external/bsd/file/dist/src/apprentice.c

   1 /*      $NetBSD: apprentice.c,v 1.1.1.1 2009/05/08 16:35:05 christos Exp $      */
   2
   3 /*
   4  * Copyright (c) Ian F. Darwin 1986-1995.
   5  * Software written by Ian F. Darwin and others;
   6  * maintained 1995-present by Christos Zoulas and others.
   7  *
   8  * Redistribution and use in source and binary forms, with or without
   9  * modification, are permitted provided that the following conditions
  10  * are met:
  11  * 1. Redistributions of source code must retain the above copyright
  12  *    notice immediately at the beginning of the file, without modification,
  13  *    this list of conditions, and the following disclaimer.
  14  * 2. Redistributions in binary form must reproduce the above copyright
  15  *    notice, this list of conditions and the following disclaimer in the
  16  *    documentation and/or other materials provided with the distribution.
  17  *
  18  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
  19  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  20  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  21  * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE FOR
  22  * ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  23  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
  24  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  25  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  26  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  27  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  28  * SUCH DAMAGE.
  29  */
  30 /*
  31  * apprentice - make one pass through /etc/magic, learning its secrets.
  32  */
  33
  34 #include "file.h"
  35
  36 #ifndef lint
  37 #if 0
  38 FILE_RCSID("@(#)$File: apprentice.c,v 1.151 2009/03/18 15:19:23 christos Exp $")
  39 #else
  40 __RCSID("$NetBSD: apprentice.c,v 1.1.1.1 2009/05/08 16:35:05 christos Exp $");
  41 #endif
  42 #endif  /* lint */
  43
  44 #include "magic.h"
  45 #include "patchlevel.h"
  46 #include <stdlib.h>
  47 #ifdef HAVE_UNISTD_H
  48 #include <unistd.h>
  49 #endif
  50 #include <string.h>
  51 #include <assert.h>
  52 #include <ctype.h>
  53 #include <fcntl.h>
  54 #ifdef QUICK
  55 #include <sys/mman.h>
  56 #endif
  57 #include <dirent.h>
  58
  59 #define EATAB {while (isascii((unsigned char) *l) && \
  60                       isspace((unsigned char) *l))  ++l;}
  61 #define LOWCASE(l) (isupper((unsigned char) (l)) ? \
  62                         tolower((unsigned char) (l)) : (l))
  63 /*
  64  * Work around a bug in headers on Digital Unix.
  65  * At least confirmed for: OSF1 V4.0 878
  66  */
  67 #if defined(__osf__) && defined(__DECC)
  68 #ifdef MAP_FAILED
  69 #undef MAP_FAILED
  70 #endif
  71 #endif
  72
  73 #ifndef MAP_FAILED
  74 #define MAP_FAILED (void *) -1
  75 #endif
  76
  77 #ifndef MAP_FILE
  78 #define MAP_FILE 0
  79 #endif
  80
  81 #ifndef MAXPATHLEN
  82 #define MAXPATHLEN      1024
  83 #endif
  84
  85 struct magic_entry {
  86         struct magic *mp;
  87         uint32_t cont_count;
  88         uint32_t max_count;
  89 };
  90
  91 int file_formats[FILE_NAMES_SIZE];
  92 const size_t file_nformats = FILE_NAMES_SIZE;
  93 const char *file_names[FILE_NAMES_SIZE];
  94 const size_t file_nnames = FILE_NAMES_SIZE;
  95
  96 private int getvalue(struct magic_set *ms, struct magic *, const char **, int);
  97 private int hextoint(int);
  98 private const char *getstr(struct magic_set *, struct magic *, const char *,
  99     int);
 100 private int parse(struct magic_set *, struct magic_entry **, uint32_t *,
 101     const char *, size_t, int);
 102 private void eatsize(const char **);
 103 private int apprentice_1(struct magic_set *, const char *, int, struct mlist *);
 104 private size_t apprentice_magic_strength(const struct magic *);
 105 private int apprentice_sort(const void *, const void *);
 106 private int apprentice_load(struct magic_set *, struct magic **, uint32_t *,
 107     const char *, int);
 108 private void byteswap(struct magic *, uint32_t);
 109 private void bs1(struct magic *);
 110 private uint16_t swap2(uint16_t);
 111 private uint32_t swap4(uint32_t);
 112 private uint64_t swap8(uint64_t);
 113 private char *mkdbname(struct magic_set *, const char *, int);
 114 private int apprentice_map(struct magic_set *, struct magic **, uint32_t *,
 115     const char *);
 116 private int apprentice_compile(struct magic_set *, struct magic **, uint32_t *,
 117     const char *);
 118 private int check_format_type(const char *, int);
 119 private int check_format(struct magic_set *, struct magic *);
 120 private int get_op(char);
 121 private int parse_mime(struct magic_set *, struct magic_entry *, const char *);
 122 private int parse_strength(struct magic_set *, struct magic_entry *, const char *);
 123 private int parse_apple(struct magic_set *, struct magic_entry *, const char *);
 124
 125
 126 private size_t maxmagic = 0;
 127 private size_t magicsize = sizeof(struct magic);
 128
 129 private const char usg_hdr[] = "cont\toffset\ttype\topcode\tmask\tvalue\tdesc";
 130
 131 private struct {
 132         const char *name;
 133         size_t len;
 134         int (*fun)(struct magic_set *, struct magic_entry *, const char *);
 135 } bang[] = {
 136 #define DECLARE_FIELD(name) { # name, sizeof(# name) - 1, parse_ ## name }
 137         DECLARE_FIELD(mime),
 138         DECLARE_FIELD(apple),
 139         DECLARE_FIELD(strength),
 140 #undef  DECLARE_FIELD
 141         { NULL, 0, NULL }
 142 };
 143
 144 #ifdef COMPILE_ONLY
 145
 146 int main(int, char *[]);
 147
 148 int
 149 main(int argc, char *argv[])
 150 {
 151         int ret;
 152         struct magic_set *ms;
 153         char *progname;
 154
 155         if ((progname = strrchr(argv[0], '/')) != NULL)
 156                 progname++;
 157         else
 158                 progname = argv[0];
 159
 160         if (argc != 2) {
 161                 (void)fprintf(stderr, "Usage: %s file\n", progname);
 162                 return 1;
 163         }
 164
 165         if ((ms = magic_open(MAGIC_CHECK)) == NULL) {
 166                 (void)fprintf(stderr, "%s: %s\n", progname, strerror(errno));
 167                 return 1;
 168         }
 169         ret = magic_compile(ms, argv[1]) == -1 ? 1 : 0;
 170         if (ret == 1)
 171                 (void)fprintf(stderr, "%s: %s\n", progname, magic_error(ms));
 172         magic_close(ms);
 173         return ret;
 174 }
 175 #endif /* COMPILE_ONLY */
 176
 177 static const struct type_tbl_s {
 178         const char name[16];
 179         const size_t len;
 180         const int type;
 181         const int format;
 182 } type_tbl[] = {
 183 # define XX(s)          s, (sizeof(s) - 1)
 184 # define XX_NULL        "", 0
 185         { XX("byte"),           FILE_BYTE,              FILE_FMT_NUM },
 186         { XX("short"),          FILE_SHORT,             FILE_FMT_NUM },
 187         { XX("default"),        FILE_DEFAULT,           FILE_FMT_STR },
 188         { XX("long"),           FILE_LONG,              FILE_FMT_NUM },
 189         { XX("string"),         FILE_STRING,            FILE_FMT_STR },
 190         { XX("date"),           FILE_DATE,              FILE_FMT_STR },
 191         { XX("beshort"),        FILE_BESHORT,           FILE_FMT_NUM },
 192         { XX("belong"),         FILE_BELONG,            FILE_FMT_NUM },
 193         { XX("bedate"),         FILE_BEDATE,            FILE_FMT_STR },
 194         { XX("leshort"),        FILE_LESHORT,           FILE_FMT_NUM },
 195         { XX("lelong"),         FILE_LELONG,            FILE_FMT_NUM },
 196         { XX("ledate"),         FILE_LEDATE,            FILE_FMT_STR },
 197         { XX("pstring"),        FILE_PSTRING,           FILE_FMT_STR },
 198         { XX("ldate"),          FILE_LDATE,             FILE_FMT_STR },
 199         { XX("beldate"),        FILE_BELDATE,           FILE_FMT_STR },
 200         { XX("leldate"),        FILE_LELDATE,           FILE_FMT_STR },
 201         { XX("regex"),          FILE_REGEX,             FILE_FMT_STR },
 202         { XX("bestring16"),     FILE_BESTRING16,        FILE_FMT_STR },
 203         { XX("lestring16"),     FILE_LESTRING16,        FILE_FMT_STR },
 204         { XX("search"),         FILE_SEARCH,            FILE_FMT_STR },
 205         { XX("medate"),         FILE_MEDATE,            FILE_FMT_STR },
 206         { XX("meldate"),        FILE_MELDATE,           FILE_FMT_STR },
 207         { XX("melong"),         FILE_MELONG,            FILE_FMT_NUM },
 208         { XX("quad"),           FILE_QUAD,              FILE_FMT_QUAD },
 209         { XX("lequad"),         FILE_LEQUAD,            FILE_FMT_QUAD },
 210         { XX("bequad"),         FILE_BEQUAD,            FILE_FMT_QUAD },
 211         { XX("qdate"),          FILE_QDATE,             FILE_FMT_STR },
 212         { XX("leqdate"),        FILE_LEQDATE,           FILE_FMT_STR },
 213         { XX("beqdate"),        FILE_BEQDATE,           FILE_FMT_STR },
 214         { XX("qldate"),         FILE_QLDATE,            FILE_FMT_STR },
 215         { XX("leqldate"),       FILE_LEQLDATE,          FILE_FMT_STR },
 216         { XX("beqldate"),       FILE_BEQLDATE,          FILE_FMT_STR },
 217         { XX("float"),          FILE_FLOAT,             FILE_FMT_FLOAT },
 218         { XX("befloat"),        FILE_BEFLOAT,           FILE_FMT_FLOAT },
 219         { XX("lefloat"),        FILE_LEFLOAT,           FILE_FMT_FLOAT },
 220         { XX("double"),         FILE_DOUBLE,            FILE_FMT_DOUBLE },
 221         { XX("bedouble"),       FILE_BEDOUBLE,          FILE_FMT_DOUBLE },
 222         { XX("ledouble"),       FILE_LEDOUBLE,          FILE_FMT_DOUBLE },
 223         { XX("leid3"),          FILE_LEID3,             FILE_FMT_NUM },
 224         { XX("beid3"),          FILE_BEID3,             FILE_FMT_NUM },
 225         { XX("indirect"),       FILE_INDIRECT,          FILE_FMT_NONE },
 226         { XX_NULL,              FILE_INVALID,           FILE_FMT_NONE },
 227 # undef XX
 228 # undef XX_NULL
 229 };
 230
 231 private int
 232 get_type(const char *l, const char **t)
 233 {
 234         const struct type_tbl_s *p;
 235
 236         for (p = type_tbl; p->len; p++) {
 237                 if (strncmp(l, p->name, p->len) == 0) {
 238                         if (t)
 239                                 *t = l + p->len;
 240                         break;
 241                 }
 242         }
 243         return p->type;
 244 }
 245
 246 private void
 247 init_file_tables(void)
 248 {
 249         static int done = 0;
 250         const struct type_tbl_s *p;
 251
 252         if (done)
 253                 return;
 254         done++;
 255
 256         for (p = type_tbl; p->len; p++) {
 257                 assert(p->type < FILE_NAMES_SIZE);
 258                 file_names[p->type] = p->name;
 259                 file_formats[p->type] = p->format;
 260         }
 261 }
 262
 263 /*
 264  * Handle one file or directory.
