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.
6 * Redistribution and use in source and binary forms, with or without
7 * modification, are permitted provided that the following conditions
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.
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
29 * ASCII magic -- file types that we know based on keywords
30 * that can appear anywhere in the file.
32 * Extensively modified by Eric Fischer <enf@pobox.com> in July, 2000,
33 * to handle character codes other than ASCII on a unified basis.
35 * Joerg Wunsch <joerg@freebsd.org> wrote the original support for 8-bit
36 * international characters, now subsumed into this file.
52 FILE_RCSID("@(#)$Id: ascmagic.c,v 1.46 2006/10/20 21:04:15 christos Exp $")
55 typedef unsigned long unichar;
57 #define MAXLINELEN 300 /* longest sane line length */
58 #define ISSPC(x) ((x) == ' ' || (x) == '\t' || (x) == '\r' || (x) == '\n' \
59 || (x) == 0x85 || (x) == '\f')
61 private int looks_ascii(const unsigned char *, size_t, unichar *, size_t *);
62 private int looks_utf8(const unsigned char *, size_t, unichar *, size_t *);
63 private int looks_unicode(const unsigned char *, size_t, unichar *, size_t *);
64 private int looks_latin1(const unsigned char *, size_t, unichar *, size_t *);
65 private int looks_extended(const unsigned char *, size_t, unichar *, size_t *);
66 private void from_ebcdic(const unsigned char *, size_t, unsigned char *);
67 private int ascmatch(const unsigned char *, const unichar *, size_t);
71 file_ascmagic(struct magic_set *ms, const unsigned char *buf, size_t nbytes)
74 unsigned char *nbuf = NULL;
80 const char *code = NULL;
81 const char *code_mime = NULL;
82 const char *type = NULL;
83 const char *subtype = NULL;
84 const char *subtype_mime = NULL;
87 int has_backspace = 0;
95 int last_line_end = -1;
96 int has_long_lines = 0;
99 * Undo the NUL-termination kindly provided by process()
100 * but leave at least one byte to look at
102 while (nbytes > 1 && buf[nbytes - 1] == '\0')
105 if ((nbuf = calloc(1, (nbytes + 1) * sizeof(nbuf[0]))) == NULL)
107 if ((ubuf = calloc(1, (nbytes + 1) * sizeof(ubuf[0]))) == NULL)
111 * Then try to determine whether it's any character code we can
112 * identify. Each of these tests, if it succeeds, will leave
113 * the text converted into one-unichar-per-character Unicode in
114 * ubuf, and the number of characters converted in ulen.
116 if (looks_ascii(buf, nbytes, ubuf, &ulen)) {
118 code_mime = "us-ascii";
120 } else if (looks_utf8(buf, nbytes, ubuf, &ulen)) {
121 code = "UTF-8 Unicode";
124 } else if ((i = looks_unicode(buf, nbytes, ubuf, &ulen)) != 0) {
126 code = "Little-endian UTF-16 Unicode";
128 code = "Big-endian UTF-16 Unicode";
130 type = "character data";
131 code_mime = "utf-16"; /* is this defined? */
132 } else if (looks_latin1(buf, nbytes, ubuf, &ulen)) {
135 code_mime = "iso-8859-1";
136 } else if (looks_extended(buf, nbytes, ubuf, &ulen)) {
137 code = "Non-ISO extended-ASCII";
139 code_mime = "unknown";
141 from_ebcdic(buf, nbytes, nbuf);
143 if (looks_ascii(nbuf, nbytes, ubuf, &ulen)) {
145 type = "character data";
146 code_mime = "ebcdic";
147 } else if (looks_latin1(nbuf, nbytes, ubuf, &ulen)) {
148 code = "International EBCDIC";
149 type = "character data";
150 code_mime = "ebcdic";
153 goto done; /* doesn't look like text at all */
163 * for troff, look for . + letter + letter or .\";
164 * this must be done to disambiguate tar archives' ./file
165 * and other trash from real troff input.
167 * I believe Plan 9 troff allows non-ASCII characters in the names
168 * of macros, so this test might possibly fail on such a file.
