![[BACK]](/icons/back.gif){kind=icon}
Return to apprentice.c CVS log ![[TXT]](/icons/text.gif){kind=icon}
![[DIR]](/icons/dir.gif){kind=icon}
Up to [local] / src / usr.bin / file|
Return to apprentice.c CVS log | Up to [local] / src / usr.bin / file |
1.25 ! chl 1: /* $OpenBSD: apprentice.c,v 1.24 2007/10/05 14:29:46 chl Exp $ */
1.1 deraadt 2: /*
1.16 ian 3: * Copyright (c) Ian F. Darwin 1986-1995.
4: * Software written by Ian F. Darwin and others;
5: * maintained 1995-present by Christos Zoulas and others.
6: *
7: * Redistribution and use in source and binary forms, with or without
8: * modification, are permitted provided that the following conditions
9: * are met:
10: * 1. Redistributions of source code must retain the above copyright
11: * notice immediately at the beginning of the file, without modification,
12: * this list of conditions, and the following disclaimer.
13: * 2. Redistributions in binary form must reproduce the above copyright
14: * notice, this list of conditions and the following disclaimer in the
15: * documentation and/or other materials provided with the distribution.
16: *
17: * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
18: * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19: * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20: * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE FOR
21: * ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22: * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23: * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24: * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25: * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26: * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27: * SUCH DAMAGE.
1.1 deraadt 28: */
1.19 tedu 29: /*
30: * apprentice - make one pass through /etc/magic, learning its secrets.
31: */
1.1 deraadt 32:
1.19 tedu 33: #include "file.h"
34: #include "magic.h"
1.1 deraadt 35: #include <stdlib.h>
1.19 tedu 36: #ifdef HAVE_UNISTD_H
37: #include <unistd.h>
38: #endif
1.1 deraadt 39: #include <string.h>
1.25 ! chl 40: #include <assert.h>
1.1 deraadt 41: #include <ctype.h>
1.19 tedu 42: #include <fcntl.h>
43: #include <sys/stat.h>
44: #include <sys/param.h>
45: #ifdef QUICK
46: #include <sys/mman.h>
47: #endif
1.1 deraadt 48:
49: #ifndef lint
1.25 ! chl 50: FILE_RCSID("@(#)$Id: apprentice.c,v 1.24 2007/10/05 14:29:46 chl Exp $")
1.1 deraadt 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))
1.19 tedu 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:
1.25 ! chl 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);
1.19 tedu 91: private int hextoint(int);
1.25 ! chl 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 **);
1.19 tedu 97: private int apprentice_1(struct magic_set *, const char *, int, struct mlist *);
1.25 ! chl 98: private size_t apprentice_magic_strength(const struct magic *);
! 99: private int apprentice_sort(const void *, const void *);
1.19 tedu 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);
1.25 ! chl 106: private uint64_t swap8(uint64_t);
! 107: private char *mkdbname(const char *, char *, size_t, int);
1.19 tedu 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 *);
1.25 ! chl 112: private int check_format_type(const char *, int);
! 113: private int check_format(struct magic_set *, struct magic *);
1.19 tedu 114:
115: private size_t maxmagic = 0;
116: private size_t magicsize = sizeof(struct magic);
117:
1.25 ! chl 118:
1.19 tedu 119: #ifdef COMPILE_ONLY
120:
121: int main(int, char *[]);
122:
123: int
124: main(int argc, char *argv[])
125: {
126: int ret;
1.25 ! chl 127: struct magic_set *ms;
! 128: char *progname;
1.19 tedu 129:
130: if ((progname = strrchr(argv[0], '/')) != NULL)
131: progname++;
132: else
133: progname = argv[0];
134:
135: if (argc != 2) {
1.25 ! chl 136: (void)fprintf(stderr, "Usage: %s file\n", progname);
! 137: return 1;
1.19 tedu 138: }
139:
1.25 ! chl 140: if ((ms = magic_open(MAGIC_CHECK)) == NULL) {
! 141: (void)fprintf(stderr, "%s: %s\n", progname, strerror(errno));
! 142: return 1;
! 143: }
! 144: ret = magic_compile(ms, argv[1]) == -1 ? 1 : 0;
! 145: if (ret == 1)
! 146: (void)fprintf(stderr, "%s: %s\n", progname, magic_error(ms));
! 147: magic_close(ms);
! 148: return ret;
1.19 tedu 149: }
150: #endif /* COMPILE_ONLY */
151:
1.25 ! chl 152: static const struct type_tbl_s {
! 153: const char *name;
! 154: const size_t len;
! 155: const int type;
! 156: const int format;
! 157: } type_tbl[] = {
! 158: # define XX(s) s, (sizeof(s) - 1)
! 159: # define XX_NULL NULL, 0
! 160: { XX("byte"), FILE_BYTE, FILE_FMT_NUM },
! 161: { XX("short"), FILE_SHORT, FILE_FMT_NUM },
! 162: { XX("default"), FILE_DEFAULT, FILE_FMT_STR },
! 163: { XX("long"), FILE_LONG, FILE_FMT_NUM },
! 164: { XX("string"), FILE_STRING, FILE_FMT_STR },
! 165: { XX("date"), FILE_DATE, FILE_FMT_STR },
! 166: { XX("beshort"), FILE_BESHORT, FILE_FMT_NUM },
! 167: { XX("belong"), FILE_BELONG, FILE_FMT_NUM },
! 168: { XX("bedate"), FILE_BEDATE, FILE_FMT_STR },
! 169: { XX("leshort"), FILE_LESHORT, FILE_FMT_NUM },
! 170: { XX("lelong"), FILE_LELONG, FILE_FMT_NUM },
! 171: { XX("ledate"), FILE_LEDATE, FILE_FMT_STR },
! 172: { XX("pstring"), FILE_PSTRING, FILE_FMT_STR },
! 173: { XX("ldate"), FILE_LDATE, FILE_FMT_STR },
! 174: { XX("beldate"), FILE_BELDATE, FILE_FMT_STR },
! 175: { XX("leldate"), FILE_LELDATE, FILE_FMT_STR },
! 176: { XX("regex"), FILE_REGEX, FILE_FMT_STR },
! 177: { XX("bestring16"), FILE_BESTRING16, FILE_FMT_STR },
! 178: { XX("lestring16"), FILE_LESTRING16, FILE_FMT_STR },
! 179: { XX("search"), FILE_SEARCH, FILE_FMT_STR },
! 180: { XX("medate"), FILE_MEDATE, FILE_FMT_STR },
! 181: { XX("meldate"), FILE_MELDATE, FILE_FMT_STR },
! 182: { XX("melong"), FILE_MELONG, FILE_FMT_NUM },
! 183: { XX("quad"), FILE_QUAD, FILE_FMT_QUAD },
! 184: { XX("lequad"), FILE_LEQUAD, FILE_FMT_QUAD },
! 185: { XX("bequad"), FILE_BEQUAD, FILE_FMT_QUAD },
! 186: { XX("qdate"), FILE_QDATE, FILE_FMT_STR },
! 187: { XX("leqdate"), FILE_LEQDATE, FILE_FMT_STR },
! 188: { XX("beqdate"), FILE_BEQDATE, FILE_FMT_STR },
! 189: { XX("qldate"), FILE_QLDATE, FILE_FMT_STR },
! 190: { XX("leqldate"), FILE_LEQLDATE, FILE_FMT_STR },
! 191: { XX("beqldate"), FILE_BEQLDATE, FILE_FMT_STR },
! 192: { XX_NULL, FILE_INVALID, FILE_FMT_NONE },
! 193: # undef XX
! 194: # undef XX_NULL
! 195: };
! 196:
! 197: private int
! 198: get_type(const char *l, const char **t)
! 199: {
! 200: const struct type_tbl_s *p;
! 201:
! 202: for (p = type_tbl; p->name; p++) {
! 203: if (strncmp(l, p->name, p->len) == 0) {
! 204: if (t)
! 205: *t = l + p->len;
! 206: break;
! 207: }
! 208: }
! 209: return p->type;
! 210: }
! 211:
! 212: private void
! 213: init_file_tables(void)
! 214: {
! 215: static int done = 0;
! 216: const struct type_tbl_s *p;
! 217:
! 218: if (done)
! 219: return;
! 220: done++;
! 221:
! 222: for (p = type_tbl; p->name; p++) {
! 223: assert(p->type < FILE_NAMES_SIZE);
! 224: file_names[p->type] = p->name;
! 225: file_formats[p->type] = p->format;
! 226: }
! 227: }
1.19 tedu 228:
229: /*
230: * Handle one file.
