contrib/file/src/ascmagic.c

   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  * ASCII magic -- try to detect text encoding.
  30  *
  31  * Extensively modified by Eric Fischer <enf@pobox.com> in July, 2000,
  32  * to handle character codes other than ASCII on a unified basis.
  33  */
  34
  35 #include "file.h"
  36
  37 #ifndef lint
  38 FILE_RCSID("@(#)$File: ascmagic.c,v 1.98 2017/11/02 20:25:39 christos Exp $")
  39 #endif  /* lint */
  40
  41 #include "magic.h"
  42 #include <string.h>
  43 #include <memory.h>
  44 #include <ctype.h>
  45 #include <stdlib.h>
  46 #ifdef HAVE_UNISTD_H
  47 #include <unistd.h>
  48 #endif
  49
  50 #define MAXLINELEN 300  /* longest sane line length */
  51 #define ISSPC(x) ((x) == ' ' || (x) == '\t' || (x) == '\r' || (x) == '\n' \
  52                   || (x) == 0x85 || (x) == '\f')
  53
  54 private unsigned char *encode_utf8(unsigned char *, size_t, unichar *, size_t);
  55 private size_t trim_nuls(const unsigned char *, size_t);
  56
  57 /*
  58  * Undo the NUL-termination kindly provided by process()
  59  * but leave at least one byte to look at
  60  */
  61 private size_t
  62 trim_nuls(const unsigned char *buf, size_t nbytes)
  63 {
  64         while (nbytes > 1 && buf[nbytes - 1] == '\0')
  65                 nbytes--;
  66
  67         return nbytes;
  68 }
  69
  70 protected int
  71 file_ascmagic(struct magic_set *ms, const struct buffer *b, int text)
  72 {
  73         unichar *ubuf = NULL;
  74         size_t ulen = 0;
  75         int rv = 1;
  76         struct buffer bb;
  77
  78         const char *code = NULL;
  79         const char *code_mime = NULL;
  80         const char *type = NULL;
  81
  82         bb = *b;
  83         bb.flen = trim_nuls(b->fbuf, b->flen);
  84
  85         /* If file doesn't look like any sort of text, give up. */
  86         if (file_encoding(ms, &bb, &ubuf, &ulen, &code, &code_mime,
  87             &type) == 0)
  88                 rv = 0;
  89         else
  90                 rv = file_ascmagic_with_encoding(ms, &bb,
  91                     ubuf, ulen, code, type, text);
  92
  93         free(ubuf);
  94
  95         return rv;
  96 }
  97
  98 protected int
  99 file_ascmagic_with_encoding(struct magic_set *ms,
 100     const struct buffer *b, unichar *ubuf, size_t ulen, const char *code,
 101     const char *type, int text)
 102 {
 103         struct buffer bb;
 104         const unsigned char *buf = b->fbuf;
 105         size_t nbytes = b->flen;
 106         unsigned char *utf8_buf = NULL, *utf8_end;
 107         size_t mlen, i;
 108         int rv = -1;
 109         int mime = ms->flags & MAGIC_MIME;
 110
 111         const char *subtype = NULL;
 112         const char *subtype_mime = NULL;
 113
 114         int has_escapes = 0;
 115         int has_backspace = 0;
 116         int seen_cr = 0;
 117
 118         int n_crlf = 0;
 119         int n_lf = 0;
 120         int n_cr = 0;
 121         int n_nel = 0;
 122         int executable = 0;
 123
 124         size_t last_line_end = (size_t)-1;
 125         int has_long_lines = 0;
 126
 127         nbytes = trim_nuls(buf, nbytes);
 128
 129         /* If we have fewer than 2 bytes, give up. */
 130         if (nbytes <= 1) {
 131                 rv = 0;
 132                 goto done;
 133         }
 134
 135         if (ulen > 0 && (ms->flags & MAGIC_NO_CHECK_SOFT) == 0) {
 136                 /* Convert ubuf to UTF-8 and try text soft magic */
 137                 /* malloc size is a conservative overestimate; could be
 138                    improved, or at least realloced after conversion. */
 139                 mlen = ulen * 6;
 140                 if ((utf8_buf = CAST(unsigned char *, malloc(mlen))) == NULL) {
 141                         file_oomem(ms, mlen);
 142                         goto done;
 143                 }
 144                 if ((utf8_end = encode_utf8(utf8_buf, mlen, ubuf, ulen))
 145                     == NULL)
 146                         goto done;
 147                 buffer_init(&bb, b->fd, utf8_buf,
 148                     (size_t)(utf8_end - utf8_buf));
 149
 150                 if ((rv = file_softmagic(ms, &bb, NULL, NULL,
 151                     TEXTTEST, text)) == 0)
 152                         rv = -1;
 153                 buffer_fini(&bb);
 154                 if ((ms->flags & (MAGIC_APPLE|MAGIC_EXTENSION))) {