171 unichar *tp = ubuf + 1;
174 ++tp; /* skip leading whitespace */
175 if ((tp[0] == '\\' && tp[1] == '\"') ||
176 (isascii((unsigned char)tp[0]) &&
177 isalnum((unsigned char)tp[0]) &&
178 isascii((unsigned char)tp[1]) &&
179 isalnum((unsigned char)tp[1]) &&
181 subtype_mime = "text/troff";
182 subtype = "troff or preprocessor input";
183 goto subtype_identified;
187 if ((*buf == 'c' || *buf == 'C') && ISSPC(buf[1])) {
188 subtype_mime = "text/fortran";
189 subtype = "fortran program";
190 goto subtype_identified;
193 /* look for tokens from names.h - this is expensive! */
200 * skip past any leading space
202 while (i < ulen && ISSPC(ubuf[i]))
208 * find the next whitespace
210 for (end = i + 1; end < nbytes; end++)
211 if (ISSPC(ubuf[end]))
215 * compare the word thus isolated against the token list
217 for (p = names; p < names + NNAMES; p++) {
218 if (ascmatch((const unsigned char *)p->name, ubuf + i,
220 subtype = types[p->type].human;
221 subtype_mime = types[p->type].mime;
222 goto subtype_identified;
232 * Now try to discover other details about the file.
234 for (i = 0; i < ulen; i++) {
235 if (ubuf[i] == '\n') {
244 seen_cr = (ubuf[i] == '\r');
248 if (ubuf[i] == 0x85) { /* X3.64/ECMA-43 "next line" character */
253 /* If this line is _longer_ than MAXLINELEN, remember it. */
254 if (i > last_line_end + MAXLINELEN)
257 if (ubuf[i] == '\033')
263 /* Beware, if the data has been truncated, the final CR could have
264 been followed by a LF. If we have HOWMANY bytes, it indicates
265 that the data might have been truncated, probably even before
266 this function was called. */
267 if (seen_cr && nbytes < HOWMANY)
270 if ((ms->flags & MAGIC_MIME)) {
272 if (file_printf(ms, subtype_mime) == -1)
275 if (file_printf(ms, "text/plain") == -1)
280 if (file_printf(ms, "; charset=") == -1)
282 if (file_printf(ms, code_mime) == -1)
286 if (file_printf(ms, code) == -1)
290 if (file_printf(ms, " ") == -1)
292 if (file_printf(ms, subtype) == -1)
296 if (file_printf(ms, " ") == -1)
298 if (file_printf(ms, type) == -1)
302 if (file_printf(ms, ", with very long lines") == -1)
306 * Only report line terminators if we find one other than LF,
307 * or if we find none at all.
309 if ((n_crlf == 0 && n_cr == 0 && n_nel == 0 && n_lf == 0) ||
310 (n_crlf != 0 || n_cr != 0 || n_nel != 0)) {
311 if (file_printf(ms, ", with") == -1)
314 if (n_crlf == 0 && n_cr == 0 && n_nel == 0 && n_lf == 0) {
315 if (file_printf(ms, " no") == -1)
319 if (file_printf(ms, " CRLF") == -1)
321 if (n_cr || n_lf || n_nel)
322 if (file_printf(ms, ",") == -1)
326 if (file_printf(ms, " CR") == -1)
329 if (file_printf(ms, ",") == -1)
333 if (file_printf(ms, " LF") == -1)
336 if (file_printf(ms, ",") == -1)
340 if (file_printf(ms, " NEL") == -1)
344 if (file_printf(ms, " line terminators") == -1)
349 if (file_printf(ms, ", with escape sequences") == -1)
352 if (file_printf(ms, ", with overstriking") == -1)
366 ascmatch(const unsigned char *s, const unichar *us, size_t ulen)
370 for (i = 0; i < ulen; i++) {
382 * This table reflects a particular philosophy about what constitutes
383 * "text," and there is room for disagreement about it.
385 * Version 3.31 of the file command considered a file to be ASCII if
386 * each of its characters was approved by either the isascii() or
387 * isalpha() function. On most systems, this would mean that any
388 * file consisting only of characters in the range 0x00 ... 0x7F
389 * would be called ASCII text, but many systems might reasonably
390 * consider some characters outside this range to be alphabetic,
391 * so the file command would call such characters ASCII. It might
392 * have been more accurate to call this "considered textual on the
393 * local system" than "ASCII."
