contrib/file-4/src/apprentice.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  * apprentice - make one pass through /etc/magic, learning its secrets.
  30  */
  31
  32 #include "file.h"
  33 #include "magic.h"
  34 #include "patchlevel.h"
  35 #include <stdlib.h>
  36 #ifdef HAVE_UNISTD_H
  37 #include <unistd.h>
  38 #endif
  39 #include <string.h>
  40 #include <assert.h>
  41 #include <ctype.h>
  42 #include <fcntl.h>
  43 #include <sys/stat.h>
  44 #include <sys/param.h>
  45 #ifdef QUICK
  46 #include <sys/mman.h>
  47 #endif
  48
  49 #ifndef lint
  50 FILE_RCSID("@(#)$File: apprentice.c,v 1.109 2007/12/27 20:52:36 christos Exp $")
  51 #endif  /* lint */
  52
  53 #define EATAB {while (isascii((unsigned char) *l) && \
  54                       isspace((unsigned char) *l))  ++l;}
  55 #define LOWCASE(l) (isupper((unsigned char) (l)) ? \
  56                         tolower((unsigned char) (l)) : (l))
  57 /*
  58  * Work around a bug in headers on Digital Unix.
  59  * At least confirmed for: OSF1 V4.0 878
  60  */
  61 #if defined(__osf__) && defined(__DECC)
  62 #ifdef MAP_FAILED
  63 #undef MAP_FAILED
  64 #endif
  65 #endif
  66
  67 #ifndef MAP_FAILED
  68 #define MAP_FAILED (void *) -1
  69 #endif
  70
  71 #ifndef MAP_FILE
  72 #define MAP_FILE 0
  73 #endif
  74
  75 #ifndef MAXPATHLEN
  76 #define MAXPATHLEN      1024
  77 #endif
  78
  79 struct magic_entry {
  80         struct magic *mp;
  81         uint32_t cont_count;
  82         uint32_t max_count;
  83 };
  84
  85 int file_formats[FILE_NAMES_SIZE];
  86 const size_t file_nformats = FILE_NAMES_SIZE;
  87 const char *file_names[FILE_NAMES_SIZE];
  88 const size_t file_nnames = FILE_NAMES_SIZE;
  89
  90 private int getvalue(struct magic_set *ms, struct magic *, const char **, int);
  91 private int hextoint(int);
  92 private const char *getstr(struct magic_set *, const char *, char *, int,
  93     int *, int);
  94 private int parse(struct magic_set *, struct magic_entry **, uint32_t *,
  95     const char *, size_t, int);
  96 private void eatsize(const char **);
  97 private int apprentice_1(struct magic_set *, const char *, int, struct mlist *);
  98 private size_t apprentice_magic_strength(const struct magic *);
  99 private int apprentice_sort(const void *, const void *);
 100 private int apprentice_file(struct magic_set *, struct magic **, uint32_t *,
 101     const char *, int);
 102 private void byteswap(struct magic *, uint32_t);
 103 private void bs1(struct magic *);
 104 private uint16_t swap2(uint16_t);
 105 private uint32_t swap4(uint32_t);
 106 private uint64_t swap8(uint64_t);
 107 private char *mkdbname(const char *, char *, size_t, int);
 108 private int apprentice_map(struct magic_set *, struct magic **, uint32_t *,
 109     const char *);
 110 private int apprentice_compile(struct magic_set *, struct magic **, uint32_t *,
 111     const char *);
 112 private int check_format_type(const char *, int);
 113 private int check_format(struct magic_set *, struct magic *);
 114 private int get_op(char);
 115
 116 private size_t maxmagic = 0;
 117 private size_t magicsize = sizeof(struct magic);
 118
 119
 120 #ifdef COMPILE_ONLY
 121
 122 int main(int, char *[]);
 123
 124 int
 125 main(int argc, char *argv[])
 126 {
 127         int ret;
 128         struct magic_set *ms;
 129         char *progname;
 130
 131         if ((progname = strrchr(argv[0], '/')) != NULL)
 132                 progname++;
 133         else
 134                 progname = argv[0];
 135
 136         if (argc != 2) {
 137                 (void)fprintf(stderr, "Usage: %s file\n", progname);
 138                 return 1;
 139         }
 140
 141         if ((ms = magic_open(MAGIC_CHECK)) == NULL) {
 142                 (void)fprintf(stderr, "%s: %s\n", progname, strerror(errno));
 143                 return 1;
 144         }
 145         ret = magic_compile(ms, argv[1]) == -1 ? 1 : 0;
 146         if (ret == 1)
 147                 (void)fprintf(stderr, "%s: %s\n", progname, magic_error(ms));
 148         magic_close(ms);
 149         return ret;
 150 }
 151 #endif /* COMPILE_ONLY */
 152
 153 static const struct type_tbl_s {
 154         const char *name;
 155         const size_t len;
 156         const int type;
 157         const int format;
 158 } type_tbl[] = {
 159 # define XX(s)          s, (sizeof(s) - 1)
 160 # define XX_NULL        NULL, 0
 161         { XX("byte"),           FILE_BYTE,              FILE_FMT_NUM },
 162         { XX("short"),          FILE_SHORT,             FILE_FMT_NUM },
 163         { XX("default"),        FILE_DEFAULT,           FILE_FMT_STR },
 164         { XX("long"),           FILE_LONG,              FILE_FMT_NUM },
 165         { XX("string"),         FILE_STRING,            FILE_FMT_STR },
 166         { XX("date"),           FILE_DATE,              FILE_FMT_STR },
 167         { XX("beshort"),        FILE_BESHORT,           FILE_FMT_NUM },
 168         { XX("belong"),         FILE_BELONG,            FILE_FMT_NUM },
 169         { XX("bedate"),         FILE_BEDATE,            FILE_FMT_STR },
 170         { XX("leshort"),        FILE_LESHORT,           FILE_FMT_NUM },
 171         { XX("lelong"),         FILE_LELONG,            FILE_FMT_NUM },
 172         { XX("ledate"),         FILE_LEDATE,            FILE_FMT_STR },
 173         { XX("pstring"),        FILE_PSTRING,           FILE_FMT_STR },
 174         { XX("ldate"),          FILE_LDATE,             FILE_FMT_STR },
 175         { XX("beldate"),        FILE_BELDATE,           FILE_FMT_STR },
 176         { XX("leldate"),        FILE_LELDATE,           FILE_FMT_STR },
 177         { XX("regex"),          FILE_REGEX,             FILE_FMT_STR },
 178         { XX("bestring16"),     FILE_BESTRING16,        FILE_FMT_STR },
 179         { XX("lestring16"),     FILE_LESTRING16,        FILE_FMT_STR },
 180         { XX("search"),         FILE_SEARCH,            FILE_FMT_STR },
 181         { XX("medate"),         FILE_MEDATE,            FILE_FMT_STR },
 182         { XX("meldate"),        FILE_MELDATE,           FILE_FMT_STR },
 183         { XX("melong"),         FILE_MELONG,            FILE_FMT_NUM },
 184         { XX("quad"),           FILE_QUAD,              FILE_FMT_QUAD },
 185         { XX("lequad"),         FILE_LEQUAD,            FILE_FMT_QUAD },
 186         { XX("bequad"),         FILE_BEQUAD,            FILE_FMT_QUAD },
 187         { XX("qdate"),          FILE_QDATE,             FILE_FMT_STR },
 188         { XX("leqdate"),        FILE_LEQDATE,           FILE_FMT_STR },
 189         { XX("beqdate"),        FILE_BEQDATE,           FILE_FMT_STR },
 190         { XX("qldate"),         FILE_QLDATE,            FILE_FMT_STR },
 191         { XX("leqldate"),       FILE_LEQLDATE,          FILE_FMT_STR },
 192         { XX("beqldate"),       FILE_BEQLDATE,          FILE_FMT_STR },
 193         { XX("float"),          FILE_FLOAT,             FILE_FMT_FLOAT },
 194         { XX("befloat"),        FILE_BEFLOAT,           FILE_FMT_FLOAT },
 195         { XX("lefloat"),        FILE_LEFLOAT,           FILE_FMT_FLOAT },
 196         { XX("double"),         FILE_DOUBLE,            FILE_FMT_DOUBLE },
 197         { XX("bedouble"),       FILE_BEDOUBLE,          FILE_FMT_DOUBLE },
 198         { XX("ledouble"),       FILE_LEDOUBLE,          FILE_FMT_DOUBLE },
 199         { XX_NULL,              FILE_INVALID,           FILE_FMT_NONE },
 200 # undef XX
 201 # undef XX_NULL
 202 };
 203
 204 private int
 205 get_type(const char *l, const char **t)
 206 {
 207         const struct type_tbl_s *p;
 208
 209         for (p = type_tbl; p->name; p++) {
 210                 if (strncmp(l, p->name, p->len) == 0) {
 211                         if (t)
 212                                 *t = l + p->len;
 213                         break;
 214                 }
 215         }
 216         return p->type;
 217 }
 218
 219 private void
 220 init_file_tables(void)
 221 {
 222         static int done = 0;
 223         const struct type_tbl_s *p;
 224
 225         if (done)
 226                 return;
 227         done++;
 228
 229         for (p = type_tbl; p->name; p++) {
 230                 assert(p->type < FILE_NAMES_SIZE);
 231                 file_names[p->type] = p->name;
 232                 file_formats[p->type] = p->format;
 233         }
 234 }
 235
 236 /*
 237  * Handle one file.
