rgen 0.5.4 → 0.6.0
Sign up to get free protection for your applications and to get access to all the features.
- data/CHANGELOG +28 -0
- data/Rakefile +3 -4
- data/lib/ea_support/uml13_ea_metamodel.rb +3 -3
- data/lib/ea_support/uml13_ea_to_uml13.rb +33 -2
- data/lib/ea_support/uml13_to_uml13_ea.rb +7 -0
- data/lib/mmgen/mm_ext/ecore_mmgen_ext.rb +4 -4
- data/lib/mmgen/templates/metamodel_generator.tpl +143 -143
- data/lib/rgen/ecore/ecore.rb +11 -1
- data/lib/rgen/ecore/ecore_interface.rb +47 -0
- data/lib/rgen/ecore/ecore_to_ruby.rb +166 -0
- data/lib/rgen/ecore/{ecore_transformer.rb → ruby_to_ecore.rb} +11 -11
- data/lib/rgen/environment.rb +15 -2
- data/lib/rgen/fragment/dump_file_cache.rb +63 -0
- data/lib/rgen/fragment/fragmented_model.rb +139 -0
- data/lib/rgen/fragment/model_fragment.rb +268 -0
- data/lib/rgen/instantiator/abstract_xml_instantiator.rb +44 -72
- data/lib/rgen/instantiator/default_xml_instantiator.rb +2 -2
- data/lib/rgen/instantiator/ecore_xml_instantiator.rb +16 -1
- data/lib/rgen/instantiator/json_instantiator.rb +16 -2
- data/lib/rgen/instantiator/nodebased_xml_instantiator.rb +118 -138
- data/lib/rgen/instantiator/qualified_name_resolver.rb +5 -1
- data/lib/rgen/instantiator/reference_resolver.rb +126 -24
- data/lib/rgen/instantiator/xmi11_instantiator.rb +6 -2
- data/lib/rgen/metamodel_builder.rb +18 -6
- data/lib/rgen/metamodel_builder/builder_extensions.rb +431 -407
- data/lib/rgen/metamodel_builder/builder_runtime.rb +8 -8
- data/lib/rgen/metamodel_builder/constant_order_helper.rb +4 -4
- data/lib/rgen/metamodel_builder/data_types.rb +5 -1
- data/lib/rgen/metamodel_builder/intermediate/feature.rb +167 -0
- data/lib/rgen/metamodel_builder/module_extension.rb +2 -2
- data/lib/rgen/model_builder.rb +10 -5
- data/lib/rgen/model_builder/builder_context.rb +17 -1
- data/lib/rgen/serializer/opposite_reference_filter.rb +18 -0
- data/lib/rgen/serializer/qualified_name_provider.rb +45 -0
- data/lib/rgen/template_language/template_container.rb +3 -1
- data/lib/rgen/{auto_class_creator.rb → util/auto_class_creator.rb} +6 -1
- data/lib/rgen/util/cached_glob.rb +67 -0
- data/lib/rgen/util/file_cache_map.rb +104 -0
- data/lib/rgen/util/file_change_detector.rb +78 -0
- data/lib/rgen/{method_delegation.rb → util/method_delegation.rb} +18 -3
- data/lib/rgen/{model_comparator.rb → util/model_comparator.rb} +17 -5
- data/lib/rgen/{model_comparator_base.rb → util/model_comparator_base.rb} +6 -1
- data/lib/rgen/{model_dumper.rb → util/model_dumper.rb} +6 -1
- data/lib/rgen/{name_helper.rb → util/name_helper.rb} +6 -1
- data/lib/rgen/util/pattern_matcher.rb +329 -0
- data/lib/transformers/uml13_to_ecore.rb +103 -60
- data/test/ecore_self_test.rb +43 -42
- data/test/json_test.rb +15 -0
- data/test/metamodel_builder_test.rb +361 -206
- data/test/metamodel_from_ecore_test.rb +45 -0
- data/test/metamodel_order_test.rb +10 -4
- data/test/metamodel_roundtrip_test.rb +2 -2
- data/test/metamodel_roundtrip_test/TestModel_Regenerated.rb +1 -1
- data/test/metamodel_roundtrip_test/houseMetamodel_Regenerated.ecore +50 -50