231: */
232: private int
233: apprentice_1(struct magic_set *ms, const char *fn, int action,
234: struct mlist *mlist)
235: {
236: struct magic *magic = NULL;
237: uint32_t nmagic = 0;
238: struct mlist *ml;
239: int rv = -1;
240: int mapped;
241:
242: if (magicsize != FILE_MAGICSIZE) {
243: file_error(ms, 0, "magic element size %lu != %lu",
244: (unsigned long)sizeof(*magic),
245: (unsigned long)FILE_MAGICSIZE);
246: return -1;
247: }
1.1 deraadt 248:
1.19 tedu 249: if (action == FILE_COMPILE) {
250: rv = apprentice_file(ms, &magic, &nmagic, fn, action);
251: if (rv != 0)
252: return -1;
253: rv = apprentice_compile(ms, &magic, &nmagic, fn);
254: free(magic);
255: return rv;
256: }
1.25 ! chl 257:
1.19 tedu 258: #ifndef COMPILE_ONLY
259: if ((rv = apprentice_map(ms, &magic, &nmagic, fn)) == -1) {
260: if (ms->flags & MAGIC_CHECK)
1.25 ! chl 261: file_magwarn(ms, "using regular magic file `%s'", fn);
1.19 tedu 262: rv = apprentice_file(ms, &magic, &nmagic, fn, action);
263: if (rv != 0)
264: return -1;
265: }
1.1 deraadt 266:
1.19 tedu 267: mapped = rv;
268:
269: if (magic == NULL || nmagic == 0) {
270: file_delmagic(magic, mapped, nmagic);
271: return -1;
272: }
1.1 deraadt 273:
1.19 tedu 274: if ((ml = malloc(sizeof(*ml))) == NULL) {
275: file_delmagic(magic, mapped, nmagic);
1.25 ! chl 276: file_oomem(ms, sizeof(*ml));
1.19 tedu 277: return -1;
278: }
1.1 deraadt 279:
1.19 tedu 280: ml->magic = magic;
281: ml->nmagic = nmagic;
282: ml->mapped = mapped;
283:
284: mlist->prev->next = ml;
285: ml->prev = mlist->prev;
286: ml->next = mlist;
287: mlist->prev = ml;
1.1 deraadt 288:
1.19 tedu 289: return 0;
290: #endif /* COMPILE_ONLY */
291: }
292:
293: protected void
294: file_delmagic(struct magic *p, int type, size_t entries)
295: {
296: if (p == NULL)
297: return;
298: switch (type) {
1.25 ! chl 299: #ifdef QUICK
1.19 tedu 300: case 2:
301: p--;
302: (void)munmap((void *)p, sizeof(*p) * (entries + 1));
303: break;
1.25 ! chl 304: #endif
1.19 tedu 305: case 1:
306: p--;
1.25 ! chl 307: /*FALLTHROUGH*/
1.19 tedu 308: case 0:
309: free(p);
310: break;
311: default:
312: abort();
313: }
314: }
315:
316: /* const char *fn: list of magic files */
317: protected struct mlist *
318: file_apprentice(struct magic_set *ms, const char *fn, int action)
1.2 deraadt 319: {
1.19 tedu 320: char *p, *mfn, *afn = NULL;
1.2 deraadt 321: int file_err, errs = -1;
1.19 tedu 322: struct mlist *mlist;
1.25 ! chl 323: static const char mime[] = ".mime";
! 324:
! 325: init_file_tables();
1.2 deraadt 326:
1.19 tedu 327: if (fn == NULL)
328: fn = getenv("MAGIC");
329: if (fn == NULL)
330: fn = MAGIC;
331:
1.25 ! chl 332: if ((mfn = strdup(fn)) == NULL) {
! 333: file_oomem(ms, strlen(fn));
1.19 tedu 334: return NULL;
335: }
1.25 ! chl 336: fn = mfn;
1.19 tedu 337:
338: if ((mlist = malloc(sizeof(*mlist))) == NULL) {
339: free(mfn);
1.25 ! chl 340: file_oomem(ms, sizeof(*mlist));
1.19 tedu 341: return NULL;
1.2 deraadt 342: }
1.19 tedu 343: mlist->next = mlist->prev = mlist;
1.17 deraadt 344:
1.2 deraadt 345: while (fn) {
1.19 tedu 346: p = strchr(fn, PATHSEP);
1.2 deraadt 347: if (p)
348: *p++ = '\0';
1.19 tedu 349: if (*fn == '\0')
350: break;
351: if (ms->flags & MAGIC_MIME) {
1.25 ! chl 352: size_t len = strlen(fn) + sizeof(mime);
1.19 tedu 353: if ((afn = malloc(len)) == NULL) {
354: free(mfn);
355: free(mlist);
1.25 ! chl 356: file_oomem(ms, len);
1.19 tedu 357: return NULL;
358: }
359: (void)strlcpy(afn, fn, len);
360: (void)strlcat(afn, ".mime", len);
361: fn = afn;
362: }
363: file_err = apprentice_1(ms, fn, action, mlist);
1.2 deraadt 364: if (file_err > errs)
365: errs = file_err;
1.19 tedu 366: if (afn) {
367: free(afn);
368: afn = NULL;
369: }
1.2 deraadt 370: fn = p;
371: }
1.19 tedu 372: if (errs == -1) {
373: free(mfn);
374: free(mlist);
375: mlist = NULL;
376: file_error(ms, 0, "could not find any magic files!");
377: return NULL;
378: }
1.2 deraadt 379: free(mfn);
1.19 tedu 380: return mlist;
1.2 deraadt 381: }
382:
1.19 tedu 383: /*
1.25 ! chl 384: * Get weight of this magic entry, for sorting purposes.
! 385: */
! 386: private size_t
! 387: apprentice_magic_strength(const struct magic *m)
! 388: {
! 389: #define MULT 10
! 390: size_t val = 2 * MULT; /* baseline strength */
! 391:
! 392: switch (m->type) {
! 393: case FILE_DEFAULT: /* make sure this sorts last */
! 394: return 0;
! 395:
! 396: case FILE_BYTE:
! 397: val += 1 * MULT;
! 398: break;
! 399:
! 400: case FILE_SHORT:
! 401: case FILE_LESHORT:
! 402: case FILE_BESHORT:
! 403: val += 2 * MULT;
! 404: break;
! 405:
! 406: case FILE_LONG:
! 407: case FILE_LELONG:
! 408: case FILE_BELONG:
! 409: case FILE_MELONG:
! 410: val += 4 * MULT;
! 411: break;
! 412:
! 413: case FILE_PSTRING:
! 414: case FILE_STRING:
! 415: val += m->vallen * MULT;
! 416: break;
! 417:
! 418: case FILE_BESTRING16:
! 419: case FILE_LESTRING16:
! 420: val += m->vallen * MULT / 2;
! 421: break;
! 422:
! 423: case FILE_SEARCH:
! 424: case FILE_REGEX:
! 425: val += m->vallen;
! 426: break;
! 427:
! 428: case FILE_DATE:
! 429: case FILE_LEDATE:
! 430: case FILE_BEDATE:
! 431: case FILE_MEDATE:
! 432: case FILE_LDATE:
! 433: case FILE_LELDATE:
! 434: case FILE_BELDATE:
! 435: case FILE_MELDATE:
! 436: val += 4 * MULT;
! 437: break;
! 438:
! 439: case FILE_QUAD:
! 440: case FILE_BEQUAD:
! 441: case FILE_LEQUAD:
! 442: case FILE_QDATE:
! 443: case FILE_LEQDATE:
! 444: case FILE_BEQDATE:
! 445: case FILE_QLDATE:
! 446: case FILE_LEQLDATE:
! 447: case FILE_BEQLDATE:
! 448: val += 8 * MULT;
! 449: break;
! 450:
! 451: default:
! 452: val = 0;
! 453: (void)fprintf(stderr, "Bad type %d\n", m->type);
! 454: abort();
! 455: }
! 456:
! 457: switch (m->reln) {
! 458: case 'x': /* matches anything penalize */
! 459: val = 0;
! 460: break;
! 461:
! 462: case '!':
! 463: case '=': /* Exact match, prefer */
! 464: val += MULT;
! 465: break;
! 466:
! 467: case '>':
! 468: case '<': /* comparison match reduce strength */
! 469: val -= 2 * MULT;
! 470: break;
! 471:
! 472: case '^':
! 473: case '&': /* masking bits, we could count them too */
! 474: val -= MULT;
! 475: break;
! 476:
! 477: default:
! 478: (void)fprintf(stderr, "Bad relation %c\n", m->reln);
! 479: abort();
! 480: }
! 481:
! 482: if (val == 0) /* ensure we only return 0 for FILE_DEFAULT */
! 483: val = 1;
! 484:
! 485: return val;
! 486: }
! 487:
! 488: /*
! 489: * Sort callback for sorting entries by "strength" (basically length)
! 490: */
! 491: private int
! 492: apprentice_sort(const void *a, const void *b)
! 493: {
! 494: const struct magic_entry *ma = a;
! 495: const struct magic_entry *mb = b;
! 496: size_t sa = apprentice_magic_strength(ma->mp);
! 497: size_t sb = apprentice_magic_strength(mb->mp);
! 498: if (sa == sb)
! 499: return 0;
! 500: else if (sa > sb)
! 501: return -1;
! 502: else
! 503: return 1;
! 504: }
! 505:
! 506: /*
1.19 tedu 507: * parse from a file
508: * const char *fn: name of magic file
509: */
510: private int
511: apprentice_file(struct magic_set *ms, struct magic **magicp, uint32_t *nmagicp,
512: const char *fn, int action)
1.1 deraadt 513: {
1.19 tedu 514: private const char hdr[] =
1.2 deraadt 515: "cont\toffset\ttype\topcode\tmask\tvalue\tdesc";
1.1 deraadt 516: FILE *f;
1.20 otto 517: char line[BUFSIZ];
1.1 deraadt 518: int errs = 0;
1.25 ! chl 519: struct magic_entry *marray;
! 520: uint32_t marraycount, i, mentrycount = 0;
! 521: size_t lineno = 0;
1.1 deraadt 522:
1.25 ! chl 523: ms->flags |= MAGIC_CHECK; /* Enable checks for parsed files */
! 524:
! 525: f = fopen(ms->file = fn, "r");
1.19 tedu 526: if (f == NULL) {
1.2 deraadt 527: if (errno != ENOENT)
1.19 tedu 528: file_error(ms, errno, "cannot read magic file `%s'",
529: fn);
1.2 deraadt 530: return -1;
1.1 deraadt 531: }
532:
1.19 tedu 533: maxmagic = MAXMAGIS;
1.25 ! chl 534: if ((marray = calloc(maxmagic, sizeof(*marray))) == NULL) {
1.19 tedu 535: (void)fclose(f);
1.25 ! chl 536: file_oomem(ms, maxmagic * sizeof(*marray));
1.19 tedu 537: return -1;
538: }
1.25 ! chl 539: marraycount = 0;
1.19 tedu 540:
541: /* print silly verbose header for USG compat. */
542: if (action == FILE_CHECK)
543: (void)fprintf(stderr, "%s\n", hdr);
544:
1.25 ! chl 545: /* read and parse this file */
! 546: for (ms->line = 1; fgets(line, sizeof(line), f) != NULL; ms->line++) {
! 547: size_t len;
! 548: len = strlen(line);
! 549: if (len == 0) /* null line, garbage, etc */
! 550: continue;
! 551: if (line[len - 1] == '\n') {
! 552: lineno++;
! 553: line[len - 1] = '\0'; /* delete newline */
! 554: }
! 555: if (line[0] == '\0') /* empty, do not parse */
1.1 deraadt 556: continue;
1.25 ! chl 557: if (line[0] == '#') /* comment, do not parse */
1.1 deraadt 558: continue;
1.25 ! chl 559: if (parse(ms, &marray, &marraycount, line, lineno, action) != 0)
! 560: errs++;
1.1 deraadt 561: }
562:
1.19 tedu 563: (void)fclose(f);
1.25 ! chl 564: if (errs)
! 565: goto out;
! 566:
! 567: #ifndef NOORDER
! 568: qsort(marray, marraycount, sizeof(*marray), apprentice_sort);
! 569: /*
! 570: * Make sure that any level 0 "default" line is last (if one exists).
! 571: */
! 572: for (i = 0; i < marraycount; i++) {
! 573: if (marray[i].mp->cont_level == 0 &&
! 574: marray[i].mp->type == FILE_DEFAULT) {
! 575: while (++i < marraycount)
! 576: if (marray[i].mp->cont_level == 0)
! 577: break;
! 578: if (i != marraycount) {
! 579: ms->line = marray[i].mp->lineno; /* XXX - Ugh! */
! 580: file_magwarn(ms,
! 581: "level 0 \"default\" did not sort last");
! 582: }
! 583: break;
! 584: }
! 585: }
! 586: #endif
! 587:
! 588: for (i = 0; i < marraycount; i++)
! 589: mentrycount += marray[i].cont_count;
! 590:
! 591: if ((*magicp = malloc(sizeof(**magicp) * mentrycount)) == NULL) {
! 592: file_oomem(ms, sizeof(**magicp) * mentrycount);
! 593: errs++;
! 594: goto out;
! 595: }
! 596:
! 597: mentrycount = 0;
! 598: for (i = 0; i < marraycount; i++) {
! 599: (void)memcpy(*magicp + mentrycount, marray[i].mp,
! 600: marray[i].cont_count * sizeof(**magicp));
! 601: mentrycount += marray[i].cont_count;
! 602: }
! 603: out:
! 604: for (i = 0; i < marraycount; i++)
! 605: free(marray[i].mp);
! 606: free(marray);
1.19 tedu 607: if (errs) {
608: *magicp = NULL;
609: *nmagicp = 0;
1.25 ! chl 610: return errs;
! 611: } else {
! 612: *nmagicp = mentrycount;
! 613: return 0;
1.19 tedu 614: }
1.25 ! chl 615:
1.1 deraadt 616: }
617:
618: /*
619: * extend the sign bit if the comparison is to be signed
620: */
1.25 ! chl 621: protected uint64_t
! 622: file_signextend(struct magic_set *ms, struct magic *m, uint64_t v)
1.1 deraadt 623: {
1.25 ! chl 624: if (!(m->flag & UNSIGNED)) {
1.1 deraadt 625: switch(m->type) {
626: /*
627: * Do not remove the casts below. They are
628: * vital. When later compared with the data,
629: * the sign extension must have happened.
