xref: /netbsd-src/external/bsd/file/dist/src/encoding.c (revision ddb176824c39fb0db5ceef3e9e40dcaa273aec38)
1 /*	$NetBSD: encoding.c,v 1.12 2023/08/18 19:00:11 christos Exp $	*/
2 
3 /*
4  * Copyright (c) Ian F. Darwin 1986-1995.
5  * Software written by Ian F. Darwin and others;
6  * maintained 1995-present by Christos Zoulas and others.
7  *
8  * Redistribution and use in source and binary forms, with or without
9  * modification, are permitted provided that the following conditions
10  * are met:
11  * 1. Redistributions of source code must retain the above copyright
12  *    notice immediately at the beginning of the file, without modification,
13  *    this list of conditions, and the following disclaimer.
14  * 2. Redistributions in binary form must reproduce the above copyright
15  *    notice, this list of conditions and the following disclaimer in the
16  *    documentation and/or other materials provided with the distribution.
17  *
18  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
19  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
20  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
21  * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE FOR
22  * ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
23  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
24  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
25  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
26  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
27  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
28  * SUCH DAMAGE.
29  */
30 /*
31  * Encoding -- determine the character encoding of a text file.
32  *
33  * Joerg Wunsch <joerg@freebsd.org> wrote the original support for 8-bit
34  * international characters.
35  */
36 
37 #include "file.h"
38 
39 #ifndef	lint
40 #if 0
41 FILE_RCSID("@(#)$File: encoding.c,v 1.42 2022/12/26 17:31:14 christos Exp $")
42 #else
43 __RCSID("$NetBSD: encoding.c,v 1.12 2023/08/18 19:00:11 christos Exp $");
44 #endif
45 #endif	/* lint */
46 
47 #include "magic.h"
48 #include <string.h>
49 #include <stdlib.h>
50 
51 
52 file_private int looks_ascii(const unsigned char *, size_t, file_unichar_t *,
53     size_t *);
54 file_private int looks_utf8_with_BOM(const unsigned char *, size_t, file_unichar_t *,
55     size_t *);
56 file_private int looks_utf7(const unsigned char *, size_t, file_unichar_t *,
57     size_t *);
58 file_private int looks_ucs16(const unsigned char *, size_t, file_unichar_t *,
59     size_t *);
60 file_private int looks_ucs32(const unsigned char *, size_t, file_unichar_t *,
61     size_t *);
62 file_private int looks_latin1(const unsigned char *, size_t, file_unichar_t *,
63     size_t *);
64 file_private int looks_extended(const unsigned char *, size_t, file_unichar_t *,
65     size_t *);
66 file_private void from_ebcdic(const unsigned char *, size_t, unsigned char *);
67 
68 #ifdef DEBUG_ENCODING
69 #define DPRINTF(a) printf a
70 #else
71 #define DPRINTF(a)
72 #endif
73 
74 /*
75  * Try to determine whether text is in some character code we can
76  * identify.  Each of these tests, if it succeeds, will leave
77  * the text converted into one-file_unichar_t-per-character Unicode in
78  * ubuf, and the number of characters converted in ulen.
