feedtools 0.2.26 → 0.2.27
Sign up to get free protection for your applications and to get access to all the features.
- data/CHANGELOG +232 -216
- data/db/migration.rb +2 -0
- data/db/schema.mysql.sql +2 -0
- data/db/schema.postgresql.sql +3 -1
- data/db/schema.sqlite.sql +3 -1
- data/lib/feed_tools.rb +37 -14
- data/lib/feed_tools/database_feed_cache.rb +13 -2
- data/lib/feed_tools/feed.rb +430 -104
- data/lib/feed_tools/feed_item.rb +533 -268
- data/lib/feed_tools/helpers/generic_helper.rb +1 -1
- data/lib/feed_tools/helpers/html_helper.rb +78 -116
- data/lib/feed_tools/helpers/retrieval_helper.rb +33 -3
- data/lib/feed_tools/helpers/uri_helper.rb +46 -54
- data/lib/feed_tools/monkey_patch.rb +27 -1
- data/lib/feed_tools/vendor/html5/History.txt +10 -0
- data/lib/feed_tools/vendor/html5/Manifest.txt +117 -0
- data/lib/feed_tools/vendor/html5/README +45 -0
- data/lib/feed_tools/vendor/html5/Rakefile.rb +33 -0
- data/lib/feed_tools/vendor/html5/bin/html5 +217 -0
- data/lib/feed_tools/vendor/html5/lib/core_ext/string.rb +17 -0
- data/lib/feed_tools/vendor/html5/lib/html5.rb +13 -0
- data/lib/feed_tools/vendor/html5/lib/html5/constants.rb +1046 -0
- data/lib/feed_tools/vendor/html5/lib/html5/filters/base.rb +10 -0
- data/lib/feed_tools/vendor/html5/lib/html5/filters/inject_meta_charset.rb +82 -0
- data/lib/feed_tools/vendor/html5/lib/html5/filters/iso639codes.rb +752 -0
- data/lib/feed_tools/vendor/html5/lib/html5/filters/optionaltags.rb +198 -0
- data/lib/feed_tools/vendor/html5/lib/html5/filters/rfc2046.rb +30 -0
- data/lib/feed_tools/vendor/html5/lib/html5/filters/rfc3987.rb +89 -0
- data/lib/feed_tools/vendor/html5/lib/html5/filters/sanitizer.rb +15 -0
- data/lib/feed_tools/vendor/html5/lib/html5/filters/validator.rb +830 -0
- data/lib/feed_tools/vendor/html5/lib/html5/filters/whitespace.rb +36 -0
- data/lib/feed_tools/vendor/html5/lib/html5/html5parser.rb +248 -0
- data/lib/feed_tools/vendor/html5/lib/html5/html5parser/after_body_phase.rb +46 -0
- data/lib/feed_tools/vendor/html5/lib/html5/html5parser/after_frameset_phase.rb +33 -0
- data/lib/feed_tools/vendor/html5/lib/html5/html5parser/after_head_phase.rb +50 -0
- data/lib/feed_tools/vendor/html5/lib/html5/html5parser/before_head_phase.rb +41 -0
- data/lib/feed_tools/vendor/html5/lib/html5/html5parser/in_body_phase.rb +613 -0
- data/lib/feed_tools/vendor/html5/lib/html5/html5parser/in_caption_phase.rb +69 -0
- data/lib/feed_tools/vendor/html5/lib/html5/html5parser/in_cell_phase.rb +78 -0
- data/lib/feed_tools/vendor/html5/lib/html5/html5parser/in_column_group_phase.rb +55 -0
- data/lib/feed_tools/vendor/html5/lib/html5/html5parser/in_frameset_phase.rb +57 -0
- data/lib/feed_tools/vendor/html5/lib/html5/html5parser/in_head_phase.rb +138 -0
- data/lib/feed_tools/vendor/html5/lib/html5/html5parser/in_row_phase.rb +89 -0
- data/lib/feed_tools/vendor/html5/lib/html5/html5parser/in_select_phase.rb +85 -0
- data/lib/feed_tools/vendor/html5/lib/html5/html5parser/in_table_body_phase.rb +86 -0
- data/lib/feed_tools/vendor/html5/lib/html5/html5parser/in_table_phase.rb +115 -0
- data/lib/feed_tools/vendor/html5/lib/html5/html5parser/initial_phase.rb +133 -0
- data/lib/feed_tools/vendor/html5/lib/html5/html5parser/phase.rb +154 -0
- data/lib/feed_tools/vendor/html5/lib/html5/html5parser/root_element_phase.rb +41 -0
- data/lib/feed_tools/vendor/html5/lib/html5/html5parser/trailing_end_phase.rb +35 -0
- data/lib/feed_tools/vendor/html5/lib/html5/inputstream.rb +648 -0
- data/lib/feed_tools/vendor/html5/lib/html5/liberalxmlparser.rb +158 -0
- data/lib/feed_tools/vendor/html5/lib/html5/sanitizer.rb +188 -0
- data/lib/feed_tools/vendor/html5/lib/html5/serializer.rb +2 -0
- data/lib/feed_tools/vendor/html5/lib/html5/serializer/htmlserializer.rb +179 -0
- data/lib/feed_tools/vendor/html5/lib/html5/serializer/xhtmlserializer.rb +20 -0
- data/lib/feed_tools/vendor/html5/lib/html5/sniffer.rb +45 -0
- data/lib/feed_tools/vendor/html5/lib/html5/tokenizer.rb +966 -0
- data/lib/feed_tools/vendor/html5/lib/html5/treebuilders.rb +24 -0
- data/lib/feed_tools/vendor/html5/lib/html5/treebuilders/base.rb +334 -0
- data/lib/feed_tools/vendor/html5/lib/html5/treebuilders/hpricot.rb +231 -0
- data/lib/feed_tools/vendor/html5/lib/html5/treebuilders/rexml.rb +209 -0
- data/lib/feed_tools/vendor/html5/lib/html5/treebuilders/simpletree.rb +185 -0
- data/lib/feed_tools/vendor/html5/lib/html5/treewalkers.rb +26 -0
- data/lib/feed_tools/vendor/html5/lib/html5/treewalkers/base.rb +162 -0
- data/lib/feed_tools/vendor/html5/lib/html5/treewalkers/hpricot.rb +48 -0
- data/lib/feed_tools/vendor/html5/lib/html5/treewalkers/rexml.rb +48 -0
- data/lib/feed_tools/vendor/html5/lib/html5/treewalkers/simpletree.rb +48 -0
- data/lib/feed_tools/vendor/html5/lib/html5/version.rb +3 -0
- data/lib/feed_tools/vendor/html5/testdata/encoding/chardet/test_big5.txt +51 -0
- data/lib/feed_tools/vendor/html5/testdata/encoding/test-yahoo-jp.dat +10 -0
- data/lib/feed_tools/vendor/html5/testdata/encoding/tests1.dat +394 -0
- data/lib/feed_tools/vendor/html5/testdata/encoding/tests2.dat +81 -0
- data/lib/feed_tools/vendor/html5/testdata/sanitizer/tests1.dat +416 -0
- data/lib/feed_tools/vendor/html5/testdata/serializer/core.test +104 -0
- data/lib/feed_tools/vendor/html5/testdata/serializer/injectmeta.test +65 -0
- data/lib/feed_tools/vendor/html5/testdata/serializer/optionaltags.test +900 -0
- data/lib/feed_tools/vendor/html5/testdata/serializer/options.test +60 -0
- data/lib/feed_tools/vendor/html5/testdata/serializer/whitespace.test +51 -0
- data/lib/feed_tools/vendor/html5/testdata/sites/google-results.htm +1 -0
- data/lib/feed_tools/vendor/html5/testdata/sites/python-ref-import.htm +1 -0
- data/lib/feed_tools/vendor/html5/testdata/sites/web-apps-old.htm +1 -0
- data/lib/feed_tools/vendor/html5/testdata/sites/web-apps.htm +34275 -0
- data/lib/feed_tools/vendor/html5/testdata/sniffer/htmlOrFeed.json +43 -0
- data/lib/feed_tools/vendor/html5/testdata/tokenizer/contentModelFlags.test +48 -0
- data/lib/feed_tools/vendor/html5/testdata/tokenizer/entities.test +2339 -0
- data/lib/feed_tools/vendor/html5/testdata/tokenizer/escapeFlag.test +21 -0
- data/lib/feed_tools/vendor/html5/testdata/tokenizer/test1.test +172 -0
- data/lib/feed_tools/vendor/html5/testdata/tokenizer/test2.test +129 -0
- data/lib/feed_tools/vendor/html5/testdata/tokenizer/test3.test +367 -0
- data/lib/feed_tools/vendor/html5/testdata/tokenizer/test4.test +198 -0
- data/lib/feed_tools/vendor/html5/testdata/tree-construction/tests1.dat +1950 -0
- data/lib/feed_tools/vendor/html5/testdata/tree-construction/tests2.dat +773 -0
- data/lib/feed_tools/vendor/html5/testdata/tree-construction/tests3.dat +270 -0
- data/lib/feed_tools/vendor/html5/testdata/tree-construction/tests4.dat +60 -0
- data/lib/feed_tools/vendor/html5/testdata/tree-construction/tests5.dat +175 -0
- data/lib/feed_tools/vendor/html5/testdata/tree-construction/tests6.dat +196 -0
- data/lib/feed_tools/vendor/html5/testdata/validator/attributes.test +1035 -0
