summaryrefslogtreecommitdiff
path: root/imap/src/c-client/utf8.c
blob: 844abefd9c49d1f14433f2a27157f4683db99cfe (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
1518
1519
1520
1521
1522
1523
1524
1525
1526
1527
1528
1529
1530
1531
1532
1533
1534
1535
1536
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
1554
1555
1556
1557
1558
1559
1560
1561
1562
1563
1564
1565
1566
1567
1568
1569
1570
1571
1572
1573
1574
1575
1576
1577
1578
1579
1580
1581
1582
1583
1584
1585
1586
1587
1588
1589
1590
1591
1592
1593
1594
1595
1596
1597
1598
1599
1600
1601
1602
1603
1604
1605
1606
1607
1608
1609
1610
1611
1612
1613
1614
1615
1616
1617
1618
1619
1620
1621
1622
1623
1624
1625
1626
1627
1628
1629
1630
1631
1632
1633
1634
1635
1636
1637
1638
1639
1640
1641
1642
1643
1644
1645
1646
1647
1648
1649
1650
1651
1652
1653
1654
1655
1656
1657
1658
1659
1660
1661
1662
1663
1664
1665
1666
1667
1668
1669
1670
1671
1672
1673
1674
1675
1676
1677
1678
1679
1680
1681
1682
1683
1684
1685
1686
1687
1688
1689
1690
1691
1692
1693
1694
1695
1696
1697
1698
1699
1700
1701
1702
1703
1704
1705
1706
1707
1708
1709
1710
1711
1712
1713
1714
1715
1716
1717
1718
1719
1720
1721
1722
1723
1724
1725
1726
1727
1728
1729
1730
1731
1732
1733
1734
1735
1736
1737
1738
1739
1740
1741
1742
1743
1744
1745
1746
1747
1748
1749
1750
1751
1752
1753
1754
1755
1756
1757
1758
1759
1760
1761
1762
1763
1764
1765
1766
1767
1768
1769
1770
1771
1772
1773
1774
1775
1776
1777
1778
1779
1780
1781
1782
1783
1784
1785
1786
1787
1788
1789
1790
1791
1792
1793
1794
1795
1796
1797
1798
1799
1800
1801
1802
1803
1804
1805
1806
1807
1808
1809
1810
1811
1812
1813
1814
1815
1816
1817
1818
1819
1820
1821
1822
1823
1824
1825
1826
1827
1828
1829
1830
1831
1832
1833
1834
1835
1836
1837
1838
1839
1840
1841
1842
1843
1844
1845
1846
1847
1848
1849
1850
1851
1852
1853
1854
1855
1856
1857
1858
1859
1860
1861
1862
1863
1864
1865
1866
1867
1868
1869
1870
1871
1872
1873
1874
1875
1876
1877
1878
1879
1880
1881
1882
1883
1884
1885
1886
1887
1888
1889
1890
1891
1892
1893
1894
1895
1896
1897
1898
1899
1900
1901
1902
1903
1904
1905
1906
1907
1908
1909
1910
1911
1912
1913
1914
1915
1916
1917
1918
1919
1920
1921
1922
1923
1924
1925
1926
1927
1928
1929
1930
1931
1932
1933
1934
1935
1936
1937
1938
1939
1940
1941
1942
1943
1944
1945
1946
1947
1948
1949
1950
1951
1952
1953
1954
1955
1956
1957
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
2034
2035
2036
2037
2038
2039
2040
2041
2042
2043
2044
2045
2046
2047
2048
2049
2050
2051
2052
2053
2054
2055
2056
2057
2058
2059
2060
2061
2062
2063
2064
2065
2066
2067
2068
2069
2070
2071
2072
2073
2074
2075
2076
2077
2078
2079
2080
2081
2082
2083
2084
2085
2086
2087
2088
2089
2090
2091
2092
2093
2094
2095
2096
2097
2098
2099
2100
2101
2102
2103
2104
2105
2106
2107
2108
2109
2110
2111
2112
2113
2114
2115
2116
2117
2118
2119
2120
2121
2122
2123
2124
2125
2126
2127
2128
2129
2130
2131
2132
2133
2134
2135
2136
2137
2138
2139
2140
2141
2142
2143
2144
2145
2146
2147
2148
2149
2150
2151
2152
2153
2154
2155
2156
2157
2158
2159
2160
2161
2162
2163
2164
2165
2166
2167
2168
2169
2170
2171
2172
2173
2174
2175
2176
2177
2178
2179
2180
2181
2182
2183
2184
2185
2186
2187
2188
2189
2190
2191
2192
2193
2194
2195
2196
2197
2198
2199
2200
2201
2202
2203
2204
2205
2206
2207
2208
2209
2210
2211
2212
2213
2214
2215
2216
2217
2218
2219
2220
2221
2222
2223
2224
2225
2226
2227
2228
2229
2230
2231
2232
2233
2234
2235
2236
2237
2238
2239
2240
2241
2242
2243
2244
2245
2246
2247
2248
2249
2250
2251
2252
2253
2254
2255
2256
2257
2258
2259
2260
2261
2262
2263
2264
2265
2266
2267
2268
2269
2270
2271
2272
2273
2274
2275
2276
2277
2278
2279
2280
2281
2282
2283
2284
2285
2286
2287
2288
2289
2290
2291
2292
2293
2294
2295
2296
2297
2298
2299
2300
2301
2302
2303
2304
2305
2306
2307
2308
2309
2310
2311
2312
2313
2314
2315
2316
2317
2318
2319
2320
2321
2322
2323
2324
2325
2326
2327
2328
2329
2330
2331
2332
2333
2334
2335
2336
2337
2338
2339
2340
2341
2342
2343
2344
2345
2346
2347
2348
2349
2350
2351
2352
2353
2354
2355
2356
2357
2358
2359
2360
2361
2362
2363
2364
2365
2366
2367
2368
2369
2370
2371
2372
2373
2374
2375
2376
2377
2378
2379
2380
2381
2382
2383
2384
2385
2386
2387
2388
2389
2390
2391
2392
2393
2394
2395
2396
2397
2398
2399
2400
2401
2402
2403
2404
2405
2406
2407
2408
2409
2410
2411
2412
2413
2414
2415
2416
2417
2418
2419
2420
2421
2422
2423
2424
2425
2426
2427
2428
2429
2430
2431
2432
2433
2434
2435
2436
2437
2438
2439
2440
2441
2442
2443
2444
2445
2446
2447
2448
2449
2450
2451
2452
2453
2454
2455
2456
2457
2458
2459
2460
2461
2462
2463
2464
2465
2466
2467
2468
2469
2470
2471
2472
2473
2474
2475
2476
2477
2478
2479
2480
2481
2482
2483
2484
2485
2486
2487
2488
2489
2490
2491
2492
2493
2494
2495
2496
2497
2498
2499
2500
2501
2502
2503
2504
2505
2506
2507
2508
2509
2510
2511
2512
2513
2514
2515
2516
2517
2518
2519
2520
2521
2522
2523
2524
2525
2526
2527
2528
2529
2530
2531
2532
2533
2534
2535
2536
2537
2538
2539
2540
2541
2542
2543
2544
2545
2546
2547
2548
2549
2550
2551
2552
2553
2554
2555
2556
2557
2558
2559
2560
2561
2562
2563
2564
2565
2566
2567
2568
2569
2570
2571
2572
2573
2574
2575
2576
2577
2578
2579
2580
2581
2582
2583
2584
2585
2586
2587
2588
2589
2590
2591
2592
2593
2594
2595
2596
2597
2598
2599
2600
2601
2602
/* ========================================================================
 * Copyright 2008-2010 Mark Crispin
 * ========================================================================
 */

/*
 * Program:	UTF-8 routines
 *
 * Author:	Mark Crispin
 *
 * Date:	11 June 1997
 * Last Edited:	28 May 2010
 * 
 * Previous versions of this file were
 *
 * Copyright 1988-2008 University of Washington
 *
 * Licensed under the Apache License, Version 2.0 (the "License");
 * you may not use this file except in compliance with the License.
 * You may obtain a copy of the License at
 *
 *     http://www.apache.org/licenses/LICENSE-2.0
 *
 */


#include <stdio.h>
#include <ctype.h>
#include "c-client.h"

/*	*** IMPORTANT ***
 *
 *  There is a very important difference between "character set" and "charset",
 * and the comments in this file reflect these differences.  A "character set"
 * (also known as "coded character set") is a mapping between codepoints and
 * characters.  A "charset" is as defined in MIME, and incorporates one or more
 * coded character sets in a character encoding scheme.  See RFC 2130 for more
 * details.
 */

static ucs4width_t ucs4width = NIL;

/* Character set conversion tables */

#include "iso_8859.c"		/* 8-bit single-byte coded graphic */
#include "koi8_r.c"		/* Cyrillic - Russia */
#include "koi8_u.c"		/* Cyrillic - Ukraine */
#include "tis_620.c"		/* Thai */
#include "viscii.c"		/* Vietnamese */
#include "windows.c"		/* Windows */
#include "ibm.c"		/* IBM */
#include "gb_2312.c"		/* Chinese (PRC) - simplified */
#include "gb_12345.c"		/* Chinese (PRC) - traditional */
#include "jis_0208.c"		/* Japanese - basic */
#include "jis_0212.c"		/* Japanese - supplementary */
#include "ksc_5601.c"		/* Korean */
#include "big5.c"		/* Taiwanese (ROC) - industrial standard */
#include "cns11643.c"		/* Taiwanese (ROC) - national standard */


#include "widths.c"		/* Unicode character widths */
#include "tmap.c"		/* Unicode titlecase mapping */
#include "decomtab.c"		/* Unicode decomposions */

/* EUC parameters */

#ifdef GBTOUNICODE		/* PRC simplified Chinese */
static const struct utf8_eucparam gb_param = {
  BASE_GB2312_KU,BASE_GB2312_TEN,MAX_GB2312_KU,MAX_GB2312_TEN,
  (void *) gb2312tab};
#endif


#ifdef GB12345TOUNICODE		/* PRC traditional Chinese */
static const struct utf8_eucparam gbt_param = {
  BASE_GB12345_KU,BASE_GB12345_TEN,MAX_GB12345_KU,MAX_GB12345_TEN,
  (void *) gb12345tab};
#endif


#ifdef BIG5TOUNICODE		/* ROC traditional Chinese */
static const struct utf8_eucparam big5_param[] = {
  {BASE_BIG5_KU,BASE_BIG5_TEN_0,MAX_BIG5_KU,MAX_BIG5_TEN_0,(void *) big5tab},
  {BASE_BIG5_KU,BASE_BIG5_TEN_1,MAX_BIG5_KU,MAX_BIG5_TEN_1,NIL}
};
#endif


#ifdef JISTOUNICODE		/* Japanese */
static const struct utf8_eucparam jis_param[] = {
  {BASE_JIS0208_KU,BASE_JIS0208_TEN,MAX_JIS0208_KU,MAX_JIS0208_TEN,
     (void *) jis0208tab},
  {MIN_KANA_8,0,MAX_KANA_8,0,(void *) KANA_8},
#ifdef JIS0212TOUNICODE		/* Japanese extended */
  {BASE_JIS0212_KU,BASE_JIS0212_TEN,MAX_JIS0212_KU,MAX_JIS0212_TEN,
     (void *) jis0212tab}
#else
  {0,0,0,0,NIL}
#endif
};
#endif


#ifdef KSCTOUNICODE		/* Korean */
static const struct utf8_eucparam ksc_param = {
  BASE_KSC5601_KU,BASE_KSC5601_TEN,MAX_KSC5601_KU,MAX_KSC5601_TEN,
  (void *) ksc5601tab};
#endif

/* List of supported charsets */

static const CHARSET utf8_csvalid[] = {
  {"US-ASCII",CT_ASCII,CF_PRIMARY | CF_DISPLAY | CF_POSTING,
   NIL,NIL,NIL},
  {"UTF-8",CT_UTF8,CF_PRIMARY | CF_DISPLAY | CF_POSTING,
   NIL,SC_UNICODE,NIL},
  {"UTF-7",CT_UTF7,CF_PRIMARY | CF_POSTING | CF_UNSUPRT,
   NIL,SC_UNICODE,"UTF-8"},
  {"ISO-8859-1",CT_1BYTE0,CF_PRIMARY | CF_DISPLAY | CF_POSTING,
   NIL,SC_LATIN_1,NIL},
  {"ISO-8859-2",CT_1BYTE,CF_PRIMARY | CF_DISPLAY | CF_POSTING,
   (void *) iso8859_2tab,SC_LATIN_2,NIL},
  {"ISO-8859-3",CT_1BYTE,CF_PRIMARY | CF_DISPLAY | CF_POSTING,
   (void *) iso8859_3tab,SC_LATIN_3,NIL},
  {"ISO-8859-4",CT_1BYTE,CF_PRIMARY | CF_DISPLAY | CF_POSTING,
   (void *) iso8859_4tab,SC_LATIN_4,NIL},
  {"ISO-8859-5",CT_1BYTE,CF_PRIMARY | CF_DISPLAY | CF_POSTING,
   (void *) iso8859_5tab,SC_CYRILLIC,"KOI8-R"},
  {"ISO-8859-6",CT_1BYTE,CF_PRIMARY | CF_DISPLAY | CF_POSTING,
   (void *) iso8859_6tab,SC_ARABIC,NIL},
  {"ISO-8859-7",CT_1BYTE,CF_PRIMARY | CF_DISPLAY | CF_POSTING,
   (void *) iso8859_7tab,SC_GREEK,NIL},
  {"ISO-8859-8",CT_1BYTE,CF_PRIMARY | CF_DISPLAY | CF_POSTING,
   (void *) iso8859_8tab,SC_HEBREW,NIL},
  {"ISO-8859-9",CT_1BYTE,CF_PRIMARY | CF_DISPLAY | CF_POSTING,
   (void *) iso8859_9tab,SC_LATIN_5,NIL},
  {"ISO-8859-10",CT_1BYTE,CF_PRIMARY | CF_DISPLAY | CF_POSTING,
   (void *) iso8859_10tab,SC_LATIN_6,NIL},
  {"ISO-8859-11",CT_1BYTE,CF_PRIMARY | CF_DISPLAY | CF_POSTING,
   (void *) iso8859_11tab,SC_THAI,NIL},
#if 0				/* ISO 8859-12 reserved for ISCII(?) */
  {"ISO-8859-12",CT_1BYTE,CF_PRIMARY | CF_DISPLAY | CF_POSTING,
   (void *) iso8859_12tab,NIL,NIL},
#endif
  {"ISO-8859-13",CT_1BYTE,CF_PRIMARY | CF_DISPLAY | CF_POSTING,
   (void *) iso8859_13tab,SC_LATIN_7,NIL},
  {"ISO-8859-14",CT_1BYTE,CF_PRIMARY | CF_DISPLAY | CF_POSTING,
   (void *) iso8859_14tab,SC_LATIN_8,NIL},
  {"ISO-8859-15",CT_1BYTE,CF_PRIMARY | CF_DISPLAY | CF_POSTING,
   (void *) iso8859_15tab,SC_LATIN_9,NIL},
  {"ISO-8859-16",CT_1BYTE,CF_PRIMARY | CF_DISPLAY | CF_POSTING,
   (void *) iso8859_16tab,SC_LATIN_10,NIL},
  {"KOI8-R",CT_1BYTE,CF_PRIMARY | CF_DISPLAY | CF_POSTING,
   (void *) koi8rtab,SC_CYRILLIC,NIL},
  {"KOI8-U",CT_1BYTE,CF_PRIMARY | CF_DISPLAY | CF_POSTING,
   (void *) koi8utab,SC_CYRILLIC | SC_UKRANIAN,NIL},
  {"KOI8-RU",CT_1BYTE,CF_DISPLAY,
   (void *) koi8utab,SC_CYRILLIC | SC_UKRANIAN,"KOI8-U"},
  {"TIS-620",CT_1BYTE,CF_PRIMARY | CF_DISPLAY | CF_POSTING,
   (void *) tis620tab,SC_THAI,"ISO-8859-11"},
  {"VISCII",CT_1BYTE8,CF_PRIMARY | CF_DISPLAY | CF_POSTING,
   (void *) visciitab,SC_VIETNAMESE,NIL},

