nokogiri 1.12.2-x86_64-linux → 1.13.0-x86_64-linux
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 +2 -0
- data/README.md +9 -7
- data/bin/nokogiri +63 -50
- data/dependencies.yml +5 -6
- data/ext/nokogiri/extconf.rb +51 -35
- data/ext/nokogiri/gumbo.c +11 -11
- data/ext/nokogiri/html4_element_description.c +1 -1
- data/ext/nokogiri/html4_sax_parser_context.c +2 -1
- data/ext/nokogiri/include/libxml2/libxml/xmlerror.h +1 -0
- data/ext/nokogiri/include/libxslt/xsltconfig.h +1 -1
- data/ext/nokogiri/nokogiri.c +1 -1
- data/ext/nokogiri/nokogiri.h +3 -0
- data/ext/nokogiri/xml_document.c +36 -36
- data/ext/nokogiri/xml_document_fragment.c +0 -2
- data/ext/nokogiri/xml_dtd.c +2 -2
- data/ext/nokogiri/xml_encoding_handler.c +25 -11
- data/ext/nokogiri/xml_namespace.c +2 -2
- data/ext/nokogiri/xml_node.c +647 -335
- data/ext/nokogiri/xml_reader.c +37 -11
- data/ext/nokogiri/xml_xpath_context.c +72 -49
- data/lib/nokogiri/2.6/nokogiri.so +0 -0
- data/lib/nokogiri/2.7/nokogiri.so +0 -0
- data/lib/nokogiri/3.0/nokogiri.so +0 -0
- data/lib/nokogiri/3.1/nokogiri.so +0 -0
- data/lib/nokogiri/class_resolver.rb +67 -0
- data/lib/nokogiri/css/node.rb +9 -8
- data/lib/nokogiri/css/parser.rb +11 -3
- data/lib/nokogiri/css/parser.y +10 -2
- data/lib/nokogiri/css/parser_extras.rb +20 -20
- data/lib/nokogiri/css/syntax_error.rb +1 -0
- data/lib/nokogiri/css/tokenizer.rb +2 -1
- data/lib/nokogiri/css/tokenizer.rex +2 -1
- data/lib/nokogiri/css/xpath_visitor.rb +174 -75
- data/lib/nokogiri/css.rb +38 -6
- data/lib/nokogiri/decorators/slop.rb +8 -7
- data/lib/nokogiri/extension.rb +1 -1
- data/lib/nokogiri/gumbo.rb +1 -0
- data/lib/nokogiri/html.rb +16 -10
- data/lib/nokogiri/html4/builder.rb +1 -0
- data/lib/nokogiri/html4/document.rb +84 -75
- data/lib/nokogiri/html4/document_fragment.rb +11 -7
- data/lib/nokogiri/html4/element_description.rb +1 -0
- data/lib/nokogiri/html4/element_description_defaults.rb +426 -520
- data/lib/nokogiri/html4/entity_lookup.rb +2 -1
- data/lib/nokogiri/html4/sax/parser.rb +2 -1
- data/lib/nokogiri/html4/sax/parser_context.rb +1 -0
- data/lib/nokogiri/html4/sax/push_parser.rb +7 -7
- data/lib/nokogiri/html4.rb +11 -5
- data/lib/nokogiri/html5/document.rb +24 -10
- data/lib/nokogiri/html5/document_fragment.rb +5 -2
- data/lib/nokogiri/html5/node.rb +6 -3
- data/lib/nokogiri/html5.rb +68 -64
- data/lib/nokogiri/jruby/dependencies.rb +10 -9
- data/lib/nokogiri/syntax_error.rb +1 -0
- data/lib/nokogiri/version/constant.rb +2 -1
- data/lib/nokogiri/version/info.rb +19 -13
- data/lib/nokogiri/version.rb +1 -0
- data/lib/nokogiri/xml/attr.rb +5 -3
