1: /*
2: * Copyright (C) 1999-2001, 2008 Free Software Foundation, Inc.
3: * This file is part of the GNU LIBICONV Library.
4: *
5: * The GNU LIBICONV Library is free software; you can redistribute it
6: * and/or modify it under the terms of the GNU Library General Public
7: * License as published by the Free Software Foundation; either version 2
8: * of the License, or (at your option) any later version.
9: *
10: * The GNU LIBICONV Library is distributed in the hope that it will be
11: * useful, but WITHOUT ANY WARRANTY; without even the implied warranty of
12: * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
13: * Library General Public License for more details.
14: *
15: * You should have received a copy of the GNU Library General Public
16: * License along with the GNU LIBICONV Library; see the file COPYING.LIB.
17: * If not, write to the Free Software Foundation, Inc., 51 Franklin Street,
18: * Fifth Floor, Boston, MA 02110-1301, USA.
19: */
20:
21: /*
22: * ISO-2022-JP-2
23: */
24:
25: /* Specification: RFC 1554 */
26: /* ESC '(' 'I' for JISX0201 Katakana is an extension not found in RFC 1554 or
27: CJK.INF, but implemented in glibc-2.1 and qt-2.0. */
28:
29: #define ESC 0x1b
30:
31: /*
32: * The state is composed of one of the following values
33: */
34: #define STATE_ASCII 0
35: #define STATE_JISX0201ROMAN 1
36: #define STATE_JISX0201KATAKANA 2
37: #define STATE_JISX0208 3
38: #define STATE_JISX0212 4
39: #define STATE_GB2312 5
40: #define STATE_KSC5601 6
41: /*
42: * and one of the following values, << 8
43: */
44: #define STATE_G2_NONE 0
45: #define STATE_G2_ISO8859_1 1
46: #define STATE_G2_ISO8859_7 2
47:
48: #define SPLIT_STATE \
49: unsigned int state1 = state & 0xff, state2 = state >> 8
50: #define COMBINE_STATE \
51: state = (state2 << 8) | state1
52:
53: static int
54: iso2022_jp2_mbtowc (conv_t conv, ucs4_t *pwc, const unsigned char *s, int n)
55: {
56: state_t state = conv->istate;
57: SPLIT_STATE;
58: int count = 0;
59: unsigned char c;
60: for (;;) {
61: c = *s;
62: if (c == ESC) {
63: if (n < count+3)
64: goto none;
65: if (s[1] == '(') {
66: if (s[2] == 'B') {
67: state1 = STATE_ASCII;
68: s += 3; count += 3;
69: if (n < count+1)
70: goto none;
71: continue;
72: }
73: if (s[2] == 'J') {
74: state1 = STATE_JISX0201ROMAN;
75: s += 3; count += 3;
76: if (n < count+1)
77: goto none;
78: continue;
79: }
80: if (s[2] == 'I') {
81: state1 = STATE_JISX0201KATAKANA;
82: s += 3; count += 3;
83: if (n < count+1)
84: goto none;
85: continue;
86: }
87: goto ilseq;
88: }
89: if (s[1] == '$') {
90: if (s[2] == '@' || s[2] == 'B') {
91: /* We don't distinguish JIS X 0208-1978 and JIS X 0208-1983. */
92: state1 = STATE_JISX0208;
93: s += 3; count += 3;
94: if (n < count+1)
95: goto none;
96: continue;
97: }
98: if (s[2] == 'A') {
99: state1 = STATE_GB2312;
100: s += 3; count += 3;
101: if (n < count+1)
102: goto none;
