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 * Encoding -- determine the character encoding of a text file.
31 * Joerg Wunsch <joerg@freebsd.org> wrote the original support for 8-bit
32 * international characters.
38 FILE_RCSID("@(#)$File: encoding.c,v 1.21 2019/06/08 20:49:14 christos Exp $")
46 private int looks_ascii(const unsigned char *, size_t, unichar *, size_t *);
47 private int looks_utf8_with_BOM(const unsigned char *, size_t, unichar *,
49 private int looks_utf7(const unsigned char *, size_t, unichar *, size_t *);
50 private int looks_ucs16(const unsigned char *, size_t, unichar *, size_t *);
51 private int looks_ucs32(const unsigned char *, size_t, unichar *, size_t *);
52 private int looks_latin1(const unsigned char *, size_t, unichar *, size_t *);
53 private int looks_extended(const unsigned char *, size_t, unichar *, size_t *);
54 private void from_ebcdic(const unsigned char *, size_t, unsigned char *);
57 #define DPRINTF(a) printf a
63 * Try to determine whether text is in some character code we can
64 * identify. Each of these tests, if it succeeds, will leave
65 * the text converted into one-unichar-per-character Unicode in
66 * ubuf, and the number of characters converted in ulen.
69 file_encoding(struct magic_set *ms, const struct buffer *b, unichar **ubuf,
70 size_t *ulen, const char **code, const char **code_mime, const char **type)
72 const unsigned char *buf = CAST(const unsigned char *, b->fbuf);
73 size_t nbytes = b->flen;
76 unsigned char *nbuf = NULL;
88 *code_mime = "binary";
90 mlen = (nbytes + 1) * sizeof((*ubuf)[0]);
91 if ((*ubuf = CAST(unichar *, calloc(CAST(size_t, 1), mlen))) == NULL) {
95 mlen = (nbytes + 1) * sizeof(nbuf[0]);
96 if ((nbuf = CAST(unsigned char *,
97 calloc(CAST(size_t, 1), mlen))) == NULL) {
102 if (looks_ascii(buf, nbytes, *ubuf, ulen)) {
103 if (looks_utf7(buf, nbytes, *ubuf, ulen) > 0) {
104 DPRINTF(("utf-7 %" SIZE_T_FORMAT "u\n", *ulen));
105 *code = "UTF-7 Unicode";
106 *code_mime = "utf-7";
108 DPRINTF(("ascii %" SIZE_T_FORMAT "u\n", *ulen));
110 *code_mime = "us-ascii";
112 } else if (looks_utf8_with_BOM(buf, nbytes, *ubuf, ulen) > 0) {
113 DPRINTF(("utf8/bom %" SIZE_T_FORMAT "u\n", *ulen));
114 *code = "UTF-8 Unicode (with BOM)";
115 *code_mime = "utf-8";
116 } else if (file_looks_utf8(buf, nbytes, *ubuf, ulen) > 1) {
117 DPRINTF(("utf8 %" SIZE_T_FORMAT "u\n", *ulen));
118 *code = "UTF-8 Unicode";
119 *code_mime = "utf-8";
120 } else if ((ucs_type = looks_ucs32(buf, nbytes, *ubuf, ulen)) != 0) {
122 *code = "Little-endian UTF-32 Unicode";
123 *code_mime = "utf-32le";
125 *code = "Big-endian UTF-32 Unicode";
126 *code_mime = "utf-32be";
128 DPRINTF(("ucs32 %" SIZE_T_FORMAT "u\n", *ulen));
129 } else if ((ucs_type = looks_ucs16(buf, nbytes, *ubuf, ulen)) != 0) {
131 *code = "Little-endian UTF-16 Unicode";
132 *code_mime = "utf-16le";
134 *code = "Big-endian UTF-16 Unicode";
135 *code_mime = "utf-16be";
137 DPRINTF(("ucs16 %" SIZE_T_FORMAT "u\n", *ulen));
138 } else if (looks_latin1(buf, nbytes, *ubuf, ulen)) {
139 DPRINTF(("latin1 %" SIZE_T_FORMAT "u\n", *ulen));
141 *code_mime = "iso-8859-1";
142 } else if (looks_extended(buf, nbytes, *ubuf, ulen)) {
143 DPRINTF(("extended %" SIZE_T_FORMAT "u\n", *ulen));
144 *code = "Non-ISO extended-ASCII";
145 *code_mime = "unknown-8bit";
147 from_ebcdic(buf, nbytes, nbuf);
149 if (looks_ascii(nbuf, nbytes, *ubuf, ulen)) {
150 DPRINTF(("ebcdic %" SIZE_T_FORMAT "u\n", *ulen));
152 *code_mime = "ebcdic";
153 } else if (looks_latin1(nbuf, nbytes, *ubuf, ulen)) {
154 DPRINTF(("ebcdic/international %" SIZE_T_FORMAT "u\n",
156 *code = "International EBCDIC";
157 *code_mime = "ebcdic";
158 } else { /* Doesn't look like text at all */
159 DPRINTF(("binary\n"));
167 if (ubuf == &udefbuf)
174 * This table reflects a particular philosophy about what constitutes
175 * "text," and there is room for disagreement about it.