- data/lib/nokogiri/xml/attribute_decl.rb +2 -1
- data/lib/nokogiri/xml/builder.rb +69 -31
- data/lib/nokogiri/xml/cdata.rb +2 -1
- data/lib/nokogiri/xml/character_data.rb +1 -0
- data/lib/nokogiri/xml/document.rb +178 -96
- data/lib/nokogiri/xml/document_fragment.rb +41 -38
- data/lib/nokogiri/xml/dtd.rb +3 -2
- data/lib/nokogiri/xml/element_content.rb +1 -0
- data/lib/nokogiri/xml/element_decl.rb +2 -1
- data/lib/nokogiri/xml/entity_decl.rb +3 -2
- data/lib/nokogiri/xml/entity_reference.rb +1 -0
- data/lib/nokogiri/xml/namespace.rb +2 -0
- data/lib/nokogiri/xml/node/save_options.rb +7 -4
- data/lib/nokogiri/xml/node.rb +512 -348
- data/lib/nokogiri/xml/node_set.rb +46 -54
- data/lib/nokogiri/xml/notation.rb +12 -0
- data/lib/nokogiri/xml/parse_options.rb +11 -7
- data/lib/nokogiri/xml/pp/character_data.rb +8 -6
- data/lib/nokogiri/xml/pp/node.rb +24 -26
- data/lib/nokogiri/xml/pp.rb +1 -0
- data/lib/nokogiri/xml/processing_instruction.rb +2 -1
- data/lib/nokogiri/xml/reader.rb +17 -19
- data/lib/nokogiri/xml/relax_ng.rb +1 -0
- data/lib/nokogiri/xml/sax/document.rb +20 -19
- data/lib/nokogiri/xml/sax/parser.rb +36 -34
- data/lib/nokogiri/xml/sax/parser_context.rb +7 -3
- data/lib/nokogiri/xml/sax/push_parser.rb +5 -5
- data/lib/nokogiri/xml/sax.rb +1 -0
- data/lib/nokogiri/xml/schema.rb +7 -6
- data/lib/nokogiri/xml/searchable.rb +42 -22
- data/lib/nokogiri/xml/syntax_error.rb +4 -4
- data/lib/nokogiri/xml/text.rb +1 -0
- data/lib/nokogiri/xml/xpath/syntax_error.rb +2 -1
- data/lib/nokogiri/xml/xpath.rb +12 -0
- data/lib/nokogiri/xml/xpath_context.rb +2 -3
- data/lib/nokogiri/xml.rb +3 -3
- data/lib/nokogiri/xslt/stylesheet.rb +1 -0
- data/lib/nokogiri/xslt.rb +3 -2
- data/lib/nokogiri.rb +19 -16
- data/lib/xsd/xmlparser/nokogiri.rb +25 -24
- metadata +124 -52
- data/lib/nokogiri/2.5/nokogiri.so +0 -0
@@ -1,4 +1,5 @@
|
|
1
1
|
# frozen_string_literal: true
|
2
|
+
|
2
3
|
module Nokogiri
|
3
4
|
module XML
|
4
5
|
module SAX
|
@@ -36,29 +37,29 @@ module Nokogiri
|
|
36
37
|
|
37
38
|
# Encodinds this parser supports
|
38
39
|
ENCODINGS = {
|
39
|
-
|
40
|
-
|
41
|
-
|
42
|
-
|
43
|
-
|
44
|
-
|
45
|
-
|
46
|
-
|
47
|
-
|
48
|
-
|
49
|
-
|
50
|
-
|
51
|
-
|
52
|
-
|
53
|
-
|
54
|
-
|
55
|
-
|
56
|
-
|
57
|
-
|
58
|
-
|
59
|
-
|
60
|
-
|
61
|
-
|
40
|
+
"NONE" => 0, # No char encoding detected
|
41
|
+
"UTF-8" => 1, # UTF-8
|
42
|
+
"UTF16LE" => 2, # UTF-16 little endian
|
43
|
+
"UTF16BE" => 3, # UTF-16 big endian
|
44
|
+
"UCS4LE" => 4, # UCS-4 little endian
|
45
|
+
"UCS4BE" => 5, # UCS-4 big endian
|
46
|
+
"EBCDIC" => 6, # EBCDIC uh!
