« back to all changes in this revision
Viewing changes to public/charEncoding.ml
- browse files at revision 1
- compare with another revision
- download diff
- download tarball
- view history from revision 1
- Committer: Bazaar Package Importer
- Author(s): Sylvain Le Gall
- Date: 2005-12-03 01:18:55 UTC
- Revision ID: james.westby@ubuntu.com-20051203011855-qzvwlld1xyqnl62t
Tags: upstream-0.6.3
Import upstream version 0.6.3
- files added:
- charmaps
- internal
- locales
- mappings
- public
- tools
- toolslib
- unidata
- unidata/tr10
added
removed
public/charEncoding.ml
1
(* $Id: charEncoding.ml,v 1.21 2004/11/03 11:21:53 yori Exp $ *)
2
(* Copyright 2001, 2002, 2003 Yamagata Yoriyuki *)
3
4
open OOChannel
5
6
let rec comp_sub_aux s1 i1 s2 i2 len =
7
if len <= 0 then true else
8
if s1.[i1] <> s2.[i2] then false else
9
comp_sub_aux s1 (i1 + 1) s2 (i2 + 1) (len - 1)
10
11
let comp_sub s1 i1 len1 s2 i2 len2 =
12
if len1 <> len2 then false else
13
comp_sub_aux s1 i1 s2 i2 len1
14
15
exception Malformed_code
16
exception Out_of_range
17
18
type 'a machine =
19
{read : 'a -> unit;
20
term : unit -> unit;}
21
22
let feed_string decoder s =
23
for i = 0 to (String.length s) - 1 do
24
decoder.read s.[i]
25
done
26
27
type t =
28
{name : string;
29
make_decoder : UChar.t machine -> char machine;
30
make_encoder : char machine -> UChar.t machine}
31
32
let null_umachine =
33
{read = (fun _ -> ());
34
term = (fun () -> ())}
35
36
type automatic_state =
37
Detection of (t * (char machine)) list
38
| Decoding of char machine
39
40
let accept_char m c =
41
try m.read c; true with Malformed_code -> false
42
43
let automatic name encs def =
44
let make_decoder um =
45
let buf = Buffer.create 0 in
46
let state = ref (Detection
47
(List.map (fun enc -> (enc, enc.make_decoder null_umachine)) encs))
48
in
49
let read c =
50
match !state with
51
Detection candidates ->
52
Buffer.add_char buf c;
53
let cs = List.filter (fun (_, m) -> accept_char m c) candidates in
54
(match cs with
55
[] -> raise Malformed_code
56
| [(enc, _)] ->
57
let m = enc.make_decoder um in
58
feed_string m (Buffer.contents buf);
59
state := Decoding m
60
| _ -> state := Detection cs)
61
| Decoding m -> m.read c
62
in
63
let term () =
64
match !state with
65
Detection ((enc, _) :: _) ->
66
let m = enc.make_decoder um in
67
feed_string m (Buffer.contents buf);
68
m.term ()
69
| Decoding m -> m.term ()
70
| _ -> raise Malformed_code
71
in
72
{read = read; term = term}
73
in
74
{name = name; make_decoder = make_decoder; make_encoder = def.make_encoder}
75
76
let table = Hashtbl.create 0
77
let holded = ref []
78
79
let install name f =
80
if Hashtbl.mem table name then
81
failwith (name ^" has been already installed : CharEncoding.install")
82
else begin
83
Hashtbl.add table name (`P f);
84
end
85
86
let new_enc name enc = install name (fun () -> enc)
87
88
let aliases : (string, string) Hashtbl.t = Hashtbl.create 0
89
90
let alias s1 s2 = Hashtbl.add aliases s1 s2
91
92
(* Charmap based encodings *)
93
module Charmap_enc =
94
struct
95
let of_charmap charmap =
96
let make_decoder um =
97
let no_char = Charmap.no_char_of charmap.Charmap.enc_to_ucs in
98
let start = Charmap.start_probe charmap.Charmap.enc_to_ucs in
99
let state = ref start in
100
let read c =
101
let i = Char.code c in
102
let n = Charmap.look_probe !state i in
103
if n <> no_char then begin
104
state := start;
105
um.read (UChar.chr_of_uint n)
106
end else
107
match Charmap.next_probe !state i with
108
None -> raise Malformed_code
109
| Some next -> state := next in
110
let term () = um.term () in
111
{read = read; term = term}
112
in
113
let make_encoder cm =
114
let read u =
115
let n = UChar.uint_code u in
116
let enc = Tbl31.get charmap.Charmap.ucs_to_enc n in
117
if enc = "" then raise Out_of_range else
118
feed_string cm enc
119
in
120
let term () = cm.term () in
121
{read = read; term = term}
122
in
123
{name = charmap.Charmap.name;
124
make_decoder = make_decoder;
125
make_encoder = make_encoder}
126
end
127
128
let of_charmap = Charmap_enc.of_charmap
129
130
let of_name alias =
131
let name = try Hashtbl.find aliases alias with Not_found -> alias in
132
let rec look name =
133
match Hashtbl.find table name with
134
`P f ->
135
let enc = f () in
136
let b = Weak.create 1 in
137
Weak.set b 0 (Some enc);
138
Hashtbl.add table name (`W b);
139
enc
140
| `W b ->
141
match Weak.get b 0 with
142
None ->
143
Hashtbl.remove table name;
144
look name
145
| Some x -> x in
146
try look name with Not_found ->
147
let charmap = Charmap.of_name name in
148
let enc = of_charmap charmap in
149
let b = Weak.create 1 in
150
Weak.set b 0 (Some enc);
151
Hashtbl.add table name (`W b);
152
enc
153
154
let name_of enc = enc.name
155
156
let buf_machine buf =
157
{read = (fun c -> Buffer.add_char buf c); term = fun () -> ()}
158
159
let recode_string ~in_enc ~out_enc s =
160
let buf = Buffer.create (String.length s) in
161
let encoder = out_enc.make_encoder (buf_machine buf) in
162
let reader = in_enc.make_decoder encoder in
163
feed_string reader s;
164
reader.term ();
165
Buffer.contents buf
166
167
let char_machine_of outchan =
168
let b = String.make 1024 '\000' in
169
let pos = ref 0 in
170
let read c =
171
b.[!pos] <- c;
172
incr pos;
173
if !pos >= 1024 then
174
let n = outchan#output b 0 1024 in
175
String.blit b n b 0 (1024 - n);
176
pos := 1024 - n
177
in
178
let flush () =
179
let n = outchan#output b 0 !pos in
180
if n < !pos then
181
failwith
182
"CharEncoding.char_machine_of: \
183
Cannot flush the entire buffer";
184
outchan#flush ();
185
in
186
let term () =
187
flush ();
188
outchan#close_out ()
189
in
190
{read = read; term = term}, flush
191
192
class uchar_output_channel_of
193
enc
194
(output : char_output_channel)
195
: [UChar.t] obj_output_channel =
196
let c, flush = char_machine_of output in
197
let m = enc.make_encoder c in
198
object
199
method put u = m.read u
200
method close_out = m.term
201
method flush = flush
202
end
203
204
class out_channel enc outchan =
205
uchar_output_channel_of enc (new OOChannel.of_out_channel outchan)
206
207
let queueing q =
208
let read u = Queue.add u q in
209
let term () = () in
210
{read = read; term = term}
211
212
class uchar_input_channel_of enc
213
(input : char_input_channel)
214
: [UChar.t] obj_input_channel =
215
let q = Queue.create () in
216
let b = String.make 1024 '\000' in
217
object(self : 'a)
218
val m = enc.make_decoder (queueing q)
219
val mutable term = false
220
method get() =
221
try Queue.take q with Queue.Empty ->
222
if term then raise End_of_file else
223
(try
224
let len = input#input b 0 1024 in
225
for i = 0 to len - 1 do m.read b.[i] done
226
with End_of_file ->
227
m.term (); term <- true);
228
self#get()
229
method close_in =
230
Queue.clear q; term <- true; input#close_in
231
end
232
233
class in_channel enc inchan =
234
uchar_input_channel_of enc (new OOChannel.of_in_channel inchan)
235
236
class in_channel_from_stream enc s =
237
uchar_input_channel_of enc
238
(new char_input_channel_of
239
(new channel_of_stream s))
240
241
let fill_string s pos len q =
242
begin
243
try while !pos < String.length s do
244
s.[!pos] <- Queue.take q;
245
incr pos;
246
done with Queue.Empty -> ()
247
end;
248
let read c =
249
if !pos < String.length s then begin
250
s.[!pos] <- c;
251
incr pos;
252
end else
253
Queue.add c q
254
in
255
let term () = () in
256
{read = read; term = term}
257
258
class convert_uchar_input enc (uinput : UChar.t obj_input_channel)
259
: char_input_channel =
260
let q = Queue.create () in
261
object (self)
262
val eof = false
263
method input s pos len =
264
let p = ref pos in
265
let m = enc.make_encoder (fill_string s p len q) in
266
try while !p < pos + len do
267
m.read (uinput#get())
268
done;
269
!p - pos
270
with End_of_file ->
271
if eof then raise End_of_file else !p - pos
272
method close_in () =
273
uinput#close_in ();
274
Queue.clear q;
275
end
276
277
class convert_uchar_output enc (uoutput : UChar.t obj_output_channel)
278
: char_output_channel =
279
let m = enc.make_decoder {read = uoutput#put; term = uoutput#close_out} in
280
object (self)
281
method output s pos len =
282
for i = pos to pos + len - 1 do m.read s.[i] done;
283
len;
284
method flush = uoutput#flush
285
method close_out = m.term
286
end
287
288
class convert_input ~in_enc ~out_enc input =
289
convert_uchar_input out_enc (new uchar_input_channel_of in_enc input)
290
291
class convert_output ~in_enc ~out_enc output =
292
convert_uchar_output in_enc (new uchar_output_channel_of out_enc output)
293
294
let ustream_of enc s = stream_of_channel (new in_channel_from_stream enc s)
295
296
class in_char_channel enc input : [char] obj_input_channel =
297
let q = Queue.create () in
298
object(self : 'a)
299
val m = enc.make_encoder (queueing q)
300
val q = q
301
val mutable term = false
302
method get() =
303
try Queue.take q with Queue.Empty ->
304
if term then raise End_of_file else
305
(try let u = input#get() in m.read u with End_of_file ->
306
m.term (); term <- true);
307
self#get()
308
method close_in () =
309
Queue.clear q; term <- true; input#close_in ()
310
end
311
312
class in_char_channel_from_ustream enc us =
313
in_char_channel enc (new channel_of_stream us)
314
315
let char_stream_of enc us =
316
stream_of_channel (new in_char_channel_from_ustream enc us)
317
318
module type Type =
319
sig
320
type text
321
val decode : t -> string -> text
322
val encode : t -> text -> string
323
end
324
325
module Make (Text : UnicodeString.Type) =
326
struct
327
type text = Text.t
328
329
let ubuf_machine ubuf =
330
{read = (fun u -> Text.Buf.add_char ubuf u); term = fun () -> ()}
331
332
let feed_text encoder s =
333
Text.iter encoder.read s
334
335
let decode enc s =
336
let buf = Text.Buf.create 0 in
337
let m = enc.make_decoder (ubuf_machine buf) in
338
feed_string m s;
339
m.term ();
340
Text.Buf.contents buf
341
342
let encode enc s =
343
let buf = Buffer.create (Text.length s) in
344
let m = enc.make_encoder (buf_machine buf) in
345
feed_text m s ;
346
m.term ();
347
Buffer.contents buf
348
end
349
350
(* Ascii *)
351
let _ =
352
let make_decoder output_machine =
353
let output = output_machine.read in
354
let close = output_machine.term in
355
let reader c =
356
let code = Char.code c in
357
if code >= 0x80 then raise Malformed_code else
358
let u = UChar.chr_of_uint code in output u
359
in
360
let term () = close () in
361
{read = reader; term = term}
362
in
363
let make_encoder output_machine =
364
let output = output_machine.read in
365
let close = output_machine.term in
366
let reader u =
367
let code = UChar.uint_code u in
368
if code < 0 || code >= 0x80 then raise Out_of_range else
369
output (Char.chr code)
370
in
371
let term () = close () in
372
{read = reader; term = term}
373
in
374
let enc = fun () ->
375
{name = "US-ASCII";
376
make_decoder = make_decoder;
377
make_encoder = make_encoder;}
378
in begin
379
install "US-ASCII" enc;
380
install "USASCII" enc;
381
install "ASCII" enc;
382
install "ISO646US" enc;
383
install "IBM367" enc;
384
install "CP367" enc;
385
install "ANSI_X3.4-1968" enc;
386
end
387
388
(* ISO-8859-1, Latin-1 *)
389
let _ =
390
let make_decoder output_machine =
391
let output = output_machine.read in
392
let close = output_machine.term in
393
let reader c =
394
let code = Char.code c in
395
let u = UChar.chr_of_uint code in output u
396
in
397
let term () = close () in
398
{read = reader; term = term}
399
in
400
let make_encoder output_machine =
401
let output = output_machine.read in
402
let close = output_machine.term in
403
let reader u =
404
let c =
405
try (UChar.char_of u)
406
with Invalid_argument _ -> raise Out_of_range
407
in
408
output c
409
in
410
let term () = close () in
411
{read = reader; term = term}
412
in
413
let enc = fun () ->
414
{name = "Latin-1";
415
make_decoder = make_decoder;
416
make_encoder = make_encoder;}
417
in begin
418
install "ISO-8859-1" enc;
419
install "Latin-1" enc;
420
end
421
422
(* UTF-8 *)
423
424
type utf8_decoder_state =
425
{mutable remain : int; mutable cur : int}
426
427
let _ =
428
let masq = 0b111111 in
429
let make_decoder output_machine =
430
let output = output_machine.read in
431
let close = output_machine.term in
432
let state = {remain = 0; cur = 0} in
433
let reader c =
