rfeedparser 0.9.8 → 0.9.9
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- data/lib/rfeedparser.rb +170 -3345
- data/lib/rfeedparser/aliases.rb +432 -0
- data/lib/rfeedparser/better_attributelist.rb +41 -0
- data/lib/rfeedparser/better_sgmlparser.rb +264 -0
- data/lib/rfeedparser/encoding_helpers.rb +257 -0
- data/lib/rfeedparser/feedparserdict.rb +93 -0
- data/lib/rfeedparser/forgiving_uri.rb +93 -0
- data/lib/rfeedparser/markup_helpers.rb +73 -0
- data/lib/rfeedparser/parser_mixin.rb +1235 -0
- data/lib/rfeedparser/parsers.rb +177 -0
- data/lib/rfeedparser/scrub.rb +207 -0
- data/lib/rfeedparser/time_helpers.rb +408 -0
- data/tests/rfeedparsertest.rb +3 -1
- metadata +3271 -3250
|
@@ -0,0 +1,432 @@
|
|
|
1
|
+
#!/usr/bin/ruby
|
|
2
|
+
|
|
3
|
+
module FeedParserUtilities
|
|
4
|
+
# Adapted from python2.4's encodings/aliases.py
|
|
5
|
+
|
|
6
|
+
Encoding_Aliases = {
|
|
7
|
+
'unicode' => 'utf-16',
|
|
8
|
+
|
|
9
|
+
# MacOSX does not have Unicode as a separate encoding nor even
|
|
10
|
+
# aliased. My Ubuntu box has it as a separate encoding but I cannot
|
|
11
|
+
# for the life of me figure out where the source code for UNICODE.so
|
|
12
|
+
# is (supposedly, in libc6 .deb but that's a damn lie), so I don't
|
|
13
|
+
# know what it expects. After some extensive research, I've decided
|
|
14
|
+
# to alias it to utf-16 much like Python does when it is built with
|
|
15
|
+
# --enable-unicode=ucs2. This could be seriously wrong. I have no idea.
|
|
16
|
+
|
|
17
|
+
# ascii codec
|
|
18
|
+
'646' => 'ascii',
|
|
19
|
+
'ansi_x3.4_1968' => 'ascii',
|
|
20
|
+
'ansi_x3_4_1968' => 'ascii', # some email headers use this non-standard name
|
|
21
|
+
'ansi_x3.4_1986' => 'ascii',
|
|
22
|
+
'cp367' => 'ascii',
|
|
23
|
+
'csascii' => 'ascii',
|
|
24
|
+
'ibm367' => 'ascii',
|
|
25
|
+
'iso646_us' => 'ascii',
|
|
26
|
+
'iso_646.irv_1991' => 'ascii',
|
|
27
|
+
'iso_ir_6' => 'ascii',
|
|
28
|
+
'us' => 'ascii',
|
|
29
|
+
'us_ascii' => 'ascii',
|
|
30
|
+
|
|
31
|
+
# big5 codec
|
|
32
|
+
'big5_tw' => 'big5',
|
|
33
|
+
'csbig5' => 'big5',
|
|
34
|
+
|
|
35
|
+
# big5hkscs codec
|
|
36
|
+
'big5_hkscs' => 'big5hkscs',
|
|
37
|
+
'hkscs' => 'big5hkscs',
|
|
38
|
+
|
|
39
|
+
# cp037 codec
|
|
40
|
+
'037' => 'cp037',
|
|
41
|
+