395 * It considered a file to be "International language text" if each
396 * of its characters was either an ASCII printing character (according
397 * to the real ASCII standard, not the above test), a character in
398 * the range 0x80 ... 0xFF, or one of the following control characters:
399 * backspace, tab, line feed, vertical tab, form feed, carriage return,
400 * escape. No attempt was made to determine the language in which files
401 * of this type were written.
404 * The table below considers a file to be ASCII if all of its characters
405 * are either ASCII printing characters (again, according to the X3.4
406 * standard, not isascii()) or any of the following controls: bell,
407 * backspace, tab, line feed, form feed, carriage return, esc, nextline.
409 * I include bell because some programs (particularly shell scripts)
410 * use it literally, even though it is rare in normal text. I exclude
411 * vertical tab because it never seems to be used in real text. I also
412 * include, with hesitation, the X3.64/ECMA-43 control nextline (0x85),
413 * because that's what the dd EBCDIC->ASCII table maps the EBCDIC newline
414 * character to. It might be more appropriate to include it in the 8859
415 * set instead of the ASCII set, but it's got to be included in *something*
416 * we recognize or EBCDIC files aren't going to be considered textual.
417 * Some old Unix source files use SO/SI (^N/^O) to shift between Greek
418 * and Latin characters, so these should possibly be allowed. But they
419 * make a real mess on VT100-style displays if they're not paired properly,
420 * so we are probably better off not calling them text.
422 * A file is considered to be ISO-8859 text if its characters are all
423 * either ASCII, according to the above definition, or printing characters
424 * from the ISO-8859 8-bit extension, characters 0xA0 ... 0xFF.
426 * Finally, a file is considered to be international text from some other
427 * character code if its characters are all either ISO-8859 (according to
428 * the above definition) or characters in the range 0x80 ... 0x9F, which
429 * ISO-8859 considers to be control characters but the IBM PC and Macintosh
430 * consider to be printing characters.
433 #define F 0 /* character never appears in text */
434 #define T 1 /* character appears in plain ASCII text */
435 #define I 2 /* character appears in ISO-8859 text */
436 #define X 3 /* character appears in non-ISO extended ASCII (Mac, IBM PC) */
438 private char text_chars[256] = {
439 /* BEL BS HT LF FF CR */
440 F, F, F, F, F, F, F, T, T, T, T, F, T, T, F, F, /* 0x0X */
442 F, F, F, F, F, F, F, F, F, F, F, T, F, F, F, F, /* 0x1X */
443 T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, /* 0x2X */
444 T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, /* 0x3X */
445 T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, /* 0x4X */
446 T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, /* 0x5X */
447 T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, /* 0x6X */
448 T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, F, /* 0x7X */
450 X, X, X, X, X, T, X, X, X, X, X, X, X, X, X, X, /* 0x8X */
451 X, X, X, X, X, X, X, X, X, X, X, X, X, X, X, X, /* 0x9X */
452 I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, /* 0xaX */
453 I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, /* 0xbX */
454 I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, /* 0xcX */
455 I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, /* 0xdX */
456 I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, /* 0xeX */
457 I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, I /* 0xfX */
461 looks_ascii(const unsigned char *buf, size_t nbytes, unichar *ubuf,
468 for (i = 0; i < nbytes; i++) {
469 int t = text_chars[buf[i]];
474 ubuf[(*ulen)++] = buf[i];
481 looks_latin1(const unsigned char *buf, size_t nbytes, unichar *ubuf, size_t *ulen)
487 for (i = 0; i < nbytes; i++) {
488 int t = text_chars[buf[i]];
490 if (t != T && t != I)
493 ubuf[(*ulen)++] = buf[i];
500 looks_extended(const unsigned char *buf, size_t nbytes, unichar *ubuf,
507 for (i = 0; i < nbytes; i++) {
508 int t = text_chars[buf[i]];
510 if (t != T && t != I && t != X)
513 ubuf[(*ulen)++] = buf[i];
520 looks_utf8(const unsigned char *buf, size_t nbytes, unichar *ubuf, size_t *ulen)
528 for (i = 0; i < nbytes; i++) {
529 if ((buf[i] & 0x80) == 0) { /* 0xxxxxxx is plain ASCII */
531 * Even if the whole file is valid UTF-8 sequences,
532 * still reject it if it uses weird control characters.
