2 * iso2022.c - support for ISO/IEC 2022 (alias ECMA-35).
4 * This isn't a complete implementation of ISO/IEC 2022, but it's
5 * close. It only handles decoding, because a fully general encoder
6 * isn't really useful. It can decode 8-bit and 7-bit versions, with
7 * support for single-byte and multi-byte character sets, all four
8 * containers (G0, G1, G2, and G3), using both single-shift and
9 * locking-shift sequences.
11 * The general principle is that any valid ISO/IEC 2022 sequence
12 * should either be correctly decoded or should emit an ERROR. The
13 * only exception to this is that the C0 and C1 sets are fixed as
14 * those of ISO/IEC 6429. Escape sequences for designating control
15 * sets are passed through, so a post-processor could fix them up if
18 * DOCS to UTF-8 works. Other DOCS sequences are ignored, which will
19 * produce surprising results.
36 enum {S4
, S6
, M4
, M6
};
38 static long int emacs_big5_1_to_unicode(int, int);
39 static long int emacs_big5_2_to_unicode(int, int);
40 static long int cns11643_1_to_unicode(int, int);
41 static long int cns11643_2_to_unicode(int, int);
42 static long int cns11643_3_to_unicode(int, int);
43 static long int cns11643_4_to_unicode(int, int);
44 static long int cns11643_5_to_unicode(int, int);
45 static long int cns11643_6_to_unicode(int, int);
46 static long int cns11643_7_to_unicode(int, int);
47 static long int null_dbcs_to_unicode(int, int);
49 const struct iso2022_subcharset
{
52 const sbcs_data
*sbcs_base
;
53 long int (*dbcs_fn
)(int, int);
54 } iso2022_subcharsets
[] = {
55 { S4
, 0, '0', 0x00, &sbcsdata_CS_DEC_GRAPHICS
},
56 { S4
, 0, '<', 0x80, &sbcsdata_CS_DEC_MCS
},
57 { S4
, 0, 'A', 0x00, &sbcsdata_CS_BS4730
},
58 { S4
, 0, 'B', 0x00, &sbcsdata_CS_ASCII
},
59 { S4
, 0, 'I', 0x80, &sbcsdata_CS_JISX0201
},
60 { S4
, 0, 'J', 0x00, &sbcsdata_CS_JISX0201
},
62 { S6
, 0, 'A', 0x80, &sbcsdata_CS_ISO8859_1
},
63 { S6
, 0, 'B', 0x80, &sbcsdata_CS_ISO8859_2
},
64 { S6
, 0, 'C', 0x80, &sbcsdata_CS_ISO8859_3
},
65 { S6
, 0, 'D', 0x80, &sbcsdata_CS_ISO8859_4
},
66 { S6
, 0, 'F', 0x80, &sbcsdata_CS_ISO8859_7
},
67 { S6
, 0, 'G', 0x80, &sbcsdata_CS_ISO8859_6
},
68 { S6
, 0, 'H', 0x80, &sbcsdata_CS_ISO8859_8
},
69 { S6
, 0, 'L', 0x80, &sbcsdata_CS_ISO8859_5
},
70 { S6
, 0, 'M', 0x80, &sbcsdata_CS_ISO8859_9
},
71 { S6
, 0, 'T', 0x80, &sbcsdata_CS_ISO8859_11
},
72 { S6
, 0, 'V', 0x80, &sbcsdata_CS_ISO8859_10
},
73 { S6
, 0, 'Y', 0x80, &sbcsdata_CS_ISO8859_13
},
74 { S6
, 0, '_', 0x80, &sbcsdata_CS_ISO8859_14
},
75 { S6
, 0, 'b', 0x80, &sbcsdata_CS_ISO8859_15
},
76 { S6
, 0, 'f', 0x80, &sbcsdata_CS_ISO8859_16
},
77 { S6
, 0, '~' }, /* empty 96-set */
79 { M4
, 0, '@' }, /* JIS C 6226-1978 */
81 { M4
, 0, '0', -0x21, 0, &emacs_big5_1_to_unicode
},
82 { M4
, 0, '1', -0x21, 0, &emacs_big5_2_to_unicode
},
83 { M4
, 0, 'A', -0x21, 0, &gb2312_to_unicode
},
84 { M4
, 0, 'B', -0x21, 0, &jisx0208_to_unicode
},
85 { M4