79  */
80 file_protected int
file_encoding(struct magic_set * ms,const struct buffer * b,file_unichar_t ** ubuf,size_t * ulen,const char ** code,const char ** code_mime,const char ** type)81 file_encoding(struct magic_set *ms, const struct buffer *b,
82     file_unichar_t **ubuf, size_t *ulen, const char **code,
83     const char **code_mime, const char **type)
84 {
85 	const unsigned char *buf = CAST(const unsigned char *, b->fbuf);
86 	size_t nbytes = b->flen;
87 	size_t mlen;
88 	int rv = 1, ucs_type;
89 	file_unichar_t *udefbuf;
90 	size_t udeflen;
91 
92 	if (ubuf == NULL)
93 		ubuf = &udefbuf;
94 	if (ulen == NULL)
95 		ulen = &udeflen;
96 
97 	*type = "text";
98 	*ulen = 0;
99 	*code = "unknown";
100 	*code_mime = "binary";
101 
102 	if (nbytes > ms->encoding_max)
103 		nbytes = ms->encoding_max;
104 
105 	mlen = (nbytes + 1) * sizeof((*ubuf)[0]);
106 	*ubuf = CAST(file_unichar_t *, calloc(CAST(size_t, 1), mlen));
107 	if (*ubuf == NULL) {
108 		file_oomem(ms, mlen);
109 		goto done;
110 	}
111 	if (looks_ascii(buf, nbytes, *ubuf, ulen)) {
112 		if (looks_utf7(buf, nbytes, *ubuf, ulen) > 0) {
113 			DPRINTF(("utf-7 %" SIZE_T_FORMAT "u\n", *ulen));
114 			*code = "Unicode text, UTF-7";
115 			*code_mime = "utf-7";
116 		} else {
117 			DPRINTF(("ascii %" SIZE_T_FORMAT "u\n", *ulen));
118 			*code = "ASCII";
119 			*code_mime = "us-ascii";
120 		}
121 	} else if (looks_utf8_with_BOM(buf, nbytes, *ubuf, ulen) > 0) {
122 		DPRINTF(("utf8/bom %" SIZE_T_FORMAT "u\n", *ulen));
123 		*code = "Unicode text, UTF-8 (with BOM)";
124 		*code_mime = "utf-8";
125 	} else if (file_looks_utf8(buf, nbytes, *ubuf, ulen) > 1) {
126 		DPRINTF(("utf8 %" SIZE_T_FORMAT "u\n", *ulen));
127 		*code = "Unicode text, UTF-8";
128 		*code_mime = "utf-8";
129 	} else if ((ucs_type = looks_ucs32(buf, nbytes, *ubuf, ulen)) != 0) {
130 		if (ucs_type == 1) {
131 			*code = "Unicode text, UTF-32, little-endian";
132 			*code_mime = "utf-32le";
133 		} else {
134 			*code = "Unicode text, UTF-32, big-endian";
135 			*code_mime = "utf-32be";
136 		}
137 		DPRINTF(("ucs32 %" SIZE_T_FORMAT "u\n", *ulen));
138 	} else if ((ucs_type = looks_ucs16(buf, nbytes, *ubuf, ulen)) != 0) {
139 		if (ucs_type == 1) {
140 			*code = "Unicode text, UTF-16, little-endian";
141 			*code_mime = "utf-16le";
142 		} else {
143 			*code = "Unicode text, UTF-16, big-endian";
144 			*code_mime = "utf-16be";
145 		}
146 		DPRINTF(("ucs16 %" SIZE_T_FORMAT "u\n", *ulen));
147 	} else if (looks_latin1(buf, nbytes, *ubuf, ulen)) {
148 		DPRINTF(("latin1 %" SIZE_T_FORMAT "u\n", *ulen));
149 		*code = "ISO-8859";
150 		*code_mime = "iso-8859-1";
151 	} else if (looks_extended(buf, nbytes, *ubuf, ulen)) {
152 		DPRINTF(("extended %" SIZE_T_FORMAT "u\n", *ulen));
153 		*code = "Non-ISO extended-ASCII";
154 		*code_mime = "unknown-8bit";
155 	} else {
156 		unsigned char *nbuf;
157 
158 		mlen = (nbytes + 1) * sizeof(nbuf[0]);
159 		if ((nbuf = CAST(unsigned char *, malloc(mlen))) == NULL) {
160 			file_oomem(ms, mlen);
161 			goto done;
162 		}
163 		from_ebcdic(buf, nbytes, nbuf);
164 
165 		if (looks_ascii(nbuf, nbytes, *ubuf, ulen)) {
166 			DPRINTF(("ebcdic %" SIZE_T_FORMAT "u\n", *ulen));
167 			*code = "EBCDIC";
168 			*code_mime = "ebcdic";
169 		} else if (looks_latin1(nbuf, nbytes, *ubuf, ulen)) {
170 			DPRINTF(("ebcdic/international %" SIZE_T_FORMAT "u\n",
171 			    *ulen));
172 			*code = "International EBCDIC";
173 			*code_mime = "ebcdic";
174 		} else { /* Doesn't look like text at all */
175 			DPRINTF(("binary\n"));
176 			rv = 0;
177 			*type = "binary";
178 		}
179 		free(nbuf);
180 	}
181 
182  done:
183 	if (ubuf == &udefbuf)
184 		free(udefbuf);
185 
186 	return rv;
187 }
188 
189 /*
190  * This table reflects a particular philosophy about what constitutes
191  * "text," and there is room for disagreement about it.