- data/test/method_delegation_test.rb +9 -9
- data/test/model_builder/ecore_internal.rb +19 -9
- data/test/model_builder/serializer_test.rb +1 -1
- data/test/reference_resolver_test.rb +79 -12
- data/test/rgen_test.rb +2 -0
- data/test/template_language_test.rb +7 -0
- data/test/template_language_test/templates/callback_indent_test/a.tpl +12 -0
- data/test/template_language_test/templates/callback_indent_test/b.tpl +5 -0
- data/test/testmodel/ea_testmodel_regenerated.xml +588 -583
- data/test/transformer_test.rb +3 -3
- data/test/util/file_cache_map_test.rb +91 -0
- data/test/util/file_cache_map_test/testdir/fileA +1 -0
- data/test/util_test.rb +4 -0
- data/test/xml_instantiator_test.rb +139 -135
- metadata +49 -104
- data/lib/rgen/ecore/ecore_instantiator.rb +0 -31
- data/lib/rgen/metamodel_builder/metamodel_description.rb +0 -232
- data/redist/xmlscan/ChangeLog +0 -1301
- data/redist/xmlscan/README +0 -34
- data/redist/xmlscan/THANKS +0 -11
- data/redist/xmlscan/doc/changes.html +0 -74
- data/redist/xmlscan/doc/changes.rd +0 -80
- data/redist/xmlscan/doc/en/conformance.html +0 -136
- data/redist/xmlscan/doc/en/conformance.rd +0 -152
- data/redist/xmlscan/doc/en/manual.html +0 -356
- data/redist/xmlscan/doc/en/manual.rd +0 -402
- data/redist/xmlscan/doc/ja/conformance.ja.html +0 -118
- data/redist/xmlscan/doc/ja/conformance.ja.rd +0 -134
- data/redist/xmlscan/doc/ja/manual.ja.html +0 -325
- data/redist/xmlscan/doc/ja/manual.ja.rd +0 -370
- data/redist/xmlscan/doc/src/Makefile +0 -41
- data/redist/xmlscan/doc/src/conformance.rd.src +0 -256
- data/redist/xmlscan/doc/src/langsplit.rb +0 -110
- data/redist/xmlscan/doc/src/manual.rd.src +0 -614
- data/redist/xmlscan/install.rb +0 -41
- data/redist/xmlscan/lib/xmlscan/encoding.rb +0 -311
- data/redist/xmlscan/lib/xmlscan/htmlscan.rb +0 -289
- data/redist/xmlscan/lib/xmlscan/namespace.rb +0 -352
- data/redist/xmlscan/lib/xmlscan/parser.rb +0 -299
- data/redist/xmlscan/lib/xmlscan/scanner.rb +0 -1109
- data/redist/xmlscan/lib/xmlscan/version.rb +0 -22
- data/redist/xmlscan/lib/xmlscan/visitor.rb +0 -158
- data/redist/xmlscan/lib/xmlscan/xmlchar.rb +0 -441
- data/redist/xmlscan/memo/CONFORMANCE +0 -1249
- data/redist/xmlscan/memo/PRODUCTIONS +0 -195
- data/redist/xmlscan/memo/contentspec.ry +0 -335
- data/redist/xmlscan/samples/chibixml.rb +0 -105
- data/redist/xmlscan/samples/getxmlchar.rb +0 -122
- data/redist/xmlscan/samples/rexml.rb +0 -159
- data/redist/xmlscan/samples/xmlbench.rb +0 -88
- data/redist/xmlscan/samples/xmlbench/parser/chibixml.rb +0 -22
- data/redist/xmlscan/samples/xmlbench/parser/nqxml.rb +0 -29
- data/redist/xmlscan/samples/xmlbench/parser/rexml.rb +0 -62
- data/redist/xmlscan/samples/xmlbench/parser/xmlparser.rb +0 -22
- data/redist/xmlscan/samples/xmlbench/parser/xmlscan-0.0.10.rb +0 -62
- data/redist/xmlscan/samples/xmlbench/parser/xmlscan-chibixml.rb +0 -22
- data/redist/xmlscan/samples/xmlbench/parser/xmlscan-rexml.rb +0 -22
- data/redist/xmlscan/samples/xmlbench/parser/xmlscan.rb +0 -99
- data/redist/xmlscan/samples/xmlbench/xmlbench-lib.rb +0 -116