177 * Version 3.31 of the file command considered a file to be ASCII if
178 * each of its characters was approved by either the isascii() or
179 * isalpha() function. On most systems, this would mean that any
180 * file consisting only of characters in the range 0x00 ... 0x7F
181 * would be called ASCII text, but many systems might reasonably
182 * consider some characters outside this range to be alphabetic,
183 * so the file command would call such characters ASCII. It might
184 * have been more accurate to call this "considered textual on the
185 * local system" than "ASCII."
187 * It considered a file to be "International language text" if each
188 * of its characters was either an ASCII printing character (according
189 * to the real ASCII standard, not the above test), a character in
190 * the range 0x80 ... 0xFF, or one of the following control characters:
191 * backspace, tab, line feed, vertical tab, form feed, carriage return,
192 * escape. No attempt was made to determine the language in which files
193 * of this type were written.
196 * The table below considers a file to be ASCII if all of its characters
197 * are either ASCII printing characters (again, according to the X3.4
198 * standard, not isascii()) or any of the following controls: bell,
199 * backspace, tab, line feed, form feed, carriage return, esc, nextline.
201 * I include bell because some programs (particularly shell scripts)
202 * use it literally, even though it is rare in normal text. I exclude
203 * vertical tab because it never seems to be used in real text. I also
204 * include, with hesitation, the X3.64/ECMA-43 control nextline (0x85),
205 * because that's what the dd EBCDIC->ASCII table maps the EBCDIC newline
206 * character to. It might be more appropriate to include it in the 8859
207 * set instead of the ASCII set, but it's got to be included in *something*
208 * we recognize or EBCDIC files aren't going to be considered textual.
209 * Some old Unix source files use SO/SI (^N/^O) to shift between Greek
210 * and Latin characters, so these should possibly be allowed. But they
211 * make a real mess on VT100-style displays if they're not paired properly,
212 * so we are probably better off not calling them text.
214 * A file is considered to be ISO-8859 text if its characters are all
215 * either ASCII, according to the above definition, or printing characters
216 * from the ISO-8859 8-bit extension, characters 0xA0 ... 0xFF.
218 * Finally, a file is considered to be international text from some other
219 * character code if its characters are all either ISO-8859 (according to
220 * the above definition) or characters in the range 0x80 ... 0x9F, which
221 * ISO-8859 considers to be control characters but the IBM PC and Macintosh
222 * consider to be printing characters.
