xref: /freebsd/contrib/file/src/encoding.c (revision c491c2db2f37399eba7a356c4bfa298d64012101)
1 /*
2  * Copyright (c) Ian F. Darwin 1986-1995.
3  * Software written by Ian F. Darwin and others;
4  * maintained 1995-present by Christos Zoulas and others.
5  *
6  * Redistribution and use in source and binary forms, with or without
7  * modification, are permitted provided that the following conditions
8  * are met:
9  * 1. Redistributions of source code must retain the above copyright
10  *    notice immediately at the beginning of the file, without modification,
11  *    this list of conditions, and the following disclaimer.
12  * 2. Redistributions in binary form must reproduce the above copyright
13  *    notice, this list of conditions and the following disclaimer in the
14  *    documentation and/or other materials provided with the distribution.
15  *
16  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
17  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
18  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
19  * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE FOR
20  * ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
21  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
22  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
23  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
24  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
25  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
26  * SUCH DAMAGE.
27  */
28 /*
29  * Encoding -- determine the character encoding of a text file.
30  *
31  * Joerg Wunsch <joerg@freebsd.org> wrote the original support for 8-bit
32  * international characters.
33  */
34 
35 #include "file.h"
36 
37 #ifndef	lint
38 FILE_RCSID("@(#)$File: encoding.c,v 1.44 2024/12/26 18:41:27 christos Exp $")
39 #endif	/* lint */
40 
41 #include "magic.h"
42 #include <string.h>
43 #include <stdlib.h>
44 
45 
46 file_private int looks_ascii(const unsigned char *, size_t, file_unichar_t *,
47     size_t *);
48 file_private int looks_utf8_with_BOM(const unsigned char *, size_t, file_unichar_t *,
49     size_t *);
50 file_private int looks_utf7(const unsigned char *, size_t, file_unichar_t *,
51     size_t *);
52 file_private int looks_ucs16(const unsigned char *, size_t, file_unichar_t *,
53     size_t *);
54 file_private int looks_ucs32(const unsigned char *, size_t, file_unichar_t *,
55     size_t *);
56 file_private int looks_latin1(const unsigned char *, size_t, file_unichar_t *,
57     size_t *);
58 file_private int looks_extended(const unsigned char *, size_t, file_unichar_t *,
59     size_t *);
60 file_private void from_ebcdic(const unsigned char *, size_t, unsigned char *);
61 
62 #ifdef DEBUG_ENCODING
63 #define DPRINTF(a) printf a
64 #else
65 #define DPRINTF(a)
66 #endif
67 
68 /*
69  * Try to determine whether text is in some character code we can
70  * identify.  Each of these tests, if it succeeds, will leave
71  * the text converted into one-file_unichar_t-per-character Unicode in
72  * ubuf, and the number of characters converted in ulen.