'csibm037' => 'cp037',
|
|
42
|
+
'ebcdic_cp_ca' => 'cp037',
|
|
43
|
+
'ebcdic_cp_nl' => 'cp037',
|
|
44
|
+
'ebcdic_cp_us' => 'cp037',
|
|
45
|
+
'ebcdic_cp_wt' => 'cp037',
|
|
46
|
+
'ibm037' => 'cp037',
|
|
47
|
+
'ibm039' => 'cp037',
|
|
48
|
+
|
|
49
|
+
# cp1026 codec
|
|
50
|
+
'1026' => 'cp1026',
|
|
51
|
+
'csibm1026' => 'cp1026',
|
|
52
|
+
'ibm1026' => 'cp1026',
|
|
53
|
+
|
|
54
|
+
# cp1140 codec
|
|
55
|
+
'1140' => 'cp1140',
|
|
56
|
+
'ibm1140' => 'cp1140',
|
|
57
|
+
|
|
58
|
+
# cp1250 codec
|
|
59
|
+
'1250' => 'cp1250',
|
|
60
|
+
'windows_1250' => 'cp1250',
|
|
61
|
+
|
|
62
|
+
# cp1251 codec
|
|
63
|
+
'1251' => 'cp1251',
|
|
64
|
+
'windows_1251' => 'cp1251',
|
|
65
|
+
|
|
66
|
+
# cp1252 codec
|
|
67
|
+
'1252' => 'cp1252',
|
|
68
|
+
'windows_1252' => 'cp1252',
|
|
69
|
+
|
|
70
|
+
# cp1253 codec
|
|
71
|
+
'1253' => 'cp1253',
|
|
72
|
+
'windows_1253' => 'cp1253',
|
|
73
|
+
|
|
74
|
+
# cp1254 codec
|
|
75
|
+
'1254' => 'cp1254',
|
|
76
|
+
'windows_1254' => 'cp1254',
|
|
77
|
+
|
|
78
|
+
# cp1255 codec
|
|
79
|
+
'1255' => 'cp1255',
|
|
80
|
+
'windows_1255' => 'cp1255',
|
|
81
|
+
|
|
82
|
+
# cp1256 codec
|
|
83
|
+
'1256' => 'cp1256',
|
|
84
|
+
'windows_1256' => 'cp1256',
|
|
85
|
+
|
|
86
|
+
# cp1257 codec
|
|
87
|
+
'1257' => 'cp1257',
|
|
88
|
+
'windows_1257' => 'cp1257',
|
|
89
|
+
|
|
90
|
+
# cp1258 codec
|
|
91
|
+
'1258' => 'cp1258',
|
|
92
|
+
'windows_1258' => 'cp1258',
|
|
93
|
+
|
|
94
|
+
# cp424 codec
|
|
95
|
+
'424' => 'cp424',
|
|
96
|
+
'csibm424' => 'cp424',
|
|
97
|
+
'ebcdic_cp_he' => 'cp424',
|
|
98
|
+
'ibm424' => 'cp424',
|
|
99
|
+
|
|
100
|
+
# cp437 codec
|
|
101
|
+
'437' => 'cp437',
|
|
102
|
+
'cspc8codepage437' => 'cp437',
|
|
103
|
+
'ibm437' => 'cp437',
|
|
104
|
+
|
|
105
|
+
# cp500 codec
|
|
106
|
+
'500' => 'cp500',
|
|
107
|
+
'csibm500' => 'cp500',
|
|
108
|
+
'ebcdic_cp_be' => 'cp500',
|
|
109
|
+
'ebcdic_cp_ch' => 'cp500',
|
|
110
|
+
'ibm500' => 'cp500',
|
|
111
|
+
|
|
112
|
+
# cp775 codec
|
|
113
|
+
'775' => 'cp775',
|
|
114
|
+
'cspc775baltic' => 'cp775',
|
|
115
|
+
'ibm775' => 'cp775',
|
|
116
|
+
|
|
117
|
+
# cp850 codec
|
|
118
|
+
'850' => 'cp850',
|
|
119
|
+
'cspc850multilingual' => 'cp850',
|
|
120
|
+
'ibm850' => 'cp850',
|
|
121
|
+
|
|
122
|
+
# cp852 codec
|