192  *
193  * Version 3.31 of the file command considered a file to be ASCII if
194  * each of its characters was approved by either the isascii() or
195  * isalpha() function.  On most systems, this would mean that any
196  * file consisting only of characters in the range 0x00 ... 0x7F
197  * would be called ASCII text, but many systems might reasonably
198  * consider some characters outside this range to be alphabetic,
199  * so the file command would call such characters ASCII.  It might
200  * have been more accurate to call this "considered textual on the
201  * local system" than "ASCII."
202  *
203  * It considered a file to be "International language text" if each
204  * of its characters was either an ASCII printing character (according
205  * to the real ASCII standard, not the above test), a character in
206  * the range 0x80 ... 0xFF, or one of the following control characters:
207  * backspace, tab, line feed, vertical tab, form feed, carriage return,
208  * escape.  No attempt was made to determine the language in which files
209  * of this type were written.
210  *
211  *
212  * The table below considers a file to be ASCII if all of its characters
213  * are either ASCII printing characters (again, according to the X3.4
214  * standard, not isascii()) or any of the following controls: bell,
215  * backspace, tab, line feed, form feed, carriage return, esc, nextline.
216  *
217  * I include bell because some programs (particularly shell scripts)
218  * use it literally, even though it is rare in normal text.  I exclude
219  * vertical tab because it never seems to be used in real text.  I also
220  * include, with hesitation, the X3.64/ECMA-43 control nextline (0x85),
221  * because that's what the dd EBCDIC->ASCII table maps the EBCDIC newline
222  * character to.  It might be more appropriate to include it in the 8859
223  * set instead of the ASCII set, but it's got to be included in *something*
224  * we recognize or EBCDIC files aren't going to be considered textual.
225  * Some old Unix source files use SO/SI (^N/^O) to shift between Greek
226  * and Latin characters, so these should possibly be allowed.  But they
227  * make a real mess on VT100-style displays if they're not paired properly,
228  * so we are probably better off not calling them text.
229  *
230  * A file is considered to be ISO-8859 text if its characters are all
231  * either ASCII, according to the above definition, or printing characters
232  * from the ISO-8859 8-bit extension, characters 0xA0 ... 0xFF.
233  *
234  * Finally, a file is considered to be international text from some other
235  * character code if its characters are all either ISO-8859 (according to
236  * the above definition) or characters in the range 0x80 ... 0x9F, which
237  * ISO-8859 considers to be control characters but the IBM PC and Macintosh
238  * consider to be printing characters.
