nokogiri 1.10.9 → 1.13.6
Sign up to get free protection for your applications and to get access to all the features.
Potentially problematic release.
This version of nokogiri might be problematic. Click here for more details.
- checksums.yaml +4 -4
- data/Gemfile +5 -0
- data/LICENSE-DEPENDENCIES.md +1173 -884
- data/LICENSE.md +1 -1
- data/README.md +178 -96
- data/bin/nokogiri +63 -50
- data/dependencies.yml +13 -64
- data/ext/nokogiri/depend +38 -358
- data/ext/nokogiri/extconf.rb +746 -424
- data/ext/nokogiri/gumbo.c +584 -0
- data/ext/nokogiri/html4_document.c +166 -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 +119 -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 +228 -91
- data/ext/nokogiri/nokogiri.h +191 -89
- data/ext/nokogiri/test_global_handlers.c +40 -0
- data/ext/nokogiri/xml_attr.c +15 -15
- data/ext/nokogiri/xml_attribute_decl.c +18 -18
- data/ext/nokogiri/xml_cdata.c +13 -18
- data/ext/nokogiri/xml_comment.c +19 -26
- data/ext/nokogiri/xml_document.c +291 -219
- data/ext/nokogiri/xml_document_fragment.c +12 -16
- data/ext/nokogiri/xml_dtd.c +56 -50
- data/ext/nokogiri/xml_element_content.c +31 -26
- data/ext/nokogiri/xml_element_decl.c +22 -22
- data/ext/nokogiri/xml_encoding_handler.c +43 -18
- data/ext/nokogiri/xml_entity_decl.c +32 -30
- data/ext/nokogiri/xml_entity_reference.c +16 -18
- data/ext/nokogiri/xml_namespace.c +60 -51
- data/ext/nokogiri/xml_node.c +1001 -610
- data/ext/nokogiri/xml_node_set.c +174 -162
- data/ext/nokogiri/xml_processing_instruction.c +17 -19
- data/ext/nokogiri/xml_reader.c +226 -175
- data/ext/nokogiri/xml_relax_ng.c +52 -28
- data/ext/nokogiri/xml_sax_parser.c +112 -112
- data/ext/nokogiri/xml_sax_parser_context.c +112 -86
- data/ext/nokogiri/xml_sax_push_parser.c +36 -27
- data/ext/nokogiri/xml_schema.c +96 -46
- data/ext/nokogiri/xml_syntax_error.c +42 -21
- data/ext/nokogiri/xml_text.c +13 -17
- data/ext/nokogiri/xml_xpath_context.c +223 -115
- data/ext/nokogiri/xslt_stylesheet.c +265 -173
- data/gumbo-parser/CHANGES.md +63 -0
- data/gumbo-parser/Makefile +101 -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/gumbo.h +943 -0
- data/gumbo-parser/src/insertion_mode.h +33 -0
- data/gumbo-parser/src/macros.h +91 -0
- data/gumbo-parser/src/parser.c +4875 -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 +222 -0
- data/gumbo-parser/src/tag_lookup.c +382 -0
- data/gumbo-parser/src/tag_lookup.gperf +169 -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 +68 -0
- data/gumbo-parser/src/util.h +30 -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 +10 -8
- data/lib/nokogiri/css/parser.rb +397 -377
- data/lib/nokogiri/css/parser.y +250 -245
- data/lib/nokogiri/css/parser_extras.rb +54 -49
- data/lib/nokogiri/css/syntax_error.rb +3 -1
- data/lib/nokogiri/css/tokenizer.rb +5 -3
- data/lib/nokogiri/css/tokenizer.rex +3 -2
- data/lib/nokogiri/css/xpath_visitor.rb +218 -91