 238  */
 239 private int
 240 apprentice_1(struct magic_set *ms, const char *fn, int action,
 241     struct mlist *mlist)
 242 {
 243         struct magic *magic = NULL;
 244         uint32_t nmagic = 0;
 245         struct mlist *ml;
 246         int rv = -1;
 247         int mapped;
 248
 249         if (magicsize != FILE_MAGICSIZE) {
 250                 file_error(ms, 0, "magic element size %lu != %lu",
 251                     (unsigned long)sizeof(*magic),
 252                     (unsigned long)FILE_MAGICSIZE);
 253                 return -1;
 254         }
 255
 256         if (action == FILE_COMPILE) {
 257                 rv = apprentice_file(ms, &magic, &nmagic, fn, action);
 258                 if (rv != 0)
 259                         return -1;
 260                 rv = apprentice_compile(ms, &magic, &nmagic, fn);
 261                 free(magic);
 262                 return rv;
 263         }
 264
 265 #ifndef COMPILE_ONLY
 266         if ((rv = apprentice_map(ms, &magic, &nmagic, fn)) == -1) {
 267                 if (ms->flags & MAGIC_CHECK)
 268                         file_magwarn(ms, "using regular magic file `%s'", fn);
 269                 rv = apprentice_file(ms, &magic, &nmagic, fn, action);
 270                 if (rv != 0)
 271                         return -1;
 272         }
 273
 274         mapped = rv;
 275
 276         if (magic == NULL || nmagic == 0) {
 277                 file_delmagic(magic, mapped, nmagic);
 278                 return -1;
 279         }
 280
 281         if ((ml = malloc(sizeof(*ml))) == NULL) {
 282                 file_delmagic(magic, mapped, nmagic);
 283                 file_oomem(ms, sizeof(*ml));
 284                 return -1;
 285         }
 286
 287         ml->magic = magic;
 288         ml->nmagic = nmagic;
 289         ml->mapped = mapped;
 290
 291         mlist->prev->next = ml;
 292         ml->prev = mlist->prev;
 293         ml->next = mlist;
 294         mlist->prev = ml;
 295
 296         return 0;
 297 #endif /* COMPILE_ONLY */
 298 }
 299
 300 protected void
 301 file_delmagic(struct magic *p, int type, size_t entries)
 302 {
 303         if (p == NULL)
 304                 return;
 305         switch (type) {
 306 #ifdef QUICK
 307         case 2:
 308                 p--;
 309                 (void)munmap((void *)p, sizeof(*p) * (entries + 1));
 310                 break;
 311 #endif
 312         case 1:
 313                 p--;
 314                 /*FALLTHROUGH*/
 315         case 0:
 316                 free(p);
 317                 break;
 318         default:
 319                 abort();
 320         }
 321 }
 322
 323 /* const char *fn: list of magic files */
 324 protected struct mlist *
 325 file_apprentice(struct magic_set *ms, const char *fn, int action)
 326 {
 327         char *p, *mfn, *afn = NULL;
 328         int file_err, errs = -1;
 329         struct mlist *mlist;
 330         static const char mime[] = ".mime";
 331
 332         init_file_tables();
 333
 334         if (fn == NULL)
 335                 fn = getenv("MAGIC");
 336         if (fn == NULL)
 337                 fn = MAGIC;
 338
 339         if ((mfn = strdup(fn)) == NULL) {
 340                 file_oomem(ms, strlen(fn));
 341                 return NULL;
 342         }
 343         fn = mfn;
 344
 345         if ((mlist = malloc(sizeof(*mlist))) == NULL) {
 346                 free(mfn);
 347                 file_oomem(ms, sizeof(*mlist));
 348                 return NULL;
 349         }
 350         mlist->next = mlist->prev = mlist;
 351
 352         while (fn) {
 353                 p = strchr(fn, PATHSEP);
 354                 if (p)
 355                         *p++ = '\0';
 356                 if (*fn == '\0')
 357                         break;
 358                 if (ms->flags & MAGIC_MIME) {
 359                         size_t len = strlen(fn) + sizeof(mime);
 360                         if ((afn = malloc(len)) == NULL) {
 361                                 free(mfn);
 362                                 free(mlist);
 363                                 file_oomem(ms, len);
 364                                 return NULL;
 365                         }
 366                         (void)strcpy(afn, fn);
 367                         (void)strcat(afn, mime);
 368                         fn = afn;
 369                 }
 370                 file_err = apprentice_1(ms, fn, action, mlist);
 371                 if (file_err > errs)
 372                         errs = file_err;
 373                 if (afn) {
 374                         free(afn);
 375                         afn = NULL;
 376                 }
 377                 fn = p;
 378         }
 379         if (errs == -1) {
 380                 free(mfn);
 381                 free(mlist);
 382                 mlist = NULL;
 383                 file_error(ms, 0, "could not find any magic files!");
 384                 return NULL;
 385         }
 386         free(mfn);
 387         return mlist;
 388 }
 389
 390 /*
 391  * Get weight of this magic entry, for sorting purposes.