- data/redist/xmlscan/samples/xmlconftest.rb +0 -200
- data/redist/xmlscan/test.rb +0 -7
- data/redist/xmlscan/tests/deftestcase.rb +0 -73
- data/redist/xmlscan/tests/runtest.rb +0 -47
- data/redist/xmlscan/tests/testall.rb +0 -14
- data/redist/xmlscan/tests/testencoding.rb +0 -438
- data/redist/xmlscan/tests/testhtmlscan.rb +0 -752
- data/redist/xmlscan/tests/testnamespace.rb +0 -457
- data/redist/xmlscan/tests/testparser.rb +0 -591
- data/redist/xmlscan/tests/testscanner.rb +0 -1749
- data/redist/xmlscan/tests/testxmlchar.rb +0 -143
- data/redist/xmlscan/tests/visitor.rb +0 -34
data/redist/xmlscan/install.rb
DELETED
@@ -1,41 +0,0 @@
|
|
1
|
-
#!/usr/bin/ruby
|
2
|
-
#
|
3
|
-
# install.rb
|
4
|
-
#
|
5
|
-
# $Id: install.rb,v 1.2 2002/12/26 21:09:38 katsu Exp $
|
6
|
-
|
7
|
-
require 'rbconfig'
|
8
|
-
require 'ftools'
|
9
|
-
require 'find'
|
10
|
-
require 'getoptlong'
|
11
|
-
|
12
|
-
DEFAULT_DESTDIR = Config::CONFIG['sitelibdir'] || Config::CONFIG['sitedir']
|
13
|
-
SRCDIR = File.dirname(__FILE__)
|
14
|
-
|
15
|
-
|
16
|
-
def install_rb(from, to)
|
17
|
-
from = SRCDIR + '/' + from
|
18
|
-
Find.find(from) { |src|
|
19
|
-
next unless File.file? src
|
20
|
-
next unless /\.rb\z/ =~ src
|
21
|
-
dst = src.sub(/\A#{Regexp.escape(from)}/, to)
|
22
|
-
File.makedirs File.dirname(dst), true
|
23
|
-
File.install src, dst, 0644, true
|
24
|
-
}
|
25
|
-
end
|
26
|
-
|
27
|
-
|
28
|
-
destdir = DEFAULT_DESTDIR
|
29
|
-
begin
|
30
|
-
GetoptLong.new([ "-d", "--destdir", GetoptLong::REQUIRED_ARGUMENT ]
|
31
|
-
).each_option { |opt, arg|
|
32
|
-
case opt
|
33
|
-
when '-d' then
|
34
|
-
destdir = arg
|
35
|
-
end
|
36
|
-
}
|
37
|
-
rescue
|
38
|
-
exit 2
|
39
|
-
end
|
40
|
-
|
41
|
-
install_rb "lib", destdir
|
@@ -1,311 +0,0 @@
|
|
1
|
-
#
|
2
|
-
# xmlscan/encoding.rb
|
3
|
-
#
|
4
|
-
# Copyright (C) Ueno Katsuhiro 2002
|
5
|
-
#
|
6
|
-
# $Id: encoding.rb,v 1.3 2003/01/12 04:10:33 katsu Exp $
|
7
|
-
#
|
8
|
-
|
9
|
-
require 'xmlscan/visitor'
|
10
|
-
|
11
|
-
|
12
|
-
module XMLScan
|
13
|
-
|
14
|
-
class EncodingError < Error ; end
|
15
|
-
|
16
|
-
|
17
|
-
class Converter
|
18
|
-
|
19
|
-
def initialize
|
20
|
-
end
|
21
|
-
|
22
|
-
def convert(s)
|
23
|
-
s
|
24
|
-
end
|
25
|
-
|
26
|
-
def finish
|
27
|
-
''
|
28
|
-
end
|
29
|
-
|
30
|
-
end
|
31
|
-
|
32
|
-
|
33
|
-
|
34
|
-
class SimpleConverter < Converter
|
35
|
-
|
36
|
-
def SimpleConverter.new_class(block)
|
37
|
-
Class.new(self).module_eval {
|
38
|
-
define_method(:convert, block)
|
39
|
-
self
|
40
|
-
}
|
41
|
-
end
|
42
|
-
|
43
|
-
# checking for Module#define_method works
|
44
|
-
begin
|
45
|
-
Class.new.module_eval{define_method(:a){};self}.new.a
|
46
|
-
rescue Exception
|
47
|
-
class << SimpleConverter
|
48
|
-
remove_method :new_class
|
49
|
-
end
|
50
|
-
def SimpleConverter.new_class(block)
|
51
|
-
Class.new(self).module_eval {
|
52
|
-
const_set :ConvProc, block
|
53
|
-
module_eval "def convert(s) ; ConvProc.call s ; end"
|
54
|
-
self
|
55
|
-
}
|
56
|
-
end
|
57
|
-
end
|
58
|
-
|
59
|
-
end
|
60
|
-
|
61
|
-
|
62
|
