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 1.5 2010/07/21 16:47:17 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 mlen = (nbytes + 1) * sizeof(nbuf[0]); 75 if ((nbuf = CAST(unsigned char *, calloc((size_t)1, mlen))) == NULL) { 76 file_oomem(ms, mlen); 77 goto done; 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 85 *type = "text"; 86 if (looks_ascii(buf, nbytes, *ubuf, ulen)) { 87 DPRINTF(("ascii %" SIZE_T_FORMAT "u\n", *ulen)); 88 *code = "ASCII"; 89 *code_mime = "us-ascii"; 90 } else if (looks_utf8_with_BOM(buf, nbytes, *ubuf, ulen) > 0) { 91 DPRINTF(("utf8/bom %" SIZE_T_FORMAT "u\n", *ulen)); 92 *code = "UTF-8 Unicode (with BOM)"; 93 *code_mime = "utf-8"; 94 } else if (file_looks_utf8(buf, nbytes, *ubuf, ulen) > 1) { 95 DPRINTF(("utf8 %" SIZE_T_FORMAT "u\n", *ulen)); 96 *code = "UTF-8 Unicode (with BOM)"; 97 *code = "UTF-8 Unicode"; 98 *code_mime = "utf-8"; 99 } else if ((ucs_type = looks_ucs16(buf, nbytes, *ubuf, ulen)) != 0) { 100 if (ucs_type == 1) { 101 *code = "Little-endian UTF-16 Unicode"; 102 *code_mime = "utf-16le"; 103 } else { 104 *code = "Big-endian UTF-16 Unicode"; 105 *code_mime = "utf-16be"; 106 } 107 DPRINTF(("ucs16 %" SIZE_T_FORMAT "u\n", *ulen)); 108 } else if (looks_latin1(buf, nbytes, *ubuf, ulen)) { 109 DPRINTF(("latin1 %" SIZE_T_FORMAT "u\n", *ulen)); 110 *code = "ISO-8859"; 111 *code_mime = "iso-8859-1"; 112 } else if (looks_extended(buf, nbytes, *ubuf, ulen)) { 113 DPRINTF(("extended %" SIZE_T_FORMAT "u\n", *ulen)); 114 *code = "Non-ISO extended-ASCII"; 115 *code_mime = "unknown-8bit"; 116 } else { 117 from_ebcdic(buf, nbytes, nbuf); 118 119 if (looks_ascii(nbuf, nbytes, *ubuf, ulen)) { 120 DPRINTF(("ebcdic %" SIZE_T_FORMAT "u\n", *ulen)); 121 *code = "EBCDIC"; 122 *code_mime = "ebcdic"; 123 } else if (looks_latin1(nbuf, nbytes, *ubuf, ulen)) { 124 DPRINTF(("ebcdic/international %" SIZE_T_FORMAT "u\n", 125 *ulen)); 126 *code = "International EBCDIC"; 127 *code_mime = "ebcdic"; 128 } else { /* Doesn't look like text at all */ 129 DPRINTF(("binary\n")); 130 rv = 0; 131 *type = "binary"; 132 } 133 } 134 135 done: 136 if (nbuf) 137 free(nbuf); 138 139 return rv; 140 } 141 142 /* 143 * This table reflects a particular philosophy about what constitutes 144 * "text," and there is room for disagreement about it. 145 * 146 * Version 3.31 of the file command considered a file to be ASCII if 147 * each of its characters was approved by either the isascii() or 148 * isalpha() function. On most systems, this would mean that any 149 * file consisting only of characters in the range 0x00 ... 0x7F 150 * would be called ASCII text, but many systems might reasonably 151 * consider some characters outside this range to be alphabetic, 152 * so the file command would call such characters ASCII. It might 153 * have been more accurate to call this "considered textual on the 154 * local system" than "ASCII." 155 * 156 * It considered a file to be "International language text" if each 157 * of its characters was either an ASCII printing character (according 158 * to the real ASCII standard, not the above test), a character in 159 * the range 0x80 ... 