 392  */
 393 private size_t
 394 apprentice_magic_strength(const struct magic *m)
 395 {
 396 #define MULT 10
 397         size_t val = 2 * MULT;  /* baseline strength */
 398
 399         switch (m->type) {
 400         case FILE_DEFAULT:      /* make sure this sorts last */
 401                 return 0;
 402
 403         case FILE_BYTE:
 404                 val += 1 * MULT;
 405                 break;
 406
 407         case FILE_SHORT:
 408         case FILE_LESHORT:
 409         case FILE_BESHORT:
 410                 val += 2 * MULT;
 411                 break;
 412
 413         case FILE_LONG:
 414         case FILE_LELONG:
 415         case FILE_BELONG:
 416         case FILE_MELONG:
 417                 val += 4 * MULT;
 418                 break;
 419
 420         case FILE_PSTRING:
 421         case FILE_STRING:
 422                 val += m->vallen * MULT;
 423                 break;
 424
 425         case FILE_BESTRING16:
 426         case FILE_LESTRING16:
 427                 val += m->vallen * MULT / 2;
 428                 break;
 429
 430         case FILE_SEARCH:
 431         case FILE_REGEX:
 432                 val += m->vallen;
 433                 break;
 434
 435         case FILE_DATE:
 436         case FILE_LEDATE:
 437         case FILE_BEDATE:
 438         case FILE_MEDATE:
 439         case FILE_LDATE:
 440         case FILE_LELDATE:
 441         case FILE_BELDATE:
 442         case FILE_MELDATE:
 443         case FILE_FLOAT:
 444         case FILE_BEFLOAT:
 445         case FILE_LEFLOAT:
 446                 val += 4 * MULT;
 447                 break;
 448
 449         case FILE_QUAD:
 450         case FILE_BEQUAD:
 451         case FILE_LEQUAD:
 452         case FILE_QDATE:
 453         case FILE_LEQDATE:
 454         case FILE_BEQDATE:
 455         case FILE_QLDATE:
 456         case FILE_LEQLDATE:
 457         case FILE_BEQLDATE:
 458         case FILE_DOUBLE:
 459         case FILE_BEDOUBLE:
 460         case FILE_LEDOUBLE:
 461                 val += 8 * MULT;
 462                 break;
 463
 464         default:
 465                 val = 0;
 466                 (void)fprintf(stderr, "Bad type %d\n", m->type);
 467                 abort();
 468         }
 469
 470         switch (m->reln) {
 471         case 'x':       /* matches anything penalize */
 472                 val = 0;
 473                 break;
 474
 475         case '!':
 476         case '=':       /* Exact match, prefer */
 477                 val += MULT;
 478                 break;
 479
 480         case '>':
 481         case '<':       /* comparison match reduce strength */
 482                 val -= 2 * MULT;
 483                 break;
 484
 485         case '^':
 486         case '&':       /* masking bits, we could count them too */
 487                 val -= MULT;
 488                 break;
 489
 490         default:
 491                 (void)fprintf(stderr, "Bad relation %c\n", m->reln);
 492                 abort();
 493         }
 494
 495         if (val == 0)   /* ensure we only return 0 for FILE_DEFAULT */
 496                 val = 1;
 497
 498         return val;
 499 }
 500
 501 /*
 502  * Sort callback for sorting entries by "strength" (basically length)
 503  */
 504 private int
 505 apprentice_sort(const void *a, const void *b)
 506 {
 507         const struct magic_entry *ma = a;
 508         const struct magic_entry *mb = b;
 509         size_t sa = apprentice_magic_strength(ma->mp);
 510         size_t sb = apprentice_magic_strength(mb->mp);
 511         if (sa == sb)
 512                 return 0;
 513         else if (sa > sb)
 514                 return -1;
 515         else
 516                 return 1;
 517 }
 518
 519 /*
 520  * parse from a file
 521  * const char *fn: name of magic file
 522  */
 523 private int
 524 apprentice_file(struct magic_set *ms, struct magic **magicp, uint32_t *nmagicp,
 525     const char *fn, int action)
 526 {
 527         private const char hdr[] =
 528                 "cont\toffset\ttype\topcode\tmask\tvalue\tdesc";
 529         FILE *f;
 530         char line[BUFSIZ];
 531         int errs = 0;
 532         struct magic_entry *marray;
 533         uint32_t marraycount, i, mentrycount = 0;
 534         size_t lineno = 0;
 535
 536         ms->flags |= MAGIC_CHECK;       /* Enable checks for parsed files */
 537
 538         f = fopen(ms->file = fn, "r");
 539         if (f == NULL) {
 540                 if (errno != ENOENT)
 541                         file_error(ms, errno, "cannot read magic file `%s'",
 542                             fn);
 543                 return -1;
 544         }
 545
 546         maxmagic = MAXMAGIS;
 547         if ((marray = calloc(maxmagic, sizeof(*marray))) == NULL) {
 548                 (void)fclose(f);
 549                 file_oomem(ms, maxmagic * sizeof(*marray));
 550                 return -1;
 551         }
 552         marraycount = 0;
 553
 554         /* print silly verbose header for USG compat. */
 555         if (action == FILE_CHECK)
 556                 (void)fprintf(stderr, "%s\n", hdr);
 557
 558         /* read and parse this file */
 559         for (ms->line = 1; fgets(line, sizeof(line), f) != NULL; ms->line++) {
 560                 size_t len;
 561                 len = strlen(line);
 562                 if (len == 0) /* null line, garbage, etc */
 563                         continue;
 564                 if (line[len - 1] == '\n') {
 565                         lineno++;
 566                         line[len - 1] = '\0'; /* delete newline */
 567                 }
 568                 if (line[0] == '\0')    /* empty, do not parse */
 569                         continue;
 570                 if (line[0] == '#')     /* comment, do not parse */
 571                         continue;
 572                 if (parse(ms, &marray, &marraycount, line, lineno, action) != 0)
 573                         errs++;
 574         }
 575
 576         (void)fclose(f);
 577         if (errs)
 578                 goto out;
 579
 580 #ifndef NOORDER
 581         qsort(marray, marraycount, sizeof(*marray), apprentice_sort);
 582         /*
 583          * Make sure that any level 0 "default" line is last (if one exists).
 584          */
 585         for (i = 0; i < marraycount; i++) {
 586                 if (marray[i].mp->cont_level == 0 &&
 587                     marray[i].mp->type == FILE_DEFAULT) {
 588                         while (++i < marraycount)
 589                                 if (marray[i].mp->cont_level == 0)
 590                                         break;
 591                         if (i != marraycount) {
 592                                 ms->line = marray[i].mp->lineno; /* XXX - Ugh! */
 593                                 file_magwarn(ms,
 594                                     "level 0 \"default\" did not sort last");
 595                         }
 596                         break;
 597                 }
 598         }
 599 #endif
 600
 601         for (i = 0; i < marraycount; i++)
 602                 mentrycount += marray[i].cont_count;
 603
 604         if ((*magicp = malloc(sizeof(**magicp) * mentrycount)) == NULL) {
 605                 file_oomem(ms, sizeof(**magicp) * mentrycount);
 606                 errs++;
 607                 goto out;
 608         }
 609
 610         mentrycount = 0;
 611         for (i = 0; i < marraycount; i++) {
 612                 (void)memcpy(*magicp + mentrycount, marray[i].mp,
 613                     marray[i].cont_count * sizeof(**magicp));
 614                 mentrycount += marray[i].cont_count;
 615         }
 616 out:
 617         for (i = 0; i < marraycount; i++)
 618                 free(marray[i].mp);
 619         free(marray);
 620         if (errs) {
 621                 *magicp = NULL;
 622                 *nmagicp = 0;
 623                 return errs;
 624         } else {
 625                 *nmagicp = mentrycount;
 626                 return 0;
 627         }
 628
 629 }
 630
 631 /*
 632  * extend the sign bit if the comparison is to be signed
 633  */
 634 protected uint64_t
 635 file_signextend(struct magic_set *ms, struct magic *m, uint64_t v)
 636 {
 637         if (!(m->flag & UNSIGNED)) {
 638                 switch(m->type) {
 639                 /*
 640                  * Do not remove the casts below.  They are
 641                  * vital.  When later compared with the data,
 642                  * the sign extension must have happened.