103: continue;
104: }
105: if (s[2] == '(') {
106: if (n < count+4)
107: goto none;
108: if (s[3] == 'D') {
109: state1 = STATE_JISX0212;
110: s += 4; count += 4;
111: if (n < count+1)
112: goto none;
113: continue;
114: }
115: if (s[3] == 'C') {
116: state1 = STATE_KSC5601;
117: s += 4; count += 4;
118: if (n < count+1)
119: goto none;
120: continue;
121: }
122: goto ilseq;
123: }
124: goto ilseq;
125: }
126: if (s[1] == '.') {
127: if (n < count+3)
128: goto none;
129: if (s[2] == 'A') {
130: state2 = STATE_G2_ISO8859_1;
131: s += 3; count += 3;
132: if (n < count+1)
133: goto none;
134: continue;
135: }
136: if (s[2] == 'F') {
137: state2 = STATE_G2_ISO8859_7;
138: s += 3; count += 3;
139: if (n < count+1)
140: goto none;
141: continue;
142: }
143: goto ilseq;
144: }
145: if (s[1] == 'N') {
146: switch (state2) {
147: case STATE_G2_NONE:
148: goto ilseq;
149: case STATE_G2_ISO8859_1:
150: if (s[2] < 0x80) {
151: unsigned char buf = s[2]+0x80;
152: int ret = iso8859_1_mbtowc(conv,pwc,&buf,1);
153: if (ret == RET_ILSEQ)
154: goto ilseq;
155: if (ret != 1) abort();
156: COMBINE_STATE;
157: conv->istate = state;
158: return count+3;
159: } else
160: goto ilseq;
161: case STATE_G2_ISO8859_7:
162: if (s[2] < 0x80) {
163: unsigned char buf = s[2]+0x80;
164: int ret = iso8859_7_mbtowc(conv,pwc,&buf,1);
165: if (ret == RET_ILSEQ)
166: goto ilseq;
167: if (ret != 1) abort();
168: COMBINE_STATE;
169: conv->istate = state;
170: return count+3;
171: } else
172: goto ilseq;
173: default: abort();
174: }
175: }
176: goto ilseq;
177: }
178: break;
179: }
180: switch (state1) {
181: case STATE_ASCII:
182: if (c < 0x80) {
183: int ret = ascii_mbtowc(conv,pwc,s,1);
184: if (ret == RET_ILSEQ)
185: goto ilseq;
186: if (ret != 1) abort();
187: if (*pwc == 0x000a || *pwc == 0x000d)
188: state2 = STATE_G2_NONE;
189: COMBINE_STATE;
190: conv->istate = state;
191: return count+1;
192: } else
193: goto ilseq;
194: case STATE_JISX0201ROMAN:
195: if (c < 0x80) {
196: int ret = jisx0201_mbtowc(conv,pwc,s,1);
197: if (ret == RET_ILSEQ)
198: goto ilseq;
199: if (ret != 1) abort();
200: if (*pwc == 0x000a || *pwc == 0x000d)
201: state2 = STATE_G2_NONE;
202: COMBINE_STATE;
203: conv->istate = state;
204: return count+1;
205: } else
206: goto ilseq;
207: case STATE_JISX0201KATAKANA:
208: if (c < 0x80) {
209: unsigned char buf = c+0x80;
210: int ret = jisx0201_mbtowc(conv,pwc,&buf,1);
211: if (ret == RET_ILSEQ)
212: goto ilseq;
213: if (ret != 1) abort();
214: COMBINE_STATE;
215: conv->istate = state;
216: return count+1;
217: } else
218: goto ilseq;
219: case STATE_JISX0208:
220: if (n < count+2)
221: goto none;
222: if (s[0] < 0x80 && s[1] < 0x80) {
223: int ret = jisx0208_mbtowc(conv,pwc,s,2);
224: if (ret == RET_ILSEQ)
225: goto ilseq;
226: if (ret != 2) abort();
227: COMBINE_STATE;
228: conv->istate = state;
229: return count+2;
230: } else
231: goto ilseq;