- data/lib/feed_tools/vendor/html5/testdata/validator/base-href-attribute.test +787 -0
- data/lib/feed_tools/vendor/html5/testdata/validator/base-target-attribute.test +35 -0
- data/lib/feed_tools/vendor/html5/testdata/validator/blockquote-cite-attribute.test +7 -0
- data/lib/feed_tools/vendor/html5/testdata/validator/classattribute.test +152 -0
- data/lib/feed_tools/vendor/html5/testdata/validator/contenteditableattribute.test +59 -0
- data/lib/feed_tools/vendor/html5/testdata/validator/contextmenuattribute.test +115 -0
- data/lib/feed_tools/vendor/html5/testdata/validator/dirattribute.test +59 -0
- data/lib/feed_tools/vendor/html5/testdata/validator/draggableattribute.test +63 -0
- data/lib/feed_tools/vendor/html5/testdata/validator/html-xmlns-attribute.test +23 -0
- data/lib/feed_tools/vendor/html5/testdata/validator/idattribute.test +115 -0
- data/lib/feed_tools/vendor/html5/testdata/validator/inputattributes.test +2795 -0
- data/lib/feed_tools/vendor/html5/testdata/validator/irrelevantattribute.test +63 -0
- data/lib/feed_tools/vendor/html5/testdata/validator/langattribute.test +5579 -0
- data/lib/feed_tools/vendor/html5/testdata/validator/li-value-attribute.test +7 -0
- data/lib/feed_tools/vendor/html5/testdata/validator/link-href-attribute.test +7 -0
- data/lib/feed_tools/vendor/html5/testdata/validator/link-hreflang-attribute.test +7 -0
- data/lib/feed_tools/vendor/html5/testdata/validator/link-rel-attribute.test +271 -0
- data/lib/feed_tools/vendor/html5/testdata/validator/ol-start-attribute.test +7 -0
- data/lib/feed_tools/vendor/html5/testdata/validator/starttags.test +375 -0
- data/lib/feed_tools/vendor/html5/testdata/validator/style-scoped-attribute.test +7 -0
- data/lib/feed_tools/vendor/html5/testdata/validator/tabindexattribute.test +79 -0
- data/lib/feed_tools/vendor/html5/tests/preamble.rb +72 -0
- data/lib/feed_tools/vendor/html5/tests/test_encoding.rb +35 -0
- data/lib/feed_tools/vendor/html5/tests/test_lxp.rb +279 -0
- data/lib/feed_tools/vendor/html5/tests/test_parser.rb +68 -0
- data/lib/feed_tools/vendor/html5/tests/test_sanitizer.rb +142 -0
- data/lib/feed_tools/vendor/html5/tests/test_serializer.rb +68 -0
- data/lib/feed_tools/vendor/html5/tests/test_sniffer.rb +27 -0
- data/lib/feed_tools/vendor/html5/tests/test_stream.rb +62 -0
- data/lib/feed_tools/vendor/html5/tests/test_tokenizer.rb +94 -0
- data/lib/feed_tools/vendor/html5/tests/test_treewalkers.rb +135 -0
- data/lib/feed_tools/vendor/html5/tests/test_validator.rb +31 -0
- data/lib/feed_tools/vendor/html5/tests/tokenizer_test_parser.rb +63 -0
- data/lib/feed_tools/vendor/uri.rb +781 -0
- data/lib/feed_tools/version.rb +1 -1
- data/rakefile +27 -6
- data/test/unit/atom_test.rb +298 -210
- data/test/unit/helper_test.rb +7 -12
- data/test/unit/rdf_test.rb +51 -1
- data/test/unit/rss_test.rb +13 -3
- metadata +239 -116
- data/lib/feed_tools/vendor/htree.rb +0 -97
- data/lib/feed_tools/vendor/htree/container.rb +0 -10
- data/lib/feed_tools/vendor/htree/context.rb +0 -67
- data/lib/feed_tools/vendor/htree/display.rb +0 -27
- data/lib/feed_tools/vendor/htree/doc.rb +0 -149
- data/lib/feed_tools/vendor/htree/elem.rb +0 -262
- data/lib/feed_tools/vendor/htree/encoder.rb +0 -163
- data/lib/feed_tools/vendor/htree/equality.rb +0 -218
- data/lib/feed_tools/vendor/htree/extract_text.rb +0 -37
- data/lib/feed_tools/vendor/htree/fstr.rb +0 -33
- data/lib/feed_tools/vendor/htree/gencode.rb +0 -97
- data/lib/feed_tools/vendor/htree/htmlinfo.rb +0 -672
- data/lib/feed_tools/vendor/htree/inspect.rb +0 -108
- data/lib/feed_tools/vendor/htree/leaf.rb +0 -94
- data/lib/feed_tools/vendor/htree/loc.rb +0 -367
- data/lib/feed_tools/vendor/htree/modules.rb +0 -48
- data/lib/feed_tools/vendor/htree/name.rb +0 -124
- data/lib/feed_tools/vendor/htree/output.rb +0 -207
- data/lib/feed_tools/vendor/htree/parse.rb +0 -409
- data/lib/feed_tools/vendor/htree/raw_string.rb +0 -124
- data/lib/feed_tools/vendor/htree/regexp-util.rb +0 -15
- data/lib/feed_tools/vendor/htree/rexml.rb +0 -130
- data/lib/feed_tools/vendor/htree/scan.rb +0 -166
- data/lib/feed_tools/vendor/htree/tag.rb +0 -111
- data/lib/feed_tools/vendor/htree/template.rb +0 -909
- data/lib/feed_tools/vendor/htree/text.rb +0 -115
- data/lib/feed_tools/vendor/htree/traverse.rb +0 -465
@@ -0,0 +1,17 @@
|
|
1
|
+
class String
|
2
|
+
alias old_format %
|
3
|
+
define_method("%") do |data|
|
4
|
+
unless data.kind_of?(Hash)
|
5
|
+
$VERBOSE = false
|
6
|
+
r = old_format(data)
|
7
|
+
$VERBOSE = true
|
8
|
+
r
|
9
|
+
else
|
10
|
+
ret = self.clone
|
11
|
+
data.each do |k,v|
|
12
|
+
ret.gsub!(/\%\(#{k}\)/, v)
|
13
|
+
end
|
14
|
+
ret
|
15
|
+
end
|
16
|
+
end
|
17
|
+
end
|
@@ -0,0 +1,1046 @@
|
|
1
|
+
module HTML5
|
2
|
+
|
3
|
+
class EOF < Exception; end
|
4
|
+
|
5
|
+
def self._(str); str end
|
6
|
+
|
7
|
+
CONTENT_MODEL_FLAGS = [
|
8
|
+
:PCDATA,
|
9
|
+
:RCDATA,
|
10
|
+
:CDATA,
|
11
|
+
:PLAINTEXT
|
12
|
+
]
|
13
|
+
|
14
|
+
SCOPING_ELEMENTS = %w[
|
15
|
+
button
|
16
|
+
caption
|
17
|
+
html
|
18
|
+
marquee
|
19
|
+
object
|
20
|
+
table
|
21
|
+
td
|
22
|
+
th
|
23
|
+
]
|
24
|
+
|
25
|
+
FORMATTING_ELEMENTS = %w[
|
26
|
+
a
|
27
|
+
b
|
28
|
+
big
|
29
|
+
em
|
30
|
+
font
|
31
|
+
i
|
32
|
+
nobr
|
33
|
+
s
|
34
|
+
small
|
35
|
+
strike
|
36
|
+
strong
|
37
|
+
tt
|
38
|
+
u
|
39
|
+
]
|
40
|
+
|
41
|
+
SPECIAL_ELEMENTS = %w[
|
42
|
+
address
|
43
|
+
area
|
44
|
+
base
|
45
|
+
basefont
|
46
|
+
bgsound
|
47
|
+
blockquote
|
48
|
+
body
|
49
|
+
br
|
50
|
+
center
|
51
|
+
col
|
52
|
+
colgroup
|
53
|
+
dd
|
54
|
+
dir
|
55
|
+
div
|
56
|
+
dl
|
57
|
+
dt
|
58
|
+
embed
|
59
|
+
fieldset
|
60
|
+
form
|
61
|
+
frame
|
62
|
+
frameset
|
63
|
+
h1
|
64
|
+
h2
|
65
|
+
h3
|
66
|
+
h4
|
67
|
+
h5
|
68
|
+
h6
|
69
|
+
head
|
70
|
+
hr
|
71
|
+
iframe
|
72
|
+
image
|
73
|
+
img
|
74
|
+
input
|
75
|
+
isindex
|
76
|
+
li
|
77
|
+
link
|
78
|
+
listing
|
79
|
+
menu
|
80
|
+
meta
|
81
|
+
noembed
|
82
|
+
noframes
|
83
|
+
noscript
|
84
|
+
ol
|
85
|
+
optgroup
|
86
|
+
option
|
87
|
+
p
|
88
|
+
param
|
89
|
+
plaintext
|
90
|
+
pre
|
91
|
+
script
|
92
|
+
select
|
93
|
+
spacer
|
94
|
+
style
|
95
|
+
tbody
|
96
|
+
textarea
|
97
|
+
tfoot
|
98
|
+
thead
|
99
|
+
title
|
100
|
+
tr
|
101
|
+
ul
|
102
|
+
wbr
|
103
|
+
]
|
104
|
+
|
105
|
+
SPACE_CHARACTERS = %W[
|
106
|
+
\t
|
107
|
+
\n
|
108
|
+
\x0B
|
109
|
+
\x0C
|
110
|
+
\x20
|
111
|
+
\r
|
112
|
+
]
|
113
|
+
|
114
|
+
TABLE_INSERT_MODE_ELEMENTS = %w[
|
115
|
+
table
|
116
|
+
tbody
|
117
|
+
tfoot
|
118
|
+
thead
|
119
|
+
tr
|
120
|
+
]
|
121
|
+
|
122
|
+
ASCII_LOWERCASE = ('a'..'z').to_a.join('')
|
123
|
+
ASCII_UPPERCASE = ('A'..'Z').to_a.join('')
|
124
|
+
ASCII_LETTERS = ASCII_LOWERCASE + ASCII_UPPERCASE
|
125
|
+
DIGITS = '0'..'9'
|
126
|
+
HEX_DIGITS = DIGITS.to_a + ('a'..'f').to_a + ('A'..'F').to_a
|
127
|
+
|
128
|
+
# Heading elements need to be ordered
|
129
|
+
HEADING_ELEMENTS = %w[
|
130
|
+
h1
|
131
|
+
h2
|
132
|
+
h3
|
133
|
+
h4
|
134
|
+
h5
|
135
|
+
h6
|
136
|
+
]