225 #define F 0 /* character never appears in text */
226 #define T 1 /* character appears in plain ASCII text */
227 #define I 2 /* character appears in ISO-8859 text */
228 #define X 3 /* character appears in non-ISO extended ASCII (Mac, IBM PC) */
230 private char text_chars[256] = {
231 /* BEL BS HT LF VT FF CR */
232 F, F, F, F, F, F, F, T, T, T, T, T, T, T, F, F, /* 0x0X */
234 F, F, F, F, F, F, F, F, F, F, F, T, F, F, F, F, /* 0x1X */
235 T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, /* 0x2X */
236 T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, /* 0x3X */
237 T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, /* 0x4X */
238 T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, /* 0x5X */
239 T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, /* 0x6X */
240 T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, F, /* 0x7X */
242 X, X, X, X, X, T, X, X, X, X, X, X, X, X, X, X, /* 0x8X */
243 X, X, X, X, X, X, X, X, X, X, X, X, X, X, X, X, /* 0x9X */
244 I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, /* 0xaX */
245 I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, /* 0xbX */
246 I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, /* 0xcX */
247 I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, /* 0xdX */
248 I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, /* 0xeX */
249 I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, I /* 0xfX */
253 looks_ascii(const unsigned char *buf, size_t nbytes, unichar *ubuf,
260 for (i = 0; i < nbytes; i++) {
261 int t = text_chars[buf[i]];
266 ubuf[(*ulen)++] = buf[i];
273 looks_latin1(const unsigned char *buf, size_t nbytes, unichar *ubuf, size_t *ulen)
279 for (i = 0; i < nbytes; i++) {
280 int t = text_chars[buf[i]];
282 if (t != T && t != I)
285 ubuf[(*ulen)++] = buf[i];
292 looks_extended(const unsigned char *buf, size_t nbytes, unichar *ubuf,
299 for (i = 0; i < nbytes; i++) {
300 int t = text_chars[buf[i]];
302 if (t != T && t != I && t != X)
305 ubuf[(*ulen)++] = buf[i];
312 * Decide whether some text looks like UTF-8. Returns:
315 * 0: uses odd control characters, so doesn't look like text
317 * 2: definitely UTF-8 text (valid high-bit set bytes)
319 * If ubuf is non-NULL on entry, text is decoded into ubuf, *ulen;
320 * ubuf must be big enough!
323 file_looks_utf8(const unsigned char *buf, size_t nbytes, unichar *ubuf, size_t *ulen)
328 int gotone = 0, ctrl = 0;
333 for (i = 0; i < nbytes; i++) {
334 if ((buf[i] & 0x80) == 0) { /* 0xxxxxxx is plain ASCII */
336 * Even if the whole file is valid UTF-8 sequences,
337 * still reject it if it uses weird control characters.
340 if (text_chars[buf[i]] != T)
344 ubuf[(*ulen)++] = buf[i];
345 } else if ((buf[i] & 0x40) == 0) { /* 10xxxxxx never 1st byte */
347 } else { /* 11xxxxxx begins UTF-8 */
350 if ((buf[i] & 0x20) == 0) { /* 110xxxxx */
353 } else if ((buf[i] & 0x10) == 0) { /* 1110xxxx */
356 } else if ((buf[i] & 0x08) == 0) { /* 11110xxx */
359 } else if ((buf[i] & 0x04) == 0) { /* 111110xx */
362 } else if ((buf[i] & 0x02) == 0) { /* 1111110x */
368 for (n = 0; n < following; n++) {
373 if ((buf[i] & 0x80) == 0 || (buf[i] & 0x40))
376 c = (c << 6) + (buf[i] & 0x3f);
385 return ctrl ? 0 : (gotone ? 2 : 1);
389 * Decide whether some text looks like UTF-8 with BOM. If there is no
390 * BOM, return -1; otherwise return the result of looks_utf8 on the
394 looks_utf8_with_BOM(const unsigned char *buf, size_t nbytes, unichar *ubuf,
397 if (nbytes > 3 && buf[0] == 0xef && buf[1] == 0xbb && buf[2] == 0xbf)
398 return file_looks_utf8(buf + 3, nbytes - 3, ubuf, ulen);
404 looks_utf7(const unsigned char *buf, size_t nbytes, unichar *ubuf, size_t *ulen)
406 if (nbytes > 4 && buf[0] == '+' && buf[1] == '/' && buf[2] == 'v')
423 looks_ucs16(const unsigned char *bf, size_t nbytes, unichar *ubf,
432 if (bf[0] == 0xff && bf[1] == 0xfe)