239  */
240 
241 #define F 0   /* character never appears in text */
242 #define T 1   /* character appears in plain ASCII text */
243 #define I 2   /* character appears in ISO-8859 text */
244 #define X 3   /* character appears in non-ISO extended ASCII (Mac, IBM PC) */
245 
246 file_private char text_chars[256] = {
247 	/*                  BEL BS HT LF VT FF CR    */
248 	F, F, F, F, F, F, F, T, T, T, T, T, T, T, F, F,  /* 0x0X */
249 	/*                              ESC          */
250 	F, F, F, F, F, F, F, F, F, F, F, T, F, F, F, F,  /* 0x1X */
251 	T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, T,  /* 0x2X */
252 	T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, T,  /* 0x3X */
253 	T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, T,  /* 0x4X */
254 	T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, T,  /* 0x5X */
255 	T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, T,  /* 0x6X */
256 	T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, F,  /* 0x7X */
257 	/*            NEL                            */
258 	X, X, X, X, X, T, X, X, X, X, X, X, X, X, X, X,  /* 0x8X */
259 	X, X, X, X, X, X, X, X, X, X, X, X, X, X, X, X,  /* 0x9X */
260 	I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, I,  /* 0xaX */
261 	I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, I,  /* 0xbX */
262 	I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, I,  /* 0xcX */
263 	I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, I,  /* 0xdX */
264 	I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, I,  /* 0xeX */
265 	I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, I   /* 0xfX */
266 };
267 
268 #define LOOKS(NAME, COND) \
269 file_private int \
270 looks_ ## NAME(const unsigned char *buf, size_t nbytes, file_unichar_t *ubuf, \
271     size_t *ulen) \
272 { \
273 	size_t i; \
274 \
275 	*ulen = 0; \
276 \
277 	for (i = 0; i < nbytes; i++) { \
278 		int t = text_chars[buf[i]]; \
279 \
280 		if (COND) \
281 			return 0; \
282 \
283 		ubuf[(*ulen)++] = buf[i]; \
284 	} \
285 	return 1; \
286 }
287 
288 LOOKS(ascii, t != T)
289 LOOKS(latin1, t != T && t != I)
290 LOOKS(extended, t != T && t != I && t != X)
291 
292 /*
293  * Decide whether some text looks like UTF-8. Returns:
294  *
295  *     -1: invalid UTF-8
296  *      0: uses odd control characters, so doesn't look like text
297  *      1: 7-bit text
298  *      2: definitely UTF-8 text (valid high-bit set bytes)
299  *
300  * If ubuf is non-NULL on entry, text is decoded into ubuf, *ulen;
301  * ubuf must be big enough!
302  */
303 
304 // from: https://golang.org/src/unicode/utf8/utf8.go
305 
306 #define	XX 0xF1 // invalid: size 1
307 #define	AS 0xF0 // ASCII: size 1
308 #define	S1 0x02 // accept 0, size 2
309 #define	S2 0x13 // accept 1, size 3
310 #define	S3 0x03 // accept 0, size 3
311 #define	S4 0x23 // accept 2, size 3
312 #define	S5 0x34 // accept 3, size 4
313 #define	S6 0x04 // accept 0, size 4
314 #define	S7 0x44 // accept 4, size 4
315 
316 #define LOCB 0x80
317 #define HICB 0xBF
318 
319 // first is information about the first byte in a UTF-8 sequence.
320 static const uint8_t first[] = {
321     //   1   2   3   4   5   6   7   8   9   A   B   C   D   E   F
322     AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, // 0x00-0x0F
323     AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, // 0x10-0x1F
324     AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, // 0x20-0x2F
325     AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, // 0x30-0x3F
326     AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, // 0x40-0x4F
327     AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, // 0x50-0x5F
328     AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, // 0x60-0x6F
329     AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, // 0x70-0x7F
330     //   1   2   3   4   5   6   7   8   9   A   B   C   D   E   F
331     XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, // 0x80-0x8F
332     XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, // 0x90-0x9F
333     XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, // 0xA0-0xAF
334     XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, // 0xB0-0xBF
335     XX, XX, S1, S1, S1, S1, S1, S1, S1, S1, S1, S1, S1, S1, S1, S1, // 0xC0-0xCF
336     S1, S1, S1, S1, S1, S1, S1, S1, S1, S1, S1, S1, S1, S1, S1, S1, // 0xD0-0xDF
337     S2, S3, S3, S3, S3, S3, S3, S3, S3, S3, S3, S3, S3, S4, S3, S3, // 0xE0-0xEF
338     S5, S6, S6, S6, S7, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, // 0xF0-0xFF
339 };
340 
341 // acceptRange gives the range of valid values for the second byte in a UTF-8
342 // sequence.
343 struct accept_range {
344 	uint8_t lo; // lowest value for second byte.
345 	uint8_t hi; // highest value for second byte.
346 } accept_ranges[16] = {
347 // acceptRanges has size 16 to avoid bounds checks in the code that uses it.
