xref: /freebsd/contrib/file/src/encoding.c (revision 898496ee09ed2b7d25f6807edc4515628196ec0a)
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 1.42 2022/12/26 17:31:14 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
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)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 file_private char text_chars[256] = {
241 	/*                  BEL BS HT LF VT FF CR    */
242 	F, F, F, F, F, F, F, T, T, T, T, T, T, T, F, F,  /* 0x0X */
243 	/*                              ESC          */
244 	F, F, F, F, F, F, F, F, F, F, F, T, F, F, F, F,  /* 0x1X */
245 	T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, T,  /* 0x2X */
246 	T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, T,  /* 0x3X */
247 	T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, T,  /* 0x4X */
248 	T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, T,  /* 0x5X */
249 	T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, T,  /* 0x6X */
250 	T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, F,  /* 0x7X */
251 	/*            NEL                            */
252 	X, X, X, X, X, T, X, X, X, X, X, X, X, X, X, X,  /* 0x8X */
253 	X, X, X, X, X, X, X, X, X, X, X, X, X, X, X, X,  /* 0x9X */
254 	I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, I,  /* 0xaX */
255 	I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, I,  /* 0xbX */
256 	I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, I,  /* 0xcX */
257 	I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, I,  /* 0xdX */
258 	I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, I,  /* 0xeX */
259 	I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, I   /* 0xfX */
260 };
261 
262 #define LOOKS(NAME, COND) \
263 file_private int \
264 looks_ ## NAME(const unsigned char *buf, size_t nbytes, file_unichar_t *ubuf, \
265     size_t *ulen) \
266 { \
267 	size_t i; \
268 \
269 	*ulen = 0; \
270 \
271 	for (i = 0; i < nbytes; i++) { \
272 		int t = text_chars[buf[i]]; \
273 \
274 		if (COND) \
275 			return 0; \
276 \
277 		ubuf[(*ulen)++] = buf[i]; \
278 	} \
279 	return 1; \
280 }
281 
282 LOOKS(ascii, t != T)
283 LOOKS(latin1, t != T && t != I)
284 LOOKS(extended, t != T && t != I && t != X)
285 
286 /*
287  * Decide whether some text looks like UTF-8. Returns:
288  *
289  *     -1: invalid UTF-8
290  *      0: uses odd control characters, so doesn't look like text
291  *      1: 7-bit text
292  *      2: definitely UTF-8 text (valid high-bit set bytes)
293  *
294  * If ubuf is non-NULL on entry, text is decoded into ubuf, *ulen;
295  * ubuf must be big enough!
296  */
297 
298 // from: https://golang.org/src/unicode/utf8/utf8.go
299 
300 #define	XX 0xF1 // invalid: size 1
301 #define	AS 0xF0 // ASCII: size 1
302 #define	S1 0x02 // accept 0, size 2
303 #define	S2 0x13 // accept 1, size 3
304 #define	S3 0x03 // accept 0, size 3
305 #define	S4 0x23 // accept 2, size 3
306 #define	S5 0x34 // accept 3, size 4
307 #define	S6 0x04 // accept 0, size 4
308 #define	S7 0x44 // accept 4, size 4
309 
310 #define LOCB 0x80
311 #define HICB 0xBF
312 
313 // first is information about the first byte in a UTF-8 sequence.
314 static const uint8_t first[] = {
315     //   1   2   3   4   5   6   7   8   9   A   B   C   D   E   F
316     AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, // 0x00-0x0F
317     AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, // 0x10-0x1F
318     AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, // 0x20-0x2F
319     AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, // 0x30-0x3F
320     AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, // 0x40-0x4F
321     AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, // 0x50-0x5F
322     AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, // 0x60-0x6F
323     AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, AS, // 0x70-0x7F
324     //   1   2   3   4   5   6   7   8   9   A   B   C   D   E   F
325     XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, // 0x80-0x8F
326     XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, // 0x90-0x9F
327     XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, // 0xA0-0xAF
328     XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, // 0xB0-0xBF
329     XX, XX, S1, S1, S1, S1, S1, S1, S1, S1, S1, S1, S1, S1, S1, S1, // 0xC0-0xCF
330     S1, S1, S1, S1, S1, S1, S1, S1, S1, S1, S1, S1, S1, S1, S1, S1, // 0xD0-0xDF
331     S2, S3, S3, S3, S3, S3, S3, S3, S3, S3, S3, S3, S3, S4, S3, S3, // 0xE0-0xEF
332     S5, S6, S6, S6, S7, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, XX, // 0xF0-0xFF
333 };
334 
335 // acceptRange gives the range of valid values for the second byte in a UTF-8
336 // sequence.
