hpricot 0.6-jruby
Sign up to get free protection for your applications and to get access to all the features.
- data/CHANGELOG +62 -0
- data/COPYING +18 -0
- data/README +284 -0
- data/Rakefile +211 -0
- data/ext/hpricot_scan/HpricotScanService.java +1340 -0
- data/ext/hpricot_scan/extconf.rb +6 -0
- data/ext/hpricot_scan/hpricot_common.rl +76 -0
- data/ext/hpricot_scan/hpricot_scan.c +5976 -0
- data/ext/hpricot_scan/hpricot_scan.h +79 -0
- data/ext/hpricot_scan/hpricot_scan.java.rl +363 -0
- data/ext/hpricot_scan/hpricot_scan.rl +273 -0
- data/extras/mingw-rbconfig.rb +176 -0
- data/lib/hpricot.rb +26 -0
- data/lib/hpricot/blankslate.rb +63 -0
- data/lib/hpricot/builder.rb +200 -0
- data/lib/hpricot/elements.rb +510 -0
- data/lib/hpricot/htmlinfo.rb +672 -0
- data/lib/hpricot/inspect.rb +107 -0
- data/lib/hpricot/modules.rb +37 -0
- data/lib/hpricot/parse.rb +297 -0
- data/lib/hpricot/tag.rb +228 -0
- data/lib/hpricot/tags.rb +164 -0
- data/lib/hpricot/traverse.rb +821 -0
- data/lib/hpricot/xchar.rb +94 -0
- data/lib/i686-linux/hpricot_scan.jar +0 -0
- data/test/files/basic.xhtml +17 -0
- data/test/files/boingboing.html +2266 -0
- data/test/files/cy0.html +3653 -0
- data/test/files/immob.html +400 -0
- data/test/files/pace_application.html +1320 -0
- data/test/files/tenderlove.html +16 -0
- data/test/files/uswebgen.html +220 -0
- data/test/files/utf8.html +1054 -0
- data/test/files/week9.html +1723 -0
- data/test/files/why.xml +19 -0
- data/test/load_files.rb +7 -0
- data/test/test_alter.rb +65 -0
- data/test/test_builder.rb +24 -0
- data/test/test_parser.rb +379 -0
- data/test/test_paths.rb +16 -0
- data/test/test_preserved.rb +66 -0
- data/test/test_xml.rb +28 -0
- metadata +98 -0
@@ -0,0 +1,107 @@
|
|
1
|
+
require 'pp'
|
2
|
+
|
3
|
+
module Hpricot
|
4
|
+
# :stopdoc:
|
5
|
+
class Elements
|
6
|
+
def pretty_print(q)
|
7
|
+
q.object_group(self) { super }
|
8
|
+
end
|
9
|
+
alias inspect pretty_print_inspect
|
10
|
+
end
|
11
|
+
|
12
|
+
class Doc
|
13
|
+
def pretty_print(q)
|
14
|
+
q.object_group(self) { @children.each {|elt| q.breakable; q.pp elt } }
|
15
|
+
end
|
16
|
+
alias inspect pretty_print_inspect
|
17
|
+
end
|
18
|
+
|
19
|
+
class Elem
|
20
|
+
def pretty_print(q)
|
21
|
+
if empty?