434
let i = Char.code c in
435
if state.remain = 0
436
then (*beginning of char*)
437
if i <= 0x7f then output (UChar.chr_of_uint i) else
438
if i >= 0xc2 && i <= 0xdf then
439
begin state.remain <- 1; state.cur <- (i - 0xc0) end
440
else if i >= 0xe0 && i <= 0xef then
441
begin state.remain <- 2; state.cur <- (i - 0xe0) end
442
else if i >= 0xf0 && i <= 0xf7 then
443
begin state.remain <- 3; state.cur <- (i - 0xf0) end
444
else if i >= 0xf8 && i <= 0xfb then
445
begin state.remain <- 4; state.cur <- (i - 0xf8) end
446
else if i >= 0xfc && i <= 0xfd then
447
begin state.remain <- 5; state.cur <- (i - 0xfc) end
448
else raise Malformed_code
449
else
450
if i < 0x80 then raise Malformed_code else
451
let i' = i - 0x80 in
452
let a = (state.cur lsl 6) + i' in
453
(* reject unnecessary long encoding *)
454
if a = 0 then raise Malformed_code else
455
state.remain <- state.remain - 1;
456
if state.remain = 0 then output (UChar.chr_of_uint a) else
457
state.cur <- a
458
in
459
let term () = close () in
460
{read = reader; term = term}
461
in
462
let make_encoder output_machine =
463
let output = output_machine.read in
464
let close = output_machine.term in
465
let reader u =
466
let k = UChar.uint_code u in
467
if k >= 0 && k <= 0x7f then output (Char.chr k) else
468
if k >= 0x80 && k <= 0x7ff then
469
let c0 = Char.chr (0xc0 + (k lsr 6)) in
470
let c1 = Char.chr (0x80 + (k land masq)) in
471
begin output c0; output c1; end
472
else if k >= 0x800 && k <= 0xffff then
473
let c0 = Char.chr (0xe0 + (k lsr 12)) in
474
let c1 = Char.chr (0x80 + ((k lsr 6) land masq)) in
475
let c2 = Char.chr (0x80 + (k land masq)) in
476
begin output c0; output c1; output c2 end
477
else if k >= 0x10000 && k <= 0x1fffff then
478
let c0 = Char.chr (0xf0 + (k lsr 18)) in
479
let c1 = Char.chr (0x80 + ((k lsr 12) land masq)) in
480
let c2 = Char.chr (0x80 + ((k lsr 6) land masq)) in
481
let c3 = Char.chr (0x80 + (k land masq)) in
482
begin output c0; output c1; output c2; output c3 end
483
else if k >= 0x200000 && k <= 0x3ffffff then
484
let c0 = Char.chr (0xf8 + (k lsr 24)) in
485
let c1 = Char.chr (0x80 + ((k lsr 18) land masq)) in
486
let c2 = Char.chr (0x80 + ((k lsr 12) land masq)) in
487
let c3 = Char.chr (0x80 + ((k lsr 6) land masq)) in
488
let c4 = Char.chr (0x80 + (k land masq)) in
489
begin output c0; output c1; output c2; output c3; output c4 end
490
else if k >= 0x4000000 || k < 0 then
491
let c0 = Char.chr (0xfc + (k lsr 30)) in
492
let c1 = Char.chr (0x80 + ((k lsr 24) land masq)) in
493
let c2 = Char.chr (0x80 + ((k lsr 18) land masq)) in
494
let c3 = Char.chr (0x80 + ((k lsr 12) land masq)) in
495
let c4 = Char.chr (0x80 + ((k lsr 6) land masq)) in
496
let c5 = Char.chr (0x80 + (k land masq)) in
497
begin
498
output c0; output c1; output c2; output c3; output c4; output c5
499
end
500
else raise Out_of_range
501
in
502
let term () = close () in
503
{read = reader; term = term}
504
in
505
let enc = fun () ->
506
{name = "UTF-8";
507
make_decoder = make_decoder;
508
make_encoder = make_encoder;}
509
in begin
510
install "UTF-8" enc;
511
end
512
513
(* UTF-16 *)
514
515
type endian = BE | LE | Unknown
516
517
type ucs2_machine =
518
{read2 : int -> unit;
519
term2 : unit -> unit}
520
521
type ucs2_buf =
522
{mutable first_char : bool;
523
mutable buf : int}
524
525
let char_machine_be m =
526
let state = {first_char = true; buf = 0} in
527
let read c =
528
if state.first_char
529
then begin state.buf <- (Char.code c); state.first_char <- false; end
530
else begin
531
m.read2 ((state.buf lsl 8) lor (Char.code c));
532
state.first_char <- true;
533
end
534
in
535
let term () = m.term2 () in
536
{read = read; term = term}
537
538
let char_machine_le m =
539
let state = {first_char = true; buf = 0} in
540
let read c =
541
if state.first_char
542
then begin state.buf <- (Char.code c); state.first_char <- false; end
543
else begin
544
m.read2 (((Char.code c) lsl 8) lor state.buf);
545
state.first_char <- true;
546
end
547
in
548
let term () = m.term2 () in
549
{read = read; term = term}
550
551
let be_outmachine m =
552
let read i =
553
m.read (Char.chr (i lsr 8));
554
m.read (Char.chr (i land 255))
555
in {read2 = read; term2 = m.term}
556
557
let le_outmachine m =
558
let read i =
559
m.read (Char.chr (i land 255));
560
m.read (Char.chr (i lsr 8))
561
in {read2 = read; term2 = m.term}
562
563
type utf16_decoder_state =
564
{mutable surrogated : bool;
565
mutable buf : int}
566
567
let _ =
568
let make_decoder m =
569
let output = m.read in
570
let close = m.term in
571
let state = {surrogated = false; buf = 0} in
572
let read i =
573
if state.surrogated
574
then
575
if i >= 0xdc00 && i <= 0xdfff then
576
let i' = (state.buf - 0xd800) lsl 10 + (i - 0xdc00) + 0x10000 in
577
begin output (UChar.chr_of_uint i'); state.surrogated <- false end
578
else raise Malformed_code
579
else
580
if i = 0xfeff then () else (*BOM is ignored*)
581
if i <= 0xd7ff || (i >= 0xe000 && i <= 0xfffd) then
582
output (UChar.chr_of_uint i)
583
else if i >= 0xd800 && i <= 0xdbff then
584
begin state.surrogated <- true; state.buf <- i end
585
else raise Malformed_code
586
in
587
let term = close in
588
{read2 = read; term2 = term}
589
in
590
let make_encoder m =
591
let output = m.read2 in
592
let close = m.term2 in
593
output 0xfeff;
594
let read u =
595
let i = UChar.uint_code u in
596
if i >= 0x0 && i < 0xd800 || i >= 0xe000 && i < 0x10000
597
then output i
598
else if i >= 0x10000 && i <= 0x10ffff then
599
let i1 = (i - 0x10000) lsr 10 + 0xd800 in
600
let i2 = (i - 0x10000) land 0x3ff + 0xdc00 in
601
output i1; output i2
602
else raise Out_of_range
603
in
604
let term = close in
605
{read = read; term = term}
606
in
607
let make_decoder_be m = char_machine_be (make_decoder m) in
608
let make_encoder_be m = make_encoder (be_outmachine m) in
609
let make_decoder_le m = char_machine_le (make_decoder m) in
610
let make_encoder_le m = make_encoder (le_outmachine m) in
611
612
let enc_be =
613
{name = "UTF-16BE";
614
make_decoder = make_decoder_be;
615
make_encoder = make_encoder_be;}
616
in
617
let enc_le =
618
{name = "UTF-16LE";
619
make_decoder = make_decoder_le;
620
make_encoder = make_encoder_le;}
621
in
622
623
let enc_unknown =
624
fun () -> automatic "UTF-16" [enc_be; enc_le] enc_be in
625
626
begin
627
install "UTF-16BE" (fun () -> enc_be);
628
install "UTF-16LE" (fun () -> enc_le);
629
install "UTF-16" enc_unknown
630
end
631
632
(* UCS-4, UTF-32, UTF-32BE, UTF-32LE*)
633
634
module UTF32 =
635
struct
636
637
type ucs4_machine =
638
{read4 : int -> unit;
639
term4 : unit -> unit}
640
641
type ucs4_buf =
642
{mutable count : int;
643
mutable buf : int}
644
645
let char_machine_be m =
646
let state = {count = 0; buf = 0} in
647
let read c =
648
let i = Char.code c in
649
if state.count = 0 && i > 0x7f then raise Malformed_code else begin
650
state.buf <- (state.buf lsl 8) lor i;
651
state.count <- state.count + 1;
652
if state.count = 4 then begin
653
m.read4 state.buf;
654
state.count <- 0; state.buf <- 0 end
655
else ()
656
end
657
in
658
let term () = m.term4 () in
659
{read = read; term = term}
660
661
let char_machine_le m =
662
let state = {count = 0; buf = 0} in
663
let read c =
664
let i = Char.code c in
665
if state.count = 3 && i > 0x7f then raise Malformed_code else begin
666
state.buf <- (i lsl (8 * state.count)) lor state.buf;
667
state.count <- state.count + 1;
668
if state.count = 4 then begin
669
m.read4 state.buf;
670
state.count <- 0; state.buf <- 0 end
671
else ()
672
end
673
in
674
let term () = m.term4 () in
675
{read = read; term = term}
676
677
let be_outmachine m =
678
let read i =
679
m.read (Char.chr (i lsr 24));
680
m.read (Char.chr ((i lsr 16) land 255));
681
m.read (Char.chr ((i lsr 8) land 255));
682
m.read (Char.chr (i land 255))
683
in {read4 = read; term4 = m.term}
684
685
let le_outmachine m =
686
let read i =
687
m.read (Char.chr (i land 255));
688
m.read (Char.chr ((i lsr 8) land 255));
689
m.read (Char.chr ((i lsr 16) land 255));
690
m.read (Char.chr (i lsr 24))
691
in {read4 = read; term4 = m.term}
692
693
let make_ucs4_decoder m =
694
let read n =
695
m.read (UChar.chr_of_uint n)
696
in
697
let term () = m.term () in
698
{read4 = read; term4 = term}
699
700
let make_ucs4_encoder m =
701
let read u =
702
let n = UChar.uint_code u in
703
m.read4 n
704
in
705
let term () = m.term4 () in
706
{read = read; term = term}
707
708
let make_utf32_decoder m =
709
let read n =
710
if n = 0xfeff then () else (*BOM is ignored*)
711
if n > 0x10ffff || n < 0 || n = 0xfffe then raise Malformed_code else
712
m.read (UChar.chr_of_uint n)
713
in
714
let term () = m.term () in
715
{read4 = read; term4 = term}
716
717
let make_utf32_encoder m =
718
m.read4 0xfeff;
719
let read u =
720
let n = UChar.uint_code u in
721
if n > 0x10ffff || n < 0 || n = 0xfffe then raise Out_of_range else
722
m.read4 n
723
in
724
let term () = m.term4 () in
725
{read = read; term = term}
726
727
let make_utf32_be_decoder m =
728
char_machine_be (make_utf32_decoder m)
729
730
let make_utf32_le_decoder m =
731
char_machine_le (make_utf32_decoder m)
732
733
let make_utf32_be_encoder m =
734
make_utf32_encoder (be_outmachine m)
735
736
let make_utf32_le_encoder m =
737
make_utf32_encoder (le_outmachine m)
738
739
(* 4-bytes encoding for unicode. All 31-bits code points are allowed.
740
* ISO 10646-1 doesn't specify endianess. We use big endian ordering
741
* without BOM.
742
*)
743
let ucs4 =
744
let make_decoder m = char_machine_be (make_ucs4_decoder m) in
745
let make_encoder m = make_ucs4_encoder (be_outmachine m) in
746
{name = "UCS-4";
747
make_decoder = make_decoder;
748
make_encoder = make_encoder}
749
750
(* The same as UCS-4, but range is limited to 21-bits code points. *)
751
let utf32_be =
752
let make_decoder m = make_utf32_be_decoder m in
753
let make_encoder m = make_utf32_be_encoder m in
754
{name = "UTF-32BE";
755
make_decoder = make_decoder;
756
make_encoder = make_encoder}
757
758
(* Little endian order. *)
759
let utf32_le =
760
let make_decoder m = make_utf32_le_decoder m in
761
let make_encoder m = make_utf32_le_encoder m in
762
{name = "UTF-32LE";
763
make_decoder = make_decoder;
764
make_encoder = make_encoder}
765
766
(* UTF-32 with unknown endianness. *)
767
let utf32 = fun () ->
768
automatic "UTF-32" [utf32_be; utf32_le] utf32_be
769
770
let init () =
771
install "UCS-4" (fun () -> ucs4);
772
install "UTF-32BE" (fun () -> utf32_be);
773
install "UTF-32LE" (fun () -> utf32_le);
774
install "UTF-32" utf32
775
776
end
777
778
let _ = UTF32.init ()
779
780
(* iso-2022-jp *)
781
782
let enc_to_ucs map enc =
783
let no_char = Unimap.no_char_ucs map in
784
let ucs = Unimap.enc_to_ucs map enc in
785
if ucs = no_char then raise Malformed_code else ucs
786
787
let ucs_to_enc map enc =
788
let no_char = Unimap.no_char_enc map in
789
let enc = Unimap.ucs_to_enc map enc in
790
if enc = no_char then raise Out_of_range else enc
791
792
module Iso2022jp =
793
struct
794
795
type charset = Ascii | Jisx0208_1978 | Jisx0208_1983 | Jisx0201_roman
796
| Jisx0201_kana | Gb2312_1980 | Ksc5601_1987 | Jisx0212_1990
797
| Iso88591 | Iso88597 | Unknown
798
799
let unibyte charset =
800
match charset with
801
Ascii -> true | Jisx0201_roman -> true | Jisx0201_kana -> true
802
| Iso88591 -> true | Iso88597 -> true
803
| Jisx0208_1978 -> false | Jisx0208_1983 -> false |Jisx0212_1990 -> false
804
| Gb2312_1980 -> false | Ksc5601_1987 -> false | Unknown -> assert false
805
806
let set94 charset =
807
match charset with
808
Ascii -> true | Jisx0201_roman -> true | Jisx0201_kana -> true
809
| Iso88591 -> false | Iso88597 -> false
810
| Jisx0208_1978 -> true | Jisx0208_1983 -> true |Jisx0212_1990 -> true
811
| Gb2312_1980 -> true | Ksc5601_1987 -> true | Unknown -> assert false
812
813
let esc = Char.chr 0x1b (* \027 *)
814
815
type state_type =
816
{mutable remain : int; (* <0 means escape sequence*)
817
mutable g0 : charset;
818
mutable g2 : charset;
819
mutable single_shift : bool;
820
buf : string}
821
822
let make_enc_2 () =
823
let gb2312 = Unimap.of_name "gb2312" in
824
let jisx0201 = Unimap.of_name "jisx0201" in
825
let jisx0208 = Unimap.of_name "jisx0208" in
826
let jisx0212 = Unimap.of_name "jisx0212" in
827
let ksc5601 = Unimap.of_name "ksc5601" in
828
let iso88597 = Unimap.of_name "iso88597" in