434 else if (bf[0] == 0xfe && bf[1] == 0xff)
441 for (i = 2; i + 1 < nbytes; i += 2) {
442 /* XXX fix to properly handle chars > 65536 */
445 ubf[(*ulen)++] = bf[i + 1]
446 | (CAST(unichar, bf[i]) << 8);
448 ubf[(*ulen)++] = bf[i]
449 | (CAST(unichar, bf[i + 1]) << 8);
451 if (ubf[*ulen - 1] == 0xfffe)
453 if (ubf[*ulen - 1] < 128 &&
454 text_chars[CAST(size_t, ubf[*ulen - 1])] != T)
462 looks_ucs32(const unsigned char *bf, size_t nbytes, unichar *ubf,
471 if (bf[0] == 0xff && bf[1] == 0xfe && bf[2] == 0 && bf[3] == 0)
473 else if (bf[0] == 0 && bf[1] == 0 && bf[2] == 0xfe && bf[3] == 0xff)
480 for (i = 4; i + 3 < nbytes; i += 4) {
481 /* XXX fix to properly handle chars > 65536 */
484 ubf[(*ulen)++] = CAST(unichar, bf[i + 3])
485 | (CAST(unichar, bf[i + 2]) << 8)
486 | (CAST(unichar, bf[i + 1]) << 16)
487 | (CAST(unichar, bf[i]) << 24);
489 ubf[(*ulen)++] = CAST(unichar, bf[i + 0])
490 | (CAST(unichar, bf[i + 1]) << 8)
491 | (CAST(unichar, bf[i + 2]) << 16)
492 | (CAST(unichar, bf[i + 3]) << 24);
494 if (ubf[*ulen - 1] == 0xfffe)
496 if (ubf[*ulen - 1] < 128 &&
497 text_chars[CAST(size_t, ubf[*ulen - 1])] != T)
509 * This table maps each EBCDIC character to an (8-bit extended) ASCII
510 * character, as specified in the rationale for the dd(1) command in
511 * draft 11.2 (September, 1991) of the POSIX P1003.2 standard.
513 * Unfortunately it does not seem to correspond exactly to any of the
514 * five variants of EBCDIC documented in IBM's _Enterprise Systems
515 * Architecture/390: Principles of Operation_, SA22-7201-06, Seventh
516 * Edition, July, 1999, pp. I-1 - I-4.
518 * Fortunately, though, all versions of EBCDIC, including this one, agree
519 * on most of the printing characters that also appear in (7-bit) ASCII.
520 * Of these, only '|', '!', '~', '^', '[', and ']' are in question at all.
522 * Fortunately too, there is general agreement that codes 0x00 through
523 * 0x3F represent control characters, 0x41 a nonbreaking space, and the
524 * remainder printing characters.
526 * This is sufficient to allow us to identify EBCDIC text and to distinguish
527 * between old-style and internationalized examples of text.
530 private unsigned char ebcdic_to_ascii[] = {
531 0, 1, 2, 3, 156, 9, 134, 127, 151, 141, 142, 11, 12, 13, 14, 15,
532 16, 17, 18, 19, 157, 133, 8, 135, 24, 25, 146, 143, 28, 29, 30, 31,
533 128, 129, 130, 131, 132, 10, 23, 27, 136, 137, 138, 139, 140, 5, 6, 7,
534 144, 145, 22, 147, 148, 149, 150, 4, 152, 153, 154, 155, 20, 21, 158, 26,
535 ' ', 160, 161, 162, 163, 164, 165, 166, 167, 168, 213, '.', '<', '(', '+', '|',
536 '&', 169, 170, 171, 172, 173, 174, 175, 176, 177, '!', '$', '*', ')', ';', '~',
537 '-', '/', 178, 179, 180, 181, 182, 183, 184, 185, 203, ',', '%', '_', '>', '?',
538 186, 187, 188, 189, 190, 191, 192, 193, 194, '`', ':', '#', '@', '\'','=', '"',
539 195, 'a', 'b', 'c', 'd', 'e', 'f', 'g', 'h', 'i', 196, 197, 198, 199, 200, 201,
540 202, 'j', 'k', 'l', 'm', 'n', 'o', 'p', 'q', 'r', '^', 204, 205, 206, 207, 208,
541 209, 229, 's', 't', 'u', 'v', 'w', 'x', 'y', 'z', 210, 211, 212, '[', 214, 215,
542 216, 217, 218, 219, 220, 221, 222, 223, 224, 225, 226, 227, 228, ']', 230, 231,
543 '{', 'A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', 'I', 232, 233, 234, 235, 236, 237,
544 '}', 'J', 'K', 'L', 'M', 'N', 'O', 'P', 'Q', 'R', 238, 239, 240, 241, 242, 243,
545 '\\',159, 'S', 'T', 'U', 'V', 'W', 'X', 'Y', 'Z', 244, 245, 246, 247, 248, 249,
546 '0', '1', '2', '3', '4', '5', '6', '7', '8', '9', 250, 251, 252, 253, 254, 255
551 * The following EBCDIC-to-ASCII table may relate more closely to reality,
552 * or at least to modern reality. It comes from
554 * http://ftp.s390.ibm.com/products/oe/bpxqp9.html
556 * and maps the characters of EBCDIC code page 1047 (the code used for
557 * Unix-derived software on IBM's 390 systems) to the corresponding
558 * characters from ISO 8859-1.
