Nokogiri_precompiled_aarch64_dedshit 1.14.5
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +7 -0
- data/Gemfile +44 -0
- data/LICENSE-DEPENDENCIES.md +2224 -0
- data/LICENSE.md +9 -0
- data/README.md +287 -0
- data/bin/nokogiri +131 -0
- data/dependencies.yml +41 -0
- data/ext/java/nokogiri/Html4Document.java +157 -0
- data/ext/java/nokogiri/Html4ElementDescription.java +133 -0
- data/ext/java/nokogiri/Html4EntityLookup.java +63 -0
- data/ext/java/nokogiri/Html4SaxParserContext.java +289 -0
- data/ext/java/nokogiri/Html4SaxPushParser.java +213 -0
- data/ext/java/nokogiri/NokogiriService.java +613 -0
- data/ext/java/nokogiri/XmlAttr.java +154 -0
- data/ext/java/nokogiri/XmlAttributeDecl.java +119 -0
- data/ext/java/nokogiri/XmlCdata.java +60 -0
- data/ext/java/nokogiri/XmlComment.java +77 -0
- data/ext/java/nokogiri/XmlDocument.java +705 -0
- data/ext/java/nokogiri/XmlDocumentFragment.java +163 -0
- data/ext/java/nokogiri/XmlDtd.java +516 -0
- data/ext/java/nokogiri/XmlElement.java +44 -0
- data/ext/java/nokogiri/XmlElementContent.java +412 -0
- data/ext/java/nokogiri/XmlElementDecl.java +148 -0
- data/ext/java/nokogiri/XmlEntityDecl.java +151 -0
- data/ext/java/nokogiri/XmlEntityReference.java +79 -0
- data/ext/java/nokogiri/XmlNamespace.java +193 -0
- data/ext/java/nokogiri/XmlNode.java +1938 -0
- data/ext/java/nokogiri/XmlNodeSet.java +463 -0
- data/ext/java/nokogiri/XmlProcessingInstruction.java +79 -0
- data/ext/java/nokogiri/XmlReader.java +615 -0
- data/ext/java/nokogiri/XmlRelaxng.java +133 -0
- data/ext/java/nokogiri/XmlSaxParserContext.java +329 -0
- data/ext/java/nokogiri/XmlSaxPushParser.java +288 -0
- data/ext/java/nokogiri/XmlSchema.java +423 -0
- data/ext/java/nokogiri/XmlSyntaxError.java +137 -0
- data/ext/java/nokogiri/XmlText.java +90 -0
- data/ext/java/nokogiri/XmlXpathContext.java +305 -0
- data/ext/java/nokogiri/XsltStylesheet.java +368 -0
- data/ext/java/nokogiri/internals/ClosedStreamException.java +13 -0
- data/ext/java/nokogiri/internals/HtmlDomParserContext.java +252 -0
- data/ext/java/nokogiri/internals/IgnoreSchemaErrorsErrorHandler.java +27 -0
- data/ext/java/nokogiri/internals/NokogiriBlockingQueueInputStream.java +178 -0
- data/ext/java/nokogiri/internals/NokogiriDomParser.java +99 -0
- data/ext/java/nokogiri/internals/NokogiriEntityResolver.java +140 -0
- data/ext/java/nokogiri/internals/NokogiriErrorHandler.java +65 -0
- data/ext/java/nokogiri/internals/NokogiriHandler.java +339 -0
- data/ext/java/nokogiri/internals/NokogiriHelpers.java +817 -0
- data/ext/java/nokogiri/internals/NokogiriNamespaceCache.java +228 -0
- data/ext/java/nokogiri/internals/NokogiriNamespaceContext.java +110 -0
- data/ext/java/nokogiri/internals/NokogiriNonStrictErrorHandler.java +86 -0
- data/ext/java/nokogiri/internals/NokogiriNonStrictErrorHandler4NekoHtml.java +107 -0