 265  */
 266 private int
 267 apprentice_1(struct magic_set *ms, const char *fn, int action,
 268     struct mlist *mlist)
 269 {
 270         struct magic *magic = NULL;
 271         uint32_t nmagic = 0;
 272         struct mlist *ml;
 273         int rv = -1;
 274         int mapped;
 275
 276         if (magicsize != FILE_MAGICSIZE) {
 277                 file_error(ms, 0, "magic element size %lu != %lu",
 278                     (unsigned long)sizeof(*magic),
 279                     (unsigned long)FILE_MAGICSIZE);
 280                 return -1;
 281         }
 282
 283         if (action == FILE_COMPILE) {
 284                 rv = apprentice_load(ms, &magic, &nmagic, fn, action);
 285                 if (rv != 0)
 286                         return -1;
 287                 rv = apprentice_compile(ms, &magic, &nmagic, fn);
 288                 free(magic);
 289                 return rv;
 290         }
 291
 292 #ifndef COMPILE_ONLY
 293         if ((rv = apprentice_map(ms, &magic, &nmagic, fn)) == -1) {
 294                 if (ms->flags & MAGIC_CHECK)
 295                         file_magwarn(ms, "using regular magic file `%s'", fn);
 296                 rv = apprentice_load(ms, &magic, &nmagic, fn, action);
 297                 if (rv != 0)
 298                         return -1;
 299         }
 300
 301         mapped = rv;
 302
 303         if (magic == NULL) {
 304                 file_delmagic(magic, mapped, nmagic);
 305                 return -1;
 306         }
 307
 308         if ((ml = CAST(struct mlist *, malloc(sizeof(*ml)))) == NULL) {
 309                 file_delmagic(magic, mapped, nmagic);
 310                 file_oomem(ms, sizeof(*ml));
 311                 return -1;
 312         }
 313
 314         ml->magic = magic;
 315         ml->nmagic = nmagic;
 316         ml->mapped = mapped;
 317
 318         mlist->prev->next = ml;
 319         ml->prev = mlist->prev;
 320         ml->next = mlist;
 321         mlist->prev = ml;
 322
 323         return 0;
 324 #endif /* COMPILE_ONLY */
 325 }
 326
 327 protected void
 328 file_delmagic(struct magic *p, int type, size_t entries)
 329 {
 330         if (p == NULL)
 331                 return;
 332         switch (type) {
 333         case 2:
 334 #ifdef QUICK
 335                 p--;
 336                 (void)munmap((void *)p, sizeof(*p) * (entries + 1));
 337                 break;
 338 #else
 339                 (void)&entries;
 340                 abort();
 341                 /*NOTREACHED*/
 342 #endif
 343         case 1:
 344                 p--;
 345                 /*FALLTHROUGH*/
 346         case 0:
 347                 free(p);
 348                 break;
 349         default:
 350                 abort();
 351         }
 352 }
 353
 354 /* const char *fn: list of magic files and directories */
 355 protected struct mlist *
 356 file_apprentice(struct magic_set *ms, const char *fn, int action)
 357 {
 358         char *p, *mfn;
 359         int file_err, errs = -1;
 360         struct mlist *mlist;
 361
 362         init_file_tables();
 363
 364         if (fn == NULL)
 365                 fn = getenv("MAGIC");
 366         if (fn == NULL)
 367                 fn = MAGIC;
 368
 369         if ((mfn = strdup(fn)) == NULL) {
 370                 file_oomem(ms, strlen(fn));
 371                 return NULL;
 372         }
 373         fn = mfn;
 374
 375         if ((mlist = CAST(struct mlist *, malloc(sizeof(*mlist)))) == NULL) {
 376                 free(mfn);
 377                 file_oomem(ms, sizeof(*mlist));
 378                 return NULL;
 379         }
 380         mlist->next = mlist->prev = mlist;
 381
 382         while (fn) {
 383                 p = strchr(fn, PATHSEP);
 384                 if (p)
 385                         *p++ = '\0';
 386                 if (*fn == '\0')
 387                         break;
 388                 file_err = apprentice_1(ms, fn, action, mlist);
 389                 errs = MAX(errs, file_err);
 390                 fn = p;
 391         }
 392         if (errs == -1) {
 393                 free(mfn);
 394                 free(mlist);
 395                 mlist = NULL;
 396                 file_error(ms, 0, "could not find any magic files!");
 397                 return NULL;
 398         }
 399         free(mfn);
 400         return mlist;
 401 }
 402
 403 /*
 404  * Get weight of this magic entry, for sorting purposes.
 405  */
 406 private size_t
 407 apprentice_magic_strength(const struct magic *m)
 408 {
 409 #define MULT 10
 410         size_t val = 2 * MULT;  /* baseline strength */
 411
 412         switch (m->type) {
 413         case FILE_DEFAULT:      /* make sure this sorts last */
 414                 if (m->factor_op != FILE_FACTOR_OP_NONE)
 415                         abort();
 416                 return 0;
 417
 418         case FILE_BYTE:
 419                 val += 1 * MULT;
 420                 break;
 421
 422         case FILE_SHORT:
 423         case FILE_LESHORT:
 424         case FILE_BESHORT:
 425                 val += 2 * MULT;
 426                 break;
 427
 428         case FILE_LONG:
 429         case FILE_LELONG:
 430         case FILE_BELONG:
 431         case FILE_MELONG:
 432                 val += 4 * MULT;
 433                 break;
 434
 435         case FILE_PSTRING:
 436         case FILE_STRING:
 437                 val += m->vallen * MULT;
 438                 break;
 439
 440         case FILE_BESTRING16:
 441         case FILE_LESTRING16:
 442                 val += m->vallen * MULT / 2;
 443                 break;
 444
 445         case FILE_SEARCH:
 446         case FILE_REGEX:
 447                 val += m->vallen * MAX(MULT / m->vallen, 1);
 448                 break;
 449
 450         case FILE_DATE:
 451         case FILE_LEDATE:
 452         case FILE_BEDATE:
 453         case FILE_MEDATE:
 454         case FILE_LDATE:
 455         case FILE_LELDATE:
 456         case FILE_BELDATE:
 457         case FILE_MELDATE:
 458         case FILE_FLOAT:
 459         case FILE_BEFLOAT:
 460         case FILE_LEFLOAT:
 461                 val += 4 * MULT;
 462                 break;
 463
 464         case FILE_QUAD:
 465         case FILE_BEQUAD:
 466         case FILE_LEQUAD:
 467         case FILE_QDATE:
 468         case FILE_LEQDATE:
 469         case FILE_BEQDATE:
 470         case FILE_QLDATE:
 471         case FILE_LEQLDATE:
 472         case FILE_BEQLDATE:
 473         case FILE_DOUBLE:
 474         case FILE_BEDOUBLE:
 475         case FILE_LEDOUBLE:
 476                 val += 8 * MULT;
 477                 break;
 478
 479         default:
 480                 val = 0;
 481                 (void)fprintf(stderr, "Bad type %d\n", m->type);
 482                 abort();
 483         }
 484
 485         switch (m->reln) {
 486         case 'x':       /* matches anything penalize */
 487         case '!':       /* matches almost anything penalize */
 488                 val = 0;
 489                 break;
 490
 491         case '=':       /* Exact match, prefer */
 492                 val += MULT;
 493                 break;
 494
 495         case '>':
 496         case '<':       /* comparison match reduce strength */
 497                 val -= 2 * MULT;
 498                 break;
 499
 500         case '^':
 501         case '&':       /* masking bits, we could count them too */
 502                 val -= MULT;
 503                 break;
 504
 505         default:
 506                 (void)fprintf(stderr, "Bad relation %c\n", m->reln);
 507                 abort();
 508         }
 509
 510         if (val == 0)   /* ensure we only return 0 for FILE_DEFAULT */
 511                 val = 1;
 512
 513         switch (m->factor_op) {
 514         case FILE_FACTOR_OP_NONE:
 515                 break;
 516         case FILE_FACTOR_OP_PLUS:
 517                 val += m->factor;
 518                 break;
 519         case FILE_FACTOR_OP_MINUS:
 520                 val -= m->factor;
 521                 break;
 522         case FILE_FACTOR_OP_TIMES:
 523                 val *= m->factor;
 524                 break;
 525         case FILE_FACTOR_OP_DIV:
 526                 val /= m->factor;
 527                 break;
 528         default:
 529                 abort();
 530         }
 531
 532         /*
 533          * Magic entries with no description get a bonus because they depend
 534          * on subsequent magic entries to print something.