#ifdef GBTOUNICODE
  {"GBK",CT_DBYTE,CF_PRIMARY | CF_DISPLAY | CF_POSTING,
     (void *) &gb_param,SC_CHINESE_SIMPLIFIED,NIL},
  {"GB2312",CT_DBYTE,CF_PRIMARY | CF_DISPLAY | CF_POSTING,
   (void *) &gb_param,SC_CHINESE_SIMPLIFIED,"GBK"},
  {"CN-GB",CT_DBYTE,CF_DISPLAY,
     (void *) &gb_param,SC_CHINESE_SIMPLIFIED,"GBK"},
#ifdef CNS1TOUNICODE
  {"ISO-2022-CN",CT_2022,CF_PRIMARY | CF_UNSUPRT,
     NIL,SC_CHINESE_SIMPLIFIED | SC_CHINESE_TRADITIONAL,
   NIL},
#endif
#endif
#ifdef GB12345TOUNICODE
  {"CN-GB-12345",CT_DBYTE,CF_PRIMARY | CF_DISPLAY,
     (void *) &gbt_param,SC_CHINESE_TRADITIONAL,"BIG5"},
#endif
#ifdef BIG5TOUNICODE
  {"BIG5",CT_DBYTE2,CF_PRIMARY | CF_DISPLAY | CF_POSTING,
     (void *) big5_param,SC_CHINESE_TRADITIONAL,NIL},
  {"CN-BIG5",CT_DBYTE2,CF_DISPLAY,
     (void *) big5_param,SC_CHINESE_TRADITIONAL,"BIG5"},
  {"BIG-5",CT_DBYTE2,CF_DISPLAY,
     (void *) big5_param,SC_CHINESE_TRADITIONAL,"BIG5"},
#endif
#ifdef JISTOUNICODE
  {"ISO-2022-JP",CT_2022,CF_PRIMARY | CF_DISPLAY | CF_POSTING,
     NIL,SC_JAPANESE,NIL},
  {"EUC-JP",CT_EUC,CF_PRIMARY | CF_DISPLAY,
     (void *) jis_param,SC_JAPANESE,"ISO-2022-JP"},
  {"SHIFT_JIS",CT_SJIS,CF_PRIMARY | CF_DISPLAY,
     NIL,SC_JAPANESE,"ISO-2022-JP"},
  {"SHIFT-JIS",CT_SJIS,CF_PRIMARY | CF_DISPLAY,
     NIL,SC_JAPANESE,"ISO-2022-JP"},
#ifdef JIS0212TOUNICODE
  {"ISO-2022-JP-1",CT_2022,CF_UNSUPRT,
     NIL,SC_JAPANESE,"ISO-2022-JP"},
#ifdef GBTOUNICODE
#ifdef KSCTOUNICODE
  {"ISO-2022-JP-2",CT_2022,CF_UNSUPRT,
     NIL,
     SC_LATIN_1 | SC_LATIN_2 | SC_LATIN_3 | SC_LATIN_4 | SC_LATIN_5 |
       SC_LATIN_6 | SC_LATIN_7 | SC_LATIN_8 | SC_LATIN_9 | SC_LATIN_10 |
	 SC_ARABIC | SC_CYRILLIC | SC_GREEK | SC_HEBREW | SC_THAI |
	   SC_VIETNAMESE | SC_CHINESE_SIMPLIFIED | SC_JAPANESE | SC_KOREAN
#ifdef CNS1TOUNICODE
	     | SC_CHINESE_TRADITIONAL
#endif
	       ,"UTF-8"},
#endif
#endif
#endif
#endif

#ifdef KSCTOUNICODE
  {"ISO-2022-KR",CT_2022,CF_PRIMARY | CF_DISPLAY | CF_UNSUPRT,
     NIL,SC_KOREAN,"EUC-KR"},
  {"EUC-KR",CT_DBYTE,CF_PRIMARY | CF_DISPLAY | CF_POSTING,
     (void *) &ksc_param,SC_KOREAN,NIL},
  {"KSC5601",CT_DBYTE,CF_PRIMARY | CF_DISPLAY,
     (void *) &ksc_param,SC_KOREAN,"EUC-KR"},
  {"KSC_5601",CT_DBYTE,CF_PRIMARY | CF_DISPLAY,
     (void *) &ksc_param,SC_KOREAN,"EUC-KR"},
  {"KS_C_5601-1987",CT_DBYTE,CF_DISPLAY,
     (void *) &ksc_param,SC_KOREAN,"EUC-KR"},
  {"KS_C_5601-1989",CT_DBYTE,CF_DISPLAY,
     (void *) &ksc_param,SC_KOREAN,"EUC-KR"},
  {"KS_C_5601-1992",CT_DBYTE,CF_DISPLAY,
     (void *) &ksc_param,SC_KOREAN,"EUC-KR"},
  {"KS_C_5601-1997",CT_DBYTE,CF_DISPLAY,
     (void *) &ksc_param,SC_KOREAN,"EUC-KR"},
#endif

				/* deep sigh */
  {"WINDOWS-874",CT_1BYTE,CF_PRIMARY | CF_DISPLAY,
     (void *) windows_874tab,SC_THAI,"ISO-8859-11"},
  {"CP874",CT_1BYTE,CF_DISPLAY,
     (void *) windows_874tab,SC_THAI,"ISO-8859-11"},
#ifdef GBTOUNICODE
  {"WINDOWS-936",CT_DBYTE,CF_PRIMARY | CF_DISPLAY,
     (void *) &gb_param,SC_CHINESE_SIMPLIFIED,"GBK"},
  {"CP936",CT_DBYTE,CF_DISPLAY,
     (void *) &gb_param,SC_CHINESE_SIMPLIFIED,"GBK"},
#endif
#ifdef KSCTOUNICODE
  {"WINDOWS-949",CT_DBYTE,CF_PRIMARY | CF_DISPLAY,
     (void *) &ksc_param,SC_KOREAN,"EUC-KR"},
  {"CP949",CT_DBYTE,CF_DISPLAY,
     (void *) &ksc_param,SC_KOREAN,"EUC-KR"},
  {"X-WINDOWS-949",CT_DBYTE,CF_PRIMARY | CF_DISPLAY,
     (void *) &ksc_param,SC_KOREAN,"EUC-KR"},
#endif
  {"WINDOWS-1250",CT_1BYTE,CF_PRIMARY | CF_DISPLAY,
     (void *) windows_1250tab,SC_LATIN_2,"ISO-8859-2"},
  {"CP1250",CT_1BYTE,CF_DISPLAY,
     (void *) windows_1250tab,SC_LATIN_2,"ISO-8859-2"},
  {"WINDOWS-1251",CT_1BYTE,CF_PRIMARY | CF_DISPLAY | CF_POSTING,
     (void *) windows_1251tab,SC_CYRILLIC,"KOI8-R"},
  {"CP1251",CT_1BYTE,CF_DISPLAY,
     (void *) windows_1251tab,SC_CYRILLIC,"KOI8-R"},
  {"WINDOWS-1252",CT_1BYTE,CF_PRIMARY | CF_DISPLAY,
     (void *) windows_1252tab,SC_LATIN_1,"ISO-8859-1"},
  {"CP1252",CT_1BYTE,CF_DISPLAY,
     (void *) windows_1252tab,SC_LATIN_1,"ISO-8859-1"},
  {"WINDOWS-1253",CT_1BYTE,CF_PRIMARY | CF_DISPLAY,
     (void *) windows_1253tab,SC_GREEK,"ISO-8859-7"},
  {"CP1253",CT_1BYTE,CF_DISPLAY,
     (void *) windows_1253tab,SC_GREEK,"ISO-8859-7"},
  {"WINDOWS-1254",CT_1BYTE,CF_PRIMARY | CF_DISPLAY,
     (void *) windows_1254tab,SC_LATIN_5,"ISO-8859-9"},
  {"CP1254",CT_1BYTE,CF_DISPLAY,
     (void *) windows_1254tab,SC_LATIN_5,"ISO-8859-9"},
  {"WINDOWS-1255",CT_1BYTE,CF_PRIMARY | CF_DISPLAY,
     (void *) windows_1255tab,SC_HEBREW,"ISO-8859-8"},
  {"CP1255",CT_1BYTE,CF_DISPLAY,
     (void *) windows_1255tab,SC_HEBREW,"ISO-8859-8"},
  {"WINDOWS-1256",CT_1BYTE,CF_PRIMARY | CF_DISPLAY,
     (void *) windows_1256tab,SC_ARABIC,"ISO-8859-6"},
  {"CP1256",CT_1BYTE,CF_DISPLAY,
     (void *) windows_1256tab,SC_ARABIC,"ISO-8859-6"},
  {"WINDOWS-1257",CT_1BYTE,CF_PRIMARY | CF_DISPLAY,
     (void *) windows_1257tab,SC_LATIN_7,"ISO-8859-13"},
  {"CP1257",CT_1BYTE,CF_DISPLAY,
     (void *) windows_1257tab,SC_LATIN_7,"ISO-8859-13"},
  {"WINDOWS-1258",CT_1BYTE,CF_PRIMARY | CF_DISPLAY,
     (void *) windows_1258tab,SC_VIETNAMESE,"VISCII"},
  {"CP1258",CT_1BYTE,CF_DISPLAY,
     (void *) windows_1258tab,SC_VIETNAMESE,"VISCII"},

				/* deeper sigh */
  {"IBM367",CT_ASCII,CF_PRIMARY | CF_DISPLAY,
     NIL,NIL,"US-ASCII"},
  {"IBM437",CT_1BYTE,CF_PRIMARY | CF_DISPLAY,
     (void *) ibm_437tab,SC_LATIN_1,"ISO-8859-1"},
  {"IBM737",CT_1BYTE,CF_PRIMARY | CF_DISPLAY,
     (void *) ibm_737tab,SC_GREEK,"ISO-8859-7"},
  {"IBM775",CT_1BYTE,CF_PRIMARY | CF_DISPLAY,
     (void *) ibm_775tab,SC_LATIN_7,"ISO-8859-13"},
  {"IBM850",CT_1BYTE,CF_PRIMARY | CF_DISPLAY,
     (void *) ibm_850tab,SC_LATIN_1,"ISO-8859-1"},
  {"IBM852",CT_1BYTE,CF_PRIMARY | CF_DISPLAY,
     (void *) ibm_852tab,SC_LATIN_2,"ISO-8859-2"},
  {"IBM855",CT_1BYTE,CF_PRIMARY | CF_DISPLAY,
     (void *) ibm_855tab,SC_CYRILLIC,"ISO-8859-5"},
  {"IBM857",CT_1BYTE,CF_PRIMARY | CF_DISPLAY,
     (void *) ibm_857tab,SC_LATIN_5,"ISO-8859-9"},
  {"IBM860",CT_1BYTE,CF_PRIMARY | CF_DISPLAY,
     (void *) ibm_860tab,SC_LATIN_1,"ISO-8859-1"},
  {"IBM861",CT_1BYTE,CF_PRIMARY | CF_DISPLAY,
     (void *) ibm_861tab,SC_LATIN_6,"ISO-8859-10"},
  {"IBM862",CT_1BYTE,CF_PRIMARY | CF_DISPLAY,
     (void *) ibm_862tab,SC_HEBREW,"ISO-8859-8"},
  {"IBM863",CT_1BYTE,CF_PRIMARY | CF_DISPLAY,
     (void *) ibm_863tab,SC_LATIN_1,"ISO-8859-1"},
  {"IBM864",CT_1BYTE,CF_PRIMARY | CF_DISPLAY,
     (void *) ibm_864tab,SC_ARABIC,"ISO-8859-6"},
  {"IBM865",CT_1BYTE,CF_PRIMARY | CF_DISPLAY,
     (void *) ibm_865tab,SC_LATIN_6,"ISO-8859-10"},
  {"IBM866",CT_1BYTE,CF_PRIMARY | CF_DISPLAY,
     (void *) ibm_866tab,SC_CYRILLIC,"KOI8-R"},
  {"IBM869",CT_1BYTE,CF_PRIMARY | CF_DISPLAY,
     (void *) ibm_869tab,SC_GREEK,"ISO-8859-7"},
  {"IBM874",CT_1BYTE,CF_PRIMARY | CF_DISPLAY,
     (void *) ibm_874tab,SC_THAI,"ISO-8859-11"},
				/* deepest sigh */
  {"ANSI_X3.4-1968",CT_ASCII,CF_DISPLAY,
     NIL,NIL,"US-ASCII"},
  {"UNICODE-1-1-UTF-7",CT_UTF7,CF_UNSUPRT,
     NIL,SC_UNICODE,"UTF-8"},
				/* these should never appear in email */
  {"UCS-2",CT_UCS2,CF_PRIMARY | CF_DISPLAY | CF_NOEMAIL,
     NIL,SC_UNICODE,"UTF-8"},
  {"UCS-4",CT_UCS4,CF_PRIMARY | CF_DISPLAY | CF_NOEMAIL,
     NIL,SC_UNICODE,"UTF-8"},
  {"UTF-16",CT_UTF16,CF_PRIMARY | CF_DISPLAY | CF_NOEMAIL,
     NIL,SC_UNICODE,"UTF-8"},
  NIL
};

/* Non-Unicode Script table */

static const SCRIPT utf8_scvalid[] = {
  {"Arabic",NIL,SC_ARABIC},
  {"Chinese Simplified","China, Singapore",SC_CHINESE_SIMPLIFIED},
  {"Chinese Traditional","Taiwan, Hong Kong, Macao",SC_CHINESE_TRADITIONAL},
  {"Cyrillic",NIL,SC_CYRILLIC},
  {"Cyrillic Ukranian",NIL,SC_UKRANIAN},
  {"Greek",NIL,SC_GREEK},
  {"Hebrew",NIL,SC_HEBREW},
  {"Japanese",NIL,SC_JAPANESE},
  {"Korean",NIL,SC_KOREAN},
  {"Latin-1","Western Europe",SC_LATIN_1},
  {"Latin-2","Eastern Europe",SC_LATIN_2},
  {"Latin-3","Southern Europe",SC_LATIN_3},
  {"Latin-4","Northern Europe",SC_LATIN_4},
  {"Latin-5","Turkish",SC_LATIN_5},
  {"Latin-6","Nordic",SC_LATIN_6},
  {"Latin-7","Baltic",SC_LATIN_7},
  {"Latin-8","Celtic",SC_LATIN_8},
  {"Latin-9","Euro",SC_LATIN_9},
  {"Latin-10","Balkan",SC_LATIN_10},
  {"Thai",NIL,SC_THAI},
  {"Vietnamese",NIL,SC_VIETNAMESE},
  NIL
};

/* Look up script name or return entire table
 * Accepts: script name or NIL
 * Returns: pointer to script table entry or NIL if unknown
 */

SCRIPT *utf8_script (char *script)
{
  unsigned long i;
  if (!script) return (SCRIPT *) &utf8_scvalid[0];
  else if (*script && (strlen (script) < 128))
    for (i = 0; utf8_scvalid[i].name; i++)
      if (!compare_cstring (script,utf8_scvalid[i].name))
	return (SCRIPT *) &utf8_scvalid[i];
  return NIL;			/* failed */
}


/* Look up charset name or return entire table
 * Accepts: charset name or NIL
 * Returns: charset table entry or NIL if unknown
 */

const CHARSET *utf8_charset (char *charset)
{
  unsigned long i;
  if (!charset) return (CHARSET *) &utf8_csvalid[0];
  else if (*charset && (strlen (charset) < 128))
    for (i = 0; utf8_csvalid[i].name; i++)
      if (!compare_cstring (charset,utf8_csvalid[i].name))
	return (CHARSET *) &utf8_csvalid[i];
  return NIL;			/* failed */
}