- data/ext/java/nokogiri/internals/NokogiriStrictErrorHandler.java +62 -0
- data/ext/java/nokogiri/internals/NokogiriXPathFunction.java +165 -0
- data/ext/java/nokogiri/internals/NokogiriXPathFunctionResolver.java +50 -0
- data/ext/java/nokogiri/internals/NokogiriXPathVariableResolver.java +37 -0
- data/ext/java/nokogiri/internals/NokogiriXsltErrorListener.java +70 -0
- data/ext/java/nokogiri/internals/ParserContext.java +262 -0
- data/ext/java/nokogiri/internals/ReaderNode.java +564 -0
- data/ext/java/nokogiri/internals/SaveContextVisitor.java +865 -0
- data/ext/java/nokogiri/internals/SchemaErrorHandler.java +50 -0
- data/ext/java/nokogiri/internals/XalanDTMManagerPatch.java +174 -0
- data/ext/java/nokogiri/internals/XmlDeclHandler.java +11 -0
- data/ext/java/nokogiri/internals/XmlDomParserContext.java +265 -0
- data/ext/java/nokogiri/internals/XmlSaxParser.java +40 -0
- data/ext/java/nokogiri/internals/c14n/AttrCompare.java +122 -0
- data/ext/java/nokogiri/internals/c14n/C14nHelper.java +178 -0
- data/ext/java/nokogiri/internals/c14n/CanonicalFilter.java +43 -0
- data/ext/java/nokogiri/internals/c14n/CanonicalizationException.java +106 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer.java +278 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer11.java +664 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer11_OmitComments.java +45 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer11_WithComments.java +45 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer20010315.java +388 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer20010315Excl.java +308 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer20010315ExclOmitComments.java +47 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer20010315ExclWithComments.java +51 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer20010315OmitComments.java +51 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer20010315WithComments.java +50 -0
- data/ext/java/nokogiri/internals/c14n/CanonicalizerBase.java +660 -0
- data/ext/java/nokogiri/internals/c14n/CanonicalizerPhysical.java +194 -0
- data/ext/java/nokogiri/internals/c14n/CanonicalizerSpi.java +77 -0
- data/ext/java/nokogiri/internals/c14n/Constants.java +45 -0
- data/ext/java/nokogiri/internals/c14n/ElementProxy.java +325 -0
- data/ext/java/nokogiri/internals/c14n/HelperNodeList.java +106 -0
- data/ext/java/nokogiri/internals/c14n/IgnoreAllErrorHandler.java +86 -0
- data/ext/java/nokogiri/internals/c14n/InclusiveNamespaces.java +181 -0
- data/ext/java/nokogiri/internals/c14n/InvalidCanonicalizerException.java +87 -0
- data/ext/java/nokogiri/internals/c14n/NameSpaceSymbTable.java +452 -0
- data/ext/java/nokogiri/internals/c14n/NodeFilter.java +52 -0
- data/ext/java/nokogiri/internals/c14n/UtfHelpper.java +190 -0
- data/ext/java/nokogiri/internals/c14n/XMLUtils.java +540 -0