73  */
74 file_protected int
75 file_encoding(struct magic_set *ms, const struct buffer *b,
76     file_unichar_t **ubuf, size_t *ulen, const char **code,
77     const char **code_mime, const char **type)
78 {
79 	const unsigned char *buf = CAST(const unsigned char *, b->fbuf);
80 	size_t nbytes = b->flen;
81 	size_t mlen;
82 	int rv = 1, ucs_type;
83 	file_unichar_t *udefbuf;
84 	size_t udeflen;
85 
86 	if (ubuf == NULL)
87 		ubuf = &udefbuf;
88 	if (ulen == NULL)
89 		ulen = &udeflen;
90 
91 	*type = "text";
92 	*ulen = 0;
93 	*code = "unknown";
94 	*code_mime = "binary";
95 
96 	if (nbytes > ms->encoding_max)
97 		nbytes = ms->encoding_max;
98 
99 	mlen = (nbytes + 1) * sizeof((*ubuf)[0]);
100 	*ubuf = CAST(file_unichar_t *, calloc(CAST(size_t, 1), mlen));
101 	if (*ubuf == NULL) {
102 		file_oomem(ms, mlen);
103 		goto done;
104 	}
105 	if (looks_ascii(buf, nbytes, *ubuf, ulen)) {
106 		if (looks_utf7(buf, nbytes, *ubuf, ulen) > 0) {
107 			DPRINTF(("utf-7 %" SIZE_T_FORMAT "u\n", *ulen));
108 			*code = "Unicode text, UTF-7";
109 			*code_mime = "utf-7";
110 		} else {
111 			DPRINTF(("ascii %" SIZE_T_FORMAT "u\n", *ulen));
112 			*code = "ASCII";
113 			*code_mime = "us-ascii";
114 		}
115 	} else if (looks_utf8_with_BOM(buf, nbytes, *ubuf, ulen) > 0) {
116 		DPRINTF(("utf8/bom %" SIZE_T_FORMAT "u\n", *ulen));
117 		*code = "Unicode text, UTF-8 (with BOM)";
118 		*code_mime = "utf-8";
119 	} else if (file_looks_utf8(buf, nbytes, *ubuf, ulen) > 1) {
120 		DPRINTF(("utf8 %" SIZE_T_FORMAT "u\n", *ulen));
121 		*code = "Unicode text, UTF-8";
122 		*code_mime = "utf-8";
123 	} else if ((ucs_type = looks_ucs32(buf, nbytes, *ubuf, ulen)) != 0) {
124 		if (ucs_type == 1) {
125 			*code = "Unicode text, UTF-32, little-endian";
126 			*code_mime = "utf-32le";
127 		} else {
128 			*code = "Unicode text, UTF-32, big-endian";
129 			*code_mime = "utf-32be";
130 		}
131 		DPRINTF(("ucs32 %" SIZE_T_FORMAT "u\n", *ulen));
132 	} else if ((ucs_type = looks_ucs16(buf, nbytes, *ubuf, ulen)) != 0) {
133 		if (ucs_type == 1) {
134 			*code = "Unicode text, UTF-16, little-endian";
135 			*code_mime = "utf-16le";
136 		} else {
137 			*code = "Unicode text, UTF-16, big-endian";
138 			*code_mime = "utf-16be";
139 		}
140 		DPRINTF(("ucs16 %" SIZE_T_FORMAT "u\n", *ulen));
141 	} else if (looks_latin1(buf, nbytes, *ubuf, ulen)) {
142 		DPRINTF(("latin1 %" SIZE_T_FORMAT "u\n", *ulen));
143 		*code = "ISO-8859";
144 		*code_mime = "iso-8859-1";
145 	} else if (looks_extended(buf, nbytes, *ubuf, ulen)) {
146 		DPRINTF(("extended %" SIZE_T_FORMAT "u\n", *ulen));
147 		*code = "Non-ISO extended-ASCII";
148 		*code_mime = "unknown-8bit";
149 	} else {
150 		unsigned char *nbuf;
151 
152 		mlen = (nbytes + 1) * sizeof(nbuf[0]);
153 		if ((nbuf = CAST(unsigned char *, malloc(mlen))) == NULL) {
154 			file_oomem(ms, mlen);
155 			goto done;
156 		}
157 		from_ebcdic(buf, nbytes, nbuf);
158 
159 		if (looks_ascii(nbuf, nbytes, *ubuf, ulen)) {
160 			DPRINTF(("ebcdic %" SIZE_T_FORMAT "u\n", *ulen));
161 			*code = "EBCDIC";
162 			*code_mime = "ebcdic";
163 		} else if (looks_latin1(nbuf, nbytes, *ubuf, ulen)) {
164 			DPRINTF(("ebcdic/international %" SIZE_T_FORMAT "u\n",
165 			    *ulen));
166 			*code = "International EBCDIC";
167 			*code_mime = "ebcdic";
168 		} else { /* Doesn't look like text at all */
169 			DPRINTF(("binary\n"));
170 			rv = 0;
171 			*type = "binary";
172 		}
173 		free(nbuf);
174 	}
175 
176  done:
177 	if (ubuf == &udefbuf)
178 		free(udefbuf);
179 
180 	return rv;
181 }
182 
183 /*
184  * This table reflects a particular philosophy about what constitutes
185  * "text," and there is room for disagreement about it.