 643                  */
 644                 case FILE_BYTE:
 645                         v = (char) v;
 646                         break;
 647                 case FILE_SHORT:
 648                 case FILE_BESHORT:
 649                 case FILE_LESHORT:
 650                         v = (short) v;
 651                         break;
 652                 case FILE_DATE:
 653                 case FILE_BEDATE:
 654                 case FILE_LEDATE:
 655                 case FILE_MEDATE:
 656                 case FILE_LDATE:
 657                 case FILE_BELDATE:
 658                 case FILE_LELDATE:
 659                 case FILE_MELDATE:
 660                 case FILE_LONG:
 661                 case FILE_BELONG:
 662                 case FILE_LELONG:
 663                 case FILE_MELONG:
 664                 case FILE_FLOAT:
 665                 case FILE_BEFLOAT:
 666                 case FILE_LEFLOAT:
 667                         v = (int32_t) v;
 668                         break;
 669                 case FILE_QUAD:
 670                 case FILE_BEQUAD:
 671                 case FILE_LEQUAD:
 672                 case FILE_QDATE:
 673                 case FILE_QLDATE:
 674                 case FILE_BEQDATE:
 675                 case FILE_BEQLDATE:
 676                 case FILE_LEQDATE:
 677                 case FILE_LEQLDATE:
 678                 case FILE_DOUBLE:
 679                 case FILE_BEDOUBLE:
 680                 case FILE_LEDOUBLE:
 681                         v = (int64_t) v;
 682                         break;
 683                 case FILE_STRING:
 684                 case FILE_PSTRING:
 685                 case FILE_BESTRING16:
 686                 case FILE_LESTRING16:
 687                 case FILE_REGEX:
 688                 case FILE_SEARCH:
 689                 case FILE_DEFAULT:
 690                         break;
 691                 default:
 692                         if (ms->flags & MAGIC_CHECK)
 693                             file_magwarn(ms, "cannot happen: m->type=%d\n",
 694                                     m->type);
 695                         return ~0U;
 696                 }
 697         }
 698         return v;
 699 }
 700
 701 private int
 702 string_modifier_check(struct magic_set *ms, struct magic const *m)
 703 {
 704         if ((ms->flags & MAGIC_CHECK) == 0)
 705                 return 0;
 706
 707         switch (m->type) {
 708         case FILE_BESTRING16:
 709         case FILE_LESTRING16:
 710                 if (m->str_flags != 0) {
 711                         file_magwarn(ms, "no modifiers allowed for 16-bit strings\n");
 712                         return -1;
 713                 }
 714                 break;
 715         case FILE_STRING:
 716         case FILE_PSTRING:
 717                 if ((m->str_flags & REGEX_OFFSET_START) != 0) {
 718                         file_magwarn(ms, "'/%c' only allowed on regex and search\n",
 719                             CHAR_REGEX_OFFSET_START);
 720                         return -1;
 721                 }
 722                 break;
 723         case FILE_SEARCH:
 724                 break;
 725         case FILE_REGEX:
 726                 if ((m->str_flags & STRING_COMPACT_BLANK) != 0) {
 727                         file_magwarn(ms, "'/%c' not allowed on regex\n",
 728                             CHAR_COMPACT_BLANK);
 729                         return -1;
 730                 }
 731                 if ((m->str_flags & STRING_COMPACT_OPTIONAL_BLANK) != 0) {
 732                         file_magwarn(ms, "'/%c' not allowed on regex\n",
 733                             CHAR_COMPACT_OPTIONAL_BLANK);
 734                         return -1;
 735                 }
 736                 break;
 737         default:
 738                 file_magwarn(ms, "coding error: m->type=%d\n",
 739                     m->type);
 740                 return -1;
 741         }
 742         return 0;
 743 }
 744
 745 private int
 746 get_op(char c)
 747 {
 748         switch (c) {
 749         case '&':
 750                 return FILE_OPAND;
 751         case '|':
 752                 return FILE_OPOR;
 753         case '^':
 754                 return FILE_OPXOR;
 755         case '+':
 756                 return FILE_OPADD;
 757         case '-':
 758                 return FILE_OPMINUS;
 759         case '*':
 760                 return FILE_OPMULTIPLY;
 761         case '/':
 762                 return FILE_OPDIVIDE;
 763         case '%':
 764                 return FILE_OPMODULO;
 765         default:
 766                 return -1;
 767         }
 768 }
 769
 770 #ifdef ENABLE_CONDITIONALS
 771 private int
 772 get_cond(const char *l, const char **t)
 773 {
 774         static struct cond_tbl_s {
 775                 const char *name;
 776                 const size_t len;
 777                 const int cond;
 778         } cond_tbl[] = {
 779                 { "if",         2,      COND_IF },
 780                 { "elif",       4,      COND_ELIF },
 781                 { "else",       4,      COND_ELSE },
 782                 { NULL,         0,      COND_NONE },
 783         };
 784         struct cond_tbl_s *p;
 785
 786         for (p = cond_tbl; p->name; p++) {
 787                 if (strncmp(l, p->name, p->len) == 0 &&
 788                     isspace((unsigned char)l[p->len])) {
 789                         if (t)
 790                                 *t = l + p->len;
 791                         break;
 792                 }
 793         }
 794         return p->cond;
 795 }
 796
 797 private int
 798 check_cond(struct magic_set *ms, int cond, uint32_t cont_level)
 799 {
 800         int last_cond;
 801         last_cond = ms->c.li[cont_level].last_cond;
 802
 803         switch (cond) {
 804         case COND_IF:
 805                 if (last_cond != COND_NONE && last_cond != COND_ELIF) {
 806                         if (ms->flags & MAGIC_CHECK)
 807                                 file_magwarn(ms, "syntax error: `if'");
 808                         return -1;
 809                 }
 810                 last_cond = COND_IF;
 811                 break;
 812
 813         case COND_ELIF:
 814                 if (last_cond != COND_IF && last_cond != COND_ELIF) {
 815                         if (ms->flags & MAGIC_CHECK)
 816                                 file_magwarn(ms, "syntax error: `elif'");
 817                         return -1;
 818                 }
 819                 last_cond = COND_ELIF;
 820                 break;
 821
 822         case COND_ELSE:
 823                 if (last_cond != COND_IF && last_cond != COND_ELIF) {
 824                         if (ms->flags & MAGIC_CHECK)
 825                                 file_magwarn(ms, "syntax error: `else'");
 826                         return -1;
 827                 }
 828                 last_cond = COND_NONE;
 829                 break;
 830
 831         case COND_NONE:
 832                 last_cond = COND_NONE;
 833                 break;
 834         }
 835
 836         ms->c.li[cont_level].last_cond = last_cond;
 837         return 0;
 838 }
 839 #endif /* ENABLE_CONDITIONALS */
 840
 841 /*
 842  * parse one line from magic file, put into magic[index++] if valid
 843  */
 844 private int
 845 parse(struct magic_set *ms, struct magic_entry **mentryp, uint32_t *nmentryp,
 846     const char *line, size_t lineno, int action)
 847 {
 848 #ifdef ENABLE_CONDITIONALS
 849         static uint32_t last_cont_level = 0;
 850 #endif
 851         size_t i;
 852         struct magic_entry *me;
 853         struct magic *m;
 854         const char *l = line;
 855         char *t;
 856         int op;
 857         uint32_t cont_level;
 858
 859         cont_level = 0;
 860
 861         while (*l == '>') {
 862                 ++l;            /* step over */
 863                 cont_level++;
 864         }
 865 #ifdef ENABLE_CONDITIONALS
 866         if (cont_level == 0 || cont_level > last_cont_level)
 867                 if (file_check_mem(ms, cont_level) == -1)
 868                         return -1;
 869         last_cont_level = cont_level;
 870 #endif
 871
 872 #define ALLOC_CHUNK     (size_t)10
 873 #define ALLOC_INCR      (size_t)200
 874
 875         if (cont_level != 0) {
 876                 if (*nmentryp == 0) {
 877                         file_error(ms, 0, "No current entry for continuation");
 878                         return -1;
 879                 }
 880                 me = &(*mentryp)[*nmentryp - 1];
 881                 if (me->cont_count == me->max_count) {
 882                         struct magic *nm;
 883                         size_t cnt = me->max_count + ALLOC_CHUNK;
 884                         if ((nm = realloc(me->mp, sizeof(*nm) * cnt)) == NULL) {
 885                                 file_oomem(ms, sizeof(*nm) * cnt);
 886                                 return -1;
 887                         }
 888                         me->mp = m = nm;
 889                         me->max_count = cnt;
 890                 }
 891                 m = &me->mp[me->cont_count++];
 892                 (void)memset(m, 0, sizeof(*m));
 893                 m->cont_level = cont_level;
 894         } else {
 895                 if (*nmentryp == maxmagic) {
 896                         struct magic_entry *mp;
 897
 898                         maxmagic += ALLOC_INCR;
 899                         if ((mp = realloc(*mentryp, sizeof(*mp) * maxmagic)) ==
 900                             NULL) {
 901                                 file_oomem(ms, sizeof(*mp) * maxmagic);
 902                                 return -1;
 903                         }
 904                         (void)memset(&mp[*nmentryp], 0, sizeof(*mp) *
 905                             ALLOC_INCR);
 906                         *mentryp = mp;
 907                 }
 908                 me = &(*mentryp)[*nmentryp];
 909                 if (me->mp == NULL) {
 910                         if ((m = malloc(sizeof(*m) * ALLOC_CHUNK)) == NULL) {
 911                                 file_oomem(ms, sizeof(*m) * ALLOC_CHUNK);
 912                                 return -1;
 913                         }
 914                         me->mp = m;
 915                         me->max_count = ALLOC_CHUNK;
 916                 } else
 917                         m = me->mp;
 918                 (void)memset(m, 0, sizeof(*m));
 919                 m->cont_level = 0;
 920                 me->cont_count = 1;
 921         }
 922         m->lineno = lineno;
 923
 924         if (*l == '&') {  /* m->cont_level == 0 checked below. */
 925                 ++l;            /* step over */