348 	{ LOCB, HICB },
349 	{ 0xA0, HICB },
350 	{ LOCB, 0x9F },
351 	{ 0x90, HICB },
352 	{ LOCB, 0x8F },
353 };
354 
355 file_protected int
file_looks_utf8(const unsigned char * buf,size_t nbytes,file_unichar_t * ubuf,size_t * ulen)356 file_looks_utf8(const unsigned char *buf, size_t nbytes, file_unichar_t *ubuf,
357     size_t *ulen)
358 {
359 	size_t i;
360 	int n;
361 	file_unichar_t c;
362 	int gotone = 0, ctrl = 0;
363 
364 	if (ubuf)
365 		*ulen = 0;
366 
367 	for (i = 0; i < nbytes; i++) {
368 		if ((buf[i] & 0x80) == 0) {	   /* 0xxxxxxx is plain ASCII */
369 			/*
370 			 * Even if the whole file is valid UTF-8 sequences,
371 			 * still reject it if it uses weird control characters.
372 			 */
373 
374 			if (text_chars[buf[i]] != T)
375 				ctrl = 1;
376 
377 			if (ubuf)
378 				ubuf[(*ulen)++] = buf[i];
379 		} else if ((buf[i] & 0x40) == 0) { /* 10xxxxxx never 1st byte */
380 			return -1;
381 		} else {			   /* 11xxxxxx begins UTF-8 */
382 			int following;
383 			uint8_t x = first[buf[i]];
384 			const struct accept_range *ar =
385 			    &accept_ranges[(unsigned int)x >> 4];
386 			if (x == XX)
387 				return -1;
388 
389 			if ((buf[i] & 0x20) == 0) {		/* 110xxxxx */
390 				c = buf[i] & 0x1f;
391 				following = 1;
392 			} else if ((buf[i] & 0x10) == 0) {	/* 1110xxxx */
393 				c = buf[i] & 0x0f;
394 				following = 2;
395 			} else if ((buf[i] & 0x08) == 0) {	/* 11110xxx */
396 				c = buf[i] & 0x07;
397 				following = 3;
398 			} else if ((buf[i] & 0x04) == 0) {	/* 111110xx */
399 				c = buf[i] & 0x03;
400 				following = 4;
401 			} else if ((buf[i] & 0x02) == 0) {	/* 1111110x */
402 				c = buf[i] & 0x01;
403 				following = 5;
404 			} else
405 				return -1;
406 
407 			for (n = 0; n < following; n++) {
408 				i++;
409 				if (i >= nbytes)
410 					goto done;
411 
412 				if (n == 0 &&
413 				     (buf[i] < ar->lo || buf[i] > ar->hi))
414 					return -1;
415 
416 				if ((buf[i] & 0x80) == 0 || (buf[i] & 0x40))
417 					return -1;
418 
419 				c = (c << 6) + (buf[i] & 0x3f);
420 			}
421 
422 			if (ubuf)
423 				ubuf[(*ulen)++] = c;
424 			gotone = 1;
425 		}
426 	}
427 done:
428 	return ctrl ? 0 : (gotone ? 2 : 1);
429 }
430 
431 /*
432  * Decide whether some text looks like UTF-8 with BOM. If there is no
433  * BOM, return -1; otherwise return the result of looks_utf8 on the
434  * rest of the text.