- data/lib/nokogiri/css.rb +50 -17
- data/lib/nokogiri/decorators/slop.rb +9 -7
- data/lib/nokogiri/extension.rb +31 -0
- data/lib/nokogiri/gumbo.rb +15 -0
- data/lib/nokogiri/html.rb +38 -27
- data/lib/nokogiri/{html → html4}/builder.rb +4 -2
- data/lib/nokogiri/{html → html4}/document.rb +103 -105
- data/lib/nokogiri/html4/document_fragment.rb +54 -0
- data/lib/nokogiri/{html → html4}/element_description.rb +3 -1
- data/lib/nokogiri/html4/element_description_defaults.rb +578 -0
- data/lib/nokogiri/{html → html4}/entity_lookup.rb +4 -2
- data/lib/nokogiri/{html → html4}/sax/parser.rb +17 -16
- data/lib/nokogiri/html4/sax/parser_context.rb +20 -0
- data/lib/nokogiri/{html → html4}/sax/push_parser.rb +12 -11
- data/lib/nokogiri/html4.rb +46 -0
- data/lib/nokogiri/html5/document.rb +91 -0
- data/lib/nokogiri/html5/document_fragment.rb +83 -0
- data/lib/nokogiri/html5/node.rb +100 -0
- data/lib/nokogiri/html5.rb +478 -0
- data/lib/nokogiri/jruby/dependencies.rb +21 -0
- data/lib/nokogiri/syntax_error.rb +2 -0
- data/lib/nokogiri/version/constant.rb +6 -0
- data/lib/nokogiri/version/info.rb +222 -0
- data/lib/nokogiri/version.rb +3 -108
- data/lib/nokogiri/xml/attr.rb +6 -3
- data/lib/nokogiri/xml/attribute_decl.rb +3 -1
- data/lib/nokogiri/xml/builder.rb +74 -33
- data/lib/nokogiri/xml/cdata.rb +3 -1
- data/lib/nokogiri/xml/character_data.rb +2 -0
- data/lib/nokogiri/xml/document.rb +224 -86
- data/lib/nokogiri/xml/document_fragment.rb +46 -44
- data/lib/nokogiri/xml/dtd.rb +4 -2
- data/lib/nokogiri/xml/element_content.rb +2 -0
- data/lib/nokogiri/xml/element_decl.rb +3 -1
- data/lib/nokogiri/xml/entity_decl.rb +4 -2
- data/lib/nokogiri/xml/entity_reference.rb +2 -0
- data/lib/nokogiri/xml/namespace.rb +3 -0
- data/lib/nokogiri/xml/node/save_options.rb +10 -5
- data/lib/nokogiri/xml/node.rb +884 -378
- data/lib/nokogiri/xml/node_set.rb +51 -54
- data/lib/nokogiri/xml/notation.rb +13 -0
- data/lib/nokogiri/xml/parse_options.rb +22 -8
- data/lib/nokogiri/xml/pp/character_data.rb +9 -6
- data/lib/nokogiri/xml/pp/node.rb +25 -26
- data/lib/nokogiri/xml/pp.rb +4 -2
- data/lib/nokogiri/xml/processing_instruction.rb +3 -1
- data/lib/nokogiri/xml/reader.rb +23 -28
- data/lib/nokogiri/xml/relax_ng.rb +8 -2
- data/lib/nokogiri/xml/sax/document.rb +45 -49
- data/lib/nokogiri/xml/sax/parser.rb +38 -34
- data/lib/nokogiri/xml/sax/parser_context.rb +8 -3
- data/lib/nokogiri/xml/sax/push_parser.rb +6 -5
- data/lib/nokogiri/xml/sax.rb +6 -4
- data/lib/nokogiri/xml/schema.rb +19 -9
- data/lib/nokogiri/xml/searchable.rb +112 -72
- data/lib/nokogiri/xml/syntax_error.rb +6 -4
- data/lib/nokogiri/xml/text.rb +2 -0
- data/lib/nokogiri/xml/xpath/syntax_error.rb +4 -2
- data/lib/nokogiri/xml/xpath.rb +15 -4
- data/lib/nokogiri/xml/xpath_context.rb +3 -3
- data/lib/nokogiri/xml.rb +38 -37
- data/lib/nokogiri/xslt/stylesheet.rb +3 -1
- data/lib/nokogiri/xslt.rb +29 -20
- data/lib/nokogiri.rb +49 -65