|
123
|
+
'852' => 'cp852',
|
|
124
|
+
'cspcp852' => 'cp852',
|
|
125
|
+
'ibm852' => 'cp852',
|
|
126
|
+
|
|
127
|
+
# cp855 codec
|
|
128
|
+
'855' => 'cp855',
|
|
129
|
+
'csibm855' => 'cp855',
|
|
130
|
+
'ibm855' => 'cp855',
|
|
131
|
+
|
|
132
|
+
# cp857 codec
|
|
133
|
+
'857' => 'cp857',
|
|
134
|
+
'csibm857' => 'cp857',
|
|
135
|
+
'ibm857' => 'cp857',
|
|
136
|
+
|
|
137
|
+
# cp860 codec
|
|
138
|
+
'860' => 'cp860',
|
|
139
|
+
'csibm860' => 'cp860',
|
|
140
|
+
'ibm860' => 'cp860',
|
|
141
|
+
|
|
142
|
+
# cp861 codec
|
|
143
|
+
'861' => 'cp861',
|
|
144
|
+
'cp_is' => 'cp861',
|
|
145
|
+
'csibm861' => 'cp861',
|
|
146
|
+
'ibm861' => 'cp861',
|
|
147
|
+
|
|
148
|
+
# cp862 codec
|
|
149
|
+
'862' => 'cp862',
|
|
150
|
+
'cspc862latinhebrew' => 'cp862',
|
|
151
|
+
'ibm862' => 'cp862',
|
|
152
|
+
|
|
153
|
+
# cp863 codec
|
|
154
|
+
'863' => 'cp863',
|
|
155
|
+
'csibm863' => 'cp863',
|
|
156
|
+
'ibm863' => 'cp863',
|
|
157
|
+
|
|
158
|
+
# cp864 codec
|
|
159
|
+
'864' => 'cp864',
|
|
160
|
+
'csibm864' => 'cp864',
|
|
161
|
+
'ibm864' => 'cp864',
|
|
162
|
+
|
|
163
|
+
# cp865 codec
|
|
164
|
+
'865' => 'cp865',
|
|
165
|
+
'csibm865' => 'cp865',
|
|
166
|
+
'ibm865' => 'cp865',
|
|
167
|
+
|
|
168
|
+
# cp866 codec
|
|
169
|
+
'866' => 'cp866',
|
|
170
|
+
'csibm866' => 'cp866',
|
|
171
|
+
'ibm866' => 'cp866',
|
|
172
|
+
|
|
173
|
+
# cp869 codec
|
|
174
|
+
'869' => 'cp869',
|
|
175
|
+
'cp_gr' => 'cp869',
|
|
176
|
+
'csibm869' => 'cp869',
|
|
177
|
+
'ibm869' => 'cp869',
|
|
178
|
+
|
|
179
|
+
# cp932 codec
|
|
180
|
+
'932' => 'cp932',
|
|
181
|
+
'ms932' => 'cp932',
|
|
182
|
+
'mskanji' => 'cp932',
|
|
183
|
+
'ms_kanji' => 'cp932',
|
|
184
|
+
|
|
185
|
+
# cp949 codec
|
|
186
|
+
'949' => 'cp949',
|
|
187
|
+
'ms949' => 'cp949',
|
|
188
|
+
'uhc' => 'cp949',
|
|
189
|
+
|
|
190
|
+
# cp950 codec
|
|
191
|
+
'950' => 'cp950',
|
|
192
|
+
'ms950' => 'cp950',
|
|
193
|
+
|
|
194
|
+
# euc_jp codec
|
|
195
|
+
'euc_jp' => 'euc-jp',
|
|
196
|
+
'eucjp' => 'euc-jp',
|
|
197
|
+
'ujis' => 'euc-jp',
|
|
198
|
+
'u_jis' => 'euc-jp',
|
|
199
|
+
|
|
200
|
+
# euc_kr codec
|
|
201
|
+
'euc_kr' => 'euc-kr',
|
|
202
|
+
'euckr' => 'euc-kr',
|
|
203
|
+
'korean' => 'euc-kr',
|
|
204
|
+
'ksc5601' => 'euc-kr',
|
|
205
|
+