630: */
1.19 tedu 631: case FILE_BYTE:
1.1 deraadt 632: v = (char) v;
633: break;
1.19 tedu 634: case FILE_SHORT:
635: case FILE_BESHORT:
636: case FILE_LESHORT:
1.1 deraadt 637: v = (short) v;
638: break;
1.19 tedu 639: case FILE_DATE:
640: case FILE_BEDATE:
641: case FILE_LEDATE:
1.25 ! chl 642: case FILE_MEDATE:
1.19 tedu 643: case FILE_LDATE:
644: case FILE_BELDATE:
645: case FILE_LELDATE:
1.25 ! chl 646: case FILE_MELDATE:
1.19 tedu 647: case FILE_LONG:
648: case FILE_BELONG:
649: case FILE_LELONG:
1.25 ! chl 650: case FILE_MELONG:
1.14 itojun 651: v = (int32_t) v;
1.1 deraadt 652: break;
1.25 ! chl 653: case FILE_QUAD:
! 654: case FILE_BEQUAD:
! 655: case FILE_LEQUAD:
! 656: case FILE_QDATE:
! 657: case FILE_QLDATE:
! 658: case FILE_BEQDATE:
! 659: case FILE_BEQLDATE:
! 660: case FILE_LEQDATE:
! 661: case FILE_LEQLDATE:
! 662: v = (int64_t) v;
! 663: break;
1.19 tedu 664: case FILE_STRING:
665: case FILE_PSTRING:
1.25 ! chl 666: case FILE_BESTRING16:
! 667: case FILE_LESTRING16:
1.19 tedu 668: case FILE_REGEX:
1.25 ! chl 669: case FILE_SEARCH:
! 670: case FILE_DEFAULT:
1.1 deraadt 671: break;
672: default:
1.19 tedu 673: if (ms->flags & MAGIC_CHECK)
1.25 ! chl 674: file_magwarn(ms, "cannot happen: m->type=%d\n",
1.19 tedu 675: m->type);
676: return ~0U;
1.1 deraadt 677: }
1.25 ! chl 678: }
1.1 deraadt 679: return v;
680: }
681:
1.25 ! chl 682: private int
! 683: string_modifier_check(struct magic_set *ms, struct magic const *m)
! 684: {
! 685: if ((ms->flags & MAGIC_CHECK) == 0)
! 686: return 0;
! 687:
! 688: switch (m->type) {
! 689: case FILE_BESTRING16:
! 690: case FILE_LESTRING16:
! 691: if (m->str_flags != 0) {
! 692: file_magwarn(ms, "no modifiers allowed for 16-bit strings\n");
! 693: return -1;
! 694: }
! 695: break;
! 696: case FILE_STRING:
! 697: case FILE_PSTRING:
! 698: if ((m->str_flags & REGEX_OFFSET_START) != 0) {
! 699: file_magwarn(ms, "'/%c' only allowed on regex and search\n",
! 700: CHAR_REGEX_OFFSET_START);
! 701: return -1;
! 702: }
! 703: break;
! 704: case FILE_SEARCH:
! 705: break;
! 706: case FILE_REGEX:
! 707: if ((m->str_flags & STRING_COMPACT_BLANK) != 0) {
! 708: file_magwarn(ms, "'/%c' not allowed on regex\n",
! 709: CHAR_COMPACT_BLANK);
! 710: return -1;
! 711: }
! 712: if ((m->str_flags & STRING_COMPACT_OPTIONAL_BLANK) != 0) {
! 713: file_magwarn(ms, "'/%c' not allowed on regex\n",
! 714: CHAR_COMPACT_OPTIONAL_BLANK);
! 715: return -1;
! 716: }
! 717: break;
! 718: default:
! 719: file_magwarn(ms, "coding error: m->type=%d\n",
! 720: m->type);
! 721: return -1;
! 722: }
! 723: return 0;
! 724: }
! 725:
! 726: private int
! 727: get_op(char c)
! 728: {
! 729: switch (c) {
! 730: case '&':
! 731: return FILE_OPAND;
! 732: case '|':
! 733: return FILE_OPOR;
! 734: case '^':
! 735: return FILE_OPXOR;
! 736: case '+':
! 737: return FILE_OPADD;
! 738: case '-':
! 739: return FILE_OPMINUS;
! 740: case '*':
! 741: return FILE_OPMULTIPLY;
! 742: case '/':
! 743: return FILE_OPDIVIDE;
! 744: case '%':
! 745: return FILE_OPMODULO;
! 746: default:
! 747: return -1;
! 748: }
! 749: }
! 750:
! 751: #ifdef ENABLE_CONDITIONALS
! 752: private int
! 753: get_cond(const char *l, const char **t)
! 754: {
! 755: static struct cond_tbl_s {
! 756: const char *name;
! 757: const size_t len;
! 758: const int cond;
! 759: } cond_tbl[] = {
! 760: { "if", 2, COND_IF },
! 761: { "elif", 4, COND_ELIF },
! 762: { "else", 4, COND_ELSE },
! 763: { NULL, 0, COND_NONE },
! 764: };
! 765: struct cond_tbl_s *p;
! 766:
! 767: for (p = cond_tbl; p->name; p++) {
! 768: if (strncmp(l, p->name, p->len) == 0 &&
! 769: isspace((unsigned char)l[p->len])) {
! 770: if (t)
! 771: *t = l + p->len;
! 772: break;
! 773: }
! 774: }
! 775: return p->cond;
! 776: }
! 777:
! 778: private int
! 779: check_cond(struct magic_set *ms, int cond, uint32_t cont_level)
! 780: {
! 781: int last_cond;
! 782: last_cond = ms->c.li[cont_level].last_cond;
! 783:
! 784: switch (cond) {
! 785: case COND_IF:
! 786: if (last_cond != COND_NONE && last_cond != COND_ELIF) {
! 787: if (ms->flags & MAGIC_CHECK)
! 788: file_magwarn(ms, "syntax error: `if'");
! 789: return -1;
! 790: }
! 791: last_cond = COND_IF;
! 792: break;
! 793:
! 794: case COND_ELIF:
! 795: if (last_cond != COND_IF && last_cond != COND_ELIF) {
! 796: if (ms->flags & MAGIC_CHECK)
! 797: file_magwarn(ms, "syntax error: `elif'");
! 798: return -1;
! 799: }
! 800: last_cond = COND_ELIF;
! 801: break;
! 802:
! 803: case COND_ELSE:
! 804: if (last_cond != COND_IF && last_cond != COND_ELIF) {
! 805: if (ms->flags & MAGIC_CHECK)
! 806: file_magwarn(ms, "syntax error: `else'");
! 807: return -1;
! 808: }
! 809: last_cond = COND_NONE;
! 810: break;
! 811:
! 812: case COND_NONE:
! 813: last_cond = COND_NONE;
! 814: break;
! 815: }
! 816:
! 817: ms->c.li[cont_level].last_cond = last_cond;
! 818: return 0;
! 819: }
! 820: #endif /* ENABLE_CONDITIONALS */
! 821:
1.1 deraadt 822: /*
823: * parse one line from magic file, put into magic[index++] if valid
824: */
1.19 tedu 825: private int
1.25 ! chl 826: parse(struct magic_set *ms, struct magic_entry **mentryp, uint32_t *nmentryp,
! 827: const char *line, size_t lineno, int action)
1.1 deraadt 828: {
1.25 ! chl 829: #ifdef ENABLE_CONDITIONALS
! 830: static uint32_t last_cont_level = 0;
! 831: #endif
! 832: size_t i;
! 833: struct magic_entry *me;
1.1 deraadt 834: struct magic *m;
1.25 ! chl 835: const char *l = line;
1.19 tedu 836: char *t;
1.25 ! chl 837: int op;
! 838: uint32_t cont_level;
1.19 tedu 839:
1.25 ! chl 840: cont_level = 0;
1.1 deraadt 841:
842: while (*l == '>') {
843: ++l; /* step over */
1.25 ! chl 844: cont_level++;
1.1 deraadt 845: }
1.25 ! chl 846: #ifdef ENABLE_CONDITIONALS
! 847: if (cont_level == 0 || cont_level > last_cont_level)
! 848: if (file_check_mem(ms, cont_level) == -1)
! 849: return -1;
! 850: last_cont_level = cont_level;
! 851: #endif
! 852:
! 853: #define ALLOC_CHUNK (size_t)10
! 854: #define ALLOC_INCR (size_t)200
1.1 deraadt 855:
1.25 ! chl 856: if (cont_level != 0) {
! 857: if (*nmentryp == 0) {
! 858: file_error(ms, 0, "No current entry for continuation");
! 859: return -1;
! 860: }
! 861: me = &(*mentryp)[*nmentryp - 1];
! 862: if (me->cont_count == me->max_count) {
! 863: struct magic *nm;
! 864: size_t cnt = me->max_count + ALLOC_CHUNK;
! 865: if ((nm = realloc(me->mp, sizeof(*nm) * cnt)) == NULL) {
! 866: file_oomem(ms, sizeof(*nm) * cnt);
! 867: return -1;
! 868: }
! 869: me->mp = m = nm;
! 870: me->max_count = cnt;
! 871: }
! 872: m = &me->mp[me->cont_count++];
! 873: (void)memset(m, 0, sizeof(*m));
! 874: m->cont_level = cont_level;
! 875: } else {
! 876: if (*nmentryp == maxmagic) {
! 877: struct magic_entry *mp;
! 878:
! 879: maxmagic += ALLOC_INCR;
! 880: if ((mp = realloc(*mentryp, sizeof(*mp) * maxmagic)) ==
! 881: NULL) {
! 882: file_oomem(ms, sizeof(*mp) * maxmagic);
! 883: return -1;
! 884: }
! 885: (void)memset(&mp[*nmentryp], 0, sizeof(*mp) *
! 886: ALLOC_INCR);
! 887: *mentryp = mp;
! 888: }
! 889: me = &(*mentryp)[*nmentryp];
! 890: if (me->mp == NULL) {
! 891: if ((m = malloc(sizeof(*m) * ALLOC_CHUNK)) == NULL) {