|
47
|
+
"UCS4-2143" => 7, # UCS-4 unusual ordering
|
48
|
+
"UCS4-3412" => 8, # UCS-4 unusual ordering
|
49
|
+
"UCS2" => 9, # UCS-2
|
50
|
+
"ISO-8859-1" => 10, # ISO-8859-1 ISO Latin 1
|
51
|
+
"ISO-8859-2" => 11, # ISO-8859-2 ISO Latin 2
|
52
|
+
"ISO-8859-3" => 12, # ISO-8859-3
|
53
|
+
"ISO-8859-4" => 13, # ISO-8859-4
|
54
|
+
"ISO-8859-5" => 14, # ISO-8859-5
|
55
|
+
"ISO-8859-6" => 15, # ISO-8859-6
|
56
|
+
"ISO-8859-7" => 16, # ISO-8859-7
|
57
|
+
"ISO-8859-8" => 17, # ISO-8859-8
|
58
|
+
"ISO-8859-9" => 18, # ISO-8859-9
|
59
|
+
"ISO-2022-JP" => 19, # ISO-2022-JP
|
60
|
+
"SHIFT-JIS" => 20, # Shift_JIS
|
61
|
+
"EUC-JP" => 21, # EUC-JP
|
62
|
+
"ASCII" => 22, # pure ASCII
|
62
63
|
}
|
63
64
|
|
64
65
|
# The Nokogiri::XML::SAX::Document where events will be sent.
|
@@ -68,7 +69,7 @@ module Nokogiri
|
|
68
69
|
attr_accessor :encoding
|
69
70
|
|
70
71
|
# Create a new Parser with +doc+ and +encoding+
|
71
|
-
def initialize
|
72
|
+
def initialize(doc = Nokogiri::XML::SAX::Document.new, encoding = "UTF-8")
|
72
73
|
@encoding = check_encoding(encoding)
|
73
74
|
@document = doc
|
74
75
|
@warned = false
|
@@ -77,7 +78,7 @@ module Nokogiri
|
|
77
78
|
###
|
78
79
|
# Parse given +thing+ which may be a string containing xml, or an
|
79
80
|
# IO object.
|
80
|
-
def parse
|
81
|
+
def parse(thing, &block)
|
81
82
|
if thing.respond_to?(:read) && thing.respond_to?(:close)
|
82
83
|
parse_io(thing, &block)
|
83
84
|
else
|
@@ -87,34 +88,35 @@ module Nokogiri
|
|
87
88
|
|
88
89
|
###
|
89
90
|
# Parse given +io+
|
90
|
-
def parse_io
|
91
|
+
def parse_io(io, encoding = "ASCII")
|
91
92
|
@encoding = check_encoding(encoding)
|
92
93
|
ctx = ParserContext.io(io, ENCODINGS[@encoding])
|
93
94
|
yield ctx if block_given?
|
94
|
-
ctx.parse_with
|
95
|
+
ctx.parse_with(self)
|
95
96
|
end
|
96
97
|
|
97
98
|
###
|
98
99
|
# Parse a file with +filename+
|
99
|
-
def parse_file
|
100
|
+
def parse_file(filename)
|
100
101
|
raise ArgumentError unless filename
|
101
102
|
raise Errno::ENOENT unless File.exist?(filename)
|
102
103
|
raise Errno::EISDIR if File.directory?(filename)
|
103
|
-
ctx = ParserContext.file
|
104
|
+
ctx = ParserContext.file(filename)
|
104
105
|
yield ctx if block_given?
|
105
|
-
ctx.parse_with
|
106
|
+
ctx.parse_with(self)
|
106
107
|
end
|
107
108
|
|
108
|
-
def parse_memory
|
109
|
-
ctx = ParserContext.memory
|
109
|
+
def parse_memory(data)
|
110
|
+
ctx = ParserContext.memory(data)
|
110
111
|
yield ctx if block_given?
