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 "php.h"
33
34 #include "file.h"
35
36 #ifndef lint
37 FILE_RCSID("@(#)$File: apprentice.c,v 1.230 2015/01/02 21:29:39 christos Exp $")
38 #endif /* lint */
39
40 #include "magic.h"
41 #include "patchlevel.h"
42 #include <stdlib.h>
43
44 #if defined(__hpux) && !defined(HAVE_STRTOULL)
45 #if SIZEOF_LONG == 8
46 # define strtoull strtoul
47 #else
48 # define strtoull __strtoull
49 #endif
50 #endif
51
52 #ifdef PHP_WIN32
53 #include "win32/unistd.h"
54 #define strtoull _strtoui64
55 #else
56 #include <unistd.h>
57 #endif
58 #include <string.h>
59 #include <assert.h>
60 #include <ctype.h>
61 #include <fcntl.h>
62
63 #ifndef SSIZE_MAX
64 #define MAXMAGIC_SIZE ((ssize_t)0x7fffffff)
65 #else
66 #define MAXMAGIC_SIZE SSIZE_MAX
67 #endif
68
69 #define EATAB {while (isascii((unsigned char) *l) && \
70 isspace((unsigned char) *l)) ++l;}
71 #define LOWCASE(l) (isupper((unsigned char) (l)) ? \
72 tolower((unsigned char) (l)) : (l))
73 /*
74 * Work around a bug in headers on Digital Unix.
75 * At least confirmed for: OSF1 V4.0 878
76 */
77 #if defined(__osf__) && defined(__DECC)
78 #ifdef MAP_FAILED
79 #undef MAP_FAILED
80 #endif
81 #endif
82
83 #ifndef offsetof
84 #define offsetof(STRUCTURE,FIELD) ((int)((char*)&((STRUCTURE*)0)->FIELD))
85 #endif
86
87 #ifndef MAP_FAILED
88 #define MAP_FAILED (void *) -1
89 #endif
90
91 #ifndef MAP_FILE
92 #define MAP_FILE 0
93 #endif
94
95 #define ALLOC_CHUNK (size_t)10
96 #define ALLOC_INCR (size_t)200
97
98 #define MAP_TYPE_MMAP 0
99 #define MAP_TYPE_MALLOC 1
100 #define MAP_TYPE_USER 2
101
102 struct magic_entry {
103 struct magic *mp;
104 uint32_t cont_count;
105 uint32_t max_count;
106 };
107
108 struct magic_entry_set {
109 struct magic_entry *me;
110 uint32_t count;
111 uint32_t max;
112 };
113
114 struct magic_map {
115 void *p;
116 size_t len;
117 int type;
118 struct magic *magic[MAGIC_SETS];
119 uint32_t nmagic[MAGIC_SETS];
120 };
121
122 int file_formats[FILE_NAMES_SIZE];
123 const size_t file_nformats = FILE_NAMES_SIZE;
124 const char *file_names[FILE_NAMES_SIZE];
125 const size_t file_nnames = FILE_NAMES_SIZE;
126
127 private int getvalue(struct magic_set *ms, struct magic *, const char **, int);
128 private int hextoint(int);
129 private const char *getstr(struct magic_set *, struct magic *, const char *,
130 int);
131 private int parse(struct magic_set *, struct magic_entry *, const char *,
132 size_t, int);
133 private void eatsize(const char **);
134 private int apprentice_1(struct magic_set *, const char *, int);
135 private size_t apprentice_magic_strength(const struct magic *);
136 private int apprentice_sort(const void *, const void *);
137 private void apprentice_list(struct mlist *, int );
138 private struct magic_map *apprentice_load(struct magic_set *,
139 const char *, int);
140 private struct mlist *mlist_alloc(void);
141 private void mlist_free(struct mlist *);
142 private void byteswap(struct magic *, uint32_t);
143 private void bs1(struct magic *);
144 private uint16_t swap2(uint16_t);
145 private uint32_t swap4(uint32_t);
146 private uint64_t swap8(uint64_t);
147 private char *mkdbname(struct magic_set *, const char *, int);
148 private struct magic_map *apprentice_buf(struct magic_set *, struct magic *,
149 size_t);
150 private struct magic_map *apprentice_map(struct magic_set *, const char *);
151 private int check_buffer(struct magic_set *, struct magic_map *, const char *);
152 private void apprentice_unmap(struct magic_map *);
153 private int apprentice_compile(struct magic_set *, struct magic_map *,
154 const char *);
155 private int check_format_type(const char *, int);
156 private int check_format(struct magic_set *, struct magic *);
157 private int get_op(char);
158 private int parse_mime(struct magic_set *, struct magic_entry *, const char *);
159 private int parse_strength(struct magic_set *, struct magic_entry *, const char *);
160 private int parse_apple(struct magic_set *, struct magic_entry *, const char *);
161
162
163 private size_t magicsize = sizeof(struct magic);
164
165 private const char usg_hdr[] = "cont\toffset\ttype\topcode\tmask\tvalue\tdesc";
166
167 private struct {
168 const char *name;
169 size_t len;
170 int (*fun)(struct magic_set *, struct magic_entry *, const char *);
171 } bang[] = {
172 #define DECLARE_FIELD(name) { # name, sizeof(# name) - 1, parse_ ## name }
173 DECLARE_FIELD(mime),
174 DECLARE_FIELD(apple),
175 DECLARE_FIELD(strength),
176 #undef DECLARE_FIELD
177 { NULL, 0, NULL }
178 };
179
180 #include "../data_file.c"
181
182 struct type_tbl_s {
183 const char name[16];
184 const size_t len;
185 const int type;
186 const int format;
187 };
188
189 /*
190 * XXX - the actual Single UNIX Specification says that "long" means "long",
191 * as in the C data type, but we treat it as meaning "4-byte integer".