|
22
|
+
q.group(1, '{emptyelem', '}') {
|
23
|
+
q.breakable; q.pp @stag
|
24
|
+
}
|
25
|
+
else
|
26
|
+
q.group(1, "{elem", "}") {
|
27
|
+
q.breakable; q.pp @stag
|
28
|
+
if @children
|
29
|
+
@children.each {|elt| q.breakable; q.pp elt }
|
30
|
+
end
|
31
|
+
if @etag
|
32
|
+
q.breakable; q.pp @etag
|
33
|
+
end
|
34
|
+
}
|
35
|
+
end
|
36
|
+
end
|
37
|
+
alias inspect pretty_print_inspect
|
38
|
+
end
|
39
|
+
|
40
|
+
module Leaf
|
41
|
+
def pretty_print(q)
|
42
|
+
q.group(1, '{', '}') {
|
43
|
+
q.text self.class.name.sub(/.*::/,'').downcase
|
44
|
+
if rs = @raw_string
|
45
|
+
rs.scan(/[^\r\n]*(?:\r\n?|\n|[^\r\n]\z)/) {|line|
|
46
|
+
q.breakable
|
47
|
+
q.pp line
|
48
|
+
}
|
49
|
+
elsif self.respond_to? :to_s
|
50
|
+
q.breakable
|
51
|
+
q.text self.to_s
|
52
|
+
end
|
53
|
+
}
|
54
|
+
end
|
55
|
+
alias inspect pretty_print_inspect
|
56
|
+
end
|
57
|
+
|
58
|
+
class STag
|
59
|
+
def pretty_print(q)
|
60
|
+
q.group(1, '<', '>') {
|
61
|
+
q.text @name
|
62
|
+
|
63
|
+
if @raw_attributes
|
64
|
+
@raw_attributes.each {|n, t|
|
65
|
+
q.breakable
|
66
|
+
if t
|
67
|
+
q.text "#{n}=\"#{Hpricot.uxs(t)}\""
|
68
|
+
else
|
69
|
+
q.text n
|
70
|
+
end
|
71
|
+
}
|
72
|
+
end
|
73
|
+
}
|
74
|
+
end
|
75
|
+
alias inspect pretty_print_inspect
|
76
|
+
end
|
77
|
+
|
78
|
+
class ETag
|
79
|
+
def pretty_print(q)
|
80
|
+
q.group(1, '</', '>') {
|
81
|
+
q.text @name
|
82
|
+
}
|
83
|
+
end
|
84
|
+
alias inspect pretty_print_inspect
|
85
|
+
end
|
86
|
+
|
87
|
+
class Text
|
88
|
+
def pretty_print(q)
|
89
|
+
q.text @content.dump
|
90
|
+
end
|
91
|
+
end
|
92
|
+
|
93
|
+
class BogusETag
|
94
|
+
def pretty_print(q)
|
95
|
+
q.group(1, '{', '}') {
|
96
|
+
q.text self.class.name.sub(/.*::/,'').downcase
|
97
|
+
if rs = @raw_string
|
98
|
+
q.breakable
|
99
|
+
q.text rs
|
100
|
+
else
|
101
|
+
q.text "</#{@name}>"
|
102
|
+
end
|
103
|
+
}
|
104
|
+
end
|
105
|
+
end
|
106
|
+
# :startdoc:
|
107
|
+
end
|
@@ -0,0 +1,37 @@
|
|
1
|
+
module Hpricot
|
2
|
+
class Name; include Hpricot end
|
3
|
+
class Context; include Hpricot end
|
4
|
+
|
5
|
+
# :stopdoc:
|
6
|
+
module Tag; include Hpricot end
|
7
|
+
class STag; include Tag end
|
8
|
+
class ETag; include Tag end
|
9
|
+
# :startdoc:
|
10
|
+
|
11
|
+
module Node; include Hpricot end
|
12
|
+
module Container; include Node end
|
13
|
+
class Doc; include Container end
|
14
|
+
class Elem; include Container end
|
15
|
+
module Leaf; include Node end
|
16
|
+
class Text; include Leaf end
|
17
|
+
class XMLDecl; include Leaf end
|
18
|
+
class DocType; include Leaf end
|
19
|
+
class ProcIns; include Leaf end
|
20
|
+
class Comment; include Leaf end
|
21
|
+
class BogusETag; include Leaf end
|
22
|
+
|
23
|
+
module Traverse end
|
24
|
+
module Container::Trav; include Traverse end
|
25
|
+
module Leaf::Trav; include Traverse end
|
26
|
+
class Doc; module Trav; include Container::Trav end; include Trav end
|
27
|
+
class Elem; module Trav; include Container::Trav end; include Trav end
|
28
|
+
class Text; module Trav; include Leaf::Trav end; include Trav end
|
29
|
+
class XMLDecl; module Trav; include Leaf::Trav end; include Trav end
|
30
|
+
class DocType; module Trav; include Leaf::Trav end; include Trav end
|
31
|
+
class ProcIns; module Trav; include Leaf::Trav end; include Trav end
|
32
|
+
class Comment; module Trav; include Leaf::Trav end; include Trav end
|
33
|
+
class BogusETag; module Trav; include Leaf::Trav end; include Trav end
|
34
|
+
|
35
|
+
class Error < StandardError; end
|
36
|
+
end
|
37
|
+
|
@@ -0,0 +1,297 @@
|
|
1
|
+
require 'hpricot/htmlinfo'
|
2
|
+
|
3
|
+
def Hpricot(input = nil, opts = {}, &blk)
|
4
|
+
Hpricot.parse(input, opts, &blk)