|
111
|
-
ctx.parse_with
|
112
|
+
ctx.parse_with(self)
|
112
113
|
end
|
113
114
|
|
114
115
|
private
|
116
|
+
|
115
117
|
def check_encoding(encoding)
|
116
118
|
encoding.upcase.tap do |enc|
|
117
|
-
raise ArgumentError
|
119
|
+
raise ArgumentError, "'#{enc}' is not a valid encoding" unless ENCODINGS[enc]
|
118
120
|
end
|
119
121
|
end
|
120
122
|
end
|
@@ -1,4 +1,5 @@
|
|
1
1
|
# frozen_string_literal: true
|
2
|
+
|
2
3
|
module Nokogiri
|
3
4
|
module XML
|
4
5
|
module SAX
|
@@ -7,9 +8,12 @@ module Nokogiri
|
|
7
8
|
# by the user. Instead, you should be looking at
|
8
9
|
# Nokogiri::XML::SAX::Parser
|
9
10
|
class ParserContext
|
10
|
-
def self.new
|
11
|
-
[:read, :close].all? { |x| thing.respond_to?(x) }
|
12
|
-
io(thing, Parser::ENCODINGS[encoding])
|
11
|
+
def self.new(thing, encoding = "UTF-8")
|
12
|
+
if [:read, :close].all? { |x| thing.respond_to?(x) }
|
13
|
+
io(thing, Parser::ENCODINGS[encoding])
|
14
|
+
else
|
15
|
+
memory(thing)
|
16
|
+
end
|
13
17
|
end
|
14
18
|
end
|
15
19
|
end
|
@@ -1,4 +1,5 @@
|
|
1
1
|
# frozen_string_literal: true
|
2
|
+
|
2
3
|
module Nokogiri
|
3
4
|
module XML
|
4
5
|
module SAX
|
@@ -24,7 +25,6 @@ module Nokogiri
|
|
24
25
|
# parser << "/div>"
|
25
26
|
# parser.finish
|
26
27
|
class PushParser
|
27
|
-
|
28
28
|
# The Nokogiri::XML::SAX::Document on which the PushParser will be
|
29
29
|
# operating
|
30
30
|
attr_accessor :document
|
@@ -32,7 +32,7 @@ module Nokogiri
|
|
32
32
|
###
|
33
33
|
# Create a new PushParser with +doc+ as the SAX Document, providing
|
34
34
|
# an optional +file_name+ and +encoding+
|
35
|
-
def initialize(doc = XML::SAX::Document.new, file_name = nil, encoding =
|
35
|
+
def initialize(doc = XML::SAX::Document.new, file_name = nil, encoding = "UTF-8")
|
36
36
|
@document = doc
|
37
37
|
@encoding = encoding
|
38
38
|
@sax_parser = XML::SAX::Parser.new(doc)
|
@@ -44,16 +44,16 @@ module Nokogiri
|
|
44
44
|
###
|
45
45
|
# Write a +chunk+ of XML to the PushParser. Any callback methods
|
46
46
|
# that can be called will be called immediately.
|
47
|
-
def write
|
47
|
+
def write(chunk, last_chunk = false)
|
48
48
|
native_write(chunk, last_chunk)
|
49
49
|
end
|
50
|
-
|
50
|
+
alias_method :<<, :write
|
51
51
|
|
52
52
|
###
|
53
53
|
# Finish the parsing. This method is only necessary for
|
54
54
|
# Nokogiri::XML::SAX::Document#end_document to be called.
|
55
55
|
def finish
|
56
|
-
write
|
56
|
+
write("", true)
|
57
57
|
end
|
58
58
|
end
|
59
59
|
end
|
data/lib/nokogiri/xml/sax.rb
CHANGED
data/lib/nokogiri/xml/schema.rb
CHANGED
@@ -1,4 +1,5 @@
|
|
1
1
|
# frozen_string_literal: true
|
2
|
+
|
2
3
|
module Nokogiri
|
3
4
|
module XML
|
4
5
|
class << self
|
@@ -42,7 +43,7 @@ module Nokogiri
|
|
42
43
|
###
|
43
44
|
# Create a new Nokogiri::XML::Schema object using a +string_or_io+
|
44
45
|
# object.
|
45
|
-
def self.new
|
46
|
+
def self.new(string_or_io, options = ParseOptions::DEFAULT_SCHEMA)
|
46
47
|
from_document(Nokogiri::XML(string_or_io), options)
|
47
48
|
end
|
48
49
|
|
@@ -51,9 +52,9 @@ module Nokogiri
|
|
51
52
|
# Nokogiri::XML::Document object, or a filename. An Array of
|
52
53
|
# Nokogiri::XML::SyntaxError objects found while validating the
|
53
54
|
# +thing+ is returned.
|
54
|
-
def validate
|
55
|
-
if thing.is_a?(Nokogiri::XML::Document)
|
56
|
-
validate_document(thing)
|
55
|
+
def validate(thing)
|
56
|
+
if thing.is_a?(Nokogiri::XML::Document)
|
57
|
+
validate_document(thing)
|
57
58
|
elsif File.file?(thing)
|
58
59
|
validate_file(thing)
|
59
60
|
else
|
@@ -64,8 +65,8 @@ module Nokogiri
|
|
64
65
|
###
|
65
66
|
# Returns true if +thing+ is a valid Nokogiri::XML::Document or
|
66
67
|
# file.
|
67
|
-
def valid?
|
68
|
-
validate(thing).