- data/lib/xsd/xmlparser/nokogiri.rb +26 -24
- data/patches/libxml2/{0002-Remove-script-macro-support.patch → 0001-Remove-script-macro-support.patch} +0 -0
- data/patches/libxml2/{0003-Update-entities-to-remove-handling-of-ssi.patch → 0002-Update-entities-to-remove-handling-of-ssi.patch} +0 -0
- data/patches/libxml2/{0004-libxml2.la-is-in-top_builddir.patch → 0003-libxml2.la-is-in-top_builddir.patch} +1 -1
- data/patches/libxml2/0004-use-glibc-strlen.patch +53 -0
- data/patches/libxml2/0005-avoid-isnan-isinf.patch +81 -0
- data/patches/libxml2/0006-update-automake-files-for-arm64.patch +3040 -0
- data/patches/libxml2/0008-htmlParseComment-handle-abruptly-closed-comments.patch +61 -0
- data/patches/libxml2/0009-allow-wildcard-namespaces.patch +77 -0
- data/patches/libxslt/0001-update-automake-files-for-arm64.patch +3037 -0
- data/ports/archives/libxml2-2.9.14.tar.xz +0 -0
- data/ports/archives/libxslt-1.1.35.tar.xz +0 -0
- metadata +196 -140
- data/ext/nokogiri/html_document.c +0 -170
- data/ext/nokogiri/html_document.h +0 -10
- data/ext/nokogiri/html_element_description.c +0 -279
- data/ext/nokogiri/html_element_description.h +0 -10
- data/ext/nokogiri/html_entity_lookup.c +0 -32
- data/ext/nokogiri/html_entity_lookup.h +0 -8
- data/ext/nokogiri/html_sax_parser_context.c +0 -116
- data/ext/nokogiri/html_sax_parser_context.h +0 -11
- data/ext/nokogiri/html_sax_push_parser.c +0 -87
- data/ext/nokogiri/html_sax_push_parser.h +0 -9
- data/ext/nokogiri/xml_attr.h +0 -9
- data/ext/nokogiri/xml_attribute_decl.h +0 -9
- data/ext/nokogiri/xml_cdata.h +0 -9
- data/ext/nokogiri/xml_comment.h +0 -9
- data/ext/nokogiri/xml_document.h +0 -23
- data/ext/nokogiri/xml_document_fragment.h +0 -10
- data/ext/nokogiri/xml_dtd.h +0 -10
- data/ext/nokogiri/xml_element_content.h +0 -10
- data/ext/nokogiri/xml_element_decl.h +0 -9
- data/ext/nokogiri/xml_encoding_handler.h +0 -8
- data/ext/nokogiri/xml_entity_decl.h +0 -10
- data/ext/nokogiri/xml_entity_reference.h +0 -9
- data/ext/nokogiri/xml_io.c +0 -61
- data/ext/nokogiri/xml_io.h +0 -11
- data/ext/nokogiri/xml_libxml2_hacks.c +0 -112
- data/ext/nokogiri/xml_libxml2_hacks.h +0 -12
- data/ext/nokogiri/xml_namespace.h +0 -14
- data/ext/nokogiri/xml_node.h +0 -13
- data/ext/nokogiri/xml_node_set.h +0 -12
- data/ext/nokogiri/xml_processing_instruction.h +0 -9
- data/ext/nokogiri/xml_reader.h +0 -10
- data/ext/nokogiri/xml_relax_ng.h +0 -9
- data/ext/nokogiri/xml_sax_parser.h +0 -39
- data/ext/nokogiri/xml_sax_parser_context.h +0 -10
- data/ext/nokogiri/xml_sax_push_parser.h +0 -9
- data/ext/nokogiri/xml_schema.h +0 -9
- data/ext/nokogiri/xml_syntax_error.h +0 -13
- data/ext/nokogiri/xml_text.h +0 -9
- data/ext/nokogiri/xml_xpath_context.h +0 -10
- data/ext/nokogiri/xslt_stylesheet.h +0 -14
- data/lib/nokogiri/html/document_fragment.rb +0 -49
- data/lib/nokogiri/html/element_description_defaults.rb +0 -671