- data/ext/java/nokogiri/internals/dom2dtm/DOM2DTM.java +1712 -0
- data/ext/java/nokogiri/internals/dom2dtm/DOM2DTMdefaultNamespaceDeclarationNode.java +737 -0
- data/ext/nokogiri/depend +38 -0
- data/ext/nokogiri/extconf.rb +1086 -0
- data/ext/nokogiri/gumbo.c +594 -0
- data/ext/nokogiri/html4_document.c +167 -0
- data/ext/nokogiri/html4_element_description.c +294 -0
- data/ext/nokogiri/html4_entity_lookup.c +37 -0
- data/ext/nokogiri/html4_sax_parser_context.c +116 -0
- data/ext/nokogiri/html4_sax_push_parser.c +95 -0
- data/ext/nokogiri/libxml2_backwards_compat.c +121 -0
- data/ext/nokogiri/nokogiri.c +265 -0
- data/ext/nokogiri/nokogiri.h +235 -0
- data/ext/nokogiri/test_global_handlers.c +42 -0
- data/ext/nokogiri/xml_attr.c +103 -0
- data/ext/nokogiri/xml_attribute_decl.c +70 -0
- data/ext/nokogiri/xml_cdata.c +57 -0
- data/ext/nokogiri/xml_comment.c +62 -0
- data/ext/nokogiri/xml_document.c +689 -0
- data/ext/nokogiri/xml_document_fragment.c +44 -0
- data/ext/nokogiri/xml_dtd.c +210 -0
- data/ext/nokogiri/xml_element_content.c +128 -0
- data/ext/nokogiri/xml_element_decl.c +69 -0
- data/ext/nokogiri/xml_encoding_handler.c +104 -0
- data/ext/nokogiri/xml_entity_decl.c +112 -0
- data/ext/nokogiri/xml_entity_reference.c +50 -0
- data/ext/nokogiri/xml_namespace.c +186 -0
- data/ext/nokogiri/xml_node.c +2426 -0
- data/ext/nokogiri/xml_node_set.c +496 -0
- data/ext/nokogiri/xml_processing_instruction.c +54 -0
- data/ext/nokogiri/xml_reader.c +794 -0
- data/ext/nokogiri/xml_relax_ng.c +164 -0
- data/ext/nokogiri/xml_sax_parser.c +316 -0
- data/ext/nokogiri/xml_sax_parser_context.c +283 -0
- data/ext/nokogiri/xml_sax_push_parser.c +166 -0
- data/ext/nokogiri/xml_schema.c +260 -0
- data/ext/nokogiri/xml_syntax_error.c +85 -0
- data/ext/nokogiri/xml_text.c +48 -0
- data/ext/nokogiri/xml_xpath_context.c +415 -0
- data/ext/nokogiri/xslt_stylesheet.c +363 -0
- data/gumbo-parser/CHANGES.md +63 -0
- data/gumbo-parser/Makefile +111 -0
- data/gumbo-parser/THANKS +27 -0
- data/gumbo-parser/src/Makefile +34 -0
- data/gumbo-parser/src/README.md +41 -0
- data/gumbo-parser/src/ascii.c +75 -0
- data/gumbo-parser/src/ascii.h +115 -0
- data/gumbo-parser/src/attribute.c +42 -0
- data/gumbo-parser/src/attribute.h +17 -0
- data/gumbo-parser/src/char_ref.c +22225 -0
- data/gumbo-parser/src/char_ref.h +29 -0
- data/gumbo-parser/src/char_ref.rl +2154 -0
- data/gumbo-parser/src/error.c +626 -0
- data/gumbo-parser/src/error.h +148 -0
- data/gumbo-parser/src/foreign_attrs.c +104 -0
- data/gumbo-parser/src/foreign_attrs.gperf +27 -0
- data/gumbo-parser/src/insertion_mode.h +33 -0
- data/gumbo-parser/src/macros.h +91 -0
- data/gumbo-parser/src/nokogiri_gumbo.h +944 -0
- data/gumbo-parser/src/parser.c +4878 -0
- data/gumbo-parser/src/parser.h +41 -0
- data/gumbo-parser/src/replacement.h +33 -0
- data/gumbo-parser/src/string_buffer.c +103 -0