232: case STATE_JISX0212:
233: if (n < count+2)
234: goto none;
235: if (s[0] < 0x80 && s[1] < 0x80) {
236: int ret = jisx0212_mbtowc(conv,pwc,s,2);
237: if (ret == RET_ILSEQ)
238: goto ilseq;
239: if (ret != 2) abort();
240: COMBINE_STATE;
241: conv->istate = state;
242: return count+2;
243: } else
244: goto ilseq;
245: case STATE_GB2312:
246: if (n < count+2)
247: goto none;
248: if (s[0] < 0x80 && s[1] < 0x80) {
249: int ret = gb2312_mbtowc(conv,pwc,s,2);
250: if (ret == RET_ILSEQ)
251: goto ilseq;
252: if (ret != 2) abort();
253: COMBINE_STATE;
254: conv->istate = state;
255: return count+2;
256: } else
257: goto ilseq;
258: case STATE_KSC5601:
259: if (n < count+2)
260: goto none;
261: if (s[0] < 0x80 && s[1] < 0x80) {
262: int ret = ksc5601_mbtowc(conv,pwc,s,2);
263: if (ret == RET_ILSEQ)
264: goto ilseq;
265: if (ret != 2) abort();
266: COMBINE_STATE;
267: conv->istate = state;
268: return count+2;
269: } else
270: goto ilseq;
271: default: abort();
272: }
273:
274: none:
275: COMBINE_STATE;
276: conv->istate = state;
277: return RET_TOOFEW(count);
278:
279: ilseq:
280: COMBINE_STATE;
281: conv->istate = state;
282: return RET_SHIFT_ILSEQ(count);
283: }
284:
285: #undef COMBINE_STATE
286: #undef SPLIT_STATE
287:
288: /*
289: * The state can also contain one of the following values, << 16.
290: * Values >= STATE_TAG_LANGUAGE are temporary tag parsing states.
291: */
292: #define STATE_TAG_NONE 0
293: #define STATE_TAG_LANGUAGE 4
294: #define STATE_TAG_LANGUAGE_j 5
295: #define STATE_TAG_LANGUAGE_ja 1
296: #define STATE_TAG_LANGUAGE_k 6
297: #define STATE_TAG_LANGUAGE_ko 2
298: #define STATE_TAG_LANGUAGE_z 7
299: #define STATE_TAG_LANGUAGE_zh 3
300:
301: #define SPLIT_STATE \
302: unsigned int state1 = state & 0xff, state2 = (state >> 8) & 0xff, state3 = state >> 16
303: #define COMBINE_STATE \
304: state = (state3 << 16) | (state2 << 8) | state1
305:
306: static int
307: iso2022_jp2_wctomb (conv_t conv, unsigned char *r, ucs4_t wc, int n)
308: {
309: state_t state = conv->ostate;
310: SPLIT_STATE;
311: unsigned char buf[2];
312: int ret;
313: /* This defines the conversion preferences depending on the current
314: langauge tag. */
315: enum conversion { none = 0, european, japanese, chinese, korean, other };
316: static const unsigned int conversion_lists[STATE_TAG_LANGUAGE] = {
317: /* STATE_TAG_NONE */
318: japanese + (european << 3) + (chinese << 6) + (korean << 9) + (other << 12),
319: /* STATE_TAG_LANGUAGE_ja */
320: japanese + (european << 3) + (chinese << 6) + (korean << 9) + (other << 12),
321: /* STATE_TAG_LANGUAGE_ko */
322: korean + (european << 3) + (japanese << 6) + (chinese << 9) + (other << 12),
323: /* STATE_TAG_LANGUAGE_zh */
324: chinese + (european << 3) + (japanese << 6) + (korean << 9) + (other << 12)
325: };
326: unsigned int conversion_list;
327:
328: /* Handle Unicode tag characters (range U+E0000..U+E007F). */
