xref: /haiku/src/kits/storage/mime/TextSnifferAddon.cpp (revision 35c76fab52eff4d5462c94fc4811c66445e4fc9b)
1 /*
2  * Copyright 2006-2013, Ingo Weinhold, ingo_weinhold@gmx.de.
3  * Distributed under the terms of the MIT License.
4  */
5 
6 
7 #include <mime/TextSnifferAddon.h>
8 
9 #include <MimeType.h>
10 
11 #include <mime/DatabaseLocation.h>
12 
13 
14 using BPrivate::Storage::Mime::DatabaseLocation;
15 
16 
17 static int file_ascmagic(DatabaseLocation* databaseLocation,
18 	const unsigned char *buf, size_t nbytes, BMimeType* mimeType);
19 
20 
21 namespace BPrivate {
22 namespace Storage {
23 namespace Mime {
24 
25 
26 // constructor
TextSnifferAddon(DatabaseLocation * databaseLocation)27 TextSnifferAddon::TextSnifferAddon(DatabaseLocation* databaseLocation)
28 	:
29 	fDatabaseLocation(databaseLocation)
30 {
31 }
32 
33 // destructor
~TextSnifferAddon()34 TextSnifferAddon::~TextSnifferAddon()
35 {
36 }
37 
38 // MinimalBufferSize
39 size_t
MinimalBufferSize()40 TextSnifferAddon::MinimalBufferSize()
41 {
42 	return 512;
43 }
44 
45 // GuessMimeType
46 float
GuessMimeType(const char * fileName,BMimeType * type)47 TextSnifferAddon::GuessMimeType(const char* fileName, BMimeType* type)
48 {
49 	// we check content only
50 	return -1;
51 }
52 
53 // GuessMimeType
54 float
GuessMimeType(BFile * file,const void * buffer,int32 length,BMimeType * type)55 TextSnifferAddon::GuessMimeType(BFile* file, const void* buffer, int32 length,
56 	BMimeType* type)
57 {
58 	if (file_ascmagic(fDatabaseLocation, (const unsigned char*)buffer, length,
59 			type)) {
60 		// If the buffer is very short, we return a lower priority. Maybe
61 		// someone else knows better.
62 		if (length < 20)
63 			return .0f;
64 		return 0.25f;
65 	}
66 
67 	return -1;
68 }
69 
70 
71 } // namespace Mime
72 } // namespace Storage
73 } // namespace BPrivate
74 
75 
76 // #pragma mark - ascmagic.c from the BSD file tool
77 /*
78  * The following code has been taken from version 4.17 of the BSD file tool,
79  * file ascmagic.c, modified for our purpose.
80  */
81 
82 /*
83  * Copyright (c) Ian F. Darwin 1986-1995.
84  * Software written by Ian F. Darwin and others;
85  * maintained 1995-present by Christos Zoulas and others.
86  *
87  * Redistribution and use in source and binary forms, with or without
88  * modification, are permitted provided that the following conditions
89  * are met:
90  * 1. Redistributions of source code must retain the above copyright
91  *    notice immediately at the beginning of the file, without modification,
92  *    this list of conditions, and the following disclaimer.
93  * 2. Redistributions in binary form must reproduce the above copyright
94  *    notice, this list of conditions and the following disclaimer in the
95  *    documentation and/or other materials provided with the distribution.
96  *
97  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
98  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
99  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
100  * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE FOR
101  * ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
102  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
103  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
104  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
105  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
106  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
107  * SUCH DAMAGE.
108  */
109 /*
110  * ASCII magic -- file types that we know based on keywords
111  * that can appear anywhere in the file.
112  *
113  * Extensively modified by Eric Fischer <enf@pobox.com> in July, 2000,
114  * to handle character codes other than ASCII on a unified basis.
115  *
116  * Joerg Wunsch <joerg@freebsd.org> wrote the original support for 8-bit
117  * international characters, now subsumed into this file.