 535          */
 536         if (m->desc[0] == '\0')
 537                 val++;
 538         return val;
 539 }
 540
 541 /*
 542  * Sort callback for sorting entries by "strength" (basically length)
 543  */
 544 private int
 545 apprentice_sort(const void *a, const void *b)
 546 {
 547         const struct magic_entry *ma = CAST(const struct magic_entry *, a);
 548         const struct magic_entry *mb = CAST(const struct magic_entry *, b);
 549         size_t sa = apprentice_magic_strength(ma->mp);
 550         size_t sb = apprentice_magic_strength(mb->mp);
 551         if (sa == sb)
 552                 return 0;
 553         else if (sa > sb)
 554                 return -1;
 555         else
 556                 return 1;
 557 }
 558
 559 private void
 560 set_test_type(struct magic *mstart, struct magic *m)
 561 {
 562         switch (m->type) {
 563         case FILE_BYTE:
 564         case FILE_SHORT:
 565         case FILE_LONG:
 566         case FILE_DATE:
 567         case FILE_BESHORT:
 568         case FILE_BELONG:
 569         case FILE_BEDATE:
 570         case FILE_LESHORT:
 571         case FILE_LELONG:
 572         case FILE_LEDATE:
 573         case FILE_LDATE:
 574         case FILE_BELDATE:
 575         case FILE_LELDATE:
 576         case FILE_MEDATE:
 577         case FILE_MELDATE:
 578         case FILE_MELONG:
 579         case FILE_QUAD:
 580         case FILE_LEQUAD:
 581         case FILE_BEQUAD:
 582         case FILE_QDATE:
 583         case FILE_LEQDATE:
 584         case FILE_BEQDATE:
 585         case FILE_QLDATE:
 586         case FILE_LEQLDATE:
 587         case FILE_BEQLDATE:
 588         case FILE_FLOAT:
 589         case FILE_BEFLOAT:
 590         case FILE_LEFLOAT:
 591         case FILE_DOUBLE:
 592         case FILE_BEDOUBLE:
 593         case FILE_LEDOUBLE:
 594         case FILE_STRING:
 595         case FILE_PSTRING:
 596         case FILE_BESTRING16:
 597         case FILE_LESTRING16:
 598                 /* binary test, set flag */
 599                 mstart->flag |= BINTEST;
 600                 break;
 601         case FILE_REGEX:
 602         case FILE_SEARCH:
 603                 /* binary test if pattern is not text */
 604                 if (file_looks_utf8(m->value.us, (size_t)m->vallen, NULL,
 605                     NULL) <= 0)
 606                         mstart->flag |= BINTEST;
 607                 break;
 608         case FILE_DEFAULT:
 609                 /* can't deduce anything; we shouldn't see this at the
 610                    top level anyway */
 611                 break;
 612         case FILE_INVALID:
 613         default:
 614                 /* invalid search type, but no need to complain here */
 615                 break;
 616         }
 617 }
 618
 619 /*
 620  * Load and parse one file.
 621  */
 622 private void
 623 load_1(struct magic_set *ms, int action, const char *fn, int *errs,
 624    struct magic_entry **marray, uint32_t *marraycount)
 625 {
 626         char line[BUFSIZ];
 627         size_t lineno = 0;
 628         FILE *f = fopen(ms->file = fn, "r");
 629         if (f == NULL) {
 630                 if (errno != ENOENT)
 631                         file_error(ms, errno, "cannot read magic file `%s'",
 632                                    fn);
 633                 (*errs)++;
 634         } else {
 635                 /* read and parse this file */
 636                 for (ms->line = 1;
 637                     fgets(line, CAST(int, sizeof(line)), f) != NULL;
 638                     ms->line++) {
 639                         size_t len;
 640                         len = strlen(line);
 641                         if (len == 0) /* null line, garbage, etc */
 642                                 continue;
 643                         if (line[len - 1] == '\n') {
 644                                 lineno++;
 645                                 line[len - 1] = '\0'; /* delete newline */
 646                         }
 647                         if (line[0] == '\0')    /* empty, do not parse */
 648                                 continue;
 649                         if (line[0] == '#')     /* comment, do not parse */
 650                                 continue;
 651                         if (line[0] == '!' && line[1] == ':') {
 652                                 size_t i;
 653
 654                                 for (i = 0; bang[i].name != NULL; i++) {
 655                                         if (len - 2 > bang[i].len &&
 656                                             memcmp(bang[i].name, line + 2,
 657                                             bang[i].len) == 0)
 658                                                 break;
 659                                 }
 660                                 if (bang[i].name == NULL) {
 661                                         file_error(ms, 0,
 662                                             "Unknown !: entry `%s'", line);
 663                                         (*errs)++;
 664                                         continue;
 665                                 }
 666                                 if (*marraycount == 0) {
 667                                         file_error(ms, 0,
 668                                             "No current entry for :!%s type",
 669                                                 bang[i].name);
 670                                         (*errs)++;
 671                                         continue;
 672                                 }
 673                                 if ((*bang[i].fun)(ms,
 674                                     &(*marray)[*marraycount - 1],
 675                                     line + bang[i].len + 2) != 0) {
 676                                         (*errs)++;
 677                                         continue;
 678                                 }
 679                                 continue;
 680                         }
 681                         if (parse(ms, marray, marraycount, line, lineno,
 682                             action) != 0)
 683                                 (*errs)++;
 684                 }
 685
 686                 (void)fclose(f);
 687         }
 688 }
 689
 690 /*
 691  * parse a file or directory of files
 692  * const char *fn: name of magic file or directory
 693  */
 694 private int
 695 apprentice_load(struct magic_set *ms, struct magic **magicp, uint32_t *nmagicp,
 696     const char *fn, int action)
 697 {
 698         int errs = 0;
 699         struct magic_entry *marray;
 700         uint32_t marraycount, i, mentrycount = 0, starttest;
 701         size_t slen;
 702         char subfn[MAXPATHLEN];
 703         struct stat st;
 704         DIR *dir;
 705         struct dirent *d;
 706
 707         ms->flags |= MAGIC_CHECK;       /* Enable checks for parsed files */
 708
 709         maxmagic = MAXMAGIS;
 710         if ((marray = CAST(struct magic_entry *, calloc(maxmagic,
 711             sizeof(*marray)))) == NULL) {
 712                 file_oomem(ms, maxmagic * sizeof(*marray));
 713                 return -1;
 714         }
 715         marraycount = 0;
 716
 717         /* print silly verbose header for USG compat. */
 718         if (action == FILE_CHECK)
 719                 (void)fprintf(stderr, "%s\n", usg_hdr);
 720
 721         /* load directory or file */
 722         /* FIXME: Read file names and sort them to prevent
 723            non-determinism. See Debian bug #488562. */
 724         if (stat(fn, &st) == 0 && S_ISDIR(st.st_mode)) {
 725                 dir = opendir(fn);
 726                 if (dir) {
 727                         while ((d = readdir(dir)) != NULL) {
 728                                 snprintf(subfn, sizeof(subfn), "%s/%s",
 729                                     fn, d->d_name);
 730                                 if (stat(subfn, &st) == 0 &&
 731                                     S_ISREG(st.st_mode)) {
 732                                         load_1(ms, action, subfn, &errs,
 733                                             &marray, &marraycount);
 734                                 }
 735                         }
 736                         closedir(dir);
 737                 } else
 738                         errs++;
 739         } else
 740                 load_1(ms, action, fn, &errs, &marray, &marraycount);
 741         if (errs)
 742                 goto out;
 743
 744         /* Set types of tests */
 745         for (i = 0; i < marraycount; ) {
 746                 if (marray[i].mp->cont_level != 0) {
 747                         i++;
 748                         continue;
 749                 }
 750
 751                 starttest = i;
 752                 do {
 753                         static const char text[] = "text";
 754                         static const char binary[] = "binary";
 755                         static const size_t len = sizeof(text);
 756                         set_test_type(marray[starttest].mp, marray[i].mp);
 757                         if ((ms->flags & MAGIC_DEBUG) == 0)
 758                                 continue;
 759                         (void)fprintf(stderr, "%s%s%s: %s\n",
 760                             marray[i].mp->mimetype,
 761                             marray[i].mp->mimetype[0] == '\0' ? "" : "; ",
 762                             marray[i].mp->desc[0] ? marray[i].mp->desc :
 763                             "(no description)",
 764                             marray[i].mp->flag & BINTEST ? binary : text);
 765                         if (marray[i].mp->flag & BINTEST) {
 766                                 char *p = strstr(marray[i].mp->desc, text);
 767                                 if (p && (p == marray[i].mp->desc ||
 768                                     isspace((unsigned char)p[-1])) &&
 769                                     (p + len - marray[i].mp->desc ==
 770                                     MAXstring || (p[len] == '\0' ||
 771                                     isspace((unsigned char)p[len]))))
 772                                         (void)fprintf(stderr, "*** Possible "
 773                                             "binary test for text type\n");
 774                         }
 775                 } while (++i < marraycount && marray[i].mp->cont_level != 0);
 776         }
 777
 778         qsort(marray, marraycount, sizeof(*marray), apprentice_sort);
 779
 780         /*
 781          * Make sure that any level 0 "default" line is last (if one exists).
 782          */
 783         for (i = 0; i < marraycount; i++) {
 784                 if (marray[i].mp->cont_level == 0 &&
 785                     marray[i].mp->type == FILE_DEFAULT) {
 786                         while (++i < marraycount)
 787                                 if (marray[i].mp->cont_level == 0)
 788                                         break;
 789                         if (i != marraycount) {
 790                                 ms->line = marray[i].mp->lineno; /* XXX - Ugh! */
 791                                 file_magwarn(ms,
 792                                     "level 0 \"default\" did not sort last");
 793                         }
 794                         break;
 795                 }
 796         }
 797
 798         for (i = 0; i < marraycount; i++)
 799                 mentrycount += marray[i].cont_count;
 800
 801         slen = sizeof(**magicp) * mentrycount;
 802         if ((*magicp = CAST(struct magic *, malloc(slen))) == NULL) {
 803                 file_oomem(ms, slen);
 804                 errs++;
 805                 goto out;
 806         }
 807
 808         mentrycount = 0;
 809         for (i = 0; i < marraycount; i++) {
 810                 (void)memcpy(*magicp + mentrycount, marray[i].mp,
 811                     marray[i].cont_count * sizeof(**magicp));
 812                 mentrycount += marray[i].cont_count;
 813         }
 814 out:
 815         for (i = 0; i < marraycount; i++)
 816                 free(marray[i].mp);
 817         free(marray);
 818         if (errs) {
 819                 *magicp = NULL;
 820                 *nmagicp = 0;
 821                 return errs;
 822         } else {
 823                 *nmagicp = mentrycount;
 824                 return 0;
 825         }
 826
 827 }
 828
 829 /*
 830  * extend the sign bit if the comparison is to be signed
 831  */
 832 protected uint64_t
 833 file_signextend(struct magic_set *ms, struct magic *m, uint64_t v)
 834 {
 835         if (!(m->flag & UNSIGNED)) {
 836                 switch(m->type) {
 837                 /*
 838                  * Do not remove the casts below.  They are
 839                  * vital.  When later compared with the data,
 840                  * the sign extension must have happened.