! 892: file_oomem(ms, sizeof(*m) * ALLOC_CHUNK);
! 893: return -1;
! 894: }
! 895: me->mp = m;
! 896: me->max_count = ALLOC_CHUNK;
! 897: } else
! 898: m = me->mp;
! 899: (void)memset(m, 0, sizeof(*m));
! 900: m->cont_level = 0;
! 901: me->cont_count = 1;
1.1 deraadt 902: }
1.25 ! chl 903: m->lineno = lineno;
! 904:
! 905: if (*l == '&') { /* m->cont_level == 0 checked below. */
1.4 millert 906: ++l; /* step over */
1.19 tedu 907: m->flag |= OFFADD;
1.4 millert 908: }
1.25 ! chl 909: if (*l == '(') {
! 910: ++l; /* step over */
! 911: m->flag |= INDIR;
! 912: if (m->flag & OFFADD)
! 913: m->flag = (m->flag & ~OFFADD) | INDIROFFADD;
! 914:
! 915: if (*l == '&') { /* m->cont_level == 0 checked below */
! 916: ++l; /* step over */
! 917: m->flag |= OFFADD;
! 918: }
! 919: }
! 920: /* Indirect offsets are not valid at level 0. */
! 921: if (m->cont_level == 0 && (m->flag & (OFFADD | INDIROFFADD)))
! 922: if (ms->flags & MAGIC_CHECK)
! 923: file_magwarn(ms, "relative offset at level 0");
1.1 deraadt 924:
925: /* get offset, then skip over it */
1.19 tedu 926: m->offset = (uint32_t)strtoul(l, &t, 0);
1.1 deraadt 927: if (l == t)
1.19 tedu 928: if (ms->flags & MAGIC_CHECK)
1.25 ! chl 929: file_magwarn(ms, "offset `%s' invalid", l);
1.1 deraadt 930: l = t;
931:
932: if (m->flag & INDIR) {
1.19 tedu 933: m->in_type = FILE_LONG;
934: m->in_offset = 0;
1.1 deraadt 935: /*
936: * read [.lbs][+-]nnnnn)
937: */
938: if (*l == '.') {
939: l++;
1.19 tedu 940: switch (*l) {
1.1 deraadt 941: case 'l':
1.19 tedu 942: m->in_type = FILE_LELONG;
943: break;
944: case 'L':
945: m->in_type = FILE_BELONG;
1.1 deraadt 946: break;
1.25 ! chl 947: case 'm':
! 948: m->in_type = FILE_MELONG;
! 949: break;
1.1 deraadt 950: case 'h':
951: case 's':
1.19 tedu 952: m->in_type = FILE_LESHORT;
953: break;
954: case 'H':
955: case 'S':
956: m->in_type = FILE_BESHORT;
1.1 deraadt 957: break;
958: case 'c':
959: case 'b':
1.19 tedu 960: case 'C':
961: case 'B':
962: m->in_type = FILE_BYTE;
1.1 deraadt 963: break;
964: default:
1.19 tedu 965: if (ms->flags & MAGIC_CHECK)
1.25 ! chl 966: file_magwarn(ms,
! 967: "indirect offset type `%c' invalid",
1.19 tedu 968: *l);
1.1 deraadt 969: break;
970: }
971: l++;
972: }
1.25 ! chl 973:
! 974: m->in_op = 0;
1.19 tedu 975: if (*l == '~') {
1.25 ! chl 976: m->in_op |= FILE_OPINVERSE;
1.19 tedu 977: l++;
978: }
1.25 ! chl 979: if ((op = get_op(*l)) != -1) {
! 980: m->in_op |= op;
1.19 tedu 981: l++;
1.25 ! chl 982: }
! 983: if (*l == '(') {
! 984: m->in_op |= FILE_OPINDIRECT;
1.19 tedu 985: l++;
1.1 deraadt 986: }
1.25 ! chl 987: if (isdigit((unsigned char)*l) || *l == '-') {
! 988: m->in_offset = (int32_t)strtol(l, &t, 0);
! 989: if (l == t)
! 990: if (ms->flags & MAGIC_CHECK)
! 991: file_magwarn(ms,
! 992: "in_offset `%s' invalid", l);
! 993: l = t;
! 994: }
! 995: if (*l++ != ')' ||
! 996: ((m->in_op & FILE_OPINDIRECT) && *l++ != ')'))
1.19 tedu 997: if (ms->flags & MAGIC_CHECK)
1.25 ! chl 998: file_magwarn(ms,
! 999: "missing ')' in indirect offset");
1.1 deraadt 1000: }
1.25 ! chl 1001: EATAB;
1.1 deraadt 1002:
1.25 ! chl 1003: #ifdef ENABLE_CONDITIONALS
! 1004: m->cond = get_cond(l, &l);
! 1005: if (check_cond(ms, m->cond, cont_level) == -1)
! 1006: return -1;
1.1 deraadt 1007:
1008: EATAB;
1.25 ! chl 1009: #endif
1.1 deraadt 1010:
1011: if (*l == 'u') {
1012: ++l;
1013: m->flag |= UNSIGNED;
1014: }
1015:
1.25 ! chl 1016: m->type = get_type(l, &l);
! 1017: if (m->type == FILE_INVALID) {
1.19 tedu 1018: if (ms->flags & MAGIC_CHECK)
1.25 ! chl 1019: file_magwarn(ms, "type `%s' invalid", l);
1.1 deraadt 1020: return -1;
1021: }
1.25 ! chl 1022:
1.1 deraadt 1023: /* New-style anding: "0 byte&0x80 =0x80 dynamically linked" */
1.19 tedu 1024: /* New and improved: ~ & | ^ + - * / % -- exciting, isn't it? */
1.25 ! chl 1025:
! 1026: m->mask_op = 0;
1.19 tedu 1027: if (*l == '~') {
1.25 ! chl 1028: if (!IS_STRING(m->type))
! 1029: m->mask_op |= FILE_OPINVERSE;
! 1030: else if (ms->flags & MAGIC_CHECK)
! 1031: file_magwarn(ms, "'~' invalid for string types");
1.1 deraadt 1032: ++l;
1.19 tedu 1033: }
1.25 ! chl 1034: m->str_count = 0;
! 1035: m->str_flags = 0;
! 1036: m->num_mask = 0;
! 1037: if ((op = get_op(*l)) != -1) {
! 1038: if (!IS_STRING(m->type)) {
! 1039: uint64_t val;
1.19 tedu 1040: ++l;
1041: m->mask_op |= op;
1.25 ! chl 1042: val = (uint64_t)strtoull(l, &t, 0);
! 1043: l = t;
! 1044: m->num_mask = file_signextend(ms, m, val);
1.19 tedu 1045: eatsize(&l);
1.25 ! chl 1046: }
! 1047: else if (op == FILE_OPDIVIDE) {
! 1048: int have_count = 0;
1.19 tedu 1049: while (!isspace((unsigned char)*++l)) {
1050: switch (*l) {
1.25 ! chl 1051: /* for portability avoid "case '0' ... '9':" */
! 1052: case '0': case '1': case '2':
! 1053: case '3': case '4': case '5':
! 1054: case '6': case '7': case '8':
! 1055: case '9': {
! 1056: if (have_count && ms->flags & MAGIC_CHECK)
! 1057: file_magwarn(ms,
! 1058: "multiple counts");
! 1059: have_count = 1;
! 1060: m->str_count = strtoul(l, &t, 0);
! 1061: l = t - 1;
1.19 tedu 1062: break;
1.25 ! chl 1063: }
1.19 tedu 1064: case CHAR_COMPACT_BLANK:
1.25 ! chl 1065: m->str_flags |= STRING_COMPACT_BLANK;
1.19 tedu 1066: break;
1067: case CHAR_COMPACT_OPTIONAL_BLANK:
1.25 ! chl 1068: m->str_flags |=
1.19 tedu 1069: STRING_COMPACT_OPTIONAL_BLANK;
1070: break;
1.25 ! chl 1071: case CHAR_IGNORE_LOWERCASE:
! 1072: m->str_flags |= STRING_IGNORE_LOWERCASE;
! 1073: break;
! 1074: case CHAR_IGNORE_UPPERCASE:
! 1075: m->str_flags |= STRING_IGNORE_UPPERCASE;
! 1076: break;
! 1077: case CHAR_REGEX_OFFSET_START:
! 1078: m->str_flags |= REGEX_OFFSET_START;
! 1079: break;
1.19 tedu 1080: default:
1081: if (ms->flags & MAGIC_CHECK)
1.25 ! chl 1082: file_magwarn(ms,
! 1083: "string extension `%c' invalid",
1.19 tedu 1084: *l);
1085: return -1;
1086: }
1.25 ! chl 1087: /* allow multiple '/' for readability */
! 1088: if (l[1] == '/' && !isspace((unsigned char)l[2]))
! 1089: l++;
1.19 tedu 1090: }
1.25 ! chl 1091: if (string_modifier_check(ms, m) == -1)
! 1092: return -1;
! 1093: }
! 1094: else {
! 1095: if (ms->flags & MAGIC_CHECK)
! 1096: file_magwarn(ms, "invalid string op: %c", *t);
! 1097: return -1;
1.19 tedu 1098: }
1099: }
1100: /*
1101: * We used to set mask to all 1's here, instead let's just not do
1102: * anything if mask = 0 (unless you have a better idea)
1103: */
1.1 deraadt 1104: EATAB;
1105:
1106: switch (*l) {
1107: case '>':
1108: case '<':
1109: /* Old-style anding: "0 byte &0x80 dynamically linked" */
1110: case '&':
1111: case '^':
1112: case '=':
1113: m->reln = *l;
1114: ++l;
1.19 tedu 1115: if (*l == '=') {
1116: /* HP compat: ignore &= etc. */
1117: ++l;
1118: }
1.1 deraadt 1119: break;
1120: case '!':
1.25 ! chl 1121: m->reln = *l;
! 1122: ++l;
! 1123: break;
1.1 deraadt 1124: default:
1.25 ! chl 1125: m->reln = '='; /* the default relation */
! 1126: if (*l == 'x' && ((isascii((unsigned char)l[1]) &&
! 1127: isspace((unsigned char)l[1])) || !l[1])) {
1.1 deraadt 1128: m->reln = *l;
1129: ++l;
1130: }
1131: break;
1132: }
1.25 ! chl 1133: /*
! 1134: * Grab the value part, except for an 'x' reln.
