113 rv = 0; 114 *type = "binary"; 115 } 116 } 117 118 done: 119 if (nbuf) 120 free(nbuf); 121 122 return rv; 123} 124 125/* 126 * This table reflects a particular philosophy about what constitutes 127 * "text," and there is room for disagreement about it. 128 * 129 * Version 3.31 of the file command considered a file to be ASCII if 130 * each of its characters was approved by either the isascii() or 131 * isalpha() function. On most systems, this would mean that any 132 * file consisting only of characters in the range 0x00 ... 0x7F 133 * would be called ASCII text, but many systems might reasonably 134 * consider some characters outside this range to be alphabetic, 135 * so the file command would call such characters ASCII. It might 136 * have been more accurate to call this "considered textual on the 137 * local system" than "ASCII." 138 * 139 * It considered a file to be "International language text" if each 140 * of its characters was either an ASCII printing character (according 141 * to the real ASCII standard, not the above test), a character in 142 * the range 0x80 ... 0xFF, or one of the following control characters: 143 * backspace, tab, line feed, vertical tab, form feed, carriage return, 144 * escape. No attempt was made to determine the language in which files 145 * of this type were written. 146 * 147 * 148 * The table below considers a file to be ASCII if all of its characters 149 * are either ASCII printing characters (again, according to the X3.4 150 * standard, not isascii()) or any of the following controls: bell, 151 * backspace, tab, line feed, form feed, carriage return, esc, nextline. 152 * 153 * I include bell because some programs (particularly shell scripts) 154 * use it literally, even though it is rare in normal text. I exclude 155 * vertical tab because it never seems to be used in real text. I also 156 * include, with hesitation, the X3.64/ECMA-43 control nextline (0x85), 157 * because that's what the dd EBCDIC->ASCII table maps the EBCDIC newline 158 * character to. It might be more appropriate to include it in the 8859 159 * set instead of the ASCII set, but it's got to be included in *something* 160 * we recognize or EBCDIC files aren't going to be considered textual. 161 * Some old Unix source files use SO/SI (^N/^O) to shift between Greek 162 * and Latin characters, so these should possibly be allowed. But they 163 * make a real mess on VT100-style displays if they're not paired properly, 164 * so we are probably better off not calling them text. 165 * 166 * A file is considered to be ISO-8859 text if its characters are all 167 * either ASCII, according to the above definition, or printing characters 168 * from the ISO-8859 8-bit extension, characters 0xA0 ... 0xFF. 169 * 170 * Finally, a file is considered to be international text from some other 171 * character code if its characters are all either ISO-8859 (according to 172 * the above definition) or characters in the range 0x80 ... 0x9F, which 173 * ISO-8859 considers to be control characters but the IBM PC and Macintosh 174 * consider to be printing characters. 