, 0, 'C', -0x21, 0, &ksx1001_to_unicode
},
86 { M4
, 0, 'D', -0x21, 0, &jisx0212_to_unicode
},
87 { M4
, 0, 'G', -0x21, 0, &cns11643_1_to_unicode
},
88 { M4
, 0, 'H', -0x21, 0, &cns11643_2_to_unicode
},
89 { M4
, 0, 'I', -0x21, 0, &cns11643_3_to_unicode
},
90 { M4
, 0, 'J', -0x21, 0, &cns11643_4_to_unicode
},
91 { M4
, 0, 'K', -0x21, 0, &cns11643_5_to_unicode
},
92 { M4
, 0, 'L', -0x21, 0, &cns11643_6_to_unicode
},
93 { M4
, 0, 'M', -0x21, 0, &cns11643_7_to_unicode
},
94 { M4
, 0, '~', 0, 0, &null_dbcs_to_unicode
}, /* empty 94^n-set */
95 { M6
, 0, '~', 0, 0, &null_dbcs_to_unicode
}, /* empty 96^n-set */
98 static long int null_dbcs_to_unicode(int r
, int c
)
104 * Emacs encodes Big5 in COMPOUND_TEXT as two 94x94 character sets.
105 * We treat Big5 as a 94x191 character set with a bunch of undefined
106 * columns in the middle, so we have to mess around a bit to make
110 static long int emacs_big5_1_to_unicode(int r
, int c
)
116 if (c
>= 64) c
+= 34; /* Skip over the gap */
117 return big5_to_unicode(r
, c
);
120 static long int emacs_big5_2_to_unicode(int r
, int c
)
126 if (c
>= 64) c
+= 34; /* Skip over the gap */
127 return big5_to_unicode(r
, c
);
130 /* Wrappers for cns11643_to_unicode() */
131 static long int cns11643_1_to_unicode(int r
, int c
)
133 return cns11643_to_unicode(0, r
, c
);
135 static long int cns11643_2_to_unicode(int r
, int c
)
137 return cns11643_to_unicode(1, r
, c
);
139 static long int cns11643_3_to_unicode(int r
, int c
)
141 return cns11643_to_unicode(2, r
, c
);
143 static long int cns11643_4_to_unicode(int r
, int c
)
145 return cns11643_to_unicode(3, r
, c
);
147 static long int cns11643_5_to_unicode(int r
, int c
)
149 return cns11643_to_unicode(4, r
, c
);
151 static long int cns11643_6_to_unicode(int r
, int c
)
153 return cns11643_to_unicode(5, r
, c
);
155 static long int cns11643_7_to_unicode(int r
, int c
)
157 return cns11643_to_unicode(6, r
, c
);
160 /* States, or "what we're currently accumulating". */
162 IDLE
, /* None of the below */
163 SS2CHAR
, /* Accumulating a character after SS2 */
164 SS3CHAR
, /* Accumulating a character after SS3 */
165 ESCSEQ
, /* Accumulating an escape sequence */
166 ESCDROP
, /* Discarding an escape sequence */
167 ESCPASS
, /* Passing through an escape sequence */
168 DOCSUTF8
, /* DOCSed into UTF-8 */
169 DOCSCTEXT
/* DOCSed into a COMPOUND_TEXT extended segment */
174 static void dump_state(charset_state
*s
)
176 unsigned s0
= s
->s0
, s1
= s
->s1
;
177 char const * const modes
[] = { "IDLE", "SS2CHAR", "SS3CHAR",
178 "ESCSEQ", "ESCDROP", "ESCPASS",
181 fprintf(stderr
, "s0: %s", modes
[s0
>> 29]);
182 fprintf(stderr
, " %02x %02x %02x ", (s0
>> 16) & 0xff, (s0
>> 8) & 0xff,
184 fprintf(stderr
, "s1: LS%d LS%dR", (s1
>> 30) & 3, (s1
>> 28) & 3);
185 fprintf(stderr
, " %d %d %d %d\n", s1
& 0x7f, (s1
>> 7) & 0x7f,
186 (s1
>> 14) & 0x7f, (s1
>> 21) & 0x7f);
190 static void designate(charset_state
*state
, int container
,