'ks_c_5601' => 'euc-kr',
|
|
206
|
+
'ks_c_5601_1987' => 'euc-kr',
|
|
207
|
+
'ksx1001' => 'euc-kr',
|
|
208
|
+
'ks_x_1001' => 'euc-kr',
|
|
209
|
+
|
|
210
|
+
# gb18030 codec
|
|
211
|
+
'gb18030_2000' => 'gb18030',
|
|
212
|
+
|
|
213
|
+
# gb2312 codec
|
|
214
|
+
'chinese' => 'gb2312',
|
|
215
|
+
'csiso58gb231280' => 'gb2312',
|
|
216
|
+
'euc_cn' => 'gb2312',
|
|
217
|
+
'euccn' => 'gb2312',
|
|
218
|
+
'eucgb2312_cn' => 'gb2312',
|
|
219
|
+
'gb2312_1980' => 'gb2312',
|
|
220
|
+
'gb2312_80' => 'gb2312',
|
|
221
|
+
'iso_ir_58' => 'gb2312',
|
|
222
|
+
|
|
223
|
+
# gbk codec
|
|
224
|
+
'936' => 'gbk',
|
|
225
|
+
'cp936' => 'gbk',
|
|
226
|
+
'ms936' => 'gbk',
|
|
227
|
+
|
|
228
|
+
# hp-roman8 codec
|
|
229
|
+
'hp_roman8' => 'hp-roman8',
|
|
230
|
+
'roman8' => 'hp-roman8',
|
|
231
|
+
'r8' => 'hp-roman8',
|
|
232
|
+
'csHPRoman8' => 'hp-roman8',
|
|
233
|
+
|
|
234
|
+
# iso2022_jp codec
|
|
235
|
+
'iso2022_jp' => 'iso-2022-jp',
|
|
236
|
+
'csiso2022jp' => 'iso-2022-jp',
|
|
237
|
+
'iso2022jp' => 'iso-2022-jp',
|
|
238
|
+
'iso_2022_jp' => 'iso-2022-jp',
|
|
239
|
+
|
|
240
|
+
# iso2022_jp_1 codec
|
|
241
|
+
'iso2002_jp_1' => 'iso-2022-jp-1',
|
|
242
|
+
'iso2022jp_1' => 'iso-2022-jp-1',
|
|
243
|
+
'iso_2022_jp_1' => 'iso-2022-jp-1',
|
|
244
|
+
|
|
245
|
+
# iso2022_jp_2 codec
|
|
246
|
+
'iso2022_jp_2' => 'iso-2002-jp-2',
|
|
247
|
+
'iso2022jp_2' => 'iso-2022-jp-2',
|
|
248
|
+
'iso_2022_jp_2' => 'iso-2022-jp-2',
|
|
249
|
+
|
|
250
|
+
# iso2022_jp_3 codec
|
|
251
|
+
'iso2002_jp_3' => 'iso-2022-jp-3',
|
|
252
|
+
'iso2022jp_3' => 'iso-2022-jp-3',
|
|
253
|
+
'iso_2022_jp_3' => 'iso-2022-jp-3',
|
|
254
|
+
|
|
255
|
+
# iso2022_kr codec
|
|
256
|
+
'iso2022_kr' => 'iso-2022-kr',
|
|
257
|
+
'csiso2022kr' => 'iso-2022-kr',
|
|
258
|
+
'iso2022kr' => 'iso-2022-kr',
|
|
259
|
+
'iso_2022_kr' => 'iso-2022-kr',
|
|
260
|
+
|
|
261
|
+
# iso8859_10 codec
|
|
262
|
+
'iso8859_10' => 'iso-8859-10',
|
|
263
|
+
'csisolatin6' => 'iso-8859-10',
|
|
264
|
+
'iso_8859_10' => 'iso-8859-10',
|
|
265
|
+
'iso_8859_10_1992' => 'iso-8859-10',
|
|
266
|
+
'iso_ir_157' => 'iso-8859-10',
|
|
267
|
+
'l6' => 'iso-8859-10',
|
|
268
|
+
'latin6' => 'iso-8859-10',
|
|
269
|
+
|
|
270
|
+
# iso8859_13 codec
|
|
271
|
+
'iso8859_13' => 'iso-8859-13',
