nokogiri-backport 1.11.0
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +7 -0
- data/Gemfile +3 -0
- data/LICENSE-DEPENDENCIES.md +1682 -0
- data/LICENSE.md +9 -0
- data/README.md +272 -0
- data/bin/nokogiri +118 -0
- data/dependencies.yml +74 -0
- data/ext/java/nokogiri/EncodingHandler.java +124 -0
- data/ext/java/nokogiri/HtmlDocument.java +178 -0
- data/ext/java/nokogiri/HtmlElementDescription.java +148 -0
- data/ext/java/nokogiri/HtmlEntityLookup.java +79 -0
- data/ext/java/nokogiri/HtmlSaxParserContext.java +282 -0
- data/ext/java/nokogiri/HtmlSaxPushParser.java +222 -0
- data/ext/java/nokogiri/NokogiriService.java +597 -0
- data/ext/java/nokogiri/XmlAttr.java +162 -0
- data/ext/java/nokogiri/XmlAttributeDecl.java +129 -0
- data/ext/java/nokogiri/XmlCdata.java +82 -0
- data/ext/java/nokogiri/XmlComment.java +97 -0
- data/ext/java/nokogiri/XmlDocument.java +633 -0
- data/ext/java/nokogiri/XmlDocumentFragment.java +185 -0
- data/ext/java/nokogiri/XmlDtd.java +481 -0
- data/ext/java/nokogiri/XmlElement.java +68 -0
- data/ext/java/nokogiri/XmlElementContent.java +382 -0
- data/ext/java/nokogiri/XmlElementDecl.java +147 -0
- data/ext/java/nokogiri/XmlEntityDecl.java +157 -0
- data/ext/java/nokogiri/XmlEntityReference.java +101 -0
- data/ext/java/nokogiri/XmlNamespace.java +199 -0
- data/ext/java/nokogiri/XmlNode.java +1684 -0
- data/ext/java/nokogiri/XmlNodeSet.java +434 -0
- data/ext/java/nokogiri/XmlProcessingInstruction.java +100 -0
- data/ext/java/nokogiri/XmlReader.java +531 -0
- data/ext/java/nokogiri/XmlRelaxng.java +151 -0
- data/ext/java/nokogiri/XmlSaxParserContext.java +374 -0
- data/ext/java/nokogiri/XmlSaxPushParser.java +286 -0
- data/ext/java/nokogiri/XmlSchema.java +388 -0
- data/ext/java/nokogiri/XmlSyntaxError.java +138 -0
- data/ext/java/nokogiri/XmlText.java +110 -0
- data/ext/java/nokogiri/XmlXpathContext.java +301 -0
- data/ext/java/nokogiri/XsltStylesheet.java +347 -0
- data/ext/java/nokogiri/internals/ClosedStreamException.java +10 -0
- data/ext/java/nokogiri/internals/HtmlDomParserContext.java +252 -0
- data/ext/java/nokogiri/internals/IgnoreSchemaErrorsErrorHandler.java +20 -0
- data/ext/java/nokogiri/internals/NokogiriBlockingQueueInputStream.java +151 -0
- data/ext/java/nokogiri/internals/NokogiriDomParser.java +116 -0
- data/ext/java/nokogiri/internals/NokogiriEntityResolver.java +121 -0
- data/ext/java/nokogiri/internals/NokogiriErrorHandler.java +69 -0
- data/ext/java/nokogiri/internals/NokogiriHandler.java +327 -0
- data/ext/java/nokogiri/internals/NokogiriHelpers.java +734 -0
- data/ext/java/nokogiri/internals/NokogiriNamespaceCache.java +217 -0
- data/ext/java/nokogiri/internals/NokogiriNamespaceContext.java +127 -0
- data/ext/java/nokogiri/internals/NokogiriNonStrictErrorHandler.java +100 -0
- data/ext/java/nokogiri/internals/NokogiriNonStrictErrorHandler4NekoHtml.java +121 -0
- data/ext/java/nokogiri/internals/NokogiriStrictErrorHandler.java +78 -0
- data/ext/java/nokogiri/internals/NokogiriXPathFunction.java +180 -0
- data/ext/java/nokogiri/internals/NokogiriXPathFunctionResolver.java +72 -0
- data/ext/java/nokogiri/internals/NokogiriXPathVariableResolver.java +60 -0
- data/ext/java/nokogiri/internals/NokogiriXsltErrorListener.java +87 -0
- data/ext/java/nokogiri/internals/ParserContext.java +259 -0
- data/ext/java/nokogiri/internals/ReaderNode.java +488 -0
- data/ext/java/nokogiri/internals/SaveContextVisitor.java +778 -0
- data/ext/java/nokogiri/internals/SchemaErrorHandler.java +73 -0
- data/ext/java/nokogiri/internals/XalanDTMManagerPatch.java +168 -0
- data/ext/java/nokogiri/internals/XmlDeclHandler.java +42 -0
- data/ext/java/nokogiri/internals/XmlDomParserContext.java +274 -0
- data/ext/java/nokogiri/internals/XmlSaxParser.java +65 -0
- data/ext/java/nokogiri/internals/c14n/AttrCompare.java +119 -0
- data/ext/java/nokogiri/internals/c14n/C14nHelper.java +159 -0
- data/ext/java/nokogiri/internals/c14n/CanonicalFilter.java +37 -0
- data/ext/java/nokogiri/internals/c14n/CanonicalizationException.java +93 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer.java +252 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer11.java +639 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer11_OmitComments.java +38 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer11_WithComments.java +38 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer20010315.java +367 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer20010315Excl.java +295 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer20010315ExclOmitComments.java +40 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer20010315ExclWithComments.java +44 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer20010315OmitComments.java +44 -0
- data/ext/java/nokogiri/internals/c14n/Canonicalizer20010315WithComments.java +43 -0
- data/ext/java/nokogiri/internals/c14n/CanonicalizerBase.java +630 -0
- data/ext/java/nokogiri/internals/c14n/CanonicalizerPhysical.java +173 -0
- data/ext/java/nokogiri/internals/c14n/CanonicalizerSpi.java +76 -0
- data/ext/java/nokogiri/internals/c14n/Constants.java +42 -0
- data/ext/java/nokogiri/internals/c14n/ElementProxy.java +293 -0
- data/ext/java/nokogiri/internals/c14n/HelperNodeList.java +93 -0
- data/ext/java/nokogiri/internals/c14n/IgnoreAllErrorHandler.java +79 -0
- data/ext/java/nokogiri/internals/c14n/InclusiveNamespaces.java +166 -0
- data/ext/java/nokogiri/internals/c14n/InvalidCanonicalizerException.java +76 -0
- data/ext/java/nokogiri/internals/c14n/NameSpaceSymbTable.java +402 -0
- data/ext/java/nokogiri/internals/c14n/NodeFilter.java +51 -0
- data/ext/java/nokogiri/internals/c14n/UtfHelpper.java +179 -0
- data/ext/java/nokogiri/internals/c14n/XMLUtils.java +507 -0
- data/ext/java/nokogiri/internals/dom2dtm/DOM2DTM.java +1745 -0
- data/ext/java/nokogiri/internals/dom2dtm/DOM2DTMdefaultNamespaceDeclarationNode.java +685 -0