191 int type
, int ibyte
, int fbyte
)
195 assert(container
>= 0 && container
<= 3);
196 assert(type
== S4
|| type
== S6
|| type
== M4
|| type
== M6
);
198 for (i
= 0; i
<= lenof(iso2022_subcharsets
); i
++) {
199 if (iso2022_subcharsets
[i
].type
== type
&&
200 iso2022_subcharsets
[i
].i
== ibyte
&&
201 iso2022_subcharsets
[i
].f
== fbyte
) {
202 state
->s1
&= ~(0x7fL
<< (container
* 7));
203 state
->s1
|= (i
<< (container
* 7));
208 * If we don't find the charset, invoke the empty one, so we
209 * output ERROR rather than garbage.
211 designate(state
, container
, type
, 0, '~');
214 static void do_utf8(long int input_chr
,
215 charset_state
*state
,
216 void (*emit
)(void *ctx
, long int output
),
219 charset_state ustate
;
220 charset_spec
const *utf8
;
223 ustate
.s0
= state
->s0
& 0x03ffffffL
;
224 read_utf8(NULL
, input_chr
, &ustate
, emit
, emitctx
);
225 state
->s0
= (state
->s0
& ~0x03ffffffL
) | (ustate
.s0
& 0x03ffffffL
);
228 static void docs_utf8(long int input_chr
,
229 charset_state
*state
,
230 void (*emit
)(void *ctx
, long int output
),
236 * Bits [25:0] of s0 are reserved for read_utf8().
237 * Bits [27:26] are a tiny state machine to recognise ESC % @.
239 retstate
= (state
->s0
& 0x0c000000L
) >> 26;
240 if (retstate
== 1 && input_chr
== '%')
242 else if (retstate
== 2 && input_chr
== '@') {
243 /* If we've got a partial UTF-8 sequence, complain. */
244 if (state
->s0
& 0x03ffffffL
)
245 emit(emitctx
, ERROR
);
249 if (retstate
>= 1) do_utf8(ESC
, state
, emit
, emitctx
);
250 if (retstate
>= 2) do_utf8('%', state
, emit
, emitctx
);
252 if (input_chr
== ESC
)
255 do_utf8(input_chr
, state
, emit
, emitctx
);
258 state
->s0
= (state
->s0
& ~0x0c000000L
) | (retstate
<< 26);
261 struct ctext_encoding
{
263 charset_spec
const *subcs
;
267 * In theory, this list is in <http://ftp.x.org/pub/docs/registry>,
268 * but XLib appears to have its own ideas, and encodes these three
272 extern charset_spec
const charset_CS_ISO8859_14
;
273 extern charset_spec
const charset_CS_ISO8859_15
;
274 extern charset_spec
const charset_CS_BIG5
;
276 static struct ctext_encoding
const ctext_encodings
[] = {
277 { "big5-0\2", &charset_CS_BIG5
},
278 { "iso8859-14\2", &charset_CS_ISO8859_14
},
279 { "iso8859-15\2", &charset_CS_ISO8859_15
}
282 static void docs_ctext(long int input_chr
,
283 charset_state
*state
,
284 void (*emit
)(void *ctx
, long int output
),
288 * s0[27:26] = first entry in ctext_encodings that matches
289 * s0[25:22] = number of characters successfully matched, 0xf if all
290 * s0[21:8] count the number of octets left in the segment
291 * s0[7:0] are for sub-charset use
293 int n
= (state
->s0
>> 22) & 0xf, i
= (state
->s0
>> 26) & 3, oi
= i
, j
;
294 int length
= (state
->s0
>> 8) & 0x3fff;
297 /* Haven't read length yet */
298 if ((state
->s0
& 0xff) == 0)
299 /* ... or even the first byte */
300 state
->s0
|= input_chr
;
302 length
= (state
->s0
& 0x7f) * 0x80 + (input_chr