 926                 m->flag |= OFFADD;
 927         }
 928         if (*l == '(') {
 929                 ++l;            /* step over */
 930                 m->flag |= INDIR;
 931                 if (m->flag & OFFADD)
 932                         m->flag = (m->flag & ~OFFADD) | INDIROFFADD;
 933
 934                 if (*l == '&') {  /* m->cont_level == 0 checked below */
 935                         ++l;            /* step over */
 936                         m->flag |= OFFADD;
 937                 }
 938         }
 939         /* Indirect offsets are not valid at level 0. */
 940         if (m->cont_level == 0 && (m->flag & (OFFADD | INDIROFFADD)))
 941                 if (ms->flags & MAGIC_CHECK)
 942                         file_magwarn(ms, "relative offset at level 0");
 943
 944         /* get offset, then skip over it */
 945         m->offset = (uint32_t)strtoul(l, &t, 0);
 946         if (l == t)
 947                 if (ms->flags & MAGIC_CHECK)
 948                         file_magwarn(ms, "offset `%s' invalid", l);
 949         l = t;
 950
 951         if (m->flag & INDIR) {
 952                 m->in_type = FILE_LONG;
 953                 m->in_offset = 0;
 954                 /*
 955                  * read [.lbs][+-]nnnnn)
 956                  */
 957                 if (*l == '.') {
 958                         l++;
 959                         switch (*l) {
 960                         case 'l':
 961                                 m->in_type = FILE_LELONG;
 962                                 break;
 963                         case 'L':
 964                                 m->in_type = FILE_BELONG;
 965                                 break;
 966                         case 'm':
 967                                 m->in_type = FILE_MELONG;
 968                                 break;
 969                         case 'h':
 970                         case 's':
 971                                 m->in_type = FILE_LESHORT;
 972                                 break;
 973                         case 'H':
 974                         case 'S':
 975                                 m->in_type = FILE_BESHORT;
 976                                 break;
 977                         case 'c':
 978                         case 'b':
 979                         case 'C':
 980                         case 'B':
 981                                 m->in_type = FILE_BYTE;
 982                                 break;
 983                         case 'e':
 984                         case 'f':
 985                         case 'g':
 986                                 m->in_type = FILE_LEDOUBLE;
 987                                 break;
 988                         case 'E':
 989                         case 'F':
 990                         case 'G':
 991                                 m->in_type = FILE_BEDOUBLE;
 992                                 break;
 993                         default:
 994                                 if (ms->flags & MAGIC_CHECK)
 995                                         file_magwarn(ms,
 996                                             "indirect offset type `%c' invalid",
 997                                             *l);
 998                                 break;
 999                         }
1000                         l++;
1001                 }
1002
1003                 m->in_op = 0;
1004                 if (*l == '~') {
1005                         m->in_op |= FILE_OPINVERSE;
1006                         l++;
1007                 }
1008                 if ((op = get_op(*l)) != -1) {
1009                         m->in_op |= op;
1010                         l++;
1011                 }
1012                 if (*l == '(') {
1013                         m->in_op |= FILE_OPINDIRECT;
1014                         l++;
1015                 }
1016                 if (isdigit((unsigned char)*l) || *l == '-') {
1017                         m->in_offset = (int32_t)strtol(l, &t, 0);
1018                         if (l == t)
1019                                 if (ms->flags & MAGIC_CHECK)
1020                                         file_magwarn(ms,
1021                                             "in_offset `%s' invalid", l);
1022                         l = t;
1023                 }
1024                 if (*l++ != ')' ||
1025                     ((m->in_op & FILE_OPINDIRECT) && *l++ != ')'))
1026                         if (ms->flags & MAGIC_CHECK)
1027                                 file_magwarn(ms,
1028                                     "missing ')' in indirect offset");
1029         }
1030         EATAB;
1031
1032 #ifdef ENABLE_CONDITIONALS
1033         m->cond = get_cond(l, &l);
1034         if (check_cond(ms, m->cond, cont_level) == -1)
1035                 return -1;
1036
1037         EATAB;
1038 #endif
1039
1040         if (*l == 'u') {
1041                 ++l;
1042                 m->flag |= UNSIGNED;
1043         }
1044
1045         m->type = get_type(l, &l);
1046         if (m->type == FILE_INVALID) {
1047                 if (ms->flags & MAGIC_CHECK)
1048                         file_magwarn(ms, "type `%s' invalid", l);
1049                 return -1;
1050         }
1051
1052         /* New-style anding: "0 byte&0x80 =0x80 dynamically linked" */
1053         /* New and improved: ~ & | ^ + - * / % -- exciting, isn't it? */
1054
1055         m->mask_op = 0;
1056         if (*l == '~') {
1057                 if (!IS_STRING(m->type))
1058                         m->mask_op |= FILE_OPINVERSE;
1059                 else if (ms->flags & MAGIC_CHECK)
1060                         file_magwarn(ms, "'~' invalid for string types");
1061                 ++l;
1062         }
1063         m->str_count = 0;
1064         m->str_flags = 0;
1065         m->num_mask = 0;
1066         if ((op = get_op(*l)) != -1) {
1067                 if (!IS_STRING(m->type)) {
1068                         uint64_t val;
1069                         ++l;
1070                         m->mask_op |= op;
1071                         val = (uint64_t)strtoull(l, &t, 0);
1072                         l = t;
1073                         m->num_mask = file_signextend(ms, m, val);
1074                         eatsize(&l);
1075                 }
1076                 else if (op == FILE_OPDIVIDE) {
1077                         int have_count = 0;
1078                         while (!isspace((unsigned char)*++l)) {
1079                                 switch (*l) {
1080                                 /* for portability avoid "case '0' ... '9':" */
1081                                 case '0':  case '1':  case '2':
1082                                 case '3':  case '4':  case '5':
1083                                 case '6':  case '7':  case '8':
1084                                 case '9': {
1085                                         if (have_count && ms->flags & MAGIC_CHECK)
1086                                                 file_magwarn(ms,
1087                                                     "multiple counts");
1088                                         have_count = 1;
1089                                         m->str_count = strtoul(l, &t, 0);
1090                                         l = t - 1;
1091                                         break;
1092                                 }
1093                                 case CHAR_COMPACT_BLANK:
1094                                         m->str_flags |= STRING_COMPACT_BLANK;
1095                                         break;
1096                                 case CHAR_COMPACT_OPTIONAL_BLANK:
1097                                         m->str_flags |=
1098                                             STRING_COMPACT_OPTIONAL_BLANK;
1099                                         break;
1100                                 case CHAR_IGNORE_LOWERCASE:
1101                                         m->str_flags |= STRING_IGNORE_LOWERCASE;
1102                                         break;
1103                                 case CHAR_IGNORE_UPPERCASE:
1104                                         m->str_flags |= STRING_IGNORE_UPPERCASE;
1105                                         break;
1106                                 case CHAR_REGEX_OFFSET_START:
1107                                         m->str_flags |= REGEX_OFFSET_START;
1108                                         break;
1109                                 default:
1110                                         if (ms->flags & MAGIC_CHECK)
1111                                                 file_magwarn(ms,
1112                                                 "string extension `%c' invalid",
1113                                                 *l);
1114                                         return -1;
1115                                 }
1116                                 /* allow multiple '/' for readability */
1117                                 if (l[1] == '/' && !isspace((unsigned char)l[2]))
1118                                         l++;
1119                         }
1120                         if (string_modifier_check(ms, m) == -1)
1121                                 return -1;
1122                 }
1123                 else {
1124                         if (ms->flags & MAGIC_CHECK)
1125                                 file_magwarn(ms, "invalid string op: %c", *t);
1126                         return -1;
1127                 }
1128         }
1129         /*
1130          * We used to set mask to all 1's here, instead let's just not do
1131          * anything if mask = 0 (unless you have a better idea)
1132          */
1133         EATAB;
1134
1135         switch (*l) {
1136         case '>':
1137         case '<':
1138         /* Old-style anding: "0 byte &0x80 dynamically linked" */
1139         case '&':
1140         case '^':
1141         case '=':
1142                 m->reln = *l;
1143                 ++l;
1144                 if (*l == '=') {
1145                    /* HP compat: ignore &= etc. */
1146                    ++l;
1147                 }
1148                 break;
1149         case '!':
1150                 m->reln = *l;
1151                 ++l;
1152                 break;
1153         default:
1154                 m->reln = '=';  /* the default relation */
1155                 if (*l == 'x' && ((isascii((unsigned char)l[1]) &&
1156                     isspace((unsigned char)l[1])) || !l[1])) {
1157                         m->reln = *l;
1158                         ++l;
1159                 }
1160                 break;
1161         }
1162         /*
1163          * Grab the value part, except for an 'x' reln.