186  *
187  * Version 3.31 of the file command considered a file to be ASCII if
188  * each of its characters was approved by either the isascii() or
189  * isalpha() function.  On most systems, this would mean that any
190  * file consisting only of characters in the range 0x00 ... 0x7F
191  * would be called ASCII text, but many systems might reasonably
192  * consider some characters outside this range to be alphabetic,
193  * so the file command would call such characters ASCII.  It might
194  * have been more accurate to call this "considered textual on the
195  * local system" than "ASCII."
196  *
197  * It considered a file to be "International language text" if each
198  * of its characters was either an ASCII printing character (according
199  * to the real ASCII standard, not the above test), a character in
200  * the range 0x80 ... 0xFF, or one of the following control characters:
201  * backspace, tab, line feed, vertical tab, form feed, carriage return,
202  * escape.  No attempt was made to determine the language in which files
203  * of this type were written.
204  *
205  *
206  * The table below considers a file to be ASCII if all of its characters
207  * are either ASCII printing characters (again, according to the X3.4
208  * standard, not isascii()) or any of the following controls: bell,
209  * backspace, tab, line feed, form feed, carriage return, esc, nextline.
210  *
211  * I include bell because some programs (particularly shell scripts)
212  * use it literally, even though it is rare in normal text.  I exclude
213  * vertical tab because it never seems to be used in real text.  I also
214  * include, with hesitation, the X3.64/ECMA-43 control nextline (0x85),
215  * because that's what the dd EBCDIC->ASCII table maps the EBCDIC newline
216  * character to.  It might be more appropriate to include it in the 8859
217  * set instead of the ASCII set, but it's got to be included in *something*
218  * we recognize or EBCDIC files aren't going to be considered textual.
219  * Some old Unix source files use SO/SI (^N/^O) to shift between Greek
220  * and Latin characters, so these should possibly be allowed.  But they
221  * make a real mess on VT100-style displays if they're not paired properly,
222  * so we are probably better off not calling them text.
223  *
224  * A file is considered to be ISO-8859 text if its characters are all
225  * either ASCII, according to the above definition, or printing characters
226  * from the ISO-8859 8-bit extension, characters 0xA0 ... 0xFF.
227  *
228  * Finally, a file is considered to be international text from some other
229  * character code if its characters are all either ISO-8859 (according to
230  * the above definition) or characters in the range 0x80 ... 0x9F, which
231  * ISO-8859 considers to be control characters but the IBM PC and Macintosh
232  * consider to be printing characters.
233  */
234 
235 #define F 0   /* character never appears in text */
236 #define T 1   /* character appears in plain ASCII text */
237 #define I 2   /* character appears in ISO-8859 text */
238 #define X 3   /* character appears in non-ISO extended ASCII (Mac, IBM PC) */
239 
240 /*
241  * SUB (substitute character ^Z) was used as EOF in DOS and early Windows
242  * NEL (next line 0x85) is considered in ECMAScript as whitespace
243  */
244 file_private char text_chars[256] = {
245 	/*                  BEL BS HT LF VT FF CR    */
246 	F, F, F, F, F, F, F, T, T, T, T, T, T, T, F, F,  /* 0x0X */
247 	/*                           SUB ESC          */
248 	F, F, F, F, F, F, F, F, F, F, T, T, F, F, F, F,  /* 0x1X */
249 	T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, T,  /* 0x2X */
250 	T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, T,  /* 0x3X */
251 	T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, T,  /* 0x4X */
252 	T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, T,  /* 0x5X */
253 	T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, T,  /* 0x6X */