- data/lib/nokogiri/html/sax/parser_context.rb +0 -16
- data/patches/libxml2/0001-Revert-Do-not-URI-escape-in-server-side-includes.patch +0 -78
- data/patches/libxml2/0005-Fix-infinite-loop-in-xmlStringLenDecodeEntities.patch +0 -32
- data/ports/archives/libxml2-2.9.10.tar.gz +0 -0
- data/ports/archives/libxslt-1.1.34.tar.gz +0 -0
@@ -0,0 +1,20 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
|
3
|
+
module Nokogiri
|
4
|
+
module HTML4
|
5
|
+
module SAX
|
6
|
+
###
|
7
|
+
# Context for HTML SAX parsers. This class is usually not instantiated by the user. Instead,
|
8
|
+
# you should be looking at Nokogiri::HTML4::SAX::Parser
|
9
|
+
class ParserContext < Nokogiri::XML::SAX::ParserContext
|
10
|
+
def self.new(thing, encoding = "UTF-8")
|
11
|
+
if [:read, :close].all? { |x| thing.respond_to?(x) }
|
12
|
+
super
|
13
|
+
else
|
14
|
+
memory(thing, encoding)
|
15
|
+
end
|
16
|
+
end
|
17
|
+
end
|
18
|
+
end
|
19
|
+
end
|
20
|
+
end
|
@@ -1,34 +1,35 @@
|
|
1
|
+
# frozen_string_literal: true
|
2
|
+
|
1
3
|
module Nokogiri
|
2
|
-
module
|
4
|
+
module HTML4
|
3
5
|
module SAX
|
4
6
|
class PushParser
|
5
|
-
|
6
|
-
# The Nokogiri::HTML::SAX::Document on which the PushParser will be
|
7
|
+
# The Nokogiri::HTML4::SAX::Document on which the PushParser will be
|
7
8
|
# operating
|
8
9
|
attr_accessor :document
|
9
|
-
|
10
|
-
def initialize(doc =
|
10
|
+
|
11
|
+
def initialize(doc = HTML4::SAX::Document.new, file_name = nil, encoding = "UTF-8")
|
11
12
|
@document = doc
|
12
13
|
@encoding = encoding
|
13
|
-
@sax_parser =
|
14
|
+
@sax_parser = HTML4::SAX::Parser.new(doc, @encoding)
|
14
15
|
|
15
16
|
## Create our push parser context
|
16
17
|
initialize_native(@sax_parser, file_name, encoding)
|
17
18
|
end
|
18
|
-
|
19
|
+
|
19
20
|
###
|
20
21
|
# Write a +chunk+ of HTML to the PushParser. Any callback methods
|
21
22
|
# that can be called will be called immediately.
|
22
|
-
def write
|
23
|
+
def write(chunk, last_chunk = false)
|
23
24
|
native_write(chunk, last_chunk)
|
24
25
|
end
|
25
|
-
|
26
|
+
alias_method :<<, :write
|
26
27
|
|
27
28
|
###
|
28
29
|
# Finish the parsing. This method is only necessary for
|
29
|
-
# Nokogiri::
|
30
|
+
# Nokogiri::HTML4::SAX::Document#end_document to be called.
|
30
31
|
def finish
|
31
|
-
write
|
32
|
+
write("", true)
|
32
33
|
end
|
33
34
|
end
|
34
35
|
end
|
@@ -0,0 +1,46 @@
|
|
1
|
+
# coding: utf-8
|
2
|
+
# frozen_string_literal: true
|
3
|
+
|
4
|
+
module Nokogiri
|
5
|
+
class << self
|
6
|
+
# :call-seq:
|
7
|
+
# HTML4(input, url = nil, encoding = nil, options = XML::ParseOptions::DEFAULT_HTML, &block) → Nokogiri::HTML4::Document
|
8
|
+
#
|
9
|
+
# Parse HTML. Convenience method for Nokogiri::HTML4::Document.parse
|
10
|
+
def HTML4(input, url = nil, encoding = nil, options = XML::ParseOptions::DEFAULT_HTML, &block)
|
11
|
+
Nokogiri::HTML4::Document.parse(input, url, encoding, options, &block)