829
830
let make_decoder_2 m =
831
let to_uchar_2 charset code =
832
let ucode =
833
match charset with
834
Ascii -> code
835
| Jisx0208_1978 -> enc_to_ucs jisx0208 code
836
| Jisx0208_1983 -> enc_to_ucs jisx0208 code
837
| Jisx0201_roman -> enc_to_ucs jisx0201 code
838
| Jisx0201_kana -> enc_to_ucs jisx0201 (code + 0x80)
839
| Gb2312_1980 -> enc_to_ucs gb2312 code
840
| Ksc5601_1987 -> enc_to_ucs ksc5601 code
841
| Jisx0212_1990 -> enc_to_ucs jisx0212 code
842
| Iso88591 -> code + 0x80
843
| Iso88597 -> enc_to_ucs iso88597 (code + 0x80)
844
| Unknown -> raise Malformed_code
845
in UChar.chr_of_uint ucode
846
in
847
let state =
848
{remain = 0;
849
g0 = Ascii;
850
g2 = Unknown;
851
single_shift = false;
852
buf = String.create 3}
853
in
854
let reader c =
855
let i = Char.code c in
856
if i >= 0x80 then raise Malformed_code else
857
if state.single_shift
858
then
859
if i >= 0x20 && i <= 0x7f
860
then
861
begin
862
m.read (to_uchar_2 state.g2 i);
863
state.single_shift <- false
864
end
865
else raise Malformed_code
866
else
867
match state.remain with
868
0 ->
869
(* escape sequence?. *)
870
if c = esc then state.remain <- -1 else
871
(* control code or white space, backspace *)
872
if i <= 0x20 || i = 0x7f then m.read (UChar.chr_of_uint i) else
873
if unibyte state.g0
874
then m.read (to_uchar_2 state.g0 i)
875
else begin state.remain <- 1; state.buf.[0] <- c end
876
| 1 ->
877
let i1 = Char.code state.buf.[0] in
878
let i2 = Char.code c in
879
if i2 >= 0x21 && i2 <= 0x7e
880
then
881
let u = to_uchar_2 state.g0 ((i1 lsl 8) + i2) in
882
m.read u
883
else raise Malformed_code;
884
state.remain <- 0
885
| _ -> (*escape seq.*)
886
state.buf.[~-(state.remain) - 1] <- c;
887
state.remain <- state.remain - 1;
888
let i = Char.code c in
889
let len = ~- (state.remain) - 1 in
890
if i >= 0x20 && i <= 0x2f then
891
if len >= 3 then raise Malformed_code else ()
892
else if i >= 0x30 && i <= 0x7e then begin
893
state.remain <- 0;
894
if comp_sub "(B" 0 2 state.buf 0 len then
895
state.g0 <- Ascii
896
else if comp_sub "(I" 0 2 state.buf 0 len then
897
state.g0 <- Jisx0201_kana
898
else if comp_sub "(J" 0 2 state.buf 0 len then
899
state.g0 <- Jisx0201_roman
900
else if comp_sub "$@" 0 2 state.buf 0 len then
901
state.g0 <- Jisx0208_1978
902
else if comp_sub "$A" 0 2 state.buf 0 len then
903
state.g0 <- Gb2312_1980
904
else if comp_sub "$B" 0 2 state.buf 0 len then
905
state.g0 <- Jisx0208_1983
906
else if comp_sub "$(C" 0 3 state.buf 0 len then
907
state.g0 <- Ksc5601_1987
908
else if comp_sub "$(D" 0 3 state.buf 0 len then
909
state.g0 <- Jisx0212_1990
910
else if comp_sub ".A" 0 2 state.buf 0 len then
911
state.g2 <- Iso88591
912
else if comp_sub ".F" 0 2 state.buf 0 len then
913
state.g2 <- Iso88597
914
else if comp_sub "N" 0 1 state.buf 0 len then
915
state.single_shift <- true
916
else raise Malformed_code;
917
end else raise Malformed_code
918
in
919
{read = reader; term = m.term} in
920
921
let make_encoder_2 m =
922
let from_uchar_2 u =
923
let n = UChar.code u in
924
(* the order of tests is *significant* *)
925
if n >= 0x00 && n <= 0x7f then Ascii, n else
926
if n >= 0xa0 && n <= 0xff then Iso88591, (n - 0x80) else
927
let enc = Unimap.ucs_to_enc jisx0208 n in
928
if enc <> Unimap.no_char_enc jisx0208 then (Jisx0208_1983, enc) else
929
let enc = Unimap.ucs_to_enc jisx0212 n in
930
if enc <> Unimap.no_char_enc jisx0212 then (Jisx0212_1990, enc) else
931
let enc = Unimap.ucs_to_enc gb2312 n in
932
if enc <> Unimap.no_char_enc gb2312 then (Gb2312_1980, enc) else
933
let enc = Unimap.ucs_to_enc ksc5601 n in
934
if enc <> Unimap.no_char_enc ksc5601 then (Ksc5601_1987, enc) else
935
let enc = Unimap.ucs_to_enc iso88597 n in
936
if enc <> Unimap.no_char_enc iso88597 then (Iso88597, enc - 0x80) else
937
let n' = Unimap.ucs_to_enc jisx0201 n in
938
if n' = Unimap.no_char_enc jisx0201 then raise Out_of_range else
939
if n' >= 0x80 then Jisx0201_kana, (n' - 0x80) else Jisx0201_roman, n'
940
in
941
(*fields other than g0, g2 are not used*)
942
let state =
943
{remain = 0;
944
g0 = Ascii;
945
g2 = Unknown;
946
single_shift = false;
947
buf = String.create 1}
948
in
949
let reader u =
950
let cs, i = from_uchar_2 u in
951
let c1 = Char.chr (i lsr 8) in
952
let c2 = Char.chr (i land 255) in
953
if c1 = esc || c2 = esc then raise Out_of_range else
954
if set94 cs
955
then
956
if state.g0 = cs
957
then
958
if unibyte cs
959
then m.read c2
960
else begin m.read c1; m.read c2 end
961
else
962
(match cs with
963
Ascii ->
964
state.g0 <- Ascii;
965
m.read esc; m.read '('; m.read 'B'; m.read c2
966
| Jisx0201_kana ->
967
state.g0 <- Jisx0201_kana;
968
m.read esc; m.read '('; m.read 'I'; m.read c2
969
| Jisx0201_roman ->
970
state.g0 <- Jisx0201_roman;
971
m.read esc; m.read '('; m.read 'J'; m.read c2
972
| Gb2312_1980 ->
973
state.g0 <- Gb2312_1980;
974
m.read esc; m.read '$'; m.read 'A'; m.read c1; m.read c2
975
| Jisx0208_1983 ->
976
state.g0 <- Jisx0208_1983;
977
m.read esc; m.read '$'; m.read 'B'; m.read c1; m.read c2
978
| Ksc5601_1987 ->
979
state.g0 <- Ksc5601_1987;
980
m.read esc; m.read '$'; m.read '('; m.read 'C';
981
m.read c1; m.read c2
982
| Jisx0212_1990 ->
983
state.g0 <- Jisx0212_1990;
984
m.read esc; m.read '$'; m.read '('; m.read 'D';
985
m.read c1; m.read c2
986
| _ -> assert false)
987
else (*set96*)
988
(* since RFC allows G2 is cleared in the begining of lines,
989
* we set G2 every times. *)
990
(match cs with
991
Iso88591 ->
992
m.read esc; m.read '.'; m.read 'A';
993
m.read esc; m.read 'N'; m.read c2
994
| Iso88597 ->
995
m.read esc; m.read '.'; m.read 'F';
996
m.read esc; m.read 'N'; m.read c2
997
| _ -> assert false)
998
in
999
let term () =
1000
if state.g0 = Ascii then () else
1001
begin m.read esc; m.read '('; m.read 'B' end;
1002
m.term ()
1003
in
1004
{read = reader; term = term} in
1005
1006
{name = "ISO-2022-JP-2";
1007
make_decoder = make_decoder_2;
1008
make_encoder = make_encoder_2}
1009
1010
let make_enc () =
1011
let jisx0201 = Unimap.of_name "jisx0201" in
1012
let jisx0208 = Unimap.of_name "jisx0208" in
1013
1014
let make_decoder m =
1015
let to_uchar charset code =
1016
let ucode =
1017
match charset with
1018
Ascii -> code
1019
| Jisx0208_1978 -> enc_to_ucs jisx0208 code
1020
| Jisx0208_1983 -> enc_to_ucs jisx0208 code
1021
| Jisx0201_roman -> enc_to_ucs jisx0201 code
1022
| _ -> raise Malformed_code
1023
in UChar.chr_of_uint ucode
1024
in
1025
let state =
1026
{remain = 0;
1027
g0 = Ascii;
1028
g2 = Unknown;
1029
single_shift = false;
1030
buf = String.create 3}
1031
in
1032
let reader c =
1033
let i = Char.code c in
1034
if i >= 0x80 then raise Malformed_code else
1035
match state.remain with
1036
0 ->
1037
(* escape sequence?. *)
1038
if c = esc then state.remain <- -1 else
1039
(* control code or white space, backspace *)
1040
if i <= 0x20 || i = 0x7f then m.read (UChar.chr_of_uint i) else
1041
if unibyte state.g0
1042
then m.read (to_uchar state.g0 i)
1043
else begin state.remain <- 1; state.buf.[0] <- c end
1044
| 1 ->
1045
let i1 = Char.code state.buf.[0] in
1046
let i2 = Char.code c in
1047
if i2 >= 0x21 && i2 <= 0x7e
1048
then
1049
let u = to_uchar state.g0 ((i1 lsl 8) + i2) in
1050
m.read u
1051
else raise Malformed_code;
1052
state.remain <- 0
1053
| _ -> (*escape seq.*)
1054
state.buf.[~-(state.remain) - 1] <- c;
1055
state.remain <- state.remain - 1;
1056
let len = ~- (state.remain) - 1 in
1057
if i >= 0x20 && i <= 0x2f
1058
then
1059
begin if len >= 3 then raise Malformed_code else () end
1060
else if i >= 0x30 && i <= 0x7e then
1061
begin
1062
state.remain <- 0;
1063
if comp_sub "(B" 0 2 state.buf 0 len then
1064
state.g0 <- Ascii
1065
else if comp_sub "(J" 0 2 state.buf 0 len then
1066
state.g0 <- Jisx0201_roman
1067
else if comp_sub "$@" 0 2 state.buf 0 len then
1068
state.g0 <- Jisx0208_1978
1069
else if comp_sub "$B" 0 2 state.buf 0 len then
1070
state.g0 <- Jisx0208_1983
1071
else raise Malformed_code
1072
end
1073
else raise Malformed_code;
1074
in
1075
{read = reader; term = m.term} in
1076
1077
let make_encoder m =
1078
let from_uchar u =
1079
let n = UChar.code u in
1080
(* the order of tests is *significant* *)
1081
if n >= 0x00 && n <= 0x7f then Ascii, n else
1082
let enc = Unimap.ucs_to_enc jisx0208 n in
1083
if enc <> Unimap.no_char_enc jisx0208 then (Jisx0208_1983, enc) else
1084
let n' = Unimap.ucs_to_enc jisx0201 n in
1085
if n' = Unimap.no_char_enc jisx0201 then raise Out_of_range else
1086
if n' >= 0x80 then raise Out_of_range else Jisx0201_roman, n'
1087
in
1088
(*fields other than g0, g2 are not used*)
1089
let state =
1090
{remain = 0;
1091
g0 = Ascii;
1092
g2 = Unknown;
1093
single_shift = false;
1094
buf = String.create 1}
1095
in
1096
let reader u =
1097
if UChar.uint_code u = 27 then raise Out_of_range else
1098
let cs, i = from_uchar u in
1099
let c1 = Char.chr (i lsr 8) in
1100
let c2 = Char.chr (i land 255) in
1101
if state.g0 = cs
1102
then
1103
if unibyte cs
1104
then m.read c2
1105
else begin m.read c1; m.read c2 end
1106
else
1107
(match cs with
1108
Ascii ->
1109
state.g0 <- Ascii;
1110
m.read esc; m.read '('; m.read 'B'; m.read c2
1111
| Jisx0201_roman ->
1112
state.g0 <- Jisx0201_roman;
1113
m.read esc; m.read '('; m.read 'J'; m.read c2
1114
| Jisx0208_1983 ->
1115
state.g0 <- Jisx0208_1983;
1116
m.read esc; m.read '$'; m.read 'B'; m.read c1; m.read c2
1117
| _ -> assert false)
1118
in
1119
let term () =
1120
if state.g0 = Ascii then () else
1121
begin m.read esc; m.read '('; m.read 'B' end;
1122
m.term ()
1123
in
1124
{read = reader; term = term} in
1125
1126
{name = "ISO-2022-JP";
1127
make_decoder = make_decoder;
1128
make_encoder = make_encoder}
1129
1130
let init () =
1131
install "ISO-2022-JP-2" make_enc_2;
1132
install "csISO2022JP2" make_enc_2;
1133
install "ISO-2022-JP" make_enc;
1134
install "csISO2022JP" make_enc
1135
end
1136
1137
let _ = Iso2022jp.init ()
1138
1139
(* Japanese auto detect *)
1140
1141
module J_auto =
1142
struct
1143
1144
let jauto =
1145
fun () ->
1146
let euc_jp = of_name "EUC-JP" in
1147
let sjis = of_name "SHIFT_JIS" in
1148
let iso2022jp = of_name "ISO-2022-JP-2" in
1149
automatic
1150
"japanese_auto_detection"
1151
[iso2022jp; euc_jp; sjis]
1152
euc_jp
1153
1154
let init () =
1155
let _ = install "japanese_auto_detection" jauto in
1156
let _ = install "jauto" jauto in ()
1157
1158
end
1159
1160
let _ = J_auto.init ()
1161
1162
(* iso-2022-kr *)
1163
1164
module Iso2022kr =
1165
struct
1166
1167
let esc = Char.chr 0x1b (* \027 *)
1168
let si = Char.chr 0x0f
1169
let so = Char.chr 0x0e
1170
1171
type charset = Ascii | Ksc5601
1172
1173
type state_type =
1174
{mutable remain : int; (* <0 means escape sequence*)
1175
mutable gl : charset;
1176
buf : string}
1177
1178
let make_enc () =
1179
let ksc5601 = Unimap.of_name "ksc5601" in
1180
1181
let make_decoder m =
1182
let state =
1183
{remain = 0;
1184
gl = Ascii;
1185
buf = String.create 3}
1186
in
1187
let reader c =
1188
let i = Char.code c in
1189
if i >= 0x80 then raise Malformed_code else
1190
match state.remain with
1191
0 ->
1192
if c = esc then state.remain <- -1 else
1193
if c = si then state.gl <- Ascii else
1194
if c = so then state.gl <- Ksc5601 else
1195
(* control code or white space, backspace *)
1196
if i <= 0x20 || i = 0x7f then m.read (UChar.chr_of_uint i) else
1197
(match state.gl with
1198
Ascii -> m.read (UChar.chr_of_uint i)
1199
| Ksc5601 -> state.remain <- 1; state.buf.[0] <- c)
1200
| 1 ->
1201
let i1 = Char.code state.buf.[0] in
1202
let i2 = Char.code c in
1203
if i2 >= 0x21 && i2 <= 0x7e
1204
then
1205
let n = enc_to_ucs ksc5601 ((i1 lsl 8) + i2) in
1206
m.read (UChar.chr_of_uint n)
1207
else raise Malformed_code;
1208
state.remain <- 0
1209
| _ -> (*escape seq.*)
1210
state.buf.[~-(state.remain) - 1] <- c;
1211
state.remain <- state.remain - 1;
1212
let i = Char.code c in
1213
let len = ~- (state.remain) - 1 in
1214
if i >= 0x20 && i <= 0x2f then
1215
if len >= 3 then raise Malformed_code else ()
1216
else if i >= 0x30 && i <= 0x7e then
1217
begin
1218
state.remain <- 0;
1219
(* designator *)
1220
if comp_sub "$)C" 0 3 state.buf 0 len then ()
1221
else raise Malformed_code
1222
end
1223
else raise Malformed_code
1224
in
1225
{read = reader; term = m.term} in
1226
1227
let make_encoder m =
1228
(*fields other than gl are not used*)
1229
let state =
1230