- data/gumbo-parser/src/string_buffer.h +68 -0
- data/gumbo-parser/src/string_piece.c +48 -0
- data/gumbo-parser/src/svg_attrs.c +174 -0
- data/gumbo-parser/src/svg_attrs.gperf +77 -0
- data/gumbo-parser/src/svg_tags.c +137 -0
- data/gumbo-parser/src/svg_tags.gperf +55 -0
- data/gumbo-parser/src/tag.c +223 -0
- data/gumbo-parser/src/tag_lookup.c +382 -0
- data/gumbo-parser/src/tag_lookup.gperf +170 -0
- data/gumbo-parser/src/tag_lookup.h +13 -0
- data/gumbo-parser/src/token_buffer.c +79 -0
- data/gumbo-parser/src/token_buffer.h +71 -0
- data/gumbo-parser/src/token_type.h +17 -0
- data/gumbo-parser/src/tokenizer.c +3463 -0
- data/gumbo-parser/src/tokenizer.h +112 -0
- data/gumbo-parser/src/tokenizer_states.h +339 -0
- data/gumbo-parser/src/utf8.c +245 -0
- data/gumbo-parser/src/utf8.h +164 -0
- data/gumbo-parser/src/util.c +66 -0
- data/gumbo-parser/src/util.h +34 -0
- data/gumbo-parser/src/vector.c +111 -0
- data/gumbo-parser/src/vector.h +45 -0
- data/lib/nokogiri/class_resolver.rb +67 -0
- data/lib/nokogiri/css/node.rb +54 -0
- data/lib/nokogiri/css/parser.rb +770 -0
- data/lib/nokogiri/css/parser.y +277 -0
- data/lib/nokogiri/css/parser_extras.rb +96 -0
- data/lib/nokogiri/css/syntax_error.rb +9 -0
- data/lib/nokogiri/css/tokenizer.rb +155 -0
- data/lib/nokogiri/css/tokenizer.rex +56 -0
- data/lib/nokogiri/css/xpath_visitor.rb +359 -0
- data/lib/nokogiri/css.rb +66 -0
- data/lib/nokogiri/decorators/slop.rb +44 -0
- data/lib/nokogiri/encoding_handler.rb +57 -0
- data/lib/nokogiri/extension.rb +32 -0
- data/lib/nokogiri/gumbo.rb +15 -0
- data/lib/nokogiri/html.rb +48 -0
- data/lib/nokogiri/html4/builder.rb +37 -0
- data/lib/nokogiri/html4/document.rb +214 -0
- data/lib/nokogiri/html4/document_fragment.rb +54 -0
- data/lib/nokogiri/html4/element_description.rb +25 -0
- data/lib/nokogiri/html4/element_description_defaults.rb +572 -0
- data/lib/nokogiri/html4/encoding_reader.rb +121 -0
- data/lib/nokogiri/html4/entity_lookup.rb +15 -0
- data/lib/nokogiri/html4/sax/parser.rb +63 -0
- data/lib/nokogiri/html4/sax/parser_context.rb +20 -0
- data/lib/nokogiri/html4/sax/push_parser.rb +37 -0
- data/lib/nokogiri/html4.rb +47 -0
- data/lib/nokogiri/html5/document.rb +168 -0
- data/lib/nokogiri/html5/document_fragment.rb +90 -0
- data/lib/nokogiri/html5/node.rb +98 -0
- data/lib/nokogiri/html5.rb +389 -0
- data/lib/nokogiri/jruby/dependencies.rb +3 -0
- data/lib/nokogiri/jruby/isorelax/isorelax/20030108/isorelax-20030108.jar +0 -0
- data/lib/nokogiri/jruby/net/sf/saxon/Saxon-HE/9.6.0-4/Saxon-HE-9.6.0-4.jar +0 -0
- data/lib/nokogiri/jruby/net/sourceforge/htmlunit/neko-htmlunit/2.63.0/neko-htmlunit-2.63.0.jar +0 -0
- data/lib/nokogiri/jruby/nokogiri_jars.rb +43 -0
- data/lib/nokogiri/jruby/nu/validator/jing/20200702VNU/jing-20200702VNU.jar +0 -0
- data/lib/nokogiri/jruby/org/nokogiri/nekodtd/0.1.11.noko2/nekodtd-0.1.11.noko2.jar +0 -0