|
|
272
|
+
'iso_8859_13' => 'iso-8859-13',
|
|
273
|
+
|
|
274
|
+
# iso8859_14 codec
|
|
275
|
+
'iso8859_14' => 'iso-8859-14',
|
|
276
|
+
'iso_8859_14' => 'iso-8859-14',
|
|
277
|
+
'iso_8859_14_1998' => 'iso-8859-14',
|
|
278
|
+
'iso_celtic' => 'iso-8859-14',
|
|
279
|
+
'iso_ir_199' => 'iso-8859-14',
|
|
280
|
+
'l8' => 'iso-8859-14',
|
|
281
|
+
'latin8' => 'iso-8859-14',
|
|
282
|
+
|
|
283
|
+
# iso8859_15 codec
|
|
284
|
+
'iso8859_15' => 'iso-8859-15',
|
|
285
|
+
'iso_8859_15' => 'iso-8859-15',
|
|
286
|
+
|
|
287
|
+
# iso8859_1 codec
|
|
288
|
+
'latin_1' => 'iso-8859-1',
|
|
289
|
+
'cp819' => 'iso-8859-1',
|
|
290
|
+
'csisolatin1' => 'iso-8859-1',
|
|
291
|
+
'ibm819' => 'iso-8859-1',
|
|
292
|
+
'iso8859' => 'iso-8859-1',
|
|
293
|
+
'iso_8859_1' => 'iso-8859-1',
|
|
294
|
+
'iso_8859_1_1987' => 'iso-8859-1',
|
|
295
|
+
'iso_ir_100' => 'iso-8859-1',
|
|
296
|
+
'l1' => 'iso-8859-1',
|
|
297
|
+
'latin' => 'iso-8859-1',
|
|
298
|
+
'latin1' => 'iso-8859-1',
|
|
299
|
+
|
|
300
|
+
# iso8859_2 codec
|
|
301
|
+
'iso8859_2' => 'iso-8859-2',
|
|
302
|
+
'csisolatin2' => 'iso-8859-2',
|
|
303
|
+
'iso_8859_2' => 'iso-8859-2',
|
|
304
|
+
'iso_8859_2_1987' => 'iso-8859-2',
|
|
305
|
+
'iso_ir_101' => 'iso-8859-2',
|
|
306
|
+
'l2' => 'iso-8859-2',
|
|
307
|
+
'latin2' => 'iso-8859-2',
|
|
308
|
+
|
|
309
|
+
# iso8859_3 codec
|
|
310
|
+
'iso8859_3' => 'iso-8859-3',
|
|
311
|
+
'csisolatin3' => 'iso-8859-3',
|
|
312
|
+
'iso_8859_3' => 'iso-8859-3',
|
|
313
|
+
'iso_8859_3_1988' => 'iso-8859-3',
|
|
314
|
+
'iso_ir_109' => 'iso-8859-3',
|
|
315
|
+
'l3' => 'iso-8859-3',
|
|
316
|
+
'latin3' => 'iso-8859-3',
|
|
317
|
+
|
|
318
|
+
# iso8859_4 codec
|
|
319
|
+
'iso8849_4' => 'iso-8859-4',
|
|
320
|
+
'csisolatin4' => 'iso-8859-4',
|
|
321
|
+
'iso_8859_4' => 'iso-8859-4',
|
|
322
|
+
'iso_8859_4_1988' => 'iso-8859-4',
|
|
323
|
+
'iso_ir_110' => 'iso-8859-4',
|
|
324
|
+
'l4' => 'iso-8859-4',
|
|
325
|
+
'latin4' => 'iso-8859-4',
|
|
326
|
+
|
|
327
|
+
# iso8859_5 codec
|
|
328
|
+
'iso8859_5' => 'iso-8859-5',
|
|
329
|
+
'csisolatincyrillic' => 'iso-8859-5',
|
|
330
|
+
'cyrillic' => 'iso-8859-5',
|
|
331
|
+
'iso_8859_5' => 'iso-8859-5',
|
|
332
|
+
'iso_8859_5_1988' => 'iso-8859-5',
|
|
333
|
+
'iso_ir_144' => 'iso-8859-5',