{remain = 0;
1231
gl = Ascii;
1232
buf = String.create 1}
1233
in
1234
feed_string m "\027$)C"; (*designator*)
1235
let reader u =
1236
let n = UChar.code u in
1237
if n <= 0x7f then
1238
if n = 0x0e || n = 0x0f || n = 0x1b then raise Out_of_range else
1239
match state.gl with
1240
Ascii -> m.read (Char.chr n)
1241
| Ksc5601 ->
1242
state.gl <- Ascii;
1243
m.read si; m.read (Char.chr n)
1244
else
1245
let i = Unimap.ucs_to_enc ksc5601 n in
1246
if i = Unimap.no_char_enc ksc5601 then raise Out_of_range else
1247
let c1 = Char.chr (i lsr 8) in
1248
let c2 = Char.chr (i land 255) in
1249
match state.gl with
1250
Ascii ->
1251
state.gl <- Ksc5601;
1252
m.read so; m.read c1; m.read c2
1253
| Ksc5601 ->
1254
m.read c1; m.read c2
1255
in
1256
let term () =
1257
(if state.gl = Ksc5601 then m.read si else ());
1258
m.term ()
1259
in
1260
{read = reader; term = term} in
1261
1262
{name = "ISO-2022-KR";
1263
make_decoder = make_decoder;
1264
make_encoder = make_encoder}
1265
1266
let init () =
1267
let _ = install "ISO-2022-KR" make_enc in
1268
let _ = install "csISO2022KR" make_enc in ()
1269
end
1270
1271
let _ = Iso2022kr.init ()
1272
1273
(* iso-2022-cn *)
1274
1275
module Iso2022cn =
1276
struct
1277
1278
let esc = Char.chr 0x1b (* \027 *)
1279
let si = Char.chr 0x0f
1280
let so = Char.chr 0x0e
1281
let uLF = UChar.chr_of_uint (Char.code '\n')
1282
1283
let cns_p1 = 0x10000
1284
let cns_p2 = 0x20000
1285
1286
type charset = Ascii | GB_2312 | CNS_11643_p1 | CNS_11643_p2 | Unknown
1287
1288
type assign = G0 | G1 | G2
1289
1290
type state_type =
1291
{mutable remain : int; (* <0 means escape sequence*)
1292
mutable gl : assign;
1293
mutable g1 : charset;
1294
mutable g2 : charset;
1295
mutable single_shift : bool;
1296
buf : string}
1297
1298
let make_enc () =
1299
let gb2312 = Unimap.of_name "gb2312" in
1300
let cns11643 = Unimap.of_name "cns11643" in
1301
1302
let make_decoder m =
1303
let to_ucs4 cs n =
1304
match cs with
1305
Ascii -> if n <= 0x7f then n else raise Malformed_code
1306
| GB_2312 -> enc_to_ucs gb2312 n
1307
| CNS_11643_p1 -> enc_to_ucs cns11643 (cns_p1 lor n)
1308
| CNS_11643_p2 -> enc_to_ucs cns11643 (cns_p2 lor n)
1309
| Unknown -> raise Malformed_code
1310
in
1311
let state =
1312
{remain = 0;
1313
gl = G0;
1314
g1 = Unknown;
1315
g2 = Unknown;
1316
single_shift = false;
1317
buf = String.create 3}
1318
in
1319
let reader c =
1320
let i = Char.code c in
1321
if i >= 0x80 then raise Malformed_code else
1322
match state.remain with
1323
0 ->
1324
if not state.single_shift then
1325
match c with
1326
'\027' -> state.remain <- ~-1
1327
| '\015' -> state.gl <- G0
1328
| '\014' -> state.gl <- G1
1329
| _ ->
1330
(* control code *)
1331
if i <= 0x20 then m.read (UChar.chr_of_uint i) else
1332
(match state.gl with
1333
G0 -> m.read (UChar.chr_of_uint i)
1334
| _ -> state.remain <- 1; state.buf.[0] <- c)
1335
else
1336
state.remain <- 1; state.buf.[0] <- c
1337
| 1 ->
1338
let i1 = Char.code state.buf.[0] in
1339
let i2 = Char.code c in
1340
let n = (i1 lsl 8) + i2 in
1341
let cs =
1342
if state.single_shift then state.g2 else
1343
match state.gl with
1344
G0 -> Ascii
1345
| G1 -> state.g1
1346
| G2 -> state.g2
1347
in
1348
m.read (UChar.chr_of_uint (to_ucs4 cs n));
1349
state.remain <- 0;
1350
if state.single_shift then state.single_shift <- false else ()
1351
| 2 -> state.remain <- 1; state.buf.[0] <- c
1352
| _ -> (*escape seq.*)
1353
state.buf.[~-(state.remain) - 1] <- c;
1354
state.remain <- state.remain - 1;
1355
let i = Char.code c in
1356
let len = ~- (state.remain) - 1 in
1357
if i >= 0x20 && i <= 0x2f then
1358
if len >= 3 then raise Malformed_code else ()
1359
else if i >= 0x30 && i <= 0x7e then
1360
begin
1361
state.remain <- 0;
1362
if comp_sub "$)A" 0 3 state.buf 0 len then
1363
state.g1 <- GB_2312
1364
else if comp_sub "$)G" 0 3 state.buf 0 len then
1365
state.g1 <- CNS_11643_p1
1366
else if comp_sub "$*H" 0 3 state.buf 0 len then
1367
state.g2 <- CNS_11643_p2
1368
else if comp_sub "N" 0 1 state.buf 0 len then
1369
begin state.single_shift <- true; state.remain <- 2 end
1370
else raise Malformed_code
1371
end
1372
else raise Malformed_code
1373
in
1374
{read = reader; term = m.term} in
1375
1376
let make_encoder m =
1377
let from_ucs4 n =
1378
if n <= 0x7f then Ascii, n else
1379
let n' = Unimap.ucs_to_enc gb2312 n in
1380
if n' <> Unimap.no_char_enc gb2312 then (GB_2312, n') else
1381
let n' = Unimap.ucs_to_enc cns11643 n in
1382
if n' = Unimap.no_char_enc cns11643 then raise Out_of_range else
1383
match n' lsr 16 with
1384
1 -> CNS_11643_p1, (n' land 0xffff)
1385
| 2 -> CNS_11643_p2, (n' land 0xffff)
1386
| _ -> raise Out_of_range
1387
in
1388
(*fields other than gl, g1, g2 are not used*)
1389
let state =
1390
{remain = 0;
1391
gl = G0;
1392
g1 = Unknown;
1393
g2 = Unknown;
1394
buf = String.create 1;
1395
single_shift = false;
1396
}
1397
in
1398
let reader u =
1399
let n = UChar.code u in
1400
if n = 0x0e || n = 0x0f || n = 0x1b then raise Out_of_range else begin
1401
if n = 0x0d || n = 0x0a || n = 0x85 then begin
1402
state.g1 <- Unknown;
1403
state.g2 <- Unknown;
1404
end else ();
1405
let cs, n' = from_ucs4 n in
1406
let i1 = n' lsr 8 in
1407
let i2 = n' land 255 in
1408
let cur_cs =
1409
match state.gl with
1410
G0 -> Ascii
1411
| G1 -> state.g1
1412
| G2 -> state.g2
1413
in
1414
if cs = cur_cs then
1415
match cs with
1416
Ascii -> m.read (Char.chr i2)
1417
| _ -> m.read (Char.chr i1); m.read (Char.chr i2)
1418
else
1419
match cs with
1420
Ascii ->
1421
m.read si;
1422
m.read (Char.chr i2);
1423
state.gl <- G0
1424
| GB_2312 ->
1425
if not (state.g1 = GB_2312) then begin
1426
m.read esc; feed_string m "$)A";
1427
state.g1 <- GB_2312;
1428
end else ();
1429
if not (state.gl = G1) then begin
1430
m.read so;
1431
state.gl <- G1
1432
end else ();
1433
m.read (Char.chr i1); m.read (Char.chr i2)
1434
| CNS_11643_p1 ->
1435
if not (state.g1 = CNS_11643_p1) then begin
1436
m.read esc; feed_string m "$)G";
1437
state.g1 <- CNS_11643_p1;
1438
end else ();
1439
if not (state.gl = G1) then begin
1440
m.read so; state.gl <- G1
1441
end else ();
1442
m.read (Char.chr i1); m.read (Char.chr i2)
1443
| CNS_11643_p2 ->
1444
if not (state.g2 = CNS_11643_p2) then begin
1445
m.read esc; feed_string m "$*H";
1446
state.g2 <- CNS_11643_p2;
1447
end else ();
1448
m.read esc; m.read 'N';
1449
m.read (Char.chr i1); m.read (Char.chr i2)
1450
| _ -> assert false
1451
end
1452
in
1453
let term () =
1454
if state.gl = G0 then () else m.read si;
1455
m.term ();
1456
in {read = reader; term = term} in
1457
1458
{name = "ISO-2022-CN";
1459
make_decoder = make_decoder;
1460
make_encoder = make_encoder}
1461
1462
let init () =
1463
let _ = install "ISO-2022-CN" make_enc in ()
1464
1465
end
1466
1467
let _ = Iso2022cn.init ()
1468
1469
(* shortcuts *)
1470
let ascii = of_name "US-ASCII"
1471
let latin1 = of_name "Latin-1"
1472
let utf8 = of_name "UTF-8"
1473
let utf16 = of_name "UTF-16"
1474
let utf16be = of_name "UTF-16BE"
1475
let utf16le = of_name "UTF-16LE"
1476
let utf32 = of_name "UTF-32"
1477
let utf32be = of_name "UTF-32BE"
1478
let utf32le = of_name "UTF-32LE"
1479
let ucs4 = of_name "UCS-4"
1480
1481
(* IANA *)
1482
(* Name: ANSI_X3.4-1968 [RFC1345,KXS2] *)
1483
(* Alias: iso-ir-6 *)
1484
(* Alias: ANSI_X3.4-1986 *)
1485
(* Alias: ISO_646.irv:1991 *)
1486
(* Alias: ASCII *)
1487
(* Alias: ISO646-US *)
1488
(* Alias: US-ASCII (preferred MIME name) *)
1489
(* Alias: us *)
1490
(* Alias: IBM367 *)
1491
(* Alias: cp367 *)
1492
(* Alias: csASCII *)
1493
let () = alias "IANA/csASCII" "ANSI_X3.4-1968"
1494
let () = alias "IANA/cp367" "ANSI_X3.4-1968"
1495
let () = alias "IANA/IBM367" "ANSI_X3.4-1968"
1496
let () = alias "IANA/us" "ANSI_X3.4-1968"
1497
let () = alias "IANA/US-ASCII" "ANSI_X3.4-1968"
1498
let () = alias "IANA/ISO646-US" "ANSI_X3.4-1968"
1499
let () = alias "IANA/ASCII" "ANSI_X3.4-1968"
1500
let () = alias "IANA/ISO_646.irv:1991" "ANSI_X3.4-1968"
1501
let () = alias "IANA/ANSI_X3.4-1986" "ANSI_X3.4-1968"
1502
let () = alias "IANA/iso-ir-6" "ANSI_X3.4-1968"
1503
let () = alias "IANA/ANSI_X3.4-1968" "ANSI_X3.4-1968"
1504
(* Name: ISO-10646-UTF-1 *)
1505
(* Alias: csISO10646UTF1 *)
1506
(* Name: ISO_646.basic:1983 [RFC1345,KXS2] *)
1507
(* Alias: ref *)
1508
(* Alias: csISO646basic1983 *)
1509
let () = alias "IANA/csISO646basic1983" "ISO_646.BASIC"
1510
let () = alias "IANA/ref" "ISO_646.BASIC"
1511
let () = alias "IANA/ISO_646.basic:1983" "ISO_646.BASIC"
1512
(* Name: INVARIANT [RFC1345,KXS2] *)
1513
(* Alias: csINVARIANT *)
1514
let () = alias "IANA/csINVARIANT" "INVARIANT"
1515
let () = alias "IANA/INVARIANT" "INVARIANT"
1516
(* Name: ISO_646.irv:1983 [RFC1345,KXS2] *)
1517
(* Alias: iso-ir-2 *)
1518
(* Alias: irv *)
1519
(* Alias: csISO2IntlRefVersion *)
1520
let () = alias "IANA/csISO2IntlRefVersion" "ISO_646.IRV"
1521
let () = alias "IANA/irv" "ISO_646.IRV"
1522
let () = alias "IANA/iso-ir-2" "ISO_646.IRV"
1523
let () = alias "IANA/ISO_646.irv:1983" "ISO_646.IRV"
1524
(* Name: BS_4730 [RFC1345,KXS2] *)
1525
(* Alias: iso-ir-4 *)
1526
(* Alias: ISO646-GB *)
1527
(* Alias: gb *)
1528
(* Alias: uk *)
1529
(* Alias: csISO4UnitedKingdom *)
1530
let () = alias "IANA/csISO4UnitedKingdom" "BS_4730"
1531
let () = alias "IANA/uk" "BS_4730"
1532
let () = alias "IANA/gb" "BS_4730"
1533
let () = alias "IANA/ISO646-GB" "BS_4730"
1534
let () = alias "IANA/iso-ir-4" "BS_4730"
1535
let () = alias "IANA/BS_4730" "BS_4730"
1536
(* Name: NATS-SEFI [RFC1345,KXS2] *)
1537
(* Alias: iso-ir-8-1 *)
1538
(* Alias: csNATSSEFI *)
1539
let () = alias "IANA/csNATSSEFI" "NATS-SEFI"
1540
let () = alias "IANA/iso-ir-8-1" "NATS-SEFI"
1541
let () = alias "IANA/NATS-SEFI" "NATS-SEFI"
1542
(* Name: NATS-SEFI-ADD [RFC1345,KXS2] *)
1543
(* Alias: iso-ir-8-2 *)
1544
(* Alias: csNATSSEFIADD *)
1545
(* Name: NATS-DANO [RFC1345,KXS2] *)
1546
(* Alias: iso-ir-9-1 *)
1547
(* Alias: csNATSDANO *)
1548
let () = alias "IANA/csNATSDANO" "NATS-DANO"
1549
let () = alias "IANA/iso-ir-9-1" "NATS-DANO"
1550
let () = alias "IANA/NATS-DANO" "NATS-DANO"
1551
(* Name: NATS-DANO-ADD [RFC1345,KXS2] *)
1552
(* Alias: iso-ir-9-2 *)
1553
(* Alias: csNATSDANOADD *)
1554
(* Name: SEN_850200_B [RFC1345,KXS2] *)
1555
(* Alias: iso-ir-10 *)
1556
(* Alias: FI *)
1557
(* Alias: ISO646-FI *)
1558
(* Alias: ISO646-SE *)
1559
(* Alias: se *)
1560
(* Alias: csISO10Swedish *)
1561
let () = alias "IANA/csISO10Swedish" "SEN_850200_B"
1562
let () = alias "IANA/se" "SEN_850200_B"
1563
let () = alias "IANA/ISO646-SE" "SEN_850200_B"
1564
let () = alias "IANA/ISO646-FI" "SEN_850200_B"
1565
let () = alias "IANA/FI" "SEN_850200_B"
1566
let () = alias "IANA/iso-ir-10" "SEN_850200_B"
1567
let () = alias "IANA/SEN_850200_B" "SEN_850200_B"
1568
(* Name: SEN_850200_C [RFC1345,KXS2] *)
1569
(* Alias: iso-ir-11 *)
1570
(* Alias: ISO646-SE2 *)
1571
(* Alias: se2 *)
1572
(* Alias: csISO11SwedishForNames *)
1573
let () = alias "IANA/csISO11SwedishForNames" "SEN_850200_C"
1574
let () = alias "IANA/se2" "SEN_850200_C"
1575
let () = alias "IANA/ISO646-SE2" "SEN_850200_C"
1576
let () = alias "IANA/iso-ir-11" "SEN_850200_C"
1577
let () = alias "IANA/SEN_850200_C" "SEN_850200_C"
1578
(* Name: KS_C_5601-1987 [RFC1345,KXS2] *)
1579
(* Alias: iso-ir-149 *)
1580
(* Alias: KS_C_5601-1989 *)
1581
(* Alias: KSC_5601 *)
1582
(* Alias: korean *)
1583
(* Alias: csKSC56011987 *)
1584
(* Name: ISO-2022-KR (preferred MIME name) [RFC1557,Choi] *)
1585
(* Alias: csISO2022KR *)
1586
let () = alias "IANA/csISO2022KR" "ISO-2022-KR"
1587
let () = alias "IANA/ISO-2022-KR" "ISO-2022-KR"
1588
(* Name: EUC-KR (preferred MIME name) [RFC1557,Choi] *)
1589
(* Alias: csEUCKR *)
1590
let () = alias "IANA/csEUCKR" "EUC-KR"
1591
let () = alias "IANA/EUC-KR" "EUC-KR"
1592
(* Name: ISO-2022-JP (preferred MIME name) [RFC1468,Murai] *)
1593
(* Alias: csISO2022JP *)
1594
let () = alias "IANA/csISO2022JP" "ISO-2022-JP"
1595
let () = alias "IANA/ISO-2022-JP" "ISO-2022-JP"
1596
(* Name: ISO-2022-JP-2 (preferred MIME name) [RFC1554,Ohta] *)
1597
(* Alias: csISO2022JP2 *)
1598
let () = alias "IANA/csISO2022JP2" "ISO-2022-JP-2"
1599
let () = alias "IANA/ISO-2022-JP-2" "ISO-2022-JP-2"
1600
(* Name: ISO-2022-CN [RFC1922] *)
1601
let () = alias "IANA/ISO-2022-CN" "ISO-2022-CN"
1602
(* Name: ISO-2022-CN-EXT [RFC1922] *)
1603
(* Name: JIS_C6220-1969-jp [RFC1345,KXS2] *)
1604