192 * Given that the OS X version of file 5.04 did the same, I guess that passes
193 * the actual test; having "long" be dependent on how big a "long" is on
194 * the machine running "file" is silly.
195 */
196 static const struct type_tbl_s type_tbl[] = {
197 # define XX(s) s, (sizeof(s) - 1)
198 # define XX_NULL "", 0
199 { XX("invalid"), FILE_INVALID, FILE_FMT_NONE },
200 { XX("byte"), FILE_BYTE, FILE_FMT_NUM },
201 { XX("short"), FILE_SHORT, FILE_FMT_NUM },
202 { XX("default"), FILE_DEFAULT, FILE_FMT_NONE },
203 { XX("long"), FILE_LONG, FILE_FMT_NUM },
204 { XX("string"), FILE_STRING, FILE_FMT_STR },
205 { XX("date"), FILE_DATE, FILE_FMT_STR },
206 { XX("beshort"), FILE_BESHORT, FILE_FMT_NUM },
207 { XX("belong"), FILE_BELONG, FILE_FMT_NUM },
208 { XX("bedate"), FILE_BEDATE, FILE_FMT_STR },
209 { XX("leshort"), FILE_LESHORT, FILE_FMT_NUM },
210 { XX("lelong"), FILE_LELONG, FILE_FMT_NUM },
211 { XX("ledate"), FILE_LEDATE, FILE_FMT_STR },
212 { XX("pstring"), FILE_PSTRING, FILE_FMT_STR },
213 { XX("ldate"), FILE_LDATE, FILE_FMT_STR },
214 { XX("beldate"), FILE_BELDATE, FILE_FMT_STR },
215 { XX("leldate"), FILE_LELDATE, FILE_FMT_STR },
216 { XX("regex"), FILE_REGEX, FILE_FMT_STR },
217 { XX("bestring16"), FILE_BESTRING16, FILE_FMT_STR },
218 { XX("lestring16"), FILE_LESTRING16, FILE_FMT_STR },
219 { XX("search"), FILE_SEARCH, FILE_FMT_STR },
220 { XX("medate"), FILE_MEDATE, FILE_FMT_STR },
221 { XX("meldate"), FILE_MELDATE, FILE_FMT_STR },
222 { XX("melong"), FILE_MELONG, FILE_FMT_NUM },
223 { XX("quad"), FILE_QUAD, FILE_FMT_QUAD },
224 { XX("lequad"), FILE_LEQUAD, FILE_FMT_QUAD },
225 { XX("bequad"), FILE_BEQUAD, FILE_FMT_QUAD },
226 { XX("qdate"), FILE_QDATE, FILE_FMT_STR },
227 { XX("leqdate"), FILE_LEQDATE, FILE_FMT_STR },
228 { XX("beqdate"), FILE_BEQDATE, FILE_FMT_STR },
229 { XX("qldate"), FILE_QLDATE, FILE_FMT_STR },
230 { XX("leqldate"), FILE_LEQLDATE, FILE_FMT_STR },
231 { XX("beqldate"), FILE_BEQLDATE, FILE_FMT_STR },
232 { XX("float"), FILE_FLOAT, FILE_FMT_FLOAT },
233 { XX("befloat"), FILE_BEFLOAT, FILE_FMT_FLOAT },
234 { XX("lefloat"), FILE_LEFLOAT, FILE_FMT_FLOAT },
235 { XX("double"), FILE_DOUBLE, FILE_FMT_DOUBLE },
236 { XX("bedouble"), FILE_BEDOUBLE, FILE_FMT_DOUBLE },
237 { XX("ledouble"), FILE_LEDOUBLE, FILE_FMT_DOUBLE },
238 { XX("leid3"), FILE_LEID3, FILE_FMT_NUM },
239 { XX("beid3"), FILE_BEID3, FILE_FMT_NUM },
240 { XX("indirect"), FILE_INDIRECT, FILE_FMT_NUM },
241 { XX("qwdate"), FILE_QWDATE, FILE_FMT_STR },
242 { XX("leqwdate"), FILE_LEQWDATE, FILE_FMT_STR },
243 { XX("beqwdate"), FILE_BEQWDATE, FILE_FMT_STR },
244 { XX("name"), FILE_NAME, FILE_FMT_NONE },
245 { XX("use"), FILE_USE, FILE_FMT_NONE },
246 { XX("clear"), FILE_CLEAR, FILE_FMT_NONE },
247 { XX_NULL, FILE_INVALID, FILE_FMT_NONE },
248 };
249
250 /*
251 * These are not types, and cannot be preceded by "u" to make them
252 * unsigned.