329: if ((wc >> 7) == (0xe0000 >> 7)) {
330: char c = wc & 0x7f;
331: if (c >= 'A' && c <= 'Z')
332: c += 'a'-'A';
333: switch (c) {
334: case 0x01:
335: state3 = STATE_TAG_LANGUAGE;
336: COMBINE_STATE;
337: conv->ostate = state;
338: return 0;
339: case 'j':
340: if (state3 == STATE_TAG_LANGUAGE) {
341: state3 = STATE_TAG_LANGUAGE_j;
342: COMBINE_STATE;
343: conv->ostate = state;
344: return 0;
345: }
346: break;
347: case 'a':
348: if (state3 == STATE_TAG_LANGUAGE_j) {
349: state3 = STATE_TAG_LANGUAGE_ja;
350: COMBINE_STATE;
351: conv->ostate = state;
352: return 0;
353: }
354: break;
355: case 'k':
356: if (state3 == STATE_TAG_LANGUAGE) {
357: state3 = STATE_TAG_LANGUAGE_k;
358: COMBINE_STATE;
359: conv->ostate = state;
360: return 0;
361: }
362: break;
363: case 'o':
364: if (state3 == STATE_TAG_LANGUAGE_k) {
365: state3 = STATE_TAG_LANGUAGE_ko;
366: COMBINE_STATE;
367: conv->ostate = state;
368: return 0;
369: }
370: break;
371: case 'z':
372: if (state3 == STATE_TAG_LANGUAGE) {
373: state3 = STATE_TAG_LANGUAGE_z;
374: COMBINE_STATE;
375: conv->ostate = state;
376: return 0;
377: }
378: break;
379: case 'h':
380: if (state3 == STATE_TAG_LANGUAGE_z) {
381: state3 = STATE_TAG_LANGUAGE_zh;
382: COMBINE_STATE;
383: conv->ostate = state;
384: return 0;
385: }
386: break;
387: case 0x7f:
388: state3 = STATE_TAG_NONE;
389: COMBINE_STATE;
390: conv->ostate = state;
391: return 0;
392: default:
393: break;
394: }
395: /* Other tag characters reset the tag parsing state or are ignored. */
396: if (state3 >= STATE_TAG_LANGUAGE)
397: state3 = STATE_TAG_NONE;
398: COMBINE_STATE;
399: conv->ostate = state;
400: return 0;
401: }
402: if (state3 >= STATE_TAG_LANGUAGE)
403: state3 = STATE_TAG_NONE;
404:
405: /* Try ASCII. */
406: ret = ascii_wctomb(conv,buf,wc,1);
407: if (ret != RET_ILUNI) {
408: if (ret != 1) abort();
409: if (buf[0] < 0x80) {
410: int count = (state1 == STATE_ASCII ? 1 : 4);
411: if (n < count)
412: return RET_TOOSMALL;
413: if (state1 != STATE_ASCII) {
414: r[0] = ESC;
415: r[1] = '(';
416: r[2] = 'B';
417: r += 3;
418: state1 = STATE_ASCII;
419: }
420: r[0] = buf[0];
421: if (wc == 0x000a || wc == 0x000d)
422: state2 = STATE_G2_NONE;
423: COMBINE_STATE;
424: conv->ostate = state;
425: return count;
426: }
427: }
428:
429: conversion_list = conversion_lists[state3];
430:
431: do {
432: switch (conversion_list & ((1 << 3) - 1)) {
433:
434: case european:
435:
436: /* Try ISO-8859-1. */
437: ret = iso8859_1_wctomb(conv,buf,wc,1);
438: if (ret != RET_ILUNI) {
439: if (ret != 1) abort();
440: if (buf[0] >= 0x80) {
441: int count = (state2 == STATE_G2_ISO8859_1 ? 3 : 6);
442: if (n < count)
443: return RET_TOOSMALL;
444: if (state2 != STATE_G2_ISO8859_1) {
445: r[0] = ESC;
446: r[1] = '.';
447: r[2] = 'A';
448: r += 3;
449: state2 = STATE_G2_ISO8859_1;
450: }
451: r[0] = ESC;
452: r[1] = 'N';
453: r[2] = buf[0]-0x80;
