rubyjedi-nokogiri_java 1.4.0.20100513161003-java
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 +330 -0
- data/CHANGELOG.rdoc +341 -0
- data/Manifest.txt +277 -0
- data/README.ja.rdoc +105 -0
- data/README.rdoc +125 -0
- data/Rakefile +307 -0
- data/bin/nokogiri +49 -0
- data/deps.rip +5 -0
- data/ext/nokogiri/extconf.rb +149 -0
- data/ext/nokogiri/html_document.c +145 -0
- data/ext/nokogiri/html_document.h +10 -0
- data/ext/nokogiri/html_element_description.c +272 -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 +92 -0
- data/ext/nokogiri/html_sax_parser_context.h +11 -0
- data/ext/nokogiri/nokogiri.c +96 -0
- data/ext/nokogiri/nokogiri.h +148 -0
- data/ext/nokogiri/xml_attr.c +92 -0
- data/ext/nokogiri/xml_attr.h +9 -0
- data/ext/nokogiri/xml_attribute_decl.c +67 -0
- data/ext/nokogiri/xml_attribute_decl.h +9 -0
- data/ext/nokogiri/xml_cdata.c +54 -0
- data/ext/nokogiri/xml_cdata.h +9 -0
- data/ext/nokogiri/xml_comment.c +52 -0
- data/ext/nokogiri/xml_comment.h +9 -0
- data/ext/nokogiri/xml_document.c +386 -0
- data/ext/nokogiri/xml_document.h +24 -0
- data/ext/nokogiri/xml_document_fragment.c +46 -0
- data/ext/nokogiri/xml_document_fragment.h +10 -0
- data/ext/nokogiri/xml_dtd.c +192 -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 +97 -0
- data/ext/nokogiri/xml_entity_decl.h +10 -0
- data/ext/nokogiri/xml_entity_reference.c +50 -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 +82 -0
- data/ext/nokogiri/xml_namespace.h +13 -0
- data/ext/nokogiri/xml_node.c +1080 -0
- data/ext/nokogiri/xml_node.h +13 -0
- data/ext/nokogiri/xml_node_set.c +405 -0
- data/ext/nokogiri/xml_node_set.h +9 -0
- data/ext/nokogiri/xml_processing_instruction.c +54 -0
- data/ext/nokogiri/xml_processing_instruction.h +9 -0
- data/ext/nokogiri/xml_reader.c +593 -0
- data/ext/nokogiri/xml_reader.h +10 -0
- data/ext/nokogiri/xml_relax_ng.c +159 -0
- data/ext/nokogiri/xml_relax_ng.h +9 -0
- data/ext/nokogiri/xml_sax_parser.c +283 -0
- data/ext/nokogiri/xml_sax_parser.h +43 -0
- data/ext/nokogiri/xml_sax_parser_context.c +157 -0
- data/ext/nokogiri/xml_sax_parser_context.h +10 -0
- data/ext/nokogiri/xml_sax_push_parser.c +114 -0
- data/ext/nokogiri/xml_sax_push_parser.h +9 -0
- data/ext/nokogiri/xml_schema.c +156 -0
- data/ext/nokogiri/xml_schema.h +9 -0
- data/ext/nokogiri/xml_syntax_error.c +52 -0
- data/ext/nokogiri/xml_syntax_error.h +13 -0
- data/ext/nokogiri/xml_text.c +48 -0
- data/ext/nokogiri/xml_text.h +9 -0
- data/ext/nokogiri/xml_xpath.c +53 -0
- data/ext/nokogiri/xml_xpath.h +11 -0
- data/ext/nokogiri/xml_xpath_context.c +239 -0
- data/ext/nokogiri/xml_xpath_context.h +9 -0
- data/ext/nokogiri/xslt_stylesheet.c +131 -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 +123 -0
- data/lib/nokogiri/css.rb +25 -0
- data/lib/nokogiri/css/generated_parser.rb +659 -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 +230 -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 +164 -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 +372 -0
- data/lib/nokogiri/ffi/structs/common_node.rb +26 -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 +108 -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 +37 -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/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 +135 -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 +27 -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 +465 -0
- data/lib/nokogiri/ffi/xml/node_set.rb +146 -0
- data/lib/nokogiri/ffi/xml/processing_instruction.rb +20 -0
- data/lib/nokogiri/ffi/xml/reader.rb +227 -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 +55 -0
- data/lib/nokogiri/ffi/xml/schema.rb +92 -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 +19 -0
- data/lib/nokogiri/ffi/xml/xpath_context.rb +135 -0