|
68
|
+
def valid?(thing)
|
69
|
+
validate(thing).empty?
|
69
70
|
end
|
70
71
|
end
|
71
72
|
end
|
@@ -1,4 +1,6 @@
|
|
1
|
+
# coding: utf-8
|
1
2
|
# frozen_string_literal: true
|
3
|
+
|
2
4
|
module Nokogiri
|
3
5
|
module XML
|
4
6
|
#
|
@@ -11,12 +13,13 @@ module Nokogiri
|
|
11
13
|
module Searchable
|
12
14
|
# Regular expression used by Searchable#search to determine if a query
|
13
15
|
# string is CSS or XPath
|
14
|
-
LOOKS_LIKE_XPATH =
|
16
|
+
LOOKS_LIKE_XPATH = %r{^(\./|/|\.\.|\.$)}
|
15
17
|
|
16
|
-
#
|
18
|
+
# :section: Searching via XPath or CSS Queries
|
17
19
|
|
18
20
|
###
|
19
|
-
# call-seq:
|
21
|
+
# call-seq:
|
22
|
+
# search(*paths, [namespace-bindings, xpath-variable-bindings, custom-handler-class])
|
20
23
|
#
|
21
24
|
# Search this object for +paths+. +paths+ must be one or more XPath or CSS queries:
|
22
25
|
#
|
@@ -52,16 +55,17 @@ module Nokogiri
|
|
52
55
|
paths, handler, ns, binds = extract_params(args)
|
53
56
|
|
54
57
|
xpaths = paths.map(&:to_s).map do |path|
|
55
|
-
(path
|
58
|
+
LOOKS_LIKE_XPATH.match?(path) ? path : xpath_query_from_css_rule(path, ns)
|
56
59
|
end.flatten.uniq
|
57
60
|
|
58
61
|
xpath(*(xpaths + [ns, handler, binds].compact))
|
59
62
|
end
|
60
63
|
|
61
|
-
|
64
|
+
alias_method :/, :search
|
62
65
|
|
63
66
|
###
|
64
|
-
# call-seq:
|
67
|
+
# call-seq:
|
68
|
+
# at(*paths, [namespace-bindings, xpath-variable-bindings, custom-handler-class])
|
65
69
|
#
|
66
70
|
# Search this object for +paths+, and return only the first
|
67
71
|
# result. +paths+ must be one or more XPath or CSS queries.
|
@@ -71,10 +75,11 @@ module Nokogiri
|
|
71
75
|
search(*args).first
|
72
76
|
end
|
73
77
|
|
74
|
-
|
78
|
+
alias_method :%, :at
|
75
79
|
|
76
80
|
###
|
77
|
-
# call-seq:
|
81
|
+
# call-seq:
|
82
|
+
# css(*rules, [namespace-bindings, custom-pseudo-class])
|
78
83
|
#
|
79
84
|
# Search this object for CSS +rules+. +rules+ must be one or more CSS
|
80
85
|
# selectors. For example:
|
@@ -109,11 +114,12 @@ module Nokogiri
|
|
109
114
|
def css(*args)
|
110
115
|
rules, handler, ns, _ = extract_params(args)
|
111
116
|
|
112
|
-
css_internal
|
117
|
+
css_internal(self, rules, handler, ns)
