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 1.9 2013/11/19 20:45:50 christos Exp $")
39 #endif /* lint */
40
41 #include "magic.h"
42 #include <string.h>
43 #include <memory.h>
44 #include <stdlib.h>
45
46
47 private int looks_ascii(const unsigned char *, size_t, unichar *, size_t *);
48 private int looks_utf8_with_BOM(const unsigned char *, size_t, unichar *,
49 size_t *);
50 private int looks_ucs16(const unsigned char *, size_t, unichar *, size_t *);
51 private int looks_latin1(const unsigned char *, size_t, unichar *, size_t *);
52 private int looks_extended(const unsigned char *, size_t, unichar *, size_t *);
53 private void from_ebcdic(const unsigned char *, size_t, unsigned char *);
54
55 #ifdef DEBUG_ENCODING
56 #define DPRINTF(a) printf a
57 #else
58 #define DPRINTF(a)
59 #endif
60
61 /*
62 * Try to determine whether text is in some character code we can
63 * identify. Each of these tests, if it succeeds, will leave
64 * the text converted into one-unichar-per-character Unicode in
65 * ubuf, and the number of characters converted in ulen.
66 */
67 protected int
68 file_encoding(struct magic_set *ms, const unsigned char *buf, size_t nbytes, unichar **ubuf, size_t *ulen, const char **code, const char **code_mime, const char **type)
69 {
70 size_t mlen;
71 int rv = 1, ucs_type;
72 unsigned char *nbuf = NULL;
73
74 *type = "text";
75 *ulen = 0;
76 *code = "unknown";
77 *code_mime = "binary";
78
79 mlen = (nbytes + 1) * sizeof((*ubuf)[0]);
80 if ((*ubuf = CAST(unichar *, calloc((size_t)1, mlen))) == NULL) {
81 file_oomem(ms, mlen);
82 goto done;
83 }
84 mlen = (nbytes + 1) * sizeof(nbuf[0]);
85 if ((nbuf = CAST(unsigned char *, calloc((size_t)1, mlen))) == NULL) {
86 file_oomem(ms, mlen);
87 goto done;
88 }
89
90 if (looks_ascii(buf, nbytes, *ubuf, ulen)) {
91 DPRINTF(("ascii %" SIZE_T_FORMAT "u\n", *ulen));
92 *code = "ASCII";
93 *code_mime = "us-ascii";
94 } else if (looks_utf8_with_BOM(buf, nbytes, *ubuf, ulen) > 0) {
95 DPRINTF(("utf8/bom %" SIZE_T_FORMAT "u\n", *ulen));
96 *code = "UTF-8 Unicode (with BOM)";
97 *code_mime = "utf-8";
98 } else if (file_looks_utf8(buf, nbytes, *ubuf, ulen) > 1) {
99 DPRINTF(("utf8 %" SIZE_T_FORMAT "u\n", *ulen));
100 *code = "UTF-8 Unicode (with BOM)";
101 *code = "UTF-8 Unicode";
102 *code_mime = "utf-8";
103 } else if ((ucs_type = looks_ucs16(buf, nbytes, *ubuf, ulen)) != 0) {
104 if (ucs_type == 1) {
105 *code = "Little-endian UTF-16 Unicode";
106 *code_mime = "utf-16le";
107 } else {
108 *code = "Big-endian UTF-16 Unicode";
109 *code_mime = "utf-16be";
110 }
111 DPRINTF(("ucs16 %" SIZE_T_FORMAT "u\n", *ulen));
112 } else if (looks_latin1(buf, nbytes, *ubuf, ulen)) {
113 DPRINTF(("latin1 %" SIZE_T_FORMAT "u\n", *ulen));
114 *code = "ISO-8859";
115 *code_mime = "iso-8859-1";
116 } else if (looks_extended(buf, nbytes, *ubuf, ulen)) {
117 DPRINTF(("extended %" SIZE_T_FORMAT "u\n", *ulen));
118 *code = "Non-ISO extended-ASCII";
119 *code_mime = "unknown-8bit";
120 } else {
121 from_ebcdic(buf, nbytes, nbuf);
122
123 if (looks_ascii(nbuf, nbytes, *ubuf, ulen)) {
124 DPRINTF(("ebcdic %" SIZE_T_FORMAT "u\n", *ulen));
125 *code = "EBCDIC";
126 *code_mime = "ebcdic";
127 } else if (looks_latin1(nbuf, nbytes, *ubuf, ulen)) {
128 DPRINTF(("ebcdic/international %" SIZE_T_FORMAT "u\n",
129 *ulen));
130 *code = "International EBCDIC";
131 *code_mime = "ebcdic";
132 } else { /* Doesn't look like text at all */
133 DPRINTF(("binary\n"));
134 rv = 0;
135 *type = "binary";
136 }
137 }
138
139 done:
140 free(nbuf);
141
142 return rv;
143 }
144
145 /*
146 * This table reflects a particular philosophy about what constitutes
147 * "text," and there is room for disagreement about it.