! 1135: */
! 1136: if (m->reln != 'x' && getvalue(ms, m, &l, action))
1.1 deraadt 1137: return -1;
1.25 ! chl 1138:
1.1 deraadt 1139: /*
1140: * TODO finish this macro and start using it!
1141: * #define offsetcheck {if (offset > HOWMANY-1)
1.19 tedu 1142: * magwarn("offset too big"); }
1.1 deraadt 1143: */
1144:
1145: /*
1.25 ! chl 1146: * Now get last part - the description
1.1 deraadt 1147: */
1148: EATAB;
1149: if (l[0] == '\b') {
1150: ++l;
1151: m->nospflag = 1;
1152: } else if ((l[0] == '\\') && (l[1] == 'b')) {
1153: ++l;
1154: ++l;
1155: m->nospflag = 1;
1156: } else
1157: m->nospflag = 0;
1.25 ! chl 1158: for (i = 0; (m->desc[i++] = *l++) != '\0' && i < sizeof(m->desc); )
! 1159: continue;
! 1160: if (i == sizeof(m->desc)) {
! 1161: m->desc[sizeof(m->desc) - 1] = '\0';
! 1162: if (ms->flags & MAGIC_CHECK)
! 1163: file_magwarn(ms, "description `%s' truncated", m->desc);
! 1164: }
1.21 pedro 1165:
1.25 ! chl 1166: /*
! 1167: * We only do this check while compiling, or if any of the magic
! 1168: * files were not compiled.
! 1169: */
! 1170: if (ms->flags & MAGIC_CHECK) {
! 1171: if (check_format(ms, m) == -1)
! 1172: return -1;
! 1173: }
1.19 tedu 1174: #ifndef COMPILE_ONLY
1175: if (action == FILE_CHECK) {
1176: file_mdump(m);
1.1 deraadt 1177: }
1.19 tedu 1178: #endif
1.25 ! chl 1179: if (m->cont_level == 0)
! 1180: ++(*nmentryp); /* make room for next */
! 1181: return 0;
! 1182: }
! 1183:
! 1184: private int
! 1185: check_format_type(const char *ptr, int type)
! 1186: {
! 1187: int quad = 0;
! 1188: if (*ptr == '\0') {
! 1189: /* Missing format string; bad */
! 1190: return -1;
! 1191: }
! 1192:
! 1193: switch (type) {
! 1194: case FILE_FMT_QUAD:
! 1195: quad = 1;
! 1196: /*FALLTHROUGH*/
! 1197: case FILE_FMT_NUM:
! 1198: if (*ptr == '-')
! 1199: ptr++;
! 1200: if (*ptr == '.')
! 1201: ptr++;
! 1202: while (isdigit((unsigned char)*ptr)) ptr++;
! 1203: if (*ptr == '.')
! 1204: ptr++;
! 1205: while (isdigit((unsigned char)*ptr)) ptr++;
! 1206: if (quad) {
! 1207: if (*ptr++ != 'l')
! 1208: return -1;
! 1209: if (*ptr++ != 'l')
! 1210: return -1;
! 1211: }
! 1212:
! 1213: switch (*ptr++) {
! 1214: case 'l':
! 1215: switch (*ptr++) {
! 1216: case 'i':
! 1217: case 'd':
! 1218: case 'u':
! 1219: case 'x':
! 1220: case 'X':
! 1221: return 0;
! 1222: default:
! 1223: return -1;
! 1224: }
! 1225:
! 1226: case 'h':
! 1227: switch (*ptr++) {
! 1228: case 'h':
! 1229: switch (*ptr++) {
! 1230: case 'i':
! 1231: case 'd':
! 1232: case 'u':
! 1233: case 'x':
! 1234: case 'X':
! 1235: return 0;
! 1236: default:
! 1237: return -1;
! 1238: }
! 1239: case 'd':
! 1240: return 0;
! 1241: default:
! 1242: return -1;
! 1243: }
! 1244:
! 1245: case 'i':
! 1246: case 'c':
! 1247: case 'd':
! 1248: case 'u':
! 1249: case 'x':
! 1250: case 'X':
! 1251: return 0;
! 1252:
! 1253: default:
! 1254: return -1;
! 1255: }
! 1256:
! 1257: case FILE_FMT_STR:
! 1258: if (*ptr == '-')
! 1259: ptr++;
! 1260: while (isdigit((unsigned char )*ptr))
! 1261: ptr++;
! 1262: if (*ptr == '.') {
! 1263: ptr++;
! 1264: while (isdigit((unsigned char )*ptr))
! 1265: ptr++;
! 1266: }
! 1267:
! 1268: switch (*ptr++) {
! 1269: case 's':
! 1270: return 0;
! 1271: default:
! 1272: return -1;
! 1273: }
! 1274:
! 1275: default:
! 1276: /* internal error */
! 1277: abort();
! 1278: }
! 1279: /*NOTREACHED*/
! 1280: return -1;
! 1281: }
! 1282:
! 1283: /*
! 1284: * Check that the optional printf format in description matches
! 1285: * the type of the magic.
! 1286: */
! 1287: private int
! 1288: check_format(struct magic_set *ms, struct magic *m)
! 1289: {
! 1290: char *ptr;
! 1291:
! 1292: for (ptr = m->desc; *ptr; ptr++)
! 1293: if (*ptr == '%')
! 1294: break;
! 1295: if (*ptr == '\0') {
! 1296: /* No format string; ok */
! 1297: return 1;
! 1298: }
! 1299:
! 1300: assert(file_nformats == file_nnames);
! 1301:
! 1302: if (m->type >= file_nformats) {
! 1303: file_error(ms, 0, "Internal error inconsistency between "
! 1304: "m->type and format strings");
! 1305: return -1;
! 1306: }
! 1307: if (file_formats[m->type] == FILE_FMT_NONE) {
! 1308: file_error(ms, 0, "No format string for `%s' with description "
! 1309: "`%s'", m->desc, file_names[m->type]);
! 1310: return -1;
! 1311: }
! 1312:
! 1313: ptr++;
! 1314: if (check_format_type(ptr, file_formats[m->type]) == -1) {
! 1315: /*
! 1316: * TODO: this error message is unhelpful if the format
! 1317: * string is not one character long
! 1318: */
! 1319: file_error(ms, 0, "Printf format `%c' is not valid for type "
! 1320: " `%s' in description `%s'",
! 1321: ptr && *ptr ? *ptr : '?',
! 1322: file_names[m->type], m->desc);
! 1323: return -1;
! 1324: }
! 1325:
! 1326: for (; *ptr; ptr++) {
! 1327: if (*ptr == '%') {
! 1328: file_error(ms, 0,
! 1329: "Too many format strings (should have at most one) "
! 1330: "for `%s' with description `%s'",
! 1331: file_names[m->type], m->desc);
! 1332: return -1;
! 1333: }
! 1334: }
1.1 deraadt 1335: return 0;
1336: }
1337:
1338: /*
1339: * Read a numeric value from a pointer, into the value union of a magic
1340: * pointer, according to the magic type. Update the string pointer to point
1341: * just after the number read. Return 0 for success, non-zero for failure.