- data/ext/nokogiri/depend +477 -0
- data/ext/nokogiri/extconf.rb +836 -0
- data/ext/nokogiri/html_document.c +171 -0
- data/ext/nokogiri/html_document.h +10 -0
- data/ext/nokogiri/html_element_description.c +279 -0
- data/ext/nokogiri/html_element_description.h +10 -0
- data/ext/nokogiri/html_entity_lookup.c +32 -0
- data/ext/nokogiri/html_entity_lookup.h +8 -0
- data/ext/nokogiri/html_sax_parser_context.c +116 -0
- data/ext/nokogiri/html_sax_parser_context.h +11 -0
- data/ext/nokogiri/html_sax_push_parser.c +87 -0
- data/ext/nokogiri/html_sax_push_parser.h +9 -0
- data/ext/nokogiri/nokogiri.c +135 -0
- data/ext/nokogiri/nokogiri.h +130 -0
- data/ext/nokogiri/xml_attr.c +103 -0
- data/ext/nokogiri/xml_attr.h +9 -0
- data/ext/nokogiri/xml_attribute_decl.c +70 -0
- data/ext/nokogiri/xml_attribute_decl.h +9 -0
- data/ext/nokogiri/xml_cdata.c +62 -0
- data/ext/nokogiri/xml_cdata.h +9 -0
- data/ext/nokogiri/xml_comment.c +69 -0
- data/ext/nokogiri/xml_comment.h +9 -0
- data/ext/nokogiri/xml_document.c +622 -0
- data/ext/nokogiri/xml_document.h +23 -0
- data/ext/nokogiri/xml_document_fragment.c +48 -0
- data/ext/nokogiri/xml_document_fragment.h +10 -0
- data/ext/nokogiri/xml_dtd.c +202 -0
- data/ext/nokogiri/xml_dtd.h +10 -0
- data/ext/nokogiri/xml_element_content.c +123 -0
- data/ext/nokogiri/xml_element_content.h +10 -0
- data/ext/nokogiri/xml_element_decl.c +69 -0
- data/ext/nokogiri/xml_element_decl.h +9 -0
- data/ext/nokogiri/xml_encoding_handler.c +79 -0
- data/ext/nokogiri/xml_encoding_handler.h +8 -0
- data/ext/nokogiri/xml_entity_decl.c +110 -0
- data/ext/nokogiri/xml_entity_decl.h +10 -0
- data/ext/nokogiri/xml_entity_reference.c +52 -0
- data/ext/nokogiri/xml_entity_reference.h +9 -0
- data/ext/nokogiri/xml_io.c +63 -0
- data/ext/nokogiri/xml_io.h +11 -0
- data/ext/nokogiri/xml_libxml2_hacks.c +112 -0
- data/ext/nokogiri/xml_libxml2_hacks.h +12 -0
- data/ext/nokogiri/xml_namespace.c +111 -0
- data/ext/nokogiri/xml_namespace.h +14 -0
- data/ext/nokogiri/xml_node.c +1773 -0
- data/ext/nokogiri/xml_node.h +13 -0
- data/ext/nokogiri/xml_node_set.c +486 -0
- data/ext/nokogiri/xml_node_set.h +12 -0
- data/ext/nokogiri/xml_processing_instruction.c +56 -0
- data/ext/nokogiri/xml_processing_instruction.h +9 -0
- data/ext/nokogiri/xml_reader.c +657 -0
- data/ext/nokogiri/xml_reader.h +10 -0
- data/ext/nokogiri/xml_relax_ng.c +179 -0
- data/ext/nokogiri/xml_relax_ng.h +9 -0
- data/ext/nokogiri/xml_sax_parser.c +305 -0
- data/ext/nokogiri/xml_sax_parser.h +39 -0
- data/ext/nokogiri/xml_sax_parser_context.c +262 -0
- data/ext/nokogiri/xml_sax_parser_context.h +10 -0
- data/ext/nokogiri/xml_sax_push_parser.c +159 -0
- data/ext/nokogiri/xml_sax_push_parser.h +9 -0
- data/ext/nokogiri/xml_schema.c +276 -0
- data/ext/nokogiri/xml_schema.h +9 -0
- data/ext/nokogiri/xml_syntax_error.c +64 -0
- data/ext/nokogiri/xml_syntax_error.h +13 -0
- data/ext/nokogiri/xml_text.c +52 -0
- data/ext/nokogiri/xml_text.h +9 -0
- data/ext/nokogiri/xml_xpath_context.c +374 -0
- data/ext/nokogiri/xml_xpath_context.h +10 -0
- data/ext/nokogiri/xslt_stylesheet.c +263 -0
- data/ext/nokogiri/xslt_stylesheet.h +14 -0
- data/lib/isorelax.jar +0 -0
- data/lib/jing.jar +0 -0
- data/lib/nekodtd.jar +0 -0
- data/lib/nekohtml.jar +0 -0
- data/lib/nokogiri/css/node.rb +53 -0
- data/lib/nokogiri/css/parser.rb +751 -0
- data/lib/nokogiri/css/parser.y +272 -0
- data/lib/nokogiri/css/parser_extras.rb +94 -0
- data/lib/nokogiri/css/syntax_error.rb +8 -0