118  */
119 
120 #include <stdio.h>
121 #include <string.h>
122 #include <memory.h>
123 #include <ctype.h>
124 #include <stdlib.h>
125 #include <unistd.h>
126 #include "names.h"
127 
128 typedef unsigned long my_unichar;
129 
130 #define MAXLINELEN 300	/* longest sane line length */
131 #define ISSPC(x) ((x) == ' ' || (x) == '\t' || (x) == '\r' || (x) == '\n' \
132 		  || (x) == 0x85 || (x) == '\f')
133 
134 static int looks_ascii(const unsigned char *, size_t, my_unichar *, size_t *);
135 static int looks_utf8(const unsigned char *, size_t, my_unichar *, size_t *);
136 static int looks_unicode(const unsigned char *, size_t, my_unichar *, size_t *);
137 static int looks_latin1(const unsigned char *, size_t, my_unichar *, size_t *);
138 static int looks_extended(const unsigned char *, size_t, my_unichar *, size_t *);
139 static void from_ebcdic(const unsigned char *, size_t, unsigned char *);
140 static int ascmatch(const unsigned char *, const my_unichar *, size_t);
141 
142 
143 static int
file_ascmagic(DatabaseLocation * databaseLocation,const unsigned char * buf,size_t nbytes,BMimeType * mimeType)144 file_ascmagic(DatabaseLocation* databaseLocation, const unsigned char *buf,
145 	size_t nbytes, BMimeType* mimeType)
146 {
147 	size_t i;
148 	unsigned char *nbuf = NULL;
149 	my_unichar *ubuf = NULL;
150 	size_t ulen;
151 	struct names *p;
152 	int rv = -1;
153 
154 	const char *code = NULL;
155 	const char *code_mime = NULL;
156 	const char *type = NULL;
157 	const char *subtype = NULL;
158 	const char *subtypeMimeGeneric = NULL;
159 	const char *subtypeMimeSpecific = NULL;
160 
161 	int has_escapes = 0;
162 	int has_backspace = 0;
163 	int seen_cr = 0;
164 
165 	int n_crlf = 0;
166 	int n_lf = 0;
167 	int n_cr = 0;
168 	int n_nel = 0;
169 
170 	int last_line_end = -1;
171 	int has_long_lines = 0;
172 
173 	if ((nbuf = (unsigned char*)malloc((nbytes + 1) * sizeof(nbuf[0]))) == NULL)
174 		goto done;
175 	if ((ubuf = (my_unichar*)malloc((nbytes + 1) * sizeof(ubuf[0]))) == NULL)
176 		goto done;
177 
178 	/*
179 	 * Then try to determine whether it's any character code we can
180 	 * identify.  Each of these tests, if it succeeds, will leave
181 	 * the text converted into one-my_unichar-per-character Unicode in
182 	 * ubuf, and the number of characters converted in ulen.
183 	 */
184 	if (looks_ascii(buf, nbytes, ubuf, &ulen)) {
185 		code = "ASCII";
186 		code_mime = "us-ascii";
187 		type = "text";
188 	} else if (looks_utf8(buf, nbytes, ubuf, &ulen)) {
189 		code = "UTF-8 Unicode";
190 		code_mime = "utf-8";
191 		type = "text";
192 	} else if ((i = looks_unicode(buf, nbytes, ubuf, &ulen)) != 0) {
193 		if (i == 1)
194 			code = "Little-endian UTF-16 Unicode";
195 		else
196 			code = "Big-endian UTF-16 Unicode";
197 
198 		type = "character data";
199 		code_mime = "utf-16";    /* is this defined? */
200 	} else if (looks_latin1(buf, nbytes, ubuf, &ulen)) {
201 		code = "ISO-8859";
202 		type = "text";
203 		code_mime = "iso-8859-1";
204 	} else if (looks_extended(buf, nbytes, ubuf, &ulen)) {
205 		code = "Non-ISO extended-ASCII";
206 		type = "text";
207 		code_mime = "unknown";
208 	} else {
209 		from_ebcdic(buf, nbytes, nbuf);
210 
211 		if (looks_ascii(nbuf, nbytes, ubuf, &ulen)) {
212 			code = "EBCDIC";
213 			type = "character data";
214 			code_mime = "ebcdic";
215 		} else if (looks_latin1(nbuf, nbytes, ubuf, &ulen)) {
216 			code = "International EBCDIC";
217 			type = "character data";
218 			code_mime = "ebcdic";
219 		} else {
220 			rv = 0;
221 			goto done;  /* doesn't look like text at all */
222 		}
223 	}
224 
225 	if (nbytes <= 1) {
226 		rv = 0;
227 		goto done;
228 	}
229 
230 	/*
231 	 * for troff, look for . + letter + letter or .\";
232 	 * this must be done to disambiguate tar archives' ./file
233 	 * and other trash from real troff input.