|
113
118
|
end
|
114
119
|
|
115
120
|
##
|
116
|
-
# call-seq:
|
121
|
+
# call-seq:
|
122
|
+
# at_css(*rules, [namespace-bindings, custom-pseudo-class])
|
117
123
|
#
|
118
124
|
# Search this object for CSS +rules+, and return only the first
|
119
125
|
# match. +rules+ must be one or more CSS selectors.
|
@@ -124,7 +130,8 @@ module Nokogiri
|
|
124
130
|
end
|
125
131
|
|
126
132
|
###
|
127
|
-
# call-seq:
|
133
|
+
# call-seq:
|
134
|
+
# xpath(*paths, [namespace-bindings, variable-bindings, custom-handler-class])
|
128
135
|
#
|
129
136
|
# Search this node for XPath +paths+. +paths+ must be one or more XPath
|
130
137
|
# queries.
|
@@ -156,11 +163,12 @@ module Nokogiri
|
|
156
163
|
def xpath(*args)
|
157
164
|
paths, handler, ns, binds = extract_params(args)
|
158
165
|
|
159
|
-
xpath_internal
|
166
|
+
xpath_internal(self, paths, handler, ns, binds)
|
160
167
|
end
|
161
168
|
|
162
169
|
##
|
163
|
-
# call-seq:
|
170
|
+
# call-seq:
|
171
|
+
# at_xpath(*paths, [namespace-bindings, variable-bindings, custom-handler-class])
|
164
172
|
#
|
165
173
|
# Search this node for XPath +paths+, and return only the first
|
166
174
|
# match. +paths+ must be one or more XPath queries.
|
@@ -170,12 +178,21 @@ module Nokogiri
|
|
170
178
|
xpath(*args).first
|
171
179
|
end
|
172
180
|
|
173
|
-
#
|
181
|
+
# :call-seq:
|
182
|
+
# >(selector) → NodeSet
|
183
|
+
#
|
184
|
+
# Search this node's immediate children using CSS selector +selector+
|
185
|
+
def >(selector) # rubocop:disable Naming/BinaryOperatorParameterName
|
186
|
+
ns = (document.root&.namespaces || {})
|
187
|
+
xpath(CSS.xpath_for(selector, prefix: "./", ns: ns).first)
|
188
|
+
end
|
189
|
+
|
190
|
+
# :section:
|
174
191
|
|
175
192
|
private
|
176
193
|
|
177
194
|
def css_internal(node, rules, handler, ns)
|
178
|
-
xpath_internal
|
195
|
+
xpath_internal(node, css_rules_to_xpath(rules, ns), handler, ns, nil)
|
179
196
|
end
|
180
197
|
|
181
198
|
def xpath_internal(node, paths, handler, ns, binds)
|
@@ -198,9 +215,9 @@ module Nokogiri
|
|
198
215
|
ctx.register_namespaces(ns)
|
199
216
|
path = path.gsub(/xmlns:/, " :") unless Nokogiri.uses_libxml?
|
200
217
|
|
201
|
-
binds
|
202
|
-
ctx.register_variable
|
203
|
-
end
|
218
|
+
binds&.each do |key, value|
|
219
|
+
ctx.register_variable(key.to_s, value)
|
220
|
+
end
|
204
221
|
|
205
222
|
ctx.evaluate(path, handler)
|
206
223
|
end
|
@@ -210,10 +227,13 @@ module Nokogiri
|
|
210
227
|
end
|
211
228
|
|
212
229
|
def xpath_query_from_css_rule(rule, ns)
|
213
|
-
visitor = Nokogiri::CSS::
|
230
|
+
visitor = Nokogiri::CSS::XPathVisitor.new(
|
231
|
+
builtins: Nokogiri::CSS::XPathVisitor::BuiltinsConfig::OPTIMAL,
|
232
|
+
doctype: document.xpath_doctype,
|
233
|
+
)
|
214
234
|
self.class::IMPLIED_XPATH_CONTEXTS.map do |implied_xpath_context|
|
215
|
-
CSS.xpath_for(rule.to_s, {:
|
216
|
-
|
235
|
+
CSS.xpath_for(rule.to_s, { prefix: implied_xpath_context, ns: ns,
|
236
|
+
visitor: visitor, })
|
217
237
|
end.join(" | ")
|
218
238
|
end
|
219
239
|
|
@@ -230,7 +250,7 @@ module Nokogiri
|
|
230
250
|
end
|
231
251
|
ns, binds = hashes.reverse
|
232
252
|
|
233
|
-
ns ||= document.root
|
253
|
+
ns ||= (document.root&.namespaces || {})
|
234
254
|
|
235
255
|
[params, handler, ns, binds]
|
236
256
|
end
|
@@ -1,4 +1,5 @@
|
|
1
1
|
# frozen_string_literal: true
|
2
|
+
|
2
3
|
module Nokogiri
|
3
4
|
module XML
|
4
5
|
###
|
@@ -42,9 +43,9 @@ module Nokogiri