175 */ 176 177#define F 0 /* character never appears in text */ 178#define T 1 /* character appears in plain ASCII text */ 179#define I 2 /* character appears in ISO-8859 text */ 180#define X 3 /* character appears in non-ISO extended ASCII (Mac, IBM PC) */ 181 182private char text_chars[256] = { 183 /* BEL BS HT LF FF CR */ 184 F, F, F, F, F, F, F, T, T, T, T, F, T, T, F, F, /* 0x0X */ 185 /* ESC */ 186 F, F, F, F, F, F, F, F, F, F, F, T, F, F, F, F, /* 0x1X */ 187 T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, /* 0x2X */ 188 T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, /* 0x3X */ 189 T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, /* 0x4X */ 190 T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, /* 0x5X */ 191 T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, /* 0x6X */ 192 T, T, T, T, T, T, T, T, T, T, T, T, T, T, T, F, /* 0x7X */ 193 /* NEL */ 194 X, X, X, X, X, T, X, X, X, X, X, X, X, X, X, X, /* 0x8X */ 195 X, X, X, X, X, X, X, X, X, X, X, X, X, X, X, X, /* 0x9X */ 196 I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, /* 0xaX */ 197 I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, /* 0xbX */ 198 I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, /* 0xcX */ 199 I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, /* 0xdX */ 200 I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, /* 0xeX */ 201 I, I, I, I, I, I, I, I, I, I, I, I, I, I, I, I /* 0xfX */ 202}; 203 204private int 205looks_ascii(const unsigned char *buf, size_t nbytes, unichar *ubuf, 206 size_t *ulen) 207{ 208 size_t i; 209 210 *ulen = 0; 211 212 for (i = 0; i < nbytes; i++) { 213 int t = text_chars[buf[i]]; 214 215 if (t != T) 216 return 0; 217 218 ubuf[(*ulen)++] = buf[i]; 219 } 220 221 return 1; 222} 223 224private int 225looks_latin1(const unsigned char *buf, size_t nbytes, unichar *ubuf, size_t *ulen) 226{ 227 size_t i; 228 229 *ulen = 0; 230 231 for (i = 0; i < nbytes; i++) { 232 int t = text_chars[buf[i]]; 233 234 if (t != T && t != I) 235 return 0; 236 237 ubuf[(*ulen)++] = buf[i]; 238 } 239 240 return 1; 241} 242 243private int 244looks_extended(const unsigned char *buf, size_t nbytes, unichar *ubuf, 245 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 && t != X) 255 return 0; 256 257 ubuf[(*ulen)++] = buf[i]; 258 } 259 260 return 1; 261} 262 263/* 264 * Decide whether some text looks like UTF-8. Returns: 265 * 266 * -1: invalid UTF-8 267 * 0: uses odd control characters, so doesn't look like text 268 * 1: 7-bit text 269 * 2: definitely UTF-8 text (valid high-bit set bytes) 270 * 271 * If ubuf is non-NULL on entry, text is decoded into ubuf, *ulen; 272 * ubuf must be big enough! 273 */ 274protected int 275file_looks_utf8(const unsigned char *buf, size_t nbytes, unichar *ubuf, size_t *ulen) 276{ 277 size_t i; 278 int n; 279 unichar c; 280 int gotone = 0, ctrl = 0; 281 282 if (ubuf) 283 *ulen = 0; 284 285 for (i = 0; i < nbytes; i++) { 286 if ((buf[i] & 0x80) == 0) { /* 0xxxxxxx is plain ASCII */ 287 /* 288 * Even if the whole file is valid UTF-8 sequences, 289 * still reject it if it uses weird control characters. 