234 	 *
235 	 * I believe Plan 9 troff allows non-ASCII characters in the names
236 	 * of macros, so this test might possibly fail on such a file.
237 	 */
238 	if (*ubuf == '.') {
239 		my_unichar *tp = ubuf + 1;
240 
241 		while (ISSPC(*tp))
242 			++tp;	/* skip leading whitespace */
243 		if ((tp[0] == '\\' && tp[1] == '\"') ||
244 		    (isascii((unsigned char)tp[0]) &&
245 		     isalnum((unsigned char)tp[0]) &&
246 		     isascii((unsigned char)tp[1]) &&
247 		     isalnum((unsigned char)tp[1]) &&
248 		     ISSPC(tp[2]))) {
249 		    subtypeMimeGeneric = "text/x-source-code";
250 			subtypeMimeSpecific = "text/troff";
251 			subtype = "troff or preprocessor input";
252 			goto subtype_identified;
253 		}
254 	}
255 
256 	if ((*buf == 'c' || *buf == 'C') && ISSPC(buf[1])) {
257 		subtypeMimeGeneric = "text/x-source-code";
258 		subtypeMimeSpecific = "text/fortran";
259 		subtype = "fortran program";
260 		goto subtype_identified;
261 	}
262 
263 	/* look for tokens from names.h - this is expensive! */
264 
265 	i = 0;
266 	while (i < ulen) {
267 		size_t end;
268 
269 		/*
270 		 * skip past any leading space
271 		 */
272 		while (i < ulen && ISSPC(ubuf[i]))
273 			i++;
274 		if (i >= ulen)
275 			break;
276 
277 		/*
278 		 * find the next whitespace
279 		 */
280 		for (end = i + 1; end < nbytes; end++)
281 			if (ISSPC(ubuf[end]))
282 				break;
283 
284 		/*
285 		 * compare the word thus isolated against the token list
286 		 */
287 		for (p = names; p < names + NNAMES; p++) {
288 			if (ascmatch((const unsigned char *)p->name, ubuf + i,
289 			    end - i)) {
290 				subtype = types[p->type].human;
291 				subtypeMimeGeneric = types[p->type].generic_mime;
292 				subtypeMimeSpecific = types[p->type].specific_mime;
293 				goto subtype_identified;
294 			}
295 		}
296 
297 		i = end;
298 	}
299 
300 	(void)code;
301 	(void)code_mime;
302 	(void)type;
303 	(void)subtype;
304 	(void)has_escapes;
305 	(void)has_backspace;
306 	(void)has_long_lines;
307 
308 subtype_identified:
309 
310 	/*
311 	 * Now try to discover other details about the file.