& 0x7f);
306 state
->s0
= (state
->s0
& 0xf0000000) | (length
<< 8);
313 /* Skipping unknown encoding. Look out for STX. */
315 state
->s0
= (state
->s0
& 0xf0000000) | (i
<< 26) | (0xf << 22);
316 } else if (n
!= 0xf) {
317 while (j
< lenof(ctext_encodings
) &&
318 !memcmp(ctext_encodings
[j
].name
,
319 ctext_encodings
[oi
].name
, n
)) {
320 if (ctext_encodings
[j
].name
[n
] < input_chr
)
325 if (i
>= lenof(ctext_encodings
) ||
326 memcmp(ctext_encodings
[i
].name
,
327 ctext_encodings
[oi
].name
, n
) ||
328 ctext_encodings
[i
].name
[n
] != input_chr
) {
329 /* Doom! We haven't heard of this encoding */
330 i
= lenof(ctext_encodings
);
334 * Otherwise, we have found an additional character in our
335 * encoding name. See if we have reached the _end_ of our
339 if (!ctext_encodings
[i
].name
[n
])
343 * Failing _that_, we simply update our encoding-name-
346 assert(i
< 4 && n
< 16);
347 state
->s0
= (state
->s0
& 0xf0000000) | (i
<< 26) | (n
<< 22);
349 if (i
>= lenof(ctext_encodings
))
350 emit(emitctx
, ERROR
);
352 charset_state substate
;
353 charset_spec
const *subcs
= ctext_encodings
[i
].subcs
;
355 substate
.s0
= state
->s0
& 0xff;
356 subcs
->read(subcs
, input_chr
, &substate
, emit
, emitctx
);
357 state
->s0
= (state
->s0
& ~0xff) | (substate
.s0
& 0xff);
363 state
->s0
= (state
->s0
&~0x003fff00) | (length
<< 8);
366 static void read_iso2022(charset_spec
const *charset
, long int input_chr
,
367 charset_state
*state
,
368 void (*emit
)(void *ctx
, long int output
),
372 /* dump_state(state); */
374 * We have to make fairly efficient use of the 64 bits of state
375 * available to us. Long-term state goes in s1, and consists of
376 * the identities of the character sets designated as G0/G1/G2/G3
377 * and the locking-shift states for GL and GR. Short-term state
378 * goes in s0: The bottom half of s0 accumulates characters for an
379 * escape sequence or a multi-byte character, while the top three
380 * bits indicate what they're being accumulated for. After DOCS,
381 * the bottom 29 bits of state are available for the DOCS function
382 * to use -- the UTF-8 one uses the bottom 26 for UTF-8 decoding
383 * and the top two to recognised ESC % @.
385 * s0[31:29] = state enum
386 * s0[24:0] = accumulated bytes
387 * s1[31:30] = GL locking-shift state
388 * s1[29:28] = GR locking-shift state
389 * s1[27:21] = G3 charset
390 * s1[20:14] = G2 charset
391 * s1[13:7] = G1 charset
392 * s1[6:0] = G0 charset
397 #define LOCKING_SHIFT(n,side) \
398 (state->s1 = (state->s1 & ~(3L<<(side))) | ((n ## L)<<(side)))
399 #define MODE ((state->s0 & 0xe0000000L) >> 29)
400 #define ENTER_MODE(m) (state->s0 = (state->s0 & ~0xe0000000L) | ((m)<<29))
401 #define SINGLE_SHIFT(n) ENTER_MODE(SS2CHAR - 2 + (n))
402 #define ASSERT_IDLE do { \
403 if (state->s0 != 0) emit(emitctx, ERROR); \
407 if (state
->s1
== 0) {
409 * Since there's no LS0R, this means we must just have started.