/* Validate charset and generate error message if invalid
 * Accepts: bad character set
 * Returns: NIL if good charset, else error message string
 */

#define BADCSS "[BADCHARSET ("
#define BADCSE ")] Unknown charset: "

char *utf8_badcharset (char *charset)
{
  char *msg = NIL;
  if (!utf8_charset (charset)) {
    char *s,*t;
    unsigned long i,j;
				/* calculate size of header, trailer, and bad
				 * charset plus charset names */
    for (i = 0, j = sizeof (BADCSS) + sizeof (BADCSE) + strlen (charset) - 2;
	 utf8_csvalid[i].name; i++)
      j += strlen (utf8_csvalid[i].name) + 1;
				/* not built right */
    if (!i) fatal ("No valid charsets!");
				/* header */
    for (s = msg = (char *) fs_get (j), t = BADCSS; *t; *s++ = *t++);
				/* each charset */
    for (i = 0; utf8_csvalid[i].name; *s++ = ' ', i++)
      for (t = utf8_csvalid[i].name; *t; *s++ = *t++);
				/* back over last space, trailer */
    for (t = BADCSE, --s; *t; *s++ = *t++);
				/* finally bogus charset */
    for (t = charset; *t; *s++ = *t++);
    *s++ = '\0';		/* finally tie off string */
    if (s != (msg + j)) fatal ("charset msg botch");
  }
  return msg;
}

/* Convert charset labelled sized text to UTF-8
 * Accepts: source sized text
 *	    charset
 *	    pointer to returned sized text if non-NIL
 *	    flags
 * Returns: T if successful, NIL if failure
 */

long utf8_text (SIZEDTEXT *text,char *charset,SIZEDTEXT *ret,long flags)
{
  ucs4cn_t cv = (flags & U8T_CASECANON) ? ucs4_titlecase : NIL;
  ucs4de_t de = (flags & U8T_DECOMPOSE) ? ucs4_decompose_recursive : NIL;
  const CHARSET *cs = (charset && *charset) ?
    utf8_charset (charset) : utf8_infercharset (text);
  if (cs) return (text && ret) ? utf8_text_cs (text,cs,ret,cv,de) : LONGT;
  if (ret) {			/* no conversion possible */
    ret->data = text->data;	/* so return source */
    ret->size = text->size;
  }
  return NIL;			/* failure */
}


/* Operations used in converting data */

#define UTF8_COUNT_BMP(count,c,cv,de) {		\
  void *more = NIL;				\
  if (cv) c = (*cv) (c);			\
  if (de) c = (*de) (c,&more);			\
  do count += UTF8_SIZE_BMP(c);			\
  while (more && (c = (*de) (U8G_ERROR,&more)));\
}

#define UTF8_WRITE_BMP(b,c,cv,de) {		\
  void *more = NIL;				\
  if (cv) c = (*cv) (c);			\
  if (de) c = (*de) (c,&more);			\
  do UTF8_PUT_BMP (b,c)				\
  while (more && (c = (*de) (U8G_ERROR,&more)));\
}

#define UTF8_COUNT(count,c,cv,de) {		\
  void *more = NIL;				\
  if (cv) c = (*cv) (c);			\
  if (de) c = (*de) (c,&more);			\
  do count += utf8_size (c);			\
  while (more && (c = (*de) (U8G_ERROR,&more)));\
}

#define UTF8_WRITE(b,c,cv,de) {			\
  void *more = NIL;				\
  if (cv) c = (*cv) (c);			\
  if (de) c = (*de) (c,&more);			\
  do b = utf8_put (b,c);			\
  while (more && (c = (*de) (U8G_ERROR,&more)));\
}

/* Convert sized text to UTF-8 given CHARSET block
 * Accepts: source sized text
 *	    CHARSET block
 *	    pointer to returned sized text 
 *	    canonicalization function
 *	    decomposition function
 * Returns: T if successful, NIL if failure
 */

long utf8_text_cs (SIZEDTEXT *text,const CHARSET *cs,SIZEDTEXT *ret,
		   ucs4cn_t cv,ucs4de_t de)
{
  ret->data = text->data;	/* default to source */
  ret->size = text->size;
  switch (cs->type) {		/* convert if type known */
  case CT_ASCII:		/* 7-bit ASCII no table */
  case CT_UTF8:			/* variable UTF-8 encoded Unicode no table */
    if (cv || de) utf8_text_utf8 (text,ret,cv,de);
    break;
  case CT_1BYTE0:		/* 1 byte no table */
    utf8_text_1byte0 (text,ret,cv,de);
    break;
  case CT_1BYTE:		/* 1 byte ASCII + table 0x80-0xff */
    utf8_text_1byte (text,ret,cs->tab,cv,de);
    break;
  case CT_1BYTE8:		/* 1 byte table 0x00 - 0xff */
    utf8_text_1byte8 (text,ret,cs->tab,cv,de);
    break;
  case CT_EUC:			/* 2 byte ASCII + utf8_eucparam base/CS2/CS3 */
    utf8_text_euc (text,ret,cs->tab,cv,de);
    break;
  case CT_DBYTE:		/* 2 byte ASCII + utf8_eucparam */
    utf8_text_dbyte (text,ret,cs->tab,cv,de);
    break;
  case CT_DBYTE2:		/* 2 byte ASCII + utf8_eucparam plane1/2 */
    utf8_text_dbyte2 (text,ret,cs->tab,cv,de);
    break;
  case CT_UTF7:			/* variable UTF-7 encoded Unicode no table */
    utf8_text_utf7 (text,ret,cv,de);
    break;
  case CT_UCS2:			/* 2 byte 16-bit Unicode no table */
    utf8_text_ucs2 (text,ret,cv,de);
    break;
  case CT_UCS4:			/* 4 byte 32-bit Unicode no table */
    utf8_text_ucs4 (text,ret,cv,de);
    break;
  case CT_UTF16:		/* variable UTF-16 encoded Unicode no table */
    utf8_text_utf16 (text,ret,cv,de);
    break;
  case CT_2022:			/* variable ISO-2022 encoded no table*/
    utf8_text_2022 (text,ret,cv,de);
    break;
  case CT_SJIS:			/* 2 byte Shift-JIS encoded JIS no table */
    utf8_text_sjis (text,ret,cv,de);
    break;
  default:			/* unknown character set type */
    return NIL;
  }
  return LONGT;			/* return success */
}

/* Reverse mapping routines
 *
 * These routines only support character sets, not all possible charsets.  In
 * particular, they do not support any Unicode encodings or ISO 2022.
 *
 * As a special dispensation, utf8_cstext() and utf8_cstocstext() support
 * support ISO-2022-JP if EUC-JP can be reverse mapped; and utf8_rmaptext()
 * will generated ISO-2022-JP using an EUC-JP rmap if flagged to do so.
 *
 * No attempt is made to map "equivalent" Unicode characters or Unicode
 * characters that have the same glyph; nor is there any attempt to handle
 * combining characters or otherwise do any stringprep.  Maybe later.
 */


/* Convert UTF-8 sized text to charset
 * Accepts: source sized text
 *	    destination charset
 *	    pointer to returned sized text
 *	    substitute character if not in cs, else NIL to return failure
 * Returns: T if successful, NIL if failure
 */


long utf8_cstext (SIZEDTEXT *text,char *charset,SIZEDTEXT *ret,
		  unsigned long errch)
{
  short iso2022jp = !compare_cstring (charset,"ISO-2022-JP");
  unsigned short *rmap = utf8_rmap (iso2022jp ? "EUC-JP" : charset);
  return rmap ? utf8_rmaptext (text,rmap,ret,errch,iso2022jp) : NIL;
}

/* Convert charset labelled sized text to another charset
 * Accepts: source sized text
 *	    source charset
 *	    pointer to returned sized text
 *	    destination charset
 *	    substitute character if not in dest cs, else NIL to return failure
 * Returns: T if successful, NIL if failure
 *
 * This routine has the same restricts as utf8_cstext().
 */

long utf8_cstocstext (SIZEDTEXT *src,char *sc,SIZEDTEXT *dst,char *dc,
		      unsigned long errch)
{
  SIZEDTEXT utf8;
  const CHARSET *scs,*dcs;
  unsigned short *rmap;
  long ret = NIL;
  long iso2022jp;
				/* lookup charsets and reverse map */
  if ((dc && (dcs = utf8_charset (dc))) &&
      (rmap = (iso2022jp = ((dcs->type == CT_2022) &&
			    !compare_cstring (dcs->name,"ISO-2022-JP"))) ?
       utf8_rmap ("EUC-JP") : utf8_rmap_cs (dcs)) &&
      (scs = (sc && *sc) ? utf8_charset (sc) : utf8_infercharset (src))) {
				/* init temporary buffer */
    memset (&utf8,NIL,sizeof (SIZEDTEXT));
				/* source cs equivalent to dest cs? */
    if ((scs->type == dcs->type) && (scs->tab == dcs->tab)) {
      dst->data = src->data;	/* yes, just copy pointers */
      dst->size = src->size;
      ret = LONGT;
    }
				/* otherwise do the conversion */
    else ret = (utf8_text_cs (src,scs,&utf8,NIL,NIL) &&
		utf8_rmaptext (&utf8,rmap,dst,errch,iso2022jp));
				/* flush temporary buffer */
    if (utf8.data && (utf8.data != src->data) && (utf8.data != dst->data))
      fs_give ((void **) &utf8.data);
  }
  return ret;
}

/* Cached rmap */

static const CHARSET *currmapcs = NIL;
static unsigned short *currmap = NIL;


/* Cache and return map for UTF-8 -> character set
 * Accepts: character set name
 * Returns: cached map if character set found, else NIL
 */

unsigned short *utf8_rmap (char *charset)
{
  return (currmapcs && !compare_cstring (charset,currmapcs->name)) ? currmap :
    utf8_rmap_cs (utf8_charset (charset));
}


/* Cache and return map for UTF-8 -> character set given CHARSET block
 * Accepts: CHARSET block
 * Returns: cached map if character set found, else NIL
 */

unsigned short *utf8_rmap_cs (const CHARSET *cs)
{
  unsigned short *ret = NIL;
  if (!cs);			/* have charset? */
  else if (cs == currmapcs) ret = currmap;
  else if ((ret = utf8_rmap_gen (cs,currmap)) != NULL) {
    currmapcs = cs;
    currmap = ret;
  }
  return ret;
}

/* Return map for UTF-8 -> character set given CHARSET block
 * Accepts: CHARSET block
 *	    old map to recycle
 * Returns: map if character set found, else NIL
 */

unsigned short *utf8_rmap_gen (const CHARSET *cs,unsigned short *oldmap)
{
  unsigned short u,*tab,*rmap;
  unsigned int i,m,ku,ten;
  struct utf8_eucparam *param,*p2;
  switch (cs->type) {		/* is a character set? */
  case CT_ASCII:		/* 7-bit ASCII no table */
  case CT_1BYTE0:		/* 1 byte no table */
  case CT_1BYTE:		/* 1 byte ASCII + table 0x80-0xff */
  case CT_1BYTE8:		/* 1 byte table 0x00 - 0xff */
  case CT_EUC:			/* 2 byte ASCII + utf8_eucparam base/CS2/CS3 */
  case CT_DBYTE:		/* 2 byte ASCII + utf8_eucparam */
  case CT_DBYTE2:		/* 2 byte ASCII + utf8_eucparam plane1/2 */
  case CT_SJIS:			/* 2 byte Shift-JIS */
    rmap = oldmap ? oldmap :	/* recycle old map if supplied else make new */
      (unsigned short *) fs_get (65536 * sizeof (unsigned short));
				/* initialize table for ASCII */
    for (i = 0; i < 128; i++) rmap[i] = (unsigned short) i;
				/* populate remainder of table with NOCHAR */
#define NOCHARBYTE (NOCHAR & 0xff)
#if NOCHAR - ((NOCHARBYTE << 8) | NOCHARBYTE)
    while (i < 65536) rmap[i++] = NOCHAR;
#else
    memset (rmap + 128,NOCHARBYTE,(65536 - 128) * sizeof (unsigned short));
#endif
    break;
  default:			/* unsupported charset type */
    rmap = NIL;			/* no map possible */
  }
  if (rmap) {			/* have a map? */
    switch (cs->type) {		/* additional reverse map actions */
    case CT_1BYTE0:		/* 1 byte no table */
      for (i = 128; i < 256; i++) rmap[i] = (unsigned short) i;
      break;
    case CT_1BYTE:		/* 1 byte ASCII + table 0x80-0xff */
      for (tab = (unsigned short *) cs->tab,i = 128; i < 256; i++)
	if (tab[i & BITS7] != UBOGON) rmap[tab[i & BITS7]] = (unsigned short)i;
      break;
    case CT_1BYTE8:		/* 1 byte table 0x00 - 0xff */
      for (tab = (unsigned short *) cs->tab,i = 0; i < 256; i++)
	if (tab[i] != UBOGON) rmap[tab[i]] = (unsigned short) i;
      break;
    case CT_EUC:		/* 2 byte ASCII + utf8_eucparam base/CS2/CS3 */
      for (param = (struct utf8_eucparam *) cs->tab,
	     tab = (unsigned short *) param->tab, ku = 0;
	   ku < param->max_ku; ku++)
	for (ten = 0; ten < param->max_ten; ten++)
	  if ((u = tab[(ku * param->max_ten) + ten]) != UBOGON)
	    rmap[u] = ((ku + param->base_ku) << 8) +
	      (ten + param->base_ten) + 0x8080;
      break;

    case CT_DBYTE:		/* 2 byte ASCII + utf8_eucparam */
      for (param = (struct utf8_eucparam *) cs->tab,
	     tab = (unsigned short *) param->tab, ku = 0;
	   ku < param->max_ku; ku++)
	for (ten = 0; ten < param->max_ten; ten++)
	  if ((u = tab[(ku * param->max_ten) + ten]) != UBOGON)
	    rmap[u] = ((ku + param->base_ku) << 8) + (ten + param->base_ten);
      break;
    case CT_DBYTE2:		/* 2 byte ASCII + utf8_eucparam plane1/2 */
      param = (struct utf8_eucparam *) cs->tab;
      p2 = param + 1;		/* plane 2 parameters */
				/* only ten parameters should differ */
      if ((param->base_ku != p2->base_ku) || (param->max_ku != p2->max_ku))
	fatal ("ku definition error for CT_DBYTE2 charset");
				/* total codepoints in each ku */
      m = param->max_ten + p2->max_ten;
      tab = (unsigned short *) param->tab;
      for (ku = 0; ku < param->max_ku; ku++) {
	for (ten = 0; ten < param->max_ten; ten++)
	  if ((u = tab[(ku * m) + ten]) != UBOGON)
	    rmap[u] = ((ku + param->base_ku) << 8) + (ten + param->base_ten);
	for (ten = 0; ten < p2->max_ten; ten++)
	  if ((u = tab[(ku * m) + param->max_ten + ten]) != UBOGON)
	    rmap[u] = ((ku + param->base_ku) << 8) + (ten + p2->base_ten);
      }
      break;
    case CT_SJIS:		/* 2 byte Shift-JIS */
      for (ku = 0; ku < MAX_JIS0208_KU; ku++)
	for (ten = 0; ten < MAX_JIS0208_TEN; ten++)
	  if ((u = jis0208tab[ku][ten]) != UBOGON) {
	    int sku = ku + BASE_JIS0208_KU;
	    int sten = ten + BASE_JIS0208_TEN;
	    rmap[u] = ((((sku + 1) >> 1) + ((sku < 95) ? 112 : 176)) << 8) +
	      sten + ((sku % 2) ? ((sten > 95) ? 32 : 31) : 126);
	  }
				/* JIS Roman */
      rmap[UCS2_YEN] = JISROMAN_YEN;
      rmap[UCS2_OVERLINE] = JISROMAN_OVERLINE;
				/* JIS hankaku katakana */
      for (u = 0; u < (MAX_KANA_8 - MIN_KANA_8); u++)
	rmap[UCS2_KATAKANA + u] = MIN_KANA_8 + u;
      break;
    }
				/* hack: map NBSP to SP if otherwise no map */
    if (rmap[0x00a0] == NOCHAR) rmap[0x00a0] = rmap[0x0020];
  }
  return rmap;			/* return map */
}