312 	 */
313 	for (i = 0; i < ulen; i++) {
314 		if (ubuf[i] == '\n') {
315 			if (seen_cr)
316 				n_crlf++;
317 			else
318 				n_lf++;
319 			last_line_end = i;
320 		} else if (seen_cr)
321 			n_cr++;
322 
323 		seen_cr = (ubuf[i] == '\r');
324 		if (seen_cr)
325 			last_line_end = i;
326 
327 		if (ubuf[i] == 0x85) { /* X3.64/ECMA-43 "next line" character */
328 			n_nel++;
329 			last_line_end = i;
330 		}
331 
332 		/* If this line is _longer_ than MAXLINELEN, remember it. */
333 		if ((int)i > last_line_end + MAXLINELEN)
334 			has_long_lines = 1;
335 
336 		if (ubuf[i] == '\033')
337 			has_escapes = 1;
338 		if (ubuf[i] == '\b')
339 			has_backspace = 1;
340 	}
341 
342 	rv = 1;
343 done:
344 	if (nbuf)
345 		free(nbuf);
346 	if (ubuf)
347 		free(ubuf);
348 
349 	if (rv) {
350 		// If we have identified the subtype, return it, otherwise just
351 		// text/plain.
352 
353 		bool found = false;
354 		if (subtypeMimeSpecific != NULL) {
355 			if (databaseLocation->IsInstalled(subtypeMimeSpecific)) {
356 				mimeType->SetTo(subtypeMimeSpecific);
357 				found = true;
358 			}
359 		}
360 		if (!found && subtypeMimeGeneric != NULL) {
361 			if (databaseLocation->IsInstalled(subtypeMimeGeneric)) {
362 				mimeType->SetTo(subtypeMimeGeneric);
363 				found = true;
364 			}
365 		}
366 		if (!found)
367 			mimeType->SetTo("text/plain");
368 	}
369 
370 	return rv;
371 }
372 
373 static int
ascmatch(const unsigned char * s,const my_unichar * us,size_t ulen)374 ascmatch(const unsigned char *s, const my_unichar *us, size_t ulen)
375 {
376 	size_t i;
377 
378 	for (i = 0; i < ulen; i++) {
379 		if (s[i] != us[i])
380 			return 0;
381 	}
382 
383 	if (s[i])
384 		return 0;
385 	else
386 		return 1;
387 }
388 
389 /*
390  * This table reflects a particular philosophy about what constitutes
391  * "text," and there is room for disagreement about it.
392  *
393  * Version 3.31 of the file command considered a file to be ASCII if
394  * each of its characters was approved by either the isascii() or
395  * isalpha() function.  On most systems, this would mean that any
396  * file consisting only of characters in the range 0x00 ... 0x7F
397  * would be called ASCII text, but many systems might reasonably
398  * consider some characters outside this range to be alphabetic,
399  * so the file command would call such characters ASCII.  It might
400  * have been more accurate to call this "considered textual on the
401  * local system" than "ASCII."
402  *
403  * It considered a file to be "International language text" if each
404  * of its characters was either an ASCII printing character (according
405  * to the real ASCII standard, not the above test), a character in
406  * the range 0x80 ... 0xFF, or one of the following control characters:
407  * backspace, tab, line feed, vertical tab, form feed, carriage return,
408  * escape.  No attempt was made to determine the language in which files
409  * of this type were written.
410  *
411  *
412  * The table below considers a file to be ASCII if all of its characters
413  * are either ASCII printing characters (again, according to the X3.4
414  * standard, not isascii()) or any of the following controls: bell,
415  * backspace, tab, line feed, form feed, carriage return, esc, nextline.
416  *
417  * I include bell because some programs (particularly shell scripts)
418  * use it literally, even though it is rare in normal text.  I exclude
419  * vertical tab because it never seems to be used in real text.  I also
420  * include, with hesitation, the X3.64/ECMA-43 control nextline (0x85),
421  * because that's what the dd EBCDIC->ASCII table maps the EBCDIC newline
422  * character to.  It might be more appropriate to include it in the 8859
423  * set instead of the ASCII set, but it's got to be included in *something*
424  * we recognize or EBCDIC files aren't going to be considered textual.