- data/lib/nokogiri/jruby/xalan/serializer/2.7.3/serializer-2.7.3.jar +0 -0
- data/lib/nokogiri/jruby/xalan/xalan/2.7.3/xalan-2.7.3.jar +0 -0
- data/lib/nokogiri/jruby/xerces/xercesImpl/2.12.2/xercesImpl-2.12.2.jar +0 -0
- data/lib/nokogiri/jruby/xml-apis/xml-apis/1.4.01/xml-apis-1.4.01.jar +0 -0
- data/lib/nokogiri/syntax_error.rb +6 -0
- data/lib/nokogiri/version/constant.rb +6 -0
- data/lib/nokogiri/version/info.rb +223 -0
- data/lib/nokogiri/version.rb +4 -0
- data/lib/nokogiri/xml/attr.rb +66 -0
- data/lib/nokogiri/xml/attribute_decl.rb +20 -0
- data/lib/nokogiri/xml/builder.rb +487 -0
- data/lib/nokogiri/xml/cdata.rb +13 -0
- data/lib/nokogiri/xml/character_data.rb +9 -0
- data/lib/nokogiri/xml/document.rb +471 -0
- data/lib/nokogiri/xml/document_fragment.rb +205 -0
- data/lib/nokogiri/xml/dtd.rb +34 -0
- data/lib/nokogiri/xml/element_content.rb +38 -0
- data/lib/nokogiri/xml/element_decl.rb +15 -0
- data/lib/nokogiri/xml/entity_decl.rb +21 -0
- data/lib/nokogiri/xml/entity_reference.rb +20 -0
- data/lib/nokogiri/xml/namespace.rb +58 -0
- data/lib/nokogiri/xml/node/save_options.rb +68 -0
- data/lib/nokogiri/xml/node.rb +1563 -0
- data/lib/nokogiri/xml/node_set.rb +447 -0
- data/lib/nokogiri/xml/notation.rb +19 -0
- data/lib/nokogiri/xml/parse_options.rb +213 -0
- data/lib/nokogiri/xml/pp/character_data.rb +21 -0
- data/lib/nokogiri/xml/pp/node.rb +57 -0
- data/lib/nokogiri/xml/pp.rb +4 -0
- data/lib/nokogiri/xml/processing_instruction.rb +11 -0
- data/lib/nokogiri/xml/reader.rb +105 -0
- data/lib/nokogiri/xml/relax_ng.rb +38 -0
- data/lib/nokogiri/xml/sax/document.rb +167 -0
- data/lib/nokogiri/xml/sax/parser.rb +125 -0
- data/lib/nokogiri/xml/sax/parser_context.rb +21 -0
- data/lib/nokogiri/xml/sax/push_parser.rb +61 -0
- data/lib/nokogiri/xml/sax.rb +6 -0
- data/lib/nokogiri/xml/schema.rb +73 -0
- data/lib/nokogiri/xml/searchable.rb +270 -0
- data/lib/nokogiri/xml/syntax_error.rb +72 -0
- data/lib/nokogiri/xml/text.rb +11 -0
- data/lib/nokogiri/xml/xpath/syntax_error.rb +13 -0
- data/lib/nokogiri/xml/xpath.rb +21 -0
- data/lib/nokogiri/xml/xpath_context.rb +16 -0
- data/lib/nokogiri/xml.rb +76 -0
- data/lib/nokogiri/xslt/stylesheet.rb +27 -0
- data/lib/nokogiri/xslt.rb +65 -0
- data/lib/nokogiri.rb +120 -0
- data/lib/xsd/xmlparser/nokogiri.rb +106 -0
- metadata +391 -0
@@ -0,0 +1,214 @@
|
|
1
|
+
# coding: utf-8
|
2
|
+
# frozen_string_literal: true
|
3
|
+
|
4
|
+
require "pathname"
|
5
|
+
|
6
|
+
module Nokogiri
|
7
|
+
module HTML4
|
8
|
+
class Document < Nokogiri::XML::Document
|
9
|
+
###
|
10
|
+
# Get the meta tag encoding for this document. If there is no meta tag,
|
11
|
+
# then nil is returned.
|
12
|
+
def meta_encoding
|
13
|
+
if (meta = at_xpath("//meta[@charset]"))
|
14
|
+
meta[:charset]
|
15
|
+
elsif (meta = meta_content_type)
|
16
|
+
meta["content"][/charset\s*=\s*([\w-]+)/i, 1]