290 */ 291 292 if (text_chars[buf[i]] != T) 293 ctrl = 1; 294 295 if (ubuf) 296 ubuf[(*ulen)++] = buf[i]; 297 } else if ((buf[i] & 0x40) == 0) { /* 10xxxxxx never 1st byte */ 298 return -1; 299 } else { /* 11xxxxxx begins UTF-8 */ 300 int following; 301 302 if ((buf[i] & 0x20) == 0) { /* 110xxxxx */ 303 c = buf[i] & 0x1f; 304 following = 1; 305 } else if ((buf[i] & 0x10) == 0) { /* 1110xxxx */ 306 c = buf[i] & 0x0f; 307 following = 2; 308 } else if ((buf[i] & 0x08) == 0) { /* 11110xxx */ 309 c = buf[i] & 0x07; 310 following = 3; 311 } else if ((buf[i] & 0x04) == 0) { /* 111110xx */ 312 c = buf[i] & 0x03; 313 following = 4; 314 } else if ((buf[i] & 0x02) == 0) { /* 1111110x */ 315 c = buf[i] & 0x01; 316 following = 5; 317 } else 318 return -1; 319 320 for (n = 0; n < following; n++) { 321 i++; 322 if (i >= nbytes) 323 goto done; 324 325 if ((buf[i] & 0x80) == 0 || (buf[i] & 0x40)) 326 return -1; 327 328 c = (c << 6) + (buf[i] & 0x3f); 329 } 330 331 if (ubuf) 332 ubuf[(*ulen)++] = c; 333 gotone = 1; 334 } 335 } 336done: 337 return ctrl ? 0 : (gotone ? 2 : 1); 338} 339 340/* 341 * Decide whether some text looks like UTF-8 with BOM. If there is no 342 * BOM, return -1; otherwise return the result of looks_utf8 on the 343 * rest of the text. 344 */ 345private int 346looks_utf8_with_BOM(const unsigned char *buf, size_t nbytes, unichar *ubuf, 347 size_t *ulen) 348{ 349 if (nbytes > 3 && buf[0] == 0xef && buf[1] == 0xbb && buf[2] == 0xbf) 350 return file_looks_utf8(buf + 3, nbytes - 3, ubuf, ulen); 351 else 352 return -1; 353} 354 355private int 356looks_ucs16(const unsigned char *buf, size_t nbytes, unichar *ubuf, 357 size_t *ulen) 358{ 359 int bigend; 360 size_t i; 361 362 if (nbytes < 2) 363 return 0; 364 365 if (buf[0] == 0xff && buf[1] == 0xfe) 366 bigend = 0; 367 else if (buf[0] == 0xfe && buf[1] == 0xff) 368 bigend = 1; 369 else 370 return 0; 371 372 *ulen = 0; 373 374 for (i = 2; i + 1 < nbytes; i += 2) { 375 /* XXX fix to properly handle chars > 65536 */ 376 377 if (bigend) 378 ubuf[(*ulen)++] = buf[i + 1] + 256 * buf[i]; 379 else 380 ubuf[(*ulen)++] = buf[i] + 256 * buf[i + 1]; 381 382 if (ubuf[*ulen - 1] == 0xfffe) 383 return 0; 384 if (ubuf[*ulen - 1] < 128 && 385 text_chars[(size_t)ubuf[*ulen - 1]] != T) 386 return 0; 387 } 388 389 return 1 + bigend; 390} 391 392#undef F 393#undef T 394#undef I 395#undef X 396 397/* 398 * This table maps each EBCDIC character to an (8-bit extended) ASCII 399 * character, as specified in the rationale for the dd(1) command in 400 * draft 11.2 (September, 1991) of the POSIX P1003.2 standard. 401 * 402 * Unfortunately it does not seem to correspond exactly to any of the 403 * five variants of EBCDIC documented in IBM's _Enterprise Systems 404 * Architecture/390: Principles of Operation_, SA22-7201-06, Seventh 405 * Edition, July, 1999, pp. I-1 - I-4. 406 * 407 * Fortunately, though, all versions of EBCDIC, including this one, agree 408 * on most of the printing characters that also appear in (7-bit) ASCII. 409 * Of these, only '|', '!', '~', '^', '[', and ']' are in question at all. 410 * 411 * Fortunately too, there is general agreement that codes 0x00 through 412 * 0x3F represent control characters, 0x41 a nonbreaking space, and the 413 * remainder printing characters. 