/* Convert UTF-8 sized text to charset using rmap
 * Accepts: source sized text
 *	    conversion rmap
 *	    pointer to returned sized text
 *	    substitute character if not in rmap, else NIL to return failure
 *	    ISO-2022-JP conversion flag
 * Returns T if successful, NIL if failure
 *
 * This routine doesn't try to convert to all possible charsets; in particular
 * it doesn't support other Unicode encodings or any ISO 2022 other than
 * ISO-2022-JP.
 */

long utf8_rmaptext (SIZEDTEXT *text,unsigned short *rmap,SIZEDTEXT *ret,
		    unsigned long errch,long iso2022jp)
{
  unsigned long i,u,c;
				/* get size of buffer */
  if ((i = utf8_rmapsize (text,rmap,errch,iso2022jp)) != 0L) {
    unsigned char *s = text->data;
    unsigned char *t = ret->data = (unsigned char *) fs_get (i);
    ret->size = i - 1;		/* number of octets in destination buffer */
				/* start non-zero ISO-2022-JP state at 1 */
    if (iso2022jp) iso2022jp = 1;
				/* convert string, ignore BOM */
    for (i = text->size; i;) if ((u = utf8_get (&s,&i)) != UCS2_BOM) {
				/* substitute error character for NOCHAR */
      if ((u & U8GM_NONBMP) || ((c = rmap[u]) == NOCHAR)) c = errch;
      switch (iso2022jp) {	/* depends upon ISO 2022 mode */
      case 0:			/* ISO 2022 not in effect */
				/* two-byte character */
	if (c > 0xff) *t++ = (unsigned char) (c >> 8);
				/* single-byte or low-byte of two-byte */
	*t++ = (unsigned char) (c & 0xff);
	break;
      case 1:			/* ISO 2022 Roman */
				/* <ch> */
	if (c < 0x80) *t++ = (unsigned char) c;
	else {			/* JIS character */
	  *t++ = I2C_ESC;	/* ESC $ B <hi> <lo> */
	  *t++ = I2C_MULTI;
	  *t++ = I2CS_94x94_JIS_NEW;
	  *t++ = (unsigned char) (c >> 8) & 0x7f;
	  *t++ = (unsigned char) c & 0x7f;
	  iso2022jp = 2;	/* shift to ISO 2022 JIS */
	}
	break;
      case 2:			/* ISO 2022 JIS */
	if (c > 0x7f) {		/* <hi> <lo> */
	  *t++ = (unsigned char) (c >> 8) & 0x7f;
	  *t++ = (unsigned char) c & 0x7f;
	}
	else {			/* ASCII character */
	  *t++ = I2C_ESC;	/* ESC ( J <ch> */
	  *t++ = I2C_G0_94;
	  *t++ = I2CS_94_JIS_ROMAN;
	  *t++ = (unsigned char) c;
	  iso2022jp = 1;	/* shift to ISO 2022 Roman */
	}
	break;
      }
    }
    if (iso2022jp == 2) {	/* ISO-2022-JP string must end in Roman */
      *t++ = I2C_ESC;		/* ESC ( J */
      *t++ = I2C_G0_94;
      *t++ = I2CS_94_JIS_ROMAN;
    }
    *t++ = NIL;			/* tie off returned data */
    return LONGT;		/* return success */
  }
  ret->data = NIL;
  ret->size = 0;
  return NIL;			/* failure */
}

/* Calculate size of convertsion of UTF-8 sized text to charset using rmap
 * Accepts: source sized text
 *	    conversion rmap
 *	    pointer to returned sized text
 *	    substitute character if not in rmap, else NIL to return failure
 *	    ISO-2022-JP conversion flag
 * Returns size+1 if successful, NIL if failure
 *
 * This routine doesn't try to handle to all possible charsets; in particular
 * it doesn't support other Unicode encodings or any ISO 2022 other than
 * ISO-2022-JP.
 */

unsigned long utf8_rmapsize (SIZEDTEXT *text,unsigned short *rmap,
			     unsigned long errch,long iso2022jp)
{
  unsigned long i,u,c;
  unsigned long ret = 1;	/* terminating NUL */
  unsigned char *s = text->data;
  if (iso2022jp) iso2022jp = 1;	/* start non-zero ISO-2022-JP state at 1 */
  for (i = text->size; i;) if ((u = utf8_get (&s,&i)) != UCS2_BOM) {
    if ((u & U8GM_NONBMP) || (((c = rmap[u]) == NOCHAR) && !(c = errch)))
      return NIL;		/* not in BMP, or NOCHAR and no err char */
    switch (iso2022jp) {	/* depends upon ISO 2022 mode */
    case 0:			/* ISO 2022 not in effect */
      ret += (c > 0xff) ? 2 : 1;
      break;
    case 1:			/* ISO 2022 Roman */
      if (c < 0x80) ret += 1;	/* <ch> */
      else {			/* JIS character */
	ret += 5;		/* ESC $ B <hi> <lo> */
	iso2022jp = 2;		/* shift to ISO 2022 JIS */
      }
      break;
    case 2:			/* ISO 2022 JIS */
      if (c > 0x7f) ret += 2;	/* <hi> <lo> */
      else {			/* ASCII character */
	ret += 4;		/* ESC ( J <ch> */
	iso2022jp = 1;		/* shift to ISO 2022 Roman */
      }
      break;
    }
  }
  if (iso2022jp == 2) {		/* ISO-2022-JP string must end in Roman */
    ret += 3;			/* ESC ( J */
    iso2022jp = 1;		/* reset state to Roman */
  }
  return ret;
}

/* Convert UCS-4 to charset using rmap
 * Accepts: source UCS-4 character(s)
 *	    numver of UCS-4 characters
 *	    conversion rmap
 *	    pointer to returned sized text
 *	    substitute character if not in rmap, else NIL to return failure
 * Returns T if successful, NIL if failure
 *
 * Currently only supports BMP characters, and does not support ISO-2022
 */

long ucs4_rmaptext (unsigned long *ucs4,unsigned long len,unsigned short *rmap,
		    SIZEDTEXT *ret,unsigned long errch)
{
  long size = ucs4_rmaplen (ucs4,len,rmap,errch);
  return (size >= 0) ?		/* build in newly-created buffer */
    ucs4_rmapbuf (ret->data = (unsigned char *) fs_get ((ret->size = size) +1),
		  ucs4,len,rmap,errch) : NIL;
}

/* Return size of UCS-4 string converted to other CS via rmap
 * Accepts: source UCS-4 character(s)
 *	    numver of UCS-4 characters
 *	    conversion rmap
 *	    substitute character if not in rmap, else NIL to return failure
 * Returns: length if success, negative if failure (no-convert)
 */

long ucs4_rmaplen (unsigned long *ucs4,unsigned long len,unsigned short *rmap,
		   unsigned long errch)
{
  long ret;
  unsigned long i,u,c;
				/* count non-BOM characters */
  for (ret = 0,i = 0; i < len; ++i) if ((u = ucs4[i]) != UCS2_BOM) {
    if ((u & U8GM_NONBMP) || (((c = rmap[u]) == NOCHAR) && !(c = errch)))
      return -1;		/* not in BMP, or NOCHAR and no err char? */
    ret += (c > 0xff) ? 2 : 1;
  }
  return ret;
}


/* Stuff buffer with UCS-4 string converted to other CS via rmap
 * Accepts: destination buffer
 *	    source UCS-4 character(s)
 *	    number of UCS-4 characters
 *	    conversion rmap
 *	    substitute character if not in rmap, else NIL to return failure
 * Returns: T, always
 */

long ucs4_rmapbuf (unsigned char *t,unsigned long *ucs4,unsigned long len,
		   unsigned short *rmap,unsigned long errch)
{
  unsigned long i,u,c;
				/* convert non-BOM characters */
  for (i = 0; i < len; ++i) if ((u = ucs4[i]) != UCS2_BOM) {
				/* substitute error character for NOCHAR */
    if ((u & U8GM_NONBMP) || ((c = rmap[u]) == NOCHAR)) c = errch;
				/* two-byte character? */
    if (c > 0xff) *t++ = (unsigned char) (c >> 8);
				/* single-byte or low-byte of two-byte */
    *t++ = (unsigned char) (c & 0xff);
  }
  *t++ = NIL;			/* tie off returned data */
  return LONGT;
}

/* Return UCS-4 Unicode character from UTF-8 string
 * Accepts: pointer to string
 *	    remaining octets in string
 * Returns: UCS-4 character with pointer and count updated
 *	    or error code with pointer and count unchanged
 */

unsigned long utf8_get (unsigned char **s,unsigned long *i)
{
  unsigned char *t = *s;
  unsigned long j = *i;
				/* decode raw UTF-8 string */
  unsigned long ret = utf8_get_raw (&t,&j);
  if (ret & U8G_ERROR);		/* invalid raw UTF-8 decoding? */
				/* no, is it surrogate? */
  else if ((ret >= UTF16_SURR) && (ret <= UTF16_MAXSURR)) ret = U8G_SURROGA;
				/* or in non-Unicode ISO 10646 space? */
  else if (ret > UCS4_MAXUNICODE) ret = U8G_NOTUNIC;
  else {
    *s = t;			/* all is well, update pointer */
    *i = j;			/* and counter */
  }
  return ret;			/* return value */
}

/* Return raw (including non-Unicode) UCS-4 character from UTF-8 string
 * Accepts: pointer to string
 *	    remaining octets in string
 * Returns: UCS-4 character with pointer and count updated
 *	    or error code with pointer and count unchanged
 */

unsigned long utf8_get_raw (unsigned char **s,unsigned long *i)
{
  unsigned char c,c1;
  unsigned char *t = *s;
  unsigned long j = *i;
  unsigned long ret = U8G_NOTUTF8;
  int more = 0;
  do {				/* make sure have source octets available */
    if (!j--) return more ? U8G_ENDSTRI : U8G_ENDSTRG;
				/* UTF-8 continuation? */
    else if (((c = *t++) > 0x7f) && (c < 0xc0)) {
				/* continuation when not in progress */
      if (!more) return U8G_BADCONT;
      --more;			/* found a continuation octet */
      ret <<= 6;		/* shift current value by 6 bits */
      ret |= c & 0x3f;		/* merge continuation octet */
    }
				/* incomplete UTF-8 character */
    else if (more) return U8G_INCMPLT;
    else {			/* start of sequence */
      c1 = j ? *t : 0xbf;	/* assume valid continuation if incomplete */
      if (c < 0x80) ret = c;	/* U+0000 - U+007f */
      else if (c < 0xc2);	/* c0 and c1 never valid */
      else if (c < 0xe0) {	/* U+0080 - U+07ff */
	c &= 0x1f;
	if (c1 >= 0x80) more = 1;
      }
      else if (c == 0xe0) {	/* U+0800 - U+0fff */
	c &= 0x0f;
	if (c1 >= 0xa0) more = 2;
      }
      else if (c < 0xed) {	/* U+1000 - U+cfff */
	c &= 0x0f;
	if (c1 >= 0x80) more = 2;
      }
      else if (c == 0xed) {	/* U+d000 - U+d7ff */
	c &= 0x0f;
	if (j == 0 || ((c1 >= 0x80) && (c1 <= 0x9f))) more = 2;
      }
      else if (c < 0xf0) {	/* U+e000 - U+ffff */
	c &= 0x0f;
	if (c1 >= 0x80) more = 2;
      }
      else if (c == 0xf0) {	/* U+10000 - U+3ffff */
	c &= 0x07;
	if (c1 >= 0x90) more = 3;
      }
      else if (c < 0xf3) {	/* U+40000 - U+fffff */
	c &= 0x07;
	if (c1 >= 0x80) more = 3;
      }
#if 0
      else if (c == 0xf4) {	/* U+100000 - U+10ffff */
	c &= 0x07;
	if (((c1 >= 0x80) && (c1 <= 0x8f))) more = 3;
      }
#else
      else if (c < 0xf8) {	/* U+100000 - U+10ffff (and 110000 - 1fffff) */
	c &= 0x07;
	if ((c1 >= 0x80)) more = 3;
      }
      else if (c < 0xfc) {	/* ISO 10646 200000 - 3ffffff */
	c &= 0x03;
	if ((c1 >= 0x80)) more = 4;
      }
      else if (c < 0xfe) {	/* ISO 10646 4000000 - 7fffffff */
	c &= 0x01;
	if ((c1 >= 0x80)) more = 5;
      }
#endif
				/* fe and ff never valid */
      if (more) {		/* multi-octet, make sure more to come */
	if (!j) return U8G_ENDSTRI;
	ret = c;		/* continuation needed, save start bits */
      }
    }
  } while (more);
  if (!(ret & U8G_ERROR)) {	/* success return? */
    *s = t;			/* yes, update pointer */
    *i = j;			/* and counter */
  }
  return ret;			/* return value */
}

/* Return UCS-4 character from named charset string
 * Accepts: charset
 *	    pointer to string
 *	    remaining octets in string
 * Returns: UCS-4 character with pointer and count updated, negative if error
 *
 * Error codes are the same as utf8_get().
 */

unsigned long ucs4_cs_get (CHARSET *cs,unsigned char **s,unsigned long *i)
{
  unsigned char c,c1,ku,ten;
  unsigned long ret,d;
  unsigned char *t = *s;
  unsigned long j = *i;
  struct utf8_eucparam *p1,*p2,*p3;
  if (j--) c = *t++;		/* get first octet */
  else return U8G_ENDSTRG;	/* empty string */
  switch (cs->type) {		/* convert if type known */
  case CT_UTF8:			/* variable UTF-8 encoded Unicode no table */
    return utf8_get (s,i);
  case CT_ASCII:		/* 7-bit ASCII no table */
    if (c >= 0x80) return U8G_NOTUTF8;
  case CT_1BYTE0:		/* 1 byte no table */
    ret = c;			/* identity */
    break;
  case CT_1BYTE:		/* 1 byte ASCII + table 0x80-0xff */
    ret = (c > 0x80) ? ((unsigned short *) cs->tab)[c & BITS7] : c;
    break;
  case CT_1BYTE8:		/* 1 byte table 0x00 - 0xff */
    ret = ((unsigned short *) cs->tab)[c];
    break;