148 *
149 * Version 3.31 of the file command considered a file to be ASCII if
150 * each of its characters was approved by either the isascii() or
151 * isalpha() function. On most systems, this would mean that any
152 * file consisting only of characters in the range 0x00 ... 0x7F
153 * would be called ASCII text, but many systems might reasonably
154 * consider some characters outside this range to be alphabetic,
155 * so the file command would call such characters ASCII. It might
156 * have been more accurate to call this "considered textual on the
157 * local system" than "ASCII."
158 *
159 * It considered a file to be "International language text" if each
160 * of its characters was either an ASCII printing character (according
161 * to the real ASCII standard, not the above test), a character in
162 * the range 0x80 ... 0xFF, or one of the following control characters:
163 * backspace, tab, line feed, vertical tab, form feed, carriage return,
164 * escape. No attempt was made to determine the language in which files
165 * of this type were written.
166 *
167 *
168 * The table below considers a file to be ASCII if all of its characters
169 * are either ASCII printing characters (again, according to the X3.4
170 * standard, not isascii()) or any of the following controls: bell,
171 * backspace, tab, line feed, form feed, carriage return, esc, nextline.
172 *
173 * I include bell because some programs (particularly shell scripts)
174 * use it literally, even though it is rare in normal text. I exclude
175 * vertical tab because it never seems to be used in real text. I also
176 * include, with hesitation, the X3.64/ECMA-43 control nextline (0x85),
177 * because that's what the dd EBCDIC->ASCII table maps the EBCDIC newline
178 * character to. It might be more appropriate to include it in the 8859
179 * set instead of the ASCII set, but it's got to be included in *something*
180 * we recognize or EBCDIC files aren't going to be considered textual.
181 * Some old Unix source files use SO/SI (^N/^O) to shift between Greek
182 * and Latin characters, so these should possibly be allowed. But they
183 * make a real mess on VT100-style displays if they're not paired properly,
184 * so we are probably better off not calling them text.
185 *
186 * A file is considered to be ISO-8859 text if its characters are all
187 * either ASCII, according to the above definition, or printing characters
188 * from the ISO-8859 8-bit extension, characters 0xA0 ... 0xFF.
189 *
190 * Finally, a file is considered to be international text from some other
191 * character code if its characters are all either ISO-8859 (according to
192 * the above definition) or characters in the range 0x80 ... 0x9F, which
193 * ISO-8859 considers to be control characters but the IBM PC and Macintosh
194 * consider to be printing characters.