- data/lib/nokogiri/ffi/xslt/stylesheet.rb +50 -0
- data/lib/nokogiri/html.rb +36 -0
- data/lib/nokogiri/html/builder.rb +35 -0
- data/lib/nokogiri/html/document.rb +88 -0
- data/lib/nokogiri/html/document_fragment.rb +15 -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 +48 -0
- data/lib/nokogiri/html/sax/parser_context.rb +16 -0
- data/lib/nokogiri/nokogiri.jar +0 -0
- data/lib/nokogiri/syntax_error.rb +4 -0
- data/lib/nokogiri/version.rb +33 -0
- data/lib/nokogiri/version_warning.rb +11 -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 +405 -0
- data/lib/nokogiri/xml/cdata.rb +11 -0
- data/lib/nokogiri/xml/character_data.rb +7 -0
- data/lib/nokogiri/xml/document.rb +163 -0
- data/lib/nokogiri/xml/document_fragment.rb +73 -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 +15 -0
- data/lib/nokogiri/xml/fragment_handler.rb +73 -0
- data/lib/nokogiri/xml/namespace.rb +13 -0
- data/lib/nokogiri/xml/node.rb +730 -0
- data/lib/nokogiri/xml/node/save_options.rb +42 -0
- data/lib/nokogiri/xml/node_set.rb +318 -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 +61 -0
- data/lib/nokogiri/xml/syntax_error.rb +43 -0
- data/lib/nokogiri/xml/xpath.rb +10 -0
- data/lib/nokogiri/xml/xpath/syntax_error.rb +8 -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/xercesImpl.jar +0 -0
- data/lib/xsd/xmlparser/nokogiri.rb +90 -0
- data/tasks/test.rb +100 -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 +76 -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 +137 -0
- data/test/html/sax/test_parser.rb +83 -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 +385 -0
- data/test/html/test_document_encoding.rb +77 -0
- data/test/html/test_document_fragment.rb +157 -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 +242 -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_jruby.rb +40 -0
- data/test/test_memory_leak.rb +87 -0
- data/test/test_nokogiri.rb +140 -0
- data/test/test_reader.rb +358 -0
- data/test/test_soap4r_sax.rb +52 -0
- data/test/test_xslt_transforms.rb +150 -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 +314 -0
- data/test/xml/sax/test_parser_context.rb +63 -0
- data/test/xml/sax/test_push_parser.rb +135 -0
- data/test/xml/test_attr.rb +38 -0
- data/test/xml/test_attribute_decl.rb +90 -0
- data/test/xml/test_builder.rb +167 -0
- data/test/xml/test_cdata.rb +38 -0
- data/test/xml/test_comment.rb +29 -0
- data/test/xml/test_document.rb +638 -0
- data/test/xml/test_document_encoding.rb +26 -0
- data/test/xml/test_document_fragment.rb +149 -0
- data/test/xml/test_dtd.rb +92 -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 +83 -0
- data/test/xml/test_entity_reference.rb +21 -0
- data/test/xml/test_namespace.rb +70 -0
- data/test/xml/test_node.rb +740 -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 +279 -0
- data/test/xml/test_node_set.rb +577 -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 +30 -0
- data/test/xml/test_unparented_node.rb +381 -0
- data/test/xml/test_xpath.rb +169 -0
- metadata +477 -0
@@ -0,0 +1,18 @@
|
|
1
|
+
module Nokogiri
|
2
|
+
module XML
|
3
|
+
module PP
|
4
|
+
module CharacterData
|
5
|
+
def pretty_print pp # :nodoc:
|
6
|
+
nice_name = self.class.name.split('::').last
|
7
|
+
pp.group(2, "#(#{nice_name} ", ')') do
|
8
|
+
pp.pp text
|
9
|
+
end
|
10
|
+
end
|
11
|
+
|
12
|
+
def inspect # :nodoc:
|
13
|
+
"#<#{self.class.name}:#{sprintf("0x%x",object_id)} #{text.inspect}>"
|
14
|
+
end
|
15
|
+
end
|
16
|
+
end
|
17
|
+
end
|
18
|
+
end
|
@@ -0,0 +1,56 @@
|
|
1
|
+
module Nokogiri
|
2
|
+
module XML
|
3
|
+
module PP
|
4
|
+
module Node
|
5
|
+
def inspect # :nodoc:
|
6
|
+
attributes = inspect_attributes.reject { |x|
|
7
|
+
begin
|
8
|
+
attribute = send x
|
9
|
+
!attribute || (attribute.respond_to?(:empty?) && attribute.empty?)