|
5
|
+
end
|
6
|
+
|
7
|
+
module Hpricot
|
8
|
+
# Exception class used for any errors related to deficiencies in the system when
|
9
|
+
# handling the character encodings of a document.
|
10
|
+
class EncodingError < StandardError; end
|
11
|
+
|
12
|
+
# Hpricot.parse parses <i>input</i> and return a document tree.
|
13
|
+
# represented by Hpricot::Doc.
|
14
|
+
def Hpricot.parse(input = nil, opts = {}, &blk)
|
15
|
+
Doc.new(make(input, opts, &blk))
|
16
|
+
end
|
17
|
+
|
18
|
+
# Hpricot::XML parses <i>input</i>, disregarding all the HTML rules
|
19
|
+
# and returning a document tree.
|
20
|
+
def Hpricot.XML(input, opts = {})
|
21
|
+
Doc.new(make(input, opts.merge(:xml => true)))
|
22
|
+
end
|
23
|
+
|
24
|
+
# :stopdoc:
|
25
|
+
|
26
|
+
def Hpricot.make(input = nil, opts = {}, &blk)
|
27
|
+
opts = {:fixup_tags => false}.merge(opts)
|
28
|
+
unless input or blk
|
29
|
+
raise ArgumentError, "An Hpricot document must be built from an input source (a String) or a block."
|
30
|
+
end
|
31
|
+
|
32
|
+
conv = opts[:xml] ? :to_s : :downcase
|
33
|
+
|
34
|
+
fragment =
|
35
|
+
if input
|
36
|
+
case opts[:encoding]
|
37
|
+
when nil
|
38
|
+
when 'utf-8'
|
39
|
+
unless defined? Encoding::Character::UTF8
|
40
|
+
raise EncodingError, "The ruby-character-encodings library could not be found for utf-8 mode."
|
41
|
+
end
|
42
|
+
else
|
43
|
+
raise EncodingError, "No encoding option `#{opts[:encoding]}' is available."
|
44
|
+
end
|
45
|
+
|
46
|
+
if opts[:xhtml_strict]
|
47
|
+
opts[:fixup_tags] = true
|
48
|
+
end
|
49
|
+
|
50
|
+
stack = [[nil, nil, [], [], [], []]]
|
51
|
+
Hpricot.scan(input) do |token|
|
52
|
+
if stack.last[5] == :CDATA and ![:procins, :comment, :cdata].include?(token[0]) and
|
53
|
+
!(token[0] == :etag and token[1].casecmp(stack.last[0]).zero?)
|
54
|
+
token[0] = :text
|
55
|
+
token[1] = token[3] if token[3]
|
56
|
+
end
|
57
|
+
|
58
|
+
if !opts[:xml] and token[0] == :emptytag
|
59
|
+
token[1] = token[1].send(conv)