425  * Some old Unix source files use SO/SI (^N/^O) to shift between Greek
426  * and Latin characters, so these should possibly be allowed.  But they
427  * make a real mess on VT100-style displays if they're not paired properly,
428  * so we are probably better off not calling them text.
429  *
430  * A file is considered to be ISO-8859 text if its characters are all
431  * either ASCII, according to the above definition, or printing characters
432  * from the ISO-8859 8-bit extension, characters 0xA0 ... 0xFF.
433  *
434  * Finally, a file is considered to be international text from some other
435  * character code if its characters are all either ISO-8859 (according to
436  * the above definition) or characters in the range 0x80 ... 0x9F, which
437  * ISO-8859 considers to be control characters but the IBM PC and Macintosh
438  * consider to be printing characters.
439  */
440 
441 #define F 0   /* character never appears in text */
442 #define T 1   /* character appears in plain ASCII text */
443 #define I 2   /* character appears in ISO-8859 text */
444 #define X 3   /* character appears in non-ISO extended ASCII (Mac, IBM PC) */
445 
446 static char text_chars[256] = {
447 	/*                  BEL BS HT LF    FF CR    */
448 	F, F, F, F, F, F, F, T, T, T, T, F, T, T, F, F,  /* 0x0X */
449         /*                              ESC          */
450 	F, F, F, F, F, F, F, F, F, F, F, T, F, F, F, F,  /* 0x1X */
451 	T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, T,  /* 0x2X */
452 	T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, T,  /* 0x3X */
453 	T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, T,  /* 0x4X */
454 	T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, T,  /* 0x5X */
455 	T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, T,  /* 0x6X */
456 	T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, F,  /* 0x7X */
457 	/*            NEL                            */
458 	X, X, X, X, X, T, X, X, X, X, X, X, X, X, X, X,  /* 0x8X */
459 	X, X, X, X, X, X, X, X, X, X, X, X, X, X, X, X,  /* 0x9X */
460 	I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, I,  /* 0xaX */
461 	I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, I,  /* 0xbX */
462 	I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, I,  /* 0xcX */
463 	I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, I,  /* 0xdX */
464 	I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, I,  /* 0xeX */
465 	I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, I   /* 0xfX */
466 };
467 
468 static int
looks_ascii(const unsigned char * buf,size_t nbytes,my_unichar * ubuf,size_t * ulen)469 looks_ascii(const unsigned char *buf, size_t nbytes, my_unichar *ubuf,
470     size_t *ulen)
471 {
472 	int i;
473 
474 	*ulen = 0;
475 
476 	for (i = 0; i < (int)nbytes; i++) {
477 		int t = text_chars[buf[i]];
478 
479 		if (t != T)
480 			return 0;
481 
482 		ubuf[(*ulen)++] = buf[i];
483 	}
484 
485 	return 1;
486 }
487 
488 static int
looks_latin1(const unsigned char * buf,size_t nbytes,my_unichar * ubuf,size_t * ulen)489 looks_latin1(const unsigned char *buf, size_t nbytes, my_unichar *ubuf, size_t *ulen)
490 {
491 	int i;
492 
493 	*ulen = 0;
494 
495 	for (i = 0; i < (int)nbytes; i++) {
496 		int t = text_chars[buf[i]];
497 
498 		if (t != T && t != I)
499 			return 0;
500 
501 		ubuf[(*ulen)++] = buf[i];
502 	}
503 
504 	return 1;
505 }
506 
507 static int
looks_extended(const unsigned char * buf,size_t nbytes,my_unichar * ubuf,size_t * ulen)508 looks_extended(const unsigned char *buf, size_t nbytes, my_unichar *ubuf,
509     size_t *ulen)
510 {
511 	int i;
512 
513 	*ulen = 0;
514 
515 	for (i = 0; i < (int)nbytes; i++) {
516 		int t = text_chars[buf[i]];
517 
518 		if (t != T && t != I && t != X)
519 			return 0;
520 
521 		ubuf[(*ulen)++] = buf[i];
522 	}
523 
524 	return 1;
525 }
526 
527 static int
looks_utf8(const unsigned char * buf,size_t nbytes,my_unichar * ubuf,size_t * ulen)528 looks_utf8(const unsigned char *buf, size_t nbytes, my_unichar *ubuf, size_t *ulen)
529 {
530 	int i, n;
531 	my_unichar c;
532 	int gotone = 0;
533 
534 	*ulen = 0;
535 
536 	for (i = 0; i < (int)nbytes; i++) {
537 		if ((buf[i] & 0x80) == 0) {	   /* 0xxxxxxx is plain ASCII */
538 			/*
539 			 * Even if the whole file is valid UTF-8 sequences,
540 			 * still reject it if it uses weird control characters.