-
|
63
|
-
class EncodingClass
|
64
|
-
|
65
|
-
KCODE_None = //n.kcode
|
66
|
-
|
67
|
-
|
68
|
-
class ConverterProperty
|
69
|
-
|
70
|
-
def inspect
|
71
|
-
"#<Conversion #{@from.name}:#{@to.name} #{@cost}>"
|
72
|
-
end
|
73
|
-
|
74
|
-
def initialize(from, to, cost, klass = nil)
|
75
|
-
@from, @to, @cost, @klass = from, to, cost, klass
|
76
|
-
end
|
77
|
-
|
78
|
-
def new_converter
|
79
|
-
@klass and @klass.new
|
80
|
-
end
|
81
|
-
|
82
|
-
attr_reader :from, :to, :cost
|
83
|
-
|
84
|
-
end
|
85
|
-
|
86
|
-
|
87
|
-
class EncodingProperty
|
88
|
-
|
89
|
-
def inspect
|
90
|
-
s = "#<Encoding #{@name}/#{@kcode}>"
|
91
|
-
end
|
92
|
-
|
93
|
-
def initialize(name)
|
94
|
-
@name = name
|
95
|
-
conv = ConverterProperty.new(self, self, 0)
|
96
|
-
@converter = { self => conv }
|
97
|
-
@convertable_from = { self => true }
|
98
|
-
@kcode_map = {}
|
99
|
-
end
|
100
|
-
|
101
|
-
attr_reader :name, :kcode_map
|
102
|
-
|
103
|
-
|
104
|
-
def convertable_from(encoding)
|
105
|
-
@convertable_from[encoding] = true
|
106
|
-
end
|
107
|
-
protected :convertable_from
|
108
|
-
|
109
|
-
def changed
|
110
|
-
@convertable_from.each_key { |i| i.update_kcode_map }
|
111
|
-
end
|
112
|
-
private :changed
|
113
|
-
|
114
|
-
|
115
|
-
def kcode?
|
116
|
-
defined? @kcode
|
117
|
-
end
|
118
|
-
|
119
|
-
def kcode
|
120
|
-
if defined? @kcode then
|
121
|
-
@kcode
|
122
|
-
else
|
123
|
-
KCODE_None
|
124
|
-
end
|
125
|
-
end
|
126
|
-
|
127
|
-
def kcode=(kcode)
|
128
|
-
if defined? @kcode then
|
129
|
-
raise EncodingError, "KCODE conflict" unless @kcode == kcode
|
130
|
-
else
|
131
|
-
@kcode = kcode
|
132
|
-
changed
|
133
|
-
end
|
134
|
-
kcode
|
135
|
-
end
|
136
|
-
|
137
|
-
|
138
|
-
def converter(to)
|
139
|
-
@converter[to]
|
140
|
-
end
|
141
|
-
|
142
|
-
def add_converter(to, cost, conv_class)
|
143
|
-
if equal? to then
|
144
|
-
raise EncodingError,"attempt to add a converter to the same encoding"
|
145
|
-
end
|
146
|
-
oldconv = @converter[to]
|
147
|
-
if not oldconv or cost <= oldconv.cost then
|
148
|
-
conv = ConverterProperty.new(self, to, cost, conv_class)
|
149
|
-
@converter[to] = conv
|
150
|
-
to.convertable_from self
|
151
|
-
changed
|
152
|
-
end
|
153
|
-
nil
|
154
|
-
end
|
155
|
-
|
156
|
-
|
157
|
-
def update_kcode_map
|
158
|
-
@kcode_map.clear
|
159
|
-
@converter.each_value { |conv|
|
160
|
-
k = conv.to.kcode
|
161
|
-
if conv.to.kcode? and k then
|
162
|
-
oldconv = @kcode_map[k]
|
163
|
-
@kcode_map[k] = conv if not oldconv or conv.cost <= oldconv.cost
|
164
|
-
end
|
165
|
-
}
|
166
|
-
end
|
167
|
-
protected :update_kcode_map
|
168
|
-
|
169
|
-
end
|
170
|
-
|
171
|
-
|
172
|
-
|
173
|
-
def initialize
|
174
|
-
@encoding = {}
|
175
|
-
end
|
176
|
-
|
177
|
-
class << self
|
178
|
-
private :new
|
179
|
-
attr_reader :instance
|
180
|
-
end
|
181
|
-
@instance = new
|
182
|
-
|
183
|
-
|
184
|
-
private
|
185
|
-
|
186
|
-
def get_encoding(name)
|
187
|
-
encoding = @encoding[name.downcase]
|
188
|
-
raise EncodingError, "undeclared encoding `#{name}'" unless encoding