|
60
|
+
if ElementContent[token[1].downcase] != :EMPTY
|
61
|
+
token[0] = :stag
|
62
|
+
end
|
63
|
+
end
|
64
|
+
|
65
|
+
# TODO: downcase instead when parsing attributes?
|
66
|
+
if !opts[:xml] and token[2].is_a?(Hash)
|
67
|
+
token[2] = token[2].inject({}) { |hsh,(k,v)| hsh[k.downcase] = v; hsh }
|
68
|
+
end
|
69
|
+
|
70
|
+
case token[0]
|
71
|
+
when :stag
|
72
|
+
case opts[:encoding] when 'utf-8'
|
73
|
+
token.map! { |str| u(str) if str.is_a? String }
|
74
|
+
end
|
75
|
+
|
76
|
+
stagname = token[0] = token[1] = token[1].send(conv)
|
77
|
+
if ElementContent[stagname] == :EMPTY and !opts[:xml]
|
78
|
+
token[0] = :emptytag
|
79
|
+
stack.last[2] << token
|
80
|
+
else
|
81
|
+
unless opts[:xml]
|
82
|
+
if opts[:fixup_tags]
|
83
|
+
# obey the tag rules set up by the current element
|
84
|
+
if ElementContent.has_key? stagname
|
85
|
+
trans = nil
|
86
|
+
(stack.length-1).downto(0) do |i|
|
87
|
+
untags = stack[i][5]
|
88
|
+
break unless untags.include? stagname
|
89
|
+
# puts "** ILLEGAL #{stagname} IN #{stack[i][0]}"
|
90
|
+
trans = i
|
91
|
+
end
|
92
|
+
if trans.to_i > 1
|
93
|
+
eles = stack.slice!(trans..-1)
|
94
|
+
stack.last[2] += eles
|
95
|
+
# puts "** TRANSPLANTED #{stagname} TO #{stack.last[0]}"
|
96
|
+
end
|
97
|
+
elsif opts[:xhtml_strict]
|
98
|
+
token[2] = {'class' => stagname}
|
99
|
+
stagname = token[0] = "div"
|
100
|
+
end
|
101
|
+
end
|
102
|
+
|
103
|
+
# setup tag rules for inside this element
|
104
|
+
if ElementContent[stagname] == :CDATA
|
105
|
+
uncontainable_tags = :CDATA
|
106
|
+
elsif opts[:fixup_tags]
|
107
|
+
possible_tags = ElementContent[stagname]
|
108
|
+
excluded_tags, included_tags = stack.last[3..4]
|
109
|
+
if possible_tags
|
110
|
+
excluded_tags = excluded_tags | (ElementExclusions[stagname] || [])
|
111
|
+
included_tags = included_tags | (ElementInclusions[stagname] || [])
|
112
|
+
containable_tags = (possible_tags | included_tags) - excluded_tags
|
113
|
+
uncontainable_tags = ElementContent.keys - containable_tags
|
114
|
+
else
|
115
|
+
# If the tagname is unknown, it is assumed that any element
|
116
|
+
# except excluded can be contained.
|
117
|
+
uncontainable_tags = excluded_tags
|
118
|
+
end
|
119
|
+
end
|
120
|
+
end
|
121
|
+
unless opts[:xml]
|
122
|
+
case token[2] when Hash
|
123
|
+
token[2] = token[2].inject({}) { |hsh,(k,v)| hsh[k.downcase] = v; hsh }
|
124
|
+
end
|
125
|
+
end
|
126
|
+
stack << [stagname, token, [], excluded_tags, included_tags, uncontainable_tags]
|
127
|
+
end
|
128
|
+
when :etag
|
129
|
+
etagname = token[0] = token[1].send(conv)
|
130
|
+
if opts[:xhtml_strict] and not ElementContent.has_key? etagname
|
131
|
+
etagname = token[0] = "div"
|
132
|
+
end
|
133
|
+
matched_elem = nil
|
134
|
+
(stack.length-1).downto(0) do |i|
|
135
|
+
stagname, = stack[i]
|
136
|
+
if stagname == etagname
|
137
|
+
matched_elem = stack[i]
|