 841                  */
 842                 case FILE_BYTE:
 843                         v = (char) v;
 844                         break;
 845                 case FILE_SHORT:
 846                 case FILE_BESHORT:
 847                 case FILE_LESHORT:
 848                         v = (short) v;
 849                         break;
 850                 case FILE_DATE:
 851                 case FILE_BEDATE:
 852                 case FILE_LEDATE:
 853                 case FILE_MEDATE:
 854                 case FILE_LDATE:
 855                 case FILE_BELDATE:
 856                 case FILE_LELDATE:
 857                 case FILE_MELDATE:
 858                 case FILE_LONG:
 859                 case FILE_BELONG:
 860                 case FILE_LELONG:
 861                 case FILE_MELONG:
 862                 case FILE_FLOAT:
 863                 case FILE_BEFLOAT:
 864                 case FILE_LEFLOAT:
 865                         v = (int32_t) v;
 866                         break;
 867                 case FILE_QUAD:
 868                 case FILE_BEQUAD:
 869                 case FILE_LEQUAD:
 870                 case FILE_QDATE:
 871                 case FILE_QLDATE:
 872                 case FILE_BEQDATE:
 873                 case FILE_BEQLDATE:
 874                 case FILE_LEQDATE:
 875                 case FILE_LEQLDATE:
 876                 case FILE_DOUBLE:
 877                 case FILE_BEDOUBLE:
 878                 case FILE_LEDOUBLE:
 879                         v = (int64_t) v;
 880                         break;
 881                 case FILE_STRING:
 882                 case FILE_PSTRING:
 883                 case FILE_BESTRING16:
 884                 case FILE_LESTRING16:
 885                 case FILE_REGEX:
 886                 case FILE_SEARCH:
 887                 case FILE_DEFAULT:
 888                 case FILE_INDIRECT:
 889                         break;
 890                 default:
 891                         if (ms->flags & MAGIC_CHECK)
 892                             file_magwarn(ms, "cannot happen: m->type=%d\n",
 893                                     m->type);
 894                         return ~0U;
 895                 }
 896         }
 897         return v;
 898 }
 899
 900 private int
 901 string_modifier_check(struct magic_set *ms, struct magic *m)
 902 {
 903         if ((ms->flags & MAGIC_CHECK) == 0)
 904                 return 0;
 905
 906         switch (m->type) {
 907         case FILE_BESTRING16:
 908         case FILE_LESTRING16:
 909                 if (m->str_flags != 0) {
 910                         file_magwarn(ms,
 911                             "no modifiers allowed for 16-bit strings\n");
 912                         return -1;
 913                 }
 914                 break;
 915         case FILE_STRING:
 916         case FILE_PSTRING:
 917                 if ((m->str_flags & REGEX_OFFSET_START) != 0) {
 918                         file_magwarn(ms,
 919                             "'/%c' only allowed on regex and search\n",
 920                             CHAR_REGEX_OFFSET_START);
 921                         return -1;
 922                 }
 923                 break;
 924         case FILE_SEARCH:
 925                 if (m->str_range == 0) {
 926                         file_magwarn(ms,
 927                             "missing range; defaulting to %d\n",
 928                             STRING_DEFAULT_RANGE);
 929                         m->str_range = STRING_DEFAULT_RANGE;
 930                         return -1;
 931                 }
 932                 break;
 933         case FILE_REGEX:
 934                 if ((m->str_flags & STRING_COMPACT_BLANK) != 0) {
 935                         file_magwarn(ms, "'/%c' not allowed on regex\n",
 936                             CHAR_COMPACT_BLANK);
 937                         return -1;
 938                 }
 939                 if ((m->str_flags & STRING_COMPACT_OPTIONAL_BLANK) != 0) {
 940                         file_magwarn(ms, "'/%c' not allowed on regex\n",
 941                             CHAR_COMPACT_OPTIONAL_BLANK);
 942                         return -1;
 943                 }
 944                 break;
 945         default:
 946                 file_magwarn(ms, "coding error: m->type=%d\n",
 947                     m->type);
 948                 return -1;
 949         }
 950         return 0;
 951 }
 952
 953 private int
 954 get_op(char c)
 955 {
 956         switch (c) {
 957         case '&':
 958                 return FILE_OPAND;
 959         case '|':
 960                 return FILE_OPOR;
 961         case '^':
 962                 return FILE_OPXOR;
 963         case '+':
 964                 return FILE_OPADD;
 965         case '-':
 966                 return FILE_OPMINUS;
 967         case '*':
 968                 return FILE_OPMULTIPLY;
 969         case '/':
 970                 return FILE_OPDIVIDE;
 971         case '%':
 972                 return FILE_OPMODULO;
 973         default:
 974                 return -1;
 975         }
 976 }
 977
 978 #ifdef ENABLE_CONDITIONALS
 979 private int
 980 get_cond(const char *l, const char **t)
 981 {
 982         static const struct cond_tbl_s {
 983                 char name[8];
 984                 size_t len;
 985                 int cond;
 986         } cond_tbl[] = {
 987                 { "if",         2,      COND_IF },
 988                 { "elif",       4,      COND_ELIF },
 989                 { "else",       4,      COND_ELSE },
 990                 { "",           0,      COND_NONE },
 991         };
 992         const struct cond_tbl_s *p;
 993
 994         for (p = cond_tbl; p->len; p++) {
 995                 if (strncmp(l, p->name, p->len) == 0 &&
 996                     isspace((unsigned char)l[p->len])) {
 997                         if (t)
 998                                 *t = l + p->len;
 999                         break;
1000                 }
1001         }
1002         return p->cond;
1003 }
1004
1005 private int
1006 check_cond(struct magic_set *ms, int cond, uint32_t cont_level)
1007 {
1008         int last_cond;
1009         last_cond = ms->c.li[cont_level].last_cond;
1010
1011         switch (cond) {
1012         case COND_IF:
1013                 if (last_cond != COND_NONE && last_cond != COND_ELIF) {
1014                         if (ms->flags & MAGIC_CHECK)
1015                                 file_magwarn(ms, "syntax error: `if'");
1016                         return -1;
1017                 }
1018                 last_cond = COND_IF;
1019                 break;
1020
1021         case COND_ELIF:
1022                 if (last_cond != COND_IF && last_cond != COND_ELIF) {
1023                         if (ms->flags & MAGIC_CHECK)
1024                                 file_magwarn(ms, "syntax error: `elif'");
1025                         return -1;
1026                 }
1027                 last_cond = COND_ELIF;
1028                 break;
1029
1030         case COND_ELSE:
1031                 if (last_cond != COND_IF && last_cond != COND_ELIF) {
1032                         if (ms->flags & MAGIC_CHECK)
1033                                 file_magwarn(ms, "syntax error: `else'");
1034                         return -1;
1035                 }
1036                 last_cond = COND_NONE;
1037                 break;
1038
1039         case COND_NONE:
1040                 last_cond = COND_NONE;
1041                 break;
1042         }
1043
1044         ms->c.li[cont_level].last_cond = last_cond;
1045         return 0;
1046 }
1047 #endif /* ENABLE_CONDITIONALS */
1048
1049 /*
1050  * parse one line from magic file, put into magic[index++] if valid
1051  */
1052 private int
1053 parse(struct magic_set *ms, struct magic_entry **mentryp, uint32_t *nmentryp,
1054     const char *line, size_t lineno, int action)
1055 {
1056 #ifdef ENABLE_CONDITIONALS
1057         static uint32_t last_cont_level = 0;
1058 #endif
1059         size_t i;
1060         struct magic_entry *me;
1061         struct magic *m;
1062         const char *l = line;
1063         char *t;
1064         int op;
1065         uint32_t cont_level;
1066
1067         cont_level = 0;
1068
1069         while (*l == '>') {
1070                 ++l;            /* step over */
1071                 cont_level++;
1072         }
1073 #ifdef ENABLE_CONDITIONALS
1074         if (cont_level == 0 || cont_level > last_cont_level)
1075                 if (file_check_mem(ms, cont_level) == -1)
1076                         return -1;
1077         last_cont_level = cont_level;
1078 #endif
1079
1080 #define ALLOC_CHUNK     (size_t)10
1081 #define ALLOC_INCR      (size_t)200
1082
1083         if (cont_level != 0) {
1084                 if (*nmentryp == 0) {
1085                         file_error(ms, 0, "No current entry for continuation");
1086                         return -1;
1087                 }
1088                 me = &(*mentryp)[*nmentryp - 1];
1089                 if (me->cont_count == me->max_count) {
1090                         struct magic *nm;
1091                         size_t cnt = me->max_count + ALLOC_CHUNK;
1092                         if ((nm = CAST(struct magic *, realloc(me->mp,
1093                             sizeof(*nm) * cnt))) == NULL) {
1094                                 file_oomem(ms, sizeof(*nm) * cnt);
1095                                 return -1;
1096                         }
1097                         me->mp = m = nm;
1098                         me->max_count = CAST(uint32_t, cnt);
1099                 }
1100                 m = &me->mp[me->cont_count++];
1101                 (void)memset(m, 0, sizeof(*m));
1102                 m->cont_level = cont_level;
1103         } else {
1104                 if (*nmentryp == maxmagic) {
1105                         struct magic_entry *mp;
1106
1107                         maxmagic += ALLOC_INCR;
1108                         if ((mp = CAST(struct magic_entry *,
1109                             realloc(*mentryp, sizeof(*mp) * maxmagic))) ==
1110                             NULL) {
1111                                 file_oomem(ms, sizeof(*mp) * maxmagic);
1112                                 return -1;
1113                         }
1114                         (void)memset(&mp[*nmentryp], 0, sizeof(*mp) *
1115                             ALLOC_INCR);
1116                         *mentryp = mp;
1117                 }
1118                 me = &(*mentryp)[*nmentryp];
1119                 if (me->mp == NULL) {
1120                         size_t len = sizeof(*m) * ALLOC_CHUNK;
1121                         if ((m = CAST(struct magic *, malloc(len))) == NULL) {
1122                                 file_oomem(ms, len);
1123                                 return -1;
1124                         }
1125                         me->mp = m;
1126                         me->max_count = ALLOC_CHUNK;
1127                 } else
1128                         m = me->mp;
1129                 (void)memset(m, 0, sizeof(*m));
1130                 m->factor_op = FILE_FACTOR_OP_NONE;
1131                 m->cont_level = 0;
1132                 me->cont_count = 1;
1133         }
1134         m->lineno = CAST(uint32_t, lineno);
1135
1136         if (*l == '&') {  /* m->cont_level == 0 checked below. */
1137                 ++l;            /* step over */
1138                 m->flag |= OFFADD;
1139         }
1140         if (*l == '(') {
1141                 ++l;            /* step over */