|
189
|
-
encoding
|
190
|
-
end
|
191
|
-
|
192
|
-
def touch_encoding(name)
|
193
|
-
name = name.downcase
|
194
|
-
encoding = @encoding[name]
|
195
|
-
encoding = @encoding[name] = EncodingProperty.new(name) unless encoding
|
196
|
-
encoding
|
197
|
-
end
|
198
|
-
|
199
|
-
|
200
|
-
public
|
201
|
-
|
202
|
-
def alias(newname, oldname)
|
203
|
-
newname = newname.downcase
|
204
|
-
if @encoding.key? newname then
|
205
|
-
raise EncodingError, "encoding `#{newname}' is already declared"
|
206
|
-
end
|
207
|
-
@encoding[newname] = get_encoding(oldname)
|
208
|
-
nil
|
209
|
-
end
|
210
|
-
|
211
|
-
|
212
|
-
def kcode(name)
|
213
|
-
encoding = @encoding[name.downcase]
|
214
|
-
if encoding then
|
215
|
-
encoding.kcode
|
216
|
-
else
|
217
|
-
KCODE_None
|
218
|
-
end
|
219
|
-
end
|
220
|
-
|
221
|
-
|
222
|
-
def set_kcode(name, kcode)
|
223
|
-
if kcode then
|
224
|
-
kcode = Regexp.new('', nil, kcode).kcode
|
225
|
-
else
|
226
|
-
kcode = nil
|
227
|
-
end
|
228
|
-
touch_encoding(name).kcode = kcode
|
229
|
-
end
|
230
|
-
|
231
|
-
|
232
|
-
def add_converter(from, to, cost, conv_class = nil, &block)
|
233
|
-
if block and conv_class then
|
234
|
-
raise ArgumentError, "multiple converters given"
|
235
|
-
elsif not block and not conv_class then
|
236
|
-
raise ArgumentError, "no converter given"
|
237
|
-
else
|
238
|
-
block = conv_class if Proc === conv_class
|
239
|
-
conv_class = SimpleConverter.new_class(block) if block
|
240
|
-
end
|
241
|
-
from = touch_encoding(from)
|
242
|
-
to = touch_encoding(to)
|
243
|
-
from.add_converter to, cost, conv_class
|
244
|
-
end
|
245
|
-
|
246
|
-
|
247
|
-
def converter(from, to)
|
248
|
-
fromenc = get_encoding(from)
|
249
|
-
toenc = get_encoding(to)
|
250
|
-
conv = fromenc.converter(toenc)
|
251
|
-
raise EncodingError, "can't convert `#{from}' to `#{to}'" unless conv
|
252
|
-
conv.new_converter
|
253
|
-
end
|
254
|
-
|
255
|
-
|
256
|
-
def converter3(from, to = nil)
|
257
|
-
to = from unless to
|
258
|
-
fromenc = get_encoding(from)
|
259
|
-
toenc = get_encoding(to)
|
260
|
-
kcode_map = fromenc.kcode_map
|
261
|
-
if kcode_map.empty? then
|
262
|
-
if fromenc.kcode and fromenc.equal? toenc then
|
263
|
-
return [ nil, fromenc.kcode, nil ]
|
264
|
-
else
|
265
|
-
raise EncodingError, "can't convert `#{from}' to any KCODE"
|
266
|
-
end
|
267
|
-
end
|
268
|
-
mincost, minkcode, minconv = nil
|
269
|
-
kcode_map.each { |kcode,conv|
|
270
|
-
conv2 = conv.to.converter(toenc)
|
271
|
-
if conv2 then
|
272
|
-
cost = conv.cost + conv2.cost
|
273
|
-
if not mincost or cost < mincost then
|
274
|
-
mincost, minkcode, minconv = cost, kcode, conv
|
275
|
-
end
|
276
|
-
end
|
277
|
-
}
|
278
|
-
unless mincost then
|
279
|
-
raise EncodingError, "can't convert `#{from}' to `#{to}' via any KCODE"
|
280
|
-
end
|
281
|
-
conv = minconv.new_converter
|
282
|
-
conv2 = minconv.to.converter(toenc)
|
283
|
-
conv2 = conv2 && conv2.new_converter
|
284
|
-
[ conv, minkcode, conv2 ]
|
285
|
-
end
|
286
|
-
|
287
|
-
end
|
288
|
-
|
289
|
-
|
290
|
-