|
12
|
+
end
|
13
|
+
end
|
14
|
+
|
15
|
+
# Since v1.12.0
|
16
|
+
#
|
17
|
+
# 💡 Before v1.12.0, Nokogiri::HTML4 did not exist, and Nokogiri::HTML was the module/namespace
|
18
|
+
# for parsing HTML.
|
19
|
+
module HTML4
|
20
|
+
class << self
|
21
|
+
###
|
22
|
+
# Parse HTML. Convenience method for Nokogiri::HTML4::Document.parse
|
23
|
+
def parse(input, url = nil, encoding = nil, options = XML::ParseOptions::DEFAULT_HTML, &block)
|
24
|
+
Document.parse(input, url, encoding, options, &block)
|
25
|
+
end
|
26
|
+
|
27
|
+
####
|
28
|
+
# Parse a fragment from +string+ in to a NodeSet.
|
29
|
+
def fragment(string, encoding = nil, options = XML::ParseOptions::DEFAULT_HTML, &block)
|
30
|
+
HTML4::DocumentFragment.parse(string, encoding, options, &block)
|
31
|
+
end
|
32
|
+
end
|
33
|
+
|
34
|
+
# Instance of Nokogiri::HTML4::EntityLookup
|
35
|
+
NamedCharacters = EntityLookup.new
|
36
|
+
end
|
37
|
+
end
|
38
|
+
|
39
|
+
require_relative "html4/entity_lookup"
|
40
|
+
require_relative "html4/document"
|
41
|
+
require_relative "html4/document_fragment"
|
42
|
+
require_relative "html4/sax/parser_context"
|
43
|
+
require_relative "html4/sax/parser"
|
44
|
+
require_relative "html4/sax/push_parser"
|
45
|
+
require_relative "html4/element_description"
|
46
|
+
require_relative "html4/element_description_defaults"
|
@@ -0,0 +1,91 @@
|
|
1
|
+
# coding: utf-8
|
2
|
+
# frozen_string_literal: true
|
3
|
+
|
4
|
+
#
|
5
|
+
# Copyright 2013-2021 Sam Ruby, Stephen Checkoway
|
6
|
+
#
|
7
|
+
# Licensed under the Apache License, Version 2.0 (the "License");
|
8
|
+
# you may not use this file except in compliance with the License.
|
9
|
+
# You may obtain a copy of the License at
|
10
|
+
#
|
11
|
+
# http://www.apache.org/licenses/LICENSE-2.0
|
12
|
+
#
|
13
|
+
# Unless required by applicable law or agreed to in writing, software
|
14
|
+
# distributed under the License is distributed on an "AS IS" BASIS,
|
15
|
+
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
16
|
+
# See the License for the specific language governing permissions and
|
17
|
+
# limitations under the License.
|
18
|
+
#
|
19
|
+
|
20
|
+
require_relative "../html4/document"