|
137
|
+
|
138
|
+
# XXX What about event-source and command?
|
139
|
+
VOID_ELEMENTS = %w[
|
140
|
+
base
|
141
|
+
link
|
142
|
+
meta
|
143
|
+
hr
|
144
|
+
br
|
145
|
+
img
|
146
|
+
embed
|
147
|
+
param
|
148
|
+
area
|
149
|
+
col
|
150
|
+
input
|
151
|
+
]
|
152
|
+
|
153
|
+
CDATA_ELEMENTS = %w[title textarea]
|
154
|
+
|
155
|
+
RCDATA_ELEMENTS = %w[
|
156
|
+
style
|
157
|
+
script
|
158
|
+
xmp
|
159
|
+
iframe
|
160
|
+
noembed
|
161
|
+
noframes
|
162
|
+
noscript
|
163
|
+
]
|
164
|
+
|
165
|
+
BOOLEAN_ATTRIBUTES = {
|
166
|
+
:global => %w[irrelevant],
|
167
|
+
'style' => %w[scoped],
|
168
|
+
'img' => %w[ismap],
|
169
|
+
'audio' => %w[autoplay controls],
|
170
|
+
'video' => %w[autoplay controls],
|
171
|
+
'script' => %w[defer async],
|
172
|
+
'details' => %w[open],
|
173
|
+
'datagrid' => %w[multiple disabled],
|
174
|
+
'command' => %w[hidden disabled checked default],
|
175
|
+
'menu' => %w[autosubmit],
|
176
|
+
'fieldset' => %w[disabled readonly],
|
177
|
+
'option' => %w[disabled readonly selected],
|
178
|
+
'optgroup' => %w[disabled readonly],
|
179
|
+
'button' => %w[disabled autofocus],
|
180
|
+
'input' => %w[disabled readonly required autofocus checked ismap],
|
181
|
+
'select' => %w[disabled readonly autofocus multiple],
|
182
|
+
'output' => %w[disabled readonly]
|
183
|
+
|
184
|
+
}
|
185
|
+
|
186
|
+
# entitiesWindows1252 has to be _ordered_ and needs to have an index.
|
187
|
+
ENTITIES_WINDOWS1252 = [
|
188
|
+
8364, # 0x80 0x20AC EURO SIGN
|
189
|
+
65533, # 0x81 UNDEFINED
|
190
|
+
8218, # 0x82 0x201A SINGLE LOW-9 QUOTATION MARK
|
191
|
+
402, # 0x83 0x0192 LATIN SMALL LETTER F WITH HOOK
|
192
|
+
8222, # 0x84 0x201E DOUBLE LOW-9 QUOTATION MARK
|
193
|
+
8230, # 0x85 0x2026 HORIZONTAL ELLIPSIS
|
194
|
+
8224, # 0x86 0x2020 DAGGER
|
195
|
+
8225, # 0x87 0x2021 DOUBLE DAGGER
|
196
|
+
710, # 0x88 0x02C6 MODIFIER LETTER CIRCUMFLEX ACCENT
|
197
|
+
8240, # 0x89 0x2030 PER MILLE SIGN
|
198
|
+
352, # 0x8A 0x0160 LATIN CAPITAL LETTER S WITH CARON
|
199
|
+
8249, # 0x8B 0x2039 SINGLE LEFT-POINTING ANGLE QUOTATION MARK
|
200
|
+
338, # 0x8C 0x0152 LATIN CAPITAL LIGATURE OE
|
201
|
+
65533, # 0x8D UNDEFINED
|
202
|
+
381, # 0x8E 0x017D LATIN CAPITAL LETTER Z WITH CARON
|
203
|
+
65533, # 0x8F UNDEFINED
|
204
|
+
65533, # 0x90 UNDEFINED
|
205
|
+
8216, # 0x91 0x2018 LEFT SINGLE QUOTATION MARK
|
206
|
+
8217, # 0x92 0x2019 RIGHT SINGLE QUOTATION MARK
|
207
|
+
8220, # 0x93 0x201C LEFT DOUBLE QUOTATION MARK
|
208
|
+
8221, # 0x94 0x201D RIGHT DOUBLE QUOTATION MARK
|
209
|
+
8226, # 0x95 0x2022 BULLET
|
210
|
+
8211, # 0x96 0x2013 EN DASH
|
211
|
+
8212, # 0x97 0x2014 EM DASH
|
212
|
+
732, # 0x98 0x02DC SMALL TILDE
|
213
|
+
8482, # 0x99 0x2122 TRADE MARK SIGN
|
214
|
+
353, # 0x9A 0x0161 LATIN SMALL LETTER S WITH CARON
|
215
|
+
8250, # 0x9B 0x203A SINGLE RIGHT-POINTING ANGLE QUOTATION MARK
|
216
|
+
339, # 0x9C 0x0153 LATIN SMALL LIGATURE OE
|
217
|
+
65533, # 0x9D UNDEFINED
|
218
|
+
382, # 0x9E 0x017E LATIN SMALL LETTER Z WITH CARON
|
219
|
+
376 # 0x9F 0x0178 LATIN CAPITAL LETTER Y WITH DIAERESIS
|
220
|
+
]
|
221
|
+
|
222
|
+
# ENTITIES was generated from Python using the following code:
|
223
|
+
#
|
224
|
+
# import constants
|
225
|
+
# entities = constants.entities.items()
|
226
|
+
# entities.sort()
|
227
|
+
# list = [ ' '.join([repr(entity), '=>', ord(value)<128 and
|
228
|
+
# repr(str(value)) or repr(value.encode('utf-8')).replace("'",'"')])
|
229
|
+
# for entity, value in entities]
|
230
|
+
# print ' ENTITIES = {\n ' + ',\n '.join(list) + '\n }'
|
231
|
+
|
232
|
+
ENTITIES = {
|
233
|
+
'AElig' => "\xc3\x86",
|
234
|
+
'AElig;' => "\xc3\x86",
|
235
|
+
'AMP' => '&',
|
236
|
+
'AMP;' => '&',
|
237
|
+
'Aacute' => "\xc3\x81",
|
238
|
+
'Aacute;' => "\xc3\x81",
|
239
|
+
'Acirc' => "\xc3\x82",
|
240
|
+
'Acirc;' => "\xc3\x82",
|
241
|
+
'Agrave' => "\xc3\x80",
|
242
|
+
'Agrave;' => "\xc3\x80",
|
243
|
+
'Alpha;' => "\xce\x91",
|
244
|
+
'Aring' => "\xc3\x85",
|
245
|
+
'Aring;' => "\xc3\x85",
|
246
|
+
'Atilde' => "\xc3\x83",
|
247
|
+
'Atilde;' => "\xc3\x83",
|
248
|
+
'Auml' => "\xc3\x84",
|
249
|
+
'Auml;' => "\xc3\x84",
|
250
|
+
'Beta;' => "\xce\x92",
|
251
|
+
'COPY' => "\xc2\xa9",
|
252
|
+
'COPY;' => "\xc2\xa9",
|
253
|
+
'Ccedil' => "\xc3\x87",
|
254
|
+
'Ccedil;' => "\xc3\x87",
|
255
|
+
'Chi;' => "\xce\xa7",
|
256
|
+
'Dagger;' => "\xe2\x80\xa1",
|
257
|
+
'Delta;' => "\xce\x94",
|
258
|
+
'ETH' => "\xc3\x90",
|
259
|
+
'ETH;' => "\xc3\x90",
|
260
|
+
'Eacute' => "\xc3\x89",
|
261
|
+
'Eacute;' => "\xc3\x89",
|
262
|
+
'Ecirc' => "\xc3\x8a",
|
263
|
+
'Ecirc;' => "\xc3\x8a",
|
264
|
+
'Egrave' => "\xc3\x88",
|
265
|
+
'Egrave;' => "\xc3\x88",
|
266
|
+
'Epsilon;' => "\xce\x95",
|
267
|
+
'Eta;' => "\xce\x97",
|
268
|
+
'Euml' => "\xc3\x8b",
|
269
|
+
'Euml;' => "\xc3\x8b",
|
270
|
+
'GT' => '>',
|
271
|
+
'GT;' => '>',
|
272
|
+
'Gamma;' => "\xce\x93",
|
273
|
+
'Iacute' => "\xc3\x8d",
|
274
|
+
'Iacute;' => "\xc3\x8d",
|
275
|
+
'Icirc' => "\xc3\x8e",
|
276
|
+
'Icirc;' => "\xc3\x8e",
|
277
|
+
'Igrave' => "\xc3\x8c",
|
278
|
+
'Igrave;' => "\xc3\x8c",
|
279
|
+
'Iota;' => "\xce\x99",
|
280
|
+
'Iuml' => "\xc3\x8f",
|
281
|
+
'Iuml;' => "\xc3\x8f",
|
282
|
+
'Kappa;' => "\xce\x9a",
|
283
|
+
'LT' => '<',
|
284
|
+
'LT;' => '<',
|
285