435  */
436 file_private int
looks_utf8_with_BOM(const unsigned char * buf,size_t nbytes,file_unichar_t * ubuf,size_t * ulen)437 looks_utf8_with_BOM(const unsigned char *buf, size_t nbytes,
438     file_unichar_t *ubuf, size_t *ulen)
439 {
440 	if (nbytes > 3 && buf[0] == 0xef && buf[1] == 0xbb && buf[2] == 0xbf)
441 		return file_looks_utf8(buf + 3, nbytes - 3, ubuf, ulen);
442 	else
443 		return -1;
444 }
445 
446 file_private int
looks_utf7(const unsigned char * buf,size_t nbytes,file_unichar_t * ubuf,size_t * ulen)447 looks_utf7(const unsigned char *buf, size_t nbytes, file_unichar_t *ubuf,
448     size_t *ulen)
449 {
450 	if (nbytes > 4 && buf[0] == '+' && buf[1] == '/' && buf[2] == 'v')
451 		switch (buf[3]) {
452 		case '8':
453 		case '9':
454 		case '+':
455 		case '/':
456 			if (ubuf)
457 				*ulen = 0;
458 			return 1;
459 		default:
460 			return -1;
461 		}
462 	else
463 		return -1;
464 }
465 
466 #define UCS16_NOCHAR(c) ((c) >= 0xfdd0 && (c) <= 0xfdef)
467 #define UCS16_HISURR(c) ((c) >= 0xd800 && (c) <= 0xdbff)
468 #define UCS16_LOSURR(c) ((c) >= 0xdc00 && (c) <= 0xdfff)
469 
470 file_private int
looks_ucs16(const unsigned char * bf,size_t nbytes,file_unichar_t * ubf,size_t * ulen)471 looks_ucs16(const unsigned char *bf, size_t nbytes, file_unichar_t *ubf,
472     size_t *ulen)
473 {
474 	int bigend;
475 	uint32_t hi;
476 	size_t i;
477 
478 	if (nbytes < 2)
479 		return 0;
480 
481 	if (bf[0] == 0xff && bf[1] == 0xfe)
482 		bigend = 0;
483 	else if (bf[0] == 0xfe && bf[1] == 0xff)
484 		bigend = 1;
485 	else
486 		return 0;
487 
488 	*ulen = 0;
489 	hi = 0;
490 
491 	for (i = 2; i + 1 < nbytes; i += 2) {
492 		uint32_t uc;
493 
494 		if (bigend)
495 			uc = CAST(uint32_t,
496 			    bf[i + 1] | (CAST(file_unichar_t, bf[i]) << 8));
497 		else
498 			uc = CAST(uint32_t,
499 			    bf[i] | (CAST(file_unichar_t, bf[i + 1]) << 8));
500 
501 		uc &= 0xffff;
502 
503 		switch (uc) {
504 		case 0xfffe:
505 		case 0xffff:
506 			return 0;
507 		default:
508 			if (UCS16_NOCHAR(uc))
509 				return 0;
510 			break;
511 		}
512 		if (hi) {
513 			if (!UCS16_LOSURR(uc))
514 				return 0;
515 			uc = 0x10000 + 0x400 * (hi - 1) + (uc - 0xdc00);
516 			hi = 0;
517 		}
518 		if (uc < 128 && text_chars[CAST(size_t, uc)] != T)
519 			return 0;
520 		ubf[(*ulen)++] = uc;
521 		if (UCS16_HISURR(uc))
522 			hi = uc - 0xd800 + 1;
523 		if (UCS16_LOSURR(uc))
524 			return 0;
525 	}
526 
527 	return 1 + bigend;
528 }
529 
530 file_private int
looks_ucs32(const unsigned char * bf,size_t nbytes,file_unichar_t * ubf,size_t * ulen)531 looks_ucs32(const unsigned char *bf, size_t nbytes, file_unichar_t *ubf,
532     size_t *ulen)
533 {
534 	int bigend;
535 	size_t i;
536 
537 	if (nbytes < 4)
538 		return 0;
539 
540 	if (bf[0] == 0xff && bf[1] == 0xfe && bf[2] == 0 && bf[3] == 0)
541 		bigend = 0;
542 	else if (bf[0] == 0 && bf[1] == 0 && bf[2] == 0xfe && bf[3] == 0xff)
543 		bigend = 1;
544 	else
545 		return 0;
546 
547 	*ulen = 0;
548 
549 	for (i = 4; i + 3 < nbytes; i += 4) {
550 		/* XXX fix to properly handle chars > 65536 */
551 
552 		if (bigend)
553 			ubf[(*ulen)++] = CAST(file_unichar_t, bf[i + 3])
554 			    | (CAST(file_unichar_t, bf[i + 2]) << 8)
555 			    | (CAST(file_unichar_t, bf[i + 1]) << 16)
556 			    | (CAST(file_unichar_t, bf[i]) << 24);
557 		else
558 			ubf[(*ulen)++] = CAST(file_unichar_t, bf[i + 0])
559 			    | (CAST(file_unichar_t, bf[i + 1]) << 8)
560 			    | (CAST(file_unichar_t, bf[i + 2]) << 16)
561 			    | (CAST(file_unichar_t, bf[i + 3]) << 24);
562 
563 		if (ubf[*ulen - 1] == 0xfffe)
564 			return 0;
565 		if (ubf[*ulen - 1] < 128 &&
566 		    text_chars[CAST(size_t, ubf[*ulen - 1])] != T)
567 			return 0;
568 	}
569 
570 	return 1 + bigend;
571 }
572 #undef F
573 #undef T
574 #undef I
575 #undef X
576 
577 /*
578  * This table maps each EBCDIC character to an (8-bit extended) ASCII
579  * character, as specified in the rationale for the dd(1) command in
580  * draft 11.2 (September, 1991) of the POSIX P1003.2 standard.