  case CT_EUC:			/* 2 byte ASCII + utf8_eucparam base/CS2/CS3 */
    if (c & BIT8) {
      p1 = (struct utf8_eucparam *) cs->tab;
      p2 = p1 + 1;
      p3 = p1 + 2;
      if (j--) c1 = *t++;	/* get second octet */
      else return U8G_ENDSTRI;
      if (!(c1 & BIT8)) return U8G_NOTUTF8;
      switch (c) {		/* check 8bit code set */
      case EUC_CS2:		/* CS2 */
	if (p2->base_ku) {	/* CS2 set up? */
	  if (p2->base_ten) {	/* yes, multibyte? */
	    if (j--) c = *t++;	/* get second octet */
	    else return U8G_ENDSTRI;
	    if ((c & BIT8) &&
		((ku = (c1 & BITS7) - p2->base_ku) < p2->max_ku) &&
		((ten = (c & BITS7) - p2->base_ten) < p2->max_ten)) {
	      ret = ((unsigned short *) p2->tab)[(ku*p2->max_ten) + ten];
	      break;
	    }
	  }
	  else if ((c1 >= p2->base_ku) && (c1 < p2->max_ku)) {
	    ret = c1 + ((unsigned long) p2->tab);
	    break;
	  }
	}
	return U8G_NOTUTF8;	/* CS2 not set up or bogus */
      case EUC_CS3:		/* CS3 */
	if (p3->base_ku) {	/* CS3 set up? */
	  if (p3->base_ten) {	/* yes, multibyte? */
	    if (j--) c = *t++;	/* get second octet */
	    else return U8G_ENDSTRI;
	    if ((c & BIT8) &&
		((ku = (c1 & BITS7) - p3->base_ku) < p3->max_ku) &&
		((ten = (c & BITS7) - p3->base_ten) < p3->max_ten)) {
	      ret = ((unsigned short *) p3->tab)[(ku*p3->max_ten) + ten];
	      break;
	    }
	  }
	  else if ((c1 >= p3->base_ku) && (c1 < p3->max_ku)) {
	    ret = c1 + ((unsigned long) p3->tab);
	    break;
	  }
	}
	return U8G_NOTUTF8;	/* CS3 not set up or bogus */
      default:
	if (((ku = (c & BITS7) - p1->base_ku) >= p1->max_ku) ||
	    ((ten = (c1 & BITS7) - p1->base_ten) >= p1->max_ten))
	  return U8G_NOTUTF8;
	ret = ((unsigned short *) p1->tab)[(ku*p1->max_ten) + ten];
		/* special hack for JIS X 0212: merge rows less than 10 */
	if ((ret == UBOGON) && ku && (ku < 10) && p3->tab && p3->base_ten)
	  ret = ((unsigned short *) p3->tab)
	    [((ku - (p3->base_ku - p1->base_ku))*p3->max_ten) + ten];
	break;
      }
    }
    else ret = c;		/* ASCII character */
    break;

  case CT_DBYTE:		/* 2 byte ASCII + utf8_eucparam */
    if (c & BIT8) {		/* double-byte character? */
      p1 = (struct utf8_eucparam *) cs->tab;
      if (j--) c1 = *t++;	/* get second octet */
      else return U8G_ENDSTRI;
      if (((ku = c - p1->base_ku) < p1->max_ku) &&
	  ((ten = c1 - p1->base_ten) < p1->max_ten))
	ret = ((unsigned short *) p1->tab)[(ku*p1->max_ten) + ten];
      else return U8G_NOTUTF8;
    }
    else ret = c;		/* ASCII character */
    break;
  case CT_DBYTE2:		/* 2 byte ASCII + utf8_eucparam plane1/2 */
    if (c & BIT8) {		/* double-byte character? */
      p1 = (struct utf8_eucparam *) cs->tab;
      p2 = p1 + 1;
      if (j--) c1 = *t++;	/* get second octet */
      else return U8G_ENDSTRI;
      if (c1 & BIT8) {		/* high vs. low plane */
	if ((ku = c - p2->base_ku) < p2->max_ku &&
	    ((ten = c1 - p2->base_ten) < p2->max_ten))
	  ret = ((unsigned short *) p1->tab)
	    [(ku*(p1->max_ten + p2->max_ten)) + p1->max_ten + ten];
	else return U8G_NOTUTF8;
      }
      else if ((ku = c - p1->base_ku) < p1->max_ku &&
	       ((ten = c1 - p1->base_ten) < p1->max_ten))
	  ret = ((unsigned short *) p1->tab)
	    [(ku*(p1->max_ten + p2->max_ten)) + ten];
      else return U8G_NOTUTF8;
    }
    else ret = c;		/* ASCII character */
    break;
  case CT_SJIS:			/* 2 byte Shift-JIS encoded JIS no table */
				/* compromise - do yen sign but not overline */
    if (!(c & BIT8)) ret = (c == JISROMAN_YEN) ? UCS2_YEN : c;
				/* half-width katakana? */
    else if ((c >= MIN_KANA_8) && (c < MAX_KANA_8)) ret = c + KANA_8;
    else {			/* Shift-JIS */
      if (j--) c1 = *t++;	/* get second octet */
      else return U8G_ENDSTRI;
      SJISTOJIS (c,c1);
      ret = JISTOUNICODE (c,c1,ku,ten);
    }
    break;

  case CT_UCS2:			/* 2 byte 16-bit Unicode no table */
    ret = c << 8;
    if (j--) c = *t++;		/* get second octet */
    else return U8G_ENDSTRI;	/* empty string */
    ret |= c;
    break;
  case CT_UCS4:			/* 4 byte 32-bit Unicode no table */
    if (c & 0x80) return U8G_NOTUTF8;
    if (j < 3) return U8G_ENDSTRI;
    j -= 3;			/* count three octets */
    ret = c << 24;
    ret |= (*t++) << 16;
    ret |= (*t++) << 8;
    ret |= (*t++);
    break;
  case CT_UTF16:		/* variable UTF-16 encoded Unicode no table */
    ret = c << 8;
    if (j--) c = *t++;		/* get second octet */
    else return U8G_ENDSTRI;	/* empty string */
    ret |= c;
				/* surrogate? */
    if ((ret >= UTF16_SURR) && (ret <= UTF16_MAXSURR)) {
				/* invalid first surrogate */
      if ((ret > UTF16_SURRHEND) || (j < 2)) return U8G_NOTUTF8;
      j -= 2;			/* count two octets */
      d = (*t++) << 8;		/* first octet of second surrogate */
      d |= *t++;		/* second octet of second surrogate */
      if ((d < UTF16_SURRL) || (d > UTF16_SURRLEND)) return U8G_NOTUTF8;
      ret = UTF16_BASE + ((ret & UTF16_MASK) << UTF16_SHIFT) +
	(d & UTF16_MASK);
    }
    break;
  default:			/* unknown/unsupported character set type */
    return U8G_NOTUTF8;
  }
  *s = t;			/* update pointer and counter */
  *i = j;
  return ret;
}

/* Produce charset validity map for BMP
 * Accepts: list of charsets to map
 * Returns: validity map, indexed by BMP codepoint
 *
 * Bit 0x1 is the "not-CJK" character bit
 */

unsigned long *utf8_csvalidmap (char *charsets[])
{
  unsigned short u,*tab;
  unsigned int m,ku,ten;
  unsigned long i,csi,csb;
  struct utf8_eucparam *param,*p2;
  char *s;
  const CHARSET *cs;
  unsigned long *ret = (unsigned long *)
    fs_get (i = 0x10000 * sizeof (unsigned long));
  memset (ret,0,i);		/* zero the entire vector */
				/* mark all the non-CJK codepoints */
	/* U+0000 - U+2E7F non-CJK */
  for (i = 0; i < 0x2E7F; ++i) ret[i] = 0x1;
	/* U+2E80 - U+2EFF CJK Radicals Supplement
	 * U+2F00 - U+2FDF Kangxi Radicals
	 * U+2FE0 - U+2FEF unassigned
	 * U+2FF0 - U+2FFF Ideographic Description Characters
	 * U+3000 - U+303F CJK Symbols and Punctuation
	 * U+3040 - U+309F Hiragana
	 * U+30A0 - U+30FF Katakana
	 * U+3100 - U+312F BoPoMoFo
	 * U+3130 - U+318F Hangul Compatibility Jamo
	 * U+3190 - U+319F Kanbun
	 * U+31A0 - U+31BF BoPoMoFo Extended
	 * U+31C0 - U+31EF CJK Strokes
	 * U+31F0 - U+31FF Katakana Phonetic Extensions
	 * U+3200 - U+32FF Enclosed CJK Letters and Months
	 * U+3300 - U+33FF CJK Compatibility
	 * U+3400 - U+4DBF CJK Unified Ideographs Extension A
	 * U+4DC0 - U+4DFF Yijing Hexagram Symbols
	 * U+4E00 - U+9FFF CJK Unified Ideographs
	 * U+A000 - U+A48F Yi Syllables
	 * U+A490 - U+A4CF Yi Radicals
	 * U+A700 - U+A71F Modifier Tone Letters
	 */
  for (i = 0xa720; i < 0xabff; ++i) ret[i] = 0x1;
	/* U+AC00 - U+D7FF Hangul Syllables */
  for (i = 0xd800; i < 0xf8ff; ++i) ret[i] = 0x1;
	/* U+F900 - U+FAFF CJK Compatibility Ideographs */
  for (i = 0xfb00; i < 0xfe2f; ++i) ret[i] = 0x1;
	/* U+FE30 - U+FE4F CJK Compatibility Forms
	 * U+FE50 - U+FE6F Small Form Variants (for CNS 11643)
	 */
  for (i = 0xfe70; i < 0xfeff; ++i) ret[i] = 0x1;
	/* U+FF00 - U+FFEF CJK Compatibility Ideographs */
  for (i = 0xfff0; i < 0x10000; ++i) ret[i] = 0x1;

				/* for each supplied charset */
  for (csi = 1; ret && charsets && (s = charsets[csi - 1]); ++csi) {
				/* substitute EUC-JP for ISO-2022-JP */
    if (!compare_cstring (s,"ISO-2022-JP")) s = "EUC-JP";
				/* look up charset */
    if ((cs = utf8_charset (s)) != NULL) {
      csb = 1 << csi;		/* charset bit */
      switch (cs->type) {
      case CT_ASCII:		/* 7-bit ASCII no table */
      case CT_1BYTE0:		/* 1 byte no table */
      case CT_1BYTE:		/* 1 byte ASCII + table 0x80-0xff */
      case CT_1BYTE8:		/* 1 byte table 0x00 - 0xff */
      case CT_EUC:		/* 2 byte ASCII + utf8_eucparam base/CS2/CS3 */
      case CT_DBYTE:		/* 2 byte ASCII + utf8_eucparam */
      case CT_DBYTE2:		/* 2 byte ASCII + utf8_eucparam plane1/2 */
      case CT_SJIS:		/* 2 byte Shift-JIS */
				/* supported charset type, all ASCII is OK */
	for (i = 0; i < 128; ++i) ret[i] |= csb;
	break;
      default:			/* unsupported charset type */
	fs_give ((void **) &ret);
	break;
      }
				/* now do additional operations */
      if (ret) switch (cs->type) {
      case CT_1BYTE0:		/* 1 byte no table */
	for (i = 128; i < 256; i++) ret[i] |= csb;
	break;
      case CT_1BYTE:		/* 1 byte ASCII + table 0x80-0xff */
	for (tab = (unsigned short *) cs->tab,i = 128; i < 256; i++)
	  if (tab[i & BITS7] != UBOGON) ret[tab[i & BITS7]] |= csb;
	break;
      case CT_1BYTE8:		/* 1 byte table 0x00 - 0xff */
	for (tab = (unsigned short *) cs->tab,i = 0; i < 256; i++)
	  if (tab[i] != UBOGON) ret[tab[i]] |= csb;
      break;
      case CT_EUC:		/* 2 byte ASCII + utf8_eucparam base/CS2/CS3 */
	for (param = (struct utf8_eucparam *) cs->tab,
	       tab = (unsigned short *) param->tab, ku = 0;
	     ku < param->max_ku; ku++)
	  for (ten = 0; ten < param->max_ten; ten++)
	    if ((u = tab[(ku * param->max_ten) + ten]) != UBOGON)
	      ret[u] |= csb;
	break;

      case CT_DBYTE:		/* 2 byte ASCII + utf8_eucparam */
	for (param = (struct utf8_eucparam *) cs->tab,
	       tab = (unsigned short *) param->tab, ku = 0;
	     ku < param->max_ku; ku++)
	  for (ten = 0; ten < param->max_ten; ten++)
	    if ((u = tab[(ku * param->max_ten) + ten]) != UBOGON)
	      ret[u] |= csb;
      break;
      case CT_DBYTE2:		/* 2 byte ASCII + utf8_eucparam plane1/2 */
	param = (struct utf8_eucparam *) cs->tab;
	p2 = param + 1;		/* plane 2 parameters */
				/* only ten parameters should differ */
	if ((param->base_ku != p2->base_ku) || (param->max_ku != p2->max_ku))
	  fatal ("ku definition error for CT_DBYTE2 charset");
				/* total codepoints in each ku */
	m = param->max_ten + p2->max_ten;
	tab = (unsigned short *) param->tab;
	for (ku = 0; ku < param->max_ku; ku++) {
	  for (ten = 0; ten < param->max_ten; ten++)
	    if ((u = tab[(ku * m) + ten]) != UBOGON)
	      ret[u] |= csb;
	  for (ten = 0; ten < p2->max_ten; ten++)
	    if ((u = tab[(ku * m) + param->max_ten + ten]) != UBOGON)
	      ret[u] |= csb;
	}
	break;
      case CT_SJIS:		/* 2 byte Shift-JIS */
	for (ku = 0; ku < MAX_JIS0208_KU; ku++)
	  for (ten = 0; ten < MAX_JIS0208_TEN; ten++)
	    if ((u = jis0208tab[ku][ten]) != UBOGON) ret[u] |= csb;
				/* JIS hankaku katakana */
	for (u = 0; u < (MAX_KANA_8 - MIN_KANA_8); u++)
	  ret[UCS2_KATAKANA + u] |= csb;
	break;
      }
    }
				/* invalid charset, punt */
    else fs_give ((void **) &ret);
  }
  return ret;
}

/* Infer charset from unlabelled sized text
 * Accepts: sized text
 * Returns: charset if one inferred, or NIL if unknown
 */

const CHARSET *utf8_infercharset (SIZEDTEXT *src)
{
  long iso2022jp = NIL;
  long eightbit = NIL;
  unsigned long i;
				/* look for ISO 2022 */
  if (src) for (i = 0; i < src->size; i++) {
				/* ESC sequence? */
    if ((src->data[i] == I2C_ESC) && (++i < src->size)) switch (src->data[i]) {
    case I2C_MULTI:		/* yes, multibyte? */
      if (++i < src->size) switch (src->data[i]) {
      case I2CS_94x94_JIS_OLD:	/* JIS X 0208-1978 */
      case I2CS_94x94_JIS_NEW:	/* JIS X 0208-1983 */
      case I2CS_94x94_JIS_EXT:	/* JIS X 0212-1990 (kludge...) */
	iso2022jp = T;		/* found an ISO-2022-JP sequence */
	break;
      default:			/* other multibyte */
	return NIL;		/* definitely invalid */
      }
      break;
    case I2C_G0_94:		/* single byte */
      if (++i < src->size) switch (src->data[i]) {
      case I2CS_94_JIS_BUGROM:	/* in case old buggy software */
      case I2CS_94_JIS_ROMAN:	/* JIS X 0201-1976 left half */
      case I2CS_94_ASCII:	/* ASCII */
      case I2CS_94_BRITISH:	/* good enough for gov't work */
	break;
      default:			/* other 94 single byte */
	return NIL;		/* definitely invalid */
      }
    }
				/* if possible UTF-8 and not ISO-2022-JP */
    else if (!iso2022jp && (eightbit >= 0) && (src->data[i] & BIT8) &&
	     (eightbit = utf8_validate (src->data + i,src->size - i)) > 0)
      i += eightbit - 1;	/* skip past all but last of UTF-8 char */
  }
				/* ISO-2022-JP overrides other guesses */
  if (iso2022jp) return utf8_charset ("ISO-2022-JP");
  if (eightbit > 0) return utf8_charset ("UTF-8");
  return eightbit ? NIL : utf8_charset ("US-ASCII");
}


/* Validate that character at this position is UTF-8
 * Accepts: string pointer
 *	    size of remaining string
 * Returns: size of UTF-8 character in octets or -1 if not UTF-8
 */

long utf8_validate (unsigned char *s,unsigned long i)
{
  unsigned long j = i;
  return (utf8_get (&s,&i) & U8G_ERROR) ? -1 : j - i;
}