337 struct accept_range {
338 	uint8_t lo; // lowest value for second byte.
339 	uint8_t hi; // highest value for second byte.
340 } accept_ranges[16] = {
341 // acceptRanges has size 16 to avoid bounds checks in the code that uses it.
342 	{ LOCB, HICB },
343 	{ 0xA0, HICB },
344 	{ LOCB, 0x9F },
345 	{ 0x90, HICB },
346 	{ LOCB, 0x8F },
347 };
348 
349 file_protected int
file_looks_utf8(const unsigned char * buf,size_t nbytes,file_unichar_t * ubuf,size_t * ulen)350 file_looks_utf8(const unsigned char *buf, size_t nbytes, file_unichar_t *ubuf,
351     size_t *ulen)
352 {
353 	size_t i;
354 	int n;
355 	file_unichar_t c;
356 	int gotone = 0, ctrl = 0;
357 
358 	if (ubuf)
359 		*ulen = 0;
360 
361 	for (i = 0; i < nbytes; i++) {
362 		if ((buf[i] & 0x80) == 0) {	   /* 0xxxxxxx is plain ASCII */
363 			/*
364 			 * Even if the whole file is valid UTF-8 sequences,
365 			 * still reject it if it uses weird control characters.
366 			 */
367 
368 			if (text_chars[buf[i]] != T)
369 				ctrl = 1;
370 
371 			if (ubuf)
372 				ubuf[(*ulen)++] = buf[i];
373 		} else if ((buf[i] & 0x40) == 0) { /* 10xxxxxx never 1st byte */
374 			return -1;
375 		} else {			   /* 11xxxxxx begins UTF-8 */
376 			int following;
377 			uint8_t x = first[buf[i]];
378 			const struct accept_range *ar =
379 			    &accept_ranges[(unsigned int)x >> 4];
380 			if (x == XX)
381 				return -1;
382 
383 			if ((buf[i] & 0x20) == 0) {		/* 110xxxxx */
384 				c = buf[i] & 0x1f;
385 				following = 1;
386 			} else if ((buf[i] & 0x10) == 0) {	/* 1110xxxx */
387 				c = buf[i] & 0x0f;
388 				following = 2;
389 			} else if ((buf[i] & 0x08) == 0) {	/* 11110xxx */
390 				c = buf[i] & 0x07;
391 				following = 3;
392 			} else if ((buf[i] & 0x04) == 0) {	/* 111110xx */
393 				c = buf[i] & 0x03;
394 				following = 4;
395 			} else if ((buf[i] & 0x02) == 0) {	/* 1111110x */
396 				c = buf[i] & 0x01;
397 				following = 5;
398 			} else
399 				return -1;
400 
401 			for (n = 0; n < following; n++) {
402 				i++;
403 				if (i >= nbytes)
404 					goto done;
405 
406 				if (n == 0 &&
407 				     (buf[i] < ar->lo || buf[i] > ar->hi))
408 					return -1;
409 
410 				if ((buf[i] & 0x80) == 0 || (buf[i] & 0x40))
411 					return -1;
412 
413 				c = (c << 6) + (buf[i] & 0x3f);
414 			}
415 
416 			if (ubuf)
417 				ubuf[(*ulen)++] = c;
418 			gotone = 1;
419 		}
420 	}
421 done:
422 	return ctrl ? 0 : (gotone ? 2 : 1);
423 }
424 
425 /*
426  * Decide whether some text looks like UTF-8 with BOM. If there is no
427  * BOM, return -1; otherwise return the result of looks_utf8 on the
428  * rest of the text.