Encoding = EncodingClass.instance
|
291
|
-
|
292
|
-
Encoding.set_kcode 'utf-8', 'U'
|
293
|
-
Encoding.set_kcode 'utf-16', nil
|
294
|
-
Encoding.alias 'iso-10646-ucs-2', 'utf-16'
|
295
|
-
Encoding.set_kcode 'iso-10646-ucs-4', nil
|
296
|
-
Encoding.set_kcode 'iso-8859-1', 'N'
|
297
|
-
Encoding.set_kcode 'iso-8859-2', 'N'
|
298
|
-
Encoding.set_kcode 'iso-8859-3', 'N'
|
299
|
-
Encoding.set_kcode 'iso-8859-4', 'N'
|
300
|
-
Encoding.set_kcode 'iso-8859-5', 'N'
|
301
|
-
Encoding.set_kcode 'iso-8859-6', 'N'
|
302
|
-
Encoding.set_kcode 'iso-8859-7', 'N'
|
303
|
-
Encoding.set_kcode 'iso-8859-8', 'N'
|
304
|
-
Encoding.set_kcode 'iso-8859-9', 'N'
|
305
|
-
Encoding.set_kcode 'iso-2022-jp', nil
|
306
|
-
Encoding.set_kcode 'shift_jis', 'S'
|
307
|
-
Encoding.set_kcode 'Windows-31J', 'S'
|
308
|
-
Encoding.set_kcode 'euc-jp', 'E'
|
309
|
-
Encoding.set_kcode 'euc-kr', 'E'
|
310
|
-
|
311
|
-
end
|
@@ -1,289 +0,0 @@
|
|
1
|
-
#
|
2
|
-
# xmlscan/htmlscan.rb
|
3
|
-
#
|
4
|
-
# Copyright (C) Ueno Katsuhiro 2002
|
5
|
-
#
|
6
|
-
# $Id: htmlscan.rb,v 1.18 2003/05/01 15:36:50 katsu Exp $
|
7
|
-
#
|
8
|
-
|
9
|
-
require 'xmlscan/scanner'
|
10
|
-
|
11
|
-
|
12
|
-
module XMLScan
|
13
|
-
|
14
|
-
class HTMLScanner < XMLScanner
|
15
|
-
|
16
|
-
private
|
17
|
-
|
18
|
-
def wellformed_error(msg)
|
19
|
-
# All wellformed error raised by XMLScanner are ignored.
|
20
|
-
# XMLScanner only raises wellformed error in stan_stag, which is a
|
21
|
-
# method completely overrided by HTMLScanner, so this method is
|
22
|
-
# never called in fact.
|
23
|
-
end
|
24
|
-
|
25
|
-
def on_xmldecl
|
26
|
-
raise "[BUG] this method must be never called"
|
27
|
-
end
|
28
|
-
|
29
|
-
def on_xmldecl_version(str)
|
30
|
-
raise "[BUG] this method must be never called"
|
31
|
-
end
|
32
|
-
|
33
|
-
def on_xmldecl_encoding(str)
|
34
|
-
raise "[BUG] this method must be never called"
|
35
|
-
end
|
36
|
-
|
37
|
-
def on_xmldecl_standalone(str)
|
38
|
-
raise "[BUG] this method must be never called"
|
39
|
-
end
|
40
|
-
|
41
|
-
def on_xmldecl_other(name, value)
|
42
|
-
raise "[BUG] this method must be never called"
|
43
|
-
end
|
44
|
-
|
45
|
-
def on_xmldecl_end
|
46
|
-
raise "[BUG] this method must be never called"
|
47
|
-
end
|
48
|
-
|
49
|
-
def on_stag_end_empty(name)
|
50
|
-
raise "[BUG] this method must be never called"
|
51
|
-
end
|
52
|
-
|
53
|
-
|
54
|
-
private
|
55
|
-
|
56
|
-
def scan_comment(s)
|
57
|
-
s[0,4] = '' # remove `<!--'
|
58
|
-
comm = ''
|
59
|
-
until /--/n =~ s
|
60
|
-
comm << s
|
61
|
-
s = @src.get_plain
|
62
|
-
unless s then
|
63
|
-
parse_error "unterminated comment meets EOF"
|
64
|
-
return on_comment(comm)
|
65
|
-
end
|
66
|
-
end
|
67
|
-
comm << $`
|
68
|
-
s = $'
|
69
|
-
until s.empty? || s.strip.empty? and @src.close_tag # --> or -- >
|
70
|
-
comm << '--'
|
71
|
-
if /\A\s*--/n =~ s then # <!--hoge-- --
|
72
|
-
comm << $&
|
73
|
-
s = $'
|
74
|
-
if s.empty? and @src.close_tag then # <!--hoge-- -->
|
75
|
-
parse_error "`-->' is found but comment must not end here"
|
76
|
-
comm.chop!.chop!