/* Convert ISO 8859-1 to UTF-8
 * Accepts: source sized text
 *	    pointer to return sized text
 *	    canonicalization function
 */

void utf8_text_1byte0 (SIZEDTEXT *text,SIZEDTEXT *ret,ucs4cn_t cv,ucs4de_t de)
{
  unsigned long i;
  unsigned char *s;
  unsigned int c;
  for (ret->size = i = 0; i < text->size;) {
    c = text->data[i++];
    UTF8_COUNT_BMP (ret->size,c,cv,de)
  }
  (s = ret->data = (unsigned char *) fs_get (ret->size + 1))[ret->size] =NIL;
  for (i = 0; i < text->size;) {
    c = text->data[i++];
    UTF8_WRITE_BMP (s,c,cv,de)	/* convert UCS-2 to UTF-8 */
  }
}


/* Convert single byte ASCII+8bit character set sized text to UTF-8
 * Accepts: source sized text
 *	    pointer to return sized text
 *	    conversion table
 *	    canonicalization function
 */

void utf8_text_1byte (SIZEDTEXT *text,SIZEDTEXT *ret,void *tab,ucs4cn_t cv,
		      ucs4de_t de)
{
  unsigned long i;
  unsigned char *s;
  unsigned int c;
  unsigned short *tbl = (unsigned short *) tab;
  for (ret->size = i = 0; i < text->size;) {
    if ((c = text->data[i++]) & BIT8) c = tbl[c & BITS7];
    UTF8_COUNT_BMP (ret->size,c,cv,de)
  }
  (s = ret->data = (unsigned char *) fs_get (ret->size + 1))[ret->size] =NIL;
  for (i = 0; i < text->size;) {
    if ((c = text->data[i++]) & BIT8) c = tbl[c & BITS7];
    UTF8_WRITE_BMP (s,c,cv,de)	/* convert UCS-2 to UTF-8 */
  }
}

/* Convert single byte 8bit character set sized text to UTF-8
 * Accepts: source sized text
 *	    pointer to return sized text
 *	    conversion table
 *	    canonicalization function
 */

void utf8_text_1byte8 (SIZEDTEXT *text,SIZEDTEXT *ret,void *tab,ucs4cn_t cv,
		       ucs4de_t de)
{
  unsigned long i;
  unsigned char *s;
  unsigned int c;
  unsigned short *tbl = (unsigned short *) tab;
  for (ret->size = i = 0; i < text->size;) {
    c = tbl[text->data[i++]];
    UTF8_COUNT_BMP (ret->size,c,cv,de)
  }
  (s = ret->data = (unsigned char *) fs_get (ret->size + 1))[ret->size] =NIL;
  for (i = 0; i < text->size;) {
    c = tbl[text->data[i++]];
    UTF8_WRITE_BMP (s,c,cv,de)	/* convert UCS-2 to UTF-8 */
  }
}

/* Convert EUC sized text to UTF-8
 * Accepts: source sized text
 *	    pointer to return sized text
 *	    EUC parameter table
 *	    canonicalization function
 */

void utf8_text_euc (SIZEDTEXT *text,SIZEDTEXT *ret,void *tab,ucs4cn_t cv,
		    ucs4de_t de)
{
  unsigned long i;
  unsigned char *s;
  unsigned int pass,c,c1,ku,ten;
  struct utf8_eucparam *p1 = (struct utf8_eucparam *) tab;
  struct utf8_eucparam *p2 = p1 + 1;
  struct utf8_eucparam *p3 = p1 + 2;
  unsigned short *t1 = (unsigned short *) p1->tab;
  unsigned short *t2 = (unsigned short *) p2->tab;
  unsigned short *t3 = (unsigned short *) p3->tab;
  for (pass = 0,s = NIL,ret->size = 0; pass <= 1; pass++) {
    for (i = 0; i < text->size;) {
				/* not CS0? */
      if ((c = text->data[i++]) & BIT8) {
				/* yes, must have another high byte */
	if ((i >= text->size) || !((c1 = text->data[i++]) & BIT8))
	  c = UBOGON;		/* out of space or bogon */
	else switch (c) {	/* check 8bit code set */
	case EUC_CS2:		/* CS2 */
	  if (p2->base_ku) {	/* CS2 set up? */
	    if (p2->base_ten)	/* yes, multibyte? */
	      c = ((i < text->size) && ((c = text->data[i++]) & BIT8) &&
		   ((ku = (c1 & BITS7) - p2->base_ku) < p2->max_ku) &&
		   ((ten = (c & BITS7) - p2->base_ten) < p2->max_ten)) ?
		     t2[(ku*p2->max_ten) + ten] : UBOGON;
	    else c = ((c1 >= p2->base_ku) && (c1 < p2->max_ku)) ?
	      c1 + ((unsigned long) p2->tab) : UBOGON;
	  }	  
	  else {		/* CS2 not set up */
	    c = UBOGON;		/* swallow byte, say bogon */
	    if (i < text->size) i++;
	  }
	  break;
	case EUC_CS3:		/* CS3 */
	  if (p3->base_ku) {	/* CS3 set up? */
	    if (p3->base_ten)	/* yes, multibyte? */
	      c = ((i < text->size) && ((c = text->data[i++]) & BIT8) &&
		   ((ku = (c1 & BITS7) - p3->base_ku) < p3->max_ku) &&
		   ((ten = (c & BITS7) - p3->base_ten) < p3->max_ten)) ?
		     t3[(ku*p3->max_ten) + ten] : UBOGON;
	    else c = ((c1 >= p3->base_ku) && (c1 < p3->max_ku)) ?
	      c1 + ((unsigned long) p3->tab) : UBOGON;
	  }	  
	  else {		/* CS3 not set up */
	    c = UBOGON;		/* swallow byte, say bogon */
	    if (i < text->size) i++;
	  }
	  break;

	default:
	  if (((ku = (c & BITS7) - p1->base_ku) >= p1->max_ku) ||
	      ((ten = (c1 & BITS7) - p1->base_ten) >= p1->max_ten)) c = UBOGON;
	  else if (((c = t1[(ku*p1->max_ten) + ten]) == UBOGON) &&
		   /* special hack for JIS X 0212: merge rows less than 10 */
		   ku && (ku < 10) && t3 && p3->base_ten)
	    c = t3[((ku - (p3->base_ku - p1->base_ku))*p3->max_ten) + ten];
	}
      }
				/* convert if second pass */
      if (pass) UTF8_WRITE_BMP (s,c,cv,de)
      else UTF8_COUNT_BMP (ret->size,c,cv,de);
    }
    if (!pass) (s = ret->data = (unsigned char *)
		fs_get (ret->size + 1))[ret->size] =NIL;
  }
}


/* Convert ASCII + double-byte sized text to UTF-8
 * Accepts: source sized text
 *	    pointer to return sized text
 *	    conversion table
 *	    canonicalization function
 */

void utf8_text_dbyte (SIZEDTEXT *text,SIZEDTEXT *ret,void *tab,ucs4cn_t cv,
		      ucs4de_t de)
{
  unsigned long i;
  unsigned char *s;
  unsigned int c,c1,ku,ten;
  struct utf8_eucparam *p1 = (struct utf8_eucparam *) tab;
  unsigned short *t1 = (unsigned short *) p1->tab;
  for (ret->size = i = 0; i < text->size;) {
    if ((c = text->data[i++]) & BIT8) {
				/* special hack for GBK: 0x80 is Euro */
      if ((c == 0x80) && (t1 == (unsigned short *) gb2312tab)) c = UCS2_EURO;
      else c = ((i < text->size) && (c1 = text->data[i++]) &&
		((ku = c - p1->base_ku) < p1->max_ku) &&
		((ten = c1 - p1->base_ten) < p1->max_ten)) ?
	     t1[(ku*p1->max_ten) + ten] : UBOGON;
    }
    UTF8_COUNT_BMP (ret->size,c,cv,de)
  }
  (s = ret->data = (unsigned char *) fs_get (ret->size + 1))[ret->size] = NIL;
  for (i = 0; i < text->size;) {
    if ((c = text->data[i++]) & BIT8) {
				/* special hack for GBK: 0x80 is Euro */
      if ((c == 0x80) && (t1 == (unsigned short *) gb2312tab)) c = UCS2_EURO;
      else c = ((i < text->size) && (c1 = text->data[i++]) &&
		((ku = c - p1->base_ku) < p1->max_ku) &&
		((ten = c1 - p1->base_ten) < p1->max_ten)) ?
	     t1[(ku*p1->max_ten) + ten] : UBOGON;
    }
    UTF8_WRITE_BMP (s,c,cv,de)	/* convert UCS-2 to UTF-8 */
  }
}

/* Convert ASCII + double byte 2 plane sized text to UTF-8
 * Accepts: source sized text
 *	    pointer to return sized text
 *	    conversion table
 *	    canonicalization function
 */

void utf8_text_dbyte2 (SIZEDTEXT *text,SIZEDTEXT *ret,void *tab,ucs4cn_t cv,
		       ucs4de_t de)
{
  unsigned long i;
  unsigned char *s;
  unsigned int c,c1,ku,ten;
  struct utf8_eucparam *p1 = (struct utf8_eucparam *) tab;
  struct utf8_eucparam *p2 = p1 + 1;
  unsigned short *t = (unsigned short *) p1->tab;
  for (ret->size = i = 0; i < text->size;) {
    if ((c = text->data[i++]) & BIT8) {
      if ((i >= text->size) || !(c1 = text->data[i++]))
	c = UBOGON;		/* out of space or bogon */
      else if (c1 & BIT8)	/* high vs. low plane */
	c = ((ku = c - p2->base_ku) < p2->max_ku &&
	     ((ten = c1 - p2->base_ten) < p2->max_ten)) ?
	       t[(ku*(p1->max_ten + p2->max_ten)) + p1->max_ten + ten] :UBOGON;
      else c = ((ku = c - p1->base_ku) < p1->max_ku &&
		((ten = c1 - p1->base_ten) < p1->max_ten)) ?
		  t[(ku*(p1->max_ten + p2->max_ten)) + ten] : UBOGON;
    }
    UTF8_COUNT_BMP (ret->size,c,cv,de)
  }
  (s = ret->data = (unsigned char *) fs_get (ret->size + 1))[ret->size] = NIL;
  for (i = 0; i < text->size;) {
    if ((c = text->data[i++]) & BIT8) {
      if ((i >= text->size) || !(c1 = text->data[i++]))
	c = UBOGON;		/* out of space or bogon */
      else if (c1 & BIT8)	/* high vs. low plane */
	c = ((ku = c - p2->base_ku) < p2->max_ku &&
	     ((ten = c1 - p2->base_ten) < p2->max_ten)) ?
	       t[(ku*(p1->max_ten + p2->max_ten)) + p1->max_ten + ten] :UBOGON;
      else c = ((ku = c - p1->base_ku) < p1->max_ku &&
		((ten = c1 - p1->base_ten) < p1->max_ten)) ?
		  t[(ku*(p1->max_ten + p2->max_ten)) + ten] : UBOGON;
    }
    UTF8_WRITE_BMP (s,c,cv,de)	/* convert UCS-2 to UTF-8 */
  }
}

#ifdef JISTOUNICODE		/* Japanese */
/* Convert Shift JIS sized text to UTF-8
 * Accepts: source sized text
 *	    pointer to return sized text
 *	    canonicalization function
 */

void utf8_text_sjis (SIZEDTEXT *text,SIZEDTEXT *ret,ucs4cn_t cv,
		     ucs4de_t de)
{
  unsigned long i;
  unsigned char *s;
  unsigned int c,c1,ku,ten;
  for (ret->size = i = 0; i < text->size;) {
    if ((c = text->data[i++]) & BIT8) {
				/* half-width katakana */
      if ((c >= MIN_KANA_8) && (c < MAX_KANA_8)) c += KANA_8;
      else if (i >= text->size) c = UBOGON;
      else {			/* Shift-JIS */
	c1 = text->data[i++];
	SJISTOJIS (c,c1);
	c = JISTOUNICODE (c,c1,ku,ten);
      }
    }
				/* compromise - do yen sign but not overline */
    else if (c == JISROMAN_YEN) c = UCS2_YEN;
    UTF8_COUNT_BMP (ret->size,c,cv,de)
  }
  (s = ret->data = (unsigned char *) fs_get (ret->size + 1))[ret->size] = NIL;
  for (i = 0; i < text->size;) {
    if ((c = text->data[i++]) & BIT8) {
				/* half-width katakana */
      if ((c >= MIN_KANA_8) && (c < MAX_KANA_8)) c += KANA_8;
      else {			/* Shift-JIS */
	c1 = text->data[i++];
	SJISTOJIS (c,c1);
	c = JISTOUNICODE (c,c1,ku,ten);
      }
    }
				/* compromise - do yen sign but not overline */
    else if (c == JISROMAN_YEN) c = UCS2_YEN;
    UTF8_WRITE_BMP (s,c,cv,de)	/* convert UCS-2 to UTF-8 */
  }
}
#endif

/* Convert ISO-2022 sized text to UTF-8
 * Accepts: source sized text
 *	    pointer to returned sized text
 *	    canonicalization function
 */

void utf8_text_2022 (SIZEDTEXT *text,SIZEDTEXT *ret,ucs4cn_t cv,ucs4de_t de)
{
  unsigned long i;
  unsigned char *s;
  unsigned int pass,state,c,co,gi,gl,gr,g[4],ku,ten;
  for (pass = 0,s = NIL,ret->size = 0; pass <= 1; pass++) {
    gi = 0;			/* quell compiler warnings */
    state = I2S_CHAR;		/* initialize engine */
    g[0]= g[2] = I2CS_ASCII;	/* G0 and G2 are ASCII */
    g[1]= g[3] = I2CS_ISO8859_1;/* G1 and G3 are ISO-8850-1 */
    gl = I2C_G0; gr = I2C_G1;	/* left is G0, right is G1 */
    for (i = 0; i < text->size;) {
      c = text->data[i++];
      switch (state) {		/* dispatch based upon engine state */
      case I2S_ESC:		/* ESC seen */
	switch (c) {		/* process intermediate character */
	case I2C_MULTI:		/* multibyte character? */
	  state = I2S_MUL;	/* mark multibyte flag seen */
	  break;
        case I2C_SS2:		/* single shift GL to G2 */
	case I2C_SS2_ALT:	/* Taiwan SeedNet */
	  gl |= I2C_SG2;
	  break;
        case I2C_SS3:		/* single shift GL to G3 */
	case I2C_SS3_ALT:	/* Taiwan SeedNet */
	  gl |= I2C_SG3;
	  break;
        case I2C_LS2:		/* shift GL to G2 */
	  gl = I2C_G2;
	  break;
        case I2C_LS3:		/* shift GL to G3 */
	  gl = I2C_G3;
	  break;
        case I2C_LS1R:		/* shift GR to G1 */
	  gr = I2C_G1;
	  break;
        case I2C_LS2R:		/* shift GR to G2 */
	  gr = I2C_G2;
	  break;
        case I2C_LS3R:		/* shift GR to G3 */
	  gr = I2C_G3;
	  break;
	case I2C_G0_94: case I2C_G1_94: case I2C_G2_94:	case I2C_G3_94:
	  g[gi = c - I2C_G0_94] = (state == I2S_MUL) ? I2CS_94x94 : I2CS_94;
	  state = I2S_INT;	/* ready for character set */
	  break;
	case I2C_G0_96:	case I2C_G1_96: case I2C_G2_96:	case I2C_G3_96:
	  g[gi = c - I2C_G0_96] = (state == I2S_MUL) ? I2CS_96x96 : I2CS_96;
	  state = I2S_INT;	/* ready for character set */
	  break;
	default:		/* bogon */
	  if (pass) *s++ = I2C_ESC,*s++ = c;
	  else ret->size += 2;
	  state = I2S_CHAR;	/* return to previous state */
	}
	break;