138
|
+
stack[i][1] += token
|
139
|
+
eles = stack.slice!((i+1)..-1)
|
140
|
+
stack.last[2] += eles
|
141
|
+
break
|
142
|
+
end
|
143
|
+
end
|
144
|
+
unless matched_elem
|
145
|
+
stack.last[2] << [:bogus_etag, token.first, token.last]
|
146
|
+
else
|
147
|
+
ele = stack.pop
|
148
|
+
stack.last[2] << ele
|
149
|
+
end
|
150
|
+
when :text
|
151
|
+
l = stack.last[2].last
|
152
|
+
if l and l[0] == :text
|
153
|
+
l[1] += token[1]
|
154
|
+
else
|
155
|
+
stack.last[2] << token
|
156
|
+
end
|
157
|
+
else
|
158
|
+
stack.last[2] << token
|
159
|
+
end
|
160
|
+
end
|
161
|
+
|
162
|
+
while 1 < stack.length
|
163
|
+
ele = stack.pop
|
164
|
+
stack.last[2] << ele
|
165
|
+
end
|
166
|
+
|
167
|
+
structure_list = stack[0][2]
|
168
|
+
structure_list.map {|s| build_node(s, opts) }
|
169
|
+
elsif blk
|
170
|
+
Hpricot.build(&blk).children
|
171
|
+
end
|
172
|
+
end
|
173
|
+
|
174
|
+
def Hpricot.build_node(structure, opts = {})
|
175
|
+
case structure[0]
|
176
|
+
when String
|
177
|
+
tagname, _, attrs, sraw, _, _, _, eraw = structure[1]
|
178
|
+
children = structure[2]
|
179
|
+
etag = eraw && ETag.parse(tagname, eraw)
|
180
|
+
stag = STag.parse(tagname, attrs, sraw, true)
|
181
|
+
if !children.empty? || etag
|
182
|
+
Elem.new(stag,
|
183
|
+
children.map {|c| build_node(c, opts) },
|
184
|
+
etag)
|
185
|
+
else
|
186
|
+
Elem.new(stag)
|
187
|
+
end
|
188
|
+
when :text
|
189
|
+
Text.parse_pcdata(structure[1])
|
190
|
+
when :emptytag
|
191
|
+
Elem.new(STag.parse(structure[1], structure[2], structure[3], false))
|
192
|
+
when :bogus_etag
|
193
|
+
BogusETag.parse(structure[1], structure[2])
|
194
|
+
when :xmldecl
|
195
|
+
XMLDecl.parse(structure[2], structure[3])
|
196
|
+
when :doctype
|
197
|
+
if opts[:xhtml_strict]
|
198
|
+
structure[2]['system_id'] = "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd"
|
199
|
+
structure[2]['public_id'] = "-//W3C//DTD XHTML 1.0 Strict//EN"
|
200
|
+
end
|
201
|
+
DocType.parse(structure[1], structure[2], structure[3])
|
202
|
+
when :procins
|
203
|
+
ProcIns.parse(structure[1])
|
204
|
+
when :comment
|
205
|
+
Comment.parse(structure[1])
|
206
|
+
when :cdata_content
|
207
|
+
Text.parse_cdata_content(structure[1])
|
208
|
+
when :cdata
|
209
|
+
Text.parse_cdata_section(structure[1])
|
210
|
+
else
|
211
|
+
raise Exception, "[bug] unknown structure: #{structure.inspect}"
|
212
|
+
end
|
213
|
+
end
|
214
|
+
|
215
|
+
def STag.parse(qname, attrs, raw_string, is_stag)
|
216
|
+
result = STag.new(qname, attrs)
|
217
|
+
result.raw_string = raw_string
|
218
|
+
result
|
219
|
+
end
|
220
|
+
|
221
|
+
def ETag.parse(qname, raw_string)
|
222
|
+
result = self.new(qname)
|
223
|
+
result.raw_string = raw_string
|
224
|
+
result
|
225
|
+
end
|
226
|
+
|
227
|
+
def BogusETag.parse(qname, raw_string)
|
228
|
+
result = self.new(qname)
|
229
|
+
result.raw_string = raw_string
|
230