|
+
'Lambda;' => "\xce\x9b",
|
286
|
+
'Mu;' => "\xce\x9c",
|
287
|
+
'Ntilde' => "\xc3\x91",
|
288
|
+
'Ntilde;' => "\xc3\x91",
|
289
|
+
'Nu;' => "\xce\x9d",
|
290
|
+
'OElig;' => "\xc5\x92",
|
291
|
+
'Oacute' => "\xc3\x93",
|
292
|
+
'Oacute;' => "\xc3\x93",
|
293
|
+
'Ocirc' => "\xc3\x94",
|
294
|
+
'Ocirc;' => "\xc3\x94",
|
295
|
+
'Ograve' => "\xc3\x92",
|
296
|
+
'Ograve;' => "\xc3\x92",
|
297
|
+
'Omega;' => "\xce\xa9",
|
298
|
+
'Omicron;' => "\xce\x9f",
|
299
|
+
'Oslash' => "\xc3\x98",
|
300
|
+
'Oslash;' => "\xc3\x98",
|
301
|
+
'Otilde' => "\xc3\x95",
|
302
|
+
'Otilde;' => "\xc3\x95",
|
303
|
+
'Ouml' => "\xc3\x96",
|
304
|
+
'Ouml;' => "\xc3\x96",
|
305
|
+
'Phi;' => "\xce\xa6",
|
306
|
+
'Pi;' => "\xce\xa0",
|
307
|
+
'Prime;' => "\xe2\x80\xb3",
|
308
|
+
'Psi;' => "\xce\xa8",
|
309
|
+
'QUOT' => '"',
|
310
|
+
'QUOT;' => '"',
|
311
|
+
'REG' => "\xc2\xae",
|
312
|
+
'REG;' => "\xc2\xae",
|
313
|
+
'Rho;' => "\xce\xa1",
|
314
|
+
'Scaron;' => "\xc5\xa0",
|
315
|
+
'Sigma;' => "\xce\xa3",
|
316
|
+
'THORN' => "\xc3\x9e",
|
317
|
+
'THORN;' => "\xc3\x9e",
|
318
|
+
'TRADE;' => "\xe2\x84\xa2",
|
319
|
+
'Tau;' => "\xce\xa4",
|
320
|
+
'Theta;' => "\xce\x98",
|
321
|
+
'Uacute' => "\xc3\x9a",
|
322
|
+
'Uacute;' => "\xc3\x9a",
|
323
|
+
'Ucirc' => "\xc3\x9b",
|
324
|
+
'Ucirc;' => "\xc3\x9b",
|
325
|
+
'Ugrave' => "\xc3\x99",
|
326
|
+
'Ugrave;' => "\xc3\x99",
|
327
|
+
'Upsilon;' => "\xce\xa5",
|
328
|
+
'Uuml' => "\xc3\x9c",
|
329
|
+
'Uuml;' => "\xc3\x9c",
|
330
|
+
'Xi;' => "\xce\x9e",
|
331
|
+
'Yacute' => "\xc3\x9d",
|
332
|
+
'Yacute;' => "\xc3\x9d",
|
333
|
+
'Yuml;' => "\xc5\xb8",
|
334
|
+
'Zeta;' => "\xce\x96",
|
335
|
+
'aacute' => "\xc3\xa1",
|
336
|
+
'aacute;' => "\xc3\xa1",
|
337
|
+
'acirc' => "\xc3\xa2",
|
338
|
+
'acirc;' => "\xc3\xa2",
|
339
|
+
'acute' => "\xc2\xb4",
|
340
|
+
'acute;' => "\xc2\xb4",
|
341
|
+
'aelig' => "\xc3\xa6",
|
342
|
+
'aelig;' => "\xc3\xa6",
|
343
|
+
'agrave' => "\xc3\xa0",
|
344
|
+
'agrave;' => "\xc3\xa0",
|
345
|
+
'alefsym;' => "\xe2\x84\xb5",
|
346
|
+
'alpha;' => "\xce\xb1",
|
347
|
+
'amp' => '&',
|
348
|
+
'amp;' => '&',
|
349
|
+
'and;' => "\xe2\x88\xa7",
|
350
|
+
'ang;' => "\xe2\x88\xa0",
|
351
|
+
'apos;' => "'",
|
352
|
+
'aring' => "\xc3\xa5",
|
353
|
+
'aring;' => "\xc3\xa5",
|
354
|
+
'asymp;' => "\xe2\x89\x88",
|
355
|
+
'atilde' => "\xc3\xa3",
|
356
|
+
'atilde;' => "\xc3\xa3",
|
357
|
+
'auml' => "\xc3\xa4",
|
358
|
+
'auml;' => "\xc3\xa4",
|
359
|
+
'bdquo;' => "\xe2\x80\x9e",
|
360
|
+
'beta;' => "\xce\xb2",
|
361
|
+
'brvbar' => "\xc2\xa6",
|
362
|
+
'brvbar;' => "\xc2\xa6",
|
363
|
+
'bull;' => "\xe2\x80\xa2",
|
364
|
+
'cap;' => "\xe2\x88\xa9",
|
365
|
+
'ccedil' => "\xc3\xa7",
|
366
|
+
'ccedil;' => "\xc3\xa7",
|
367
|
+
'cedil' => "\xc2\xb8",
|
368
|
+
'cedil;' => "\xc2\xb8",
|
369
|
+
'cent' => "\xc2\xa2",
|
370
|
+
'cent;' => "\xc2\xa2",
|
371
|
+
'chi;' => "\xcf\x87",
|
372
|
+
'circ;' => "\xcb\x86",
|
373
|
+
'clubs;' => "\xe2\x99\xa3",
|
374
|
+
'cong;' => "\xe2\x89\x85",
|
375
|
+
'copy' => "\xc2\xa9",
|
376
|
+
'copy;' => "\xc2\xa9",
|
377
|
+
'crarr;' => "\xe2\x86\xb5",
|
378
|
+
'cup;' => "\xe2\x88\xaa",
|
379
|
+
'curren' => "\xc2\xa4",
|
380
|
+
'curren;' => "\xc2\xa4",
|
381
|
+
'dArr;' => "\xe2\x87\x93",
|
382
|
+
'dagger;' => "\xe2\x80\xa0",
|
383
|
+
'darr;' => "\xe2\x86\x93",
|
384
|
+
'deg' => "\xc2\xb0",
|
385
|
+
'deg;' => "\xc2\xb0",
|
386
|
+
'delta;' => "\xce\xb4",
|
387
|
+
'diams;' => "\xe2\x99\xa6",
|
388
|
+
'divide' => "\xc3\xb7",
|
389
|
+
'divide;' => "\xc3\xb7",
|
390
|
+
'eacute' => "\xc3\xa9",
|
391
|
+
'eacute;' => "\xc3\xa9",
|
392
|
+
'ecirc' => "\xc3\xaa",
|
393
|
+
'ecirc;' => "\xc3\xaa",
|
394
|
+
'egrave' => "\xc3\xa8",
|
395
|
+
'egrave;' => "\xc3\xa8",
|
396
|
+
'empty;' => "\xe2\x88\x85",
|
397
|
+
'emsp;' => "\xe2\x80\x83",
|
398
|
+
'ensp;' => "\xe2\x80\x82",
|
399
|
+
'epsilon;' => "\xce\xb5",
|
400
|
+
'equiv;' => "\xe2\x89\xa1",
|
401
|
+
'eta;' => "\xce\xb7",
|
402
|
+
'eth' => "\xc3\xb0",
|
403
|
+
'eth;' => "\xc3\xb0",
|
404
|
+
'euml' => "\xc3\xab",
|
405
|
+
'euml;' => "\xc3\xab",
|
406
|
+
'euro;' => "\xe2\x82\xac",
|
407
|
+
'exist;' => "\xe2\x88\x83",
|
408
|
+
'fnof;' => "\xc6\x92",
|
409
|
+
'forall;' => "\xe2\x88\x80",
|
410
|
+
'frac12' => "\xc2\xbd",
|
411
|
+
'frac12;' => "\xc2\xbd",
|
412
|
+
'frac14' => "\xc2\xbc",
|
413
|
+
'frac14;' => "\xc2\xbc",
|
414
|
+
'frac34' => "\xc2\xbe",
|
415
|
+
'frac34;' => "\xc2\xbe",
|
416
|
+
'frasl;' => "\xe2\x81\x84",
|
417
|
+
'gamma;' => "\xce\xb3",
|
418
|
+
'ge;' => "\xe2\x89\xa5",
|
419
|
+
'gt' => '>',
|
420
|
+
'gt;' => '>',
|
421
|
+
'hArr;' => "\xe2\x87\x94",
|
422
|
+
'harr;' => "\xe2\x86\x94",
|
423
|
+
'hearts;' => "\xe2\x99\xa5",
|
424
|
+
'hellip;' => "\xe2\x80\xa6",
|
425
|
+
'iacute' => "\xc3\xad",
|
426
|
+
'iacute;' => "\xc3\xad",
|
427
|
+
'icirc' => "\xc3\xae",
|
428
|
+
'icirc;' => "\xc3\xae",
|
429
|
+
'iexcl' => "\xc2\xa1",
|
430
|
+
'iexcl;' => "\xc2\xa1",
|
431
|
+
'igrave' => "\xc3\xac",
|
432
|
+
'igrave;' => "\xc3\xac",
|
433
|
+
'image;' => "\xe2\x84\x91",
|
434
|
+
'infin;' => "\xe2\x88\x9e",
|
435
|
+
'int;' => "\xe2\x88\xab",
|
436
|
+
'iota;' => "\xce\xb9",
|
437
|
+
'iquest' => "\xc2\xbf",
|
438
|
+
'iquest;' => "\xc2\xbf",
|
439
|
+
'isin;' => "\xe2\x88\x88",
|
440
|
+
'iuml' => "\xc3\xaf",