541 			 */
542 
543 			if (text_chars[buf[i]] != T)
544 				return 0;
545 
546 			ubuf[(*ulen)++] = buf[i];
547 		} else if ((buf[i] & 0x40) == 0) { /* 10xxxxxx never 1st byte */
548 			return 0;
549 		} else {			   /* 11xxxxxx begins UTF-8 */
550 			int following;
551 
552 			if ((buf[i] & 0x20) == 0) {		/* 110xxxxx */
553 				c = buf[i] & 0x1f;
554 				following = 1;
555 			} else if ((buf[i] & 0x10) == 0) {	/* 1110xxxx */
556 				c = buf[i] & 0x0f;
557 				following = 2;
558 			} else if ((buf[i] & 0x08) == 0) {	/* 11110xxx */
559 				c = buf[i] & 0x07;
560 				following = 3;
561 			} else if ((buf[i] & 0x04) == 0) {	/* 111110xx */
562 				c = buf[i] & 0x03;
563 				following = 4;
564 			} else if ((buf[i] & 0x02) == 0) {	/* 1111110x */
565 				c = buf[i] & 0x01;
566 				following = 5;
567 			} else
568 				return 0;
569 
570 			for (n = 0; n < following; n++) {
571 				i++;
572 				if (i >= (int)nbytes)
573 					goto done;
574 
575 				if ((buf[i] & 0x80) == 0 || (buf[i] & 0x40))
576 					return 0;
577 
578 				c = (c << 6) + (buf[i] & 0x3f);
579 			}
580 
581 			ubuf[(*ulen)++] = c;
582 			gotone = 1;
583 		}
584 	}
585 done:
586 	return gotone;   /* don't claim it's UTF-8 if it's all 7-bit */
587 }
588 
589 static int
looks_unicode(const unsigned char * buf,size_t nbytes,my_unichar * ubuf,size_t * ulen)590 looks_unicode(const unsigned char *buf, size_t nbytes, my_unichar *ubuf,
591     size_t *ulen)
592 {
593 	int bigend;
594 	int i;
595 
596 	if (nbytes < 2)
597 		return 0;
598 
599 	if (buf[0] == 0xff && buf[1] == 0xfe)
600 		bigend = 0;
601 	else if (buf[0] == 0xfe && buf[1] == 0xff)
602 		bigend = 1;
603 	else
604 		return 0;
605 
606 	*ulen = 0;
607 
608 	for (i = 2; i + 1 < (int)nbytes; i += 2) {
609 		/* XXX fix to properly handle chars > 65536 */
610 
611 		if (bigend)
612 			ubuf[(*ulen)++] = buf[i + 1] + 256 * buf[i];
613 		else
614 			ubuf[(*ulen)++] = buf[i] + 256 * buf[i + 1];
615 
616 		if (ubuf[*ulen - 1] == 0xfffe)
617 			return 0;
618 		if (ubuf[*ulen - 1] < 128 &&
619 		    text_chars[(size_t)ubuf[*ulen - 1]] != T)
620 			return 0;
621 	}
622 
623 	return 1 + bigend;
624 }
625 
626 #undef F
627 #undef T
628 #undef I
629 #undef X
630 
631 /*
632  * This table maps each EBCDIC character to an (8-bit extended) ASCII
633  * character, as specified in the rationale for the dd(1) command in
634  * draft 11.2 (September, 1991) of the POSIX P1003.2 standard.