(* Alias: JIS_C6220-1969 *)
1605
(* Alias: iso-ir-13 *)
1606
(* Alias: katakana *)
1607
(* Alias: x0201-7 *)
1608
(* Alias: csISO13JISC6220jp *)
1609
(* Name: JIS_C6220-1969-ro [RFC1345,KXS2] *)
1610
(* Alias: iso-ir-14 *)
1611
(* Alias: jp *)
1612
(* Alias: ISO646-JP *)
1613
(* Alias: csISO14JISC6220ro *)
1614
let () = alias "IANA/csISO14JISC6220ro" "JIS_C6220-1969-RO"
1615
let () = alias "IANA/ISO646-JP" "JIS_C6220-1969-RO"
1616
let () = alias "IANA/jp" "JIS_C6220-1969-RO"
1617
let () = alias "IANA/iso-ir-14" "JIS_C6220-1969-RO"
1618
let () = alias "IANA/JIS_C6220-1969-ro" "JIS_C6220-1969-RO"
1619
(* Name: IT [RFC1345,KXS2] *)
1620
(* Alias: iso-ir-15 *)
1621
(* Alias: ISO646-IT *)
1622
(* Alias: csISO15Italian *)
1623
let () = alias "IANA/csISO15Italian" "IT"
1624
let () = alias "IANA/ISO646-IT" "IT"
1625
let () = alias "IANA/iso-ir-15" "IT"
1626
let () = alias "IANA/IT" "IT"
1627
(* Name: PT [RFC1345,KXS2] *)
1628
(* Alias: iso-ir-16 *)
1629
(* Alias: ISO646-PT *)
1630
(* Alias: csISO16Portuguese *)
1631
let () = alias "IANA/csISO16Portuguese" "PT"
1632
let () = alias "IANA/ISO646-PT" "PT"
1633
let () = alias "IANA/iso-ir-16" "PT"
1634
let () = alias "IANA/PT" "PT"
1635
(* Name: ES [RFC1345,KXS2] *)
1636
(* Alias: iso-ir-17 *)
1637
(* Alias: ISO646-ES *)
1638
(* Alias: csISO17Spanish *)
1639
let () = alias "IANA/csISO17Spanish" "ES"
1640
let () = alias "IANA/ISO646-ES" "ES"
1641
let () = alias "IANA/iso-ir-17" "ES"
1642
let () = alias "IANA/ES" "ES"
1643
(* Name: greek7-old [RFC1345,KXS2] *)
1644
(* Alias: iso-ir-18 *)
1645
(* Alias: csISO18Greek7Old *)
1646
let () = alias "IANA/csISO18Greek7Old" "GREEK7-OLD"
1647
let () = alias "IANA/iso-ir-18" "GREEK7-OLD"
1648
let () = alias "IANA/greek7-old" "GREEK7-OLD"
1649
(* Name: latin-greek [RFC1345,KXS2] *)
1650
(* Alias: iso-ir-19 *)
1651
(* Alias: csISO19LatinGreek *)
1652
let () = alias "IANA/csISO19LatinGreek" "LATIN-GREEK"
1653
let () = alias "IANA/iso-ir-19" "LATIN-GREEK"
1654
let () = alias "IANA/latin-greek" "LATIN-GREEK"
1655
(* Name: DIN_66003 [RFC1345,KXS2] *)
1656
(* Alias: iso-ir-21 *)
1657
(* Alias: de *)
1658
(* Alias: ISO646-DE *)
1659
(* Alias: csISO21German *)
1660
let () = alias "IANA/csISO21German" "DIN_66003"
1661
let () = alias "IANA/ISO646-DE" "DIN_66003"
1662
let () = alias "IANA/de" "DIN_66003"
1663
let () = alias "IANA/iso-ir-21" "DIN_66003"
1664
let () = alias "IANA/DIN_66003" "DIN_66003"
1665
(* Name: NF_Z_62-010_(1973) [RFC1345,KXS2] *)
1666
(* Alias: iso-ir-25 *)
1667
(* Alias: ISO646-FR1 *)
1668
(* Alias: csISO25French *)
1669
let () = alias "IANA/csISO25French" "NF_Z_62-010_(1973)"
1670
let () = alias "IANA/ISO646-FR1" "NF_Z_62-010_(1973)"
1671
let () = alias "IANA/iso-ir-25" "NF_Z_62-010_(1973)"
1672
let () = alias "IANA/NF_Z_62-010_(1973)" "NF_Z_62-010_(1973)"
1673
(* Name: Latin-greek-1 [RFC1345,KXS2] *)
1674
(* Alias: iso-ir-27 *)
1675
(* Alias: csISO27LatinGreek1 *)
1676
let () = alias "IANA/csISO27LatinGreek1" "LATIN-GREEK-1"
1677
let () = alias "IANA/iso-ir-27" "LATIN-GREEK-1"
1678
let () = alias "IANA/Latin-greek-1" "LATIN-GREEK-1"
1679
(* Name: ISO_5427 [RFC1345,KXS2] *)
1680
(* Alias: iso-ir-37 *)
1681
(* Alias: csISO5427Cyrillic *)
1682
let () = alias "IANA/csISO5427Cyrillic" "ISO_5427"
1683
let () = alias "IANA/iso-ir-37" "ISO_5427"
1684
let () = alias "IANA/ISO_5427" "ISO_5427"
1685
(* Name: JIS_C6226-1978 [RFC1345,KXS2] *)
1686
(* Alias: iso-ir-42 *)
1687
(* Alias: csISO42JISC62261978 *)
1688
(* Name: BS_viewdata [RFC1345,KXS2] *)
1689
(* Alias: iso-ir-47 *)
1690
(* Alias: csISO47BSViewdata *)
1691
let () = alias "IANA/csISO47BSViewdata" "BS_VIEWDATA"
1692
let () = alias "IANA/iso-ir-47" "BS_VIEWDATA"
1693
let () = alias "IANA/BS_viewdata" "BS_VIEWDATA"
1694
(* Name: INIS [RFC1345,KXS2] *)
1695
(* Alias: iso-ir-49 *)
1696
(* Alias: csISO49INIS *)
1697
let () = alias "IANA/csISO49INIS" "INIS"
1698
let () = alias "IANA/iso-ir-49" "INIS"
1699
let () = alias "IANA/INIS" "INIS"
1700
(* Name: INIS-8 [RFC1345,KXS2] *)
1701
(* Alias: iso-ir-50 *)
1702
(* Alias: csISO50INIS8 *)
1703
let () = alias "IANA/csISO50INIS8" "INIS-8"
1704
let () = alias "IANA/iso-ir-50" "INIS-8"
1705
let () = alias "IANA/INIS-8" "INIS-8"
1706
(* Name: INIS-cyrillic [RFC1345,KXS2] *)
1707
(* Alias: iso-ir-51 *)
1708
(* Alias: csISO51INISCyrillic *)
1709
let () = alias "IANA/csISO51INISCyrillic" "INIS-CYRILLIC"
1710
let () = alias "IANA/iso-ir-51" "INIS-CYRILLIC"
1711
let () = alias "IANA/INIS-cyrillic" "INIS-CYRILLIC"
1712
(* Name: ISO_5427:1981 [RFC1345,KXS2] *)
1713
(* Alias: iso-ir-54 *)
1714
(* Alias: ISO5427Cyrillic1981 *)
1715
let () = alias "IANA/ISO5427Cyrillic1981" "ISO_5427-EXT"
1716
let () = alias "IANA/iso-ir-54" "ISO_5427-EXT"
1717
let () = alias "IANA/ISO_5427:1981" "ISO_5427-EXT"
1718
(* Name: ISO_5428:1980 [RFC1345,KXS2] *)
1719
(* Alias: iso-ir-55 *)
1720
(* Alias: csISO5428Greek *)
1721
let () = alias "IANA/csISO5428Greek" "ISO_5428"
1722
let () = alias "IANA/iso-ir-55" "ISO_5428"
1723
let () = alias "IANA/ISO_5428:1980" "ISO_5428"
1724
(* Name: GB_1988-80 [RFC1345,KXS2] *)
1725
(* Alias: iso-ir-57 *)
1726
(* Alias: cn *)
1727
(* Alias: ISO646-CN *)
1728
(* Alias: csISO57GB1988 *)
1729
let () = alias "IANA/csISO57GB1988" "GB_1988-80"
1730
let () = alias "IANA/ISO646-CN" "GB_1988-80"
1731
let () = alias "IANA/cn" "GB_1988-80"
1732
let () = alias "IANA/iso-ir-57" "GB_1988-80"
1733
let () = alias "IANA/GB_1988-80" "GB_1988-80"
1734
(* Name: GB_2312-80 [RFC1345,KXS2] *)
1735
(* Alias: iso-ir-58 *)
1736
(* Alias: chinese *)
1737
(* Alias: csISO58GB231280 *)
1738
(* Name: NS_4551-1 [RFC1345,KXS2] *)
1739
(* Alias: iso-ir-60 *)
1740
(* Alias: ISO646-NO *)
1741
(* Alias: no *)
1742
(* Alias: csISO60DanishNorwegian *)
1743
(* Alias: csISO60Norwegian1 *)
1744
let () = alias "IANA/csISO60Norwegian1" "NS_4551-1"
1745
let () = alias "IANA/csISO60DanishNorwegian" "NS_4551-1"
1746
let () = alias "IANA/no" "NS_4551-1"
1747
let () = alias "IANA/ISO646-NO" "NS_4551-1"
1748
let () = alias "IANA/iso-ir-60" "NS_4551-1"
1749
let () = alias "IANA/NS_4551-1" "NS_4551-1"
1750
(* Name: NS_4551-2 [RFC1345,KXS2] *)
1751
(* Alias: ISO646-NO2 *)
1752
(* Alias: iso-ir-61 *)
1753
(* Alias: no2 *)
1754
(* Alias: csISO61Norwegian2 *)
1755
let () = alias "IANA/csISO61Norwegian2" "NS_4551-2"
1756
let () = alias "IANA/no2" "NS_4551-2"
1757
let () = alias "IANA/iso-ir-61" "NS_4551-2"
1758
let () = alias "IANA/ISO646-NO2" "NS_4551-2"
1759
let () = alias "IANA/NS_4551-2" "NS_4551-2"
1760
(* Name: NF_Z_62-010 [RFC1345,KXS2] *)
1761
(* Alias: iso-ir-69 *)
1762
(* Alias: ISO646-FR *)
1763
(* Alias: fr *)
1764
(* Alias: csISO69French *)
1765
let () = alias "IANA/csISO69French" "NF_Z_62-010"
1766
let () = alias "IANA/fr" "NF_Z_62-010"
1767
let () = alias "IANA/ISO646-FR" "NF_Z_62-010"
1768
let () = alias "IANA/iso-ir-69" "NF_Z_62-010"
1769
let () = alias "IANA/NF_Z_62-010" "NF_Z_62-010"
1770
(* Name: videotex-suppl [RFC1345,KXS2] *)
1771
(* Alias: iso-ir-70 *)
1772
(* Alias: csISO70VideotexSupp1 *)
1773
let () = alias "IANA/csISO70VideotexSupp1" "VIDEOTEX-SUPPL"
1774
let () = alias "IANA/iso-ir-70" "VIDEOTEX-SUPPL"
1775
let () = alias "IANA/videotex-suppl" "VIDEOTEX-SUPPL"
1776
(* Name: PT2 [RFC1345,KXS2] *)
1777
(* Alias: iso-ir-84 *)
1778
(* Alias: ISO646-PT2 *)
1779
(* Alias: csISO84Portuguese2 *)
1780
let () = alias "IANA/csISO84Portuguese2" "PT2"
1781
let () = alias "IANA/ISO646-PT2" "PT2"
1782
let () = alias "IANA/iso-ir-84" "PT2"
1783
let () = alias "IANA/PT2" "PT2"
1784
(* Name: ES2 [RFC1345,KXS2] *)
1785
(* Alias: iso-ir-85 *)
1786
(* Alias: ISO646-ES2 *)
1787
(* Alias: csISO85Spanish2 *)
1788
let () = alias "IANA/csISO85Spanish2" "ES2"
1789
let () = alias "IANA/ISO646-ES2" "ES2"
1790
let () = alias "IANA/iso-ir-85" "ES2"
1791
let () = alias "IANA/ES2" "ES2"
1792
(* Name: MSZ_7795.3 [RFC1345,KXS2] *)
1793
(* Alias: iso-ir-86 *)
1794
(* Alias: ISO646-HU *)
1795
(* Alias: hu *)
1796
(* Alias: csISO86Hungarian *)
1797
let () = alias "IANA/csISO86Hungarian" "MSZ_7795.3"
1798
let () = alias "IANA/hu" "MSZ_7795.3"
1799
let () = alias "IANA/ISO646-HU" "MSZ_7795.3"
1800
let () = alias "IANA/iso-ir-86" "MSZ_7795.3"
1801
let () = alias "IANA/MSZ_7795.3" "MSZ_7795.3"
1802
(* Name: JIS_C6226-1983 [RFC1345,KXS2] *)
1803
(* Alias: iso-ir-87 *)
1804
(* Alias: x0208 *)
1805
(* Alias: JIS_X0208-1983 *)
1806
(* Alias: csISO87JISX0208 *)
1807
(* Name: greek7 [RFC1345,KXS2] *)
1808
(* Alias: iso-ir-88 *)
1809
(* Alias: csISO88Greek7 *)
1810
let () = alias "IANA/csISO88Greek7" "GREEK7"
1811
let () = alias "IANA/iso-ir-88" "GREEK7"
1812
let () = alias "IANA/greek7" "GREEK7"
1813
(* Name: ASMO_449 [RFC1345,KXS2] *)
1814
(* Alias: ISO_9036 *)
1815
(* Alias: arabic7 *)
1816
(* Alias: iso-ir-89 *)
1817
(* Alias: csISO89ASMO449 *)
1818
let () = alias "IANA/csISO89ASMO449" "ASMO_449"
1819
let () = alias "IANA/iso-ir-89" "ASMO_449"
1820
let () = alias "IANA/arabic7" "ASMO_449"
1821
let () = alias "IANA/ISO_9036" "ASMO_449"
1822
let () = alias "IANA/ASMO_449" "ASMO_449"
1823
(* Name: iso-ir-90 [RFC1345,KXS2] *)
1824
(* Alias: csISO90 *)
1825
let () = alias "IANA/csISO90" "ISO-IR-90"
1826
let () = alias "IANA/iso-ir-90" "ISO-IR-90"
1827
(* Name: JIS_C6229-1984-a [RFC1345,KXS2] *)
1828
(* Alias: iso-ir-91 *)
1829
(* Alias: jp-ocr-a *)
1830
(* Alias: csISO91JISC62291984a *)
1831
(* Name: JIS_C6229-1984-b [RFC1345,KXS2] *)
1832
(* Alias: iso-ir-92 *)
1833
(* Alias: ISO646-JP-OCR-B *)
1834
(* Alias: jp-ocr-b *)
1835
(* Alias: csISO92JISC62991984b *)
1836
let () = alias "IANA/csISO92JISC62991984b" "JIS_C6229-1984-B"
1837
let () = alias "IANA/jp-ocr-b" "JIS_C6229-1984-B"
1838
let () = alias "IANA/ISO646-JP-OCR-B" "JIS_C6229-1984-B"
1839
let () = alias "IANA/iso-ir-92" "JIS_C6229-1984-B"
1840
let () = alias "IANA/JIS_C6229-1984-b" "JIS_C6229-1984-B"
1841
(* Name: JIS_C6229-1984-b-add [RFC1345,KXS2] *)
1842
(* Alias: iso-ir-93 *)
1843
(* Alias: jp-ocr-b-add *)
1844
(* Alias: csISO93JIS62291984badd *)
1845
(* Name: JIS_C6229-1984-hand [RFC1345,KXS2] *)
1846
(* Alias: iso-ir-94 *)
1847
(* Alias: jp-ocr-hand *)
1848
(* Alias: csISO94JIS62291984hand *)
1849
(* Name: JIS_C6229-1984-hand-add [RFC1345,KXS2] *)
1850
(* Alias: iso-ir-95 *)
1851
(* Alias: jp-ocr-hand-add *)
1852
(* Alias: csISO95JIS62291984handadd *)
1853
(* Name: JIS_C6229-1984-kana [RFC1345,KXS2] *)
1854
(* Alias: iso-ir-96 *)
1855
(* Alias: csISO96JISC62291984kana *)
1856
(* Name: ISO_2033-1983 [RFC1345,KXS2] *)
1857
(* Alias: iso-ir-98 *)
1858
(* Alias: e13b *)
1859
(* Alias: csISO2033 *)
1860
let () = alias "IANA/csISO2033" "ISO_2033-1983"
1861
let () = alias "IANA/e13b" "ISO_2033-1983"
1862
let () = alias "IANA/iso-ir-98" "ISO_2033-1983"
1863
let () = alias "IANA/ISO_2033-1983" "ISO_2033-1983"
1864
(* Name: ANSI_X3.110-1983 [RFC1345,KXS2] *)
1865
(* Alias: iso-ir-99 *)
1866
(* Alias: CSA_T500-1983 *)
1867
(* Alias: NAPLPS *)
1868
(* Alias: csISO99NAPLPS *)
1869
let () = alias "IANA/csISO99NAPLPS" "ANSI_X3.110-1983"
1870
let () = alias "IANA/NAPLPS" "ANSI_X3.110-1983"
1871
let () = alias "IANA/CSA_T500-1983" "ANSI_X3.110-1983"
1872
let () = alias "IANA/iso-ir-99" "ANSI_X3.110-1983"
1873
let () = alias "IANA/ANSI_X3.110-1983" "ANSI_X3.110-1983"
1874
(* Name: ISO_8859-1:1987 [RFC1345,KXS2] *)
1875
(* Alias: iso-ir-100 *)
1876
(* Alias: ISO_8859-1 *)
1877
(* Alias: ISO-8859-1 (preferred MIME name) *)
1878
(* Alias: latin1 *)
1879
(* Alias: l1 *)
1880
(* Alias: IBM819 *)
1881
(* Alias: CP819 *)
1882
(* Alias: csISOLatin1 *)
1883
let () = alias "IANA/csISOLatin1" "ISO-8859-1"
1884
let () = alias "IANA/CP819" "ISO-8859-1"
1885
let () = alias "IANA/IBM819" "ISO-8859-1"
1886
let () = alias "IANA/l1" "ISO-8859-1"
1887
let () = alias "IANA/latin1" "ISO-8859-1"
1888
let () = alias "IANA/ISO-8859-1" "ISO-8859-1"
1889
let () = alias "IANA/ISO_8859-1" "ISO-8859-1"
1890
let () = alias "IANA/iso-ir-100" "ISO-8859-1"
1891
let () = alias "IANA/ISO_8859-1:1987" "ISO-8859-1"
1892
(* Name: ISO_8859-2:1987 [RFC1345,KXS2] *)
1893
(* Alias: iso-ir-101 *)
1894
(* Alias: ISO_8859-2 *)
1895
(* Alias: ISO-8859-2 (preferred MIME name) *)
1896
(* Alias: latin2 *)
1897
(* Alias: l2 *)
1898
(* Alias: csISOLatin2 *)
1899
let () = alias "IANA/csISOLatin2" "ISO-8859-2"
1900
let () = alias "IANA/l2" "ISO-8859-2"
1901
let () = alias "IANA/latin2" "ISO-8859-2"
1902
let () = alias "IANA/ISO-8859-2" "ISO-8859-2"