- data/lib/nokogiri/css/tokenizer.rb +154 -0
- data/lib/nokogiri/css/tokenizer.rex +55 -0
- data/lib/nokogiri/css/xpath_visitor.rb +260 -0
- data/lib/nokogiri/css.rb +28 -0
- data/lib/nokogiri/decorators/slop.rb +43 -0
- data/lib/nokogiri/html/builder.rb +36 -0
- data/lib/nokogiri/html/document.rb +322 -0
- data/lib/nokogiri/html/document_fragment.rb +50 -0
- data/lib/nokogiri/html/element_description.rb +24 -0
- data/lib/nokogiri/html/element_description_defaults.rb +672 -0
- data/lib/nokogiri/html/entity_lookup.rb +14 -0
- data/lib/nokogiri/html/sax/parser.rb +63 -0
- data/lib/nokogiri/html/sax/parser_context.rb +17 -0
- data/lib/nokogiri/html/sax/push_parser.rb +37 -0
- data/lib/nokogiri/html.rb +38 -0
- data/lib/nokogiri/jruby/dependencies.rb +20 -0
- data/lib/nokogiri/syntax_error.rb +5 -0
- data/lib/nokogiri/version/constant.rb +5 -0
- data/lib/nokogiri/version/info.rb +182 -0
- data/lib/nokogiri/version.rb +3 -0
- data/lib/nokogiri/xml/attr.rb +15 -0
- data/lib/nokogiri/xml/attribute_decl.rb +19 -0
- data/lib/nokogiri/xml/builder.rb +447 -0
- data/lib/nokogiri/xml/cdata.rb +12 -0
- data/lib/nokogiri/xml/character_data.rb +8 -0
- data/lib/nokogiri/xml/document.rb +290 -0
- data/lib/nokogiri/xml/document_fragment.rb +159 -0
- data/lib/nokogiri/xml/dtd.rb +33 -0
- data/lib/nokogiri/xml/element_content.rb +37 -0
- data/lib/nokogiri/xml/element_decl.rb +14 -0
- data/lib/nokogiri/xml/entity_decl.rb +20 -0
- data/lib/nokogiri/xml/entity_reference.rb +19 -0
- data/lib/nokogiri/xml/namespace.rb +14 -0
- data/lib/nokogiri/xml/node/save_options.rb +62 -0
- data/lib/nokogiri/xml/node.rb +1240 -0
- data/lib/nokogiri/xml/node_set.rb +372 -0
- data/lib/nokogiri/xml/notation.rb +7 -0
- data/lib/nokogiri/xml/parse_options.rb +127 -0
- data/lib/nokogiri/xml/pp/character_data.rb +19 -0
- data/lib/nokogiri/xml/pp/node.rb +57 -0
- data/lib/nokogiri/xml/pp.rb +3 -0
- data/lib/nokogiri/xml/processing_instruction.rb +9 -0
- data/lib/nokogiri/xml/reader.rb +116 -0
- data/lib/nokogiri/xml/relax_ng.rb +37 -0
- data/lib/nokogiri/xml/sax/document.rb +172 -0
- data/lib/nokogiri/xml/sax/parser.rb +123 -0
- data/lib/nokogiri/xml/sax/parser_context.rb +17 -0
- data/lib/nokogiri/xml/sax/push_parser.rb +61 -0
- data/lib/nokogiri/xml/sax.rb +5 -0
- data/lib/nokogiri/xml/schema.rb +72 -0
- data/lib/nokogiri/xml/searchable.rb +239 -0
- data/lib/nokogiri/xml/syntax_error.rb +71 -0
- data/lib/nokogiri/xml/text.rb +10 -0
- data/lib/nokogiri/xml/xpath/syntax_error.rb +12 -0
- data/lib/nokogiri/xml/xpath.rb +11 -0
- data/lib/nokogiri/xml/xpath_context.rb +17 -0
- data/lib/nokogiri/xml.rb +76 -0
- data/lib/nokogiri/xslt/stylesheet.rb +26 -0
- data/lib/nokogiri/xslt.rb +57 -0
- data/lib/nokogiri.rb +144 -0
- data/lib/serializer.jar +0 -0
- data/lib/xalan.jar +0 -0
- data/lib/xercesImpl.jar +0 -0
- data/lib/xml-apis.jar +0 -0
- data/lib/xsd/xmlparser/nokogiri.rb +103 -0
- metadata +531 -0
@@ -0,0 +1,116 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
module Nokogiri
|
3
|
+
module XML
|
4
|
+
###
|
5
|
+
# Nokogiri::XML::Reader parses an XML document similar to the way a cursor
|
6
|
+
# would move. The Reader is given an XML document, and yields nodes
|
7
|
+
# to an each block.
|
8
|
+
#
|
9
|
+
# Here is an example of usage:
|
10
|
+
#
|
11
|
+
# reader = Nokogiri::XML::Reader(<<-eoxml)
|
12
|
+
# <x xmlns:tenderlove='http://tenderlovemaking.com/'>
|
13
|
+
# <tenderlove:foo awesome='true'>snuggles!</tenderlove:foo>
|
14