1342: */
1.19 tedu 1343: private int
1.25 ! chl 1344: getvalue(struct magic_set *ms, struct magic *m, const char **p, int action)
1.1 deraadt 1345: {
1346: int slen;
1347:
1.19 tedu 1348: switch (m->type) {
1.25 ! chl 1349: case FILE_BESTRING16:
! 1350: case FILE_LESTRING16:
1.19 tedu 1351: case FILE_STRING:
1352: case FILE_PSTRING:
1353: case FILE_REGEX:
1.25 ! chl 1354: case FILE_SEARCH:
! 1355: *p = getstr(ms, *p, m->value.s, sizeof(m->value.s), &slen, action);
1.19 tedu 1356: if (*p == NULL) {
1357: if (ms->flags & MAGIC_CHECK)
1.25 ! chl 1358: file_magwarn(ms, "cannot get string from `%s'",
1.19 tedu 1359: m->value.s);
1360: return -1;
1361: }
1.1 deraadt 1362: m->vallen = slen;
1.19 tedu 1363: return 0;
1364: default:
1.1 deraadt 1365: if (m->reln != 'x') {
1.25 ! chl 1366: char *ep;
! 1367: m->value.q = file_signextend(ms, m,
! 1368: (uint64_t)strtoull(*p, &ep, 0));
! 1369: *p = ep;
1.1 deraadt 1370: eatsize(p);
1371: }
1.19 tedu 1372: return 0;
1373: }
1.1 deraadt 1374: }
1375:
1376: /*
1377: * Convert a string containing C character escapes. Stop at an unescaped
1378: * space or tab.
1379: * Copy the converted version to "p", returning its length in *slen.
1380: * Return updated scan pointer as function result.
1381: */
1.25 ! chl 1382: private const char *
! 1383: getstr(struct magic_set *ms, const char *s, char *p, int plen, int *slen, int action)
1.1 deraadt 1384: {
1.25 ! chl 1385: const char *origs = s;
! 1386: char *origp = p;
1.1 deraadt 1387: char *pmax = p + plen - 1;
1.11 mpech 1388: int c;
1389: int val;
1.1 deraadt 1390:
1391: while ((c = *s++) != '\0') {
1392: if (isspace((unsigned char) c))
1393: break;
1394: if (p >= pmax) {
1.19 tedu 1395: file_error(ms, 0, "string too long: `%s'", origs);
1396: return NULL;
1.1 deraadt 1397: }
1.25 ! chl 1398: if (c == '\\') {
1.1 deraadt 1399: switch(c = *s++) {
1400:
1401: case '\0':
1.25 ! chl 1402: if (action == FILE_COMPILE)
! 1403: file_magwarn(ms, "incomplete escape");
1.1 deraadt 1404: goto out;
1405:
1.25 ! chl 1406: case '\t':
! 1407: if (action == FILE_COMPILE) {
! 1408: file_magwarn(ms,
! 1409: "escaped tab found, use \\t instead");
! 1410: action++;
! 1411: }
! 1412: /*FALLTHROUGH*/
1.1 deraadt 1413: default:
1.25 ! chl 1414: if (action == FILE_COMPILE) {
! 1415: if (isprint((unsigned char)c))
! 1416: file_magwarn(ms,
! 1417: "no need to escape `%c'", c);
! 1418: else
! 1419: file_magwarn(ms,
! 1420: "unknown escape sequence: \\%03o", c);
! 1421: }
! 1422: /*FALLTHROUGH*/
! 1423: /* space, perhaps force people to use \040? */
! 1424: case ' ':
! 1425: #if 0
! 1426: /*
! 1427: * Other things people escape, but shouldn't need to,
! 1428: * so we disallow them
! 1429: */
! 1430: case '\'':
! 1431: case '"':
! 1432: case '?':
! 1433: #endif
! 1434: /* Relations */
! 1435: case '>':
! 1436: case '<':
! 1437: case '&':
! 1438: case '^':
! 1439: case '=':
! 1440: case '!':
! 1441: /* and baskslash itself */
! 1442: case '\\':
1.1 deraadt 1443: *p++ = (char) c;
1444: break;
1445:
1.25 ! chl 1446: case 'a':
! 1447: *p++ = '\a';
! 1448: break;
! 1449:
! 1450: case 'b':
! 1451: *p++ = '\b';
! 1452: break;
! 1453:
! 1454: case 'f':
! 1455: *p++ = '\f';
! 1456: break;
! 1457:
1.1 deraadt 1458: case 'n':
1459: *p++ = '\n';
1460: break;
1461:
1462: case 'r':
1463: *p++ = '\r';
1464: break;
1465:
1466: case 't':
1467: *p++ = '\t';
1468: break;
1469:
1470: case 'v':
1471: *p++ = '\v';
1472: break;
1473:
1474: /* \ and up to 3 octal digits */
1475: case '0':
1476: case '1':
1477: case '2':
1478: case '3':
1479: case '4':
1480: case '5':
1481: case '6':
1482: case '7':
1483: val = c - '0';
1484: c = *s++; /* try for 2 */
1.25 ! chl 1485: if (c >= '0' && c <= '7') {
! 1486: val = (val << 3) | (c - '0');
1.1 deraadt 1487: c = *s++; /* try for 3 */
1.25 ! chl 1488: if (c >= '0' && c <= '7')
! 1489: val = (val << 3) | (c-'0');
1.1 deraadt 1490: else
1491: --s;
1492: }
1493: else
1494: --s;
1495: *p++ = (char)val;
1496: break;
1497:
1.4 millert 1498: /* \x and up to 2 hex digits */
1.1 deraadt 1499: case 'x':
1500: val = 'x'; /* Default if no digits */
1501: c = hextoint(*s++); /* Get next char */
1502: if (c >= 0) {
1503: val = c;
1504: c = hextoint(*s++);
1.4 millert 1505: if (c >= 0)
1.1 deraadt 1506: val = (val << 4) + c;
1.4 millert 1507: else
1.1 deraadt 1508: --s;
1509: } else
1510: --s;
1511: *p++ = (char)val;
1512: break;
1513: }
1514: } else
1515: *p++ = (char)c;
1516: }
1517: out:
1518: *p = '\0';
1519: *slen = p - origp;
1520: return s;
1521: }
1522:
1523:
1524: /* Single hex char to int; -1 if not a hex char. */
1.19 tedu 1525: private int
1526: hextoint(int c)
1527: {
1528: if (!isascii((unsigned char) c))
1529: return -1;
1530: if (isdigit((unsigned char) c))
1531: return c - '0';
1.25 ! chl 1532: if ((c >= 'a') && (c <= 'f'))
1.19 tedu 1533: return c + 10 - 'a';
1.25 ! chl 1534: if (( c>= 'A') && (c <= 'F'))
1.19 tedu 1535: return c + 10 - 'A';
1536: return -1;
1.1 deraadt 1537: }
1538:
1539:
1540: /*
1541: * Print a string containing C character escapes.
1542: */
1.19 tedu 1543: protected void
1544: file_showstr(FILE *fp, const char *s, size_t len)
1.1 deraadt 1545: {
1.11 mpech 1546: char c;
1.1 deraadt 1547:
1548: for (;;) {
1549: c = *s++;
1.19 tedu 1550: if (len == ~0U) {
1.1 deraadt 1551: if (c == '\0')
1552: break;
1553: }
1554: else {
1555: if (len-- == 0)
1556: break;
1557: }
1.25 ! chl 1558: if (c >= 040 && c <= 0176) /* TODO isprint && !iscntrl */
1.1 deraadt 1559: (void) fputc(c, fp);
1560: else {
1561: (void) fputc('\\', fp);
1562: switch (c) {
1.25 ! chl 1563: case '\a':
! 1564: (void) fputc('a', fp);
! 1565: break;
! 1566:
! 1567: case '\b':
! 1568: (void) fputc('b', fp);
! 1569: break;
! 1570:
! 1571: case '\f':
! 1572: (void) fputc('f', fp);
! 1573: break;
! 1574:
1.1 deraadt 1575: case '\n':
1576: (void) fputc('n', fp);
1577: break;
1578:
1579: case '\r':
1580: (void) fputc('r', fp);
1581: break;
1582:
1583: case '\t':
1584: (void) fputc('t', fp);
1585: break;
1586:
1587: case '\v':
1588: (void) fputc('v', fp);
1589: break;
1590:
1591: default:
1592: (void) fprintf(fp, "%.3o", c & 0377);
1593: break;
1594: }
1595: }
1596: }
1597: }
1598:
1599: /*
1600: * eatsize(): Eat the size spec from a number [eg. 10UL]
1601: */
1.19 tedu 1602: private void
1.25 ! chl 1603: eatsize(const char **p)
1.1 deraadt 1604: {
1.25 ! chl 1605: const char *l = *p;
1.1 deraadt 1606:
1607: if (LOWCASE(*l) == 'u')
1608: l++;
1609:
1610: switch (LOWCASE(*l)) {
1611: case 'l': /* long */
1612: case 's': /* short */
1613: case 'h': /* short */
1614: case 'b': /* char/byte */
1615: case 'c': /* char/byte */
1616: l++;
1617: /*FALLTHROUGH*/
1618: default:
1619: break;
1620: }
1621:
1622: *p = l;
1.19 tedu 1623: }
1624:
1625: /*
1626: * handle a compiled file.