195 */
196
197 #define F 0 /* character never appears in text */
198 #define T 1 /* character appears in plain ASCII text */
199 #define I 2 /* character appears in ISO-8859 text */
200 #define X 3 /* character appears in non-ISO extended ASCII (Mac, IBM PC) */
201
202 private char text_chars[256] = {
203 /* BEL BS HT LF FF CR */
204 F, F, F, F, F, F, F, T, T, T, T, F, T, T, F, F, /* 0x0X */
205 /* ESC */
206 F, F, F, F, F, F, F, F, F, F, F, T, F, F, F, F, /* 0x1X */
207 T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, /* 0x2X */
208 T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, /* 0x3X */
209 T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, /* 0x4X */
210 T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, /* 0x5X */
211 T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, /* 0x6X */
212 T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, F, /* 0x7X */
213 /* NEL */
214 X, X, X, X, X, T, X, X, X, X, X, X, X, X, X, X, /* 0x8X */
215 X, X, X, X, X, X, X, X, X, X, X, X, X, X, X, X, /* 0x9X */
216 I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, /* 0xaX */
217 I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, /* 0xbX */
218 I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, /* 0xcX */
219 I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, /* 0xdX */
220 I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, /* 0xeX */
221 I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, I /* 0xfX */
222 };
223
224 private int
225 looks_ascii(const unsigned char *buf, size_t nbytes, unichar *ubuf,
226 size_t *ulen)
227 {
228 size_t i;
229
230 *ulen = 0;
231
232 for (i = 0; i < nbytes; i++) {
233 int t = text_chars[buf[i]];
234
235 if (t != T)
236 return 0;
237
238 ubuf[(*ulen)++] = buf[i];
239 }
240
241 return 1;
242 }
243
244 private int
245 looks_latin1(const unsigned char *buf, size_t nbytes, unichar *ubuf, size_t *ulen)
246 {
247 size_t i;
248
249 *ulen = 0;
250
251 for (i = 0; i < nbytes; i++) {
252 int t = text_chars[buf[i]];
253
254 if (t != T && t != I)
255 return 0;
256
257 ubuf[(*ulen)++] = buf[i];
258 }
259
260 return 1;
261 }
262
263 private int
264 looks_extended(const unsigned char *buf, size_t nbytes, unichar *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 (t != T && t != I && t != X)
275 return 0;
276
277 ubuf[(*ulen)++] = buf[i];
278 }
279
280 return 1;
281 }
282
283 /*
284 * Decide whether some text looks like UTF-8. Returns:
285 *
286 * -1: invalid UTF-8
287 * 0: uses odd control characters, so doesn't look like text
288 * 1: 7-bit text
289 * 2: definitely UTF-8 text (valid high-bit set bytes)
290 *
291 * If ubuf is non-NULL on entry, text is decoded into ubuf, *ulen;
292 * ubuf must be big enough!
293 */
294 protected int
295 file_looks_utf8(const unsigned char *buf, size_t nbytes, unichar *ubuf, size_t *ulen)
296 {
297 size_t i;
298 int n;
299 unichar c;
300 int gotone = 0, ctrl = 0;
301
302 if (ubuf)
303 *ulen = 0;
304
305 for (i = 0; i < nbytes; i++) {
306 if ((buf[i] & 0x80) == 0) { /* 0xxxxxxx is plain ASCII */
307 /*
308 * Even if the whole file is valid UTF-8 sequences,
309 * still reject it if it uses weird control characters.
310 */
311
312 if (text_chars[buf[i]] != T)
313 ctrl = 1;
314
315 if (ubuf)
316 ubuf[(*ulen)++] = buf[i];
317 } else if ((buf[i] & 0x40) == 0) { /* 10xxxxxx never 1st byte */
318 return -1;
319 } else { /* 11xxxxxx begins UTF-8 */
320 int following;
321
322 if ((buf[i] & 0x20) == 0) { /* 110xxxxx */
323 c = buf[i] & 0x1f;
324 following = 1;
325 } else if ((buf[i] & 0x10) == 0) { /* 1110xxxx */
326 c = buf[i] & 0x0f;
327 following = 2;
328 } else if ((buf[i] & 0x08) == 0) { /* 11110xxx */
329 c = buf[i] & 0x07;
330 following = 3;
331 } else if ((buf[i] & 0x04) == 0) { /* 111110xx */
332 c = buf[i] & 0x03;
333 following = 4;
334 } else if ((buf[i] & 0x02) == 0) { /* 1111110x */
335 c = buf[i] & 0x01;
336 following = 5;
337 } else
338 return -1;
339
340 for (n = 0; n < following; n++) {
341 i++;
342 if (i >= nbytes)
343 goto done;
344
345 if ((buf[i] & 0x80) == 0 || (buf[i] & 0x40))
346 return -1;
347
348 c = (c << 6) + (buf[i] & 0x3f);
349 }
350
351 if (ubuf)
352 ubuf[(*ulen)++] = c;
353 gotone = 1;
354 }
355 }
356 done:
357 return ctrl ? 0 : (gotone ? 2 : 1);
358 }
359
360 /*
361 * Decide whether some text looks like UTF-8 with BOM. If there is no
362 * BOM, return -1; otherwise return the result of looks_utf8 on the
363 * rest of the text.