|
|
334
|
+
|
|
335
|
+
# iso8859_6 codec
|
|
336
|
+
'iso8859_6' => 'iso-8859-6',
|
|
337
|
+
'arabic' => 'iso-8859-6',
|
|
338
|
+
'asmo_708' => 'iso-8859-6',
|
|
339
|
+
'csisolatinarabic' => 'iso-8859-6',
|
|
340
|
+
'ecma_114' => 'iso-8859-6',
|
|
341
|
+
'iso_8859_6' => 'iso-8859-6',
|
|
342
|
+
'iso_8859_6_1987' => 'iso-8859-6',
|
|
343
|
+
'iso_ir_127' => 'iso-8859-6',
|
|
344
|
+
|
|
345
|
+
# iso8859_7 codec
|
|
346
|
+
'iso8859_7' => 'iso-8859-7',
|
|
347
|
+
'csisolatingreek' => 'iso-8859-7',
|
|
348
|
+
'ecma_118' => 'iso-8859-7',
|
|
349
|
+
'elot_928' => 'iso-8859-7',
|
|
350
|
+
'greek' => 'iso-8859-7',
|
|
351
|
+
'greek8' => 'iso-8859-7',
|
|
352
|
+
'iso_8859_7' => 'iso-8859-7',
|
|
353
|
+
'iso_8859_7_1987' => 'iso-8859-7',
|
|
354
|
+
'iso_ir_126' => 'iso-8859-7',
|
|
355
|
+
|
|
356
|
+
# iso8859_8 codec
|
|
357
|
+
'iso8859_9' => 'iso8859_8',
|
|
358
|
+
'csisolatinhebrew' => 'iso-8859-8',
|
|
359
|
+
'hebrew' => 'iso-8859-8',
|
|
360
|
+
'iso_8859_8' => 'iso-8859-8',
|
|
361
|
+
'iso_8859_8_1988' => 'iso-8859-8',
|
|
362
|
+
'iso_ir_138' => 'iso-8859-8',
|
|
363
|
+
|
|
364
|
+
# iso8859_9 codec
|
|
365
|
+
'iso8859_9' => 'iso-8859-9',
|
|
366
|
+
'csisolatin5' => 'iso-8859-9',
|
|
367
|
+
'iso_8859_9' => 'iso-8859-9',
|
|
368
|
+
'iso_8859_9_1989' => 'iso-8859-9',
|
|
369
|
+
'iso_ir_148' => 'iso-8859-9',
|
|
370
|
+
'l5' => 'iso-8859-9',
|
|
371
|
+
'latin5' => 'iso-8859-9',
|
|
372
|
+
|
|
373
|
+
# iso8859_11 codec
|
|
374
|
+
'iso8859_11' => 'iso-8859-11',
|
|
375
|
+
'thai' => 'iso-8859-11',
|
|
376
|
+
'iso_8859_11' => 'iso-8859-11',
|
|
377
|
+
'iso_8859_11_2001' => 'iso-8859-11',
|
|
378
|
+
|
|
379
|
+
# iso8859_16 codec
|
|
380
|
+
'iso8859_16' => 'iso-8859-16',
|
|
381
|
+
'iso_8859_16' => 'iso-8859-16',
|
|
382
|
+
'iso_8859_16_2001' => 'iso-8859-16',
|
|
383
|
+
'iso_ir_226' => 'iso-8859-16',
|
|
384
|
+
'l10' => 'iso-8859-16',
|
|
385
|
+
'latin10' => 'iso-8859-16',
|
|
386
|
+
|
|
387
|
+
# cskoi8r codec
|
|
388
|
+
'koi8_r' => 'cskoi8r',
|
|
389
|
+
|
|
390
|
+
# mac_cyrillic codec
|
|
391
|
+
'mac_cyrillic' => 'maccyrillic',
|
|
392
|
+
|
|
393
|
+
# shift_jis codec
|
|
394
|
+
'csshiftjis' => 'shift_jis',
|
|
395
|
+
'shiftjis' => 'shift_jis',
|
|
396
|
+
'sjis' => 'shift_jis',
|
|
397
|
+
's_jis' => 'shift_jis',