0xFF, or one of the following control characters: 160 * backspace, tab, line feed, vertical tab, form feed, carriage return, 161 * escape. No attempt was made to determine the language in which files 162 * of this type were written. 163 * 164 * 165 * The table below considers a file to be ASCII if all of its characters 166 * are either ASCII printing characters (again, according to the X3.4 167 * standard, not isascii()) or any of the following controls: bell, 168 * backspace, tab, line feed, form feed, carriage return, esc, nextline. 169 * 170 * I include bell because some programs (particularly shell scripts) 171 * use it literally, even though it is rare in normal text. I exclude 172 * vertical tab because it never seems to be used in real text. I also 173 * include, with hesitation, the X3.64/ECMA-43 control nextline (0x85), 174 * because that's what the dd EBCDIC->ASCII table maps the EBCDIC newline 175 * character to. It might be more appropriate to include it in the 8859 176 * set instead of the ASCII set, but it's got to be included in *something* 177 * we recognize or EBCDIC files aren't going to be considered textual. 178 * Some old Unix source files use SO/SI (^N/^O) to shift between Greek 179 * and Latin characters, so these should possibly be allowed. But they 180 * make a real mess on VT100-style displays if they're not paired properly, 181 * so we are probably better off not calling them text. 182 * 183 * A file is considered to be ISO-8859 text if its characters are all 184 * either ASCII, according to the above definition, or printing characters 185 * from the ISO-8859 8-bit extension, characters 0xA0 ... 0xFF. 186 * 187 * Finally, a file is considered to be international text from some other 188 * character code if its characters are all either ISO-8859 (according to 189 * the above definition) or characters in the range 0x80 ... 0x9F, which 190 * ISO-8859 considers to be control characters but the IBM PC and Macintosh 191 * consider to be printing characters. 192 */ 193 194 #define F 0 /* character never appears in text */ 195 #define T 1 /* character appears in plain ASCII text */ 196 #define I 2 /* character appears in ISO-8859 text */ 197 #define X 3 /* character appears in non-ISO extended ASCII (Mac, IBM PC) */ 198 199 private char text_chars[256] = { 200 /* BEL BS HT LF FF CR */ 201 F, F, F, F, F, F, F, T, T, T, T, F, T, T, F, F, /* 0x0X */ 202 /* ESC */ 203 F, F, F, F, F, F, F, F, F, F, F, T, F, F, F, F, /* 0x1X */ 204 T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, /* 0x2X */ 205 T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, /* 0x3X */ 206 T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, /* 0x4X */ 207 T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, /* 0x5X */ 208 T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, /* 0x6X */ 209 T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, F, /* 0x7X */ 210 /* NEL */ 211 X, X, X, X, X, T, X, X, X, X, X, X, X, X, X, X, /* 0x8X */ 212 X, X, X, X, X, X, X, X, X, X, X, X, X, X, X, X, /* 0x9X */ 213 