|
10
|
+
rescue NoMethodError
|
11
|
+
true
|
12
|
+
end
|
13
|
+
}.map { |attribute|
|
14
|
+
"#{attribute.to_s.sub(/_\w+/, 's')}=#{send(attribute).inspect}"
|
15
|
+
}.join ' '
|
16
|
+
"#<#{self.class.name}:#{sprintf("0x%x", object_id)} #{attributes}>"
|
17
|
+
end
|
18
|
+
|
19
|
+
def pretty_print pp # :nodoc:
|
20
|
+
nice_name = self.class.name.split('::').last
|
21
|
+
pp.group(2, "#(#{nice_name}:#{sprintf("0x%x", object_id)} {", '})') do
|
22
|
+
|
23
|
+
pp.breakable
|
24
|
+
attrs = inspect_attributes.map { |t|
|
25
|
+
[t, send(t)] if respond_to?(t)
|
26
|
+
}.compact.find_all { |x|
|
27
|
+
if x.last
|
28
|
+
if [:attribute_nodes, :children].include? x.first
|
29
|
+
!x.last.empty?
|
30
|
+
else
|
31
|
+
true
|
32
|
+
end
|
33
|
+
end
|
34
|
+
}
|
35
|
+
|
36
|
+
pp.seplist(attrs) do |v|
|
37
|
+
if [:attribute_nodes, :children].include? v.first
|
38
|
+
pp.group(2, "#{v.first.to_s.sub(/_\w+$/, 's')} = [", "]") do
|
39
|
+
pp.breakable
|
40
|
+
pp.seplist(v.last) do |item|
|
41
|
+
pp.pp item
|
42
|
+
end
|
43
|
+
end
|
44
|
+
else
|
45
|
+
pp.text "#{v.first} = "