 155                         rv = rv == -1 ? 0 : 1;
 156                         goto done;
 157                 }
 158         }
 159         if ((ms->flags & (MAGIC_APPLE|MAGIC_EXTENSION)))
 160                 return 0;
 161
 162         /* Now try to discover other details about the file. */
 163         for (i = 0; i < ulen; i++) {
 164                 if (ubuf[i] == '\n') {
 165                         if (seen_cr)
 166                                 n_crlf++;
 167                         else
 168                                 n_lf++;
 169                         last_line_end = i;
 170                 } else if (seen_cr)
 171                         n_cr++;
 172
 173                 seen_cr = (ubuf[i] == '\r');
 174                 if (seen_cr)
 175                         last_line_end = i;
 176
 177                 if (ubuf[i] == 0x85) { /* X3.64/ECMA-43 "next line" character */
 178                         n_nel++;
 179                         last_line_end = i;
 180                 }
 181
 182                 /* If this line is _longer_ than MAXLINELEN, remember it. */
 183                 if (i > last_line_end + MAXLINELEN)
 184                         has_long_lines = 1;
 185
 186                 if (ubuf[i] == '\033')
 187                         has_escapes = 1;
 188                 if (ubuf[i] == '\b')
 189                         has_backspace = 1;
 190         }
 191
 192         /* Beware, if the data has been truncated, the final CR could have
 193            been followed by a LF.  If we have ms->bytes_max bytes, it indicates
 194            that the data might have been truncated, probably even before
 195            this function was called. */
 196         if (seen_cr && nbytes < ms->bytes_max)
 197                 n_cr++;
 198
 199         if (strcmp(type, "binary") == 0) {
 200                 rv = 0;
 201                 goto done;
 202         }
 203         if (mime) {
 204                 if (!file_printedlen(ms) && (mime & MAGIC_MIME_TYPE) != 0) {
 205                         if (subtype_mime) {
 206                                 if (file_printf(ms, "%s", subtype_mime) == -1)
 207                                         goto done;
 208                         } else {
 209                                 if (file_printf(ms, "text/plain") == -1)
 210                                         goto done;
 211                         }
 212                 }
 213         } else {
 214                 if (file_printedlen(ms)) {
 215                         switch (file_replace(ms, " text$", ", ")) {
 216                         case 0:
 217                                 switch (file_replace(ms, " text executable$",
 218                                     ", ")) {
 219                                 case 0:
 220                                         if (file_printf(ms, ", ") == -1)
 221                                                 goto done;
 222                                         break;
 223                                 case -1:
 224                                         goto done;
 225                                 default:
 226                                         executable = 1;
 227                                         break;
 228                                 }
 229                                 break;
 230                         case -1:
 231                                 goto done;
 232                         default:
 233                                 break;
 234                         }
 235                 }
 236
 237                 if (file_printf(ms, "%s", code) == -1)
 238                         goto done;
 239
 240                 if (subtype) {
 241                         if (file_printf(ms, " %s", subtype) == -1)
 242                                 goto done;
 243                 }
 244
 245                 if (file_printf(ms, " %s", type) == -1)
 246                         goto done;
 247
 248                 if (executable)
 249                         if (file_printf(ms, " executable") == -1)
 250                                 goto done;
 251
 252                 if (has_long_lines)
 253                         if (file_printf(ms, ", with very long lines") == -1)
 254                                 goto done;
 255
 256                 /*
 257                  * Only report line terminators if we find one other than LF,
 258                  * or if we find none at all.