      case I2S_MUL:		/* ESC $ */
	switch (c) {		/* process multibyte intermediate character */
	case I2C_G0_94: case I2C_G1_94: case I2C_G2_94:	case I2C_G3_94:
	  g[gi = c - I2C_G0_94] = I2CS_94x94;
	  state = I2S_INT;	/* ready for character set */
	  break;
	case I2C_G0_96:	case I2C_G1_96: case I2C_G2_96:	case I2C_G3_96:
	  g[gi = c - I2C_G0_96] = I2CS_96x96;
	  state = I2S_INT;	/* ready for character set */
	  break;
	default:		/* probably omitted I2CS_94x94 */
	  g[gi = I2C_G0] = I2CS_94x94 | c;
	  state = I2S_CHAR;	/* return to character state */
	}
	break;
      case I2S_INT:
	state = I2S_CHAR;	/* return to character state */
	g[gi] |= c;		/* set character set */
	break;

      case I2S_CHAR:		/* character data */
	switch (c) {
	case I2C_ESC:		/* ESC character */
	  state = I2S_ESC;	/* see if ISO-2022 prefix */
	  break;
	case I2C_SI:		/* shift GL to G0 */
	  gl = I2C_G0;
	  break;
	case I2C_SO:		/* shift GL to G1 */
	  gl = I2C_G1;
	  break;
        case I2C_SS2_ALT:	/* single shift GL to G2 */
	case I2C_SS2_ALT_7:
	  gl |= I2C_SG2;
	  break;
        case I2C_SS3_ALT:	/* single shift GL to G3 */
	case I2C_SS3_ALT_7:
	  gl |= I2C_SG3;
	  break;

	default:		/* ordinary character */
	  co = c;		/* note original character */
	  if (gl & (3 << 2)) {	/* single shifted? */
	    gi = g[gl >> 2];	/* get shifted character set */
	    gl &= 0x3;		/* cancel shift */
	  }
				/* select left or right half */
	  else gi = (c & BIT8) ? g[gr] : g[gl];
	  c &= BITS7;		/* make 7-bit */
	  switch (gi) {		/* interpret in character set */
	  case I2CS_ASCII:	/* ASCII */
	    break;		/* easy! */
	  case I2CS_BRITISH:	/* British ASCII */
				/* Pound sterling sign */
	    if (c == BRITISH_POUNDSTERLING) c = UCS2_POUNDSTERLING;
	    break;
	  case I2CS_JIS_ROMAN:	/* JIS Roman */
	  case I2CS_JIS_BUGROM:	/* old bugs */
	    switch (c) {	/* two exceptions to ASCII */
	    case JISROMAN_YEN:	/* Yen sign */
	      c = UCS2_YEN;
	      break;
				/* overline */
	    case JISROMAN_OVERLINE:
	      c = UCS2_OVERLINE;
	      break;
	    }
	    break;
	  case I2CS_JIS_KANA:	/* JIS hankaku katakana */
	    if ((c >= MIN_KANA_7) && (c < MAX_KANA_7)) c += KANA_7;
	    break;

	  case I2CS_ISO8859_1:	/* Latin-1 (West European) */
	    c |= BIT8;		/* just turn on high bit */
	    break;
	  case I2CS_ISO8859_2:	/* Latin-2 (Czech, Slovak) */
	    c = iso8859_2tab[c];
	    break;
	  case I2CS_ISO8859_3:	/* Latin-3 (Dutch, Turkish) */
	    c = iso8859_3tab[c];
	    break;
	  case I2CS_ISO8859_4:	/* Latin-4 (Scandinavian) */
	    c = iso8859_4tab[c];
	    break;
	  case I2CS_ISO8859_5:	/* Cyrillic */
	    c = iso8859_5tab[c];
	    break;
	  case I2CS_ISO8859_6:	/* Arabic */
	    c = iso8859_6tab[c];
	    break;
	  case I2CS_ISO8859_7:	/* Greek */
	    c = iso8859_7tab[c];
	    break;
	  case I2CS_ISO8859_8:	/* Hebrew */
	    c = iso8859_8tab[c];
	    break;
	  case I2CS_ISO8859_9:	/* Latin-5 (Finnish, Portuguese) */
	    c = iso8859_9tab[c];
	    break;
	  case I2CS_TIS620:	/* Thai */
	    c = tis620tab[c];
	    break;
	  case I2CS_ISO8859_10:	/* Latin-6 (Northern Europe) */
	    c = iso8859_10tab[c];
	    break;
	  case I2CS_ISO8859_13:	/* Latin-7 (Baltic) */
	    c = iso8859_13tab[c];
	    break;
	  case I2CS_VSCII:	/* Vietnamese */
	    c = visciitab[c];
	    break;
	  case I2CS_ISO8859_14:	/* Latin-8 (Celtic) */
	    c = iso8859_14tab[c];
	    break;
	  case I2CS_ISO8859_15:	/* Latin-9 (Euro) */
	    c = iso8859_15tab[c];
	    break;
	  case I2CS_ISO8859_16:	/* Latin-10 (Baltic) */
	    c = iso8859_16tab[c];
	    break;

	  default:		/* all other character sets */
				/* multibyte character set */
	    if ((gi & I2CS_MUL) && !(c & BIT8) && isgraph (c)) {
	      c = (i < text->size) ? text->data[i++] : 0;
	      switch (gi) {
#ifdef GBTOUNICODE
	      case I2CS_GB:	/* GB 2312 */
		co |= BIT8;	/* make into EUC */
		c |= BIT8;
		c = GBTOUNICODE (co,c,ku,ten);
		break;
#endif
#ifdef JISTOUNICODE
	      case I2CS_JIS_OLD:/* JIS X 0208-1978 */
	      case I2CS_JIS_NEW:/* JIS X 0208-1983 */
		c = JISTOUNICODE (co,c,ku,ten);
		break;
#endif
#ifdef JIS0212TOUNICODE
	      case I2CS_JIS_EXT:/* JIS X 0212-1990 */
		c = JIS0212TOUNICODE (co,c,ku,ten);
		break;
#endif
#ifdef KSCTOUNICODE
	      case I2CS_KSC:	/* KSC 5601 */
		co |= BIT8;	/* make into EUC */
		c |= BIT8;
		c = KSCTOUNICODE (co,c,ku,ten);
		break;
#endif
#ifdef CNS1TOUNICODE
	      case I2CS_CNS1:	/* CNS 11643 plane 1 */
		c = CNS1TOUNICODE (co,c,ku,ten);
		break;
#endif
#ifdef CNS2TOUNICODE
	      case I2CS_CNS2:	/* CNS 11643 plane 2 */
		c = CNS2TOUNICODE (co,c,ku,ten);
		break;
#endif
#ifdef CNS3TOUNICODE
	      case I2CS_CNS3:	/* CNS 11643 plane 3 */
		c = CNS3TOUNICODE (co,c,ku,ten);
		break;
#endif
#ifdef CNS4TOUNICODE
	      case I2CS_CNS4:	/* CNS 11643 plane 4 */
		c = CNS4TOUNICODE (co,c,ku,ten);
		break;
#endif
#ifdef CNS5TOUNICODE
	      case I2CS_CNS5:	/* CNS 11643 plane 5 */
		c = CNS5TOUNICODE (co,c,ku,ten);
		break;
#endif
#ifdef CNS6TOUNICODE
	      case I2CS_CNS6:	/* CNS 11643 plane 6 */
		c = CNS6TOUNICODE (co,c,ku,ten);
		break;
#endif
#ifdef CNS7TOUNICODE
	      case I2CS_CNS7:	/* CNS 11643 plane 7 */
		c = CNS7TOUNICODE (co,c,ku,ten);
		break;
#endif
	      default:		/* unknown multibyte, treat as UCS-2 */
		c |= (co << 8);	/* wrong, but nothing else to do */
		break;
	      }
	    }
	    else c = co;	/* unknown single byte, treat as 8859-1 */
	  }
				/* convert if second pass */
	  if (pass) UTF8_WRITE_BMP (s,c,cv,de)
	  else UTF8_COUNT_BMP (ret->size,c,cv,de);
	}
      }
    }
    if (!pass) (s = ret->data = (unsigned char *)
		fs_get (ret->size + 1))[ret->size] = NIL;
    else if (((unsigned long) (s - ret->data)) != ret->size)
      fatal ("ISO-2022 to UTF-8 botch");
  }
}

/* Convert UTF-7 sized text to UTF-8
 * Accepts: source sized text
 *	    pointer to returned sized text
 *	    canonicalization function
 */

void utf8_text_utf7 (SIZEDTEXT *text,SIZEDTEXT *ret,ucs4cn_t cv,ucs4de_t de)
{
  unsigned long i;
  unsigned char *s;
  unsigned int c,c1,d,uc,pass,e,e1,state,surrh;
  for (pass = 0,s = NIL,ret->size = 0; pass <= 1; pass++) {
    c1 = d = uc = e = e1 = 0;
    for (i = 0,state = NIL; i < text->size;) {
      c = text->data[i++];	/* get next byte */
      switch (state) {
      case U7_PLUS:		/* previous character was + */
	if (c == '-') {		/* +- means textual + */
	  c = '+';
	  state = U7_ASCII;	/* revert to ASCII */
	  break;
	}
	state = U7_UNICODE;	/* enter Unicode state */
	e = e1 = 0;		/* initialize Unicode quantum position */
      case U7_UNICODE:		/* Unicode state */
	if (c == '-') state = U7_MINUS;
	else {			/* decode Unicode */
	  /* don't use isupper/islower since this is ASCII only */
	  if ((c >= 'A') && (c <= 'Z')) c -= 'A';
	  else if ((c >= 'a') && (c <= 'z')) c -= 'a' - 26;
	  else if (isdigit (c)) c -= '0' - 52;
	  else if (c == '+') c = 62;
	  else if (c == '/') c = 63;
	  else state = U7_ASCII;/* end of modified BASE64 */
	}
	break;
      case U7_MINUS:		/* previous character was absorbed - */
	state = U7_ASCII;	/* revert to ASCII */
      case U7_ASCII:		/* ASCII state */
	if (c == '+') state = U7_PLUS;
	break;
      }

      switch (state) {		/* store character if in character mode */
      case U7_UNICODE:		/* Unicode */
	switch (e++) {		/* install based on BASE64 state */
	case 0:
	  c1 = c << 2;		/* byte 1: high 6 bits */
	  break;
	case 1:
	  d = c1 | (c >> 4);	/* byte 1: low 2 bits */
	  c1 = c << 4;		/* byte 2: high 4 bits */
	  break;
	case 2:
	  d = c1 | (c >> 2);	/* byte 2: low 4 bits */
	  c1 = c << 6;		/* byte 3: high 2 bits */
	  break;
	case 3:
	  d = c | c1;		/* byte 3: low 6 bits */
	  e = 0;		/* reinitialize mechanism */
	  break;
	}
	if (e == 1) break;	/* done if first BASE64 state */
	if (!e1) {		/* first byte of UCS-2 character */
	  uc = (d & 0xff) << 8;	/* note first byte */
	  e1 = T;		/* enter second UCS-2 state */
	  break;		/* done */
	}
	c = uc | (d & 0xff);	/* build UCS-2 character */
	e1 = NIL;		/* back to first UCS-2 state, drop in */
				/* surrogate pair?  */
	if ((c >= UTF16_SURR) && (c <= UTF16_MAXSURR)) {
				/* save high surrogate for later */
	  if (c < UTF16_SURRL) surrh = c;
	  else c = UTF16_BASE + ((surrh & UTF16_MASK) << UTF16_SHIFT) +
		 (c & UTF16_MASK);
	  break;		/* either way with surrogates, we're done */
	}
      case U7_ASCII:		/* just install if ASCII */
				/* convert if second pass */
	if (pass) UTF8_WRITE_BMP (s,c,cv,de)
	else UTF8_COUNT_BMP (ret->size,c,cv,de);
      }
    }
    if (!pass) (s = ret->data = (unsigned char *)
		fs_get (ret->size + 1))[ret->size] = NIL;
    else if (((unsigned long) (s - ret->data)) != ret->size)
      fatal ("UTF-7 to UTF-8 botch");
  }
}


/* Convert UTF-8 sized text to UTF-8
 * Accepts: source sized text
 *	    pointer to returned sized text
 *	    canonicalization function
 */

void utf8_text_utf8 (SIZEDTEXT *text,SIZEDTEXT *ret,ucs4cn_t cv,ucs4de_t de)
{
  unsigned long i,c;
  unsigned char *s,*t;
  for (ret->size = 0, t = text->data, i = text->size; i;) {
    if ((c = utf8_get (&t,&i)) & U8G_ERROR) {
      ret->data = text->data;	/* conversion failed */
      ret->size = text->size;
      return;
    }
    UTF8_COUNT (ret->size,c,cv,de)
  }
  (s = ret->data = (unsigned char *) fs_get (ret->size + 1))[ret->size] =NIL;
  for (t = text->data, i = text->size; i;) {
    c = utf8_get (&t,&i);
    UTF8_WRITE (s,c,cv,de)	/* convert UCS-4 to UTF-8 */
  }
  if (((unsigned long) (s - ret->data)) != ret->size)
    fatal ("UTF-8 to UTF-8 botch");
}

/* Convert UCS-2 sized text to UTF-8
 * Accepts: source sized text
 *	    pointer to returned sized text
 *	    canonicalization function
 */

void utf8_text_ucs2 (SIZEDTEXT *text,SIZEDTEXT *ret,ucs4cn_t cv,ucs4de_t de)
{
  unsigned long i;
  unsigned char *s,*t;
  unsigned int c;
  for (ret->size = 0, t = text->data, i = text->size / 2; i; --i) {
    c = *t++ << 8;
    c |= *t++;
    UTF8_COUNT_BMP (ret->size,c,cv,de);
  }
  (s = ret->data = (unsigned char *) fs_get (ret->size + 1))[ret->size] = NIL;
  for (t = text->data, i = text->size / 2; i; --i) {
    c = *t++ << 8;
    c |= *t++;
    UTF8_WRITE_BMP (s,c,cv,de)	/* convert UCS-2 to UTF-8 */
  }
  if (((unsigned long) (s - ret->data)) != ret->size)
    fatal ("UCS-2 to UTF-8 botch");
}


/* Convert UCS-4 sized text to UTF-8
 * Accepts: source sized text
 *	    pointer to returned sized text
 *	    canonicalization function
 */

void utf8_text_ucs4 (SIZEDTEXT *text,SIZEDTEXT *ret,ucs4cn_t cv,ucs4de_t de)
{
  unsigned long i;
  unsigned char *s,*t;
  unsigned long c;
  for (ret->size = 0, t = text->data, i = text->size / 4; i; --i) {
    c = *t++ << 24; c |= *t++ << 16; c |= *t++ << 8; c |= *t++;
    UTF8_COUNT (ret->size,c,cv,de);
  }
  (s = ret->data = (unsigned char *) fs_get (ret->size + 1))[ret->size] = NIL;
  for (t = text->data, i = text->size / 2; i; --i) {
    c = *t++ << 24; c |= *t++ << 16; c |= *t++ << 8; c |= *t++;
    UTF8_WRITE (s,c,cv,de)	/* convert UCS-4 to UTF-8 */
  }
  if (((unsigned long) (s - ret->data)) != ret->size)
    fatal ("UCS-4 to UTF-8 botch");
}