|
441
|
+
'iuml;' => "\xc3\xaf",
|
442
|
+
'kappa;' => "\xce\xba",
|
443
|
+
'lArr;' => "\xe2\x87\x90",
|
444
|
+
'lambda;' => "\xce\xbb",
|
445
|
+
'lang;' => "\xe3\x80\x88",
|
446
|
+
'laquo' => "\xc2\xab",
|
447
|
+
'laquo;' => "\xc2\xab",
|
448
|
+
'larr;' => "\xe2\x86\x90",
|
449
|
+
'lceil;' => "\xe2\x8c\x88",
|
450
|
+
'ldquo;' => "\xe2\x80\x9c",
|
451
|
+
'le;' => "\xe2\x89\xa4",
|
452
|
+
'lfloor;' => "\xe2\x8c\x8a",
|
453
|
+
'lowast;' => "\xe2\x88\x97",
|
454
|
+
'loz;' => "\xe2\x97\x8a",
|
455
|
+
'lrm;' => "\xe2\x80\x8e",
|
456
|
+
'lsaquo;' => "\xe2\x80\xb9",
|
457
|
+
'lsquo;' => "\xe2\x80\x98",
|
458
|
+
'lt' => '<',
|
459
|
+
'lt;' => '<',
|
460
|
+
'macr' => "\xc2\xaf",
|
461
|
+
'macr;' => "\xc2\xaf",
|
462
|
+
'mdash;' => "\xe2\x80\x94",
|
463
|
+
'micro' => "\xc2\xb5",
|
464
|
+
'micro;' => "\xc2\xb5",
|
465
|
+
'middot' => "\xc2\xb7",
|
466
|
+
'middot;' => "\xc2\xb7",
|
467
|
+
'minus;' => "\xe2\x88\x92",
|
468
|
+
'mu;' => "\xce\xbc",
|
469
|
+
'nabla;' => "\xe2\x88\x87",
|
470
|
+
'nbsp' => "\xc2\xa0",
|
471
|
+
'nbsp;' => "\xc2\xa0",
|
472
|
+
'ndash;' => "\xe2\x80\x93",
|
473
|
+
'ne;' => "\xe2\x89\xa0",
|
474
|
+
'ni;' => "\xe2\x88\x8b",
|
475
|
+
'not' => "\xc2\xac",
|
476
|
+
'not;' => "\xc2\xac",
|
477
|
+
'notin;' => "\xe2\x88\x89",
|
478
|
+
'nsub;' => "\xe2\x8a\x84",
|
479
|
+
'ntilde' => "\xc3\xb1",
|
480
|
+
'ntilde;' => "\xc3\xb1",
|
481
|
+
'nu;' => "\xce\xbd",
|
482
|
+
'oacute' => "\xc3\xb3",
|
483
|
+
'oacute;' => "\xc3\xb3",
|
484
|
+
'ocirc' => "\xc3\xb4",
|
485
|
+
'ocirc;' => "\xc3\xb4",
|
486
|
+
'oelig;' => "\xc5\x93",
|
487
|
+
'ograve' => "\xc3\xb2",
|
488
|
+
'ograve;' => "\xc3\xb2",
|
489
|
+
'oline;' => "\xe2\x80\xbe",
|
490
|
+
'omega;' => "\xcf\x89",
|
491
|
+
'omicron;' => "\xce\xbf",
|
492
|
+
'oplus;' => "\xe2\x8a\x95",
|
493
|
+
'or;' => "\xe2\x88\xa8",
|
494
|
+
'ordf' => "\xc2\xaa",
|
495
|
+
'ordf;' => "\xc2\xaa",
|
496
|
+
'ordm' => "\xc2\xba",
|
497
|
+
'ordm;' => "\xc2\xba",
|
498
|
+
'oslash' => "\xc3\xb8",
|
499
|
+
'oslash;' => "\xc3\xb8",
|
500
|
+
'otilde' => "\xc3\xb5",
|
501
|
+
'otilde;' => "\xc3\xb5",
|
502
|
+
'otimes;' => "\xe2\x8a\x97",
|
503
|
+
'ouml' => "\xc3\xb6",
|
504
|
+
'ouml;' => "\xc3\xb6",
|
505
|
+
'para' => "\xc2\xb6",
|
506
|
+
'para;' => "\xc2\xb6",
|
507
|
+
'part;' => "\xe2\x88\x82",
|
508
|
+
'permil;' => "\xe2\x80\xb0",
|
509
|
+
'perp;' => "\xe2\x8a\xa5",
|
510
|
+
'phi;' => "\xcf\x86",
|
511
|
+
'pi;' => "\xcf\x80",
|
512
|
+
'piv;' => "\xcf\x96",
|
513
|
+
'plusmn' => "\xc2\xb1",
|
514
|
+
'plusmn;' => "\xc2\xb1",
|
515
|
+
'pound' => "\xc2\xa3",
|
516
|
+
'pound;' => "\xc2\xa3",
|
517
|
+
'prime;' => "\xe2\x80\xb2",
|
518
|
+
'prod;' => "\xe2\x88\x8f",
|
519
|
+
'prop;' => "\xe2\x88\x9d",
|
520
|
+
'psi;' => "\xcf\x88",
|
521
|
+
'quot' => '"',
|
522
|
+
'quot;' => '"',
|
523
|
+
'rArr;' => "\xe2\x87\x92",
|
524
|
+
'radic;' => "\xe2\x88\x9a",
|
525
|
+
'rang;' => "\xe3\x80\x89",
|
526
|
+
'raquo' => "\xc2\xbb",
|
527
|
+
'raquo;' => "\xc2\xbb",
|
528
|
+
'rarr;' => "\xe2\x86\x92",
|
529
|
+
'rceil;' => "\xe2\x8c\x89",
|
530
|
+
'rdquo;' => "\xe2\x80\x9d",
|
531
|
+
'real;' => "\xe2\x84\x9c",
|
532
|
+
'reg' => "\xc2\xae",
|
533
|
+
'reg;' => "\xc2\xae",
|
534
|
+
'rfloor;' => "\xe2\x8c\x8b",
|
535
|
+
'rho;' => "\xcf\x81",
|
536
|
+
'rlm;' => "\xe2\x80\x8f",
|
537
|
+
'rsaquo;' => "\xe2\x80\xba",
|
538
|
+
'rsquo;' => "\xe2\x80\x99",
|
539
|
+
'sbquo;' => "\xe2\x80\x9a",
|
540
|
+
'scaron;' => "\xc5\xa1",
|
541
|
+
'sdot;' => "\xe2\x8b\x85",
|
542
|
+
'sect' => "\xc2\xa7",
|
543
|
+
'sect;' => "\xc2\xa7",
|
544
|
+
'shy' => "\xc2\xad",
|
545
|
+
'shy;' => "\xc2\xad",
|
546
|
+
'sigma;' => "\xcf\x83",
|
547
|
+
'sigmaf;' => "\xcf\x82",
|
548
|
+
'sim;' => "\xe2\x88\xbc",
|
549
|
+
'spades;' => "\xe2\x99\xa0",
|
550
|
+
'sub;' => "\xe2\x8a\x82",
|
551
|
+
'sube;' => "\xe2\x8a\x86",
|
552
|
+
'sum;' => "\xe2\x88\x91",
|
553
|
+
'sup1' => "\xc2\xb9",
|
554
|
+
'sup1;' => "\xc2\xb9",
|
555
|
+
'sup2' => "\xc2\xb2",
|
556
|
+
'sup2;' => "\xc2\xb2",
|
557
|
+
'sup3' => "\xc2\xb3",
|
558
|
+
'sup3;' => "\xc2\xb3",
|
559
|
+
'sup;' => "\xe2\x8a\x83",
|
560
|
+
'supe;' => "\xe2\x8a\x87",
|
561
|
+
'szlig' => "\xc3\x9f",
|
562
|
+
'szlig;' => "\xc3\x9f",
|
563
|
+
'tau;' => "\xcf\x84",
|
564
|
+
'there4;' => "\xe2\x88\xb4",
|
565
|
+
'theta;' => "\xce\xb8",
|
566
|
+
'thetasym;' => "\xcf\x91",
|
567
|
+
'thinsp;' => "\xe2\x80\x89",
|
568
|
+
'thorn' => "\xc3\xbe",
|
569
|
+
'thorn;' => "\xc3\xbe",
|
570
|
+
'tilde;' => "\xcb\x9c",
|
571
|
+
'times' => "\xc3\x97",
|
572
|
+
'times;' => "\xc3\x97",
|
573
|
+
'trade;' => "\xe2\x84\xa2",
|
574
|
+
'uArr;' => "\xe2\x87\x91",
|
575
|
+
'uacute' => "\xc3\xba",
|
576
|
+
'uacute;' => "\xc3\xba",
|
577
|
+
'uarr;' => "\xe2\x86\x91",
|
578
|
+
'ucirc' => "\xc3\xbb",
|
579
|
+
'ucirc;' => "\xc3\xbb",
|
580
|
+
'ugrave' => "\xc3\xb9",
|
581
|
+
'ugrave;' => "\xc3\xb9",
|
582
|
+
'uml' => "\xc2\xa8",
|
583
|
+
'uml;' => "\xc2\xa8",
|
584
|
+
'upsih;' => "\xcf\x92",
|
585
|
+
'upsilon;' => "\xcf\x85",
|
586
|
+
'uuml' => "\xc3\xbc",
|
587
|
+
'uuml;' => "\xc3\xbc",
|
588
|
+
'weierp;' => "\xe2\x84\x98",
|
589
|
+
'xi;' => "\xce\xbe",
|
590
|
+
'yacute' => "\xc3\xbd",
|
591
|
+
'yacute;' => "\xc3\xbd",
|
592
|
+
'yen' => "\xc2\xa5",
|
593
|
+
'yen;' => "\xc2\xa5",
|