|
+
# </x>
|
15
|
+
# eoxml
|
16
|
+
#
|
17
|
+
# reader.each do |node|
|
18
|
+
#
|
19
|
+
# # node is an instance of Nokogiri::XML::Reader
|
20
|
+
# puts node.name
|
21
|
+
#
|
22
|
+
# end
|
23
|
+
#
|
24
|
+
# Note that Nokogiri::XML::Reader#each can only be called once!! Once
|
25
|
+
# the cursor moves through the entire document, you must parse the
|
26
|
+
# document again. So make sure that you capture any information you
|
27
|
+
# need during the first iteration.
|
28
|
+
#
|
29
|
+
# The Reader parser is good for when you need the speed of a SAX parser,
|
30
|
+
# but do not want to write a Document handler.
|
31
|
+
class Reader
|
32
|
+
include Enumerable
|
33
|
+
|
34
|
+
TYPE_NONE = 0
|
35
|
+
# Element node type
|
36
|
+
TYPE_ELEMENT = 1
|
37
|
+
# Attribute node type
|
38
|
+
TYPE_ATTRIBUTE = 2
|
39
|
+
# Text node type
|
40
|
+
TYPE_TEXT = 3
|
41
|
+
# CDATA node type
|
42
|
+
TYPE_CDATA = 4
|
43
|
+
# Entity Reference node type
|
44
|
+
TYPE_ENTITY_REFERENCE = 5
|
45
|
+
# Entity node type
|
46
|
+
TYPE_ENTITY = 6
|
47
|
+
# PI node type
|
48
|
+
TYPE_PROCESSING_INSTRUCTION = 7
|
49
|
+
# Comment node type
|
50
|
+
TYPE_COMMENT = 8
|
51
|
+
# Document node type
|
52
|
+
TYPE_DOCUMENT = 9
|
53
|
+
# Document Type node type
|
54
|
+
TYPE_DOCUMENT_TYPE = 10
|
55
|
+
# Document Fragment node type
|
56
|
+
TYPE_DOCUMENT_FRAGMENT = 11
|
57
|
+
# Notation node type
|
58
|
+
TYPE_NOTATION = 12
|
59
|
+
# Whitespace node type
|
60
|
+
TYPE_WHITESPACE = 13
|
61
|
+
# Significant Whitespace node type
|
62
|
+
TYPE_SIGNIFICANT_WHITESPACE = 14
|
63
|
+
# Element end node type
|
64
|
+
TYPE_END_ELEMENT = 15
|
65
|
+
# Entity end node type
|
66
|
+
TYPE_END_ENTITY = 16
|
67
|
+
# XML Declaration node type
|
68
|
+
TYPE_XML_DECLARATION = 17
|
69
|
+
|
70
|
+
# A list of errors encountered while parsing
|
71
|
+
attr_accessor :errors
|
72
|
+
|
73
|
+
# The encoding for the document
|
74
|
+
attr_reader :encoding
|
75
|
+
|
76
|
+
# The XML source
|
77
|
+
attr_reader :source
|
78
|
+
|
79
|
+
alias :self_closing? :empty_element?
|
80
|
+
|
81
|
+
def initialize source, url = nil, encoding = nil # :nodoc:
|
82
|
+
@source = source
|
83
|
+
@errors = []
|
84
|
+
@encoding = encoding
|
85
|
+
end
|
86
|
+
private :initialize
|
87
|
+
|
88
|
+
###
|
89
|
+
# Get a list of attributes for the current node.
|
90
|
+
def attributes
|
91
|
+
attrs_hash = attribute_nodes.each_with_object({}) do |node, hash|
|
92
|
+
hash[node.name] = node.to_s
|
93
|
+
end
|
94
|
+
ns = namespaces
|
95
|
+
attrs_hash.merge!(ns) if ns
|
96
|
+
attrs_hash
|
97
|
+
end
|
98
|
+
|
99
|
+
###
|
100
|
+
# Get a list of attributes for the current node
|
101
|
+
def attribute_nodes
|
102
|
+
nodes = attr_nodes
|
103
|
+
nodes.each { |v| v.instance_variable_set(:@_r, self) }
|
104
|
+
nodes
|
105
|
+
end
|
106
|
+
|
107
|
+
###
|
108
|
+
# Move the cursor through the document yielding the cursor to the block
|
109
|
+
def each
|
110
|
+
while cursor = self.read
|
111
|
+
yield cursor
|
112
|
+
end
|
113
|
+
end
|
114
|
+
end
|
115
|
+
end
|
116
|
+
end
|
@@ -0,0 +1,37 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
module Nokogiri
|
3
|
+
module XML
|
4
|
+
class << self
|
5
|
+
###
|
6
|
+
# Create a new Nokogiri::XML::RelaxNG document from +string_or_io+.
|
7
|
+
# See Nokogiri::XML::RelaxNG for an example.
|
8
|
+
def RelaxNG(string_or_io, options = ParseOptions::DEFAULT_SCHEMA)
|
9
|
+
RelaxNG.new(string_or_io, options)