|
+
result
|
231
|
+
end
|
232
|
+
|
233
|
+
def Text.parse_pcdata(raw_string)
|
234
|
+
result = Text.new(raw_string)
|
235
|
+
result
|
236
|
+
end
|
237
|
+
|
238
|
+
def Text.parse_cdata_content(raw_string)
|
239
|
+
result = CData.new(raw_string)
|
240
|
+
result
|
241
|
+
end
|
242
|
+
|
243
|
+
def Text.parse_cdata_section(content)
|
244
|
+
result = CData.new(content)
|
245
|
+
result
|
246
|
+
end
|
247
|
+
|
248
|
+
def XMLDecl.parse(attrs, raw_string)
|
249
|
+
attrs ||= {}
|
250
|
+
version = attrs['version']
|
251
|
+
encoding = attrs['encoding']
|
252
|
+
case attrs['standalone']
|
253
|
+
when 'yes'
|
254
|
+
standalone = true
|
255
|
+
when 'no'
|
256
|
+
standalone = false
|
257
|
+
else
|
258
|
+
standalone = nil
|
259
|
+
end
|
260
|
+
|
261
|
+
result = XMLDecl.new(version, encoding, standalone)
|
262
|
+
result.raw_string = raw_string
|
263
|
+
result
|
264
|
+
end
|
265
|
+
|
266
|
+
def DocType.parse(root_element_name, attrs, raw_string)
|
267
|
+
if attrs
|
268
|
+
public_identifier = attrs['public_id']
|
269
|
+
system_identifier = attrs['system_id']
|
270
|
+
end
|
271
|
+
|
272
|
+
root_element_name = root_element_name.downcase
|
273
|
+
|
274
|
+
result = DocType.new(root_element_name, public_identifier, system_identifier)
|
275
|
+
result.raw_string = raw_string
|
276
|
+
result
|
277
|
+
end
|
278
|
+
|
279
|
+
def ProcIns.parse(raw_string)
|
280
|
+
_, target, content = *raw_string.match(/\A<\?(\S+)\s+(.+)/m)
|
281
|
+
result = ProcIns.new(target, content)
|
282
|
+
result
|
283
|
+
end
|
284
|
+
|
285
|
+
def Comment.parse(content)
|
286
|
+
result = Comment.new(content)
|
287
|
+
result
|
288
|
+
end
|
289
|
+
|
290
|
+
module Pat
|
291
|
+
NameChar = /[-A-Za-z0-9._:]/
|
292
|
+
Name = /[A-Za-z_:]#{NameChar}*/
|
293
|
+
Nmtoken = /#{NameChar}+/
|
294
|
+
end
|
295
|
+
|
296
|
+
# :startdoc:
|
297
|
+
end
|
data/lib/hpricot/tag.rb
ADDED
@@ -0,0 +1,228 @@
|
|
1
|
+
module Hpricot
|
2
|
+
# :stopdoc:
|
3
|
+
|
4
|
+
class Doc
|
5
|
+
attr_accessor :children
|
6
|
+
def initialize(children = [])
|
7
|
+
@children = children ? children.each { |c| c.parent = self } : []
|
8
|
+
end
|
9
|
+
def output(out, opts = {})
|
10
|
+
@children.each do |n|
|
11
|
+
n.output(out, opts)
|
12
|
+
end
|
13
|
+
out
|
14
|
+
end
|
15
|
+
def altered!; end
|
16
|
+
end
|
17
|
+
|
18
|
+
class BaseEle
|
19
|
+
attr_accessor :raw_string, :parent
|
20
|
+
def html_quote(str)
|
21
|
+
"\"" + str.gsub('"', '\\"') + "\""
|
22
|
+
end
|
23
|
+
def if_output(opts)
|
24
|
+
if opts[:preserve] and not @raw_string.nil?
|
25
|
+
@raw_string
|
26
|
+
else
|
27
|
+
yield opts
|
28
|
+
end
|
29
|
+
end
|
30
|
+
def pathname; self.name end
|
31
|
+
def altered!
|
32
|
+
@raw_string = nil
|
33
|
+
end
|
34
|
+
def self.alterable(*fields)
|
35
|
+
attr_accessor(*fields)
|
36
|
+
fields.each do |f|
|
37
|
+
define_method("#{f}=") do |v|
|
38
|
+
altered!