454: COMBINE_STATE;
455: conv->ostate = state;
456: return count;
457: }
458: }
459:
460: /* Try ISO-8859-7. */
461: ret = iso8859_7_wctomb(conv,buf,wc,1);
462: if (ret != RET_ILUNI) {
463: if (ret != 1) abort();
464: if (buf[0] >= 0x80) {
465: int count = (state2 == STATE_G2_ISO8859_7 ? 3 : 6);
466: if (n < count)
467: return RET_TOOSMALL;
468: if (state2 != STATE_G2_ISO8859_7) {
469: r[0] = ESC;
470: r[1] = '.';
471: r[2] = 'F';
472: r += 3;
473: state2 = STATE_G2_ISO8859_7;
474: }
475: r[0] = ESC;
476: r[1] = 'N';
477: r[2] = buf[0]-0x80;
478: COMBINE_STATE;
479: conv->ostate = state;
480: return count;
481: }
482: }
483:
484: break;
485:
486: case japanese:
487:
488: /* Try JIS X 0201-1976 Roman. */
489: ret = jisx0201_wctomb(conv,buf,wc,1);
490: if (ret != RET_ILUNI) {
491: if (ret != 1) abort();
492: if (buf[0] < 0x80) {
493: int count = (state1 == STATE_JISX0201ROMAN ? 1 : 4);
494: if (n < count)
495: return RET_TOOSMALL;
496: if (state1 != STATE_JISX0201ROMAN) {
497: r[0] = ESC;
498: r[1] = '(';
499: r[2] = 'J';
500: r += 3;
501: state1 = STATE_JISX0201ROMAN;
502: }
503: r[0] = buf[0];
504: if (wc == 0x000a || wc == 0x000d)
505: state2 = STATE_G2_NONE;
506: COMBINE_STATE;
507: conv->ostate = state;
508: return count;
509: }
510: }
511:
512: /* Try JIS X 0208-1990 in place of JIS X 0208-1978 and
513: JIS X 0208-1983. */
514: ret = jisx0208_wctomb(conv,buf,wc,2);
515: if (ret != RET_ILUNI) {
516: if (ret != 2) abort();
517: if (buf[0] < 0x80 && buf[1] < 0x80) {
518: int count = (state1 == STATE_JISX0208 ? 2 : 5);
519: if (n < count)
520: return RET_TOOSMALL;
521: if (state1 != STATE_JISX0208) {
522: r[0] = ESC;
523: r[1] = '$';
524: r[2] = 'B';
525: r += 3;
526: state1 = STATE_JISX0208;
527: }
528: r[0] = buf[0];
529: r[1] = buf[1];
530: COMBINE_STATE;
531: conv->ostate = state;
532: return count;
533: }
534: }
535:
536: /* Try JIS X 0212-1990. */
537: ret = jisx0212_wctomb(conv,buf,wc,2);
538: if (ret != RET_ILUNI) {
539: if (ret != 2) abort();
540: if (buf[0] < 0x80 && buf[1] < 0x80) {
541: int count = (state1 == STATE_JISX0212 ? 2 : 6);
542: if (n < count)
543: return RET_TOOSMALL;
544: if (state1 != STATE_JISX0212) {
545: r[0] = ESC;
546: r[1] = '$';
547: r[2] = '(';
548: r[3] = 'D';
549: r += 4;
550: state1 = STATE_JISX0212;
551: }
552: r[0] = buf[0];
553: r[1] = buf[1];
554: COMBINE_STATE;
555: conv->ostate = state;
556: return count;
557: }
558: }
559:
560: break;
561:
562: case chinese:
563:
564: /* Try GB 2312-1980. */
565: ret = gb2312_wctomb(conv,buf,wc,2);
566: if (ret != RET_ILUNI) {
567: if (ret != 2) abort();
568: if (buf[0] < 0x80 && buf[1] < 0x80) {
569: int count = (state1 == STATE_GB2312 ? 2 : 5);
570: if (n < count)
571: return RET_TOOSMALL;
572: if (state1 != STATE_GB2312) {
573: r[0] = ESC;
574: r[1] = '$';
575: r[2] = 'A';
576: r += 3;
577: state1 = STATE_GB2312;
578: }
579: r[0] = buf[0];