410 * Set up a sane initial state (LS0, LS1R, ASCII in G0/G1/G2/G3).
412 LOCKING_SHIFT(0, LEFT
);
413 LOCKING_SHIFT(1, RIGHT
);
414 designate(state
, 0, S4
, 0, 'B');
415 designate(state
, 1, S4
, 0, 'B');
416 designate(state
, 2, S4
, 0, 'B');
417 designate(state
, 3, S4
, 0, 'B');
420 if (MODE
== DOCSUTF8
) {
421 docs_utf8(input_chr
, state
, emit
, emitctx
);
424 if (MODE
== DOCSCTEXT
) {
425 docs_ctext(input_chr
, state
, emit
, emitctx
);
429 if ((input_chr
& 0x60) == 0x00) {
430 /* C0 or C1 control */
437 LOCKING_SHIFT(0, LEFT
);
440 LOCKING_SHIFT(1, LEFT
);
449 emit(emitctx
, input_chr
);
452 } else if ((input_chr
& 0x80) || MODE
< ESCSEQ
) {
454 struct iso2022_subcharset
const *subcs
;
459 * Force idle state if we're in mid escape sequence, or in a
460 * multi-byte character with a different top bit.
462 if (MODE
>= ESCSEQ
||
463 ((state
->s0
& 0x00ff0000L
) != 0 &&
464 (((state
->s0
>> 16) ^ input_chr
) & 0x80)))
466 if (MODE
== SS2CHAR
|| MODE
== SS3CHAR
) /* Single-shift */
467 container
= MODE
- SS2CHAR
+ 2;
468 else if (input_chr
>= 0x80) /* GR */
469 container
= (state
->s1
>> 28) & 3;
471 container
= state
->s1
>> 30;
474 input_7bit
= input_chr
& ~0x80;
475 subcs
= &iso2022_subcharsets
[(state
->s1
>> (container
* 7)) & 0x7f];
476 if ((subcs
->type
== S4
|| subcs
->type
== M4
) &&
477 (input_7bit
== 0x20 || input_7bit
== 0x7f)) {
478 /* characters not in 94-char set */
479 if (is_gl
) emit(emitctx
, input_7bit
);
480 else emit(emitctx
, ERROR
);
481 } else if (subcs
->type
== M4
|| subcs
->type
== M6
) {
482 if ((state
->s0
& 0x00ff0000L
) == 0) {
483 state
->s0
|= input_chr
<< 16;
487 subcs
->dbcs_fn(((state
->s0
>> 16) & 0x7f) + subcs
->offset
,
488 input_7bit
+ subcs
->offset
));
491 if ((state
->s0
& 0x00ff0000L
) != 0)
492 emit(emitctx
, ERROR
);
493 emit(emitctx
, subcs
->sbcs_base ?
494 sbcs_to_unicode(subcs
->sbcs_base
, input_7bit
+ subcs
->offset
):
500 if (MODE
== ESCPASS
) {
501 emit(emitctx
, input_chr
);
502 if ((input_chr
& 0xf0) != 0x20)
508 * Intermediate bytes shall be any of the 16 positions of
509 * column 02 of the code table; they are denoted by the symbol
512 if ((input_chr
& 0xf0) == 0x20) {
513 if (((state
->s0
>> 16) & 0xff) == 0)
514 state
->s0
|= input_chr
<< 16;
515 else if (((state
->s0
>> 8) & 0xff) == 0)
516 state
->s0
|= input_chr
<< 8;
518 /* Long escape sequence. Switch to ESCPASS or ESCDROP. */
519 i1
= (state
->s0
>> 16) & 0xff;
520 i2
= (state
->s0
>> 8) & 0xff;
522 case '(': case ')': case '*': case '+':
523 case '-': case '.': case '/':
531 emit(emitctx
, input_chr
);
541 * Final bytes shall be any of the 79 positions of columns 03
542 * to 07 of the code table excluding position 07/15; they are
543 * denoted by the symbol F.