581  *
582  * Unfortunately it does not seem to correspond exactly to any of the
583  * five variants of EBCDIC documented in IBM's _Enterprise Systems
584  * Architecture/390: Principles of Operation_, SA22-7201-06, Seventh
585  * Edition, July, 1999, pp. I-1 - I-4.
586  *
587  * Fortunately, though, all versions of EBCDIC, including this one, agree
588  * on most of the printing characters that also appear in (7-bit) ASCII.
589  * Of these, only '|', '!', '~', '^', '[', and ']' are in question at all.
590  *
591  * Fortunately too, there is general agreement that codes 0x00 through
592  * 0x3F represent control characters, 0x41 a nonbreaking space, and the
593  * remainder printing characters.
594  *
595  * This is sufficient to allow us to identify EBCDIC text and to distinguish
596  * between old-style and internationalized examples of text.
597  */
598 
599 file_private unsigned char ebcdic_to_ascii[] = {
600   0,   1,   2,   3, 156,   9, 134, 127, 151, 141, 142,  11,  12,  13,  14,  15,
601  16,  17,  18,  19, 157, 133,   8, 135,  24,  25, 146, 143,  28,  29,  30,  31,
602 128, 129, 130, 131, 132,  10,  23,  27, 136, 137, 138, 139, 140,   5,   6,   7,
603 144, 145,  22, 147, 148, 149, 150,   4, 152, 153, 154, 155,  20,  21, 158,  26,
604 ' ', 160, 161, 162, 163, 164, 165, 166, 167, 168, 213, '.', '<', '(', '+', '|',
605 '&', 169, 170, 171, 172, 173, 174, 175, 176, 177, '!', '$', '*', ')', ';', '~',
606 '-', '/', 178, 179, 180, 181, 182, 183, 184, 185, 203, ',', '%', '_', '>', '?',
607 186, 187, 188, 189, 190, 191, 192, 193, 194, '`', ':', '#', '@', '\'','=', '"',
608 195, 'a', 'b', 'c', 'd', 'e', 'f', 'g', 'h', 'i', 196, 197, 198, 199, 200, 201,
609 202, 'j', 'k', 'l', 'm', 'n', 'o', 'p', 'q', 'r', '^', 204, 205, 206, 207, 208,
610 209, 229, 's', 't', 'u', 'v', 'w', 'x', 'y', 'z', 210, 211, 212, '[', 214, 215,
611 216, 217, 218, 219, 220, 221, 222, 223, 224, 225, 226, 227, 228, ']', 230, 231,
612 '{', 'A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', 'I', 232, 233, 234, 235, 236, 237,
613 '}', 'J', 'K', 'L', 'M', 'N', 'O', 'P', 'Q', 'R', 238, 239, 240, 241, 242, 243,
614 '\\',159, 'S', 'T', 'U', 'V', 'W', 'X', 'Y', 'Z', 244, 245, 246, 247, 248, 249,
615 '0', '1', '2', '3', '4', '5', '6', '7', '8', '9', 250, 251, 252, 253, 254, 255
616 };
617 
618 #ifdef notdef
619 /*
620  * The following EBCDIC-to-ASCII table may relate more closely to reality,
621  * or at least to modern reality.  It comes from
622  *
623  *   http://ftp.s390.ibm.com/products/oe/bpxqp9.html
624  *
625  * and maps the characters of EBCDIC code page 1047 (the code used for
626  * Unix-derived software on IBM's 390 systems) to the corresponding
627  * characters from ISO 8859-1.