1903
let () = alias "IANA/ISO_8859-2" "ISO-8859-2"
1904
let () = alias "IANA/iso-ir-101" "ISO-8859-2"
1905
let () = alias "IANA/ISO_8859-2:1987" "ISO-8859-2"
1906
(* Name: T.61-7bit [RFC1345,KXS2] *)
1907
(* Alias: iso-ir-102 *)
1908
(* Alias: csISO102T617bit *)
1909
let () = alias "IANA/csISO102T617bit" "T.61-7BIT"
1910
let () = alias "IANA/iso-ir-102" "T.61-7BIT"
1911
let () = alias "IANA/T.61-7bit" "T.61-7BIT"
1912
(* Name: T.61-8bit [RFC1345,KXS2] *)
1913
(* Alias: T.61 *)
1914
(* Alias: iso-ir-103 *)
1915
(* Alias: csISO103T618bit *)
1916
let () = alias "IANA/csISO103T618bit" "T.61-8BIT"
1917
let () = alias "IANA/iso-ir-103" "T.61-8BIT"
1918
let () = alias "IANA/T.61" "T.61-8BIT"
1919
let () = alias "IANA/T.61-8bit" "T.61-8BIT"
1920
(* Name: ISO_8859-3:1988 [RFC1345,KXS2] *)
1921
(* Alias: iso-ir-109 *)
1922
(* Alias: ISO_8859-3 *)
1923
(* Alias: ISO-8859-3 (preferred MIME name) *)
1924
(* Alias: latin3 *)
1925
(* Alias: l3 *)
1926
(* Alias: csISOLatin3 *)
1927
let () = alias "IANA/csISOLatin3" "ISO-8859-3"
1928
let () = alias "IANA/l3" "ISO-8859-3"
1929
let () = alias "IANA/latin3" "ISO-8859-3"
1930
let () = alias "IANA/ISO-8859-3" "ISO-8859-3"
1931
let () = alias "IANA/ISO_8859-3" "ISO-8859-3"
1932
let () = alias "IANA/iso-ir-109" "ISO-8859-3"
1933
let () = alias "IANA/ISO_8859-3:1988" "ISO-8859-3"
1934
(* Name: ISO_8859-4:1988 [RFC1345,KXS2] *)
1935
(* Alias: iso-ir-110 *)
1936
(* Alias: ISO_8859-4 *)
1937
(* Alias: ISO-8859-4 (preferred MIME name) *)
1938
(* Alias: latin4 *)
1939
(* Alias: l4 *)
1940
(* Alias: csISOLatin4 *)
1941
let () = alias "IANA/csISOLatin4" "ISO-8859-4"
1942
let () = alias "IANA/l4" "ISO-8859-4"
1943
let () = alias "IANA/latin4" "ISO-8859-4"
1944
let () = alias "IANA/ISO-8859-4" "ISO-8859-4"
1945
let () = alias "IANA/ISO_8859-4" "ISO-8859-4"
1946
let () = alias "IANA/iso-ir-110" "ISO-8859-4"
1947
let () = alias "IANA/ISO_8859-4:1988" "ISO-8859-4"
1948
(* Name: ECMA-cyrillic [RFC1345,KXS2] *)
1949
(* Alias: iso-ir-111 *)
1950
(* Alias: csISO111ECMACyrillic *)
1951
let () = alias "IANA/csISO111ECMACyrillic" "ECMA-CYRILLIC"
1952
let () = alias "IANA/iso-ir-111" "ECMA-CYRILLIC"
1953
let () = alias "IANA/ECMA-cyrillic" "ECMA-CYRILLIC"
1954
(* Name: CSA_Z243.4-1985-1 [RFC1345,KXS2] *)
1955
(* Alias: iso-ir-121 *)
1956
(* Alias: ISO646-CA *)
1957
(* Alias: csa7-1 *)
1958
(* Alias: ca *)
1959
(* Alias: csISO121Canadian1 *)
1960
let () = alias "IANA/csISO121Canadian1" "CSA_Z243.4-1985-1"
1961
let () = alias "IANA/ca" "CSA_Z243.4-1985-1"
1962
let () = alias "IANA/csa7-1" "CSA_Z243.4-1985-1"
1963
let () = alias "IANA/ISO646-CA" "CSA_Z243.4-1985-1"
1964
let () = alias "IANA/iso-ir-121" "CSA_Z243.4-1985-1"
1965
let () = alias "IANA/CSA_Z243.4-1985-1" "CSA_Z243.4-1985-1"
1966
(* Name: CSA_Z243.4-1985-2 [RFC1345,KXS2] *)
1967
(* Alias: iso-ir-122 *)
1968
(* Alias: ISO646-CA2 *)
1969
(* Alias: csa7-2 *)
1970
(* Alias: csISO122Canadian2 *)
1971
let () = alias "IANA/csISO122Canadian2" "CSA_Z243.4-1985-2"
1972
let () = alias "IANA/csa7-2" "CSA_Z243.4-1985-2"
1973
let () = alias "IANA/ISO646-CA2" "CSA_Z243.4-1985-2"
1974
let () = alias "IANA/iso-ir-122" "CSA_Z243.4-1985-2"
1975
let () = alias "IANA/CSA_Z243.4-1985-2" "CSA_Z243.4-1985-2"
1976
(* Name: CSA_Z243.4-1985-gr [RFC1345,KXS2] *)
1977
(* Alias: iso-ir-123 *)
1978
(* Alias: csISO123CSAZ24341985gr *)
1979
let () = alias "IANA/csISO123CSAZ24341985gr" "CSA_Z243.4-1985-GR"
1980
let () = alias "IANA/iso-ir-123" "CSA_Z243.4-1985-GR"
1981
let () = alias "IANA/CSA_Z243.4-1985-gr" "CSA_Z243.4-1985-GR"
1982
(* Name: ISO_8859-6:1987 [RFC1345,KXS2] *)
1983
(* Alias: iso-ir-127 *)
1984
(* Alias: ISO_8859-6 *)
1985
(* Alias: ISO-8859-6 (preferred MIME name) *)
1986
(* Alias: ECMA-114 *)
1987
(* Alias: ASMO-708 *)
1988
(* Alias: arabic *)
1989
(* Alias: csISOLatinArabic *)
1990
let () = alias "IANA/csISOLatinArabic" "ISO-8859-6"
1991
let () = alias "IANA/arabic" "ISO-8859-6"
1992
let () = alias "IANA/ASMO-708" "ISO-8859-6"
1993
let () = alias "IANA/ECMA-114" "ISO-8859-6"
1994
let () = alias "IANA/ISO-8859-6" "ISO-8859-6"
1995
let () = alias "IANA/ISO_8859-6" "ISO-8859-6"
1996
let () = alias "IANA/iso-ir-127" "ISO-8859-6"
1997
let () = alias "IANA/ISO_8859-6:1987" "ISO-8859-6"
1998
(* Name: ISO_8859-6-E [RFC1556,IANA] *)
1999
(* Alias: csISO88596E *)
2000
(* Alias: ISO-8859-6-E (preferred MIME name) *)
2001
(* Name: ISO_8859-6-I [RFC1556,IANA] *)
2002
(* Alias: csISO88596I *)
2003
(* Alias: ISO-8859-6-I (preferred MIME name) *)
2004
(* Name: ISO_8859-7:1987 [RFC1947,RFC1345,KXS2] *)
2005
(* Alias: iso-ir-126 *)
2006
(* Alias: ISO_8859-7 *)
2007
(* Alias: ISO-8859-7 (preferred MIME name) *)
2008
(* Alias: ELOT_928 *)
2009
(* Alias: ECMA-118 *)
2010
(* Alias: greek *)
2011
(* Alias: greek8 *)
2012
(* Alias: csISOLatinGreek *)
2013
let () = alias "IANA/csISOLatinGreek" "ISO-8859-7"
2014
let () = alias "IANA/greek8" "ISO-8859-7"
2015
let () = alias "IANA/greek" "ISO-8859-7"
2016
let () = alias "IANA/ECMA-118" "ISO-8859-7"
2017
let () = alias "IANA/ELOT_928" "ISO-8859-7"
2018
let () = alias "IANA/ISO-8859-7" "ISO-8859-7"
2019
let () = alias "IANA/ISO_8859-7" "ISO-8859-7"
2020
let () = alias "IANA/iso-ir-126" "ISO-8859-7"
2021
let () = alias "IANA/ISO_8859-7:1987" "ISO-8859-7"
2022
(* Name: T.101-G2 [RFC1345,KXS2] *)
2023
(* Alias: iso-ir-128 *)
2024
(* Alias: csISO128T101G2 *)
2025
let () = alias "IANA/csISO128T101G2" "T.101-G2"
2026
let () = alias "IANA/iso-ir-128" "T.101-G2"
2027
let () = alias "IANA/T.101-G2" "T.101-G2"
2028
(* Name: ISO_8859-8:1988 [RFC1345,KXS2] *)
2029
(* Alias: iso-ir-138 *)
2030
(* Alias: ISO_8859-8 *)
2031
(* Alias: ISO-8859-8 (preferred MIME name) *)
2032
(* Alias: hebrew *)
2033
(* Alias: csISOLatinHebrew *)
2034
let () = alias "IANA/csISOLatinHebrew" "ISO-8859-8"
2035
let () = alias "IANA/hebrew" "ISO-8859-8"
2036
let () = alias "IANA/ISO-8859-8" "ISO-8859-8"
2037
let () = alias "IANA/ISO_8859-8" "ISO-8859-8"
2038
let () = alias "IANA/iso-ir-138" "ISO-8859-8"
2039
let () = alias "IANA/ISO_8859-8:1988" "ISO-8859-8"
2040
(* Name: ISO_8859-8-E [RFC1556,Nussbacher] *)
2041
(* Alias: csISO88598E *)
2042
(* Alias: ISO-8859-8-E (preferred MIME name) *)
2043
(* Name: ISO_8859-8-I [RFC1556,Nussbacher] *)
2044
(* Alias: csISO88598I *)
2045
(* Alias: ISO-8859-8-I (preferred MIME name) *)
2046
(* Name: CSN_369103 [RFC1345,KXS2] *)
2047
(* Alias: iso-ir-139 *)
2048
(* Alias: csISO139CSN369103 *)
2049
let () = alias "IANA/csISO139CSN369103" "CSN_369103"
2050
let () = alias "IANA/iso-ir-139" "CSN_369103"
2051
let () = alias "IANA/CSN_369103" "CSN_369103"
2052
(* Name: JUS_I.B1.002 [RFC1345,KXS2] *)
2053
(* Alias: iso-ir-141 *)
2054
(* Alias: ISO646-YU *)
2055
(* Alias: js *)
2056
(* Alias: yu *)
2057
(* Alias: csISO141JUSIB1002 *)
2058
let () = alias "IANA/csISO141JUSIB1002" "JUS_I.B1.002"
2059
let () = alias "IANA/yu" "JUS_I.B1.002"
2060
let () = alias "IANA/js" "JUS_I.B1.002"
2061
let () = alias "IANA/ISO646-YU" "JUS_I.B1.002"
2062
let () = alias "IANA/iso-ir-141" "JUS_I.B1.002"
2063
let () = alias "IANA/JUS_I.B1.002" "JUS_I.B1.002"
2064
(* Name: ISO_6937-2-add [RFC1345,KXS2] *)
2065
(* Alias: iso-ir-142 *)
2066
(* Alias: csISOTextComm *)
2067
let () = alias "IANA/csISOTextComm" "ISO_6937-2-ADD"
2068
let () = alias "IANA/iso-ir-142" "ISO_6937-2-ADD"
2069
let () = alias "IANA/ISO_6937-2-add" "ISO_6937-2-ADD"
2070
(* Name: IEC_P27-1 [RFC1345,KXS2] *)
2071
(* Alias: iso-ir-143 *)
2072
(* Alias: csISO143IECP271 *)
2073
let () = alias "IANA/csISO143IECP271" "IEC_P27-1"
2074
let () = alias "IANA/iso-ir-143" "IEC_P27-1"
2075
let () = alias "IANA/IEC_P27-1" "IEC_P27-1"
2076
(* Name: ISO_8859-5:1988 [RFC1345,KXS2] *)
2077
(* Alias: iso-ir-144 *)
2078
(* Alias: ISO_8859-5 *)
2079
(* Alias: ISO-8859-5 (preferred MIME name) *)
2080
(* Alias: cyrillic *)
2081
(* Alias: csISOLatinCyrillic *)
2082
let () = alias "IANA/csISOLatinCyrillic" "ISO-8859-5"
2083
let () = alias "IANA/cyrillic" "ISO-8859-5"
2084
let () = alias "IANA/ISO-8859-5" "ISO-8859-5"
2085
let () = alias "IANA/ISO_8859-5" "ISO-8859-5"
2086
let () = alias "IANA/iso-ir-144" "ISO-8859-5"
2087
let () = alias "IANA/ISO_8859-5:1988" "ISO-8859-5"
2088
(* Name: JUS_I.B1.003-serb [RFC1345,KXS2] *)
2089
(* Alias: iso-ir-146 *)
2090
(* Alias: serbian *)
2091
(* Alias: csISO146Serbian *)
2092
let () = alias "IANA/csISO146Serbian" "JUS_I.B1.003-SERB"
2093
let () = alias "IANA/serbian" "JUS_I.B1.003-SERB"
2094
let () = alias "IANA/iso-ir-146" "JUS_I.B1.003-SERB"
2095
let () = alias "IANA/JUS_I.B1.003-serb" "JUS_I.B1.003-SERB"
2096
(* Name: JUS_I.B1.003-mac [RFC1345,KXS2] *)
2097
(* Alias: macedonian *)
2098
(* Alias: iso-ir-147 *)
2099
(* Alias: csISO147Macedonian *)
2100
let () = alias "IANA/csISO147Macedonian" "JUS_I.B1.003-MAC"
2101
let () = alias "IANA/iso-ir-147" "JUS_I.B1.003-MAC"
2102
let () = alias "IANA/macedonian" "JUS_I.B1.003-MAC"
2103
let () = alias "IANA/JUS_I.B1.003-mac" "JUS_I.B1.003-MAC"
2104
(* Name: ISO_8859-9:1989 [RFC1345,KXS2] *)
2105
(* Alias: iso-ir-148 *)
2106
(* Alias: ISO_8859-9 *)
2107
(* Alias: ISO-8859-9 (preferred MIME name) *)
2108
(* Alias: latin5 *)
2109
(* Alias: l5 *)
2110
(* Alias: csISOLatin5 *)
2111
let () = alias "IANA/csISOLatin5" "ISO-8859-9"
2112
let () = alias "IANA/l5" "ISO-8859-9"
2113
let () = alias "IANA/latin5" "ISO-8859-9"
2114
let () = alias "IANA/ISO-8859-9" "ISO-8859-9"
2115
let () = alias "IANA/ISO_8859-9" "ISO-8859-9"
2116
let () = alias "IANA/iso-ir-148" "ISO-8859-9"
2117
let () = alias "IANA/ISO_8859-9:1989" "ISO-8859-9"
2118
(* Name: greek-ccitt [RFC1345,KXS2] *)
2119
(* Alias: iso-ir-150 *)
2120
(* Alias: csISO150 *)
2121
(* Alias: csISO150GreekCCITT *)
2122
let () = alias "IANA/csISO150GreekCCITT" "GREEK-CCITT"
2123
let () = alias "IANA/csISO150" "GREEK-CCITT"
2124
let () = alias "IANA/iso-ir-150" "GREEK-CCITT"
2125
let () = alias "IANA/greek-ccitt" "GREEK-CCITT"
2126
(* Name: NC_NC00-10:81 [RFC1345,KXS2] *)
2127
(* Alias: cuba *)
2128
(* Alias: iso-ir-151 *)
2129
(* Alias: ISO646-CU *)
2130
(* Alias: csISO151Cuba *)
2131
let () = alias "IANA/csISO151Cuba" "NC_NC00-10"
2132
let () = alias "IANA/ISO646-CU" "NC_NC00-10"
2133
let () = alias "IANA/iso-ir-151" "NC_NC00-10"
2134
let () = alias "IANA/cuba" "NC_NC00-10"
2135
let () = alias "IANA/NC_NC00-10:81" "NC_NC00-10"
2136
(* Name: ISO_6937-2-25 [RFC1345,KXS2] *)
2137
(* Alias: iso-ir-152 *)
2138
(* Alias: csISO6937Add *)
2139
let () = alias "IANA/csISO6937Add" "ISO_6937-2-25"
2140
let () = alias "IANA/iso-ir-152" "ISO_6937-2-25"
2141
let () = alias "IANA/ISO_6937-2-25" "ISO_6937-2-25"
2142
(* Name: GOST_19768-74 [RFC1345,KXS2] *)
2143
(* Alias: ST_SEV_358-88 *)
2144
(* Alias: iso-ir-153 *)
2145
(* Alias: csISO153GOST1976874 *)
2146
let () = alias "IANA/csISO153GOST1976874" "GOST_19768-74"
2147
let () = alias "IANA/iso-ir-153" "GOST_19768-74"
2148
let () = alias "IANA/ST_SEV_358-88" "GOST_19768-74"
2149
let () = alias "IANA/GOST_19768-74" "GOST_19768-74"
2150
(* Name: ISO_8859-supp [RFC1345,KXS2] *)
2151
(* Alias: iso-ir-154 *)
2152
(* Alias: latin1-2-5 *)
2153
(* Alias: csISO8859Supp *)
2154
let () = alias "IANA/csISO8859Supp" "ISO_8859-SUPP"
2155
let () = alias "IANA/latin1-2-5" "ISO_8859-SUPP"
2156
let () = alias "IANA/iso-ir-154" "ISO_8859-SUPP"
2157
let () = alias "IANA/ISO_8859-supp" "ISO_8859-SUPP"
2158
(* Name: ISO_10367-box [RFC1345,KXS2] *)
2159
(* Alias: iso-ir-155 *)
2160
(* Alias: csISO10367Box *)
2161
let () = alias "IANA/csISO10367Box" "ISO_10367-BOX"
2162
let () = alias "IANA/iso-ir-155" "ISO_10367-BOX"
2163
let () = alias "IANA/ISO_10367-box" "ISO_10367-BOX"
2164
(* Name: ISO-8859-10 (preferred MIME name) [RFC1345,KXS2] *)
2165
(* Alias: iso-ir-157 *)
2166
(* Alias: l6 *)
2167
(* Alias: ISO_8859-10:1992 *)
2168
(* Alias: csISOLatin6 *)
2169
(* Alias: latin6 *)
2170
let () = alias "IANA/latin6" "ISO-8859-10"
2171
let () = alias "IANA/csISOLatin6" "ISO-8859-10"
2172
let () = alias "IANA/ISO_8859-10:1992" "ISO-8859-10"
2173
let () = alias "IANA/l6" "ISO-8859-10"
2174
let () = alias "IANA/iso-ir-157" "ISO-8859-10"
2175
let () = alias "IANA/ISO-8859-10" "ISO-8859-10"
2176
(* Name: latin-lap [RFC1345,KXS2] *)
2177
(* Alias: lap *)
2178
(* Alias: iso-ir-158 *)
2179
(* Alias: csISO158Lap *)
2180
let () = alias "IANA/csISO158Lap" "SAMI"
2181
let () = alias "IANA/iso-ir-158" "SAMI"
2182
let () = alias "IANA/lap" "SAMI"
2183