|
21
|
+
|
22
|
+
module Nokogiri
|
23
|
+
module HTML5
|
24
|
+
# Since v1.12.0
|
25
|
+
#
|
26
|
+
# 💡 HTML5 functionality is not available when running JRuby.
|
27
|
+
class Document < Nokogiri::HTML4::Document
|
28
|
+
def self.parse(string_or_io, url = nil, encoding = nil, **options, &block)
|
29
|
+
yield options if block
|
30
|
+
string_or_io = "" unless string_or_io
|
31
|
+
|
32
|
+
if string_or_io.respond_to?(:encoding) && string_or_io.encoding.name != "ASCII-8BIT"
|
33
|
+
encoding ||= string_or_io.encoding.name
|
34
|
+
end
|
35
|
+
|
36
|
+
if string_or_io.respond_to?(:read) && string_or_io.respond_to?(:path)
|
37
|
+
url ||= string_or_io.path
|
38
|
+
end
|
39
|
+
unless string_or_io.respond_to?(:read) || string_or_io.respond_to?(:to_str)
|
40
|
+
raise ArgumentError, "not a string or IO object"
|
41
|
+
end
|
42
|
+
|
43
|
+
do_parse(string_or_io, url, encoding, options)
|
44
|
+
end
|
45
|
+
|
46
|
+
def self.read_io(io, url = nil, encoding = nil, **options)
|
47
|
+
raise ArgumentError, "io object doesn't respond to :read" unless io.respond_to?(:read)
|
48
|
+
|
49
|
+
do_parse(io, url, encoding, options)
|
50
|
+
end
|
51
|
+
|
52
|
+
def self.read_memory(string, url = nil, encoding = nil, **options)
|
53
|
+
raise ArgumentError, "string object doesn't respond to :to_str" unless string.respond_to?(:to_str)
|
54
|
+
|
55
|
+
do_parse(string, url, encoding, options)
|
56
|
+
end
|
57
|
+
|
58
|
+
def fragment(tags = nil)
|
59
|
+
DocumentFragment.new(self, tags, root)
|
60
|
+
end
|
61
|
+
|
62
|
+
def to_xml(options = {}, &block)
|
63
|
+
# Bypass XML::Document#to_xml which doesn't add
|
64
|
+
# XML::Node::SaveOptions::AS_XML like XML::Node#to_xml does.
|
65
|
+
XML::Node.instance_method(:to_xml).bind(self).call(options, &block)
|
66
|
+
end
|
67
|
+
|
68
|
+
# :call-seq:
|
69
|
+
# xpath_doctype() → Nokogiri::CSS::XPathVisitor::DoctypeConfig
|
70
|
+
#
|
71
|
+
# [Returns] The document type which determines CSS-to-XPath translation.
|
72
|
+
#
|
73
|
+
# See XPathVisitor for more information.
|
74
|
+
def xpath_doctype
|
75
|
+
Nokogiri::CSS::XPathVisitor::DoctypeConfig::HTML5
|
76
|
+
end
|
77
|
+
|
78
|
+
private
|
79
|
+
|
80
|
+
def self.do_parse(string_or_io, url, encoding, options)
|
81
|
+
string = HTML5.read_and_encode(string_or_io, encoding)
|
82
|
+
max_attributes = options[:max_attributes] || Nokogiri::Gumbo::DEFAULT_MAX_ATTRIBUTES
|
83
|
+
max_errors = options[:max_errors] || options[:max_parse_errors] || Nokogiri::Gumbo::DEFAULT_MAX_ERRORS
|
84
|
+
max_depth = options[:max_tree_depth] || Nokogiri::Gumbo::DEFAULT_MAX_TREE_DEPTH
|
85
|
+
doc = Nokogiri::Gumbo.parse(string, url, max_attributes, max_errors, max_depth)
|
86
|
+
doc.encoding = "UTF-8"
|
87
|
+
doc
|
88
|
+
end
|
89
|
+
end
|
90
|
+
end
|
91
|
+
end
|
@@ -0,0 +1,83 @@
|
|
1
|
+
# coding: utf-8
|
2
|
+
# frozen_string_literal: true
|
3
|
+
|
4
|
+
#
|
5
|
+
# Copyright 2013-2021 Sam Ruby, Stephen Checkoway
|
6
|
+
#
|
7
|
+
# Licensed under the Apache License, Version 2.0 (the "License");
|
8
|
+
# you may not use this file except in compliance with the License.
|
9
|
+
# You may obtain a copy of the License at
|
10
|
+
#
|
11
|
+
# http://www.apache.org/licenses/LICENSE-2.0
|
12
|
+
#
|
13
|
+
# Unless required by applicable law or agreed to in writing, software
|
14
|
+
# distributed under the License is distributed on an "AS IS" BASIS,
|
15
|
+
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
16
|
+
# See the License for the specific language governing permissions and
|
17
|
+
# limitations under the License.
|
18
|
+
#
|
19
|
+
|
20
|
+
require_relative "../html4/document_fragment"