|
|
398
|
+
|
|
399
|
+
# shift_jisx0213 codec
|
|
400
|
+
'shiftjisx0213' => 'shift_jisx0213',
|
|
401
|
+
'sjisx0213' => 'shift_jisx0213',
|
|
402
|
+
's_jisx0213' => 'shift_jisx0213',
|
|
403
|
+
|
|
404
|
+
# utf_16 codec
|
|
405
|
+
'utf_16' => 'utf-16',
|
|
406
|
+
'u16' => 'utf-16',
|
|
407
|
+
'utf16' => 'utf-16',
|
|
408
|
+
|
|
409
|
+
# utf_16_be codec
|
|
410
|
+
'utf_16_be' => 'utf-16be',
|
|
411
|
+
'unicodebigunmarked' => 'utf-16be',
|
|
412
|
+
'utf_16be' => 'utf-16be',
|
|
413
|
+
|
|
414
|
+
# utf_16_le codec
|
|
415
|
+
'utf_16_le' => 'utf-16le',
|
|
416
|
+
'unicodelittleunmarked' => 'utf-16le',
|
|
417
|
+
'utf_16le' => 'utf-16le',
|
|
418
|
+
|
|
419
|
+
# utf_7 codec
|
|
420
|
+
'utf_7' => 'utf-7',
|
|
421
|
+
'u7' => 'utf-7',
|
|
422
|
+
'utf7' => 'utf-7',
|
|
423
|
+
|
|
424
|
+
# utf_8 codec
|
|
425
|
+
'utf_8' => 'utf-8',
|
|
426
|
+
'u8' => 'utf-8',
|
|
427
|
+
'utf' => 'utf-8',
|
|
428
|
+
'utf8' => 'utf-8',
|
|
429
|
+
'utf8_ucs2' => 'utf-8',
|
|
430
|
+
'utf8_ucs4' => 'utf-8',
|
|
431
|
+
}
|
|
432
|
+
end
|
|
@@ -0,0 +1,41 @@
|
|
|
1
|
+
#!/usr/bin/ruby
|
|
2
|
+
|
|
3
|
+
# Add some helper methods to make AttributeList (all of those damn attrs
|
|
4
|
+
# and attrsD used by StrictFeedParser) act more like a Hash.
|
|
5
|
+
# NOTE AttributeList is still Read-Only (AFAICT).
|
|
6
|
+
# Monkey patching is terrible, and I have an addiction.
|
|
7
|
+
module XML
|
|
8
|
+
module SAX
|
|
9
|
+
module AttributeList # in xml/sax.rb
|
|
10
|
+
def [](key)
|
|
11
|
+
getValue(key)
|
|
12
|
+
end
|
|
13
|
+
|
|
14
|
+
def each(&blk)
|
|
15
|
+
(0...getLength).each{|pos| yield [getName(pos), getValue(pos)]}
|
|
16
|
+
end
|
|
17
|
+
|
|
18
|
+
def each_key(&blk)
|
|
19
|
+
(0...getLength).each{|pos| yield getName(pos) }
|
|
20
|
+
end
|
|
21
|
+
|
|
22
|
+
def each_value(&blk)
|
|
23
|
+
(0...getLength).each{|pos| yield getValue(pos) }
|
|
24
|
+
end
|
|
25
|
+
|
|
26
|
+
def to_a # Rather use collect? grep for to_a.collect
|
|
27
|
+
l = []
|
|
28
|
+
each{|k,v| l << [k,v]}
|
|
29
|
+
return l
|
|
30
|
+
end
|
|
31
|
+
|
|
32
|
+
def to_s
|
|
33
|
+
l = []
|
|
34
|
+
each{|k,v| l << "#{k} => #{v}"}
|
|
35
|
+
"{ "+l.join(", ")+" }"
|
|
36
|
+
end
|
|
37
|
+
end
|
|
38
|
+
end
|
|
39
|
+
end
|
|
40
|
+
|
|
41
|
+
|