1142                 m->flag |= INDIR;
1143                 if (m->flag & OFFADD)
1144                         m->flag = (m->flag & ~OFFADD) | INDIROFFADD;
1145
1146                 if (*l == '&') {  /* m->cont_level == 0 checked below */
1147                         ++l;            /* step over */
1148                         m->flag |= OFFADD;
1149                 }
1150         }
1151         /* Indirect offsets are not valid at level 0. */
1152         if (m->cont_level == 0 && (m->flag & (OFFADD | INDIROFFADD)))
1153                 if (ms->flags & MAGIC_CHECK)
1154                         file_magwarn(ms, "relative offset at level 0");
1155
1156         /* get offset, then skip over it */
1157         m->offset = (uint32_t)strtoul(l, &t, 0);
1158         if (l == t)
1159                 if (ms->flags & MAGIC_CHECK)
1160                         file_magwarn(ms, "offset `%s' invalid", l);
1161         l = t;
1162
1163         if (m->flag & INDIR) {
1164                 m->in_type = FILE_LONG;
1165                 m->in_offset = 0;
1166                 /*
1167                  * read [.lbs][+-]nnnnn)
1168                  */
1169                 if (*l == '.') {
1170                         l++;
1171                         switch (*l) {
1172                         case 'l':
1173                                 m->in_type = FILE_LELONG;
1174                                 break;
1175                         case 'L':
1176                                 m->in_type = FILE_BELONG;
1177                                 break;
1178                         case 'm':
1179                                 m->in_type = FILE_MELONG;
1180                                 break;
1181                         case 'h':
1182                         case 's':
1183                                 m->in_type = FILE_LESHORT;
1184                                 break;
1185                         case 'H':
1186                         case 'S':
1187                                 m->in_type = FILE_BESHORT;
1188                                 break;
1189                         case 'c':
1190                         case 'b':
1191                         case 'C':
1192                         case 'B':
1193                                 m->in_type = FILE_BYTE;
1194                                 break;
1195                         case 'e':
1196                         case 'f':
1197                         case 'g':
1198                                 m->in_type = FILE_LEDOUBLE;
1199                                 break;
1200                         case 'E':
1201                         case 'F':
1202                         case 'G':
1203                                 m->in_type = FILE_BEDOUBLE;
1204                                 break;
1205                         case 'i':
1206                                 m->in_type = FILE_LEID3;
1207                                 break;
1208                         case 'I':
1209                                 m->in_type = FILE_BEID3;
1210                                 break;
1211                         default:
1212                                 if (ms->flags & MAGIC_CHECK)
1213                                         file_magwarn(ms,
1214                                             "indirect offset type `%c' invalid",
1215                                             *l);
1216                                 break;
1217                         }
1218                         l++;
1219                 }
1220
1221                 m->in_op = 0;
1222                 if (*l == '~') {
1223                         m->in_op |= FILE_OPINVERSE;
1224                         l++;
1225                 }
1226                 if ((op = get_op(*l)) != -1) {
1227                         m->in_op |= op;
1228                         l++;
1229                 }
1230                 if (*l == '(') {
1231                         m->in_op |= FILE_OPINDIRECT;
1232                         l++;
1233                 }
1234                 if (isdigit((unsigned char)*l) || *l == '-') {
1235                         m->in_offset = (int32_t)strtol(l, &t, 0);
1236                         if (l == t)
1237                                 if (ms->flags & MAGIC_CHECK)
1238                                         file_magwarn(ms,
1239                                             "in_offset `%s' invalid", l);
1240                         l = t;
1241                 }
1242                 if (*l++ != ')' ||
1243                     ((m->in_op & FILE_OPINDIRECT) && *l++ != ')'))
1244                         if (ms->flags & MAGIC_CHECK)
1245                                 file_magwarn(ms,
1246                                     "missing ')' in indirect offset");
1247         }
1248         EATAB;
1249
1250 #ifdef ENABLE_CONDITIONALS
1251         m->cond = get_cond(l, &l);
1252         if (check_cond(ms, m->cond, cont_level) == -1)
1253                 return -1;
1254
1255         EATAB;
1256 #endif
1257
1258         if (*l == 'u') {
1259                 ++l;
1260                 m->flag |= UNSIGNED;
1261         }
1262
1263         m->type = get_type(l, &l);
1264         if (m->type == FILE_INVALID) {
1265                 if (ms->flags & MAGIC_CHECK)
1266                         file_magwarn(ms, "type `%s' invalid", l);
1267                 return -1;
1268         }
1269
1270         /* New-style anding: "0 byte&0x80 =0x80 dynamically linked" */
1271         /* New and improved: ~ & | ^ + - * / % -- exciting, isn't it? */
1272
1273         m->mask_op = 0;
1274         if (*l == '~') {
1275                 if (!IS_STRING(m->type))
1276                         m->mask_op |= FILE_OPINVERSE;
1277                 else if (ms->flags & MAGIC_CHECK)
1278                         file_magwarn(ms, "'~' invalid for string types");
1279                 ++l;
1280         }
1281         m->str_range = 0;
1282         m->str_flags = 0;
1283         m->num_mask = 0;
1284         if ((op = get_op(*l)) != -1) {
1285                 if (!IS_STRING(m->type)) {
1286                         uint64_t val;
1287                         ++l;
1288                         m->mask_op |= op;
1289                         val = (uint64_t)strtoull(l, &t, 0);
1290                         l = t;
1291                         m->num_mask = file_signextend(ms, m, val);
1292                         eatsize(&l);
1293                 }
1294                 else if (op == FILE_OPDIVIDE) {
1295                         int have_range = 0;
1296                         while (!isspace((unsigned char)*++l)) {
1297                                 switch (*l) {
1298                                 case '0':  case '1':  case '2':
1299                                 case '3':  case '4':  case '5':
1300                                 case '6':  case '7':  case '8':
1301                                 case '9':
1302                                         if (have_range &&
1303                                             (ms->flags & MAGIC_CHECK))
1304                                                 file_magwarn(ms,
1305                                                     "multiple ranges");
1306                                         have_range = 1;
1307                                         m->str_range = CAST(uint32_t,
1308                                             strtoul(l, &t, 0));
1309                                         if (m->str_range == 0)
1310                                                 file_magwarn(ms,
1311                                                     "zero range");
1312                                         l = t - 1;
1313                                         break;
1314                                 case CHAR_COMPACT_BLANK:
1315                                         m->str_flags |= STRING_COMPACT_BLANK;
1316                                         break;
1317                                 case CHAR_COMPACT_OPTIONAL_BLANK:
1318                                         m->str_flags |=
1319                                             STRING_COMPACT_OPTIONAL_BLANK;
1320                                         break;
1321                                 case CHAR_IGNORE_LOWERCASE:
1322                                         m->str_flags |= STRING_IGNORE_LOWERCASE;
1323                                         break;
1324                                 case CHAR_IGNORE_UPPERCASE:
1325                                         m->str_flags |= STRING_IGNORE_UPPERCASE;
1326                                         break;
1327                                 case CHAR_REGEX_OFFSET_START:
1328                                         m->str_flags |= REGEX_OFFSET_START;
1329                                         break;
1330                                 default:
1331                                         if (ms->flags & MAGIC_CHECK)
1332                                                 file_magwarn(ms,
1333                                                 "string extension `%c' invalid",
1334                                                 *l);
1335                                         return -1;
1336                                 }
1337                                 /* allow multiple '/' for readability */
1338                                 if (l[1] == '/' &&
1339                                     !isspace((unsigned char)l[2]))
1340                                         l++;
1341                         }
1342                         if (string_modifier_check(ms, m) == -1)
1343                                 return -1;
1344                 }
1345                 else {
1346                         if (ms->flags & MAGIC_CHECK)
1347                                 file_magwarn(ms, "invalid string op: %c", *t);
1348                         return -1;
1349                 }
1350         }
1351         /*
1352          * We used to set mask to all 1's here, instead let's just not do
1353          * anything if mask = 0 (unless you have a better idea)
1354          */
1355         EATAB;
1356
1357         switch (*l) {
1358         case '>':
1359         case '<':
1360                 m->reln = *l;
1361                 ++l;
1362                 if (*l == '=') {
1363                         if (ms->flags & MAGIC_CHECK) {
1364                                 file_magwarn(ms, "%c= not supported",
1365                                     m->reln);
1366                                 return -1;
1367                         }
1368                    ++l;
1369                 }
1370                 break;
1371         /* Old-style anding: "0 byte &0x80 dynamically linked" */
1372         case '&':
1373         case '^':
1374         case '=':
1375                 m->reln = *l;
1376                 ++l;
1377                 if (*l == '=') {
1378                    /* HP compat: ignore &= etc. */
1379                    ++l;
1380                 }
1381                 break;
1382         case '!':
1383                 m->reln = *l;
1384                 ++l;
1385                 break;
1386         default:
1387                 m->reln = '=';  /* the default relation */
1388                 if (*l == 'x' && ((isascii((unsigned char)l[1]) &&
1389                     isspace((unsigned char)l[1])) || !l[1])) {
1390                         m->reln = *l;
1391                         ++l;
1392                 }
1393                 break;
1394         }
1395         /*
1396          * Grab the value part, except for an 'x' reln.