I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, /* 0xaX */ 214 I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, /* 0xbX */ 215 I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, /* 0xcX */ 216 I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, /* 0xdX */ 217 I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, /* 0xeX */ 218 I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, I /* 0xfX */ 219 }; 220 221 private int 222 looks_ascii(const unsigned char *buf, size_t nbytes, unichar *ubuf, 223 size_t *ulen) 224 { 225 size_t i; 226 227 *ulen = 0; 228 229 for (i = 0; i < nbytes; i++) { 230 int t = text_chars[buf[i]]; 231 232 if (t != T) 233 return 0; 234 235 ubuf[(*ulen)++] = buf[i]; 236 } 237 238 return 1; 239 } 240 241 private int 242 looks_latin1(const unsigned char *buf, size_t nbytes, unichar *ubuf, size_t *ulen) 243 { 244 size_t i; 245 246 *ulen = 0; 247 248 for (i = 0; i < nbytes; i++) { 249 int t = text_chars[buf[i]]; 250 251 if (t != T && t != I) 252 return 0; 253 254 ubuf[(*ulen)++] = buf[i]; 255 } 256 257 return 1; 258 } 259 260 private int 261 looks_extended(const unsigned char *buf, size_t nbytes, unichar *ubuf, 262 size_t *ulen) 263 { 264 size_t i; 265 266 *ulen = 0; 267 268 for (i = 0; i < nbytes; i++) { 269 int t = text_chars[buf[i]]; 270 271 if (t != T && t != I && t != X) 272 return 0; 273 274 ubuf[(*ulen)++] = buf[i]; 275 } 276 277 return 1; 278 } 279 280 /* 281 * Decide whether some text looks like UTF-8. Returns: 282 * 283 * -1: invalid UTF-8 284 * 0: uses odd control characters, so doesn't look like text 285 * 1: 7-bit text 286 * 2: definitely UTF-8 text (valid high-bit set bytes) 287 * 288 * If ubuf is non-NULL on entry, text is decoded into ubuf, *ulen; 289 * ubuf must be big enough! 290 */ 291 protected int 292 file_looks_utf8(const unsigned char *buf, size_t nbytes, unichar *ubuf, size_t *ulen) 293 { 294 size_t i; 295 int n; 296 unichar c; 297 int gotone = 0, ctrl = 0; 298 299 if (ubuf) 300 *ulen = 0; 301 302 for (i = 0; i < nbytes; i++) { 303 if ((buf[i] & 0x80) == 0) { /* 0xxxxxxx is plain ASCII */ 304 /* 305 * Even if the whole file is valid UTF-8 sequences, 306 * still reject it if it uses weird control characters. 307 */ 308 309 if (text_chars[buf[i]] != T) 310 ctrl = 1; 311 312 if (ubuf) 313 ubuf[(*ulen)++] = buf[i]; 314 } else if ((buf[i] & 0x40) == 0) { /* 10xxxxxx never 1st byte */ 315 return -1; 316 } else { /* 11xxxxxx begins UTF-8 */ 317 int following; 318 319 if ((buf[i] & 0x20) == 0) { /* 110xxxxx */ 320 c = buf[i] & 0x1f; 321 following = 1; 322 } else if ((buf[i] & 0x10) == 0) { /* 1110xxxx */ 323 c = buf[i] & 0x0f; 324 following = 2; 325 } else if ((buf[i] & 0x08) == 0) { /* 11110xxx */ 326 c = buf[i] & 0x07; 327 following = 3; 328 } else if ((buf[i] & 0x04) == 0) { /* 111110xx */ 329 c = buf[i] & 0x03; 330 following = 4; 331 } else if ((buf[i] & 0x02) == 0) { /* 1111110x */ 332 c = buf[i] & 0x01; 333 following = 5; 334 } else 335 return -1; 336 337 for (n = 0; n < following; n++) { 338 i++; 339 if (i >= nbytes) 340 goto done; 341 342 if ((buf[i] & 0x80) == 0 || (buf[i] & 0x40)) 343 return -1; 344 345 c = (c << 6) + (buf[i] & 0x3f); 346 } 347 348 if (ubuf) 349 ubuf[(*ulen)++] = c; 350 gotone = 1; 351 } 352 } 353 done: 354 return ctrl ? 