254 	T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, F,  /* 0x7X */
255 	/*            NEL                            */
256 	X, X, X, X, X, T, X, X, X, X, X, X, X, X, X, X,  /* 0x8X */
257 	X, X, X, X, X, X, X, X, X, X, X, X, X, X, X, X,  /* 0x9X */
258 	I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, I,  /* 0xaX */
259 	I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, I,  /* 0xbX */
260 	I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, I,  /* 0xcX */
261 	I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, I,  /* 0xdX */
262 	I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, I,  /* 0xeX */
263 	I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, I   /* 0xfX */
264 };
265 
266 #define LOOKS(NAME, COND) \
267 file_private int \
268 looks_ ## NAME(const unsigned char *buf, size_t nbytes, file_unichar_t *ubuf, \
269     size_t *ulen) \
270 { \
271 	size_t i; \
272 \
273 	*ulen = 0; \
274 \
275 	for (i = 0; i < nbytes; i++) { \
276 		int t = text_chars[buf[i]]; \
277 \
278 		if (COND) \
279 			return 0; \
280 \
281 		ubuf[(*ulen)++] = buf[i]; \
282 	} \
283 	return 1; \
284 }
285 
286 LOOKS(ascii, t != T)
287 LOOKS(latin1, t != T && t != I)
288 LOOKS(extended, t != T && t != I && t != X)
289 
290 /*
291  * Decide whether some text looks like UTF-8. Returns:
292  *
293  *     -1: invalid UTF-8
294  *      0: uses odd control characters, so doesn't look like text
295  *      1: 7-bit text
296  *      2: definitely UTF-8 text (valid high-bit set bytes)
297  *
298  * If ubuf is non-NULL on entry, text is decoded into ubuf, *ulen;
299  * ubuf must be big enough!
300  */
301 
302 // from: https://golang.org/src/unicode/utf8/utf8.go
303 
304 #define	XX 0xF1 // invalid: size 1
305 #define	AS 0xF0 // ASCII: size 1
306 #define	S1 0x02 // accept 0, size 2
307 #define	S2 0x13 // accept 1, size 3
308 #define	S3 0x03 // accept 0, size 3
309 #define	S4 0x23 // accept 2, size 3
310 #define	S5 0x34 // accept 3, size 4
311 #define	S6 0x04 // accept 0, size 4
312 #define	S7 0x44 // accept 4, size 4
313 
314 #define LOCB 0x80
315 #define HICB 0xBF
316 
317 // first is information about the first byte in a UTF-8 sequence.
318 static const uint8_t first[] = {
319     //   1   2   3   4   5   6   7   8   9   A   B   C   D   E   F
320     AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, // 0x00-0x0F
321     AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, // 0x10-0x1F
322     AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, // 0x20-0x2F
323     AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, // 0x30-0x3F
324     AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, // 0x40-0x4F
325     AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, // 0x50-0x5F
326     AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, // 0x60-0x6F
327     AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, // 0x70-0x7F
328     //   1   2   3   4   5   6   7   8   9   A   B   C   D   E   F
329     XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, // 0x80-0x8F
330     XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, // 0x90-0x9F
331     XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, // 0xA0-0xAF
332     XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, // 0xB0-0xBF
333     XX, XX, S1, S1, S1, S1, S1, S1, S1, S1, S1, S1, S1, S1, S1, S1, // 0xC0-0xCF
334     S1, S1, S1, S1, S1, S1, S1, S1, S1, S1, S1, S1, S1, S1, S1, S1, // 0xD0-0xDF
335     S2, S3, S3, S3, S3, S3, S3, S3, S3, S3, S3, S3, S3, S4, S3, S3, // 0xE0-0xEF
336     S5, S6, S6, S6, S7, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, // 0xF0-0xFF
337 };
338 
339 // acceptRange gives the range of valid values for the second byte in a UTF-8
340 // sequence.
341 static struct accept_range {
342 	uint8_t lo; // lowest value for second byte.
343 	uint8_t hi; // highest value for second byte.
344 } accept_ranges[16] = {
345 // acceptRanges has size 16 to avoid bounds checks in the code that uses it.