|
77
|
-
break
|
78
|
-
end
|
79
|
-
else # <!--hoge-- fuga
|
80
|
-
parse_error "only whitespace can appear between two comments"
|
81
|
-
end
|
82
|
-
if /\A-\s*\z/n =~ s and @src.close_tag then # <!--hoge--->
|
83
|
-
parse_error "`-->' is found but comment must not end here"
|
84
|
-
comm.chop!
|
85
|
-
break
|
86
|
-
end
|
87
|
-
until /--/n =~ s # copy & paste for performance
|
88
|
-
comm << s
|
89
|
-
s = @src.get_plain
|
90
|
-
unless s then
|
91
|
-
parse_error "unterminated comment meets EOF"
|
92
|
-
return on_comment(comm)
|
93
|
-
end
|
94
|
-
end
|
95
|
-
comm << $`
|
96
|
-
s = $'
|
97
|
-
end
|
98
|
-
on_comment comm
|
99
|
-
end
|
100
|
-
|
101
|
-
|
102
|
-
alias scan_xml_pi scan_pi # PIO "<?" PIC "?>" -- <? PI ?> --
|
103
|
-
|
104
|
-
|
105
|
-
def scan_pi(s) # <?PI > this is default in SGML.
|
106
|
-
s[0,2] = '' # remove `<?'
|
107
|
-
pi = s
|
108
|
-
until @src.close_tag
|
109
|
-
s = @src.get_plain
|
110
|
-
unless s then
|
111
|
-
parse_error "unterminated PI meets EOF"
|
112
|
-
break
|
113
|
-
end
|
114
|
-
pi << s
|
115
|
-
end
|
116
|
-
on_pi '', pi
|
117
|
-
end
|
118
|
-
|
119
|
-
|
120
|
-
def scan_stag(s)
|
121
|
-
unless /(?=[\/\s='"])/n =~ s then
|
122
|
-
name = s
|
123
|
-
name[0,1] = '' # remove `<'
|
124
|
-
if name.empty? then # <> or <<
|
125
|
-
if @src.close_tag then
|
126
|
-
return found_empty_stag
|
127
|
-
else
|
128
|
-
parse_error "parse error at `<'"
|
129
|
-
return on_chardata('<')
|
130
|
-
end
|
131
|
-
end
|
132
|
-
on_stag name
|
133
|
-
found_unclosed_stag name unless @src.close_tag
|
134
|
-
on_stag_end name
|
135
|
-
else
|
136
|
-
name = $`
|
137
|
-
s = $'
|
138
|
-
name[0,1] = '' # remove `<'
|
139
|
-
if name.empty? then # `< tag' or `<=`
|
140
|
-
parse_error "parse error at `<'"
|
141
|
-
if @src.close_tag then
|
142
|
-
s << '>'
|
143
|
-
end
|
144
|
-
return on_chardata('<' << s)
|
145
|
-
end
|
146
|
-
on_stag name
|
147
|
-
begin
|
148
|
-
continue = false
|
149
|
-
s.scan(
|
150
|
-
/([^\s=\/'"]+)(?:\s*=\s*(?:('[^']*'?|"[^"]*"?)|([^\s='"]+)))?|(\S)/n
|
151
|
-
) { |key,val,val2,error|
|
152
|
-
if key then
|
153
|
-
if val then # key="value"
|
154
|
-
on_attribute key
|
155
|
-
qmark = val.slice!(0,1)
|
156
|
-
if val[-1] == qmark[0] then
|
157
|
-
val.chop!
|
158
|
-
scan_attvalue val unless val.empty?
|
159
|
-
else
|
160
|
-
scan_attvalue val unless val.empty?