1397          */
1398         if (m->reln != 'x' && getvalue(ms, m, &l, action))
1399                 return -1;
1400
1401         /*
1402          * TODO finish this macro and start using it!
1403          * #define offsetcheck {if (offset > HOWMANY-1)
1404          *      magwarn("offset too big"); }
1405          */
1406
1407         /*
1408          * Now get last part - the description
1409          */
1410         EATAB;
1411         if (l[0] == '\b') {
1412                 ++l;
1413                 m->flag |= NOSPACE;
1414         } else if ((l[0] == '\\') && (l[1] == 'b')) {
1415                 ++l;
1416                 ++l;
1417                 m->flag |= NOSPACE;
1418         }
1419         for (i = 0; (m->desc[i++] = *l++) != '\0' && i < sizeof(m->desc); )
1420                 continue;
1421         if (i == sizeof(m->desc)) {
1422                 m->desc[sizeof(m->desc) - 1] = '\0';
1423                 if (ms->flags & MAGIC_CHECK)
1424                         file_magwarn(ms, "description `%s' truncated", m->desc);
1425         }
1426
1427         /*
1428          * We only do this check while compiling, or if any of the magic
1429          * files were not compiled.
1430          */
1431         if (ms->flags & MAGIC_CHECK) {
1432                 if (check_format(ms, m) == -1)
1433                         return -1;
1434         }
1435 #ifndef COMPILE_ONLY
1436         if (action == FILE_CHECK) {
1437                 file_mdump(m);
1438         }
1439 #endif
1440         m->mimetype[0] = '\0';          /* initialise MIME type to none */
1441         if (m->cont_level == 0)
1442                 ++(*nmentryp);          /* make room for next */
1443         return 0;
1444 }
1445
1446 /*
1447  * parse a STRENGTH annotation line from magic file, put into magic[index - 1]
1448  * if valid
1449  */
1450 private int
1451 parse_strength(struct magic_set *ms, struct magic_entry *me, const char *line)
1452 {
1453         const char *l = line;
1454         char *el;
1455         unsigned long factor;
1456         struct magic *m = &me->mp[0];
1457
1458         if (m->factor_op != FILE_FACTOR_OP_NONE) {
1459                 file_magwarn(ms,
1460                     "Current entry already has a strength type: %c %d",
1461                     m->factor_op, m->factor);
1462                 return -1;
1463         }
1464         EATAB;
1465         switch (*l) {
1466         case FILE_FACTOR_OP_NONE:
1467         case FILE_FACTOR_OP_PLUS:
1468         case FILE_FACTOR_OP_MINUS:
1469         case FILE_FACTOR_OP_TIMES:
1470         case FILE_FACTOR_OP_DIV:
1471                 m->factor_op = *l++;
1472                 break;
1473         default:
1474                 file_magwarn(ms, "Unknown factor op `%c'", *l);
1475                 return -1;
1476         }
1477         EATAB;
1478         factor = strtoul(l, &el, 0);
1479         if (factor > 255) {
1480                 file_magwarn(ms, "Too large factor `%lu'", factor);
1481                 goto out;
1482         }
1483         if (*el && !isspace((unsigned char)*el)) {
1484                 file_magwarn(ms, "Bad factor `%s'", l);
1485                 goto out;
1486         }
1487         m->factor = (uint8_t)factor;
1488         if (m->factor == 0 && m->factor_op == FILE_FACTOR_OP_DIV) {
1489                 file_magwarn(ms, "Cannot have factor op `%c' and factor %u",
1490                     m->factor_op, m->factor);
1491                 goto out;
1492         }
1493         return 0;
1494 out:
1495         m->factor_op = FILE_FACTOR_OP_NONE;
1496         m->factor = 0;
1497         return -1;
1498 }
1499
1500 /*
1501  * Parse an Apple CREATOR/TYPE annotation from magic file and put it into magic[index - 1]
1502  */
1503 private int
1504 parse_apple(struct magic_set *ms, struct magic_entry *me, const char *line)
1505 {
1506         size_t i;
1507         const char *l = line;
1508         struct magic *m = &me->mp[me->cont_count == 0 ? 0 : me->cont_count - 1];
1509
1510         if (m->apple[0] != '\0') {
1511                 file_magwarn(ms, "Current entry already has a APPLE type `%.8s',"
1512                     " new type `%s'", m->mimetype, l);
1513                 return -1;
1514         }
1515
1516         EATAB;
1517         for (i = 0; *l && ((isascii((unsigned char)*l) && isalnum((unsigned char)*l))
1518              || strchr("-+/.", *l)) && i < sizeof(m->apple); m->apple[i++] = *l++)
1519                 continue;
1520         if (i == sizeof(m->apple) && *l) {
1521                 if (ms->flags & MAGIC_CHECK)
1522                         file_magwarn(ms, "APPLE type `%s' truncated %zu",
1523                             line, i);
1524         }
1525
1526         if (i > 0)
1527                 return 0;
1528         else
1529                 return -1;
1530 }
1531
1532 /*
1533  * parse a MIME annotation line from magic file, put into magic[index - 1]
1534  * if valid
1535  */
1536 private int
1537 parse_mime(struct magic_set *ms, struct magic_entry *me, const char *line)
1538 {
1539         size_t i;
1540         const char *l = line;
1541         struct magic *m = &me->mp[me->cont_count == 0 ? 0 : me->cont_count - 1];
1542
1543         if (m->mimetype[0] != '\0') {
1544                 file_magwarn(ms, "Current entry already has a MIME type `%s',"
1545                     " new type `%s'", m->mimetype, l);
1546                 return -1;
1547         }
1548
1549         EATAB;
1550         for (i = 0; *l && ((isascii((unsigned char)*l) && isalnum((unsigned char)*l))
1551              || strchr("-+/.", *l)) && i < sizeof(m->mimetype); m->mimetype[i++] = *l++)
1552                 continue;
1553         if (i == sizeof(m->mimetype)) {
1554                 m->desc[sizeof(m->mimetype) - 1] = '\0';
1555                 if (ms->flags & MAGIC_CHECK)
1556                         file_magwarn(ms, "MIME type `%s' truncated %zu",
1557                             m->mimetype, i);
1558         } else
1559                 m->mimetype[i] = '\0';
1560
1561         if (i > 0)
1562                 return 0;
1563         else
1564                 return -1;
1565 }
1566
1567 private int
1568 check_format_type(const char *ptr, int type)
1569 {
1570         int quad = 0;
1571         if (*ptr == '\0') {
1572                 /* Missing format string; bad */
1573                 return -1;
1574         }
1575
1576         switch (type) {
1577         case FILE_FMT_QUAD:
1578                 quad = 1;
1579                 /*FALLTHROUGH*/
1580         case FILE_FMT_NUM:
1581                 if (*ptr == '-')
1582                         ptr++;
1583                 if (*ptr == '.')
1584                         ptr++;
1585                 while (isdigit((unsigned char)*ptr)) ptr++;
1586                 if (*ptr == '.')
1587                         ptr++;
1588                 while (isdigit((unsigned char)*ptr)) ptr++;
1589                 if (quad) {
1590                         if (*ptr++ != 'l')
1591                                 return -1;
1592                         if (*ptr++ != 'l')
1593                                 return -1;
1594                 }
1595
1596                 switch (*ptr++) {
1597                 case 'l':
1598                         switch (*ptr++) {
1599                         case 'i':
1600                         case 'd':
1601                         case 'u':
1602                         case 'x':
1603                         case 'X':
1604                                 return 0;
1605                         default:
1606                                 return -1;
1607                         }
1608
1609                 case 'h':
1610                         switch (*ptr++) {
1611                         case 'h':
1612                                 switch (*ptr++) {
1613                                 case 'i':
1614                                 case 'd':
1615                                 case 'u':
1616                                 case 'x':
1617                                 case 'X':
1618                                         return 0;
1619                                 default:
1620                                         return -1;
1621                                 }
1622                         case 'd':
1623                                 return 0;
1624                         default:
1625                                 return -1;
1626                         }
1627
1628                 case 'i':
1629                 case 'c':
1630                 case 'd':
1631                 case 'u':
1632                 case 'x':
1633                 case 'X':
1634                         return 0;
1635
1636                 default:
1637                         return -1;
1638                 }
1639
1640         case FILE_FMT_FLOAT:
1641         case FILE_FMT_DOUBLE:
1642                 if (*ptr == '-')
1643                         ptr++;
1644                 if (*ptr == '.')
1645                         ptr++;
1646                 while (isdigit((unsigned char)*ptr)) ptr++;
1647                 if (*ptr == '.')
1648                         ptr++;
1649                 while (isdigit((unsigned char)*ptr)) ptr++;
1650
1651                 switch (*ptr++) {
1652                 case 'e':
1653                 case 'E':
1654                 case 'f':
1655                 case 'F':
1656                 case 'g':
1657                 case 'G':
1658                         return 0;
1659
1660                 default:
1661                         return -1;
1662                 }
1663
1664
1665         case FILE_FMT_STR:
1666                 if (*ptr == '-')
1667                         ptr++;
1668                 while (isdigit((unsigned char )*ptr))
1669                         ptr++;
1670                 if (*ptr == '.') {
1671                         ptr++;
1672                         while (isdigit((unsigned char )*ptr))
1673                                 ptr++;
1674                 }
1675
1676                 switch (*ptr++) {
1677                 case 's':
1678                         return 0;
1679                 default:
1680                         return -1;
1681                 }
1682
1683         default:
1684                 /* internal error */
1685                 abort();
1686         }
1687         /*NOTREACHED*/
1688         return -1;
1689 }
1690
1691 /*
1692  * Check that the optional printf format in description matches
1693  * the type of the magic.