0 : (gotone ? 2 : 1); 355 } 356 357 /* 358 * Decide whether some text looks like UTF-8 with BOM. If there is no 359 * BOM, return -1; otherwise return the result of looks_utf8 on the 360 * rest of the text. 361 */ 362 private int 363 looks_utf8_with_BOM(const unsigned char *buf, size_t nbytes, unichar *ubuf, 364 size_t *ulen) 365 { 366 if (nbytes > 3 && buf[0] == 0xef && buf[1] == 0xbb && buf[2] == 0xbf) 367 return file_looks_utf8(buf + 3, nbytes - 3, ubuf, ulen); 368 else 369 return -1; 370 } 371 372 private int 373 looks_ucs16(const unsigned char *buf, size_t nbytes, unichar *ubuf, 374 size_t *ulen) 375 { 376 int bigend; 377 size_t i; 378 379 if (nbytes < 2) 380 return 0; 381 382 if (buf[0] == 0xff && buf[1] == 0xfe) 383 bigend = 0; 384 else if (buf[0] == 0xfe && buf[1] == 0xff) 385 bigend = 1; 386 else 387 return 0; 388 389 *ulen = 0; 390 391 for (i = 2; i + 1 < nbytes; i += 2) { 392 /* XXX fix to properly handle chars > 65536 */ 393 394 if (bigend) 395 ubuf[(*ulen)++] = buf[i + 1] + 256 * buf[i]; 396 else 397 ubuf[(*ulen)++] = buf[i] + 256 * buf[i + 1]; 398 399 if (ubuf[*ulen - 1] == 0xfffe) 400 return 0; 401 if (ubuf[*ulen - 1] < 128 && 402 text_chars[(size_t)ubuf[*ulen - 1]] != T) 403 return 0; 404 } 405 406 return 1 + bigend; 407 } 408 409 #undef F 410 #undef T 411 #undef I 412 #undef X 413 414 /* 415 * This table maps each EBCDIC character to an (8-bit extended) ASCII 416 * character, as specified in the rationale for the dd(1) command in 417 * draft 11.2 (September, 1991) of the POSIX P1003.2 standard. 418 * 419 * Unfortunately it does not seem to correspond exactly to any of the 420 * five variants of EBCDIC documented in IBM's _Enterprise Systems 421 * Architecture/390: Principles of Operation_, SA22-7201-06, Seventh 422 * Edition, July, 1999, pp. I-1 - I-4. 423 * 424 * Fortunately, though, all versions of EBCDIC, including this one, agree 425 * on most of the printing characters that also appear in (7-bit) ASCII. 426 * Of these, only '|', '!', '~', '^', '[', and ']' are in question at all. 427 * 428 * Fortunately too, there is general agreement that codes 0x00 through 429 * 0x3F represent control characters, 0x41 a nonbreaking space, and the 430 * remainder printing characters. 431 * 432 * This is sufficient to allow us to identify EBCDIC text and to distinguish 433 * between old-style and internationalized examples of text. 434 */ 435 436 private unsigned char ebcdic_to_ascii[] = { 437 0, 1, 2, 3, 156, 9, 134, 127, 151, 141, 142, 11, 12, 13, 14, 15, 438 16, 17, 18, 19, 157, 133, 8, 135, 24, 25, 146, 143, 28, 29, 30, 31, 439 128, 129, 130, 131, 132, 10, 23, 27, 136, 137, 138, 139, 140, 5, 6, 7, 440 144, 145, 22, 147, 148, 149, 150, 4, 152, 153, 154, 155, 20, 21, 158, 26, 441 ' ', 160, 161, 162, 163, 164, 165, 166, 167, 168, 213, '.', '<', '(', '+', '|', 442 '&', 169, 170, 171, 172, 173, 174, 175, 176, 177, '!', '$', '*', ')', ';', '~', 443 '-', '/', 178, 179, 180, 181, 182, 183, 184, 