346 	{ LOCB, HICB },
347 	{ 0xA0, HICB },
348 	{ LOCB, 0x9F },
349 	{ 0x90, HICB },
350 	{ LOCB, 0x8F },
351 };
352 
353 file_protected int
354 file_looks_utf8(const unsigned char *buf, size_t nbytes, file_unichar_t *ubuf,
355     size_t *ulen)
356 {
357 	size_t i;
358 	int n;
359 	file_unichar_t c;
360 	int gotone = 0, ctrl = 0;
361 
362 	if (ubuf)
363 		*ulen = 0;
364 
365 	for (i = 0; i < nbytes; i++) {
366 		if ((buf[i] & 0x80) == 0) {	   /* 0xxxxxxx is plain ASCII */
367 			/*
368 			 * Even if the whole file is valid UTF-8 sequences,
369 			 * still reject it if it uses weird control characters.
370 			 */
371 
372 			if (text_chars[buf[i]] != T)
373 				ctrl = 1;
374 
375 			if (ubuf)
376 				ubuf[(*ulen)++] = buf[i];
377 		} else if ((buf[i] & 0x40) == 0) { /* 10xxxxxx never 1st byte */
378 			return -1;
379 		} else {			   /* 11xxxxxx begins UTF-8 */
380 			int following;
381 			uint8_t x = first[buf[i]];
382 			const struct accept_range *ar =
383 			    &accept_ranges[(unsigned int)x >> 4];
384 			if (x == XX)
385 				return -1;
386 
387 			if ((buf[i] & 0x20) == 0) {		/* 110xxxxx */
388 				c = buf[i] & 0x1f;
389 				following = 1;
390 			} else if ((buf[i] & 0x10) == 0) {	/* 1110xxxx */
391 				c = buf[i] & 0x0f;
392 				following = 2;
393 			} else if ((buf[i] & 0x08) == 0) {	/* 11110xxx */
394 				c = buf[i] & 0x07;
395 				following = 3;
396 			} else if ((buf[i] & 0x04) == 0) {	/* 111110xx */
397 				c = buf[i] & 0x03;
398 				following = 4;
399 			} else if ((buf[i] & 0x02) == 0) {	/* 1111110x */
400 				c = buf[i] & 0x01;
401 				following = 5;
402 			} else
403 				return -1;
404 
405 			for (n = 0; n < following; n++) {
406 				i++;
407 				if (i >= nbytes)
408 					goto done;
409 
410 				if (n == 0 &&
411 				     (buf[i] < ar->lo || buf[i] > ar->hi))
412 					return -1;
413 
414 				if ((buf[i] & 0x80) == 0 || (buf[i] & 0x40))
415 					return -1;
416 
417 				c = (c << 6) + (buf[i] & 0x3f);
418 			}
419 
420 			if (ubuf)
421 				ubuf[(*ulen)++] = c;
422 			gotone = 1;
423 		}
424 	}
425 done:
426 	return ctrl ? 0 : (gotone ? 2 : 1);
427 }
428 
429 /*
430  * Decide whether some text looks like UTF-8 with BOM. If there is no
431  * BOM, return -1; otherwise return the result of looks_utf8 on the
432  * rest of the text.