1164          */
1165         if (m->reln != 'x' && getvalue(ms, m, &l, action))
1166                 return -1;
1167
1168         /*
1169          * TODO finish this macro and start using it!
1170          * #define offsetcheck {if (offset > HOWMANY-1)
1171          *      magwarn("offset too big"); }
1172          */
1173
1174         /*
1175          * Now get last part - the description
1176          */
1177         EATAB;
1178         if (l[0] == '\b') {
1179                 ++l;
1180                 m->nospflag = 1;
1181         } else if ((l[0] == '\\') && (l[1] == 'b')) {
1182                 ++l;
1183                 ++l;
1184                 m->nospflag = 1;
1185         } else
1186                 m->nospflag = 0;
1187         for (i = 0; (m->desc[i++] = *l++) != '\0' && i < sizeof(m->desc); )
1188                 continue;
1189         if (i == sizeof(m->desc)) {
1190                 m->desc[sizeof(m->desc) - 1] = '\0';
1191                 if (ms->flags & MAGIC_CHECK)
1192                         file_magwarn(ms, "description `%s' truncated", m->desc);
1193         }
1194
1195         /*
1196          * We only do this check while compiling, or if any of the magic
1197          * files were not compiled.
1198          */
1199         if (ms->flags & MAGIC_CHECK) {
1200                 if (check_format(ms, m) == -1)
1201                         return -1;
1202         }
1203 #ifndef COMPILE_ONLY
1204         if (action == FILE_CHECK) {
1205                 file_mdump(m);
1206         }
1207 #endif
1208         if (m->cont_level == 0)
1209                 ++(*nmentryp);          /* make room for next */
1210         return 0;
1211 }
1212
1213 private int
1214 check_format_type(const char *ptr, int type)
1215 {
1216         int quad = 0;
1217         if (*ptr == '\0') {
1218                 /* Missing format string; bad */
1219                 return -1;
1220         }
1221
1222         switch (type) {
1223         case FILE_FMT_QUAD:
1224                 quad = 1;
1225                 /*FALLTHROUGH*/
1226         case FILE_FMT_NUM:
1227                 if (*ptr == '-')
1228                         ptr++;
1229                 if (*ptr == '.')
1230                         ptr++;
1231                 while (isdigit((unsigned char)*ptr)) ptr++;
1232                 if (*ptr == '.')
1233                         ptr++;
1234                 while (isdigit((unsigned char)*ptr)) ptr++;
1235                 if (quad) {
1236                         if (*ptr++ != 'l')
1237                                 return -1;
1238                         if (*ptr++ != 'l')
1239                                 return -1;
1240                 }
1241
1242                 switch (*ptr++) {
1243                 case 'l':
1244                         switch (*ptr++) {
1245                         case 'i':
1246                         case 'd':
1247                         case 'u':
1248                         case 'x':
1249                         case 'X':
1250                                 return 0;
1251                         default:
1252                                 return -1;
1253                         }
1254
1255                 case 'h':
1256                         switch (*ptr++) {
1257                         case 'h':
1258                                 switch (*ptr++) {
1259                                 case 'i':
1260                                 case 'd':
1261                                 case 'u':
1262                                 case 'x':
1263                                 case 'X':
1264                                         return 0;
1265                                 default:
1266                                         return -1;
1267                                 }
1268                         case 'd':
1269                                 return 0;
1270                         default:
1271                                 return -1;
1272                         }
1273
1274                 case 'i':
1275                 case 'c':
1276                 case 'd':
1277                 case 'u':
1278                 case 'x':
1279                 case 'X':
1280                         return 0;
1281
1282                 default:
1283                         return -1;
1284                 }
1285
1286         case FILE_FMT_FLOAT:
1287         case FILE_FMT_DOUBLE:
1288                 if (*ptr == '-')
1289                         ptr++;
1290                 if (*ptr == '.')
1291                         ptr++;
1292                 while (isdigit((unsigned char)*ptr)) ptr++;
1293                 if (*ptr == '.')
1294                         ptr++;
1295                 while (isdigit((unsigned char)*ptr)) ptr++;
1296
1297                 switch (*ptr++) {
1298                 case 'e':
1299                 case 'E':
1300                 case 'f':
1301                 case 'F':
1302                 case 'g':
1303                 case 'G':
1304                         return 0;
1305
1306                 default:
1307                         return -1;
1308                 }
1309
1310
1311         case FILE_FMT_STR:
1312                 if (*ptr == '-')
1313                         ptr++;
1314                 while (isdigit((unsigned char )*ptr))
1315                         ptr++;
1316                 if (*ptr == '.') {
1317                         ptr++;
1318                         while (isdigit((unsigned char )*ptr))
1319                                 ptr++;
1320                 }
1321
1322                 switch (*ptr++) {
1323                 case 's':
1324                         return 0;
1325                 default:
1326                         return -1;
1327                 }
1328
1329         default:
1330                 /* internal error */
1331                 abort();
1332         }
1333         /*NOTREACHED*/
1334         return -1;
1335 }
1336
1337 /*
1338  * Check that the optional printf format in description matches
1339  * the type of the magic.
1340  */
1341 private int
1342 check_format(struct magic_set *ms, struct magic *m)
1343 {
1344         char *ptr;
1345
1346         for (ptr = m->desc; *ptr; ptr++)
1347                 if (*ptr == '%')
1348                         break;
1349         if (*ptr == '\0') {
1350                 /* No format string; ok */
1351                 return 1;
1352         }
1353
1354         assert(file_nformats == file_nnames);
1355
1356         if (m->type >= file_nformats) {
1357                 file_error(ms, 0, "Internal error inconsistency between "
1358                     "m->type and format strings");
1359                 return -1;
1360         }
1361         if (file_formats[m->type] == FILE_FMT_NONE) {
1362                 file_error(ms, 0, "No format string for `%s' with description "
1363                     "`%s'", m->desc, file_names[m->type]);
1364                 return -1;
1365         }
1366
1367         ptr++;
1368         if (check_format_type(ptr, file_formats[m->type]) == -1) {
1369                 /*
1370                  * TODO: this error message is unhelpful if the format
1371                  * string is not one character long
1372                  */
1373                 file_error(ms, 0, "Printf format `%c' is not valid for type "
1374                     " `%s' in description `%s'", *ptr,
1375                     file_names[m->type], m->desc);
1376                 return -1;
1377         }
1378
1379         for (; *ptr; ptr++) {
1380                 if (*ptr == '%') {
1381                         file_error(ms, 0,
1382                             "Too many format strings (should have at most one) "
1383                             "for `%s' with description `%s'",
1384                             file_names[m->type], m->desc);
1385                         return -1;
1386                 }
1387         }
1388         return 0;
1389 }
1390
1391 /*
1392  * Read a numeric value from a pointer, into the value union of a magic
1393  * pointer, according to the magic type.  Update the string pointer to point
1394  * just after the number read.  Return 0 for success, non-zero for failure.
