equal "<div>foo</div>", fragment.inner_html
|
|
155
|
+
assert_equal "<div>foo</div>", fragment.to_html
|
|
156
|
+
assert_equal "<div>foo</div>", fragment.to_s
|
|
157
|
+
end
|
|
158
|
+
|
|
159
|
+
def test_to_html_does_not_contain_entities
|
|
160
|
+
return unless defined?(NKF) # NKF is not implemented on Rubinius as of 2009-11-23
|
|
161
|
+
html = NKF.nkf("-e --msdos", <<-EOH)
|
|
162
|
+
<html><body>
|
|
163
|
+
<p> test paragraph
|
|
164
|
+
foo bar </p>
|
|
165
|
+
</body></html>
|
|
166
|
+
EOH
|
|
167
|
+
nokogiri = Nokogiri::HTML.parse(html)
|
|
168
|
+
|
|
169
|
+
if RUBY_PLATFORM =~ /java/
|
|
170
|
+
# NKF linebreak modes are not supported as of jruby 1.2
|
|
171
|
+
# see http://jira.codehaus.org/browse/JRUBY-3602 for status
|
|
172
|
+
assert_equal "<p>testparagraph\nfoobar</p>",
|
|
173
|
+
nokogiri.at("p").to_html.gsub(/ /, '')
|
|
174
|
+
else
|
|
175
|
+
assert_equal "<p>testparagraph\r\nfoobar</p>",
|
|
176
|
+
nokogiri.at("p").to_html.gsub(/ /, '')
|
|
177
|
+
end
|
|
178
|
+
end
|
|
179
|
+
end
|
|
180
|
+
end
|
|
181
|
+
end
|