|
161
|
-
begin
|
162
|
-
s = @src.get
|
163
|
-
unless s then
|
164
|
-
parse_error "unterminated attribute `#{key}' meets EOF"
|
165
|
-
break
|
166
|
-
end
|
167
|
-
c = s[0]
|
168
|
-
val, s = s.split(qmark, 2)
|
169
|
-
scan_attvalue '>' unless c == ?< or c == ?>
|
170
|
-
scan_attvalue val if c
|
171
|
-
end until s
|
172
|
-
continue = s
|
173
|
-
end
|
174
|
-
on_attribute_end key
|
175
|
-
elsif val2 then # key=value
|
176
|
-
on_attribute key
|
177
|
-
on_attr_value val2
|
178
|
-
on_attribute_end key
|
179
|
-
else # value
|
180
|
-
on_attribute nil
|
181
|
-
on_attr_value key
|
182
|
-
on_attribute_end nil
|
183
|
-
end
|
184
|
-
else
|
185
|
-
parse_error "parse error at `#{error}'"
|
186
|
-
end
|
187
|
-
}
|
188
|
-
end while continue
|
189
|
-
found_unclosed_stag name unless @src.close_tag
|
190
|
-
on_stag_end name
|
191
|
-
end
|
192
|
-
end
|
193
|
-
|
194
|
-
|
195
|
-
# This method should be called only from on_stag_end.
|
196
|
-
def get_cdata_content
|
197
|
-
unless not s = @src.test or s[0] == ?< && s[1] == ?/ then
|
198
|
-
dst = @src.get
|
199
|
-
until not s = @src.test or s[0] == ?< && s[1] == ?/
|
200
|
-
dst << @src.get_plain
|
201
|
-
end
|
202
|
-
dst
|
203
|
-
else
|
204
|
-
''
|
205
|
-
end
|
206
|
-
end
|
207
|
-
public :get_cdata_content
|
208
|
-
|
209
|
-
|
210
|
-
def scan_bang_tag(s)
|
211
|
-
if s == '<!' and @src.close_tag then # <!>
|
212
|
-
on_comment ''
|
213
|
-
else
|
214
|
-
parse_error "parse error at `<!'"
|
215
|
-
while s and not @src.close_tag # skip entire
|
216
|
-
s = @src.get_plain
|
217
|
-
end
|
218
|
-
end
|
219
|
-
end
|
220
|
-
|
221
|
-
|
222
|
-
def scan_internal_dtd(s)
|
223
|
-
parse_error "DTD subset is found but it is not permitted in HTML"
|
224
|
-
skip_internal_dtd s
|
225
|
-
end
|
226
|
-
|
227
|
-
|
228
|
-
def found_invalid_pubsys(pubsys)
|
229
|
-
s = pubsys.upcase
|
230
|
-
return s if s == 'PUBLIC' or s == 'SYSTEM'
|
231
|
-
super
|
232
|
-
end
|
233
|
-
|
234
|
-
|
235
|
-
def scan_prolog(s)
|
236
|
-
doctype = 0
|
237
|
-
while s
|
238
|
-
if s[0] == ?< then
|
239
|
-
if (c = s[1]) == ?! then
|
240
|
-
if s[2] == ?- and s[3] == ?- then
|
241
|
-
scan_comment s
|
242
|
-
elsif /\A<!doctype(?=\s)/in =~ s then
|
243
|
-
doctype += 1
|
244
|
-
if doctype > 1 then
|
245
|
-
parse_error "another document type declaration is found"
|
246
|
-
end
|
247
|
-
scan_doctype $'
|
248
|
-
else
|
249
|
-
break
|
250
|
-
end
|
251
|
-
elsif c == ?? then
|
252
|
-
scan_pi s
|
253
|
-
else
|
254
|
-
break
|
255
|
-
end
|
256
|
-
elsif s.strip.empty? then
|
257
|
-
on_prolog_space s
|
258
|
-
else
|
259
|
-
break
|
260
|
-
end
|
261
|
-
s = @src.get
|
262
|
-
end
|
263
|
-
scan_content(s || @src.get)
|
264
|
-
end
|
265
|
-
|
266
|
-
end
|
267
|
-
|
268
|
-
end
|
269
|
-
|
270
|
-
|
271
|
-
|
272
|
-
|
273
|
-
|
274
|
-
if $0 == __FILE__ then
|
275
|
-
class TestVisitor
|
276
|
-
include XMLScan::Visitor
|
277
|
-
def parse_error(msg)
|
278
|
-
STDERR.printf("%s:%d: %s\n", $s.path, $s.lineno, msg) if $VERBOSE
|
279
|
-
end
|
280
|
-
end
|
281
|
-
|
282
|
-
$s = scan = XMLScan::HTMLScanner.new(TestVisitor.new)
|
283
|
-
src = ARGF
|
284
|
-
def src.path; filename; end
|
285
|
-
t1 = Time.times.utime
|
286
|
-
scan.parse src
|
287
|
-
t2 = Time.times.utime
|
288
|
-
STDERR.printf "%2.3f sec\n", t2 - t1
|
289
|
-
end
|