580: r[1] = buf[1];
581: COMBINE_STATE;
582: conv->ostate = state;
583: return count;
584: }
585: }
586:
587: break;
588:
589: case korean:
590:
591: /* Try KS C 5601-1992. */
592: ret = ksc5601_wctomb(conv,buf,wc,2);
593: if (ret != RET_ILUNI) {
594: if (ret != 2) abort();
595: if (buf[0] < 0x80 && buf[1] < 0x80) {
596: int count = (state1 == STATE_KSC5601 ? 2 : 6);
597: if (n < count)
598: return RET_TOOSMALL;
599: if (state1 != STATE_KSC5601) {
600: r[0] = ESC;
601: r[1] = '$';
602: r[2] = '(';
603: r[3] = 'C';
604: r += 4;
605: state1 = STATE_KSC5601;
606: }
607: r[0] = buf[0];
608: r[1] = buf[1];
609: COMBINE_STATE;
610: conv->ostate = state;
611: return count;
612: }
613: }
614:
615: break;
616:
617: case other:
618:
619: /* Try JIS X 0201-1976 Kana. This is not officially part of
620: ISO-2022-JP-2, according to RFC 1554. Therefore we try this
621: only after all other attempts. */
622: ret = jisx0201_wctomb(conv,buf,wc,1);
623: if (ret != RET_ILUNI) {
624: if (ret != 1) abort();
625: if (buf[0] >= 0x80) {
626: int count = (state1 == STATE_JISX0201KATAKANA ? 1 : 4);
627: if (n < count)
628: return RET_TOOSMALL;
629: if (state1 != STATE_JISX0201KATAKANA) {
630: r[0] = ESC;
631: r[1] = '(';
632: r[2] = 'I';
633: r += 3;
634: state1 = STATE_JISX0201KATAKANA;
635: }
636: r[0] = buf[0]-0x80;
637: COMBINE_STATE;
638: conv->ostate = state;
639: return count;
640: }
641: }
642:
643: break;
644:
645: default:
646: abort();
647: }
648:
649: conversion_list = conversion_list >> 3;
650: } while (conversion_list != 0);
651:
652: return RET_ILUNI;
653: }
654:
655: static int
656: iso2022_jp2_reset (conv_t conv, unsigned char *r, int n)
657: {
658: state_t state = conv->ostate;
659: SPLIT_STATE;
660: (void)state2;
661: (void)state3;
662: if (state1 != STATE_ASCII) {
663: if (n < 3)
664: return RET_TOOSMALL;
665: r[0] = ESC;
666: r[1] = '(';
667: r[2] = 'B';
668: /* conv->ostate = 0; will be done by the caller */
669: return 3;
670: } else
671: return 0;
672: }
673:
674: #undef COMBINE_STATE
675: #undef SPLIT_STATE
676: #undef STATE_TAG_LANGUAGE_zh
677: #undef STATE_TAG_LANGUAGE_z
678: #undef STATE_TAG_LANGUAGE_ko
679: #undef STATE_TAG_LANGUAGE_k
680: #undef STATE_TAG_LANGUAGE_ja
681: #undef STATE_TAG_LANGUAGE_j
682: #undef STATE_TAG_LANGUAGE
683: #undef STATE_TAG_NONE
684: #undef STATE_G2_ISO8859_7
685: #undef STATE_G2_ISO8859_1
686: #undef STATE_G2_NONE
687: #undef STATE_KSC5601
688: #undef STATE_GB2312
689: #undef STATE_JISX0212
690: #undef STATE_JISX0208
691: #undef STATE_JISX0201KATAKANA
692: #undef STATE_JISX0201ROMAN
693: #undef STATE_ASCII
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>
![[BACK]](/icons/cvsweb/back.gif){kind=icon}
Return to iso2022_jp2.h CVS log ![[TXT]](/icons/cvsweb/text.gif){kind=icon}
![[DIR]](/icons/cvsweb/dir.gif){kind=icon}
Up to [ELWIX - Embedded LightWeight unIX -] / embedaddon / libiconv / lib