|
46
|
+
pp.pp v.last
|
47
|
+
end
|
48
|
+
end
|
49
|
+
pp.breakable
|
50
|
+
|
51
|
+
end
|
52
|
+
end
|
53
|
+
end
|
54
|
+
end
|
55
|
+
end
|
56
|
+
end
|
@@ -0,0 +1,74 @@
|
|
1
|
+
module Nokogiri
|
2
|
+
module XML
|
3
|
+
###
|
4
|
+
# Nokogiri::XML::Reader parses an XML document similar to the way a cursor
|
5
|
+
# would move. The Reader is given an XML document, and yields nodes
|
6
|
+
# to an each block.
|
7
|
+
#
|
8
|
+
# Here is an example of usage:
|
9
|
+
#
|
10
|
+
# reader = Nokogiri::XML::Reader(<<-eoxml)
|
11
|
+
# <x xmlns:tenderlove='http://tenderlovemaking.com/'>
|
12
|
+
# <tenderlove:foo awesome='true'>snuggles!</tenderlove:foo>
|
13
|
+
# </x>
|
14
|
+
# eoxml
|
15
|
+
#
|
16
|
+
# reader.each do |node|
|
17
|
+
#
|
18
|
+
# # node is an instance of Nokogiri::XML::Reader
|
19
|
+
# puts node.name
|
20
|
+
#
|
21
|
+
# end
|
22
|
+
#
|
23
|
+
# Note that Nokogiri::XML::Reader#each can only be called once!! Once
|
24
|
+
# the cursor moves through the entire document, you must parse the
|
25
|
+
# document again. So make sure that you capture any information you
|
26
|
+
# need during the first iteration.
|
27
|
+
#
|
28
|
+
# The Reader parser is good for when you need the speed of a SAX parser,
|
29
|
+
# but do not want to write a Document handler.
|
30
|
+
class Reader
|
31
|
+
include Enumerable
|
32
|
+
|
33
|
+
# A list of errors encountered while parsing
|
34
|
+
attr_accessor :errors
|
35
|
+
|
36
|
+
# The encoding for the document
|
37
|
+
attr_reader :encoding
|
38
|
+
|
39
|
+
# The XML source
|
40
|
+
attr_reader :source
|
41
|
+
|
42
|
+
def initialize source, url = nil, encoding = nil # :nodoc:
|
43
|
+
@source = source
|
44
|
+
@errors = []
|
45
|
+
@encoding = encoding
|
46
|
+
end
|
47
|
+
private :initialize
|
48
|
+
|
49
|
+
###
|
50
|
+
# Get a list of attributes for the current node.
|
51
|
+
def attributes
|
52
|
+
Hash[*(attribute_nodes.map { |node|
|
53
|
+
[node.name, node.to_s]
|
54
|
+
}.flatten)].merge(namespaces || {})
|
55
|
+
end
|
56
|
+
|
57
|
+
###
|
58
|
+
# Get a list of attributes for the current node
|
59
|
+
def attribute_nodes
|
60
|
+
nodes = attr_nodes
|
61
|
+
nodes.each { |v| v.instance_variable_set(:@_r, self) }
|
62
|
+
nodes
|
63
|
+
end
|
64
|
+
|
65
|
+
###
|
66
|
+
# Move the cursor through the document yielding each node to the block
|
67
|
+
def each(&block)
|
68
|
+
while node = self.read
|
69
|
+
block.call(node)
|
70
|
+
end
|
71
|
+
end
|
72
|
+
end
|
73
|
+
end
|
74
|
+
end
|
@@ -0,0 +1,32 @@
|
|
1
|
+
module Nokogiri
|
2
|
+
module XML
|
3
|
+
class << self
|
4
|
+
###
|
5
|
+
# Create a new Nokogiri::XML::RelaxNG document from +string_or_io+.
|
6
|
+
# See Nokogiri::XML::RelaxNG for an example.
|
7
|
+
def RelaxNG string_or_io
|
8
|
+
RelaxNG.new(string_or_io)