535 if (text_chars[buf[i]] != T)
538 ubuf[(*ulen)++] = buf[i];
539 } else if ((buf[i] & 0x40) == 0) { /* 10xxxxxx never 1st byte */
541 } else { /* 11xxxxxx begins UTF-8 */
544 if ((buf[i] & 0x20) == 0) { /* 110xxxxx */
547 } else if ((buf[i] & 0x10) == 0) { /* 1110xxxx */
550 } else if ((buf[i] & 0x08) == 0) { /* 11110xxx */
553 } else if ((buf[i] & 0x04) == 0) { /* 111110xx */
556 } else if ((buf[i] & 0x02) == 0) { /* 1111110x */
562 for (n = 0; n < following; n++) {
567 if ((buf[i] & 0x80) == 0 || (buf[i] & 0x40))
570 c = (c << 6) + (buf[i] & 0x3f);
578 return gotone; /* don't claim it's UTF-8 if it's all 7-bit */
582 looks_unicode(const unsigned char *buf, size_t nbytes, unichar *ubuf,
591 if (buf[0] == 0xff && buf[1] == 0xfe)
593 else if (buf[0] == 0xfe && buf[1] == 0xff)
600 for (i = 2; i + 1 < nbytes; i += 2) {
601 /* XXX fix to properly handle chars > 65536 */
604 ubuf[(*ulen)++] = buf[i + 1] + 256 * buf[i];
606 ubuf[(*ulen)++] = buf[i] + 256 * buf[i + 1];
608 if (ubuf[*ulen - 1] == 0xfffe)
610 if (ubuf[*ulen - 1] < 128 &&
611 text_chars[(size_t)ubuf[*ulen - 1]] != T)
624 * This table maps each EBCDIC character to an (8-bit extended) ASCII
625 * character, as specified in the rationale for the dd(1) command in
626 * draft 11.2 (September, 1991) of the POSIX P1003.2 standard.
628 * Unfortunately it does not seem to correspond exactly to any of the
629 * five variants of EBCDIC documented in IBM's _Enterprise Systems
630 * Architecture/390: Principles of Operation_, SA22-7201-06, Seventh
631 * Edition, July, 1999, pp. I-1 - I-4.
633 * Fortunately, though, all versions of EBCDIC, including this one, agree
634 * on most of the printing characters that also appear in (7-bit) ASCII.
635 * Of these, only '|', '!', '~', '^', '[', and ']' are in question at all.
637 * Fortunately too, there is general agreement that codes 0x00 through
638 * 0x3F represent control characters, 0x41 a nonbreaking space, and the
639 * remainder printing characters.
641 * This is sufficient to allow us to identify EBCDIC text and to distinguish
642 * between old-style and internationalized examples of text.
645 private unsigned char ebcdic_to_ascii[] = {
646 0, 1, 2, 3, 156, 9, 134, 127, 151, 141, 142, 11, 12, 13, 14, 15,
647 16, 17, 18, 19, 157, 133, 8, 135, 24, 25, 146, 143, 28, 29, 30, 31,
648 128, 129, 130, 131, 132, 10, 23, 27, 136, 137, 138, 139, 140, 5, 6, 7,
649 144, 145, 22, 147, 148, 149, 150, 4, 152, 153, 154, 155, 20, 21, 158, 26,
650 ' ', 160, 161, 162, 163, 164, 165, 166, 167, 168, 213, '.', '<', '(', '+', '|',
651 '&', 169, 170, 171, 172, 173, 174, 175, 176, 177, '!', '$', '*', ')', ';', '~',
652 '-', '/', 178, 179, 180, 181, 182, 183, 184, 185, 203, ',', '%', '_', '>', '?',
653 186, 187, 188, 189, 190, 191, 192, 193, 194, '`', ':', '#', '@', '\'','=', '"',
654 195, 'a', 'b', 'c', 'd', 'e', 'f', 'g', 'h', 'i', 196, 197, 198, 199, 200, 201,
655 202, 'j', 'k', 'l', 'm', 'n', 'o', 'p', 'q', 'r', '^', 204, 205, 206, 207, 208,
656 209, 229, 's', 't', 'u', 'v', 'w', 'x', 'y', 'z', 210, 211, 212, '[', 214, 215,
657 216, 217, 218, 219, 220, 221, 222, 223, 224, 225, 226, 227, 228, ']', 230, 231,
658 '{', 'A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', 'I', 232, 233, 234, 235, 236, 237,
659 '}', 'J', 'K', 'L', 'M', 'N', 'O', 'P', 'Q', 'R', 238, 239, 240, 241, 242, 243,
660 '\\',159, 'S', 'T', 'U', 'V', 'W', 'X', 'Y', 'Z', 244, 245, 246, 247, 248, 249,
661 '0', '1', '2', '3', '4', '5', '6', '7', '8', '9', 250, 251, 252, 253, 254, 255
666 * The following EBCDIC-to-ASCII table may relate more closely to reality,
667 * or at least to modern reality. It comes from
669 * http://ftp.s390.ibm.com/products/oe/bpxqp9.html
671 * and maps the characters of EBCDIC code page 1047 (the code used for
672 * Unix-derived software on IBM's 390 systems) to the corresponding
673 * characters from ISO 8859-1.