|
21
|
+
|
22
|
+
module Nokogiri
|
23
|
+
module HTML5
|
24
|
+
# Since v1.12.0
|
25
|
+
#
|
26
|
+
# 💡 HTML5 functionality is not available when running JRuby.
|
27
|
+
class DocumentFragment < Nokogiri::HTML4::DocumentFragment
|
28
|
+
attr_accessor :document
|
29
|
+
attr_accessor :errors
|
30
|
+
|
31
|
+
# Create a document fragment.
|
32
|
+
def initialize(doc, tags = nil, ctx = nil, options = {})
|
33
|
+
self.document = doc
|
34
|
+
self.errors = []
|
35
|
+
return self unless tags
|
36
|
+
|
37
|
+
max_attributes = options[:max_attributes] || Nokogiri::Gumbo::DEFAULT_MAX_ATTRIBUTES
|
38
|
+
max_errors = options[:max_errors] || Nokogiri::Gumbo::DEFAULT_MAX_ERRORS
|
39
|
+
max_depth = options[:max_tree_depth] || Nokogiri::Gumbo::DEFAULT_MAX_TREE_DEPTH
|
40
|
+
tags = Nokogiri::HTML5.read_and_encode(tags, nil)
|
41
|
+
Nokogiri::Gumbo.fragment(self, tags, ctx, max_attributes, max_errors, max_depth)
|
42
|
+
end
|
43
|
+
|
44
|
+
def serialize(options = {}, &block)
|
45
|
+
# Bypass XML::Document.serialize which doesn't support options even
|
46
|
+
# though XML::Node.serialize does!
|
47
|
+
XML::Node.instance_method(:serialize).bind(self).call(options, &block)
|
48
|
+
end
|
49
|
+
|
50
|
+
# Parse a document fragment from +tags+, returning a Nodeset.
|
51
|
+
def self.parse(tags, encoding = nil, options = {})
|
52
|
+
doc = HTML5::Document.new
|
53
|
+
tags = HTML5.read_and_encode(tags, encoding)
|
54
|
+
doc.encoding = "UTF-8"
|
55
|
+
new(doc, tags, nil, options)
|
56
|
+
end
|
57
|
+
|
58
|
+
def extract_params(params) # :nodoc:
|
59
|
+
handler = params.find do |param|
|
60
|
+
![Hash, String, Symbol].include?(param.class)
|
61
|
+
end
|
62
|
+
params -= [handler] if handler
|
63
|
+
|
64
|
+
hashes = []
|
65
|
+
while Hash === params.last || params.last.nil?
|
66
|
+
hashes << params.pop
|
67
|
+
break if params.empty?
|
68
|
+
end
|
69
|
+
ns, binds = hashes.reverse
|
70
|
+
|
71
|
+
ns ||=
|
72
|
+
begin
|
73
|
+
ns = {}
|
74
|
+
children.each { |child| ns.merge!(child.namespaces) }
|
75
|
+
ns
|
76
|
+
end
|
77
|
+
|
78
|
+
[params, handler, ns, binds]
|
79
|
+
end
|
80
|
+
end
|
81
|
+
end
|
82
|
+
end
|
83
|
+
# vim: set shiftwidth=2 softtabstop=2 tabstop=8 expandtab:
|
@@ -0,0 +1,100 @@
|
|
1
|
+
# coding: utf-8
|
2
|
+
# frozen_string_literal: true
|
3
|
+
|
4
|
+
#
|
5
|
+
# Copyright 2013-2021 Sam Ruby, Stephen Checkoway
|
6
|
+
#
|
7
|
+
# Licensed under the Apache License, Version 2.0 (the "License");
|
8
|
+
# you may not use this file except in compliance with the License.
|
9
|
+
# You may obtain a copy of the License at
|
10
|
+
#
|
11
|
+
# http://www.apache.org/licenses/LICENSE-2.0
|
12
|
+
#
|
13
|
+
# Unless required by applicable law or agreed to in writing, software
|
14
|
+
# distributed under the License is distributed on an "AS IS" BASIS,
|
15
|
+
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
16
|
+
# See the License for the specific language governing permissions and
|
17
|
+
# limitations under the License.
|
18
|
+
#
|
19
|
+
|
20
|
+
require_relative "../xml/node"