|
9
|
+
end
|
10
|
+
end
|
11
|
+
|
12
|
+
###
|
13
|
+
# Nokogiri::XML::RelaxNG is used for validating XML against a
|
14
|
+
# RelaxNG schema.
|
15
|
+
#
|
16
|
+
# == Synopsis
|
17
|
+
#
|
18
|
+
# Validate an XML document against a RelaxNG schema. Loop over the errors
|
19
|
+
# that are returned and print them out:
|
20
|
+
#
|
21
|
+
# schema = Nokogiri::XML::RelaxNG(File.open(ADDRESS_SCHEMA_FILE))
|
22
|
+
# doc = Nokogiri::XML(File.open(ADDRESS_XML_FILE))
|
23
|
+
#
|
24
|
+
# schema.validate(doc).each do |error|
|
25
|
+
# puts error.message
|
26
|
+
# end
|
27
|
+
#
|
28
|
+
# The list of errors are Nokogiri::XML::SyntaxError objects.
|
29
|
+
class RelaxNG < Nokogiri::XML::Schema
|
30
|
+
end
|
31
|
+
end
|
32
|
+
end
|
@@ -0,0 +1,160 @@
|
|
1
|
+
module Nokogiri
|
2
|
+
module XML
|
3
|
+
###
|
4
|
+
# SAX Parsers are event driven parsers. Nokogiri provides two different
|
5
|
+
# event based parsers when dealing with XML. If you want to do SAX style
|
6
|
+
# parsing using HTML, check out Nokogiri::HTML::SAX.
|
7
|
+
#
|
8
|
+
# The basic way a SAX style parser works is by creating a parser,
|
9
|
+
# telling the parser about the events we're interested in, then giving
|
10
|
+
# the parser some XML to process. The parser will notify you when
|
11
|
+
# it encounters events your said you would like to know about.
|
12
|
+
#
|
13
|
+
# To register for events, you simply subclass Nokogiri::XML::SAX::Document,
|
14
|
+
# and implement the methods for which you would like notification.
|
15
|
+
#
|
16
|
+
# For example, if I want to be notified when a document ends, and when an
|
17
|
+
# element starts, I would write a class like this:
|
18
|
+
#
|
19
|
+
# class MyDocument < Nokogiri::XML::SAX::Document
|
20
|
+
# def end_document
|
21
|
+
# puts "the document has ended"
|
22
|
+
# end
|
23
|
+
#
|
24
|
+
# def start_element name, attributes = []
|
25
|
+
# puts "#{name} started"
|
26
|
+
# end
|
27
|
+
# end
|
28
|
+
#
|
29
|
+
# Then I would instantiate a SAX parser with this document, and feed the
|
30
|
+
# parser some XML
|
31
|
+
#
|
32
|
+
# # Create a new parser
|
33
|
+
# parser = Nokogiri::XML::SAX::Parser.new(MyDocument.new)
|
34
|
+
#
|
35
|
+
# # Feed the parser some XML
|
36
|
+
# parser.parse(File.read(ARGV[0], 'rb'))
|
37
|
+
#
|
38
|
+
# Now my document handler will be called when each node starts, and when
|
39
|
+
# then document ends. To see what kinds of events are available, take
|
40
|
+
# a look at Nokogiri::XML::SAX::Document.
|
41
|
+
#
|
42
|
+
# Two SAX parsers for XML are available, a parser that reads from a string
|
43
|
+
# or IO object as it feels necessary, and a parser that lets you spoon
|
44
|
+
# feed it XML. If you want to let Nokogiri deal with reading your XML,
|
45
|
+
# use the Nokogiri::XML::SAX::Parser. If you want to have fine grain
|
46
|
+
# control over the XML input, use the Nokogiri::XML::SAX::PushParser.
|
47
|
+
module SAX
|
48
|
+
###
|
49
|
+
# This class is used for registering types of events you are interested
|
50
|
+
# in handling. All of the methods on this class are available as
|
51
|
+
# possible events while parsing an XML document. To register for any
|
52
|
+
# particular event, just subclass this class and implement the methods
|
53
|
+
# you are interested in knowing about.
|
54
|
+
#
|
55
|
+
# To only be notified about start and end element events, write a class
|
56
|
+
# like this:
|
57
|
+
#
|
58
|
+
# class MyDocument < Nokogiri::XML::SAX::Document
|
59
|
+
# def start_element name, attrs = []
|
60
|
+
# puts "#{name} started!"
|
61
|
+
# end
|
62
|
+
#
|
63
|
+
# def end_element name
|
64
|
+
# puts "#{name} ended"
|
65
|
+
# end
|
66
|
+
# end
|
67
|
+
#
|
68
|
+
# You can use this event handler for any SAX style parser included with
|
69
|
+
# Nokogiri. See Nokogiri::XML::SAX, and Nokogiri::HTML::SAX.
|
70
|
+
class Document
|
71
|
+
###
|
72
|
+
# Called when an XML declaration is parsed
|
73
|
+
def xmldecl version, encoding, standalone
|
74
|
+
end
|
75
|
+
|
76
|
+
###
|
77
|
+
# Called when document starts parsing
|
78
|
+
def start_document
|
79
|
+
end
|
80
|
+
|
81
|
+
###
|
82
|