1627: */
1628: private int
1629: apprentice_map(struct magic_set *ms, struct magic **magicp, uint32_t *nmagicp,
1630: const char *fn)
1631: {
1632: int fd;
1633: struct stat st;
1634: uint32_t *ptr;
1635: uint32_t version;
1636: int needsbyteswap;
1637: char buf[MAXPATHLEN];
1.25 ! chl 1638: char *dbname = mkdbname(fn, buf, sizeof(buf), 0);
1.19 tedu 1639: void *mm = NULL;
1640:
1641: if (dbname == NULL)
1642: return -1;
1643:
1.25 ! chl 1644: if ((fd = open(dbname, O_RDONLY|O_BINARY)) == -1)
1.19 tedu 1645: return -1;
1646:
1647: if (fstat(fd, &st) == -1) {
1648: file_error(ms, errno, "cannot stat `%s'", dbname);
1649: goto error;
1650: }
1651: if (st.st_size < 16) {
1652: file_error(ms, 0, "file `%s' is too small", dbname);
1653: goto error;
1654: }
1655:
1656: #ifdef QUICK
1657: if ((mm = mmap(0, (size_t)st.st_size, PROT_READ|PROT_WRITE,
1658: MAP_PRIVATE|MAP_FILE, fd, (off_t)0)) == MAP_FAILED) {
1659: file_error(ms, errno, "cannot map `%s'", dbname);
1660: goto error;
1661: }
1662: #define RET 2
1663: #else
1664: if ((mm = malloc((size_t)st.st_size)) == NULL) {
1.25 ! chl 1665: file_oomem(ms, (size_t)st.st_size);
1.19 tedu 1666: goto error;
1667: }
1668: if (read(fd, mm, (size_t)st.st_size) != (size_t)st.st_size) {
1669: file_badread(ms);
1670: goto error;
1671: }
1672: #define RET 1
1673: #endif
1674: *magicp = mm;
1675: (void)close(fd);
1676: fd = -1;
1677: ptr = (uint32_t *)(void *)*magicp;
1678: if (*ptr != MAGICNO) {
1679: if (swap4(*ptr) != MAGICNO) {
1680: file_error(ms, 0, "bad magic in `%s'");
1681: goto error;
1682: }
1683: needsbyteswap = 1;
1684: } else
1685: needsbyteswap = 0;
1686: if (needsbyteswap)
1687: version = swap4(ptr[1]);
1688: else
1689: version = ptr[1];
1690: if (version != VERSIONNO) {
1691: file_error(ms, 0, "version mismatch (%d != %d) in `%s'",
1692: version, VERSIONNO, dbname);
1693: goto error;
1694: }
1695: *nmagicp = (uint32_t)(st.st_size / sizeof(struct magic)) - 1;
1696: (*magicp)++;
1697: if (needsbyteswap)
1698: byteswap(*magicp, *nmagicp);
1699: return RET;
1700:
1701: error:
1702: if (fd != -1)
1703: (void)close(fd);
1704: if (mm) {
1705: #ifdef QUICK
1706: (void)munmap((void *)mm, (size_t)st.st_size);
1707: #else
1708: free(mm);
1709: #endif
1710: } else {
1711: *magicp = NULL;
1712: *nmagicp = 0;
1713: }
1714: return -1;
1715: }
1716:
1717: private const uint32_t ar[] = {
1718: MAGICNO, VERSIONNO
1719: };
1720: /*
1721: * handle an mmaped file.
1722: */
1723: private int
1724: apprentice_compile(struct magic_set *ms, struct magic **magicp,
1725: uint32_t *nmagicp, const char *fn)
1726: {
1727: int fd;
1728: char buf[MAXPATHLEN];
1.25 ! chl 1729: char *dbname = mkdbname(fn, buf, sizeof(buf), 1);
1.19 tedu 1730:
1731: if (dbname == NULL)
1732: return -1;
1733:
1.25 ! chl 1734: if ((fd = open(dbname, O_WRONLY|O_CREAT|O_TRUNC|O_BINARY, 0644)) == -1) {
1.19 tedu 1735: file_error(ms, errno, "cannot open `%s'", dbname);
1736: return -1;
1737: }
1738:
1739: if (write(fd, ar, sizeof(ar)) != (ssize_t)sizeof(ar)) {
1740: file_error(ms, errno, "error writing `%s'", dbname);
1741: return -1;
1742: }
1743:
1744: if (lseek(fd, (off_t)sizeof(struct magic), SEEK_SET)
1745: != sizeof(struct magic)) {
1746: file_error(ms, errno, "error seeking `%s'", dbname);
1747: return -1;
1748: }
1749:
1750: if (write(fd, *magicp, (sizeof(struct magic) * *nmagicp))
1751: != (ssize_t)(sizeof(struct magic) * *nmagicp)) {
1752: file_error(ms, errno, "error writing `%s'", dbname);
1753: return -1;
1754: }
1755:
1756: (void)close(fd);
1757: return 0;
1758: }
1759:
1760: private const char ext[] = ".mgc";
1761: /*
1762: * make a dbname
1763: */
1764: private char *
1.25 ! chl 1765: mkdbname(const char *fn, char *buf, size_t bufsiz, int strip)
1.19 tedu 1766: {
1.25 ! chl 1767: if (strip) {
! 1768: const char *p;
! 1769: if ((p = strrchr(fn, '/')) != NULL)
! 1770: fn = ++p;
! 1771: }
! 1772:
1.19 tedu 1773: (void)snprintf(buf, bufsiz, "%s%s", fn, ext);
1774: return buf;
1775: }
1776:
1777: /*
1778: * Byteswap an mmap'ed file if needed
1779: */
1780: private void
1781: byteswap(struct magic *magic, uint32_t nmagic)
1782: {
1783: uint32_t i;
1784: for (i = 0; i < nmagic; i++)
1785: bs1(&magic[i]);
1786: }
1787:
1788: /*
1789: * swap a short
1790: */
1791: private uint16_t
1792: swap2(uint16_t sv)
1793: {
1794: uint16_t rv;
1795: uint8_t *s = (uint8_t *)(void *)&sv;
1796: uint8_t *d = (uint8_t *)(void *)&rv;
1797: d[0] = s[1];
1798: d[1] = s[0];
1799: return rv;
1800: }
1801:
1802: /*
1803: * swap an int
1804: */
1805: private uint32_t
1806: swap4(uint32_t sv)
1807: {
1808: uint32_t rv;
1809: uint8_t *s = (uint8_t *)(void *)&sv;
1810: uint8_t *d = (uint8_t *)(void *)&rv;
1811: d[0] = s[3];
1812: d[1] = s[2];
1813: d[2] = s[1];
1814: d[3] = s[0];
1815: return rv;
1816: }
1817:
1818: /*
1.25 ! chl 1819: * swap a quad
! 1820: */
! 1821: private uint64_t
! 1822: swap8(uint64_t sv)
! 1823: {
! 1824: uint32_t rv;
! 1825: uint8_t *s = (uint8_t *)(void *)&sv;
! 1826: uint8_t *d = (uint8_t *)(void *)&rv;
! 1827: d[0] = s[3];
! 1828: d[1] = s[2];
! 1829: d[2] = s[1];
! 1830: d[3] = s[0];
! 1831: d[4] = s[7];
! 1832: d[5] = s[6];
! 1833: d[6] = s[5];
! 1834: d[7] = s[4];
! 1835: return rv;
! 1836: }
! 1837:
! 1838: /*
1.19 tedu 1839: * byteswap a single magic entry
1840: */
1841: private void
1842: bs1(struct magic *m)
1843: {
1844: m->cont_level = swap2(m->cont_level);
1845: m->offset = swap4((uint32_t)m->offset);
1846: m->in_offset = swap4((uint32_t)m->in_offset);
1.25 ! chl 1847: m->lineno = swap4((uint32_t)m->lineno);
! 1848: if (IS_STRING(m->type)) {
! 1849: m->str_count = swap4(m->str_count);
! 1850: m->str_flags = swap4(m->str_flags);
! 1851: }
! 1852: else {
! 1853: m->value.q = swap8(m->value.q);
! 1854: m->num_mask = swap8(m->num_mask);
! 1855: }
1.1 deraadt 1856: }