|
10
|
+
end
|
11
|
+
end
|
12
|
+
|
13
|
+
###
|
14
|
+
# Nokogiri::XML::RelaxNG is used for validating XML against a
|
15
|
+
# RelaxNG schema.
|
16
|
+
#
|
17
|
+
# == Synopsis
|
18
|
+
#
|
19
|
+
# Validate an XML document against a RelaxNG schema. Loop over the errors
|
20
|
+
# that are returned and print them out:
|
21
|
+
#
|
22
|
+
# schema = Nokogiri::XML::RelaxNG(File.open(ADDRESS_SCHEMA_FILE))
|
23
|
+
# doc = Nokogiri::XML(File.open(ADDRESS_XML_FILE))
|
24
|
+
#
|
25
|
+
# schema.validate(doc).each do |error|
|
26
|
+
# puts error.message
|
27
|
+
# end
|
28
|
+
#
|
29
|
+
# The list of errors are Nokogiri::XML::SyntaxError objects.
|
30
|
+
#
|
31
|
+
# NOTE: RelaxNG input is always treated as TRUSTED documents, meaning that they will cause the
|
32
|
+
# underlying parsing libraries to access network resources. This is counter to Nokogiri's
|
33
|
+
# "untrusted by default" security policy, but is a limitation of the underlying libraries.
|
34
|
+
class RelaxNG < Nokogiri::XML::Schema
|
35
|
+
end
|
36
|
+
end
|
37
|
+
end
|
@@ -0,0 +1,172 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
module Nokogiri
|
3
|
+
module XML
|
4
|
+
###
|
5
|
+
# SAX Parsers are event driven parsers. Nokogiri provides two different
|
6
|
+
# event based parsers when dealing with XML. If you want to do SAX style
|
7
|
+
# parsing using HTML, check out Nokogiri::HTML::SAX.
|
8
|
+
#
|
9
|
+
# The basic way a SAX style parser works is by creating a parser,
|
10
|
+
# telling the parser about the events we're interested in, then giving
|
11
|
+
# the parser some XML to process. The parser will notify you when
|
12
|
+
# it encounters events you said you would like to know about.
|
13
|
+
#
|
14
|
+
# To register for events, you simply subclass Nokogiri::XML::SAX::Document,
|
15
|
+
# and implement the methods for which you would like notification.
|
16
|
+
#
|
17
|
+
# For example, if I want to be notified when a document ends, and when an
|
18
|
+
# element starts, I would write a class like this:
|
19
|
+
#
|
20
|
+
# class MyDocument < Nokogiri::XML::SAX::Document
|
21
|
+
# def end_document
|
22
|
+
# puts "the document has ended"
|
23
|
+
# end
|
24
|
+
#
|
25
|
+
# def start_element name, attributes = []
|
26
|
+
# puts "#{name} started"
|
27
|
+
# end
|
28
|
+
# end
|
29
|
+
#
|
30
|
+
# Then I would instantiate a SAX parser with this document, and feed the
|
31
|
+
# parser some XML
|
32
|
+
#
|
33
|
+
# # Create a new parser
|
34
|
+
# parser = Nokogiri::XML::SAX::Parser.new(MyDocument.new)
|
35
|
+
#
|
36
|
+
# # Feed the parser some XML
|
37
|
+
# parser.parse(File.open(ARGV[0]))
|
38
|
+
#
|
39
|
+
# Now my document handler will be called when each node starts, and when
|
40
|
+
# then document ends. To see what kinds of events are available, take
|
41
|
+
# a look at Nokogiri::XML::SAX::Document.
|
42
|
+
#
|
43
|
+
# Two SAX parsers for XML are available, a parser that reads from a string
|
44
|
+
# or IO object as it feels necessary, and a parser that lets you spoon
|
45
|
+
# feed it XML. If you want to let Nokogiri deal with reading your XML,
|
46
|
+
# use the Nokogiri::XML::SAX::Parser. If you want to have fine grain
|
47
|
+
# control over the XML input, use the Nokogiri::XML::SAX::PushParser.
|
48
|
+
module SAX
|
49
|
+
###
|
50
|
+
# This class is used for registering types of events you are interested
|
51
|
+
# in handling. All of the methods on this class are available as
|
52
|
+
# possible events while parsing an XML document. To register for any
|
53
|
+
# particular event, just subclass this class and implement the methods
|
54
|
+
# you are interested in knowing about.
|
55
|
+
#
|
56
|
+
# To only be notified about start and end element events, write a class
|
57
|
+
# like this:
|
58
|
+
#
|
59
|
+
# class MyDocument < Nokogiri::XML::SAX::Document
|
60
|
+
# def start_element name, attrs = []
|
61
|
+
# puts "#{name} started!"
|
62
|
+
# end
|
63
|
+
#
|
64
|
+
# def end_element name
|
65
|
+
# puts "#{name} ended"
|
66
|
+
# end
|
67
|
+
# end
|
68
|
+
#
|
69
|
+
# You can use this event handler for any SAX style parser included with
|
70
|
+
# Nokogiri. See Nokogiri::XML::SAX, and Nokogiri::HTML::SAX.
|
71
|
+
class Document
|
72
|
+
###
|
73
|
+
# Called when an XML declaration is parsed
|
74
|
+
def xmldecl version, encoding, standalone
|
75
|
+
end
|
76
|
+
|
77
|
+
###
|
78
|
+
# Called when document starts parsing
|
79
|
+
def start_document
|
80
|
+
end
|
81
|
+
|
82
|
+
###
|
83
|
+
# Called when document ends parsing
|
84
|
+
def end_document
|
85
|
+
end
|
86
|
+