414 * 415 * This is sufficient to allow us to identify EBCDIC text and to distinguish 416 * between old-style and internationalized examples of text. 417 */ 418 419private unsigned char ebcdic_to_ascii[] = { 420 0, 1, 2, 3, 156, 9, 134, 127, 151, 141, 142, 11, 12, 13, 14, 15, 421 16, 17, 18, 19, 157, 133, 8, 135, 24, 25, 146, 143, 28, 29, 30, 31, 422128, 129, 130, 131, 132, 10, 23, 27, 136, 137, 138, 139, 140, 5, 6, 7, 423144, 145, 22, 147, 148, 149, 150, 4, 152, 153, 154, 155, 20, 21, 158, 26, 424' ', 160, 161, 162, 163, 164, 165, 166, 167, 168, 213, '.', '<', '(', '+', '|', 425'&', 169, 170, 171, 172, 173, 174, 175, 176, 177, '!', '$', '*', ')', ';', '~', 426'-', '/', 178, 179, 180, 181, 182, 183, 184, 185, 203, ',', '%', '_', '>', '?', 427186, 187, 188, 189, 190, 191, 192, 193, 194, '`', ':', '#', '@', '\'','=', '"', 428195, 'a', 'b', 'c', 'd', 'e', 'f', 'g', 'h', 'i', 196, 197, 198, 199, 200, 201, 429202, 'j', 'k', 'l', 'm', 'n', 'o', 'p', 'q', 'r', '^', 204, 205, 206, 207, 208, 430209, 229, 's', 't', 'u', 'v', 'w', 'x', 'y', 'z', 210, 211, 212, '[', 214, 215, 431216, 217, 218, 219, 220, 221, 222, 223, 224, 225, 226, 227, 228, ']', 230, 231, 432'{', 'A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', 'I', 232, 233, 234, 235, 236, 237, 433'}', 'J', 'K', 'L', 'M', 'N', 'O', 'P', 'Q', 'R', 238, 239, 240, 241, 242, 243, 434'\\',159, 'S', 'T', 'U', 'V', 'W', 'X', 'Y', 'Z', 244, 245, 246, 247, 248, 249, 435'0', '1', '2', '3', '4', '5', '6', '7', '8', '9', 250, 251, 252, 253, 254, 255 436}; 437 438#ifdef notdef 439/* 440 * The following EBCDIC-to-ASCII table may relate more closely to reality, 441 * or at least to modern reality. It comes from 442 * 443 * http://ftp.s390.ibm.com/products/oe/bpxqp9.html 444 * 445 * and maps the characters of EBCDIC code page 1047 (the code used for 446 * Unix-derived software on IBM's 390 systems) to the corresponding 447 * characters from ISO 8859-1. 448 * 449 * If this table is used instead of the above one, some of the special 450 * cases for the NEL character can be taken out of the code. 451 */ 452 453private unsigned char ebcdic_1047_to_8859[] = { 4540x00,0x01,0x02,0x03,0x9C,0x09,0x86,0x7F,0x97,0x8D,0x8E,0x0B,0x0C,0x0D,0x0E,0x0F, 4550x10,0x11,0x12,0x13,0x9D,0x0A,0x08,0x87,0x18,0x19,0x92,0x8F,0x1C,0x1D,0x1E,0x1F, 4560x80,0x81,0x82,0x83,0x84,0x85,0x17,0x1B,0x88,0x89,0x8A,0x8B,0x8C,0x05,0x06,0x07, 4570x90,0x91,0x16,0x93,0x94,0x95,0x96,0x04,0x98,0x99,0x9A,0x9B,0x14,0x15,0x9E,0x1A, 4580x20,0xA0,0xE2,0xE4,0xE0,0xE1,0xE3,0xE5,0xE7,0xF1,0xA2,0x2E,0x3C,0x28,0x2B,0x7C, 4590x26,0xE9,0xEA,0xEB,0xE8,0xED,0xEE,0xEF,0xEC,0xDF,0x21,0x24,0x2A,0x29,0x3B,0x5E, 4600x2D,0x2F,0xC2,0xC4,0xC0,0xC1,0xC3,0xC5,0xC7,0xD1,0xA6,0x2C,0x25,0x5F,0x3E,0x3F, 4610xF8,0xC9,0xCA,0xCB,0xC8,0xCD,0xCE,0xCF,0xCC,0x60,0x3A,0x23,0x40,0x27,0x3D,0x22, 4620xD8,0x61,0x62,0x63,0x64,0x65,0x66,0x67,0x68,0x69,0xAB,0xBB,0xF0,0xFD,0xFE,0xB1, 4630xB0,0x6A,0x6B,0x6C,0x6D,0x6E,0x6F,0x70,0x71,0x72,0xAA,0xBA,0xE6,0xB8,0xC6,0xA4, 4640xB5,0x7E,0x73,0x74,0x75,0x76,0x77,0x78,0x79,0x7A,0xA1,0xBF,0xD0,0x5B,0xDE,0xAE, 4650xAC,0xA3,0xA5,0xB7,0xA9,0xA7,0xB6,0xBC,0xBD,0xBE,0xDD,0xA8,0xAF,0x5D,0xB4,0xD7, 4660x7B,0x41,0x42,0x43,0x44,0x45,0x46,0x47,0x48,0x49,0xAD,0xF4,0xF6,0xF2,0xF3,0xF5, 4670x7D,0x4A,0x4B,0x4C,0x4D,0x4E,0x4F,0x50,0x51,0x52,0xB9,0xFB,0xFC,0xF9,0xFA,0xFF, 4680x5C,0xF7,0x53,0x54,0x55,0x56,0x57,0x58,0x59,0x5A,0xB2,0xD4,0xD6,0xD2,0xD3,0xD5, 4690x30,0x31,0x32,0x33,0x34,0x35,0x36,0x37,0x38,0x39,0xB3,0xDB,0xDC,0xD9,0xDA,0x9F 470}; 471#endif 472 473/* 474 * Copy buf[0 ... nbytes-1] into out[], translating EBCDIC to ASCII. 475 */ 476private void 477from_ebcdic(const unsigned char *buf, size_t nbytes, unsigned char *out) 478{ 479 size_t i; 480 481 for (i = 0; i < nbytes; i++) { 482 out[i] = ebcdic_to_ascii[buf[i]]; 483 } 484}