|
39
|
+
instance_variable_set("@#{f}", v)
|
40
|
+
end
|
41
|
+
end
|
42
|
+
end
|
43
|
+
end
|
44
|
+
|
45
|
+
class Elem
|
46
|
+
attr_accessor :stag, :etag, :children
|
47
|
+
def initialize(stag, children=nil, etag=nil)
|
48
|
+
@stag, @etag = stag, etag
|
49
|
+
@children = children ? children.each { |c| c.parent = self } : []
|
50
|
+
end
|
51
|
+
def empty?; @children.empty? end
|
52
|
+
[:name, :raw_attributes, :parent, :altered!].each do |m|
|
53
|
+
[m, "#{m}="].each { |m2| define_method(m2) { |*a| [@etag, @stag].inject { |_,t| t.send(m2, *a) if t and t.respond_to?(m2) } } }
|
54
|
+
end
|
55
|
+
def attributes
|
56
|
+
if raw_attributes
|
57
|
+
raw_attributes.inject({}) do |hsh, (k, v)|
|
58
|
+
hsh[k] = Hpricot.uxs(v)
|
59
|
+
hsh
|
60
|
+
end
|
61
|
+
end
|
62
|
+
end
|
63
|
+
def to_plain_text
|
64
|
+
if self.name == 'br'
|
65
|
+
"\n"
|
66
|
+
elsif self.name == 'p'
|
67
|
+
"\n\n" + super + "\n\n"
|
68
|
+
elsif self.name == 'a' and self.has_attribute?('href')
|
69
|
+
"#{super} [#{self['href']}]"
|
70
|
+
elsif self.name == 'img' and self.has_attribute?('src')
|
71
|
+
"[img:#{self['src']}]"
|
72
|
+
else
|
73
|
+
super
|
74
|
+
end
|
75
|
+
end
|
76
|
+
def pathname; self.name end
|
77
|
+
def output(out, opts = {})
|
78
|
+
if empty? and ElementContent[@stag.name] == :EMPTY
|
79
|
+
@stag.output(out, opts.merge(:style => :empty))
|
80
|
+
else
|
81
|
+
@stag.output(out, opts)
|
82
|
+
@children.each { |n| n.output(out, opts) }
|
83
|
+
if @etag
|
84
|
+
@etag.output(out, opts)
|
85
|
+
elsif !opts[:preserve]
|
86
|
+
ETag.new(@stag.name).output(out, opts)
|
87
|
+
end
|
88
|
+
end
|
89
|
+
out
|
90
|
+
end
|
91
|
+
end
|
92
|
+
|
93
|
+
class STag < BaseEle
|
94
|
+
def initialize(name, attributes=nil)
|
95
|
+
@name = name.to_s
|
96
|
+
@raw_attributes = attributes || {}
|
97
|
+
end
|
98
|
+
alterable :name, :raw_attributes
|
99
|
+
def attributes_as_html
|
100
|
+
if @raw_attributes
|
101
|
+
@raw_attributes.map do |aname, aval|
|
102
|
+
" #{aname}" +
|
103
|
+
(aval ? "=\"#{aval}\"" : "")
|
104
|
+
end.join
|
105
|
+
end
|
106
|
+
end
|
107
|
+
def output(out, opts = {})
|
108
|
+
out <<
|
109
|
+
if_output(opts) do
|
110
|
+
"<#{@name}#{attributes_as_html}" +
|
111
|
+
(opts[:style] == :empty ? " /" : "") +
|
112
|
+
">"
|
113
|
+
end
|
114
|
+
end
|
115
|
+
end
|
116
|
+
|
117
|
+
class ETag < BaseEle
|
118
|
+
def initialize(qualified_name)
|
119
|
+
@name = qualified_name.to_s
|
120
|
+
end
|
121
|
+
alterable :name
|
122
|
+
def output(out, opts = {})
|
123
|
+
out <<
|
124
|
+
if_output(opts) do
|
125
|
+
"</#{@name}>"
|
126
|
+
end
|
127
|
+
end
|
128
|
+
end
|
129
|
+
|
130
|
+
class BogusETag < ETag
|
131
|
+
def output(out, opts = {}); out << if_output(opts) { '' }; end
|
132