429  */
430 file_private int
looks_utf8_with_BOM(const unsigned char * buf,size_t nbytes,file_unichar_t * ubuf,size_t * ulen)431 looks_utf8_with_BOM(const unsigned char *buf, size_t nbytes,
432     file_unichar_t *ubuf, size_t *ulen)
433 {
434 	if (nbytes > 3 && buf[0] == 0xef && buf[1] == 0xbb && buf[2] == 0xbf)
435 		return file_looks_utf8(buf + 3, nbytes - 3, ubuf, ulen);
436 	else
437 		return -1;
438 }
439 
440 file_private int
looks_utf7(const unsigned char * buf,size_t nbytes,file_unichar_t * ubuf,size_t * ulen)441 looks_utf7(const unsigned char *buf, size_t nbytes, file_unichar_t *ubuf,
442     size_t *ulen)
443 {
444 	if (nbytes > 4 && buf[0] == '+' && buf[1] == '/' && buf[2] == 'v')
445 		switch (buf[3]) {
446 		case '8':
447 		case '9':
448 		case '+':
449 		case '/':
450 			if (ubuf)
451 				*ulen = 0;
452 			return 1;
453 		default:
454 			return -1;
455 		}
456 	else
457 		return -1;
458 }
459 
460 #define UCS16_NOCHAR(c) ((c) >= 0xfdd0 && (c) <= 0xfdef)
461 #define UCS16_HISURR(c) ((c) >= 0xd800 && (c) <= 0xdbff)
462 #define UCS16_LOSURR(c) ((c) >= 0xdc00 && (c) <= 0xdfff)
463 
464 file_private int
looks_ucs16(const unsigned char * bf,size_t nbytes,file_unichar_t * ubf,size_t * ulen)465 looks_ucs16(const unsigned char *bf, size_t nbytes, file_unichar_t *ubf,
466     size_t *ulen)
467 {
468 	int bigend;
469 	uint32_t hi;
470 	size_t i;
471 
472 	if (nbytes < 2)
473 		return 0;
474 
475 	if (bf[0] == 0xff && bf[1] == 0xfe)
476 		bigend = 0;
477 	else if (bf[0] == 0xfe && bf[1] == 0xff)
478 		bigend = 1;
479 	else
480 		return 0;
481 
482 	*ulen = 0;
483 	hi = 0;
484 
485 	for (i = 2; i + 1 < nbytes; i += 2) {
486 		uint32_t uc;
487 
488 		if (bigend)
489 			uc = CAST(uint32_t,
490 			    bf[i + 1] | (CAST(file_unichar_t, bf[i]) << 8));
491 		else
492 			uc = CAST(uint32_t,
493 			    bf[i] | (CAST(file_unichar_t, bf[i + 1]) << 8));
494 
495 		uc &= 0xffff;
496 
497 		switch (uc) {
498 		case 0xfffe:
499 		case 0xffff:
500 			return 0;
501 		default:
502 			if (UCS16_NOCHAR(uc))
503 				return 0;
504 			break;
505 		}
506 		if (hi) {
507 			if (!UCS16_LOSURR(uc))
508 				return 0;
509 			uc = 0x10000 + 0x400 * (hi - 1) + (uc - 0xdc00);
510 			hi = 0;
511 		}
512 		if (uc < 128 && text_chars[CAST(size_t, uc)] != T)
513 			return 0;
514 		ubf[(*ulen)++] = uc;
515 		if (UCS16_HISURR(uc))
516 			hi = uc - 0xd800 + 1;
517 		if (UCS16_LOSURR(uc))
518 			return 0;
519 	}
520 
521 	return 1 + bigend;
522 }
523 
524 file_private int
looks_ucs32(const unsigned char * bf,size_t nbytes,file_unichar_t * ubf,size_t * ulen)525 looks_ucs32(const unsigned char *bf, size_t nbytes, file_unichar_t *ubf,
526     size_t *ulen)
527 {
528 	int bigend;
529 	size_t i;
530 
531 	if (nbytes < 4)
532 		return 0;
533 
534 	if (bf[0] == 0xff && bf[1] == 0xfe && bf[2] == 0 && bf[3] == 0)
535 		bigend = 0;
536 	else if (bf[0] == 0 && bf[1] == 0 && bf[2] == 0xfe && bf[3] == 0xff)
537 		bigend = 1;
538 	else
539 		return 0;
540 
541 	*ulen = 0;
542 
543 	for (i = 4; i + 3 < nbytes; i += 4) {
544 		/* XXX fix to properly handle chars > 65536 */
545 
546 		if (bigend)
547 			ubf[(*ulen)++] = CAST(file_unichar_t, bf[i + 3])
548 			    | (CAST(file_unichar_t, bf[i + 2]) << 8)
549 			    | (CAST(file_unichar_t, bf[i + 1]) << 16)
550 			    | (CAST(file_unichar_t, bf[i]) << 24);
551 		else
552 			ubf[(*ulen)++] = CAST(file_unichar_t, bf[i + 0])
553 			    | (CAST(file_unichar_t, bf[i + 1]) << 8)
554 			    | (CAST(file_unichar_t, bf[i + 2]) << 16)
555 			    | (CAST(file_unichar_t, bf[i + 3]) << 24);
556 
557 		if (ubf[*ulen - 1] == 0xfffe)
558 			return 0;
559 		if (ubf[*ulen - 1] < 128 &&
560 		    text_chars[CAST(size_t, ubf[*ulen - 1])] != T)
561 			return 0;
562 	}
563 
564 	return 1 + bigend;
565 }
566 #undef F
567 #undef T
568 #undef I
569 #undef X
570 
571 /*
572  * This table maps each EBCDIC character to an (8-bit extended) ASCII
573  * character, as specified in the rationale for the dd(1) command in
574  * draft 11.2 (September, 1991) of the POSIX P1003.2 standard.