364 */
365 private int
366 looks_utf8_with_BOM(const unsigned char *buf, size_t nbytes, unichar *ubuf,
367 size_t *ulen)
368 {
369 if (nbytes > 3 && buf[0] == 0xef && buf[1] == 0xbb && buf[2] == 0xbf)
370 return file_looks_utf8(buf + 3, nbytes - 3, ubuf, ulen);
371 else
372 return -1;
373 }
374
375 private int
376 looks_ucs16(const unsigned char *buf, size_t nbytes, unichar *ubuf,
377 size_t *ulen)
378 {
379 int bigend;
380 size_t i;
381
382 if (nbytes < 2)
383 return 0;
384
385 if (buf[0] == 0xff && buf[1] == 0xfe)
386 bigend = 0;
387 else if (buf[0] == 0xfe && buf[1] == 0xff)
388 bigend = 1;
389 else
390 return 0;
391
392 *ulen = 0;
393
394 for (i = 2; i + 1 < nbytes; i += 2) {
395 /* XXX fix to properly handle chars > 65536 */
396
397 if (bigend)
398 ubuf[(*ulen)++] = buf[i + 1] + 256 * buf[i];
399 else
400 ubuf[(*ulen)++] = buf[i] + 256 * buf[i + 1];
401
402 if (ubuf[*ulen - 1] == 0xfffe)
403 return 0;
404 if (ubuf[*ulen - 1] < 128 &&
405 text_chars[(size_t)ubuf[*ulen - 1]] != T)
406 return 0;
407 }
408
409 return 1 + bigend;
410 }
411
412 #undef F
413 #undef T
414 #undef I
415 #undef X
416
417 /*
418 * This table maps each EBCDIC character to an (8-bit extended) ASCII
419 * character, as specified in the rationale for the dd(1) command in
420 * draft 11.2 (September, 1991) of the POSIX P1003.2 standard.
421 *
422 * Unfortunately it does not seem to correspond exactly to any of the
423 * five variants of EBCDIC documented in IBM's _Enterprise Systems
424 * Architecture/390: Principles of Operation_, SA22-7201-06, Seventh
425 * Edition, July, 1999, pp. I-1 - I-4.
426 *
427 * Fortunately, though, all versions of EBCDIC, including this one, agree
428 * on most of the printing characters that also appear in (7-bit) ASCII.
429 * Of these, only '|', '!', '~', '^', '[', and ']' are in question at all.
430 *
431 * Fortunately too, there is general agreement that codes 0x00 through
432 * 0x3F represent control characters, 0x41 a nonbreaking space, and the
433 * remainder printing characters.
434 *
435 * This is sufficient to allow us to identify EBCDIC text and to distinguish
436 * between old-style and internationalized examples of text.
437 */
438
439 private unsigned char ebcdic_to_ascii[] = {
440 0, 1, 2, 3, 156, 9, 134, 127, 151, 141, 142, 11, 12, 13, 14, 15,
441 16, 17, 18, 19, 157, 133, 8, 135, 24, 25, 146, 143, 28, 29, 30, 31,
442 128, 129, 130, 131, 132, 10, 23, 27, 136, 137, 138, 139, 140, 5, 6, 7,
443 144, 145, 22, 147, 148, 149, 150, 4, 152, 153, 154, 155, 20, 21, 158, 26,
444 ' ', 160, 161, 162, 163, 164, 165, 166, 167, 168, 213, '.', '<', '(', '+', '|',
445 '&', 169, 170, 171, 172, 173, 174, 175, 176, 177, '!', '$', '*', ')', ';', '~',
446 '-', '/', 178, 179, 180, 181, 182, 183, 184, 185, 203, ',', '%', '_', '>', '?',
447 186, 187, 188, 189, 190, 191, 192, 193, 194, '`', ':', '#', '@', '\'','=', '"',
448 195, 'a', 'b', 'c', 'd', 'e', 'f', 'g', 'h', 'i', 196, 197, 198, 199, 200, 201,
449 202, 'j', 'k', 'l', 'm', 'n', 'o', 'p', 'q', 'r', '^', 204, 205, 206, 207, 208,
450 209, 229, 's', 't', 'u', 'v', 'w', 'x', 'y', 'z', 210, 211, 212, '[', 214, 215,
451 216, 217, 218, 219, 220, 221, 222, 223, 224, 225, 226, 227, 228, ']', 230, 231,
452 '{', 'A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', 'I', 232, 233, 234, 235, 236, 237,
453 '}', 'J', 'K', 'L', 'M', 'N', 'O', 'P', 'Q', 'R', 238, 239, 240, 241, 242, 243,
454 '\\',159, 'S', 'T', 'U', 'V', 'W', 'X', 'Y', 'Z', 244, 245, 246, 247, 248, 249,
455 '0', '1', '2', '3', '4', '5', '6', '7', '8', '9', 250, 251, 252, 253, 254, 255
456 };
457
458 #ifdef notdef
459 /*
460 * The following EBCDIC-to-ASCII table may relate more closely to reality,
461 * or at least to modern reality. It comes from
462 *
463 * http://ftp.s390.ibm.com/products/oe/bpxqp9.html
464 *
465 * and maps the characters of EBCDIC code page 1047 (the code used for
466 * Unix-derived software on IBM's 390 systems) to the corresponding
467 * characters from ISO 8859-1.