1395  */
1396 private int
1397 getvalue(struct magic_set *ms, struct magic *m, const char **p, int action)
1398 {
1399         int slen;
1400
1401         switch (m->type) {
1402         case FILE_BESTRING16:
1403         case FILE_LESTRING16:
1404         case FILE_STRING:
1405         case FILE_PSTRING:
1406         case FILE_REGEX:
1407         case FILE_SEARCH:
1408                 *p = getstr(ms, *p, m->value.s, sizeof(m->value.s), &slen, action);
1409                 if (*p == NULL) {
1410                         if (ms->flags & MAGIC_CHECK)
1411                                 file_magwarn(ms, "cannot get string from `%s'",
1412                                     m->value.s);
1413                         return -1;
1414                 }
1415                 m->vallen = slen;
1416                 return 0;
1417         case FILE_FLOAT:
1418         case FILE_BEFLOAT:
1419         case FILE_LEFLOAT:
1420                 if (m->reln != 'x') {
1421                         char *ep;
1422 #ifdef HAVE_STRTOF
1423                         m->value.f = strtof(*p, &ep);
1424 #else
1425                         m->value.f = (float)strtod(*p, &ep);
1426 #endif
1427                         *p = ep;
1428                 }
1429                 return 0;
1430         case FILE_DOUBLE:
1431         case FILE_BEDOUBLE:
1432         case FILE_LEDOUBLE:
1433                 if (m->reln != 'x') {
1434                         char *ep;
1435                         m->value.d = strtod(*p, &ep);
1436                         *p = ep;
1437                 }
1438                 return 0;
1439         default:
1440                 if (m->reln != 'x') {
1441                         char *ep;
1442                         m->value.q = file_signextend(ms, m,
1443                             (uint64_t)strtoull(*p, &ep, 0));
1444                         *p = ep;
1445                         eatsize(p);
1446                 }
1447                 return 0;
1448         }
1449 }
1450
1451 /*
1452  * Convert a string containing C character escapes.  Stop at an unescaped
1453  * space or tab.
1454  * Copy the converted version to "p", returning its length in *slen.
1455  * Return updated scan pointer as function result.
1456  */
1457 private const char *
1458 getstr(struct magic_set *ms, const char *s, char *p, int plen, int *slen, int action)
1459 {
1460         const char *origs = s;
1461         char    *origp = p;
1462         char    *pmax = p + plen - 1;
1463         int     c;
1464         int     val;
1465
1466         while ((c = *s++) != '\0') {
1467                 if (isspace((unsigned char) c))
1468                         break;
1469                 if (p >= pmax) {
1470                         file_error(ms, 0, "string too long: `%s'", origs);
1471                         return NULL;
1472                 }
1473                 if (c == '\\') {
1474                         switch(c = *s++) {
1475
1476                         case '\0':
1477                                 if (action == FILE_COMPILE)
1478                                         file_magwarn(ms, "incomplete escape");
1479                                 goto out;
1480
1481                         case '\t':
1482                                 if (action == FILE_COMPILE) {
1483                                         file_magwarn(ms,
1484                                             "escaped tab found, use \\t instead");
1485                                         action++;
1486                                 }
1487                                 /*FALLTHROUGH*/
1488                         default:
1489                                 if (action == FILE_COMPILE) {
1490                                         if (isprint((unsigned char)c))
1491                                             file_magwarn(ms,
1492                                                 "no need to escape `%c'", c);
1493                                         else
1494                                             file_magwarn(ms,
1495                                                 "unknown escape sequence: \\%03o", c);
1496                                 }
1497                                 /*FALLTHROUGH*/
1498                         /* space, perhaps force people to use \040? */
1499                         case ' ':
1500 #if 0
1501                         /*
1502                          * Other things people escape, but shouldn't need to,
1503                          * so we disallow them
1504                          */
1505                         case '\'':
1506                         case '"':
1507                         case '?':
1508 #endif
1509                         /* Relations */
1510                         case '>':
1511                         case '<':
1512                         case '&':
1513                         case '^':
1514                         case '=':
1515                         case '!':
1516                         /* and baskslash itself */
1517                         case '\\':
1518                                 *p++ = (char) c;
1519                                 break;
1520
1521                         case 'a':
1522                                 *p++ = '\a';
1523                                 break;
1524
1525                         case 'b':
1526                                 *p++ = '\b';
1527                                 break;
1528
1529                         case 'f':
1530                                 *p++ = '\f';
1531                                 break;
1532
1533                         case 'n':
1534                                 *p++ = '\n';
1535                                 break;
1536
1537                         case 'r':
1538                                 *p++ = '\r';
1539                                 break;
1540
1541                         case 't':
1542                                 *p++ = '\t';
1543                                 break;
1544
1545                         case 'v':
1546                                 *p++ = '\v';
1547                                 break;
1548
1549                         /* \ and up to 3 octal digits */
1550                         case '0':
1551                         case '1':
1552                         case '2':
1553                         case '3':
1554                         case '4':
1555                         case '5':
1556                         case '6':
1557                         case '7':
1558                                 val = c - '0';
1559                                 c = *s++;  /* try for 2 */
1560                                 if (c >= '0' && c <= '7') {
1561                                         val = (val << 3) | (c - '0');
1562                                         c = *s++;  /* try for 3 */
1563                                         if (c >= '0' && c <= '7')
1564                                                 val = (val << 3) | (c-'0');
1565                                         else
1566                                                 --s;
1567                                 }
1568                                 else
1569                                         --s;
1570                                 *p++ = (char)val;
1571                                 break;
1572
1573                         /* \x and up to 2 hex digits */
1574                         case 'x':
1575                                 val = 'x';      /* Default if no digits */
1576                                 c = hextoint(*s++);     /* Get next char */
1577                                 if (c >= 0) {
1578                                         val = c;
1579                                         c = hextoint(*s++);
1580                                         if (c >= 0)
1581                                                 val = (val << 4) + c;
1582                                         else
1583                                                 --s;
1584                                 } else
1585                                         --s;
1586                                 *p++ = (char)val;
1587                                 break;
1588                         }
1589                 } else
1590                         *p++ = (char)c;
1591         }
1592 out:
1593         *p = '\0';
1594         *slen = p - origp;
1595         return s;
1596 }
1597
1598
1599 /* Single hex char to int; -1 if not a hex char. */
1600 private int
1601 hextoint(int c)
1602 {
1603         if (!isascii((unsigned char) c))
1604                 return -1;
1605         if (isdigit((unsigned char) c))
1606                 return c - '0';
1607         if ((c >= 'a') && (c <= 'f'))
1608                 return c + 10 - 'a';
1609         if (( c>= 'A') && (c <= 'F'))
1610                 return c + 10 - 'A';
1611         return -1;
1612 }
1613
1614
1615 /*
1616  * Print a string containing C character escapes.
1617  */
1618 protected void
1619 file_showstr(FILE *fp, const char *s, size_t len)
1620 {
1621         char    c;
1622
1623         for (;;) {
1624                 c = *s++;
1625                 if (len == ~0U) {
1626                         if (c == '\0')
1627                                 break;
1628                 }
1629                 else  {
1630                         if (len-- == 0)
1631                                 break;
1632                 }
1633                 if (c >= 040 && c <= 0176)      /* TODO isprint && !iscntrl */
1634                         (void) fputc(c, fp);
1635                 else {
1636                         (void) fputc('\\', fp);
1637                         switch (c) {
1638                         case '\a':
1639                                 (void) fputc('a', fp);
1640                                 break;
1641
1642                         case '\b':
1643                                 (void) fputc('b', fp);
1644                                 break;
1645
1646                         case '\f':
1647                                 (void) fputc('f', fp);
1648                                 break;
1649
1650                         case '\n':
1651                                 (void) fputc('n', fp);
1652                                 break;
1653
1654                         case '\r':
1655                                 (void) fputc('r', fp);
1656                                 break;
1657
1658                         case '\t':
1659                                 (void) fputc('t', fp);
1660                                 break;
1661
1662                         case '\v':
1663                                 (void) fputc('v', fp);
1664                                 break;
1665
1666                         default:
1667                                 (void) fprintf(fp, "%.3o", c & 0377);
1668                                 break;
1669                         }
1670                 }
1671         }
1672 }
1673
1674 /*
1675  * eatsize(): Eat the size spec from a number [eg. 10UL]
1676  */
1677 private void
1678 eatsize(const char **p)
1679 {
1680         const char *l = *p;
1681
1682         if (LOWCASE(*l) == 'u')
1683                 l++;
1684
1685         switch (LOWCASE(*l)) {
1686         case 'l':    /* long */
1687         case 's':    /* short */
1688         case 'h':    /* short */
1689         case 'b':    /* char/byte */
1690         case 'c':    /* char/byte */
1691                 l++;
1692                 /*FALLTHROUGH*/
1693         default:
1694                 break;
1695         }
1696
1697         *p = l;
1698 }
1699
1700 /*
1701  * handle a compiled file.