594
|
+
'yuml' => "\xc3\xbf",
|
595
|
+
'yuml;' => "\xc3\xbf",
|
596
|
+
'zeta;' => "\xce\xb6",
|
597
|
+
'zwj;' => "\xe2\x80\x8d",
|
598
|
+
'zwnj;' => "\xe2\x80\x8c"
|
599
|
+
}
|
600
|
+
|
601
|
+
ENCODINGS = %w[
|
602
|
+
ansi_x3.4-1968
|
603
|
+
iso-ir-6
|
604
|
+
ansi_x3.4-1986
|
605
|
+
iso_646.irv:1991
|
606
|
+
ascii
|
607
|
+
iso646-us
|
608
|
+
us-ascii
|
609
|
+
us
|
610
|
+
ibm367
|
611
|
+
cp367
|
612
|
+
csascii
|
613
|
+
ks_c_5601-1987
|
614
|
+
korean
|
615
|
+
iso-2022-kr
|
616
|
+
csiso2022kr
|
617
|
+
euc-kr
|
618
|
+
iso-2022-jp
|
619
|
+
csiso2022jp
|
620
|
+
iso-2022-jp-2
|
621
|
+
iso-ir-58
|
622
|
+
chinese
|
623
|
+
csiso58gb231280
|
624
|
+
iso_8859-1:1987
|
625
|
+
iso-ir-100
|
626
|
+
iso_8859-1
|
627
|
+
iso-8859-1
|
628
|
+
latin1
|
629
|
+
l1
|
630
|
+
ibm819
|
631
|
+
cp819
|
632
|
+
csisolatin1
|
633
|
+
iso_8859-2:1987
|
634
|
+
iso-ir-101
|
635
|
+
iso_8859-2
|
636
|
+
iso-8859-2
|
637
|
+
latin2
|
638
|
+
l2
|
639
|
+
csisolatin2
|
640
|
+
iso_8859-3:1988
|
641
|
+
iso-ir-109
|
642
|
+
iso_8859-3
|
643
|
+
iso-8859-3
|
644
|
+
latin3
|
645
|
+
l3
|
646
|
+
csisolatin3
|
647
|
+
iso_8859-4:1988
|
648
|
+
iso-ir-110
|
649
|
+
iso_8859-4
|
650
|
+
iso-8859-4
|
651
|
+
latin4
|
652
|
+
l4
|
653
|
+
csisolatin4
|
654
|
+
iso_8859-6:1987
|
655
|
+
iso-ir-127
|
656
|
+
iso_8859-6
|
657
|
+
iso-8859-6
|
658
|
+
ecma-114
|
659
|
+
asmo-708
|
660
|
+
arabic
|
661
|
+
csisolatinarabic
|
662
|
+
iso_8859-7:1987
|
663
|
+
iso-ir-126
|
664
|
+
iso_8859-7
|
665
|
+
iso-8859-7
|
666
|
+
elot_928
|
667
|
+
ecma-118
|
668
|
+
greek
|
669
|
+
greek8
|
670
|
+
csisolatingreek
|
671
|
+
iso_8859-8:1988
|
672
|
+
iso-ir-138
|
673
|
+
iso_8859-8
|
674
|
+
iso-8859-8
|
675
|
+
hebrew
|
676
|
+
csisolatinhebrew
|
677
|
+
iso_8859-5:1988
|
678
|
+
iso-ir-144
|
679
|
+
iso_8859-5
|
680
|
+
iso-8859-5
|
681
|
+
cyrillic
|
682
|
+
csisolatincyrillic
|
683
|
+
iso_8859-9:1989
|
684
|
+
iso-ir-148
|
685
|
+
iso_8859-9
|
686
|
+
iso-8859-9
|
687
|
+
latin5
|
688
|
+
l5
|
689
|
+
csisolatin5
|
690
|
+
iso-8859-10
|
691
|
+
iso-ir-157
|
692
|
+
l6
|
693
|
+
iso_8859-10:1992
|
694
|
+
csisolatin6
|
695
|
+
latin6
|
696
|
+
hp-roman8
|
697
|
+
roman8
|
698
|
+
r8
|
699
|
+
ibm037
|
700
|
+
cp037
|
701
|
+
csibm037
|
702
|
+
ibm424
|
703
|
+
cp424
|
704
|
+
csibm424
|
705
|
+
ibm437
|
706
|
+
cp437
|
707
|
+
437
|
708
|
+
cspc8codepage437
|
709
|
+
ibm500
|
710
|
+
cp500
|
711
|
+
csibm500
|
712
|
+
ibm775
|
713
|
+
cp775
|
714
|
+
cspc775baltic
|
715
|
+
ibm850
|
716
|
+
cp850
|
717
|
+
850
|
718
|
+
cspc850multilingual
|
719
|
+
ibm852
|
720
|
+
cp852
|
721
|
+
852
|
722
|
+
cspcp852
|
723
|
+
ibm855
|
724
|
+
cp855
|
725
|
+
855
|
726
|
+
csibm855
|
727
|
+
ibm857
|
728
|
+
cp857
|
729
|
+
857
|
730
|
+
csibm857
|
731
|
+
ibm860
|
732
|
+
cp860
|
733
|
+
860
|
734
|
+
csibm860
|
735
|
+
ibm861
|
736
|
+
cp861
|
737
|
+
861
|
738
|
+
cp-is
|
739
|
+
csibm861
|
740
|
+
ibm862
|
741
|
+
cp862
|
742
|
+
862
|
743
|
+
cspc862latinhebrew
|
744
|
+
ibm863
|
745
|
+
cp863
|
746
|
+
863
|
747
|
+
csibm863
|
748
|
+
ibm864
|
749
|
+
cp864
|
750
|
+
csibm864
|
751
|
+
ibm865
|
752
|
+
cp865
|
753
|
+
865
|
754
|
+
csibm865
|
755
|
+
ibm866
|
756
|
+
cp866
|
757
|
+
866
|
758
|
+
csibm866
|
759
|
+
ibm869
|
760
|
+
cp869
|
761
|
+
869
|
762
|
+
cp-gr
|
763
|
+
csibm869
|
764
|
+
ibm1026
|
765
|
+
cp1026
|
766
|
+
csibm1026
|
767
|
+
koi8-r
|
768
|
+
cskoi8r
|
769
|
+
koi8-u
|
770
|
+
big5-hkscs
|
771
|
+
ptcp154
|
772
|
+
csptcp154
|
773
|
+
pt154
|
774
|
+
cp154
|
775
|
+
utf-7
|
776
|
+
utf-16be
|
777
|
+
utf-16le
|
778
|
+
utf-16
|
779
|
+
utf-8
|
780
|
+
iso-8859-13
|
781
|
+
iso-8859-14
|
782
|
+
iso-ir-199
|
783
|
+
iso_8859-14:1998
|
784
|
+
iso_8859-14
|
785
|
+
latin8
|
786
|
+
iso-celtic
|
787
|
+
l8
|
788
|
+
iso-8859-15
|
789
|
+
iso_8859-15
|
790
|
+
iso-8859-16
|
791
|
+
iso-ir-226
|
792
|
+
iso_8859-16:2001
|
793
|
+
iso_8859-16
|
794
|
+
latin10
|
795
|
+
l10
|
796
|
+
gbk
|
797
|
+
cp936
|
798
|
+
ms936
|
799
|
+
gb18030
|
800
|
+
shift_jis
|
801
|
+
ms_kanji
|
802
|
+
csshiftjis
|
803
|
+
euc-jp
|
804
|
+
gb2312
|
805
|
+
big5
|
806
|
+
csbig5
|
807
|
+
windows-1250
|
808
|
+
windows-1251
|
809
|
+
windows-1252
|
810
|
+
windows-1253
|
811
|
+
windows-1254
|
812
|
+
windows-1255
|
813
|
+
windows-1256
|
814
|
+
windows-1257
|
815
|
+
windows-1258
|
816
|
+
tis-620
|
817
|
+
hz-gb-2312
|
818
|
+
]
|
819
|
+
|
820
|
+
E = {
|
821
|
+
"null-character" =>
|
822
|
+
_("Null character in input stream, replaced with U+FFFD."),
|
823
|
+
"incorrectly-placed-solidus" =>
|
824
|
+
_("Solidus (/) incorrectly placed in tag."),
|
825
|
+
"incorrect-cr-newline-entity" =>
|
826
|
+
_("Incorrect CR newline entity, replaced with LF."),
|
827
|
+
"illegal-windows-1252-entity" =>
|
828
|
+
_("Entity used with illegal number (windows-1252 reference)."),
|
829
|
+
"cant-convert-numeric-entity" =>
|
830
|
+
_("Numeric entity couldn't be converted to character " +
|
831
|
+
"(codepoint U+%(charAsInt)08x)."),
|
832
|
+
"illegal-codepoint-for-numeric-entity" =>
|
833
|
+
_("Numeric entity represents an illegal codepoint=> " +
|
834
|
+
"U+%(charAsInt)08x."),
|
835
|
+
"numeric-entity-without-semicolon" =>
|
836
|
+
_("Numeric entity didn't end with ';'."),
|
837
|
+
"expected-numeric-entity-but-got-eof" =>
|
838