675 * If this table is used instead of the above one, some of the special
676 * cases for the NEL character can be taken out of the code.
679 private unsigned char ebcdic_1047_to_8859[] = {
680 0x00,0x01,0x02,0x03,0x9C,0x09,0x86,0x7F,0x97,0x8D,0x8E,0x0B,0x0C,0x0D,0x0E,0x0F,
681 0x10,0x11,0x12,0x13,0x9D,0x0A,0x08,0x87,0x18,0x19,0x92,0x8F,0x1C,0x1D,0x1E,0x1F,
682 0x80,0x81,0x82,0x83,0x84,0x85,0x17,0x1B,0x88,0x89,0x8A,0x8B,0x8C,0x05,0x06,0x07,
683 0x90,0x91,0x16,0x93,0x94,0x95,0x96,0x04,0x98,0x99,0x9A,0x9B,0x14,0x15,0x9E,0x1A,
684 0x20,0xA0,0xE2,0xE4,0xE0,0xE1,0xE3,0xE5,0xE7,0xF1,0xA2,0x2E,0x3C,0x28,0x2B,0x7C,
685 0x26,0xE9,0xEA,0xEB,0xE8,0xED,0xEE,0xEF,0xEC,0xDF,0x21,0x24,0x2A,0x29,0x3B,0x5E,
686 0x2D,0x2F,0xC2,0xC4,0xC0,0xC1,0xC3,0xC5,0xC7,0xD1,0xA6,0x2C,0x25,0x5F,0x3E,0x3F,
687 0xF8,0xC9,0xCA,0xCB,0xC8,0xCD,0xCE,0xCF,0xCC,0x60,0x3A,0x23,0x40,0x27,0x3D,0x22,
688 0xD8,0x61,0x62,0x63,0x64,0x65,0x66,0x67,0x68,0x69,0xAB,0xBB,0xF0,0xFD,0xFE,0xB1,
689 0xB0,0x6A,0x6B,0x6C,0x6D,0x6E,0x6F,0x70,0x71,0x72,0xAA,0xBA,0xE6,0xB8,0xC6,0xA4,
690 0xB5,0x7E,0x73,0x74,0x75,0x76,0x77,0x78,0x79,0x7A,0xA1,0xBF,0xD0,0x5B,0xDE,0xAE,
691 0xAC,0xA3,0xA5,0xB7,0xA9,0xA7,0xB6,0xBC,0xBD,0xBE,0xDD,0xA8,0xAF,0x5D,0xB4,0xD7,
692 0x7B,0x41,0x42,0x43,0x44,0x45,0x46,0x47,0x48,0x49,0xAD,0xF4,0xF6,0xF2,0xF3,0xF5,
693 0x7D,0x4A,0x4B,0x4C,0x4D,0x4E,0x4F,0x50,0x51,0x52,0xB9,0xFB,0xFC,0xF9,0xFA,0xFF,
694 0x5C,0xF7,0x53,0x54,0x55,0x56,0x57,0x58,0x59,0x5A,0xB2,0xD4,0xD6,0xD2,0xD3,0xD5,
695 0x30,0x31,0x32,0x33,0x34,0x35,0x36,0x37,0x38,0x39,0xB3,0xDB,0xDC,0xD9,0xDA,0x9F
700 * Copy buf[0 ... nbytes-1] into out[], translating EBCDIC to ASCII.
703 from_ebcdic(const unsigned char *buf, size_t nbytes, unsigned char *out)
707 for (i = 0; i < nbytes; i++) {
708 out[i] = ebcdic_to_ascii[buf[i]];