/* Convert UTF-16 sized text to UTF-8
 * Accepts: source sized text
 *	    pointer to returned sized text
 *	    canonicalization function
 */

void utf8_text_utf16 (SIZEDTEXT *text,SIZEDTEXT *ret,ucs4cn_t cv,ucs4de_t de)
{
  unsigned long i;
  unsigned char *s,*t;
  unsigned long c,d;
  for (ret->size = 0, t = text->data, i = text->size / 2; i; --i) {
    c = *t++ << 8;
    c |= *t++;
				/* possible surrogate? */
    if ((c >= UTF16_SURR) && (c <= UTF16_MAXSURR)) {
				/* invalid first surrogate */
      if ((c > UTF16_SURRHEND) || !i) c = UBOGON;
      else {			/* get second surrogate */
	d = *t++ << 8;
	d |= *t++;
	--i;			/* swallowed another 16-bits */
				/* invalid second surrogate */
	if ((d < UTF16_SURRL) || (d > UTF16_SURRLEND)) c = UBOGON;
	else c = UTF16_BASE + ((c & UTF16_MASK) << UTF16_SHIFT) +
	       (d & UTF16_MASK);
      }
    }
    UTF8_COUNT (ret->size,c,cv,de);
  }
  (s = ret->data = (unsigned char *) fs_get (ret->size + 1))[ret->size] = NIL;
  for (t = text->data, i = text->size / 2; i; --i) {
    c = *t++ << 8;
    c |= *t++;
				/* possible surrogate? */
    if ((c >= UTF16_SURR) && (c <= UTF16_MAXSURR)) {
				/* invalid first surrogate */
      if ((c > UTF16_SURRHEND) || !i) c = UBOGON;
      else {			/* get second surrogate */
	d = *t++ << 8;
	d |= *t++;
	--i;			/* swallowed another 16-bits */
				/* invalid second surrogate */
	if ((d < UTF16_SURRL) || (d > UTF16_SURRLEND)) c = UBOGON;
	else c = UTF16_BASE + ((c & UTF16_MASK) << UTF16_SHIFT) +
	       (d & UTF16_MASK);
      }
    }
    UTF8_WRITE (s,c,cv,de)	/* convert UCS-4 to UTF-8 */
  }
  if (((unsigned long) (s - ret->data)) != ret->size)
    fatal ("UTF-16 to UTF-8 botch");
}

/* Size of UCS-4 character, possibly not in BMP, as UTF-8 octets
 * Accepts: character
 * Returns: size (0 means bogon)
 *
 * Use UTF8_SIZE macro if known to be in the BMP
 */

unsigned long utf8_size (unsigned long c)
{
  if (c < 0x80) return 1;
  else if (c < 0x800) return 2;
  else if (c < 0x10000) return 3;
  else if (c < 0x200000) return 4;
  else if (c < 0x4000000) return 5;
  else if (c < 0x80000000) return 6;
  return 0;
}


/* Put UCS-4 character, possibly not in BMP, as UTF-8 octets
 * Accepts: destination string pointer
 *	    character
 * Returns: updated destination pointer
 *
 * Use UTF8_PUT_BMP macro if known to be in the BMP
 */

unsigned char *utf8_put (unsigned char *s,unsigned long c)
{
  unsigned char mark[6] = {0x00,0xc0,0xe0,0xf0,0xf8,0xfc};
  unsigned long size = utf8_size (c);
  switch (size) {
  case 6:
    s[5] = 0x80 | (unsigned char) (c & 0x3f);
    c >>= 6;
  case 5:
    s[4] = 0x80 | (unsigned char) (c & 0x3f);
    c >>= 6;
  case 4:
    s[3] = 0x80 | (unsigned char) (c & 0x3f);
    c >>= 6;
  case 3:
    s[2] = 0x80 | (unsigned char) (c & 0x3f);
    c >>= 6;
  case 2:
    s[1] = 0x80 | (unsigned char) (c & 0x3f);
    c >>= 6;
  case 1:
    *s = mark[size-1] | (unsigned char) (c & 0x7f);
    break;
  }
  return s + size;
}

/* Return title case of a fixed-width UCS-4 character
 * Accepts: character
 * Returns: title case of character
 */

unsigned long ucs4_titlecase (unsigned long c)
{
  if (c <= UCS4_TMAPMAX) return ucs4_tmaptab[c];
  if (c < UCS4_TMAPHIMIN) return c;
  if (c <= UCS4_TMAPHIMAX) return c - UCS4_TMAPHIMAP;
  if (c < UCS4_TMAPDESERETMIN) return c;
  if (c <= UCS4_TMAPDESERETMAX) return c - UCS4_TMAPDESERETMAP;
  return c;
}


/* Return width of a fixed-width UCS-4 character in planes 0-2
 * Accepts: character
 * Returns: width (0, 1, 2) or negative error condition if not valid
 */

long ucs4_width (unsigned long c)
{
  long ret;
  ucs4width_t uw = (ucs4width_t) utf8_parameters(GET_UCS4WIDTH, NIL);
				/* out of range, not-a-char, or surrogates */
  if ((c > UCS4_MAXUNICODE) || ((c & 0xfffe) == 0xfffe) ||
      ((c >= UTF16_SURR) && (c <= UTF16_MAXSURR))) ret = U4W_NOTUNCD;
				/* private-use */
  else if (c >= UCS4_PVTBASE) ret = U4W_PRIVATE;
				/* SSP are not printing characters */
  else if (c >= UCS4_SSPBASE) ret = U4W_SSPCHAR;
				/* unassigned planes */
  else if (c >= UCS4_UNABASE) ret = U4W_UNASSGN;
				/* SIP and reserved plane 3 are wide */
  else if (c >= UCS4_SIPBASE) ret = 2;
#if (UCS4_WIDLEN != UCS4_SIPBASE)
#error "UCS4_WIDLEN != UCS4_SIPBASE"
#endif
				/* C0/C1 controls */
  else if ((c <= UCS2_C0CONTROLEND) ||
	   ((c >= UCS2_C1CONTROL) && (c <= UCS2_C1CONTROLEND)))
    ret = U4W_CONTROL;
				/* BMP and SMP get value from table */
  else switch (ret = (ucs4_widthtab[(c >> 2)] >> ((3 - (c & 0x3)) << 1)) &0x3){
  case 0:			/* zero-width */
    if (c == 0x00ad) ret = 1;	/* force U+00ad (SOFT HYPHEN) to width 1 */
  case 1:			/* single-width */
  case 2:			/* double-width */
    break;
  case 3:			/* ambiguous width */
    ret = uw ? (*uw)(c)		/* this is better than the line below */
	     : (c >= 0x2100) ? 2 : 1;/* need to do something better than this */
    break;
  }
  return ret;
}

/* Return screen width of UTF-8 string
 * Accepts: string
 * Returns: width or negative if not valid UTF-8
 */

long utf8_strwidth (unsigned char *s)
{
  unsigned long c,i,ret;
				/* go through string */
  for (ret = 0; *s; ret += ucs4_width (c)) {
    /* It's alright to give a fake value for the byte count to utf8_get()
     * since the null of a null-terminated string will stop processing anyway.
     */
    i = 6;			/* fake value */
    if ((c = utf8_get (&s,&i)) & U8G_ERROR) return -1;
  }
  return ret;
}


/* Return screen width of UTF-8 text
 * Accepts: SIZEDTEXT to string
 * Returns: width or negative if not valid UTF-8
 */

long utf8_textwidth (SIZEDTEXT *utf8)
{
  unsigned long c;
  unsigned char *s = utf8->data;
  unsigned long i = utf8->size;
  unsigned long ret = 0;
  while (i) {			/* while there's a string to process */
    if ((c = utf8_get (&s,&i)) & U8G_ERROR) return -1;
    ret += ucs4_width (c);
  }
  return ret;
}

/* Decomposition (phew!) */

#define MORESINGLE 1		/* single UCS-4 tail value */
#define MOREMULTIPLE 2		/* multiple UCS-2 tail values */

struct decomposemore {
  short type;			/* type of more */
  union {
    unsigned long single;	/* single decomposed value */
    struct {			/* multiple BMP values */
      unsigned short *next;
      unsigned long count;
    } multiple;
  } data;
};

#define RECURSIVEMORE struct recursivemore

RECURSIVEMORE {
  struct decomposemore *more;
  RECURSIVEMORE *next;
};


/* Return decomposition of a UCS-4 character
 * Accepts: character or U8G_ERROR to return next from "more"
 *	    pointer to returned more
 * Returns: [next] decomposed value, more set if still more decomposition
 */

unsigned long ucs4_decompose (unsigned long c,void **more)
{
  unsigned long i,ix,ret;
  struct decomposemore *m;
  if (c & U8G_ERROR) {		/* want to chase more? */
				/* do sanity check */
    if ((m = (struct decomposemore *) *more) != NULL) switch (m->type) {
    case MORESINGLE:		/* single value */
      ret = m->data.single;
      fs_give (more);		/* no more decomposition */
      break;
    case MOREMULTIPLE:		/* multiple value */
      ret = *m->data.multiple.next++;
      if (!--m->data.multiple.count) fs_give (more);
      break;
    default:			/* uh-oh */
      fatal ("invalid more block argument to ucs4_decompose!");
    }
    else fatal ("no more block provided to ucs4_decompose!");
  }

  else {			/* start decomposition */
    *more = NIL;		/* initially set no more */
				/* BMP low decompositions */
    if (c < UCS4_BMPLOMIN) ret = c;
				/* fix this someday */
    else if (c == UCS4_BMPLOMIN) ret = ucs4_dbmplotab[0];
    else if (c <= UCS4_BMPLOMAX) {
				/* within range - have a decomposition? */
      if ((i = ucs4_dbmploixtab[c - UCS4_BMPLOMIN]) != 0L) {
				/* get first value of decomposition */
	ret = ucs4_dbmplotab[ix = i & UCS4_BMPLOIXMASK];
				/* has continuation? */
	if (i & UCS4_BMPLOSIZEMASK) {
	  m = (struct decomposemore *)
	    (*more = memset (fs_get (sizeof (struct decomposemore)),0,
			    sizeof (struct decomposemore)));
	  m->type = MOREMULTIPLE;
	  m->data.multiple.next = &ucs4_dbmplotab[++ix];
	  m->data.multiple.count = i >> UCS4_BMPLOSIZESHIFT;
	}
      }
      else ret = c;		/* in range but doesn't decompose */
    }
				/* BMP CJK compatibility */
    else if (c < UCS4_BMPCJKMIN) ret = c;
    else if (c <= UCS4_BMPCJKMAX) {
      if (!(ret = ucs4_bmpcjk1decomptab[c - UCS4_BMPCJKMIN])) ret = c;
    }
				/* BMP CJK compatibility - some not in BMP */
#if UCS4_BMPCJK2MIN - (UCS4_BMPCJKMAX + 1)
    else if (c < UCS4_BMPCJK2MIN) ret = c;
#endif
    else if (c <= UCS4_BMPCJK2MAX)
      ret = ucs4_bmpcjk2decomptab[c - UCS4_BMPCJK2MIN];
				/* BMP high decompositions */
    else if (c < UCS4_BMPHIMIN) ret = c;
    else if (c <= UCS4_BMPHIMAX) {
				/* within range - have a decomposition? */
      if ((i = ucs4_dbmphiixtab[c - UCS4_BMPHIMIN]) != 0L) {
				/* get first value of decomposition */
	ret = ucs4_dbmphitab[ix = i & UCS4_BMPHIIXMASK];
				/* has continuation? */
	if (i & UCS4_BMPHISIZEMASK) {
	  m = (struct decomposemore *)
	    (*more = memset (fs_get (sizeof (struct decomposemore)),0,
			    sizeof (struct decomposemore)));
	  m->type = MOREMULTIPLE;
	  m->data.multiple.next = &ucs4_dbmphitab[++ix];
	  m->data.multiple.count = i >> UCS4_BMPHISIZESHIFT;
	}
      }
      else ret = c;		/* in range but doesn't decompose */
    }

				/* BMP half and full width forms */
    else if (c < UCS4_BMPHALFFULLMIN) ret = c;
    else if (c <= UCS4_BMPHALFFULLMAX) {
      if (!(ret = ucs4_bmphalffulldecomptab[c - UCS4_BMPHALFFULLMIN])) ret = c;
    }
				/* SMP music */
    else if (c < UCS4_SMPMUSIC1MIN) ret = c;
    else if (c <= UCS4_SMPMUSIC1MAX) {
      ret = ucs4_smpmusic1decomptab[c -= UCS4_SMPMUSIC1MIN][0];
      m = (struct decomposemore *)
	(*more = memset (fs_get (sizeof (struct decomposemore)),0,
			 sizeof (struct decomposemore)));
      m->type = MORESINGLE;
      m->data.single = ucs4_smpmusic1decomptab[c][1];
    }
    else if (c < UCS4_SMPMUSIC2MIN) ret = c;
    else if (c <= UCS4_SMPMUSIC2MAX) {
      ret = ucs4_smpmusic2decomptab[c -= UCS4_SMPMUSIC2MIN][0];
      m = (struct decomposemore *)
	(*more = memset (fs_get (sizeof (struct decomposemore)),0,
			 sizeof (struct decomposemore)));
      m->type = MORESINGLE;
      m->data.single = ucs4_smpmusic2decomptab[c][1];
    }
				/* SMP mathematical forms */
    else if (c < UCS4_SMPMATHMIN) ret = c;
    else if (c <= UCS4_SMPMATHMAX) {
      if (!(ret = ucs4_smpmathdecomptab[c - UCS4_SMPMATHMIN])) ret = c;
    }
				/* CJK compatibility ideographs in SIP */
    else if (!(ret = ((c >= UCS4_SIPMIN) && (c <= UCS4_SIPMAX)) ?
	       ucs4_sipdecomptab[c - UCS4_SIPMIN] : c)) ret = c;
  }
  return ret;
}

/* Return recursive decomposition of a UCS-4 character
 * Accepts: character or U8G_ERROR to return next from "more"
 *	    pointer to returned more
 * Returns: [next] decomposed value, more set if still more decomposition
 */

unsigned long ucs4_decompose_recursive (unsigned long c,void **more)
{
  unsigned long c1;
  void *m,*mn;
  RECURSIVEMORE *mr;
  if (c & U8G_ERROR) {		/* want to chase more? */
    mn = NIL;
    if ((mr = (RECURSIVEMORE *) *more) != NULL) switch (mr->more->type) {
    case MORESINGLE:		/* decompose single value */
      c = ucs4_decompose_recursive (mr->more->data.single,&mn);
      *more = mr->next;		/* done with this more, remove it */
      fs_give ((void **) &mr->more);
      fs_give ((void **) &mr);
      break;
    case MOREMULTIPLE:		/* decompose current value in multiple */
      c = ucs4_decompose_recursive (*mr->more->data.multiple.next++,&mn);
				/* if done with this multiple decomposition */
      if (!--mr->more->data.multiple.count) {
	*more = mr->next;	/* done with this more, remove it */
	fs_give ((void **) &mr->more);
	fs_give ((void **) &mr);
      }
      break;
    default:			/* uh-oh */
      fatal ("invalid more block argument to ucs4_decompose_recursive!");
    }
    else fatal ("no more block provided to ucs4_decompose_recursive!");
    if ((mr = mn) != NULL) {	/* did this value recurse on us? */
      mr->next = *more;		/* yes, insert new more at head */
      *more = mr;
    }
  }
  else {			/* start decomposition */
    *more = NIL;		/* initially set no more */
    mr = NIL;
    do {			/* repeatedly decompose this codepoint */
      c = ucs4_decompose (c1 = c,&m);
      if (m) {			/* multi-byte decomposition */
	if (c1 == c) fatal ("endless multiple decomposition!");
				/* create a block to stash this more */
	mr = memset (fs_get (sizeof (RECURSIVEMORE)),0,sizeof (RECURSIVEMORE));
	mr->more = m;		/* note the expansion */
	mr->next = *more;	/* old list is the tail */
	*more = mr;		/* and this is the new head */
      }
    } while (c1 != c);		/* until nothing more to decompose */
  }
  return c;
}

void *utf8_parameters (long function,void *value)
{
  void *ret;

  switch(function){
     case SET_UCS4WIDTH:
	ucs4width = (ucs4width_t) value;
     case GET_UCS4WIDTH:
	ret = (void *) ucs4width;
	break;
     default: ret = NULL;
	break;
  }
  return ret;
}