 259                  */
 260                 if ((n_crlf == 0 && n_cr == 0 && n_nel == 0 && n_lf == 0) ||
 261                     (n_crlf != 0 || n_cr != 0 || n_nel != 0)) {
 262                         if (file_printf(ms, ", with") == -1)
 263                                 goto done;
 264
 265                         if (n_crlf == 0 && n_cr == 0 && n_nel == 0 && n_lf == 0) {
 266                                 if (file_printf(ms, " no") == -1)
 267                                         goto done;
 268                         } else {
 269                                 if (n_crlf) {
 270                                         if (file_printf(ms, " CRLF") == -1)
 271                                                 goto done;
 272                                         if (n_cr || n_lf || n_nel)
 273                                                 if (file_printf(ms, ",") == -1)
 274                                                         goto done;
 275                                 }
 276                                 if (n_cr) {
 277                                         if (file_printf(ms, " CR") == -1)
 278                                                 goto done;
 279                                         if (n_lf || n_nel)
 280                                                 if (file_printf(ms, ",") == -1)
 281                                                         goto done;
 282                                 }
 283                                 if (n_lf) {
 284                                         if (file_printf(ms, " LF") == -1)
 285                                                 goto done;
 286                                         if (n_nel)
 287                                                 if (file_printf(ms, ",") == -1)
 288                                                         goto done;
 289                                 }
 290                                 if (n_nel)
 291                                         if (file_printf(ms, " NEL") == -1)
 292                                                 goto done;
 293                         }
 294
 295                         if (file_printf(ms, " line terminators") == -1)
 296                                 goto done;
 297                 }
 298
 299                 if (has_escapes)
 300                         if (file_printf(ms, ", with escape sequences") == -1)
 301                                 goto done;
 302                 if (has_backspace)
 303                         if (file_printf(ms, ", with overstriking") == -1)
 304                                 goto done;
 305         }
 306         rv = 1;
 307 done:
 308         free(utf8_buf);
 309
 310         return rv;
 311 }
 312
 313 /*
 314  * Encode Unicode string as UTF-8, returning pointer to character
 315  * after end of string, or NULL if an invalid character is found.
 316  */
 317 private unsigned char *
 318 encode_utf8(unsigned char *buf, size_t len, unichar *ubuf, size_t ulen)
 319 {
 320         size_t i;
 321         unsigned char *end = buf + len;
 322
 323         for (i = 0; i < ulen; i++) {
 324                 if (ubuf[i] <= 0x7f) {
 325                         if (end - buf < 1)
 326                                 return NULL;
 327                         *buf++ = (unsigned char)ubuf[i];
 328                 } else if (ubuf[i] <= 0x7ff) {
 329                         if (end - buf < 2)
 330                                 return NULL;
 331                         *buf++ = (unsigned char)((ubuf[i] >> 6) + 0xc0);
 332                         *buf++ = (unsigned char)((ubuf[i] & 0x3f) + 0x80);
 333                 } else if (ubuf[i] <= 0xffff) {
 334                         if (end - buf < 3)
 335                                 return NULL;
 336                         *buf++ = (unsigned char)((ubuf[i] >> 12) + 0xe0);
 337                         *buf++ = (unsigned char)(((ubuf[i] >> 6) & 0x3f) + 0x80);
 338                         *buf++ = (unsigned char)((ubuf[i] & 0x3f) + 0x80);
 339                 } else if (ubuf[i] <= 0x1fffff) {
 340                         if (end - buf < 4)
 341                                 return NULL;
 342                         *buf++ = (unsigned char)((ubuf[i] >> 18) + 0xf0);
 343                         *buf++ = (unsigned char)(((ubuf[i] >> 12) & 0x3f) + 0x80);
 344                         *buf++ = (unsigned char)(((ubuf[i] >>  6) & 0x3f) + 0x80);
 345                         *buf++ = (unsigned char)((ubuf[i] & 0x3f) + 0x80);
 346                 } else if (ubuf[i] <= 0x3ffffff) {
 347                         if (end - buf < 5)
 348                                 return NULL;
 349                         *buf++ = (unsigned char)((ubuf[i] >> 24) + 0xf8);
 350                         *buf++ = (unsigned char)(((ubuf[i] >> 18) & 0x3f) + 0x80);
 351                         *buf++ = (unsigned char)(((ubuf[i] >> 12) & 0x3f) + 0x80);
 352                         *buf++ = (unsigned char)(((ubuf[i] >>  6) & 0x3f) + 0x80);
 353                         *buf++ = (unsigned char)((ubuf[i] & 0x3f) + 0x80);
 354                 } else if (ubuf[i] <= 0x7fffffff) {
 355                         if (end - buf < 6)
 356                                 return NULL;
 357                         *buf++ = (unsigned char)((ubuf[i] >> 30) + 0xfc);
 358                         *buf++ = (unsigned char)(((ubuf[i] >> 24) & 0x3f) + 0x80);
 359                         *buf++ = (unsigned char)(((ubuf[i] >> 18) & 0x3f) + 0x80);
 360                         *buf++ = (unsigned char)(((ubuf[i] >> 12) & 0x3f) + 0x80);
 361                         *buf++ = (unsigned char)(((ubuf[i] >>  6) & 0x3f) + 0x80);
 362                         *buf++ = (unsigned char)((ubuf[i] & 0x3f) + 0x80);
 363                 } else /* Invalid character */
 364                         return NULL;
 365         }
 366
 367         return buf;
 368 }