468 *
469 * If this table is used instead of the above one, some of the special
470 * cases for the NEL character can be taken out of the code.
471 */
472
473 private unsigned char ebcdic_1047_to_8859[] = {
474 0x00,0x01,0x02,0x03,0x9C,0x09,0x86,0x7F,0x97,0x8D,0x8E,0x0B,0x0C,0x0D,0x0E,0x0F,
475 0x10,0x11,0x12,0x13,0x9D,0x0A,0x08,0x87,0x18,0x19,0x92,0x8F,0x1C,0x1D,0x1E,0x1F,
476 0x80,0x81,0x82,0x83,0x84,0x85,0x17,0x1B,0x88,0x89,0x8A,0x8B,0x8C,0x05,0x06,0x07,
477 0x90,0x91,0x16,0x93,0x94,0x95,0x96,0x04,0x98,0x99,0x9A,0x9B,0x14,0x15,0x9E,0x1A,
478 0x20,0xA0,0xE2,0xE4,0xE0,0xE1,0xE3,0xE5,0xE7,0xF1,0xA2,0x2E,0x3C,0x28,0x2B,0x7C,
479 0x26,0xE9,0xEA,0xEB,0xE8,0xED,0xEE,0xEF,0xEC,0xDF,0x21,0x24,0x2A,0x29,0x3B,0x5E,
480 0x2D,0x2F,0xC2,0xC4,0xC0,0xC1,0xC3,0xC5,0xC7,0xD1,0xA6,0x2C,0x25,0x5F,0x3E,0x3F,
481 0xF8,0xC9,0xCA,0xCB,0xC8,0xCD,0xCE,0xCF,0xCC,0x60,0x3A,0x23,0x40,0x27,0x3D,0x22,
482 0xD8,0x61,0x62,0x63,0x64,0x65,0x66,0x67,0x68,0x69,0xAB,0xBB,0xF0,0xFD,0xFE,0xB1,
483 0xB0,0x6A,0x6B,0x6C,0x6D,0x6E,0x6F,0x70,0x71,0x72,0xAA,0xBA,0xE6,0xB8,0xC6,0xA4,
484 0xB5,0x7E,0x73,0x74,0x75,0x76,0x77,0x78,0x79,0x7A,0xA1,0xBF,0xD0,0x5B,0xDE,0xAE,
485 0xAC,0xA3,0xA5,0xB7,0xA9,0xA7,0xB6,0xBC,0xBD,0xBE,0xDD,0xA8,0xAF,0x5D,0xB4,0xD7,
486 0x7B,0x41,0x42,0x43,0x44,0x45,0x46,0x47,0x48,0x49,0xAD,0xF4,0xF6,0xF2,0xF3,0xF5,
487 0x7D,0x4A,0x4B,0x4C,0x4D,0x4E,0x4F,0x50,0x51,0x52,0xB9,0xFB,0xFC,0xF9,0xFA,0xFF,
488 0x5C,0xF7,0x53,0x54,0x55,0x56,0x57,0x58,0x59,0x5A,0xB2,0xD4,0xD6,0xD2,0xD3,0xD5,
489 0x30,0x31,0x32,0x33,0x34,0x35,0x36,0x37,0x38,0x39,0xB3,0xDB,0xDC,0xD9,0xDA,0x9F
490 };
491 #endif
492
493 /*
494 * Copy buf[0 ... nbytes-1] into out[], translating EBCDIC to ASCII.
495 */
496 private void
497 from_ebcdic(const unsigned char *buf, size_t nbytes, unsigned char *out)
498 {
499 size_t i;
500
501 for (i = 0; i < nbytes; i++) {
502 out[i] = ebcdic_to_ascii[buf[i]];
503 }
504 }