433  */
434 file_private int
435 looks_utf8_with_BOM(const unsigned char *buf, size_t nbytes,
436     file_unichar_t *ubuf, size_t *ulen)
437 {
438 	if (nbytes > 3 && buf[0] == 0xef && buf[1] == 0xbb && buf[2] == 0xbf)
439 		return file_looks_utf8(buf + 3, nbytes - 3, ubuf, ulen);
440 	else
441 		return -1;
442 }
443 
444 file_private int
445 looks_utf7(const unsigned char *buf, size_t nbytes, file_unichar_t *ubuf,
446     size_t *ulen)
447 {
448 	if (nbytes > 4 && buf[0] == '+' && buf[1] == '/' && buf[2] == 'v')
449 		switch (buf[3]) {
450 		case '8':
451 		case '9':
452 		case '+':
453 		case '/':
454 			if (ubuf)
455 				*ulen = 0;
456 			return 1;
457 		default:
458 			return -1;
459 		}
460 	else
461 		return -1;
462 }
463 
464 #define UCS16_NOCHAR(c) ((c) >= 0xfdd0 && (c) <= 0xfdef)
465 #define UCS16_HISURR(c) ((c) >= 0xd800 && (c) <= 0xdbff)
466 #define UCS16_LOSURR(c) ((c) >= 0xdc00 && (c) <= 0xdfff)
467 
468 file_private int
469 looks_ucs16(const unsigned char *bf, size_t nbytes, file_unichar_t *ubf,
470     size_t *ulen)
471 {
472 	int bigend;
473 	uint32_t hi;
474 	size_t i;
475 
476 	if (nbytes < 2)
477 		return 0;
478 
479 	if (bf[0] == 0xff && bf[1] == 0xfe)
480 		bigend = 0;
481 	else if (bf[0] == 0xfe && bf[1] == 0xff)
482 		bigend = 1;
483 	else
484 		return 0;
485 
486 	*ulen = 0;
487 	hi = 0;
488 
489 	for (i = 2; i + 1 < nbytes; i += 2) {
490 		uint32_t uc;
491 
492 		if (bigend)
493 			uc = CAST(uint32_t,
494 			    bf[i + 1] | (CAST(file_unichar_t, bf[i]) << 8));
495 		else
496 			uc = CAST(uint32_t,
497 			    bf[i] | (CAST(file_unichar_t, bf[i + 1]) << 8));
498 
499 		uc &= 0xffff;
500 
501 		switch (uc) {
502 		case 0xfffe:
503 		case 0xffff:
504 			return 0;
505 		default:
506 			if (UCS16_NOCHAR(uc))
507 				return 0;
508 			break;
509 		}
510 		if (hi) {
511 			if (!UCS16_LOSURR(uc))
512 				return 0;
513 			uc = 0x10000 + 0x400 * (hi - 1) + (uc - 0xdc00);
514 			hi = 0;
515 		}
516 		if (uc < 128 && text_chars[CAST(size_t, uc)] != T)
517 			return 0;
518 		ubf[(*ulen)++] = uc;
519 		if (UCS16_HISURR(uc))
520 			hi = uc - 0xd800 + 1;
521 		if (UCS16_LOSURR(uc))
522 			return 0;
523 	}
524 
525 	return 1 + bigend;
526 }
527 
528 file_private int
529 looks_ucs32(const unsigned char *bf, size_t nbytes, file_unichar_t *ubf,
530     size_t *ulen)
531 {
532 	int bigend;
533 	size_t i;
534 
535 	if (nbytes < 4)
536 		return 0;
537 
538 	if (bf[0] == 0xff && bf[1] == 0xfe && bf[2] == 0 && bf[3] == 0)
539 		bigend = 0;
540 	else if (bf[0] == 0 && bf[1] == 0 && bf[2] == 0xfe && bf[3] == 0xff)
541 		bigend = 1;
542 	else
543 		return 0;
544 
545 	*ulen = 0;
546 
547 	for (i = 4; i + 3 < nbytes; i += 4) {
548 		/* XXX fix to properly handle chars > 65536 */
549 
550 		if (bigend)
551 			ubf[(*ulen)++] = CAST(file_unichar_t, bf[i + 3])
552 			    | (CAST(file_unichar_t, bf[i + 2]) << 8)
553 			    | (CAST(file_unichar_t, bf[i + 1]) << 16)
554 			    | (CAST(file_unichar_t, bf[i]) << 24);
555 		else
556 			ubf[(*ulen)++] = CAST(file_unichar_t, bf[i + 0])
557 			    | (CAST(file_unichar_t, bf[i + 1]) << 8)
558 			    | (CAST(file_unichar_t, bf[i + 2]) << 16)
559 			    | (CAST(file_unichar_t, bf[i + 3]) << 24);
560 
561 		if (ubf[*ulen - 1] == 0xfffe)
562 			return 0;
563 		if (ubf[*ulen - 1] < 128 &&
564 		    text_chars[CAST(size_t, ubf[*ulen - 1])] != T)
565 			return 0;
566 	}
567 
568 	return 1 + bigend;
569 }
570 #undef F
571 #undef T
572 #undef I
573 #undef X
574 
575 /*
576  * This table maps each EBCDIC character to an (8-bit extended) ASCII
577  * character, as specified in the rationale for the dd(1) command in
578  * draft 11.2 (September, 1991) of the POSIX P1003.2 standard.