let () = alias "IANA/latin-lap" "SAMI"
2184
(* Name: JIS_X0212-1990 [RFC1345,KXS2] *)
2185
(* Alias: x0212 *)
2186
(* Alias: iso-ir-159 *)
2187
(* Alias: csISO159JISX02121990 *)
2188
(* Name: DS_2089 [RFC1345,KXS2] *)
2189
(* Alias: DS2089 *)
2190
(* Alias: ISO646-DK *)
2191
(* Alias: dk *)
2192
(* Alias: csISO646Danish *)
2193
let () = alias "IANA/csISO646Danish" "DS_2089"
2194
let () = alias "IANA/dk" "DS_2089"
2195
let () = alias "IANA/ISO646-DK" "DS_2089"
2196
let () = alias "IANA/DS2089" "DS_2089"
2197
let () = alias "IANA/DS_2089" "DS_2089"
2198
(* Name: us-dk [RFC1345,KXS2] *)
2199
(* Alias: csUSDK *)
2200
(* Name: dk-us [RFC1345,KXS2] *)
2201
(* Alias: csDKUS *)
2202
(* Name: JIS_X0201 [RFC1345,KXS2] *)
2203
(* Alias: X0201 *)
2204
(* Alias: csHalfWidthKatakana *)
2205
let () = alias "IANA/csHalfWidthKatakana" "JIS_X0201"
2206
let () = alias "IANA/X0201" "JIS_X0201"
2207
let () = alias "IANA/JIS_X0201" "JIS_X0201"
2208
(* Name: KSC5636 [RFC1345,KXS2] *)
2209
(* Alias: ISO646-KR *)
2210
(* Alias: csKSC5636 *)
2211
let () = alias "IANA/csKSC5636" "KSC5636"
2212
let () = alias "IANA/ISO646-KR" "KSC5636"
2213
let () = alias "IANA/KSC5636" "KSC5636"
2214
(* Name: ISO-10646-UCS-2 *)
2215
(* Alias: csUnicode *)
2216
let () = alias "IANA/ISO-10646-UCS-2" "UTF-16"
2217
let () = alias "IANA/csUnicode" "UTF-16"
2218
(* Name: ISO-10646-UCS-4 *)
2219
(* Alias: csUCS4 *)
2220
let () = alias "IANA/ISO-10646-UCS-4" "UCS4"
2221
let () = alias "IANA/csUCS4" "UCS4"
2222
(* Name: DEC-MCS [RFC1345,KXS2] *)
2223
(* Alias: dec *)
2224
(* Alias: csDECMCS *)
2225
let () = alias "IANA/csDECMCS" "DEC-MCS"
2226
let () = alias "IANA/dec" "DEC-MCS"
2227
let () = alias "IANA/DEC-MCS" "DEC-MCS"
2228
(* Name: hp-roman8 [HP-PCL5,RFC1345,KXS2] *)
2229
(* Alias: roman8 *)
2230
(* Alias: r8 *)
2231
(* Alias: csHPRoman8 *)
2232
let () = alias "IANA/csHPRoman8" "HP-ROMAN8"
2233
let () = alias "IANA/r8" "HP-ROMAN8"
2234
let () = alias "IANA/roman8" "HP-ROMAN8"
2235
let () = alias "IANA/hp-roman8" "HP-ROMAN8"
2236
(* Name: macintosh [RFC1345,KXS2] *)
2237
(* Alias: mac *)
2238
(* Alias: csMacintosh *)
2239
let () = alias "IANA/csMacintosh" "MACINTOSH"
2240
let () = alias "IANA/mac" "MACINTOSH"
2241
let () = alias "IANA/macintosh" "MACINTOSH"
2242
(* Name: IBM037 [RFC1345,KXS2] *)
2243
(* Alias: cp037 *)
2244
(* Alias: ebcdic-cp-us *)
2245
(* Alias: ebcdic-cp-ca *)
2246
(* Alias: ebcdic-cp-wt *)
2247
(* Alias: ebcdic-cp-nl *)
2248
(* Alias: csIBM037 *)
2249
let () = alias "IANA/csIBM037" "IBM037"
2250
let () = alias "IANA/ebcdic-cp-nl" "IBM037"
2251
let () = alias "IANA/ebcdic-cp-wt" "IBM037"
2252
let () = alias "IANA/ebcdic-cp-ca" "IBM037"
2253
let () = alias "IANA/ebcdic-cp-us" "IBM037"
2254
let () = alias "IANA/cp037" "IBM037"
2255
let () = alias "IANA/IBM037" "IBM037"
2256
(* Name: IBM038 [RFC1345,KXS2] *)
2257
(* Alias: EBCDIC-INT *)
2258
(* Alias: cp038 *)
2259
(* Alias: csIBM038 *)
2260
let () = alias "IANA/csIBM038" "IBM038"
2261
let () = alias "IANA/cp038" "IBM038"
2262
let () = alias "IANA/EBCDIC-INT" "IBM038"
2263
let () = alias "IANA/IBM038" "IBM038"
2264
(* Name: IBM273 [RFC1345,KXS2] *)
2265
(* Alias: CP273 *)
2266
(* Alias: csIBM273 *)
2267
let () = alias "IANA/csIBM273" "IBM273"
2268
let () = alias "IANA/CP273" "IBM273"
2269
let () = alias "IANA/IBM273" "IBM273"
2270
(* Name: IBM274 [RFC1345,KXS2] *)
2271
(* Alias: EBCDIC-BE *)
2272
(* Alias: CP274 *)
2273
(* Alias: csIBM274 *)
2274
let () = alias "IANA/csIBM274" "IBM274"
2275
let () = alias "IANA/CP274" "IBM274"
2276
let () = alias "IANA/EBCDIC-BE" "IBM274"
2277
let () = alias "IANA/IBM274" "IBM274"
2278
(* Name: IBM275 [RFC1345,KXS2] *)
2279
(* Alias: EBCDIC-BR *)
2280
(* Alias: cp275 *)
2281
(* Alias: csIBM275 *)
2282
let () = alias "IANA/csIBM275" "IBM275"
2283
let () = alias "IANA/cp275" "IBM275"
2284
let () = alias "IANA/EBCDIC-BR" "IBM275"
2285
let () = alias "IANA/IBM275" "IBM275"
2286
(* Name: IBM277 [RFC1345,KXS2] *)
2287
(* Alias: EBCDIC-CP-DK *)
2288
(* Alias: EBCDIC-CP-NO *)
2289
(* Alias: csIBM277 *)
2290
let () = alias "IANA/csIBM277" "IBM277"
2291
let () = alias "IANA/EBCDIC-CP-NO" "IBM277"
2292
let () = alias "IANA/EBCDIC-CP-DK" "IBM277"
2293
let () = alias "IANA/IBM277" "IBM277"
2294
(* Name: IBM278 [RFC1345,KXS2] *)
2295
(* Alias: CP278 *)
2296
(* Alias: ebcdic-cp-fi *)
2297
(* Alias: ebcdic-cp-se *)
2298
(* Alias: csIBM278 *)
2299
let () = alias "IANA/csIBM278" "IBM278"
2300
let () = alias "IANA/ebcdic-cp-se" "IBM278"
2301
let () = alias "IANA/ebcdic-cp-fi" "IBM278"
2302
let () = alias "IANA/CP278" "IBM278"
2303
let () = alias "IANA/IBM278" "IBM278"
2304
(* Name: IBM280 [RFC1345,KXS2] *)
2305
(* Alias: CP280 *)
2306
(* Alias: ebcdic-cp-it *)
2307
(* Alias: csIBM280 *)
2308
let () = alias "IANA/csIBM280" "IBM280"
2309
let () = alias "IANA/ebcdic-cp-it" "IBM280"
2310
let () = alias "IANA/CP280" "IBM280"
2311
let () = alias "IANA/IBM280" "IBM280"
2312
(* Name: IBM281 [RFC1345,KXS2] *)
2313
(* Alias: EBCDIC-JP-E *)
2314
(* Alias: cp281 *)
2315
(* Alias: csIBM281 *)
2316
let () = alias "IANA/csIBM281" "IBM281"
2317
let () = alias "IANA/cp281" "IBM281"
2318
let () = alias "IANA/EBCDIC-JP-E" "IBM281"
2319
let () = alias "IANA/IBM281" "IBM281"
2320
(* Name: IBM284 [RFC1345,KXS2] *)
2321
(* Alias: CP284 *)
2322
(* Alias: ebcdic-cp-es *)
2323
(* Alias: csIBM284 *)
2324
let () = alias "IANA/csIBM284" "IBM284"
2325
let () = alias "IANA/ebcdic-cp-es" "IBM284"
2326
let () = alias "IANA/CP284" "IBM284"
2327
let () = alias "IANA/IBM284" "IBM284"
2328
(* Name: IBM285 [RFC1345,KXS2] *)
2329
(* Alias: CP285 *)
2330
(* Alias: ebcdic-cp-gb *)
2331
(* Alias: csIBM285 *)
2332
let () = alias "IANA/csIBM285" "IBM285"
2333
let () = alias "IANA/ebcdic-cp-gb" "IBM285"
2334
let () = alias "IANA/CP285" "IBM285"
2335
let () = alias "IANA/IBM285" "IBM285"
2336
(* Name: IBM290 [RFC1345,KXS2] *)
2337
(* Alias: cp290 *)
2338
(* Alias: EBCDIC-JP-kana *)
2339
(* Alias: csIBM290 *)
2340
let () = alias "IANA/csIBM290" "IBM290"
2341
let () = alias "IANA/EBCDIC-JP-kana" "IBM290"
2342
let () = alias "IANA/cp290" "IBM290"
2343
let () = alias "IANA/IBM290" "IBM290"
2344
(* Name: IBM297 [RFC1345,KXS2] *)
2345
(* Alias: cp297 *)
2346
(* Alias: ebcdic-cp-fr *)
2347
(* Alias: csIBM297 *)
2348
let () = alias "IANA/csIBM297" "IBM297"
2349
let () = alias "IANA/ebcdic-cp-fr" "IBM297"
2350
let () = alias "IANA/cp297" "IBM297"
2351
let () = alias "IANA/IBM297" "IBM297"
2352
(* Name: IBM420 [RFC1345,KXS2] *)
2353
(* Alias: cp420 *)
2354
(* Alias: ebcdic-cp-ar1 *)
2355
(* Alias: csIBM420 *)
2356
let () = alias "IANA/csIBM420" "IBM420"
2357
let () = alias "IANA/ebcdic-cp-ar1" "IBM420"
2358
let () = alias "IANA/cp420" "IBM420"
2359
let () = alias "IANA/IBM420" "IBM420"
2360
(* Name: IBM423 [RFC1345,KXS2] *)
2361
(* Alias: cp423 *)
2362
(* Alias: ebcdic-cp-gr *)
2363
(* Alias: csIBM423 *)
2364
let () = alias "IANA/csIBM423" "IBM423"
2365
let () = alias "IANA/ebcdic-cp-gr" "IBM423"
2366
let () = alias "IANA/cp423" "IBM423"
2367
let () = alias "IANA/IBM423" "IBM423"
2368
(* Name: IBM424 [RFC1345,KXS2] *)
2369
(* Alias: cp424 *)
2370
(* Alias: ebcdic-cp-he *)
2371
(* Alias: csIBM424 *)
2372
let () = alias "IANA/csIBM424" "IBM424"
2373
let () = alias "IANA/ebcdic-cp-he" "IBM424"
2374
let () = alias "IANA/cp424" "IBM424"
2375
let () = alias "IANA/IBM424" "IBM424"
2376
(* Name: IBM437 [RFC1345,KXS2] *)
2377
(* Alias: cp437 *)
2378
(* Alias: 437 *)
2379
(* Alias: csPC8CodePage437 *)
2380
let () = alias "IANA/csPC8CodePage437" "IBM437"
2381
let () = alias "IANA/437" "IBM437"
2382
let () = alias "IANA/cp437" "IBM437"
2383
let () = alias "IANA/IBM437" "IBM437"
2384
(* Name: IBM500 [RFC1345,KXS2] *)
2385
(* Alias: CP500 *)
2386
(* Alias: ebcdic-cp-be *)
2387
(* Alias: ebcdic-cp-ch *)
2388
(* Alias: csIBM500 *)
2389
let () = alias "IANA/csIBM500" "IBM500"
2390
let () = alias "IANA/ebcdic-cp-ch" "IBM500"
2391
let () = alias "IANA/ebcdic-cp-be" "IBM500"
2392
let () = alias "IANA/CP500" "IBM500"
2393
let () = alias "IANA/IBM500" "IBM500"
2394
(* Name: IBM775 [HP-PCL5] *)
2395
(* Alias: cp775 *)
2396
(* Alias: csPC775Baltic *)
2397
(* Name: IBM850 [RFC1345,KXS2] *)
2398
(* Alias: cp850 *)
2399
(* Alias: 850 *)
2400
(* Alias: csPC850Multilingual *)
2401
let () = alias "IANA/csPC850Multilingual" "IBM850"
2402
let () = alias "IANA/850" "IBM850"
2403
let () = alias "IANA/cp850" "IBM850"
2404
let () = alias "IANA/IBM850" "IBM850"
2405
(* Name: IBM851 [RFC1345,KXS2] *)
2406
(* Alias: cp851 *)
2407
(* Alias: 851 *)
2408
(* Alias: csIBM851 *)
2409
let () = alias "IANA/csIBM851" "IBM851"
2410
let () = alias "IANA/851" "IBM851"
2411
let () = alias "IANA/cp851" "IBM851"
2412
let () = alias "IANA/IBM851" "IBM851"
2413
(* Name: IBM852 [RFC1345,KXS2] *)
2414
(* Alias: cp852 *)
2415
(* Alias: 852 *)
2416
(* Alias: csPCp852 *)
2417
let () = alias "IANA/csPCp852" "IBM852"
2418
let () = alias "IANA/852" "IBM852"
2419
let () = alias "IANA/cp852" "IBM852"
2420
let () = alias "IANA/IBM852" "IBM852"
2421
(* Name: IBM855 [RFC1345,KXS2] *)
2422
(* Alias: cp855 *)
2423
(* Alias: 855 *)
2424
(* Alias: csIBM855 *)
2425
let () = alias "IANA/csIBM855" "IBM855"
2426
let () = alias "IANA/855" "IBM855"
2427
let () = alias "IANA/cp855" "IBM855"
2428
let () = alias "IANA/IBM855" "IBM855"
2429
(* Name: IBM857 [RFC1345,KXS2] *)
2430
(* Alias: cp857 *)
2431
(* Alias: 857 *)
2432
(* Alias: csIBM857 *)
2433
let () = alias "IANA/csIBM857" "IBM857"
2434
let () = alias "IANA/857" "IBM857"
2435
let () = alias "IANA/cp857" "IBM857"
2436
let () = alias "IANA/IBM857" "IBM857"
2437
(* Name: IBM860 [RFC1345,KXS2] *)
2438
(* Alias: cp860 *)
2439
(* Alias: 860 *)
2440
(* Alias: csIBM860 *)
2441
let () = alias "IANA/csIBM860" "IBM860"
2442
let () = alias "IANA/860" "IBM860"
2443
let () = alias "IANA/cp860" "IBM860"
2444
let () = alias "IANA/IBM860" "IBM860"
2445
(* Name: IBM861 [RFC1345,KXS2] *)
2446
(* Alias: cp861 *)
2447
(* Alias: 861 *)
2448
(* Alias: cp-is *)
2449
(* Alias: csIBM861 *)
2450
let () = alias "IANA/csIBM861" "IBM861"
2451
let () = alias "IANA/cp-is" "IBM861"
2452
let () = alias "IANA/861" "IBM861"
2453
let () = alias "IANA/cp861" "IBM861"
2454
let () = alias "IANA/IBM861" "IBM861"
2455
(* Name: IBM862 [RFC1345,KXS2] *)
2456
(* Alias: cp862 *)
2457
(* Alias: 862 *)
2458
(* Alias: csPC862LatinHebrew *)
2459
let () = alias "IANA/csPC862LatinHebrew" "IBM862"
2460
let () = alias "IANA/862" "IBM862"
2461
let () = alias "IANA/cp862" "IBM862"
2462
let () = alias "IANA/IBM862" "IBM862"
2463
(* Name: IBM863 [RFC1345,KXS2] *)
2464
(* Alias: cp863 *)
2465
(* Alias: 863 *)
2466
(* Alias: csIBM863 *)
2467
let () = alias "IANA/csIBM863" "IBM863"
2468
let () = alias "IANA/863" "IBM863"
2469
let () = alias "IANA/cp863" "IBM863"
2470
let () = alias "IANA/IBM863" "IBM863"
2471
(* Name: IBM864 [RFC1345,KXS2] *)
2472
(* Alias: cp864 *)
2473
(* Alias: csIBM864 *)
2474
let () = alias "IANA/csIBM864" "IBM864"
2475
let () = alias "IANA/cp864" "IBM864"
2476
let () = alias "IANA/IBM864" "IBM864"
2477
(* Name: IBM865 [RFC1345,KXS2] *)
2478
(* Alias: cp865 *)
2479
(* Alias: 865 *)
2480
(* Alias: csIBM865 *)
2481
let () = alias "IANA/csIBM865" "IBM865"
2482
let () = alias "IANA/865" "IBM865"
2483
let () = alias "IANA/cp865" "IBM865"
2484
let () = alias "IANA/IBM865" "IBM865"
2485
(* Name: IBM866 [Pond] *)
2486
(* Alias: cp866 *)
2487
(* Alias: 866 *)
2488
(* Alias: csIBM866 *)
2489
let () = alias "IANA/csIBM866" "IBM866"
2490
let () = alias "IANA/866" "IBM866"
2491
let () = alias "IANA/cp866" "IBM866"
2492
let () = alias "IANA/IBM866" "IBM866"
2493
(* Name: IBM868 [RFC1345,KXS2] *)
2494
(* Alias: CP868 *)
2495
(* Alias: cp-ar *)
2496
(* Alias: csIBM868 *)
2497
let () = alias "IANA/csIBM868" "IBM868"
2498
let () = alias "IANA/cp-ar" "IBM868"
2499
let () = alias "IANA/CP868" "IBM868"
2500
let () = alias "IANA/IBM868" "IBM868"
2501
(* Name: IBM869 [RFC1345,KXS2] *)
2502
(* Alias: cp869 *)
2503
(* Alias: 869 *)
2504
(* Alias: cp-gr *)
2505
(* Alias: csIBM869 *)
2506
let () = alias "IANA/csIBM869" "IBM869"
2507
let () = alias "IANA/cp-gr" "IBM869"
2508
let () = alias "IANA/869" "IBM869"
2509
let () = alias "IANA/cp869" "IBM869"
2510
let () = alias "IANA/IBM869" "IBM869"
2511
(* Name: IBM870 [RFC1345,KXS2] *)
2512
(* Alias: CP870 *)
2513
(* Alias: ebcdic-cp-roece *)
2514
(* Alias: ebcdic-cp-yu *)
2515
(* Alias: csIBM870 *)
2516