|
17
|
+
end
|
18
|
+
end
|
19
|
+
|
20
|
+
###
|
21
|
+
# Set the meta tag encoding for this document.
|
22
|
+
#
|
23
|
+
# If an meta encoding tag is already present, its content is
|
24
|
+
# replaced with the given text.
|
25
|
+
#
|
26
|
+
# Otherwise, this method tries to create one at an appropriate
|
27
|
+
# place supplying head and/or html elements as necessary, which
|
28
|
+
# is inside a head element if any, and before any text node or
|
29
|
+
# content element (typically <body>) if any.
|
30
|
+
#
|
31
|
+
# The result when trying to set an encoding that is different
|
32
|
+
# from the document encoding is undefined.
|
33
|
+
#
|
34
|
+
# Beware in CRuby, that libxml2 automatically inserts a meta tag
|
35
|
+
# into a head element.
|
36
|
+
def meta_encoding=(encoding)
|
37
|
+
if (meta = meta_content_type)
|
38
|
+
meta["content"] = format("text/html; charset=%s", encoding)
|
39
|
+
encoding
|
40
|
+
elsif (meta = at_xpath("//meta[@charset]"))
|
41
|
+
meta["charset"] = encoding
|
42
|
+
else
|
43
|
+
meta = XML::Node.new("meta", self)
|
44
|
+
if (dtd = internal_subset) && dtd.html5_dtd?
|
45
|
+
meta["charset"] = encoding
|
46
|
+
else
|
47
|
+
meta["http-equiv"] = "Content-Type"
|
48
|
+
meta["content"] = format("text/html; charset=%s", encoding)
|
49
|
+
end
|
50
|
+
|
51
|
+
if (head = at_xpath("//head"))
|
52
|
+
head.prepend_child(meta)
|
53
|
+
else
|
54
|
+
set_metadata_element(meta)
|
55
|
+
end
|
56
|
+
encoding
|
57
|
+
end
|
58
|
+
end
|
59
|
+
|
60
|
+
def meta_content_type
|
61
|
+
xpath("//meta[@http-equiv and boolean(@content)]").find do |node|
|
62
|
+
node["http-equiv"] =~ /\AContent-Type\z/i
|
63
|
+
end
|
64
|
+
end
|
65
|
+
private :meta_content_type
|
66
|
+
|
67
|
+
###
|
68
|
+
# Get the title string of this document. Return nil if there is
|
69
|
+
# no title tag.
|
70
|
+
def title
|
71
|
+
(title = at_xpath("//title")) && title.inner_text
|
72
|
+
end
|
73
|
+
|
74
|
+
###
|
75
|
+
# Set the title string of this document.
|
76
|
+
#
|
77
|
+
# If a title element is already present, its content is replaced
|
78
|
+
# with the given text.
|
79
|
+
#
|
80
|
+
# Otherwise, this method tries to create one at an appropriate
|
81
|
+
# place supplying head and/or html elements as necessary, which
|
82
|
+
# is inside a head element if any, right after a meta
|
83
|
+
# encoding/charset tag if any, and before any text node or
|
84
|
+
# content element (typically <body>) if any.
|
85
|
+
def title=(text)
|
86
|
+
tnode = XML::Text.new(text, self)
|
87
|
+
if (title = at_xpath("//title"))
|
88
|
+
title.children = tnode
|
89
|
+
return text
|
90
|
+
end
|
91
|
+
|
92
|
+
title = XML::Node.new("title", self) << tnode
|
93
|
+
if (head = at_xpath("//head"))
|
94
|
+
head << title
|
95
|
+
elsif (meta = (at_xpath("//meta[@charset]") || meta_content_type))
|
96
|
+
# better put after charset declaration
|
97
|
+
meta.add_next_sibling(title)
|
98
|
+
else
|
99
|
+
set_metadata_element(title)
|
100
|
+
end
|
101
|
+
end
|
102
|
+
|
103
|
+
def set_metadata_element(element) # rubocop:disable Naming/AccessorMethodName
|
104
|
+
if (head = at_xpath("//head"))
|
105
|
+
head << element
|
106
|
+
elsif (html = at_xpath("//html"))
|
107
|
+
head = html.prepend_child(XML::Node.new("head", self))
|
108
|
+
head.prepend_child(element)
|
109
|
+
elsif (first = children.find do |node|
|
110
|
+
case node
|
111
|
+
when XML::Element, XML::Text
|
112
|
+
true
|
113
|
+
end
|
114
|
+
end)