|
+
_("Numeric entity expected. Got end of file instead."),
|
839
|
+
"expected-numeric-entity" =>
|
840
|
+
_("Numeric entity expected but none found."),
|
841
|
+
"named-entity-without-semicolon" =>
|
842
|
+
_("Named entity didn't end with ';'."),
|
843
|
+
"expected-named-entity" =>
|
844
|
+
_("Named entity expected. Got none."),
|
845
|
+
"attributes-in-end-tag" =>
|
846
|
+
_("End tag contains unexpected attributes."),
|
847
|
+
"expected-tag-name-but-got-right-bracket" =>
|
848
|
+
_("Expected tag name. Got '>' instead."),
|
849
|
+
"expected-tag-name-but-got-question-mark" =>
|
850
|
+
_("Expected tag name. Got '?' instead. (HTML doesn't " +
|
851
|
+
"support processing instructions.)"),
|
852
|
+
"expected-tag-name" =>
|
853
|
+
_("Expected tag name. Got something else instead"),
|
854
|
+
"expected-closing-tag-but-got-right-bracket" =>
|
855
|
+
_("Expected closing tag. Got '>' instead. Ignoring '</>'."),
|
856
|
+
"expected-closing-tag-but-got-eof" =>
|
857
|
+
_("Expected closing tag. Unexpected end of file."),
|
858
|
+
"expected-closing-tag-but-got-char" =>
|
859
|
+
_("Expected closing tag. Unexpected character '%(data)' found."),
|
860
|
+
"eof-in-tag-name" =>
|
861
|
+
_("Unexpected end of file in the tag name."),
|
862
|
+
"expected-attribute-name-but-got-eof" =>
|
863
|
+
_("Unexpected end of file. Expected attribute name instead."),
|
864
|
+
"eof-in-attribute-name" =>
|
865
|
+
_("Unexpected end of file in attribute name."),
|
866
|
+
"duplicate-attribute" =>
|
867
|
+
_("Dropped duplicate attribute on tag."),
|
868
|
+
"expected-end-of-tag-name-but-got-eof" =>
|
869
|
+
_("Unexpected end of file. Expected = or end of tag."),
|
870
|
+
"expected-attribute-value-but-got-eof" =>
|
871
|
+
_("Unexpected end of file. Expected attribute value."),
|
872
|
+
"eof-in-attribute-value-double-quote" =>
|
873
|
+
_("Unexpected end of file in attribute value (\")."),
|
874
|
+
"eof-in-attribute-value-single-quote" =>
|
875
|
+
_("Unexpected end of file in attribute value (')."),
|
876
|
+
"eof-in-attribute-value-no-quotes" =>
|
877
|
+
_("Unexpected end of file in attribute value."),
|
878
|
+
"expected-dashes-or-doctype" =>
|
879
|
+
_("Expected '--' or 'DOCTYPE'. Not found."),
|
880
|
+
"incorrect-comment" =>
|
881
|
+
_("Incorrect comment."),
|
882
|
+
"eof-in-comment" =>
|
883
|
+
_("Unexpected end of file in comment."),
|
884
|
+
"eof-in-comment-end-dash" =>
|
885
|
+
_("Unexpected end of file in comment (-)"),
|
886
|
+
"unexpected-dash-after-double-dash-in-comment" =>
|
887
|
+
_("Unexpected '-' after '--' found in comment."),
|
888
|
+
"eof-in-comment-double-dash" =>
|
889
|
+
_("Unexpected end of file in comment (--)."),
|
890
|
+
"unexpected-char-in-comment" =>
|
891
|
+
_("Unexpected character in comment found."),
|
892
|
+
"need-space-after-doctype" =>
|
893
|
+
_("No space after literal string 'DOCTYPE'."),
|
894
|
+
"expected-doctype-name-but-got-right-bracket" =>
|
895
|
+
_("Unexpected > character. Expected DOCTYPE name."),
|
896
|
+
"expected-doctype-name-but-got-eof" =>
|
897
|
+
_("Unexpected end of file. Expected DOCTYPE name."),
|
898
|
+
"eof-in-doctype-name" =>
|
899
|
+
_("Unexpected end of file in DOCTYPE name."),
|
900
|
+
"eof-in-doctype" =>
|
901
|
+
_("Unexpected end of file in DOCTYPE."),
|
902
|
+
"expected-space-or-right-bracket-in-doctype" =>
|
903
|
+
_("Expected space or '>'. Got '%(data)'"),
|
904
|
+
"unexpected-end-of-doctype" =>
|
905
|
+
_("Unexpected end of DOCTYPE."),
|
906
|
+
"unexpected-char-in-doctype" =>
|
907
|
+
_("Unexpected character in DOCTYPE."),
|
908
|
+
"eof-in-bogus-doctype" =>
|
909
|
+
_("Unexpected end of file in bogus doctype."),
|
910
|
+
"eof-in-innerhtml" =>
|
911
|
+
_("XXX innerHTML EOF"),
|
912
|
+
"unexpected-doctype" =>
|
913
|
+
_("Unexpected DOCTYPE. Ignored."),
|
914
|
+
"non-html-root" =>
|
915
|
+
_("html needs to be the first start tag."),
|
916
|
+
"expected-doctype-but-got-eof" =>
|
917
|
+
_("Unexpected End of file. Expected DOCTYPE."),
|
918
|
+
"unknown-doctype" =>
|
919
|
+
_("Erroneous DOCTYPE."),
|
920
|
+
"expected-doctype-but-got-chars" =>
|
921
|
+
_("Unexpected non-space characters. Expected DOCTYPE."),
|
922
|
+
"expected-doctype-but-got-start-tag" =>
|
923
|
+
_("Unexpected start tag (%(name)). Expected DOCTYPE."),
|
924
|
+
"expected-doctype-but-got-end-tag" =>
|
925
|
+
_("Unexpected end tag (%(name)). Expected DOCTYPE."),
|
926
|
+
"end-tag-after-implied-root" =>
|
927
|
+
_("Unexpected end tag (%(name)) after the (implied) root element."),
|
928
|
+
"expected-named-closing-tag-but-got-eof" =>
|
929
|
+
_("Unexpected end of file. Expected end tag (%(name))."),
|
930
|
+
"two-heads-are-not-better-than-one" =>
|
931
|
+
_("Unexpected start tag head in existing head. Ignored."),
|
932
|
+
"unexpected-end-tag" =>
|
933
|
+
_("Unexpected end tag (%(name)). Ignored."),
|
934
|
+
"unexpected-start-tag-out-of-my-head" =>
|
935
|
+
_("Unexpected start tag (%(name)) that can be in head. Moved."),
|
936
|
+
"unexpected-start-tag" =>
|
937
|
+
_("Unexpected start tag (%(name))."),
|
938
|
+
"missing-end-tag" =>
|
939
|
+
_("Missing end tag (%(name))."),
|
940
|
+
"missing-end-tags" =>
|
941
|
+
_("Missing end tags (%(name))."),
|
942
|
+
"unexpected-start-tag-implies-end-tag" =>
|
943
|
+