+
# Called when document ends parsing
|
83
|
+
def end_document
|
84
|
+
end
|
85
|
+
|
86
|
+
###
|
87
|
+
# Called at the beginning of an element
|
88
|
+
# +name+ is the name of the tag with +attrs+ as attributes
|
89
|
+
def start_element name, attrs = []
|
90
|
+
end
|
91
|
+
|
92
|
+
###
|
93
|
+
# Called at the end of an element
|
94
|
+
# +name+ is the tag name
|
95
|
+
def end_element name
|
96
|
+
end
|
97
|
+
|
98
|
+
###
|
99
|
+
# Called at the beginning of an element
|
100
|
+
# +name+ is the element name
|
101
|
+
# +attrs+ is a list of attributes
|
102
|
+
# +prefix+ is the namespace prefix for the element
|
103
|
+
# +uri+ is the associated namespace URI
|
104
|
+
# +ns+ is a hash of namespace prefix:urls associated with the element
|
105
|
+
def start_element_namespace name, attrs = [], prefix = nil, uri = nil, ns = []
|
106
|
+
###
|
107
|
+
# Deal with SAX v1 interface
|
108
|
+
name = [prefix, name].compact.join(':')
|
109
|
+
attributes = ns.map { |ns_prefix,ns_uri|
|
110
|
+
[['xmlns', ns_prefix].compact.join(':'), ns_uri]
|
111
|
+
} + attrs.map { |attr|
|
112
|
+
[[attr.prefix, attr.localname].compact.join(':'), attr.value]
|
113
|
+
}.flatten
|
114
|
+
start_element name, attributes
|
115
|
+
end
|
116
|
+
|
117
|
+
###
|
118
|
+
# Called at the end of an element
|
119
|
+
# +name+ is the element's name
|
120
|
+
# +prefix+ is the namespace prefix associated with the element
|
121
|
+
# +uri+ is the associated namespace URI
|
122
|
+
def end_element_namespace name, prefix = nil, uri = nil
|
123
|
+
###
|
124
|
+
# Deal with SAX v1 interface
|
125
|
+
end_element [prefix, name].compact.join(':')
|
126
|
+
end
|
127
|
+
|
128
|
+
###
|
129
|
+
# Characters read between a tag
|
130
|
+
# +string+ contains the character data
|
131
|
+
def characters string
|
132
|
+
end
|
133
|
+
|
134
|
+
###
|
135
|
+
# Called when comments are encountered
|
136
|
+
# +string+ contains the comment data
|
137
|
+
def comment string
|
138
|
+
end
|
139
|
+
|
140
|
+
###
|
141
|
+
# Called on document warnings
|
142
|
+
# +string+ contains the warning
|
143
|
+
def warning string
|
144
|
+
end
|
145
|
+
|
146
|
+
###
|
147
|
+
# Called on document errors
|
148
|
+
# +string+ contains the error
|
149
|
+
def error string
|
150
|
+
end
|
151
|
+
|
152
|
+
###
|
153
|
+
# Called when cdata blocks are found
|
154
|
+
# +string+ contains the cdata content
|
155
|
+
def cdata_block string
|
156
|
+
end
|
157
|
+
end
|
158
|
+
end
|
159
|
+
end
|
160
|
+
end
|
@@ -0,0 +1,115 @@
|
|
1
|
+
module Nokogiri
|
2
|
+
module XML
|
3
|
+
module SAX
|
4
|
+
###
|
5
|
+
# This parser is a SAX style parser that reads it's input as it
|
6
|
+
# deems necessary. The parser takes a Nokogiri::XML::SAX::Document,
|
7
|
+
# an optional encoding, then given an XML input, sends messages to
|
8
|
+
# the Nokogiri::XML::SAX::Document.
|
9
|
+
#
|
10
|
+
# Here is an example of using this parser:
|
11
|
+
#
|
12
|
+
# # Create a subclass of Nokogiri::XML::SAX::Document and implement
|
13
|
+
# # the events we care about:
|
14
|
+
# class MyDoc < Nokogiri::XML::SAX::Document
|
15
|
+
# def start_element name, attrs = []
|
16
|
+
# puts "starting: #{name}"
|
17
|
+
# end
|
18
|
+
#
|
19
|
+
# def end_element name
|
20
|
+
# puts "ending: #{name}"
|
21
|
+
# end
|
22
|
+
# end
|
23
|
+
#
|
24
|
+
# # Create our parser
|
25
|
+
# parser = Nokogiri::XML::SAX::Parser.new(MyDoc.new)
|
26
|
+
#
|
27
|
+
# # Send some XML to the parser
|
28
|
+
# parser.parse(File.read(ARGV[0]))
|
29
|
+
#
|
30
|
+
# For more information about SAX parsers, see Nokogiri::XML::SAX. Also
|
31
|
+
# see Nokogiri::XML::SAX::Document for the available events.
|
32
|
+
class Parser
|
33
|
+
class Attribute < Struct.new(:localname, :prefix, :uri, :value)
|
34
|
+
end
|
35
|
+
|
36
|
+
# Encodinds this parser supports
|
37
|
+
ENCODINGS = {
|
38
|
+
'NONE' => 0, # No char encoding detected
|
39
|
+
'UTF-8' => 1, # UTF-8
|
40
|
+
'UTF16LE' => 2, # UTF-16 little endian
|
41
|
+
'UTF16BE' => 3, # UTF-16 big endian
|
42
|
+
'UCS4LE' => 4, # UCS-4 little endian
|
43
|
+
'UCS4BE' => 5, # UCS-4 big endian
|
44
|
+
'EBCDIC' => 6, # EBCDIC uh!