|
115
|
+
# We reach here only if the underlying document model
|
116
|
+
# allows <html>/<head> elements to be omitted and does not
|
117
|
+
# automatically supply them.
|
118
|
+
first.add_previous_sibling(element)
|
119
|
+
else
|
120
|
+
html = add_child(XML::Node.new("html", self))
|
121
|
+
head = html.add_child(XML::Node.new("head", self))
|
122
|
+
head.prepend_child(element)
|
123
|
+
end
|
124
|
+
end
|
125
|
+
private :set_metadata_element
|
126
|
+
|
127
|
+
####
|
128
|
+
# Serialize Node using +options+. Save options can also be set using a block.
|
129
|
+
#
|
130
|
+
# See also Nokogiri::XML::Node::SaveOptions and Node@Serialization+and+Generating+Output.
|
131
|
+
#
|
132
|
+
# These two statements are equivalent:
|
133
|
+
#
|
134
|
+
# node.serialize(:encoding => 'UTF-8', :save_with => FORMAT | AS_XML)
|
135
|
+
#
|
136
|
+
# or
|
137
|
+
#
|
138
|
+
# node.serialize(:encoding => 'UTF-8') do |config|
|
139
|
+
# config.format.as_xml
|
140
|
+
# end
|
141
|
+
#
|
142
|
+
def serialize(options = {})
|
143
|
+
options[:save_with] ||= XML::Node::SaveOptions::DEFAULT_HTML
|
144
|
+
super
|
145
|
+
end
|
146
|
+
|
147
|
+
####
|
148
|
+
# Create a Nokogiri::XML::DocumentFragment from +tags+
|
149
|
+
def fragment(tags = nil)
|
150
|
+
DocumentFragment.new(self, tags, root)
|
151
|
+
end
|
152
|
+
|
153
|
+
# :call-seq:
|
154
|
+
# xpath_doctype() → Nokogiri::CSS::XPathVisitor::DoctypeConfig
|
155
|
+
#
|
156
|
+
# [Returns] The document type which determines CSS-to-XPath translation.
|
157
|
+
#
|
158
|
+
# See XPathVisitor for more information.
|
159
|
+
def xpath_doctype
|
160
|
+
Nokogiri::CSS::XPathVisitor::DoctypeConfig::HTML4
|
161
|
+
end
|
162
|
+
|
163
|
+
class << self
|
164
|
+
###
|
165
|
+
# Parse HTML. +string_or_io+ may be a String, or any object that
|
166
|
+
# responds to _read_ and _close_ such as an IO, or StringIO.
|
167
|
+
# +url+ is resource where this document is located. +encoding+ is the
|
168
|
+
# encoding that should be used when processing the document. +options+
|
169
|
+
# is a number that sets options in the parser, such as
|
170
|
+
# Nokogiri::XML::ParseOptions::RECOVER. See the constants in
|
171
|
+
# Nokogiri::XML::ParseOptions.
|
172
|
+
def parse(string_or_io, url = nil, encoding = nil, options = XML::ParseOptions::DEFAULT_HTML)
|
173
|
+
options = Nokogiri::XML::ParseOptions.new(options) if Integer === options
|
174
|
+
yield options if block_given?
|
175
|
+
|
176
|
+
url ||= string_or_io.respond_to?(:path) ? string_or_io.path : nil
|
177
|
+
|
178
|
+
if string_or_io.respond_to?(:encoding)
|
179
|
+
unless string_or_io.encoding == Encoding::ASCII_8BIT
|
180
|
+
encoding ||= string_or_io.encoding.name
|
181
|
+
end
|
182
|
+
end
|
183
|
+
|
184
|
+
if string_or_io.respond_to?(:read)
|
185
|
+
if string_or_io.is_a?(Pathname)
|
186
|
+
# resolve the Pathname to the file and open it as an IO object, see #2110
|
187
|
+
string_or_io = string_or_io.expand_path.open
|
188
|
+
url ||= string_or_io.path
|
189
|
+
end
|
190
|
+
|
191
|
+
unless encoding
|
192
|
+
string_or_io = EncodingReader.new(string_or_io)
|
193
|
+
begin
|
194
|
+
return read_io(string_or_io, url, encoding, options.to_i)
|
195
|
+
rescue EncodingReader::EncodingFound => e
|
196
|
+
encoding = e.found_encoding
|
197
|
+
end
|
198
|
+
end
|
199
|
+
return read_io(string_or_io, url, encoding, options.to_i)
|
200
|
+
end
|
201
|
+
|
202
|
+
# read_memory pukes on empty docs
|
203
|
+
if string_or_io.nil? || string_or_io.empty?