185, 203, ',', '%', '_', '>', '?', 444 186, 187, 188, 189, 190, 191, 192, 193, 194, '`', ':', '#', '@', '\'','=', '"', 445 195, 'a', 'b', 'c', 'd', 'e', 'f', 'g', 'h', 'i', 196, 197, 198, 199, 200, 201, 446 202, 'j', 'k', 'l', 'm', 'n', 'o', 'p', 'q', 'r', '^', 204, 205, 206, 207, 208, 447 209, 229, 's', 't', 'u', 'v', 'w', 'x', 'y', 'z', 210, 211, 212, '[', 214, 215, 448 216, 217, 218, 219, 220, 221, 222, 223, 224, 225, 226, 227, 228, ']', 230, 231, 449 '{', 'A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', 'I', 232, 233, 234, 235, 236, 237, 450 '}', 'J', 'K', 'L', 'M', 'N', 'O', 'P', 'Q', 'R', 238, 239, 240, 241, 242, 243, 451 '\\',159, 'S', 'T', 'U', 'V', 'W', 'X', 'Y', 'Z', 244, 245, 246, 247, 248, 249, 452 '0', '1', '2', '3', '4', '5', '6', '7', '8', '9', 250, 251, 252, 253, 254, 255 453 }; 454 455 #ifdef notdef 456 /* 457 * The following EBCDIC-to-ASCII table may relate more closely to reality, 458 * or at least to modern reality. It comes from 459 * 460 * http://ftp.s390.ibm.com/products/oe/bpxqp9.html 461 * 462 * and maps the characters of EBCDIC code page 1047 (the code used for 463 * Unix-derived software on IBM's 390 systems) to the corresponding 464 * characters from ISO 8859-1. 465 * 466 * If this table is used instead of the above one, some of the special 467 * cases for the NEL character can be taken out of the code. 468 */ 469 470 private unsigned char ebcdic_1047_to_8859[] = { 471 0x00,0x01,0x02,0x03,0x9C,0x09,0x86,0x7F,0x97,0x8D,0x8E,0x0B,0x0C,0x0D,0x0E,0x0F, 472 0x10,0x11,0x12,0x13,0x9D,0x0A,0x08,0x87,0x18,0x19,0x92,0x8F,0x1C,0x1D,0x1E,0x1F, 473 0x80,0x81,0x82,0x83,0x84,0x85,0x17,0x1B,0x88,0x89,0x8A,0x8B,0x8C,0x05,0x06,0x07, 474 0x90,0x91,0x16,0x93,0x94,0x95,0x96,0x04,0x98,0x99,0x9A,0x9B,0x14,0x15,0x9E,0x1A, 475 0x20,0xA0,0xE2,0xE4,0xE0,0xE1,0xE3,0xE5,0xE7,0xF1,0xA2,0x2E,0x3C,0x28,0x2B,0x7C, 476 0x26,0xE9,0xEA,0xEB,0xE8,0xED,0xEE,0xEF,0xEC,0xDF,0x21,0x24,0x2A,0x29,0x3B,0x5E, 477 0x2D,0x2F,0xC2,0xC4,0xC0,0xC1,0xC3,0xC5,0xC7,0xD1,0xA6,0x2C,0x25,0x5F,0x3E,0x3F, 478 0xF8,0xC9,0xCA,0xCB,0xC8,0xCD,0xCE,0xCF,0xCC,0x60,0x3A,0x23,0x40,0x27,0x3D,0x22, 479 0xD8,0x61,0x62,0x63,0x64,0x65,0x66,0x67,0x68,0x69,0xAB,0xBB,0xF0,0xFD,0xFE,0xB1, 480 0xB0,0x6A,0x6B,0x6C,0x6D,0x6E,0x6F,0x70,0x71,0x72,0xAA,0xBA,0xE6,0xB8,0xC6,0xA4, 481 0xB5,0x7E,0x73,0x74,0x75,0x76,0x77,0x78,0x79,0x7A,0xA1,0xBF,0xD0,0x5B,0xDE,0xAE, 482 0xAC,0xA3,0xA5,0xB7,0xA9,0xA7,0xB6,0xBC,0xBD,0xBE,0xDD,0xA8,0xAF,0x5D,0xB4,0xD7, 483 0x7B,0x41,0x42,0x43,0x44,0x45,0x46,0x47,0x48,0x49,0xAD,0xF4,0xF6,0xF2,0xF3,0xF5, 484 0x7D,0x4A,0x4B,0x4C,0x4D,0x4E,0x4F,0x50,0x51,0x52,0xB9,0xFB,0xFC,0xF9,0xFA,0xFF, 485 0x5C,0xF7,0x53,0x54,0x55,0x56,0x57,0x58,0x59,0x5A,0xB2,0xD4,0xD6,0xD2,0xD3,0xD5, 486 0x30,0x31,0x32,0x33,0x34,0x35,0x36,0x37,0x38,0x39,0xB3,0xDB,0xDC,0xD9,0xDA,0x9F 487 }; 488 #endif 489 490 /* 491 * Copy buf[0 ... nbytes-1] into out[], translating EBCDIC to ASCII. 492 */ 493 private void 494 from_ebcdic(const unsigned char *buf, size_t nbytes, unsigned char *out) 495 { 496 size_t i; 497 498 for (i = 0; i < nbytes; i++) { 499 out[i] = ebcdic_to_ascii[buf[i]]; 500 } 501 } 502