579  *
580  * Unfortunately it does not seem to correspond exactly to any of the
581  * five variants of EBCDIC documented in IBM's _Enterprise Systems
582  * Architecture/390: Principles of Operation_, SA22-7201-06, Seventh
583  * Edition, July, 1999, pp. I-1 - I-4.
584  *
585  * Fortunately, though, all versions of EBCDIC, including this one, agree
586  * on most of the printing characters that also appear in (7-bit) ASCII.
587  * Of these, only '|', '!', '~', '^', '[', and ']' are in question at all.
588  *
589  * Fortunately too, there is general agreement that codes 0x00 through
590  * 0x3F represent control characters, 0x41 a nonbreaking space, and the
591  * remainder printing characters.
592  *
593  * This is sufficient to allow us to identify EBCDIC text and to distinguish
594  * between old-style and internationalized examples of text.
595  */
596 
597 file_private unsigned char ebcdic_to_ascii[] = {
598   0,   1,   2,   3, 156,   9, 134, 127, 151, 141, 142,  11,  12,  13,  14,  15,
599  16,  17,  18,  19, 157, 133,   8, 135,  24,  25, 146, 143,  28,  29,  30,  31,
600 128, 129, 130, 131, 132,  10,  23,  27, 136, 137, 138, 139, 140,   5,   6,   7,
601 144, 145,  22, 147, 148, 149, 150,   4, 152, 153, 154, 155,  20,  21, 158,  26,
602 ' ', 160, 161, 162, 163, 164, 165, 166, 167, 168, 213, '.', '<', '(', '+', '|',
603 '&', 169, 170, 171, 172, 173, 174, 175, 176, 177, '!', '$', '*', ')', ';', '~',
604 '-', '/', 178, 179, 180, 181, 182, 183, 184, 185, 203, ',', '%', '_', '>', '?',
605 186, 187, 188, 189, 190, 191, 192, 193, 194, '`', ':', '#', '@', '\'','=', '"',
606 195, 'a', 'b', 'c', 'd', 'e', 'f', 'g', 'h', 'i', 196, 197, 198, 199, 200, 201,
607 202, 'j', 'k', 'l', 'm', 'n', 'o', 'p', 'q', 'r', '^', 204, 205, 206, 207, 208,
608 209, 229, 's', 't', 'u', 'v', 'w', 'x', 'y', 'z', 210, 211, 212, '[', 214, 215,
609 216, 217, 218, 219, 220, 221, 222, 223, 224, 225, 226, 227, 228, ']', 230, 231,
610 '{', 'A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', 'I', 232, 233, 234, 235, 236, 237,
611 '}', 'J', 'K', 'L', 'M', 'N', 'O', 'P', 'Q', 'R', 238, 239, 240, 241, 242, 243,
612 '\\',159, 'S', 'T', 'U', 'V', 'W', 'X', 'Y', 'Z', 244, 245, 246, 247, 248, 249,
613 '0', '1', '2', '3', '4', '5', '6', '7', '8', '9', 250, 251, 252, 253, 254, 255
614 };
615 
616 #ifdef notdef
617 /*
618  * The following EBCDIC-to-ASCII table may relate more closely to reality,
619  * or at least to modern reality.  It comes from
620  *
621  *   http://ftp.s390.ibm.com/products/oe/bpxqp9.html
622  *
623  * and maps the characters of EBCDIC code page 1047 (the code used for
624  * Unix-derived software on IBM's 390 systems) to the corresponding
625  * characters from ISO 8859-1.