|
21
|
+
|
22
|
+
module Nokogiri
|
23
|
+
module HTML5
|
24
|
+
# Since v1.12.0
|
25
|
+
#
|
26
|
+
# 💡 HTML5 functionality is not available when running JRuby.
|
27
|
+
module Node
|
28
|
+
def inner_html(options = {})
|
29
|
+
return super(options) unless document.is_a?(HTML5::Document)
|
30
|
+
|
31
|
+
result = options[:preserve_newline] && HTML5.prepend_newline?(self) ? +"\n" : +""
|
32
|
+
result << children.map { |child| child.to_html(options) }.join
|
33
|
+
result
|
34
|
+
end
|
35
|
+
|
36
|
+
def write_to(io, *options)
|
37
|
+
return super(io, *options) unless document.is_a?(HTML5::Document)
|
38
|
+
|
39
|
+
options = options.first.is_a?(Hash) ? options.shift : {}
|
40
|
+
encoding = options[:encoding] || options[0]
|
41
|
+
if Nokogiri.jruby?
|
42
|
+
save_options = options[:save_with] || options[1]
|
43
|
+
indent_times = options[:indent] || 0
|
44
|
+
else
|
45
|
+
save_options = options[:save_with] || options[1] || XML::Node::SaveOptions::FORMAT
|
46
|
+
indent_times = options[:indent] || 2
|
47
|
+
end
|
48
|
+
indent_string = (options[:indent_text] || " ") * indent_times
|
49
|
+
|
50
|
+
config = XML::Node::SaveOptions.new(save_options.to_i)
|
51
|
+
yield config if block_given?
|
52
|
+
|
53
|
+
config_options = config.options
|
54
|
+
if config_options & (XML::Node::SaveOptions::AS_XML | XML::Node::SaveOptions::AS_XHTML) != 0
|
55
|
+
# Use Nokogiri's serializing code.
|
56
|
+
native_write_to(io, encoding, indent_string, config_options)
|
57
|
+
else
|
58
|
+
# Serialize including the current node.
|
59
|
+
encoding ||= document.encoding || Encoding::UTF_8
|
60
|
+
internal_ops = {
|
61
|
+
preserve_newline: options[:preserve_newline] || false,
|
62
|
+
}
|
63
|
+
HTML5.serialize_node_internal(self, io, encoding, internal_ops)
|
64
|
+
end
|
65
|
+
end
|
66
|
+
|
67
|
+
def fragment(tags)
|
68
|
+
return super(tags) unless document.is_a?(HTML5::Document)
|
69
|
+
|
70
|
+
DocumentFragment.new(document, tags, self)
|
71
|
+
end
|
72
|
+
|
73
|
+
private
|
74
|
+
|
75
|
+
# HTML elements can have attributes that contain colons.
|
76
|
+
# Nokogiri::XML::Node#[]= treats names with colons as a prefixed QName
|
77
|
+
# and tries to create an attribute in a namespace. This is especially
|
78
|
+
# annoying with attribute names like xml:lang since libxml2 will
|
79
|
+
# actually create the xml namespace if it doesn't exist already.
|
80
|
+
def add_child_node_and_reparent_attrs(node)
|
81
|
+
return super(node) unless document.is_a?(HTML5::Document)
|
82
|
+
|
83
|
+
# I'm not sure what this method is supposed to do. Reparenting
|
84
|
+
# namespaces is handled by libxml2, including child namespaces which
|
85
|
+
# this method wouldn't handle.
|
86
|
+
# https://github.com/sparklemotion/nokogiri/issues/1790
|
87
|
+
add_child_node(node)
|
88
|
+
# node.attribute_nodes.find_all { |a| a.namespace }.each do |attr|
|
89
|
+
# attr.remove
|
90
|
+
# ns = attr.namespace
|
91
|
+
# a["#{ns.prefix}:#{attr.name}"] = attr.value
|
92
|
+
# end
|
93
|
+
end
|
94
|
+
end
|
95
|
+
# Monkey patch
|
96
|
+
XML::Node.prepend(HTML5::Node)
|
97
|
+
end
|
98
|
+
end
|
99
|
+
|
100
|
+
# vim: set shiftwidth=2 softtabstop=2 tabstop=8 expandtab:
|