let () = alias "IANA/csIBM870" "IBM870"
2517
let () = alias "IANA/ebcdic-cp-yu" "IBM870"
2518
let () = alias "IANA/ebcdic-cp-roece" "IBM870"
2519
let () = alias "IANA/CP870" "IBM870"
2520
let () = alias "IANA/IBM870" "IBM870"
2521
(* Name: IBM871 [RFC1345,KXS2] *)
2522
(* Alias: CP871 *)
2523
(* Alias: ebcdic-cp-is *)
2524
(* Alias: csIBM871 *)
2525
let () = alias "IANA/csIBM871" "IBM871"
2526
let () = alias "IANA/ebcdic-cp-is" "IBM871"
2527
let () = alias "IANA/CP871" "IBM871"
2528
let () = alias "IANA/IBM871" "IBM871"
2529
(* Name: IBM880 [RFC1345,KXS2] *)
2530
(* Alias: cp880 *)
2531
(* Alias: EBCDIC-Cyrillic *)
2532
(* Alias: csIBM880 *)
2533
let () = alias "IANA/csIBM880" "IBM880"
2534
let () = alias "IANA/EBCDIC-Cyrillic" "IBM880"
2535
let () = alias "IANA/cp880" "IBM880"
2536
let () = alias "IANA/IBM880" "IBM880"
2537
(* Name: IBM891 [RFC1345,KXS2] *)
2538
(* Alias: cp891 *)
2539
(* Alias: csIBM891 *)
2540
let () = alias "IANA/csIBM891" "IBM891"
2541
let () = alias "IANA/cp891" "IBM891"
2542
let () = alias "IANA/IBM891" "IBM891"
2543
(* Name: IBM903 [RFC1345,KXS2] *)
2544
(* Alias: cp903 *)
2545
(* Alias: csIBM903 *)
2546
let () = alias "IANA/csIBM903" "IBM903"
2547
let () = alias "IANA/cp903" "IBM903"
2548
let () = alias "IANA/IBM903" "IBM903"
2549
(* Name: IBM904 [RFC1345,KXS2] *)
2550
(* Alias: cp904 *)
2551
(* Alias: 904 *)
2552
(* Alias: csIBBM904 *)
2553
let () = alias "IANA/csIBBM904" "IBM904"
2554
let () = alias "IANA/904" "IBM904"
2555
let () = alias "IANA/cp904" "IBM904"
2556
let () = alias "IANA/IBM904" "IBM904"
2557
(* Name: IBM905 [RFC1345,KXS2] *)
2558
(* Alias: CP905 *)
2559
(* Alias: ebcdic-cp-tr *)
2560
(* Alias: csIBM905 *)
2561
let () = alias "IANA/csIBM905" "IBM905"
2562
let () = alias "IANA/ebcdic-cp-tr" "IBM905"
2563
let () = alias "IANA/CP905" "IBM905"
2564
let () = alias "IANA/IBM905" "IBM905"
2565
(* Name: IBM918 [RFC1345,KXS2] *)
2566
(* Alias: CP918 *)
2567
(* Alias: ebcdic-cp-ar2 *)
2568
(* Alias: csIBM918 *)
2569
let () = alias "IANA/csIBM918" "IBM918"
2570
let () = alias "IANA/ebcdic-cp-ar2" "IBM918"
2571
let () = alias "IANA/CP918" "IBM918"
2572
let () = alias "IANA/IBM918" "IBM918"
2573
(* Name: IBM1026 [RFC1345,KXS2] *)
2574
(* Alias: CP1026 *)
2575
(* Alias: csIBM1026 *)
2576
let () = alias "IANA/csIBM1026" "IBM1026"
2577
let () = alias "IANA/CP1026" "IBM1026"
2578
let () = alias "IANA/IBM1026" "IBM1026"
2579
(* Name: EBCDIC-AT-DE [RFC1345,KXS2] *)
2580
(* Alias: csIBMEBCDICATDE *)
2581
let () = alias "IANA/csIBMEBCDICATDE" "EBCDIC-AT-DE"
2582
let () = alias "IANA/EBCDIC-AT-DE" "EBCDIC-AT-DE"
2583
(* Name: EBCDIC-AT-DE-A [RFC1345,KXS2] *)
2584
(* Alias: csEBCDICATDEA *)
2585
let () = alias "IANA/csEBCDICATDEA" "EBCDIC-AT-DE-A"
2586
let () = alias "IANA/EBCDIC-AT-DE-A" "EBCDIC-AT-DE-A"
2587
(* Name: EBCDIC-CA-FR [RFC1345,KXS2] *)
2588
(* Alias: csEBCDICCAFR *)
2589
let () = alias "IANA/csEBCDICCAFR" "EBCDIC-CA-FR"
2590
let () = alias "IANA/EBCDIC-CA-FR" "EBCDIC-CA-FR"
2591
(* Name: EBCDIC-DK-NO [RFC1345,KXS2] *)
2592
(* Alias: csEBCDICDKNO *)
2593
let () = alias "IANA/csEBCDICDKNO" "EBCDIC-DK-NO"
2594
let () = alias "IANA/EBCDIC-DK-NO" "EBCDIC-DK-NO"
2595
(* Name: EBCDIC-DK-NO-A [RFC1345,KXS2] *)
2596
(* Alias: csEBCDICDKNOA *)
2597
let () = alias "IANA/csEBCDICDKNOA" "EBCDIC-DK-NO-A"
2598
let () = alias "IANA/EBCDIC-DK-NO-A" "EBCDIC-DK-NO-A"
2599
(* Name: EBCDIC-FI-SE [RFC1345,KXS2] *)
2600
(* Alias: csEBCDICFISE *)
2601
let () = alias "IANA/csEBCDICFISE" "EBCDIC-FI-SE"
2602
let () = alias "IANA/EBCDIC-FI-SE" "EBCDIC-FI-SE"
2603
(* Name: EBCDIC-FI-SE-A [RFC1345,KXS2] *)
2604
(* Alias: csEBCDICFISEA *)
2605
let () = alias "IANA/csEBCDICFISEA" "EBCDIC-FI-SE-A"
2606
let () = alias "IANA/EBCDIC-FI-SE-A" "EBCDIC-FI-SE-A"
2607
(* Name: EBCDIC-FR [RFC1345,KXS2] *)
2608
(* Alias: csEBCDICFR *)
2609
let () = alias "IANA/csEBCDICFR" "EBCDIC-FR"
2610
let () = alias "IANA/EBCDIC-FR" "EBCDIC-FR"
2611
(* Name: EBCDIC-IT [RFC1345,KXS2] *)
2612
(* Alias: csEBCDICIT *)
2613
let () = alias "IANA/csEBCDICIT" "EBCDIC-IT"
2614
let () = alias "IANA/EBCDIC-IT" "EBCDIC-IT"
2615
(* Name: EBCDIC-PT [RFC1345,KXS2] *)
2616
(* Alias: csEBCDICPT *)
2617
let () = alias "IANA/csEBCDICPT" "EBCDIC-PT"
2618
let () = alias "IANA/EBCDIC-PT" "EBCDIC-PT"
2619
(* Name: EBCDIC-ES [RFC1345,KXS2] *)
2620
(* Alias: csEBCDICES *)
2621
let () = alias "IANA/csEBCDICES" "EBCDIC-ES"
2622
let () = alias "IANA/EBCDIC-ES" "EBCDIC-ES"
2623
(* Name: EBCDIC-ES-A [RFC1345,KXS2] *)
2624
(* Alias: csEBCDICESA *)
2625
let () = alias "IANA/csEBCDICESA" "EBCDIC-ES-A"
2626
let () = alias "IANA/EBCDIC-ES-A" "EBCDIC-ES-A"
2627
(* Name: EBCDIC-ES-S [RFC1345,KXS2] *)
2628
(* Alias: csEBCDICESS *)
2629
let () = alias "IANA/csEBCDICESS" "EBCDIC-ES-S"
2630
let () = alias "IANA/EBCDIC-ES-S" "EBCDIC-ES-S"
2631
(* Name: EBCDIC-UK [RFC1345,KXS2] *)
2632
(* Alias: csEBCDICUK *)
2633
let () = alias "IANA/csEBCDICUK" "EBCDIC-UK"
2634
let () = alias "IANA/EBCDIC-UK" "EBCDIC-UK"
2635
(* Name: EBCDIC-US [RFC1345,KXS2] *)
2636
(* Alias: csEBCDICUS *)
2637
let () = alias "IANA/csEBCDICUS" "EBCDIC-US"
2638
let () = alias "IANA/EBCDIC-US" "EBCDIC-US"
2639
(* Name: UNKNOWN-8BIT [RFC1428] *)
2640
(* Alias: csUnknown8BiT *)
2641
(* Name: MNEMONIC [RFC1345,KXS2] *)
2642
(* Alias: csMnemonic *)
2643
(* Name: MNEM [RFC1345,KXS2] *)
2644
(* Alias: csMnem *)
2645
(* Name: VISCII [RFC1456] *)
2646
(* Alias: csVISCII *)
2647
let () = alias "IANA/csVISCII" "VISCII"
2648
let () = alias "IANA/VISCII" "VISCII"
2649
(* Name: VIQR [RFC1456] *)
2650
(* Alias: csVIQR *)
2651
(* Name: KOI8-R (preferred MIME name) [RFC1489] *)
2652
(* Alias: csKOI8R *)
2653
let () = alias "IANA/csKOI8R" "KOI8-R"
2654
let () = alias "IANA/KOI8-R" "KOI8-R"
2655
(* Name: KOI8-U [RFC2319] *)
2656
let () = alias "IANA/KOI8-U" "KOI8-U"
2657
(* Name: IBM00858 *)
2658
(* Alias: CCSID00858 *)
2659
(* Alias: CP00858 *)
2660
(* Alias: PC-Multilingual-850+euro *)
2661
(* Name: IBM00924 *)
2662
(* Alias: CCSID00924 *)
2663
(* Alias: CP00924 *)
2664
(* Alias: ebcdic-Latin9--euro *)
2665
(* Name: IBM01140 *)
2666
(* Alias: CCSID01140 *)
2667
(* Alias: CP01140 *)
2668
(* Alias: ebcdic-us-37+euro *)
2669
(* Name: IBM01141 *)
2670
(* Alias: CCSID01141 *)
2671
(* Alias: CP01141 *)
2672
(* Alias: ebcdic-de-273+euro *)
2673
(* Name: IBM01142 *)
2674
(* Alias: CCSID01142 *)
2675
(* Alias: CP01142 *)
2676
(* Alias: ebcdic-dk-277+euro *)
2677
(* Alias: ebcdic-no-277+euro *)
2678
(* Name: IBM01143 *)
2679
(* Alias: CCSID01143 *)
2680
(* Alias: CP01143 *)
2681
(* Alias: ebcdic-fi-278+euro *)
2682
(* Alias: ebcdic-se-278+euro *)
2683
(* Name: IBM01144 *)
2684
(* Alias: CCSID01144 *)
2685
(* Alias: CP01144 *)
2686
(* Alias: ebcdic-it-280+euro *)
2687
(* Name: IBM01145 *)
2688
(* Alias: CCSID01145 *)
2689
(* Alias: CP01145 *)
2690
(* Alias: ebcdic-es-284+euro *)
2691
(* Name: IBM01146 *)
2692
(* Alias: CCSID01146 *)
2693
(* Alias: CP01146 *)
2694
(* Alias: ebcdic-gb-285+euro *)
2695
(* Name: IBM01147 *)
2696
(* Alias: CCSID01147 *)
2697
(* Alias: CP01147 *)
2698
(* Alias: ebcdic-fr-297+euro *)
2699
(* Name: IBM01148 *)
2700
(* Alias: CCSID01148 *)
2701
(* Alias: CP01148 *)
2702
(* Alias: ebcdic-international-500+euro *)
2703
(* Name: IBM01149 *)
2704
(* Alias: CCSID01149 *)
2705
(* Alias: CP01149 *)
2706
(* Alias: ebcdic-is-871+euro *)
2707
(* Name: Big5-HKSCS *)
2708
let () = alias "IANA/Big5-HKSCS" "BIG5-HKSCS"
2709
(* Name: UNICODE-1-1 [RFC1641] *)
2710
(* Alias: csUnicode11 *)
2711
(* Name: SCSU *)
2712
(* Name: UTF-7 [RFC2152] *)
2713
(* Name: UTF-16BE [RFC2781] *)
2714
let () = alias "IANA/UTF-16BE" "UTF-16BE"
2715
(* Name: UTF-16LE [RFC2781] *)
2716
let () = alias "IANA/UTF-16LE" "UTF-16LE"
2717
(* Name: UTF-16 [RFC2781] *)
2718
let () = alias "IANA/UTF-16" "UTF-16"
2719
(* Name: UNICODE-1-1-UTF-7 [RFC1642] *)
2720
(* Alias: csUnicode11UTF7 *)
2721
(* Name: UTF-8 [RFC2279] *)
2722
let () = alias "IANA/UTF-8" "UTF-8"
2723
(* Name: ISO-8859-13 *)
2724
let () = alias "IANA/ISO-8859-13" "ISO-8859-13"
2725
(* Name: ISO-8859-14 *)
2726
(* Alias: iso-ir-199 *)
2727
(* Alias: ISO_8859-14:1998 *)
2728
(* Alias: ISO_8859-14 *)
2729
(* Alias: latin8 *)
2730
(* Alias: iso-celtic *)
2731
(* Alias: l8 *)
2732
let () = alias "IANA/l8" "ISO-8859-14"
2733
let () = alias "IANA/iso-celtic" "ISO-8859-14"
2734
let () = alias "IANA/latin8" "ISO-8859-14"
2735
let () = alias "IANA/ISO_8859-14" "ISO-8859-14"
2736
let () = alias "IANA/ISO_8859-14:1998" "ISO-8859-14"
2737
let () = alias "IANA/iso-ir-199" "ISO-8859-14"
2738
let () = alias "IANA/ISO-8859-14" "ISO-8859-14"
2739
(* Name: ISO-8859-15 *)
2740
(* Alias: ISO_8859-15 *)
2741
let () = alias "IANA/ISO_8859-15" "ISO-8859-15"
2742
let () = alias "IANA/ISO-8859-15" "ISO-8859-15"
2743
(* Name: ISO-8859-16 *)
2744
let () = alias "IANA/ISO-8859-16" "ISO-8859-16"
2745
(* Name: JIS_Encoding *)
2746
(* Alias: csJISEncoding *)
2747
(* Name: Shift_JIS (preferred MIME name) *)
2748
(* Alias: MS_Kanji *)
2749
(* Alias: csShiftJIS *)
2750
let () = alias "IANA/csShiftJIS" "SHIFT_JIS"
2751
let () = alias "IANA/MS_Kanji" "SHIFT_JIS"
2752
let () = alias "IANA/Shift_JIS" "SHIFT_JIS"
2753
(* Name: Extended_UNIX_Code_Packed_Format_for_Japanese *)
2754
(* Alias: csEUCPkdFmtJapanese *)
2755
(* Alias: EUC-JP (preferred MIME name) *)
2756
let () = alias "IANA/EUC-JP" "EUC-JP"
2757
let () = alias "IANA/csEUCPkdFmtJapanese" "EUC-JP"
2758
let () = alias "IANA/Extended_UNIX_Code_Packed_Format_for_Japanese" "EUC-JP"
2759
(* Name: Extended_UNIX_Code_Fixed_Width_for_Japanese *)
2760
(* Alias: csEUCFixWidJapanese *)
2761
(* Name: ISO-10646-UCS-Basic *)
2762
(* Alias: csUnicodeASCII *)
2763
(* Name: ISO-10646-Unicode-Latin1 *)
2764
(* Alias: csUnicodeLatin1 *)
2765
(* Alias: ISO-10646 *)
2766
(* Name: ISO-10646-J-1 *)
2767
(* Name: ISO-Unicode-IBM-1261 *)
2768
(* Alias: csUnicodeIBM1261 *)
2769
(* Name: ISO-Unicode-IBM-1268 *)
2770
(* Alias: csUnidoceIBM1268 *)
2771
(* Name: ISO-Unicode-IBM-1276 *)
2772
(* Alias: csUnicodeIBM1276 *)
2773
(* Name: ISO-Unicode-IBM-1264 *)
2774
(* Alias: csUnicodeIBM1264 *)
2775
(* Name: ISO-Unicode-IBM-1265 *)
2776
(* Alias: csUnicodeIBM1265 *)
2777
(* Name: ISO-8859-1-Windows-3.0-Latin-1 [HP-PCL5] *)
2778
(* Alias: csWindows30Latin1 *)
2779
(* Name: ISO-8859-1-Windows-3.1-Latin-1 [HP-PCL5] *)
2780
(* Alias: csWindows31Latin1 *)
2781
(* Name: ISO-8859-2-Windows-Latin-2 [HP-PCL5] *)
2782
(* Alias: csWindows31Latin2 *)
2783
(* Name: ISO-8859-9-Windows-Latin-5 [HP-PCL5] *)
2784
(* Alias: csWindows31Latin5 *)
2785
(* Name: Adobe-Standard-Encoding [Adobe] *)
2786
(* Alias: csAdobeStandardEncoding *)
2787
(* Name: Ventura-US [HP-PCL5] *)
2788
(* Alias: csVenturaUS *)
2789
(* Name: Ventura-International [HP-PCL5] *)
2790
(* Alias: csVenturaInternational *)
2791
(* Name: PC8-Danish-Norwegian [HP-PCL5] *)
2792
(* Alias: csPC8DanishNorwegian *)
2793
(* Name: PC8-Turkish [HP-PCL5] *)
2794
(* Alias: csPC8Turkish *)
2795
(* Name: IBM-Symbols [IBM-CIDT] *)
2796
(* Alias: csIBMSymbols *)
2797
(* Name: IBM-Thai [IBM-CIDT] *)
2798
(* Alias: csIBMThai *)
2799
(* Name: HP-Legal [HP-PCL5] *)
2800
(* Alias: csHPLegal *)
2801
(* Name: HP-Pi-font [HP-PCL5] *)
2802
(* Alias: csHPPiFont *)
2803
(* Name: HP-Math8 [HP-PCL5] *)
2804
(* Alias: csHPMath8 *)
2805
(* Name: Adobe-Symbol-Encoding [Adobe] *)
2806
(* Alias: csHPPSMath *)
2807
(* Name: HP-DeskTop [HP-PCL5] *)
2808
(* Alias: csHPDesktop *)
2809
(* Name: Ventura-Math [HP-PCL5] *)
2810
(* Alias: csVenturaMath *)
2811
(* Name: Microsoft-Publishing [HP-PCL5] *)
2812
(* Alias: csMicrosoftPublishing *)
2813
(* Name: Windows-31J *)
2814
(* Alias: csWindows31J *)
2815
(* Name: GB2312 (preferred MIME name) *)
2816
(* Alias: csGB2312 *)
2817
let () = alias "IANA/csGB2312" "GB2312"
2818
let () = alias "IANA/GB2312" "GB2312"
2819
(* Name: Big5 (preferred MIME name) *)
2820
(* Alias: csBig5 *)
2821
let () = alias "IANA/csBig5" "BIG5"
2822
let () = alias "IANA/Big5" "BIG5"
2823
(* Name: windows-1250 *)
2824
(* Name: windows-1251 *)
2825
(* Name: windows-1252 *)
2826
(* Name: windows-1253 *)
2827
(* Name: windows-1254 *)
2828
(* Name: windows-1255 *)
2829
(* Name: windows-1256 *)
2830
(* Name: windows-1257 *)
2831
(* Name: windows-1258 *)
2832
(* Name: TIS-620 *)
2833
let () = alias "IANA/TIS-620" "TIS-620"
2834
(* Name: HZ-GB-2312 *)
Loggerhead is a web-based interface for Breezy
Version: 2.0.1