|
87
|
+
###
|
88
|
+
# Called at the beginning of an element
|
89
|
+
# * +name+ is the name of the tag
|
90
|
+
# * +attrs+ are an assoc list of namespaces and attributes, e.g.:
|
91
|
+
# [ ["xmlns:foo", "http://sample.net"], ["size", "large"] ]
|
92
|
+
def start_element name, attrs = []
|
93
|
+
end
|
94
|
+
|
95
|
+
###
|
96
|
+
# Called at the end of an element
|
97
|
+
# +name+ is the tag name
|
98
|
+
def end_element name
|
99
|
+
end
|
100
|
+
|
101
|
+
###
|
102
|
+
# Called at the beginning of an element
|
103
|
+
# +name+ is the element name
|
104
|
+
# +attrs+ is a list of attributes
|
105
|
+
# +prefix+ is the namespace prefix for the element
|
106
|
+
# +uri+ is the associated namespace URI
|
107
|
+
# +ns+ is a hash of namespace prefix:urls associated with the element
|
108
|
+
def start_element_namespace name, attrs = [], prefix = nil, uri = nil, ns = []
|
109
|
+
###
|
110
|
+
# Deal with SAX v1 interface
|
111
|
+
name = [prefix, name].compact.join(':')
|
112
|
+
attributes = ns.map { |ns_prefix,ns_uri|
|
113
|
+
[['xmlns', ns_prefix].compact.join(':'), ns_uri]
|
114
|
+
} + attrs.map { |attr|
|
115
|
+
[[attr.prefix, attr.localname].compact.join(':'), attr.value]
|
116
|
+
}
|
117
|
+
start_element name, attributes
|
118
|
+
end
|
119
|
+
|
120
|
+
###
|
121
|
+
# Called at the end of an element
|
122
|
+
# +name+ is the element's name
|
123
|
+
# +prefix+ is the namespace prefix associated with the element
|
124
|
+
# +uri+ is the associated namespace URI
|
125
|
+
def end_element_namespace name, prefix = nil, uri = nil
|
126
|
+
###
|
127
|
+
# Deal with SAX v1 interface
|
128
|
+
end_element [prefix, name].compact.join(':')
|
129
|
+
end
|
130
|
+
|
131
|
+
###
|
132
|
+
# Characters read between a tag. This method might be called multiple
|
133
|
+
# times given one contiguous string of characters.
|
134
|
+
#
|
135
|
+
# +string+ contains the character data
|
136
|
+
def characters string
|
137
|
+
end
|
138
|
+
|
139
|
+
###
|
140
|
+
# Called when comments are encountered
|
141
|
+
# +string+ contains the comment data
|
142
|
+
def comment string
|
143
|
+
end
|
144
|
+
|
145
|
+
###
|
146
|
+
# Called on document warnings
|
147
|
+
# +string+ contains the warning
|
148
|
+
def warning string
|
149
|
+
end
|
150
|
+
|
151
|
+
###
|
152
|
+
# Called on document errors
|
153
|
+
# +string+ contains the error
|
154
|
+
def error string
|
155
|
+
end
|
156
|
+
|
157
|
+
###
|
158
|
+
# Called when cdata blocks are found
|
159
|
+
# +string+ contains the cdata content
|
160
|
+
def cdata_block string
|
161
|
+
end
|
162
|
+
|
163
|
+
###
|
164
|
+
# Called when processing instructions are found
|
165
|
+
# +name+ is the target of the instruction
|
166
|
+
# +content+ is the value of the instruction
|
167
|
+
def processing_instruction name, content
|
168
|
+
end
|
169
|
+
end
|
170
|
+
end
|
171
|
+
end
|
172
|
+
end
|
@@ -0,0 +1,123 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
module Nokogiri
|
3
|
+
module XML
|
4
|
+
module SAX
|
5
|
+
###
|
6
|
+
# This parser is a SAX style parser that reads it's input as it
|
7
|
+
# deems necessary. The parser takes a Nokogiri::XML::SAX::Document,
|
8
|
+
# an optional encoding, then given an XML input, sends messages to
|
9
|
+
# the Nokogiri::XML::SAX::Document.
|
10
|
+
#
|
11
|
+
# Here is an example of using this parser:
|
12
|
+
#
|
13
|
+
# # Create a subclass of Nokogiri::XML::SAX::Document and implement
|
14
|
+
# # the events we care about:
|
15
|
+
# class MyDoc < Nokogiri::XML::SAX::Document
|
16
|
+
# def start_element name, attrs = []
|
17
|
+
# puts "starting: #{name}"
|
18
|
+
# end
|
19
|
+
#
|
20
|
+
# def end_element name
|
21
|
+
# puts "ending: #{name}"
|
22
|
+
# end
|
23
|
+
# end
|
24
|
+
#
|
25
|
+
# # Create our parser
|
26
|
+
# parser = Nokogiri::XML::SAX::Parser.new(MyDoc.new)
|
27
|
+
#
|
28
|
+
# # Send some XML to the parser
|
29
|
+
# parser.parse(File.open(ARGV[0]))
|
30
|
+
#
|
31
|
+
# For more information about SAX parsers, see Nokogiri::XML::SAX. Also
|
32
|
+
# see Nokogiri::XML::SAX::Document for the available events.
|
33
|
+
class Parser
|
34
|
+
class Attribute < Struct.new(:localname, :prefix, :uri, :value)
|
35
|
+
end
|
36
|
+
|
37
|
+
# Encodinds this parser supports
|
38
|
+
ENCODINGS = {
|
39
|
+
'NONE' => 0, # No char encoding detected
|
40
|
+
'UTF-8' => 1, # UTF-8
|
41
|
+
'UTF16LE' => 2, # UTF-16 little endian
|
42
|
+
'UTF16BE' => 3, # UTF-16 big endian
|
43
|
+
'UCS4LE' => 4, # UCS-4 little endian
|
44
|
+
'UCS4BE' => 5, # UCS-4 big endian
|
45
|
+
'EBCDIC' => 6, # EBCDIC uh!