|
+
end
|
133
|
+
|
134
|
+
class Text < BaseEle
|
135
|
+
def initialize(text)
|
136
|
+
@content = text
|
137
|
+
end
|
138
|
+
alterable :content
|
139
|
+
def pathname; "text()" end
|
140
|
+
def to_s
|
141
|
+
Hpricot.uxs(@content)
|
142
|
+
end
|
143
|
+
alias_method :inner_text, :to_s
|
144
|
+
alias_method :to_plain_text, :to_s
|
145
|
+
def output(out, opts = {})
|
146
|
+
out <<
|
147
|
+
if_output(opts) do
|
148
|
+
@content
|
149
|
+
end
|
150
|
+
end
|
151
|
+
end
|
152
|
+
|
153
|
+
class CData < Text
|
154
|
+
alias_method :to_s, :content
|
155
|
+
alias_method :to_plain_text, :content
|
156
|
+
def output(out, opts = {})
|
157
|
+
out <<
|
158
|
+
if_output(opts) do
|
159
|
+
"<![CDATA[#@content]]>"
|
160
|
+
end
|
161
|
+
end
|
162
|
+
end
|
163
|
+
|
164
|
+
class XMLDecl < BaseEle
|
165
|
+
def initialize(version, encoding, standalone)
|
166
|
+
@version, @encoding, @standalone = version, encoding, standalone
|
167
|
+
end
|
168
|
+
alterable :version, :encoding, :standalone
|
169
|
+
def pathname; "xmldecl()" end
|
170
|
+
def output(out, opts = {})
|
171
|
+
out <<
|
172
|
+
if_output(opts) do
|
173
|
+
"<?xml version=\"#{@version}\"" +
|
174
|
+
(@encoding ? " encoding=\"#{encoding}\"" : "") +
|
175
|
+
(@standalone != nil ? " standalone=\"#{standalone ? 'yes' : 'no'}\"" : "") +
|
176
|
+
"?>"
|
177
|
+
end
|
178
|
+
end
|
179
|
+
end
|
180
|
+
|
181
|
+
class DocType < BaseEle
|
182
|
+
def initialize(target, pubid, sysid)
|
183
|
+
@target, @public_id, @system_id = target, pubid, sysid
|
184
|
+
end
|
185
|
+
alterable :target, :public_id, :system_id
|
186
|
+
def pathname; "doctype()" end
|
187
|
+
def output(out, opts = {})
|
188
|
+
out <<
|
189
|
+
if_output(opts) do
|
190
|
+
"<!DOCTYPE #{@target} " +
|
191
|
+
(@public_id ? "PUBLIC \"#{@public_id}\"" : "SYSTEM") +
|
192
|
+
(@system_id ? " #{html_quote(@system_id)}" : "") + ">"
|
193
|
+
end
|
194
|
+
end
|
195
|
+
end
|
196
|
+
|
197
|
+
class ProcIns < BaseEle
|
198
|
+
def initialize(target, content)
|
199
|
+
@target, @content = target, content
|
200
|
+
end
|
201
|
+
def pathname; "procins()" end
|
202
|
+
alterable :target, :content
|
203
|
+
def output(out, opts = {})
|
204
|
+
out <<
|
205
|
+
if_output(opts) do
|
206
|
+
"<?#{@target}" +
|
207
|
+
(@content ? " #{@content}" : "") +
|
208
|
+
"?>"
|
209
|
+
end
|
210
|
+
end
|
211
|
+
end
|
212
|
+
|
213
|
+
class Comment < BaseEle
|
214
|
+
def initialize(content)
|
215
|
+
@content = content
|
216
|
+
end
|
217
|
+
def pathname; "comment()" end
|
218
|
+
alterable :content
|
219
|
+
def output(out, opts = {})
|
220
|
+
out <<
|
221
|
+
if_output(opts) do
|
222
|
+
"<!--#{@content}-->"
|
223
|
+
end
|
224
|
+
end
|
225
|
+
end
|
226
|
+
|
227
|
+
# :startdoc:
|
228
|
+
end
|