560 * If this table is used instead of the above one, some of the special
561 * cases for the NEL character can be taken out of the code.
564 private unsigned char ebcdic_1047_to_8859[] = {
565 0x00,0x01,0x02,0x03,0x9C,0x09,0x86,0x7F,0x97,0x8D,0x8E,0x0B,0x0C,0x0D,0x0E,0x0F,
566 0x10,0x11,0x12,0x13,0x9D,0x0A,0x08,0x87,0x18,0x19,0x92,0x8F,0x1C,0x1D,0x1E,0x1F,
567 0x80,0x81,0x82,0x83,0x84,0x85,0x17,0x1B,0x88,0x89,0x8A,0x8B,0x8C,0x05,0x06,0x07,
568 0x90,0x91,0x16,0x93,0x94,0x95,0x96,0x04,0x98,0x99,0x9A,0x9B,0x14,0x15,0x9E,0x1A,
569 0x20,0xA0,0xE2,0xE4,0xE0,0xE1,0xE3,0xE5,0xE7,0xF1,0xA2,0x2E,0x3C,0x28,0x2B,0x7C,
570 0x26,0xE9,0xEA,0xEB,0xE8,0xED,0xEE,0xEF,0xEC,0xDF,0x21,0x24,0x2A,0x29,0x3B,0x5E,
571 0x2D,0x2F,0xC2,0xC4,0xC0,0xC1,0xC3,0xC5,0xC7,0xD1,0xA6,0x2C,0x25,0x5F,0x3E,0x3F,
572 0xF8,0xC9,0xCA,0xCB,0xC8,0xCD,0xCE,0xCF,0xCC,0x60,0x3A,0x23,0x40,0x27,0x3D,0x22,
573 0xD8,0x61,0x62,0x63,0x64,0x65,0x66,0x67,0x68,0x69,0xAB,0xBB,0xF0,0xFD,0xFE,0xB1,
574 0xB0,0x6A,0x6B,0x6C,0x6D,0x6E,0x6F,0x70,0x71,0x72,0xAA,0xBA,0xE6,0xB8,0xC6,0xA4,
575 0xB5,0x7E,0x73,0x74,0x75,0x76,0x77,0x78,0x79,0x7A,0xA1,0xBF,0xD0,0x5B,0xDE,0xAE,
576 0xAC,0xA3,0xA5,0xB7,0xA9,0xA7,0xB6,0xBC,0xBD,0xBE,0xDD,0xA8,0xAF,0x5D,0xB4,0xD7,
577 0x7B,0x41,0x42,0x43,0x44,0x45,0x46,0x47,0x48,0x49,0xAD,0xF4,0xF6,0xF2,0xF3,0xF5,
578 0x7D,0x4A,0x4B,0x4C,0x4D,0x4E,0x4F,0x50,0x51,0x52,0xB9,0xFB,0xFC,0xF9,0xFA,0xFF,
579 0x5C,0xF7,0x53,0x54,0x55,0x56,0x57,0x58,0x59,0x5A,0xB2,0xD4,0xD6,0xD2,0xD3,0xD5,
580 0x30,0x31,0x32,0x33,0x34,0x35,0x36,0x37,0x38,0x39,0xB3,0xDB,0xDC,0xD9,0xDA,0x9F
585 * Copy buf[0 ... nbytes-1] into out[], translating EBCDIC to ASCII.
588 from_ebcdic(const unsigned char *buf, size_t nbytes, unsigned char *out)
592 for (i = 0; i < nbytes; i++) {
593 out[i] = ebcdic_to_ascii[buf[i]];