|
46
|
+
'UCS4-2143' => 7, # UCS-4 unusual ordering
|
47
|
+
'UCS4-3412' => 8, # UCS-4 unusual ordering
|
48
|
+
'UCS2' => 9, # UCS-2
|
49
|
+
'ISO-8859-1' => 10, # ISO-8859-1 ISO Latin 1
|
50
|
+
'ISO-8859-2' => 11, # ISO-8859-2 ISO Latin 2
|
51
|
+
'ISO-8859-3' => 12, # ISO-8859-3
|
52
|
+
'ISO-8859-4' => 13, # ISO-8859-4
|
53
|
+
'ISO-8859-5' => 14, # ISO-8859-5
|
54
|
+
'ISO-8859-6' => 15, # ISO-8859-6
|
55
|
+
'ISO-8859-7' => 16, # ISO-8859-7
|
56
|
+
'ISO-8859-8' => 17, # ISO-8859-8
|
57
|
+
'ISO-8859-9' => 18, # ISO-8859-9
|
58
|
+
'ISO-2022-JP' => 19, # ISO-2022-JP
|
59
|
+
'SHIFT-JIS' => 20, # Shift_JIS
|
60
|
+
'EUC-JP' => 21, # EUC-JP
|
61
|
+
'ASCII' => 22, # pure ASCII
|
62
|
+
}
|
63
|
+
|
64
|
+
# The Nokogiri::XML::SAX::Document where events will be sent.
|
65
|
+
attr_accessor :document
|
66
|
+
|
67
|
+
# The encoding beings used for this document.
|
68
|
+
attr_accessor :encoding
|
69
|
+
|
70
|
+
# Create a new Parser with +doc+ and +encoding+
|
71
|
+
def initialize doc = Nokogiri::XML::SAX::Document.new, encoding = 'UTF-8'
|
72
|
+
@encoding = check_encoding(encoding)
|
73
|
+
@document = doc
|
74
|
+
@warned = false
|
75
|
+
end
|
76
|
+
|
77
|
+
###
|
78
|
+
# Parse given +thing+ which may be a string containing xml, or an
|
79
|
+
# IO object.
|
80
|
+
def parse thing, &block
|
81
|
+
if thing.respond_to?(:read) && thing.respond_to?(:close)
|
82
|
+
parse_io(thing, &block)
|
83
|
+
else
|
84
|
+
parse_memory(thing, &block)
|
85
|
+
end
|
86
|
+
end
|
87
|
+
|
88
|
+
###
|
89
|
+
# Parse given +io+
|
90
|
+
def parse_io io, encoding = 'ASCII'
|
91
|
+
@encoding = check_encoding(encoding)
|
92
|
+
ctx = ParserContext.io(io, ENCODINGS[@encoding])
|
93
|
+
yield ctx if block_given?
|
94
|
+
ctx.parse_with self
|
95
|
+
end
|
96
|
+
|
97
|
+
###
|
98
|
+
# Parse a file with +filename+
|
99
|
+
def parse_file filename
|
100
|
+
raise ArgumentError unless filename
|
101
|
+
raise Errno::ENOENT unless File.exist?(filename)
|
102
|
+
raise Errno::EISDIR if File.directory?(filename)
|
103
|
+
ctx = ParserContext.file filename
|
104
|
+
yield ctx if block_given?
|
105
|
+
ctx.parse_with self
|
106
|
+
end
|
107
|
+
|
108
|
+
def parse_memory data
|
109
|
+
ctx = ParserContext.memory data
|
110
|
+
yield ctx if block_given?
|
111
|
+
ctx.parse_with self
|
112
|
+
end
|
113
|
+
|
114
|
+
private
|
115
|
+
def check_encoding(encoding)
|
116
|
+
encoding.upcase.tap do |enc|
|
117
|
+
raise ArgumentError.new("'#{enc}' is not a valid encoding") unless ENCODINGS[enc]
|
118
|
+
end
|
119
|
+
end
|
120
|
+
end
|
121
|
+
end
|
122
|
+
end
|
123
|
+
end
|
@@ -0,0 +1,17 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
module Nokogiri
|
3
|
+
module XML
|
4
|
+
module SAX
|
5
|
+
###
|
6
|
+
# Context for XML SAX parsers. This class is usually not instantiated
|
7
|
+
# by the user. Instead, you should be looking at
|
8
|
+
# Nokogiri::XML::SAX::Parser
|
9
|
+
class ParserContext
|
10
|
+
def self.new thing, encoding = 'UTF-8'
|
11
|
+
[:read, :close].all? { |x| thing.respond_to?(x) } ?
|
12
|
+
io(thing, Parser::ENCODINGS[encoding]) : memory(thing)