575  *
576  * Unfortunately it does not seem to correspond exactly to any of the
577  * five variants of EBCDIC documented in IBM's _Enterprise Systems
578  * Architecture/390: Principles of Operation_, SA22-7201-06, Seventh
579  * Edition, July, 1999, pp. I-1 - I-4.
580  *
581  * Fortunately, though, all versions of EBCDIC, including this one, agree
582  * on most of the printing characters that also appear in (7-bit) ASCII.
583  * Of these, only '|', '!', '~', '^', '[', and ']' are in question at all.
584  *
585  * Fortunately too, there is general agreement that codes 0x00 through
586  * 0x3F represent control characters, 0x41 a nonbreaking space, and the
587  * remainder printing characters.
588  *
589  * This is sufficient to allow us to identify EBCDIC text and to distinguish
590  * between old-style and internationalized examples of text.
591  */
592 
593 file_private unsigned char ebcdic_to_ascii[] = {
594   0,   1,   2,   3, 156,   9, 134, 127, 151, 141, 142,  11,  12,  13,  14,  15,
595  16,  17,  18,  19, 157, 133,   8, 135,  24,  25, 146, 143,  28,  29,  30,  31,
596 128, 129, 130, 131, 132,  10,  23,  27, 136, 137, 138, 139, 140,   5,   6,   7,
597 144, 145,  22, 147, 148, 149, 150,   4, 152, 153, 154, 155,  20,  21, 158,  26,
598 ' ', 160, 161, 162, 163, 164, 165, 166, 167, 168, 213, '.', '<', '(', '+', '|',
599 '&', 169, 170, 171, 172, 173, 174, 175, 176, 177, '!', '$', '*', ')', ';', '~',
600 '-', '/', 178, 179, 180, 181, 182, 183, 184, 185, 203, ',', '%', '_', '>', '?',
601 186, 187, 188, 189, 190, 191, 192, 193, 194, '`', ':', '#', '@', '\'','=', '"',
602 195, 'a', 'b', 'c', 'd', 'e', 'f', 'g', 'h', 'i', 196, 197, 198, 199, 200, 201,
603 202, 'j', 'k', 'l', 'm', 'n', 'o', 'p', 'q', 'r', '^', 204, 205, 206, 207, 208,
604 209, 229, 's', 't', 'u', 'v', 'w', 'x', 'y', 'z', 210, 211, 212, '[', 214, 215,
605 216, 217, 218, 219, 220, 221, 222, 223, 224, 225, 226, 227, 228, ']', 230, 231,
606 '{', 'A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', 'I', 232, 233, 234, 235, 236, 237,
607 '}', 'J', 'K', 'L', 'M', 'N', 'O', 'P', 'Q', 'R', 238, 239, 240, 241, 242, 243,
608 '\\',159, 'S', 'T', 'U', 'V', 'W', 'X', 'Y', 'Z', 244, 245, 246, 247, 248, 249,
609 '0', '1', '2', '3', '4', '5', '6', '7', '8', '9', 250, 251, 252, 253, 254, 255
610 };
611 
612 #ifdef notdef
613 /*
614  * The following EBCDIC-to-ASCII table may relate more closely to reality,
615  * or at least to modern reality.  It comes from
616  *
617  *   http://ftp.s390.ibm.com/products/oe/bpxqp9.html
618  *
619  * and maps the characters of EBCDIC code page 1047 (the code used for
620  * Unix-derived software on IBM's 390 systems) to the corresponding
621  * characters from ISO 8859-1.