1694  */
1695 private int
1696 check_format(struct magic_set *ms, struct magic *m)
1697 {
1698         char *ptr;
1699
1700         for (ptr = m->desc; *ptr; ptr++)
1701                 if (*ptr == '%')
1702                         break;
1703         if (*ptr == '\0') {
1704                 /* No format string; ok */
1705                 return 1;
1706         }
1707
1708         assert(file_nformats == file_nnames);
1709
1710         if (m->type >= file_nformats) {
1711                 file_magwarn(ms, "Internal error inconsistency between "
1712                     "m->type and format strings");
1713                 return -1;
1714         }
1715         if (file_formats[m->type] == FILE_FMT_NONE) {
1716                 file_magwarn(ms, "No format string for `%s' with description "
1717                     "`%s'", m->desc, file_names[m->type]);
1718                 return -1;
1719         }
1720
1721         ptr++;
1722         if (check_format_type(ptr, file_formats[m->type]) == -1) {
1723                 /*
1724                  * TODO: this error message is unhelpful if the format
1725                  * string is not one character long
1726                  */
1727                 file_magwarn(ms, "Printf format `%c' is not valid for type "
1728                     "`%s' in description `%s'", *ptr ? *ptr : '?',
1729                     file_names[m->type], m->desc);
1730                 return -1;
1731         }
1732
1733         for (; *ptr; ptr++) {
1734                 if (*ptr == '%') {
1735                         file_magwarn(ms,
1736                             "Too many format strings (should have at most one) "
1737                             "for `%s' with description `%s'",
1738                             file_names[m->type], m->desc);
1739                         return -1;
1740                 }
1741         }
1742         return 0;
1743 }
1744
1745 /*
1746  * Read a numeric value from a pointer, into the value union of a magic
1747  * pointer, according to the magic type.  Update the string pointer to point
1748  * just after the number read.  Return 0 for success, non-zero for failure.
1749  */
1750 private int
1751 getvalue(struct magic_set *ms, struct magic *m, const char **p, int action)
1752 {
1753         switch (m->type) {
1754         case FILE_BESTRING16:
1755         case FILE_LESTRING16:
1756         case FILE_STRING:
1757         case FILE_PSTRING:
1758         case FILE_REGEX:
1759         case FILE_SEARCH:
1760                 *p = getstr(ms, m, *p, action == FILE_COMPILE);
1761                 if (*p == NULL) {
1762                         if (ms->flags & MAGIC_CHECK)
1763                                 file_magwarn(ms, "cannot get string from `%s'",
1764                                     m->value.s);
1765                         return -1;
1766                 }
1767                 return 0;
1768         case FILE_FLOAT:
1769         case FILE_BEFLOAT:
1770         case FILE_LEFLOAT:
1771                 if (m->reln != 'x') {
1772                         char *ep;
1773 #ifdef HAVE_STRTOF
1774                         m->value.f = strtof(*p, &ep);
1775 #else
1776                         m->value.f = (float)strtod(*p, &ep);
1777 #endif
1778                         *p = ep;
1779                 }
1780                 return 0;
1781         case FILE_DOUBLE:
1782         case FILE_BEDOUBLE:
1783         case FILE_LEDOUBLE:
1784                 if (m->reln != 'x') {
1785                         char *ep;
1786                         m->value.d = strtod(*p, &ep);
1787                         *p = ep;
1788                 }
1789                 return 0;
1790         default:
1791                 if (m->reln != 'x') {
1792                         char *ep;
1793                         m->value.q = file_signextend(ms, m,
1794                             (uint64_t)strtoull(*p, &ep, 0));
1795                         *p = ep;
1796                         eatsize(p);
1797                 }
1798                 return 0;
1799         }
1800 }
1801
1802 /*
1803  * Convert a string containing C character escapes.  Stop at an unescaped
1804  * space or tab.
1805  * Copy the converted version to "m->value.s", and the length in m->vallen.
1806  * Return updated scan pointer as function result. Warn if set.
1807  */
1808 private const char *
1809 getstr(struct magic_set *ms, struct magic *m, const char *s, int warn)
1810 {
1811         const char *origs = s;
1812         char    *p = m->value.s;
1813         size_t  plen = sizeof(m->value.s);
1814         char    *origp = p;
1815         char    *pmax = p + plen - 1;
1816         int     c;
1817         int     val;
1818
1819         while ((c = *s++) != '\0') {
1820                 if (isspace((unsigned char) c))
1821                         break;
1822                 if (p >= pmax) {
1823                         file_error(ms, 0, "string too long: `%s'", origs);
1824                         return NULL;
1825                 }
1826                 if (c == '\\') {
1827                         switch(c = *s++) {
1828
1829                         case '\0':
1830                                 if (warn)
1831                                         file_magwarn(ms, "incomplete escape");
1832                                 goto out;
1833
1834                         case '\t':
1835                                 if (warn) {
1836                                         file_magwarn(ms,
1837                                             "escaped tab found, use \\t instead");
1838                                         warn = 0;       /* already did */
1839                                 }
1840                                 /*FALLTHROUGH*/
1841                         default:
1842                                 if (warn) {
1843                                         if (isprint((unsigned char)c)) {
1844                                                 /* Allow escaping of
1845                                                  * ``relations'' */
1846                                                 if (strchr("<>&^=!", c)
1847                                                     == NULL) {
1848                                                         file_magwarn(ms, "no "
1849                                                             "need to escape "
1850                                                             "`%c'", c);
1851                                                 }
1852                                         } else {
1853                                                 file_magwarn(ms,
1854                                                     "unknown escape sequence: "
1855                                                     "\\%03o", c);
1856                                         }
1857                                 }
1858                                 /*FALLTHROUGH*/
1859                         /* space, perhaps force people to use \040? */
1860                         case ' ':
1861 #if 0
1862                         /*
1863                          * Other things people escape, but shouldn't need to,
1864                          * so we disallow them
1865                          */
1866                         case '\'':
1867                         case '"':
1868                         case '?':
1869 #endif
1870                         /* Relations */
1871                         case '>':
1872                         case '<':
1873                         case '&':
1874                         case '^':
1875                         case '=':
1876                         case '!':
1877                         /* and baskslash itself */
1878                         case '\\':
1879                                 *p++ = (char) c;
1880                                 break;
1881
1882                         case 'a':
1883                                 *p++ = '\a';
1884                                 break;
1885
1886                         case 'b':
1887                                 *p++ = '\b';
1888                                 break;
1889
1890                         case 'f':
1891                                 *p++ = '\f';
1892                                 break;
1893
1894                         case 'n':
1895                                 *p++ = '\n';
1896                                 break;
1897
1898                         case 'r':
1899                                 *p++ = '\r';
1900                                 break;
1901
1902                         case 't':
1903                                 *p++ = '\t';
1904                                 break;
1905
1906                         case 'v':
1907                                 *p++ = '\v';
1908                                 break;
1909
1910                         /* \ and up to 3 octal digits */
1911                         case '0':
1912                         case '1':
1913                         case '2':
1914                         case '3':
1915                         case '4':
1916                         case '5':
1917                         case '6':
1918                         case '7':
1919                                 val = c - '0';
1920                                 c = *s++;  /* try for 2 */
1921                                 if (c >= '0' && c <= '7') {
1922                                         val = (val << 3) | (c - '0');
1923                                         c = *s++;  /* try for 3 */
1924                                         if (c >= '0' && c <= '7')
1925                                                 val = (val << 3) | (c-'0');
1926                                         else
1927                                                 --s;
1928                                 }
1929                                 else
1930                                         --s;
1931                                 *p++ = (char)val;
1932                                 break;
1933
1934                         /* \x and up to 2 hex digits */
1935                         case 'x':
1936                                 val = 'x';      /* Default if no digits */
1937                                 c = hextoint(*s++);     /* Get next char */
1938                                 if (c >= 0) {
1939                                         val = c;
1940                                         c = hextoint(*s++);
1941                                         if (c >= 0)
1942                                                 val = (val << 4) + c;
1943                                         else
1944                                                 --s;
1945                                 } else
1946                                         --s;
1947                                 *p++ = (char)val;
1948                                 break;
1949                         }
1950                 } else
1951                         *p++ = (char)c;
1952         }
1953 out:
1954         *p = '\0';
1955         m->vallen = CAST(unsigned char, (p - origp));
1956         if (m->type == FILE_PSTRING)
1957                 m->vallen++;
1958         return s;
1959 }
1960
1961
1962 /* Single hex char to int; -1 if not a hex char. */
1963 private int
1964 hextoint(int c)
1965 {
1966         if (!isascii((unsigned char) c))
1967                 return -1;
1968         if (isdigit((unsigned char) c))
1969                 return c - '0';
1970         if ((c >= 'a') && (c <= 'f'))
1971                 return c + 10 - 'a';
1972         if (( c>= 'A') && (c <= 'F'))
1973                 return c + 10 - 'A';
1974         return -1;
1975 }
1976
1977
1978 /*
1979  * Print a string containing C character escapes.
1980  */
1981 protected void
1982 file_showstr(FILE *fp, const char *s, size_t len)
1983 {
1984         char    c;
1985
1986         for (;;) {
1987                 c = *s++;
1988                 if (len == ~0U) {
1989                         if (c == '\0')
1990                                 break;
1991                 }
1992                 else  {
1993                         if (len-- == 0)
1994                                 break;
1995                 }
1996                 if (c >= 040 && c <= 0176)      /* TODO isprint && !iscntrl */
1997                         (void) fputc(c, fp);
1998                 else {
1999                         (void) fputc('\\', fp);
2000                         switch (c) {
2001                         case '\a':
2002                                 (void) fputc('a', fp);
2003                                 break;
2004
2005                         case '\b':
2006                                 (void) fputc('b', fp);
2007                                 break;
2008
2009                         case '\f':
2010                                 (void) fputc('f', fp);
2011                                 break;
2012
2013                         case '\n':
2014                                 (void) fputc('n', fp);
2015                                 break;
2016
2017                         case '\r':
2018                                 (void) fputc('r', fp);
2019                                 break;
2020
2021                         case '\t':
2022                                 (void) fputc('t', fp);
2023                                 break;
2024
2025                         case '\v':
2026                                 (void) fputc('v', fp);
2027                                 break;
2028
2029                         default:
2030                                 (void) fprintf(fp, "%.3o", c & 0377);
2031                                 break;
2032                         }
2033                 }
2034         }
2035 }
2036
2037 /*
2038  * eatsize(): Eat the size spec from a number [eg. 10UL]
2039  */
2040 private void
2041 eatsize(const char **p)
2042 {
2043         const char *l = *p;
2044
2045         if (LOWCASE(*l) == 'u')
2046                 l++;
2047
2048         switch (LOWCASE(*l)) {
2049         case 'l':    /* long */
2050         case 's':    /* short */
2051         case 'h':    /* short */
2052         case 'b':    /* char/byte */
2053         case 'c':    /* char/byte */
2054                 l++;
2055                 /*FALLTHROUGH*/
2056         default:
2057                 break;
2058         }
2059
2060         *p = l;
2061 }
2062
2063 /*
2064  * handle a compiled file.