628  *
629  * If this table is used instead of the above one, some of the special
630  * cases for the NEL character can be taken out of the code.
631  */
632 
633 file_private unsigned char ebcdic_1047_to_8859[] = {
634 0x00,0x01,0x02,0x03,0x9C,0x09,0x86,0x7F,0x97,0x8D,0x8E,0x0B,0x0C,0x0D,0x0E,0x0F,
635 0x10,0x11,0x12,0x13,0x9D,0x0A,0x08,0x87,0x18,0x19,0x92,0x8F,0x1C,0x1D,0x1E,0x1F,
636 0x80,0x81,0x82,0x83,0x84,0x85,0x17,0x1B,0x88,0x89,0x8A,0x8B,0x8C,0x05,0x06,0x07,
637 0x90,0x91,0x16,0x93,0x94,0x95,0x96,0x04,0x98,0x99,0x9A,0x9B,0x14,0x15,0x9E,0x1A,
638 0x20,0xA0,0xE2,0xE4,0xE0,0xE1,0xE3,0xE5,0xE7,0xF1,0xA2,0x2E,0x3C,0x28,0x2B,0x7C,
639 0x26,0xE9,0xEA,0xEB,0xE8,0xED,0xEE,0xEF,0xEC,0xDF,0x21,0x24,0x2A,0x29,0x3B,0x5E,
640 0x2D,0x2F,0xC2,0xC4,0xC0,0xC1,0xC3,0xC5,0xC7,0xD1,0xA6,0x2C,0x25,0x5F,0x3E,0x3F,
641 0xF8,0xC9,0xCA,0xCB,0xC8,0xCD,0xCE,0xCF,0xCC,0x60,0x3A,0x23,0x40,0x27,0x3D,0x22,
642 0xD8,0x61,0x62,0x63,0x64,0x65,0x66,0x67,0x68,0x69,0xAB,0xBB,0xF0,0xFD,0xFE,0xB1,
643 0xB0,0x6A,0x6B,0x6C,0x6D,0x6E,0x6F,0x70,0x71,0x72,0xAA,0xBA,0xE6,0xB8,0xC6,0xA4,
644 0xB5,0x7E,0x73,0x74,0x75,0x76,0x77,0x78,0x79,0x7A,0xA1,0xBF,0xD0,0x5B,0xDE,0xAE,
645 0xAC,0xA3,0xA5,0xB7,0xA9,0xA7,0xB6,0xBC,0xBD,0xBE,0xDD,0xA8,0xAF,0x5D,0xB4,0xD7,
646 0x7B,0x41,0x42,0x43,0x44,0x45,0x46,0x47,0x48,0x49,0xAD,0xF4,0xF6,0xF2,0xF3,0xF5,
647 0x7D,0x4A,0x4B,0x4C,0x4D,0x4E,0x4F,0x50,0x51,0x52,0xB9,0xFB,0xFC,0xF9,0xFA,0xFF,
648 0x5C,0xF7,0x53,0x54,0x55,0x56,0x57,0x58,0x59,0x5A,0xB2,0xD4,0xD6,0xD2,0xD3,0xD5,
649 0x30,0x31,0x32,0x33,0x34,0x35,0x36,0x37,0x38,0x39,0xB3,0xDB,0xDC,0xD9,0xDA,0x9F
650 };
651 #endif
652 
653 /*
654  * Copy buf[0 ... nbytes-1] into out[], translating EBCDIC to ASCII.
655  */
656 file_private void
from_ebcdic(const unsigned char * buf,size_t nbytes,unsigned char * out)657 from_ebcdic(const unsigned char *buf, size_t nbytes, unsigned char *out)
658 {
659 	size_t i;
660 
661 	for (i = 0; i < nbytes; i++) {
662 		out[i] = ebcdic_to_ascii[buf[i]];
663 	}
664 }
665