635  *
636  * Unfortunately it does not seem to correspond exactly to any of the
637  * five variants of EBCDIC documented in IBM's _Enterprise Systems
638  * Architecture/390: Principles of Operation_, SA22-7201-06, Seventh
639  * Edition, July, 1999, pp. I-1 - I-4.
640  *
641  * Fortunately, though, all versions of EBCDIC, including this one, agree
642  * on most of the printing characters that also appear in (7-bit) ASCII.
643  * Of these, only '|', '!', '~', '^', '[', and ']' are in question at all.
644  *
645  * Fortunately too, there is general agreement that codes 0x00 through
646  * 0x3F represent control characters, 0x41 a nonbreaking space, and the
647  * remainder printing characters.
648  *
649  * This is sufficient to allow us to identify EBCDIC text and to distinguish
650  * between old-style and internationalized examples of text.
651  */
652 
653 static unsigned char ebcdic_to_ascii[] = {
654   0,   1,   2,   3, 156,   9, 134, 127, 151, 141, 142,  11,  12,  13,  14,  15,
655  16,  17,  18,  19, 157, 133,   8, 135,  24,  25, 146, 143,  28,  29,  30,  31,
656 128, 129, 130, 131, 132,  10,  23,  27, 136, 137, 138, 139, 140,   5,   6,   7,
657 144, 145,  22, 147, 148, 149, 150,   4, 152, 153, 154, 155,  20,  21, 158,  26,
658 ' ', 160, 161, 162, 163, 164, 165, 166, 167, 168, 213, '.', '<', '(', '+', '|',
659 '&', 169, 170, 171, 172, 173, 174, 175, 176, 177, '!', '$', '*', ')', ';', '~',
660 '-', '/', 178, 179, 180, 181, 182, 183, 184, 185, 203, ',', '%', '_', '>', '?',
661 186, 187, 188, 189, 190, 191, 192, 193, 194, '`', ':', '#', '@', '\'','=', '"',
662 195, 'a', 'b', 'c', 'd', 'e', 'f', 'g', 'h', 'i', 196, 197, 198, 199, 200, 201,
663 202, 'j', 'k', 'l', 'm', 'n', 'o', 'p', 'q', 'r', '^', 204, 205, 206, 207, 208,
664 209, 229, 's', 't', 'u', 'v', 'w', 'x', 'y', 'z', 210, 211, 212, '[', 214, 215,
665 216, 217, 218, 219, 220, 221, 222, 223, 224, 225, 226, 227, 228, ']', 230, 231,
666 '{', 'A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', 'I', 232, 233, 234, 235, 236, 237,
667 '}', 'J', 'K', 'L', 'M', 'N', 'O', 'P', 'Q', 'R', 238, 239, 240, 241, 242, 243,
668 '\\',159, 'S', 'T', 'U', 'V', 'W', 'X', 'Y', 'Z', 244, 245, 246, 247, 248, 249,
669 '0', '1', '2', '3', '4', '5', '6', '7', '8', '9', 250, 251, 252, 253, 254, 255
670 };
671 
672 #ifdef notdef
673 /*
674  * The following EBCDIC-to-ASCII table may relate more closely to reality,
675  * or at least to modern reality.  It comes from
676  *
677  *   http://ftp.s390.ibm.com/products/oe/bpxqp9.html
678  *
679  * and maps the characters of EBCDIC code page 1047 (the code used for
680  * Unix-derived software on IBM's 390 systems) to the corresponding
681  * characters from ISO 8859-1.