| 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 199private 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 221private int 222looks_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 241private int 242looks_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 260private int 261looks_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 */ 291protected int 292file_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 } 353done: 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 */ 362private int 363looks_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 372private int 373looks_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 436private 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, 439128, 129, 130, 131, 132, 10, 23, 27, 136, 137, 138, 139, 140, 5, 6, 7, 440144, 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, ',', '%', '_', '>', '?', 444186, 187, 188, 189, 190, 191, 192, 193, 194, '`', ':', '#', '@', '\'','=', '"', 445195, 'a', 'b', 'c', 'd', 'e', 'f', 'g', 'h', 'i', 196, 197, 198, 199, 200, 201, 446202, 'j', 'k', 'l', 'm', 'n', 'o', 'p', 'q', 'r', '^', 204, 205, 206, 207, 208, 447209, 229, 's', 't', 'u', 'v', 'w', 'x', 'y', 'z', 210, 211, 212, '[', 214, 215, 448216, 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 470private unsigned char ebcdic_1047_to_8859[] = { 4710x00,0x01,0x02,0x03,0x9C,0x09,0x86,0x7F,0x97,0x8D,0x8E,0x0B,0x0C,0x0D,0x0E,0x0F, 4720x10,0x11,0x12,0x13,0x9D,0x0A,0x08,0x87,0x18,0x19,0x92,0x8F,0x1C,0x1D,0x1E,0x1F, 4730x80,0x81,0x82,0x83,0x84,0x85,0x17,0x1B,0x88,0x89,0x8A,0x8B,0x8C,0x05,0x06,0x07, 4740x90,0x91,0x16,0x93,0x94,0x95,0x96,0x04,0x98,0x99,0x9A,0x9B,0x14,0x15,0x9E,0x1A, 4750x20,0xA0,0xE2,0xE4,0xE0,0xE1,0xE3,0xE5,0xE7,0xF1,0xA2,0x2E,0x3C,0x28,0x2B,0x7C, 4760x26,0xE9,0xEA,0xEB,0xE8,0xED,0xEE,0xEF,0xEC,0xDF,0x21,0x24,0x2A,0x29,0x3B,0x5E, 4770x2D,0x2F,0xC2,0xC4,0xC0,0xC1,0xC3,0xC5,0xC7,0xD1,0xA6,0x2C,0x25,0x5F,0x3E,0x3F, 4780xF8,0xC9,0xCA,0xCB,0xC8,0xCD,0xCE,0xCF,0xCC,0x60,0x3A,0x23,0x40,0x27,0x3D,0x22, 4790xD8,0x61,0x62,0x63,0x64,0x65,0x66,0x67,0x68,0x69,0xAB,0xBB,0xF0,0xFD,0xFE,0xB1, 4800xB0,0x6A,0x6B,0x6C,0x6D,0x6E,0x6F,0x70,0x71,0x72,0xAA,0xBA,0xE6,0xB8,0xC6,0xA4, 4810xB5,0x7E,0x73,0x74,0x75,0x76,0x77,0x78,0x79,0x7A,0xA1,0xBF,0xD0,0x5B,0xDE,0xAE, 4820xAC,0xA3,0xA5,0xB7,0xA9,0xA7,0xB6,0xBC,0xBD,0xBE,0xDD,0xA8,0xAF,0x5D,0xB4,0xD7, 4830x7B,0x41,0x42,0x43,0x44,0x45,0x46,0x47,0x48,0x49,0xAD,0xF4,0xF6,0xF2,0xF3,0xF5, 4840x7D,0x4A,0x4B,0x4C,0x4D,0x4E,0x4F,0x50,0x51,0x52,0xB9,0xFB,0xFC,0xF9,0xFA,0xFF, 4850x5C,0xF7,0x53,0x54,0x55,0x56,0x57,0x58,0x59,0x5A,0xB2,0xD4,0xD6,0xD2,0xD3,0xD5, 4860x30,0x31,0x32,0x33,0x34,0x35,0x36,0x37,0x38,0x39,0xB3,0xDB,0xDC,0xD9,0xDA,0x9F 487}; 488#endif 489 490/* 491 * Copy buf[0 ... nbytes-1] into out[], translating EBCDIC to ASCII. 492 */ 493private void 494from_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}
|