2065  */
2066 private int
2067 apprentice_map(struct magic_set *ms, struct magic **magicp, uint32_t *nmagicp,
2068     const char *fn)
2069 {
2070         int fd;
2071         struct stat st;
2072         uint32_t *ptr;
2073         uint32_t version;
2074         int needsbyteswap;
2075         char *dbname = NULL;
2076         void *mm = NULL;
2077
2078         dbname = mkdbname(ms, fn, 0);
2079         if (dbname == NULL)
2080                 goto error2;
2081
2082         if ((fd = open(dbname, O_RDONLY|O_BINARY)) == -1)
2083                 goto error2;
2084
2085         if (fstat(fd, &st) == -1) {
2086                 file_error(ms, errno, "cannot stat `%s'", dbname);
2087                 goto error1;
2088         }
2089         if (st.st_size < 8) {
2090                 file_error(ms, 0, "file `%s' is too small", dbname);
2091                 goto error1;
2092         }
2093
2094 #ifdef QUICK
2095         if ((mm = mmap(0, (size_t)st.st_size, PROT_READ|PROT_WRITE,
2096             MAP_PRIVATE|MAP_FILE, fd, (off_t)0)) == MAP_FAILED) {
2097                 file_error(ms, errno, "cannot map `%s'", dbname);
2098                 goto error1;
2099         }
2100 #define RET     2
2101 #else
2102         if ((mm = CAST(void *, malloc((size_t)st.st_size))) == NULL) {
2103                 file_oomem(ms, (size_t)st.st_size);
2104                 goto error1;
2105         }
2106         if (read(fd, mm, (size_t)st.st_size) != (ssize_t)st.st_size) {
2107                 file_badread(ms);
2108                 goto error1;
2109         }
2110 #define RET     1
2111 #endif
2112         *magicp = CAST(struct magic *, mm);
2113         (void)close(fd);
2114         fd = -1;
2115         ptr = (uint32_t *)(void *)*magicp;
2116         if (*ptr != MAGICNO) {
2117                 if (swap4(*ptr) != MAGICNO) {
2118                         file_error(ms, 0, "bad magic in `%s'", dbname);
2119                         goto error1;
2120                 }
2121                 needsbyteswap = 1;
2122         } else
2123                 needsbyteswap = 0;
2124         if (needsbyteswap)
2125                 version = swap4(ptr[1]);
2126         else
2127                 version = ptr[1];
2128         if (version != VERSIONNO) {
2129                 file_error(ms, 0, "File %d.%d supports only version %d magic "
2130                     "files. `%s' is version %d", FILE_VERSION_MAJOR, patchlevel,
2131                     VERSIONNO, dbname, version);
2132                 goto error1;
2133         }
2134         *nmagicp = (uint32_t)(st.st_size / sizeof(struct magic));
2135         if (*nmagicp > 0)
2136                 (*nmagicp)--;
2137         (*magicp)++;
2138         if (needsbyteswap)
2139                 byteswap(*magicp, *nmagicp);
2140         free(dbname);
2141         return RET;
2142
2143 error1:
2144         if (fd != -1)
2145                 (void)close(fd);
2146         if (mm) {
2147 #ifdef QUICK
2148                 (void)munmap((void *)mm, (size_t)st.st_size);
2149 #else
2150                 free(mm);
2151 #endif
2152         } else {
2153                 *magicp = NULL;
2154                 *nmagicp = 0;
2155         }
2156 error2:
2157         free(dbname);
2158         return -1;
2159 }
2160
2161 private const uint32_t ar[] = {
2162     MAGICNO, VERSIONNO
2163 };
2164 /*
2165  * handle an mmaped file.
2166  */
2167 private int
2168 apprentice_compile(struct magic_set *ms, struct magic **magicp,
2169     uint32_t *nmagicp, const char *fn)
2170 {
2171         int fd;
2172         char *dbname;
2173         int rv = -1;
2174
2175         dbname = mkdbname(ms, fn, 1);
2176
2177         if (dbname == NULL)
2178                 goto out;
2179
2180         if ((fd = open(dbname, O_WRONLY|O_CREAT|O_TRUNC|O_BINARY, 0644)) == -1) {
2181                 file_error(ms, errno, "cannot open `%s'", dbname);
2182                 goto out;
2183         }
2184
2185         if (write(fd, ar, sizeof(ar)) != (ssize_t)sizeof(ar)) {
2186                 file_error(ms, errno, "error writing `%s'", dbname);
2187                 goto out;
2188         }
2189
2190         if (lseek(fd, (off_t)sizeof(struct magic), SEEK_SET)
2191             != sizeof(struct magic)) {
2192                 file_error(ms, errno, "error seeking `%s'", dbname);
2193                 goto out;
2194         }
2195
2196         if (write(fd, *magicp, (sizeof(struct magic) * *nmagicp))
2197             != (ssize_t)(sizeof(struct magic) * *nmagicp)) {
2198                 file_error(ms, errno, "error writing `%s'", dbname);
2199                 goto out;
2200         }
2201
2202         (void)close(fd);
2203         rv = 0;
2204 out:
2205         free(dbname);
2206         return rv;
2207 }
2208
2209 private const char ext[] = ".mgc";
2210 /*
2211  * make a dbname
2212  */
2213 private char *
2214 mkdbname(struct magic_set *ms, const char *fn, int strip)
2215 {
2216         const char *p, *q;
2217         char *buf;
2218
2219         if (strip) {
2220                 if ((p = strrchr(fn, '/')) != NULL)
2221                         fn = ++p;
2222         }
2223
2224         for (q = fn; *q; q++)
2225                 continue;
2226         /* Look for .mgc */
2227         for (p = ext + sizeof(ext) - 1; p >= ext && q >= fn; p--, q--)
2228                 if (*p != *q)
2229                         break;
2230
2231         /* Did not find .mgc, restore q */
2232         if (p >= ext)
2233                 while (*q)
2234                         q++;
2235
2236         q++;
2237         /* Compatibility with old code that looked in .mime */
2238         if (ms->flags & MAGIC_MIME) {
2239                 asprintf(&buf, "%.*s.mime%s", (int)(q - fn), fn, ext);
2240                 if (access(buf, R_OK) != -1) {
2241                         ms->flags &= MAGIC_MIME_TYPE;
2242                         return buf;
2243                 }
2244                 free(buf);
2245         }
2246         asprintf(&buf, "%.*s%s", (int)(q - fn), fn, ext);
2247
2248         /* Compatibility with old code that looked in .mime */
2249         if (strstr(p, ".mime") != NULL)
2250                 ms->flags &= MAGIC_MIME_TYPE;
2251         return buf;
2252 }
2253
2254 /*
2255  * Byteswap an mmap'ed file if needed
2256  */
2257 private void
2258 byteswap(struct magic *magic, uint32_t nmagic)
2259 {
2260         uint32_t i;
2261         for (i = 0; i < nmagic; i++)
2262                 bs1(&magic[i]);
2263 }
2264
2265 /*
2266  * swap a short
2267  */
2268 private uint16_t
2269 swap2(uint16_t sv)
2270 {
2271         uint16_t rv;
2272         uint8_t *s = (uint8_t *)(void *)&sv;
2273         uint8_t *d = (uint8_t *)(void *)&rv;
2274         d[0] = s[1];
2275         d[1] = s[0];
2276         return rv;
2277 }
2278
2279 /*
2280  * swap an int
2281  */
2282 private uint32_t
2283 swap4(uint32_t sv)
2284 {
2285         uint32_t rv;
2286         uint8_t *s = (uint8_t *)(void *)&sv;
2287         uint8_t *d = (uint8_t *)(void *)&rv;
2288         d[0] = s[3];
2289         d[1] = s[2];
2290         d[2] = s[1];
2291         d[3] = s[0];
2292         return rv;
2293 }
2294
2295 /*
2296  * swap a quad
2297  */
2298 private uint64_t
2299 swap8(uint64_t sv)
2300 {
2301         uint64_t rv;
2302         uint8_t *s = (uint8_t *)(void *)&sv;
2303         uint8_t *d = (uint8_t *)(void *)&rv;
2304 #if 0
2305         d[0] = s[3];
2306         d[1] = s[2];
2307         d[2] = s[1];
2308         d[3] = s[0];
2309         d[4] = s[7];
2310         d[5] = s[6];
2311         d[6] = s[5];
2312         d[7] = s[4];
2313 #else
2314         d[0] = s[7];
2315         d[1] = s[6];
2316         d[2] = s[5];
2317         d[3] = s[4];
2318         d[4] = s[3];
2319         d[5] = s[2];
2320         d[6] = s[1];
2321         d[7] = s[0];
2322 #endif
2323         return rv;
2324 }
2325
2326 /*
2327  * byteswap a single magic entry
2328  */
2329 private void
2330 bs1(struct magic *m)
2331 {
2332         m->cont_level = swap2(m->cont_level);
2333         m->offset = swap4((uint32_t)m->offset);
2334         m->in_offset = swap4((uint32_t)m->in_offset);
2335         m->lineno = swap4((uint32_t)m->lineno);
2336         if (IS_STRING(m->type)) {
2337                 m->str_range = swap4(m->str_range);
2338                 m->str_flags = swap4(m->str_flags);
2339         }
2340         else {
2341                 m->value.q = swap8(m->value.q);
2342                 m->num_mask = swap8(m->num_mask);
2343         }
2344 }