|
|
42
43
|
|
43
44
|
def to_s
|
44
45
|
message = super.chomp
|
45
|
-
[location_to_s, level_to_s, message]
|
46
|
-
compact.join(": ")
|
47
|
-
force_encoding(message.encoding)
|
46
|
+
[location_to_s, level_to_s, message]
|
47
|
+
.compact.join(": ")
|
48
|
+
.force_encoding(message.encoding)
|
48
49
|
end
|
49
50
|
|
50
51
|
private
|
@@ -54,7 +55,6 @@ module Nokogiri
|
|
54
55
|
when 3 then "FATAL"
|
55
56
|
when 2 then "ERROR"
|
56
57
|
when 1 then "WARNING"
|
57
|
-
else nil
|
58
58
|
end
|
59
59
|
end
|
60
60
|
|
data/lib/nokogiri/xml/text.rb
CHANGED
data/lib/nokogiri/xml/xpath.rb
CHANGED
@@ -1,7 +1,19 @@
|
|
1
1
|
# frozen_string_literal: true
|
2
|
+
|
2
3
|
module Nokogiri
|
3
4
|
module XML
|
4
5
|
module XPath
|
6
|
+
# The XPath search prefix to search globally, +//+
|
7
|
+
GLOBAL_SEARCH_PREFIX = "//"
|
8
|
+
|
9
|
+
# The XPath search prefix to search direct descendants of the root element, +/+
|
10
|
+
ROOT_SEARCH_PREFIX = "/"
|
11
|
+
|
12
|
+
# The XPath search prefix to search direct descendants of the current element, +./+
|
13
|
+
CURRENT_SEARCH_PREFIX = "./"
|
14
|
+
|
15
|
+
# The XPath search prefix to search anywhere in the current element's subtree, +.//+
|
16
|
+
SUBTREE_SEARCH_PREFIX = ".//"
|
5
17
|
end
|
6
18
|
end
|
7
19
|
end
|
@@ -1,17 +1,16 @@
|
|
1
1
|
# frozen_string_literal: true
|
2
|
+
|
2
3
|
module Nokogiri
|
3
4
|
module XML
|
4
5
|
class XPathContext
|
5
|
-
|
6
6
|
###
|
7
7
|
# Register namespaces in +namespaces+
|
8
8
|
def register_namespaces(namespaces)
|
9
9
|
namespaces.each do |k, v|
|
10
|
-
k = k.to_s.gsub(/.*:/,
|
10
|
+
k = k.to_s.gsub(/.*:/, "") # strip off 'xmlns:' or 'xml:'
|
11
11
|
register_ns(k, v)
|
12
12
|
end
|
13
13
|
end
|
14
|
-
|
15
14
|
end
|
16
15
|
end
|
17
16
|
end
|
data/lib/nokogiri/xml.rb
CHANGED
@@ -1,4 +1,5 @@
|
|
1
1
|
# frozen_string_literal: true
|
2
|
+
|
2
3
|
module Nokogiri
|
3
4
|
class << self
|
4
5
|
###
|
@@ -21,7 +22,6 @@ module Nokogiri
|
|
21
22
|
# Nokogiri::XML::Reader for mor information
|
22
23
|
def Reader(string_or_io, url = nil, encoding = nil, options = ParseOptions::STRICT)
|
23
24
|
options = Nokogiri::XML::ParseOptions.new(options) if Integer === options
|
24
|
-
# Give the options to the user
|
25
25
|
yield options if block_given?
|
26
26
|
|
27
27
|
if string_or_io.respond_to?(:read)
|
@@ -38,8 +38,8 @@ module Nokogiri
|
|
38
38
|
|
39
39
|
####
|
40
40
|
# Parse a fragment from +string+ in to a NodeSet.
|
41
|
-
def fragment(string)
|
42
|
-
XML::DocumentFragment.parse(string)
|
41
|
+
def fragment(string, options = ParseOptions::DEFAULT_XML, &block)
|
42
|
+
XML::DocumentFragment.parse(string, options, &block)
|
43
43
|
end
|
44
44
|
end
|
45
45
|
end
|
data/lib/nokogiri/xslt.rb
CHANGED
@@ -1,4 +1,5 @@
|
|
1
1
|
# frozen_string_literal: true
|
2
|
+
|
2
3
|
module Nokogiri
|
3
4
|
class << self
|
4
5
|
###
|
@@ -8,7 +9,7 @@ module Nokogiri
|
|
8
9
|
#
|
9
10
|
# xslt = Nokogiri::XSLT(File.read(ARGV[0]))
|
10
11
|
#
|
11
|
-
def XSLT
|
12
|
+
def XSLT(stylesheet, modules = {})
|
12
13
|
XSLT.parse(stylesheet, modules)
|
13
14
|
end
|
14
15
|
end
|
@@ -39,7 +40,7 @@ module Nokogiri
|
|
39
40
|
parray = (params.instance_of?(Hash) ? params.to_a.flatten : params).dup
|
40
41
|
parray.each_with_index do |v, i|
|
41
42
|
parray[i] = if i % 2 > 0
|
42
|
-
if
|
43
|
+
if /'/.match?(v)
|
43
44
|
"concat('#{v.gsub(/'/, %q{', "'", '})}')"
|
44
45
|
else
|
45
46
|
"'#{v}'"