|
45
|
+
'UCS4-2143' => 7, # UCS-4 unusual ordering
|
46
|
+
'UCS4-3412' => 8, # UCS-4 unusual ordering
|
47
|
+
'UCS2' => 9, # UCS-2
|
48
|
+
'ISO-8859-1' => 10, # ISO-8859-1 ISO Latin 1
|
49
|
+
'ISO-8859-2' => 11, # ISO-8859-2 ISO Latin 2
|
50
|
+
'ISO-8859-3' => 12, # ISO-8859-3
|
51
|
+
'ISO-8859-4' => 13, # ISO-8859-4
|
52
|
+
'ISO-8859-5' => 14, # ISO-8859-5
|
53
|
+
'ISO-8859-6' => 15, # ISO-8859-6
|
54
|
+
'ISO-8859-7' => 16, # ISO-8859-7
|
55
|
+
'ISO-8859-8' => 17, # ISO-8859-8
|
56
|
+
'ISO-8859-9' => 18, # ISO-8859-9
|
57
|
+
'ISO-2022-JP' => 19, # ISO-2022-JP
|
58
|
+
'SHIFT-JIS' => 20, # Shift_JIS
|
59
|
+
'EUC-JP' => 21, # EUC-JP
|
60
|
+
'ASCII' => 22, # pure ASCII
|
61
|
+
}
|
62
|
+
|
63
|
+
# The Nokogiri::XML::SAX::Document where events will be sent.
|
64
|
+
attr_accessor :document
|
65
|
+
|
66
|
+
# The encoding beings used for this document.
|
67
|
+
attr_accessor :encoding
|
68
|
+
|
69
|
+
# Create a new Parser with +doc+ and +encoding+
|
70
|
+
def initialize doc = Nokogiri::XML::SAX::Document.new, encoding = 'UTF-8'
|
71
|
+
@encoding = encoding
|
72
|
+
@document = doc
|
73
|
+
@warned = false
|
74
|
+
end
|
75
|
+
|
76
|
+
###
|
77
|
+
# Parse given +thing+ which may be a string containing xml, or an
|
78
|
+
# IO object.
|
79
|
+
def parse thing, &block
|
80
|
+
if thing.respond_to?(:read) && thing.respond_to?(:close)
|
81
|
+
parse_io(thing, &block)
|
82
|
+
else
|
83
|
+
parse_memory(thing, &block)
|
84
|
+
end
|
85
|
+
end
|
86
|
+
|
87
|
+
###
|
88
|
+
# Parse given +io+
|
89
|
+
def parse_io io, encoding = 'ASCII'
|
90
|
+
@encoding = encoding
|
91
|
+
ctx = ParserContext.io(io, ENCODINGS[encoding])
|
92
|
+
yield ctx if block_given?
|
93
|
+
ctx.parse_with self
|
94
|
+
end
|
95
|
+
|
96
|
+
###
|
97
|
+
# Parse a file with +filename+
|
98
|
+
def parse_file filename
|
99
|
+
raise ArgumentError unless filename
|
100
|
+
raise Errno::ENOENT unless File.exists?(filename)
|
101
|
+
raise Errno::EISDIR if File.directory?(filename)
|
102
|
+
ctx = ParserContext.file filename
|
103
|
+
yield ctx if block_given?
|
104
|
+
ctx.parse_with self
|
105
|
+
end
|
106
|
+
|
107
|
+
def parse_memory data
|
108
|
+
ctx = ParserContext.memory data
|
109
|
+
yield ctx if block_given?
|
110
|
+
ctx.parse_with self
|
111
|
+
end
|
112
|
+
end
|
113
|
+
end
|
114
|
+
end
|
115
|
+
end
|