1702  */
1703 private int
1704 apprentice_map(struct magic_set *ms, struct magic **magicp, uint32_t *nmagicp,
1705     const char *fn)
1706 {
1707         int fd;
1708         struct stat st;
1709         uint32_t *ptr;
1710         uint32_t version;
1711         int needsbyteswap;
1712         char buf[MAXPATHLEN];
1713         char *dbname = mkdbname(fn, buf, sizeof(buf), 0);
1714         void *mm = NULL;
1715
1716         if (dbname == NULL)
1717                 return -1;
1718
1719         if ((fd = open(dbname, O_RDONLY|O_BINARY)) == -1)
1720                 return -1;
1721
1722         if (fstat(fd, &st) == -1) {
1723                 file_error(ms, errno, "cannot stat `%s'", dbname);
1724                 goto error;
1725         }
1726         if (st.st_size < 16) {
1727                 file_error(ms, 0, "file `%s' is too small", dbname);
1728                 goto error;
1729         }
1730
1731 #ifdef QUICK
1732         if ((mm = mmap(0, (size_t)st.st_size, PROT_READ|PROT_WRITE,
1733             MAP_PRIVATE|MAP_FILE, fd, (off_t)0)) == MAP_FAILED) {
1734                 file_error(ms, errno, "cannot map `%s'", dbname);
1735                 goto error;
1736         }
1737 #define RET     2
1738 #else
1739         if ((mm = malloc((size_t)st.st_size)) == NULL) {
1740                 file_oomem(ms, (size_t)st.st_size);
1741                 goto error;
1742         }
1743         if (read(fd, mm, (size_t)st.st_size) != (size_t)st.st_size) {
1744                 file_badread(ms);
1745                 goto error;
1746         }
1747 #define RET     1
1748 #endif
1749         *magicp = mm;
1750         (void)close(fd);
1751         fd = -1;
1752         ptr = (uint32_t *)(void *)*magicp;
1753         if (*ptr != MAGICNO) {
1754                 if (swap4(*ptr) != MAGICNO) {
1755                         file_error(ms, 0, "bad magic in `%s'");
1756                         goto error;
1757                 }
1758                 needsbyteswap = 1;
1759         } else
1760                 needsbyteswap = 0;
1761         if (needsbyteswap)
1762                 version = swap4(ptr[1]);
1763         else
1764                 version = ptr[1];
1765         if (version != VERSIONNO) {
1766                 file_error(ms, 0, "File %d.%d supports only %d version magic "
1767                     "files. `%s' is version %d", FILE_VERSION_MAJOR, patchlevel,
1768                     VERSIONNO, dbname, version);
1769                 goto error;
1770         }
1771         *nmagicp = (uint32_t)(st.st_size / sizeof(struct magic)) - 1;
1772         (*magicp)++;
1773         if (needsbyteswap)
1774                 byteswap(*magicp, *nmagicp);
1775         return RET;
1776
1777 error:
1778         if (fd != -1)
1779                 (void)close(fd);
1780         if (mm) {
1781 #ifdef QUICK
1782                 (void)munmap((void *)mm, (size_t)st.st_size);
1783 #else
1784                 free(mm);
1785 #endif
1786         } else {
1787                 *magicp = NULL;
1788                 *nmagicp = 0;
1789         }
1790         return -1;
1791 }
1792
1793 private const uint32_t ar[] = {
1794     MAGICNO, VERSIONNO
1795 };
1796 /*
1797  * handle an mmaped file.
1798  */
1799 private int
1800 apprentice_compile(struct magic_set *ms, struct magic **magicp,
1801     uint32_t *nmagicp, const char *fn)
1802 {
1803         int fd;
1804         char buf[MAXPATHLEN];
1805         char *dbname = mkdbname(fn, buf, sizeof(buf), 1);
1806
1807         if (dbname == NULL)
1808                 return -1;
1809
1810         if ((fd = open(dbname, O_WRONLY|O_CREAT|O_TRUNC|O_BINARY, 0644)) == -1) {
1811                 file_error(ms, errno, "cannot open `%s'", dbname);
1812                 return -1;
1813         }
1814
1815         if (write(fd, ar, sizeof(ar)) != (ssize_t)sizeof(ar)) {
1816                 file_error(ms, errno, "error writing `%s'", dbname);
1817                 return -1;
1818         }
1819
1820         if (lseek(fd, (off_t)sizeof(struct magic), SEEK_SET)
1821             != sizeof(struct magic)) {
1822                 file_error(ms, errno, "error seeking `%s'", dbname);
1823                 return -1;
1824         }
1825
1826         if (write(fd, *magicp, (sizeof(struct magic) * *nmagicp))
1827             != (ssize_t)(sizeof(struct magic) * *nmagicp)) {
1828                 file_error(ms, errno, "error writing `%s'", dbname);
1829                 return -1;
1830         }
1831
1832         (void)close(fd);
1833         return 0;
1834 }
1835
1836 private const char ext[] = ".mgc";
1837 /*
1838  * make a dbname
1839  */
1840 private char *
1841 mkdbname(const char *fn, char *buf, size_t bufsiz, int strip)
1842 {
1843         if (strip) {
1844                 const char *p;
1845                 if ((p = strrchr(fn, '/')) != NULL)
1846                         fn = ++p;
1847         }
1848
1849         (void)snprintf(buf, bufsiz, "%s%s", fn, ext);
1850         return buf;
1851 }
1852
1853 /*
1854  * Byteswap an mmap'ed file if needed
1855  */
1856 private void
1857 byteswap(struct magic *magic, uint32_t nmagic)
1858 {
1859         uint32_t i;
1860         for (i = 0; i < nmagic; i++)
1861                 bs1(&magic[i]);
1862 }
1863
1864 /*
1865  * swap a short
1866  */
1867 private uint16_t
1868 swap2(uint16_t sv)
1869 {
1870         uint16_t rv;
1871         uint8_t *s = (uint8_t *)(void *)&sv;
1872         uint8_t *d = (uint8_t *)(void *)&rv;
1873         d[0] = s[1];
1874         d[1] = s[0];
1875         return rv;
1876 }
1877
1878 /*
1879  * swap an int
1880  */
1881 private uint32_t
1882 swap4(uint32_t sv)
1883 {
1884         uint32_t rv;
1885         uint8_t *s = (uint8_t *)(void *)&sv;
1886         uint8_t *d = (uint8_t *)(void *)&rv;
1887         d[0] = s[3];
1888         d[1] = s[2];
1889         d[2] = s[1];
1890         d[3] = s[0];
1891         return rv;
1892 }
1893
1894 /*
1895  * swap a quad
1896  */
1897 private uint64_t
1898 swap8(uint64_t sv)
1899 {
1900         uint32_t rv;
1901         uint8_t *s = (uint8_t *)(void *)&sv;
1902         uint8_t *d = (uint8_t *)(void *)&rv;
1903         d[0] = s[3];
1904         d[1] = s[2];
1905         d[2] = s[1];
1906         d[3] = s[0];
1907         d[4] = s[7];
1908         d[5] = s[6];
1909         d[6] = s[5];
1910         d[7] = s[4];
1911         return rv;
1912 }
1913
1914 /*
1915  * byteswap a single magic entry
1916  */
1917 private void
1918 bs1(struct magic *m)
1919 {
1920         m->cont_level = swap2(m->cont_level);
1921         m->offset = swap4((uint32_t)m->offset);
1922         m->in_offset = swap4((uint32_t)m->in_offset);
1923         m->lineno = swap4((uint32_t)m->lineno);
1924         if (IS_STRING(m->type)) {
1925                 m->str_count = swap4(m->str_count);
1926                 m->str_flags = swap4(m->str_flags);
1927         }
1928         else {
1929                 m->value.q = swap8(m->value.q);
1930                 m->num_mask = swap8(m->num_mask);
1931         }
1932 }