|
data/lib/nokogiri.rb
CHANGED
@@ -1,6 +1,5 @@
|
|
1
|
-
#
|
1
|
+
# coding: utf-8
|
2
2
|
# frozen_string_literal: true
|
3
|
-
# Modify the PATH on windows so that the external DLLs will get loaded.
|
4
3
|
|
5
4
|
require "rbconfig"
|
6
5
|
|
@@ -19,30 +18,32 @@ require_relative "nokogiri/extension"
|
|
19
18
|
#
|
20
19
|
# Here is an example:
|
21
20
|
#
|
22
|
-
#
|
23
|
-
#
|
21
|
+
# require 'nokogiri'
|
22
|
+
# require 'open-uri'
|
24
23
|
#
|
25
|
-
#
|
24
|
+
# # Get a Nokogiri::HTML4::Document for the page we’re interested in...
|
26
25
|
#
|
27
|
-
#
|
26
|
+
# doc = Nokogiri::HTML4(URI.open('http://www.google.com/search?q=tenderlove'))
|
28
27
|
#
|
29
|
-
#
|
28
|
+
# # Do funky things with it using Nokogiri::XML::Node methods...
|
30
29
|
#
|
31
|
-
#
|
32
|
-
#
|
33
|
-
#
|
34
|
-
#
|
35
|
-
#
|
30
|
+
# ####
|
31
|
+
# # Search for nodes by css
|
32
|
+
# doc.css('h3.r a.l').each do |link|
|
33
|
+
# puts link.content
|
34
|
+
# end
|
36
35
|
#
|
37
|
-
# See
|
38
|
-
#
|
36
|
+
# See also:
|
37
|
+
#
|
38
|
+
# - Nokogiri::XML::Searchable#css for more information about CSS searching
|
39
|
+
# - Nokogiri::XML::Searchable#xpath for more information about XPath searching
|
39
40
|
module Nokogiri
|
40
41
|
class << self
|
41
42
|
###
|
42
43
|
# Parse an HTML or XML document. +string+ contains the document.
|
43
44
|
def parse(string, url = nil, encoding = nil, options = nil)
|
44
45
|
if string.respond_to?(:read) ||
|
45
|
-
/^\s*<(?:!DOCTYPE\s+)?html[\s>]/i
|
46
|
+
/^\s*<(?:!DOCTYPE\s+)?html[\s>]/i.match?(string[0, 512])
|
46
47
|
# Expect an HTML indicator to appear within the first 512
|
47
48
|
# characters of a document. (<?xml ?> + <?xml-stylesheet ?>
|
48
49
|
# shouldn't be that long)
|
@@ -85,6 +86,7 @@ module Nokogiri
|
|
85
86
|
Nokogiri(*args, &block).slop!
|
86
87
|
end
|
87
88
|
|
89
|
+
# :nodoc:
|
88
90
|
def install_default_aliases
|
89
91
|
# Make sure to support some popular encoding aliases not known by
|
90
92
|
# all iconv implementations.
|
@@ -105,7 +107,7 @@ end
|
|
105
107
|
#
|
106
108
|
# To specify the type of document, use {Nokogiri.XML}, {Nokogiri.HTML4}, or {Nokogiri.HTML5}.
|
107
109
|
def Nokogiri(*args, &block)
|
108
|
-
if
|
110
|
+
if block
|
109
111
|
Nokogiri::HTML4::Builder.new(&block).doc.root
|
110
112
|
else
|
111
113
|
Nokogiri.parse(*args)
|
@@ -113,6 +115,7 @@ def Nokogiri(*args, &block)
|
|
113
115
|
end
|
114
116
|
|
115
117
|
require_relative "nokogiri/version"
|
118
|
+
require_relative "nokogiri/class_resolver"
|
116
119
|
require_relative "nokogiri/syntax_error"
|
117
120
|
require_relative "nokogiri/xml"
|
118
121
|
require_relative "nokogiri/xslt"
|