545 i1
= (state
->s0
>> 16) & 0xff;
546 i2
= (state
->s0
>> 8) & 0xff;
548 input_chr
= 0; /* Make sure it won't match. */
551 case 0: /* No intermediate bytes */
560 LOCKING_SHIFT(2, LEFT
);
563 LOCKING_SHIFT(3, LEFT
);
566 LOCKING_SHIFT(3, RIGHT
);
569 LOCKING_SHIFT(2, RIGHT
);
572 LOCKING_SHIFT(1, RIGHT
);
575 /* Unsupported escape sequence. Spit it back out. */
577 emit(emitctx
, input_chr
);
582 * Various coding structure facilities specify that designating
583 * a code element also invokes it. As far as I can see, invoking
584 * it now will have the same practical effect, since those
585 * facilities also ban the use of locking shifts.
588 case 'A': /* G0 element used and invoked into GL */
589 LOCKING_SHIFT(0, LEFT
);
591 case 'C': /* G0 in GL, G1 in GR */
592 case 'D': /* Ditto, at least for 8-bit codes */
593 case 'L': /* ISO 4873 (ECMA-43) level 1 */
594 case 'M': /* ISO 4873 (ECMA-43) level 2 */
595 LOCKING_SHIFT(0, LEFT
);
596 LOCKING_SHIFT(1, RIGHT
);
602 * IRR (Identify Revised Registration) is ignored here,
603 * since any revised registration must be
604 * upward-compatible with the old one, so either we'll
605 * support the new one or we'll emit ERROR when we run
606 * into a new character. In either case, there's nothing
610 case '(': /* GZD4 */ case ')': /* G1D4 */
611 case '*': /* G2D4 */ case '+': /* G3D4 */
612 designate(state
, i1
- '(', S4
, i2
, input_chr
);
614 case '-': /* G1D6 */ case '.': /* G2D6 */ case '/': /* G3D6 */
615 designate(state
, i1
- ',', S6
, i2
, input_chr
);
617 case '$': /* G?DM? */
619 case 0: /* Obsolete version of GZDM4 */
621 case '(': /* GZDM4 */ case ')': /* G1DM4 */
622 case '*': /* G2DM4 */ case '+': /* G3DM4 */
623 designate(state
, i2
- '(', M4
, 0, input_chr
);
625 case '-': /* G1DM6 */
626 case '.': /* G2DM6 */ case '/': /* G3DM6 */
627 designate(state
, i2
- ',', M6
, 0, input_chr
);
630 emit(emitctx
, ERROR
);
634 /* XXX What's a reasonable way to handle an unrecognised DOCS? */
639 ENTER_MODE(DOCSUTF8
);
646 ENTER_MODE(DOCSCTEXT
);
653 /* Unsupported nF escape sequence. Re-emit it. */
656 if (i2
) emit(emitctx
, i2
);
657 emit(emitctx
, input_chr
);
663 static int write_iso2022(charset_spec
const *charset
, long int input_chr
,
664 charset_state
*state
,
665 void (*emit
)(void *ctx
, long int output
),
671 const charset_spec charset_CS_ISO2022
= {
672 CS_ISO2022
, read_iso2022
, write_iso2022
, NULL
683 void iso2022_emit(void *ctx
, long output
)
685 wchar_t **p
= (wchar_t **)ctx
;
689 void iso2022_read_test(int line
, char *input
, int inlen
, ...)