622  *
623  * If this table is used instead of the above one, some of the special
624  * cases for the NEL character can be taken out of the code.
625  */
626 
627 file_private unsigned char ebcdic_1047_to_8859[] = {
628 0x00,0x01,0x02,0x03,0x9C,0x09,0x86,0x7F,0x97,0x8D,0x8E,0x0B,0x0C,0x0D,0x0E,0x0F,
629 0x10,0x11,0x12,0x13,0x9D,0x0A,0x08,0x87,0x18,0x19,0x92,0x8F,0x1C,0x1D,0x1E,0x1F,
630 0x80,0x81,0x82,0x83,0x84,0x85,0x17,0x1B,0x88,0x89,0x8A,0x8B,0x8C,0x05,0x06,0x07,
631 0x90,0x91,0x16,0x93,0x94,0x95,0x96,0x04,0x98,0x99,0x9A,0x9B,0x14,0x15,0x9E,0x1A,
632 0x20,0xA0,0xE2,0xE4,0xE0,0xE1,0xE3,0xE5,0xE7,0xF1,0xA2,0x2E,0x3C,0x28,0x2B,0x7C,
633 0x26,0xE9,0xEA,0xEB,0xE8,0xED,0xEE,0xEF,0xEC,0xDF,0x21,0x24,0x2A,0x29,0x3B,0x5E,
634 0x2D,0x2F,0xC2,0xC4,0xC0,0xC1,0xC3,0xC5,0xC7,0xD1,0xA6,0x2C,0x25,0x5F,0x3E,0x3F,
635 0xF8,0xC9,0xCA,0xCB,0xC8,0xCD,0xCE,0xCF,0xCC,0x60,0x3A,0x23,0x40,0x27,0x3D,0x22,
636 0xD8,0x61,0x62,0x63,0x64,0x65,0x66,0x67,0x68,0x69,0xAB,0xBB,0xF0,0xFD,0xFE,0xB1,
637 0xB0,0x6A,0x6B,0x6C,0x6D,0x6E,0x6F,0x70,0x71,0x72,0xAA,0xBA,0xE6,0xB8,0xC6,0xA4,
638 0xB5,0x7E,0x73,0x74,0x75,0x76,0x77,0x78,0x79,0x7A,0xA1,0xBF,0xD0,0x5B,0xDE,0xAE,
639 0xAC,0xA3,0xA5,0xB7,0xA9,0xA7,0xB6,0xBC,0xBD,0xBE,0xDD,0xA8,0xAF,0x5D,0xB4,0xD7,
640 0x7B,0x41,0x42,0x43,0x44,0x45,0x46,0x47,0x48,0x49,0xAD,0xF4,0xF6,0xF2,0xF3,0xF5,
641 0x7D,0x4A,0x4B,0x4C,0x4D,0x4E,0x4F,0x50,0x51,0x52,0xB9,0xFB,0xFC,0xF9,0xFA,0xFF,
642 0x5C,0xF7,0x53,0x54,0x55,0x56,0x57,0x58,0x59,0x5A,0xB2,0xD4,0xD6,0xD2,0xD3,0xD5,
643 0x30,0x31,0x32,0x33,0x34,0x35,0x36,0x37,0x38,0x39,0xB3,0xDB,0xDC,0xD9,0xDA,0x9F
644 };
645 #endif
646 
647 /*
648  * Copy buf[0 ... nbytes-1] into out[], translating EBCDIC to ASCII.
649  */
650 file_private void
from_ebcdic(const unsigned char * buf,size_t nbytes,unsigned char * out)651 from_ebcdic(const unsigned char *buf, size_t nbytes, unsigned char *out)
652 {
653 	size_t i;
654 
655 	for (i = 0; i < nbytes; i++) {
656 		out[i] = ebcdic_to_ascii[buf[i]];
657 	}
658 }
659