682  *
683  * If this table is used instead of the above one, some of the special
684  * cases for the NEL character can be taken out of the code.
685  */
686 
687 static unsigned char ebcdic_1047_to_8859[] = {
688 0x00,0x01,0x02,0x03,0x9C,0x09,0x86,0x7F,0x97,0x8D,0x8E,0x0B,0x0C,0x0D,0x0E,0x0F,
689 0x10,0x11,0x12,0x13,0x9D,0x0A,0x08,0x87,0x18,0x19,0x92,0x8F,0x1C,0x1D,0x1E,0x1F,
690 0x80,0x81,0x82,0x83,0x84,0x85,0x17,0x1B,0x88,0x89,0x8A,0x8B,0x8C,0x05,0x06,0x07,
691 0x90,0x91,0x16,0x93,0x94,0x95,0x96,0x04,0x98,0x99,0x9A,0x9B,0x14,0x15,0x9E,0x1A,
692 0x20,0xA0,0xE2,0xE4,0xE0,0xE1,0xE3,0xE5,0xE7,0xF1,0xA2,0x2E,0x3C,0x28,0x2B,0x7C,
693 0x26,0xE9,0xEA,0xEB,0xE8,0xED,0xEE,0xEF,0xEC,0xDF,0x21,0x24,0x2A,0x29,0x3B,0x5E,
694 0x2D,0x2F,0xC2,0xC4,0xC0,0xC1,0xC3,0xC5,0xC7,0xD1,0xA6,0x2C,0x25,0x5F,0x3E,0x3F,
695 0xF8,0xC9,0xCA,0xCB,0xC8,0xCD,0xCE,0xCF,0xCC,0x60,0x3A,0x23,0x40,0x27,0x3D,0x22,
696 0xD8,0x61,0x62,0x63,0x64,0x65,0x66,0x67,0x68,0x69,0xAB,0xBB,0xF0,0xFD,0xFE,0xB1,
697 0xB0,0x6A,0x6B,0x6C,0x6D,0x6E,0x6F,0x70,0x71,0x72,0xAA,0xBA,0xE6,0xB8,0xC6,0xA4,
698 0xB5,0x7E,0x73,0x74,0x75,0x76,0x77,0x78,0x79,0x7A,0xA1,0xBF,0xD0,0x5B,0xDE,0xAE,
699 0xAC,0xA3,0xA5,0xB7,0xA9,0xA7,0xB6,0xBC,0xBD,0xBE,0xDD,0xA8,0xAF,0x5D,0xB4,0xD7,
700 0x7B,0x41,0x42,0x43,0x44,0x45,0x46,0x47,0x48,0x49,0xAD,0xF4,0xF6,0xF2,0xF3,0xF5,
701 0x7D,0x4A,0x4B,0x4C,0x4D,0x4E,0x4F,0x50,0x51,0x52,0xB9,0xFB,0xFC,0xF9,0xFA,0xFF,
702 0x5C,0xF7,0x53,0x54,0x55,0x56,0x57,0x58,0x59,0x5A,0xB2,0xD4,0xD6,0xD2,0xD3,0xD5,
703 0x30,0x31,0x32,0x33,0x34,0x35,0x36,0x37,0x38,0x39,0xB3,0xDB,0xDC,0xD9,0xDA,0x9F
704 };
705 #endif
706 
707 /*
708  * Copy buf[0 ... nbytes-1] into out[], translating EBCDIC to ASCII.
709  */
710 static void
from_ebcdic(const unsigned char * buf,size_t nbytes,unsigned char * out)711 from_ebcdic(const unsigned char *buf, size_t nbytes, unsigned char *out)
712 {
713 	int i;
714 
715 	for (i = 0; i < (int)nbytes; i++) {
716 		out[i] = ebcdic_to_ascii[buf[i]];
717 	}
718 }
719