692 wchar_t *p
, str
[512];
697 state
.s0
= state
.s1
= 0;
700 for (i
= 0; i
< inlen
; i
++)
701 read_iso2022(NULL
, input
[i
] & 0xFF, &state
, iso2022_emit
, &p
);
705 for (i
= 0; i
< p
- str
; i
++) {
706 l
= va_arg(ap
, long int);
708 printf("%d: correct string shorter than output\n", line
);
713 printf("%d: char %d came out as %08x, should be %08lx\n",
719 l
= va_arg(ap
, long int);
721 printf("%d: correct string longer than output\n", line
);
728 /* Macro to concoct the first three parameters of iso2022_read_test. */
729 #define TESTSTR(x) __LINE__, x, lenof(x)
733 printf("read tests beginning\n");
734 /* Simple test (Emacs sample text for Japanese, in ISO-2022-JP) */
735 iso2022_read_test(TESTSTR("Japanese (\x1b$BF|K\\8l\x1b(B)\t"
736 "\x1b$B$3$s$K$A$O\x1b(B, "
737 "\x1b$B%3%s%K%A%O\x1b(B\n"),
738 'J','a','p','a','n','e','s','e',' ','(',
739 0x65E5, 0x672C, 0x8A9E, ')', '\t',
740 0x3053, 0x3093, 0x306b, 0x3061, 0x306f, ',', ' ',
741 0x30b3, 0x30f3, 0x30cb, 0x30c1, 0x30cf, '\n', 0, -1);
742 /* Same thing in EUC-JP (with designations, and half-width katakana) */
743 iso2022_read_test(TESTSTR("\x1b$)B\x1b*I\x1b$+D"
744 "Japanese (\xc6\xfc\xcb\xdc\xb8\xec)\t"
745 "\xa4\xb3\xa4\xf3\xa4\xcb\xa4\xc1\xa4\xcf, "
746 "\x8e\xba\x8e\xdd\x8e\xc6\x8e\xc1\x8e\xca\n"),
747 'J','a','p','a','n','e','s','e',' ','(',
748 0x65E5, 0x672C, 0x8A9E, ')', '\t',
749 0x3053, 0x3093, 0x306b, 0x3061, 0x306f, ',', ' ',
750 0xff7a, 0xff9d, 0xff86, 0xff81, 0xff8a, '\n', 0, -1);
751 /* Multibyte single-shift */
752 iso2022_read_test(TESTSTR("\x1b$)B\x1b*I\x1b$+D\x8f\"/!"),
754 /* Non-existent SBCS */
755 iso2022_read_test(TESTSTR("\x1b(!Zfnord\n"),
756 ERROR
, ERROR
, ERROR
, ERROR
, ERROR
, '\n', 0, -1);
757 /* Pass-through of ordinary escape sequences, including a long one */
758 iso2022_read_test(TESTSTR("\x1b""b\x1b#5\x1b#!!!5"),
759 0x1B, 'b', 0x1B, '#', '5',
760 0x1B, '#', '!', '!', '!', '5', 0, -1);
761 /* Non-existent DBCS (also 5-byte escape sequence) */
762 iso2022_read_test(TESTSTR("\x1b$(!Bfnord!"),
763 ERROR
, ERROR
, ERROR
, 0, -1);
764 /* Incomplete DB characters */
765 iso2022_read_test(TESTSTR("\x1b$B(,(\x1b(BHi\x1b$B(,(\n"),
766 0x2501, ERROR
, 'H', 'i', 0x2501, ERROR
, '\n', 0, -1);
767 iso2022_read_test(TESTSTR("\x1b$)B\x1b*I\x1b$+D\xa4""B"),
769 iso2022_read_test(TESTSTR("\x1b$)B\x1b*I\x1b$+D\x0e\x1b|$\xa2\xaf"),
770 ERROR
, 0x02D8, 0, -1);
771 /* Incomplete escape sequence */
772 iso2022_read_test(TESTSTR("\x1b\n"), ERROR
, '\n', 0, -1);