_("Unexpected start tag (%(startName)) " +
|
944
|
+
"implies end tag (%(endName))."),
|
945
|
+
"unexpected-start-tag-treated-as" =>
|
946
|
+
_("Unexpected start tag (%(originalName)). Treated as %(newName)."),
|
947
|
+
"deprecated-tag" =>
|
948
|
+
_("Unexpected start tag %(name). Don't use it!"),
|
949
|
+
"unexpected-start-tag-ignored" =>
|
950
|
+
_("Unexpected start tag %(name). Ignored."),
|
951
|
+
"expected-one-end-tag-but-got-another" =>
|
952
|
+
_("Unexpected end tag (%(gotName)). " +
|
953
|
+
"Missing end tag (%(expectedName))."),
|
954
|
+
"end-tag-too-early" =>
|
955
|
+
_("End tag (%(name)) seen too early. Expected other end tag."),
|
956
|
+
"end-tag-too-early-named" =>
|
957
|
+
_("Unexpected end tag (%(gotName)). Expected end tag (%(expectedName))."),
|
958
|
+
"end-tag-too-early-ignored" =>
|
959
|
+
_("End tag (%(name)) seen too early. Ignored."),
|
960
|
+
"adoption-agency-1.1" =>
|
961
|
+
_("End tag (%(name)) violates step 1, " +
|
962
|
+
"paragraph 1 of the adoption agency algorithm."),
|
963
|
+
"adoption-agency-1.2" =>
|
964
|
+
_("End tag (%(name)) violates step 1, " +
|
965
|
+
"paragraph 2 of the adoption agency algorithm."),
|
966
|
+
"adoption-agency-1.3" =>
|
967
|
+
_("End tag (%(name)) violates step 1, " +
|
968
|
+
"paragraph 3 of the adoption agency algorithm."),
|
969
|
+
"unexpected-end-tag-treated-as" =>
|
970
|
+
_("Unexpected end tag (%(originalName)). Treated as %(newName)."),
|
971
|
+
"no-end-tag" =>
|
972
|
+
_("This element (%(name)) has no end tag."),
|
973
|
+
"unexpected-implied-end-tag-in-table" =>
|
974
|
+
_("Unexpected implied end tag (%(name)) in the table phase."),
|
975
|
+
"unexpected-implied-end-tag-in-table-body" =>
|
976
|
+
_("Unexpected implied end tag (%(name)) in the table body phase."),
|
977
|
+
"unexpected-char-implies-table-voodoo" =>
|
978
|
+
_("Unexpected non-space characters in " +
|
979
|
+
"table context caused voodoo mode."),
|
980
|
+
"unexpected-start-tag-implies-table-voodoo" =>
|
981
|
+
_("Unexpected start tag (%(name)) in " +
|
982
|
+
"table context caused voodoo mode."),
|
983
|
+
"unexpected-end-tag-implies-table-voodoo" =>
|
984
|
+
_("Unexpected end tag (%(name)) in " +
|
985
|
+
"table context caused voodoo mode."),
|
986
|
+
"unexpected-cell-in-table-body" =>
|
987
|
+
_("Unexpected table cell start tag (%(name)) " +
|
988
|
+
"in the table body phase."),
|
989
|
+
"unexpected-cell-end-tag" =>
|
990
|
+
_("Got table cell end tag (%(name)) " +
|
991
|
+
"while required end tags are missing."),
|
992
|
+
"unexpected-end-tag-in-table-body" =>
|
993
|
+
_("Unexpected end tag (%(name)) in the table body phase. Ignored."),
|
994
|
+
"unexpected-implied-end-tag-in-table-row" =>
|
995
|
+
_("Unexpected implied end tag (%(name)) in the table row phase."),
|
996
|
+
"unexpected-end-tag-in-table-row" =>
|
997
|
+
_("Unexpected end tag (%(name)) in the table row phase. Ignored."),
|
998
|
+
"unexpected-select-in-select" =>
|
999
|
+
_("Unexpected select start tag in the select phase " +
|
1000
|
+
"implies select start tag."),
|
1001
|
+
"unexpected-start-tag-in-select" =>
|
1002
|
+
_("Unexpected start tag token (%(name) in the select phase. " +
|
1003
|
+
"Ignored."),
|
1004
|
+
"unexpected-end-tag-in-select" =>
|
1005
|
+
_("Unexpected end tag (%(name)) in the select phase. Ignored."),
|
1006
|
+
"unexpected-char-after-body" =>
|
1007
|
+
_("Unexpected non-space characters in the after body phase."),
|
1008
|
+
"unexpected-start-tag-after-body" =>
|
1009
|
+
_("Unexpected start tag token (%(name))" +
|
1010
|
+
" in the after body phase."),
|
1011
|
+
"unexpected-end-tag-after-body" =>
|
1012
|
+
_("Unexpected end tag token (%(name))" +
|
1013
|
+
" in the after body phase."),
|
1014
|
+
"unexpected-char-in-frameset" =>
|
1015
|
+
_("Unepxected characters in the frameset phase. Characters ignored."),
|
1016
|
+
"unexpected-start-tag-in-frameset" =>
|
1017
|
+
_("Unexpected start tag token (%(name))" +
|
1018
|
+
" in the frameset phase. Ignored."),
|
1019
|
+
"unexpected-frameset-in-frameset-innerhtml" =>
|
1020
|
+
_("Unexpected end tag token (frameset) " +
|
1021
|
+
"in the frameset phase (innerHTML)."),
|
1022
|
+
"unexpected-end-tag-in-frameset" =>
|
1023
|
+
_("Unexpected end tag token (%(name))" +
|
1024
|
+
" in the frameset phase. Ignored."),
|
1025
|
+
"unexpected-char-after-frameset" =>
|
1026
|
+
_("Unexpected non-space characters in the " +
|
1027
|
+
"after frameset phase. Ignored."),
|
1028
|
+
"unexpected-start-tag-after-frameset" =>
|
1029
|
+
_("Unexpected start tag (%(name))" +
|
1030
|
+
" in the after frameset phase. Ignored."),
|
1031
|
+
"unexpected-end-tag-after-frameset" =>
|
1032
|
+
_("Unexpected end tag (%(name))" +
|
1033
|
+
" in the after frameset phase. Ignored."),
|
1034
|
+
"expected-eof-but-got-char" =>
|
1035
|
+
_("Unexpected non-space characters. Expected end of file."),
|
1036
|
+
"expected-eof-but-got-start-tag" =>
|
1037
|
+
_("Unexpected start tag (%(name))" +
|
1038
|
+
". Expected end of file."),
|
1039
|
+
"expected-eof-but-got-end-tag" =>
|
1040
|
+
_("Unexpected end tag (%(name))" +
|
1041
|
+
". Expected end of file."),
|
1042
|
+
"unexpected-end-table-in-caption" =>
|
1043
|
+
_("Unexpected end table tag in caption. Generates implied end caption.")
|
1044
|
+
}
|
1045
|
+
|
1046
|
+
end
|