|
13
|
+
end
|
14
|
+
end
|
15
|
+
end
|
16
|
+
end
|
17
|
+
end
|
@@ -0,0 +1,61 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
module Nokogiri
|
3
|
+
module XML
|
4
|
+
module SAX
|
5
|
+
###
|
6
|
+
# PushParser can parse a document that is fed to it manually. It
|
7
|
+
# must be given a SAX::Document object which will be called with
|
8
|
+
# SAX events as the document is being parsed.
|
9
|
+
#
|
10
|
+
# Calling PushParser#<< writes XML to the parser, calling any SAX
|
11
|
+
# callbacks it can.
|
12
|
+
#
|
13
|
+
# PushParser#finish tells the parser that the document is finished
|
14
|
+
# and calls the end_document SAX method.
|
15
|
+
#
|
16
|
+
# Example:
|
17
|
+
#
|
18
|
+
# parser = PushParser.new(Class.new(XML::SAX::Document) {
|
19
|
+
# def start_document
|
20
|
+
# puts "start document called"
|
21
|
+
# end
|
22
|
+
# }.new)
|
23
|
+
# parser << "<div>hello<"
|
24
|
+
# parser << "/div>"
|
25
|
+
# parser.finish
|
26
|
+
class PushParser
|
27
|
+
|
28
|
+
# The Nokogiri::XML::SAX::Document on which the PushParser will be
|
29
|
+
# operating
|
30
|
+
attr_accessor :document
|
31
|
+
|
32
|
+
###
|
33
|
+
# Create a new PushParser with +doc+ as the SAX Document, providing
|
34
|
+
# an optional +file_name+ and +encoding+
|
35
|
+
def initialize(doc = XML::SAX::Document.new, file_name = nil, encoding = 'UTF-8')
|
36
|
+
@document = doc
|
37
|
+
@encoding = encoding
|
38
|
+
@sax_parser = XML::SAX::Parser.new(doc)
|
39
|
+
|
40
|
+
## Create our push parser context
|
41
|
+
initialize_native(@sax_parser, file_name)
|
42
|
+
end
|
43
|
+
|
44
|
+
###
|
45
|
+
# Write a +chunk+ of XML to the PushParser. Any callback methods
|
46
|
+
# that can be called will be called immediately.
|
47
|
+
def write chunk, last_chunk = false
|
48
|
+
native_write(chunk, last_chunk)
|
49
|
+
end
|
50
|
+
alias :<< :write
|
51
|
+
|
52
|
+
###
|
53
|
+
# Finish the parsing. This method is only necessary for
|
54
|
+
# Nokogiri::XML::SAX::Document#end_document to be called.
|
55
|
+
def finish
|
56
|
+
write '', true
|
57
|
+
end
|
58
|
+
end
|
59
|
+
end
|
60
|
+
end
|
61
|
+
end
|
@@ -0,0 +1,72 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
module Nokogiri
|
3
|
+
module XML
|
4
|
+
class << self
|
5
|
+
###
|
6
|
+
# Create a new Nokogiri::XML::Schema object using a +string_or_io+
|
7
|
+
# object.
|
8
|
+
def Schema(string_or_io, options = ParseOptions::DEFAULT_SCHEMA)
|
9
|
+
Schema.new(string_or_io, options)
|
10
|
+
end
|
11
|
+
end
|
12
|
+
|
13
|
+
###
|
14
|
+
# Nokogiri::XML::Schema is used for validating XML against a schema
|
15
|
+
# (usually from an xsd file).
|
16
|
+
#
|
17
|
+
# == Synopsis
|
18
|
+
#
|
19
|
+
# Validate an XML document against a Schema. Loop over the errors that
|
20
|
+
# are returned and print them out:
|
21
|
+
#
|
22
|
+
# xsd = Nokogiri::XML::Schema(File.read(PO_SCHEMA_FILE))
|
23
|
+
# doc = Nokogiri::XML(File.read(PO_XML_FILE))
|
24
|
+
#
|
25
|
+
# xsd.validate(doc).each do |error|
|
26
|
+
# puts error.message
|
27
|
+
# end
|
28
|
+
#
|
29
|
+
# The list of errors are Nokogiri::XML::SyntaxError objects.
|
30
|
+
#
|
31
|
+
# NOTE: As of v1.11.0, Schema treats inputs as UNTRUSTED by default, and so external entities
|
32
|
+
# are not resolved from the network (`http://` or `ftp://`). Previously, parsing treated
|
33
|
+
# documents as "trusted" by default which was counter to Nokogiri's "untrusted by default"
|
34
|
+
# security policy. If a document is trusted, then the caller may turn off the NONET option via
|
35
|
+
# the ParseOptions to re-enable external entity resolution over a network connection.
|
36
|
+
class Schema
|
37
|
+
# Errors while parsing the schema file
|
38
|
+
attr_accessor :errors
|
39
|
+
# The Nokogiri::XML::ParseOptions used to parse the schema
|
40
|
+
attr_accessor :parse_options
|
41
|
+
|
42
|
+
###
|
43
|
+
# Create a new Nokogiri::XML::Schema object using a +string_or_io+
|
44
|
+
# object.
|
45
|
+
def self.new string_or_io, options = ParseOptions::DEFAULT_SCHEMA
|
46
|
+
from_document(Nokogiri::XML(string_or_io), options)
|
47
|
+
end
|
48
|
+
|
49
|
+
###
|
50
|
+
# Validate +thing+ against this schema. +thing+ can be a
|
51
|
+
# Nokogiri::XML::Document object, or a filename. An Array of
|
52
|
+
# Nokogiri::XML::SyntaxError objects found while validating the
|
53
|
+
# +thing+ is returned.
|
54
|
+
def validate thing
|
55
|
+
if thing.is_a?(Nokogiri::XML::Document)
|
56
|
+
validate_document(thing)
|
57
|
+
elsif File.file?(thing)
|
58
|
+
validate_file(thing)
|
59
|
+
else
|
60
|
+
raise ArgumentError, "Must provide Nokogiri::Xml::Document or the name of an existing file"
|
61
|
+
end
|
62
|
+
end
|
63
|
+
|
64
|
+
###
|
65
|
+
# Returns true if +thing+ is a valid Nokogiri::XML::Document or
|
66
|
+
# file.
|
67
|
+
def valid? thing
|
68
|
+
validate(thing).length == 0
|
69
|
+
end
|
70
|
+
end
|
71
|
+
end
|
72
|
+
end
|