773 iso2022_read_test(TESTSTR("\x1b-A\x1b~\x1b\xa1"), ERROR
, 0xa1, 0, -1);
774 /* Incomplete single-shift */
775 iso2022_read_test(TESTSTR("\x8e\n"), ERROR
, '\n', 0, -1);
776 iso2022_read_test(TESTSTR("\x1b$*B\x8e(\n"), ERROR
, '\n', 0, -1);
777 /* Corner cases (02/00 and 07/15) */
778 iso2022_read_test(TESTSTR("\x1b(B\x20\x7f"), 0x20, 0x7f, 0, -1);
779 iso2022_read_test(TESTSTR("\x1b(I\x20\x7f"), 0x20, 0x7f, 0, -1);
780 iso2022_read_test(TESTSTR("\x1b$B\x20\x7f"), 0x20, 0x7f, 0, -1);
781 iso2022_read_test(TESTSTR("\x1b-A\x0e\x20\x7f"), 0xa0, 0xff, 0, -1);
782 iso2022_read_test(TESTSTR("\x1b$-~\x0e\x20\x7f"), ERROR
, 0, -1);
783 iso2022_read_test(TESTSTR("\x1b)B\xa0\xff"), ERROR
, ERROR
, 0, -1);
784 iso2022_read_test(TESTSTR("\x1b)I\xa0\xff"), ERROR
, ERROR
, 0, -1);
785 iso2022_read_test(TESTSTR("\x1b$)B\xa0\xff"), ERROR
, ERROR
, 0, -1);
786 iso2022_read_test(TESTSTR("\x1b-A\x1b~\xa0\xff"), 0xa0, 0xff, 0, -1);
787 iso2022_read_test(TESTSTR("\x1b$-~\x1b~\xa0\xff"), ERROR
, 0, -1);
788 /* Designate control sets */
789 iso2022_read_test(TESTSTR("\x1b!@"), 0x1b, '!', '@', 0, -1);
790 /* Designate other coding system (UTF-8) */
791 iso2022_read_test(TESTSTR("\x1b%G"
792 "\xCE\xBA\xE1\xBD\xB9\xCF\x83\xCE\xBC\xCE\xB5"),
793 0x03BA, 0x1F79, 0x03C3, 0x03BC, 0x03B5, 0, -1);
794 iso2022_read_test(TESTSTR("\x1b-A\x1b%G\xCE\xBA\x1b%@\xa0"),
795 0x03BA, 0xA0, 0, -1);
796 iso2022_read_test(TESTSTR("\x1b%G\xCE\x1b%@"), ERROR
, 0, -1);
797 iso2022_read_test(TESTSTR("\x1b%G\xCE\xBA\x1b%\x1b%@"),
798 0x03BA, 0x1B, '%', 0, -1);
799 /* DOCS (COMPOUND_TEXT extended segment) */
800 iso2022_read_test(TESTSTR("\x1b%/1\x80\x80"), 0, -1);
801 iso2022_read_test(TESTSTR("\x1b%/1\x80\x8fiso-8859-15\2xyz\x1b(B"),
802 ERROR
, ERROR
, ERROR
, 0, -1);
803 iso2022_read_test(TESTSTR("\x1b%/1\x80\x8eiso8859-15\2xyz\x1b(B"),
804 'x', 'y', 'z', 0, -1);
805 iso2022_read_test(TESTSTR("\x1b-A\x1b%/2\x80\x89"
806 "big5-0\2\xa1\x40\xa1\x40"),
807 0x3000, 0xa1, 0x40, 0, -1);
808 /* Emacs Big5-in-ISO-2022 mapping */
809 iso2022_read_test(TESTSTR("\x1b$(0&x86\x1b(B \x1b$(0DeBv"),
810 0x5143, 0x6c23, ' ', ' ', 0x958b, 0x767c, 0, -1);
811 /* Test from RFC 1922 (ISO-2022-CN) */
812 iso2022_read_test(TESTSTR("\x1b$)A\x0e=;;;\x1b$)GG(_P\x0f"),
813 0x4EA4, 0x6362, 0x4EA4, 0x63db, 0, -1);
815 printf("read tests completed\n");
816 printf("total: %d errors\n", total_errs
);
817 return (total_errs
!= 0);
820 #endif /* TESTMODE */
822 #else /* ENUM_CHARSETS */
824 ENUM_CHARSET(CS_ISO2022
)