|
204
|
+
return encoding ? new.tap { |i| i.encoding = encoding } : new
|
205
|
+
end
|
206
|
+
|
207
|
+
encoding ||= EncodingReader.detect_encoding(string_or_io)
|
208
|
+
|
209
|
+
read_memory(string_or_io, url, encoding, options.to_i)
|
210
|
+
end
|
211
|
+
end
|
212
|
+
end
|
213
|
+
end
|
214
|
+
end
|
@@ -0,0 +1,54 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
|
3
|
+
module Nokogiri
|
4
|
+
module HTML4
|
5
|
+
class DocumentFragment < Nokogiri::XML::DocumentFragment
|
6
|
+
####
|
7
|
+
# Create a Nokogiri::XML::DocumentFragment from +tags+, using +encoding+
|
8
|
+
def self.parse(tags, encoding = nil, options = XML::ParseOptions::DEFAULT_HTML, &block)
|
9
|
+
doc = HTML4::Document.new
|
10
|
+
|
11
|
+
encoding ||= if tags.respond_to?(:encoding)
|
12
|
+
encoding = tags.encoding
|
13
|
+
if encoding == ::Encoding::ASCII_8BIT
|
14
|
+
"UTF-8"
|
15
|
+
else
|
16
|
+
encoding.name
|
17
|
+
end
|
18
|
+
else
|
19
|
+
"UTF-8"
|
20
|
+
end
|
21
|
+
|
22
|
+
doc.encoding = encoding
|
23
|
+
|
24
|
+
new(doc, tags, nil, options, &block)
|
25
|
+
end
|
26
|
+
|
27
|
+
def initialize(document, tags = nil, ctx = nil, options = XML::ParseOptions::DEFAULT_HTML)
|
28
|
+
return self unless tags
|
29
|
+
|
30
|
+
options = Nokogiri::XML::ParseOptions.new(options) if Integer === options
|
31
|
+
yield options if block_given?
|
32
|
+
|
33
|
+
if ctx
|
34
|
+
preexisting_errors = document.errors.dup
|
35
|
+
node_set = ctx.parse("<div>#{tags}</div>", options)
|
36
|
+
node_set.first.children.each { |child| child.parent = self } unless node_set.empty?
|
37
|
+
self.errors = document.errors - preexisting_errors
|
38
|
+
else
|
39
|
+
# This is a horrible hack, but I don't care
|
40
|
+
path = if /^\s*?<body/i.match?(tags)
|
41
|
+
"/html/body"
|
42
|
+
else
|
43
|
+
"/html/body/node()"
|
44
|
+
end
|
45
|
+
|
46
|
+
temp_doc = HTML4::Document.parse("<html><body>#{tags}", nil, document.encoding, options)
|
47
|
+
temp_doc.xpath(path).each { |child| child.parent = self }
|
48
|
+
self.errors = temp_doc.errors
|
49
|
+
end
|
50
|
+
children
|
51
|
+
end
|
52
|
+
end
|
53
|
+
end
|
54
|
+
end
|
@@ -0,0 +1,25 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
|
3
|
+
module Nokogiri
|
4
|
+
module HTML4
|
5
|
+
class ElementDescription
|
6
|
+
###
|
7
|
+
# Is this element a block element?
|
8
|
+
def block?
|
9
|
+
!inline?
|
10
|
+
end
|
11
|
+
|
12
|
+
###
|
13
|
+
# Convert this description to a string
|
14
|
+
def to_s
|
15
|
+
"#{name}: #{description}"
|
16
|
+
end
|
17
|
+
|
18
|
+
###
|
19
|
+
# Inspection information
|
20
|
+
def inspect
|
21
|
+
"#<#{self.class.name}: #{name} #{description}>"
|
22
|
+
end
|
23
|
+
end
|
24
|
+
end
|
25
|
+
end
|