626  *
627  * If this table is used instead of the above one, some of the special
628  * cases for the NEL character can be taken out of the code.
629  */
630 
631 file_private unsigned char ebcdic_1047_to_8859[] = {
632 0x00,0x01,0x02,0x03,0x9C,0x09,0x86,0x7F,0x97,0x8D,0x8E,0x0B,0x0C,0x0D,0x0E,0x0F,
633 0x10,0x11,0x12,0x13,0x9D,0x0A,0x08,0x87,0x18,0x19,0x92,0x8F,0x1C,0x1D,0x1E,0x1F,
634 0x80,0x81,0x82,0x83,0x84,0x85,0x17,0x1B,0x88,0x89,0x8A,0x8B,0x8C,0x05,0x06,0x07,
635 0x90,0x91,0x16,0x93,0x94,0x95,0x96,0x04,0x98,0x99,0x9A,0x9B,0x14,0x15,0x9E,0x1A,
636 0x20,0xA0,0xE2,0xE4,0xE0,0xE1,0xE3,0xE5,0xE7,0xF1,0xA2,0x2E,0x3C,0x28,0x2B,0x7C,
637 0x26,0xE9,0xEA,0xEB,0xE8,0xED,0xEE,0xEF,0xEC,0xDF,0x21,0x24,0x2A,0x29,0x3B,0x5E,
638 0x2D,0x2F,0xC2,0xC4,0xC0,0xC1,0xC3,0xC5,0xC7,0xD1,0xA6,0x2C,0x25,0x5F,0x3E,0x3F,
639 0xF8,0xC9,0xCA,0xCB,0xC8,0xCD,0xCE,0xCF,0xCC,0x60,0x3A,0x23,0x40,0x27,0x3D,0x22,
640 0xD8,0x61,0x62,0x63,0x64,0x65,0x66,0x67,0x68,0x69,0xAB,0xBB,0xF0,0xFD,0xFE,0xB1,
641 0xB0,0x6A,0x6B,0x6C,0x6D,0x6E,0x6F,0x70,0x71,0x72,0xAA,0xBA,0xE6,0xB8,0xC6,0xA4,
642 0xB5,0x7E,0x73,0x74,0x75,0x76,0x77,0x78,0x79,0x7A,0xA1,0xBF,0xD0,0x5B,0xDE,0xAE,
643 0xAC,0xA3,0xA5,0xB7,0xA9,0xA7,0xB6,0xBC,0xBD,0xBE,0xDD,0xA8,0xAF,0x5D,0xB4,0xD7,
644 0x7B,0x41,0x42,0x43,0x44,0x45,0x46,0x47,0x48,0x49,0xAD,0xF4,0xF6,0xF2,0xF3,0xF5,
645 0x7D,0x4A,0x4B,0x4C,0x4D,0x4E,0x4F,0x50,0x51,0x52,0xB9,0xFB,0xFC,0xF9,0xFA,0xFF,
646 0x5C,0xF7,0x53,0x54,0x55,0x56,0x57,0x58,0x59,0x5A,0xB2,0xD4,0xD6,0xD2,0xD3,0xD5,
647 0x30,0x31,0x32,0x33,0x34,0x35,0x36,0x37,0x38,0x39,0xB3,0xDB,0xDC,0xD9,0xDA,0x9F
648 };
649 #endif
650 
651 /*
652  * Copy buf[0 ... nbytes-1] into out[], translating EBCDIC to ASCII.
653  */
654 file_private void
655 from_ebcdic(const unsigned char *buf, size_t nbytes, unsigned char *out)
656 {
657 	size_t i;
658 
659 	for (i = 0; i < nbytes; i++) {
660 		out[i] = ebcdic_to_ascii[buf[i]];
661 	}
662 }
663