253 */
254 static const struct type_tbl_s special_tbl[] = {
255 { XX("name"), FILE_NAME, FILE_FMT_STR },
256 { XX("use"), FILE_USE, FILE_FMT_STR },
257 { XX_NULL, FILE_INVALID, FILE_FMT_NONE },
258 };
259 # undef XX
260 # undef XX_NULL
261
262 #ifndef S_ISDIR
263 #define S_ISDIR(mode) ((mode) & _S_IFDIR)
264 #endif
265
266 private int
get_type(const struct type_tbl_s * tbl,const char * l,const char ** t)267 get_type(const struct type_tbl_s *tbl, const char *l, const char **t)
268 {
269 const struct type_tbl_s *p;
270
271 for (p = tbl; p->len; p++) {
272 if (strncmp(l, p->name, p->len) == 0) {
273 if (t)
274 *t = l + p->len;
275 break;
276 }
277 }
278 return p->type;
279 }
280
281 private int
get_standard_integer_type(const char * l,const char ** t)282 get_standard_integer_type(const char *l, const char **t)
283 {
284 int type;
285
286 if (isalpha((unsigned char)l[1])) {
287 switch (l[1]) {
288 case 'C':
289 /* "dC" and "uC" */
290 type = FILE_BYTE;
291 break;
292 case 'S':
293 /* "dS" and "uS" */
294 type = FILE_SHORT;
295 break;
296 case 'I':
297 case 'L':
298 /*
299 * "dI", "dL", "uI", and "uL".
300 *
301 * XXX - the actual Single UNIX Specification says
302 * that "L" means "long", as in the C data type,
303 * but we treat it as meaning "4-byte integer".
304 * Given that the OS X version of file 5.04 did
305 * the same, I guess that passes the actual SUS
306 * validation suite; having "dL" be dependent on
307 * how big a "long" is on the machine running
308 * "file" is silly.
309 */
310 type = FILE_LONG;
311 break;
312 case 'Q':
313 /* "dQ" and "uQ" */
314 type = FILE_QUAD;
315 break;
316 default:
317 /* "d{anything else}", "u{anything else}" */
318 return FILE_INVALID;
319 }
320 l += 2;
321 } else if (isdigit((unsigned char)l[1])) {
322 /*
323 * "d{num}" and "u{num}"; we only support {num} values
324 * of 1, 2, 4, and 8 - the Single UNIX Specification
325 * doesn't say anything about whether arbitrary
326 * values should be supported, but both the Solaris 10
327 * and OS X Mountain Lion versions of file passed the
328 * Single UNIX Specification validation suite, and
329 * neither of them support values bigger than 8 or
330 * non-power-of-2 values.
331 */
332 if (isdigit((unsigned char)l[2])) {
333 /* Multi-digit, so > 9 */
334 return FILE_INVALID;
335 }
336 switch (l[1]) {
337 case '1':
338 type = FILE_BYTE;
339 break;
340 case '2':
341 type = FILE_SHORT;
342 break;
343 case '4':
344 type = FILE_LONG;
345 break;
346 case '8':
347 type = FILE_QUAD;
348 break;
349 default:
350 /* XXX - what about 3, 5, 6, or 7? */
351 return FILE_INVALID;
352 }
353 l += 2;
354 } else {
355 /*
356 * "d" or "u" by itself.
357 */
358 type = FILE_LONG;
359 ++l;
360 }
361 if (t)
362 *t = l;
363 return type;
364 }
365
366 private void
init_file_tables(void)367 init_file_tables(void)
368 {
369 static int done = 0;
370 const struct type_tbl_s *p;
371
372 if (done)
373 return;
374 done++;
375
376 for (p = type_tbl; p->len; p++) {
377 assert(p->type < FILE_NAMES_SIZE);
378 file_names[p->type] = p->name;
379 file_formats[p->type] = p->format;
380 }
381 assert(p - type_tbl == FILE_NAMES_SIZE);
382 }
383
384 private int
add_mlist(struct mlist * mlp,struct magic_map * map,size_t idx)385 add_mlist(struct mlist *mlp, struct magic_map *map, size_t idx)
386 {
387 struct mlist *ml;
388
389 mlp->map = idx == 0 ? map : NULL;
390 if ((ml = CAST(struct mlist *, emalloc(sizeof(*ml)))) == NULL)
391 return -1;
392
393 ml->map = NULL;
394 ml->magic = map->magic[idx];
395 ml->nmagic = map->nmagic[idx];
396
397 mlp->prev->next = ml;
398 ml->prev = mlp->prev;
399 ml->next = mlp;
400 mlp->prev = ml;
401 return 0;
402 }
403
404 /*
405 * Handle one file or directory.
406 */
407 private int
apprentice_1(struct magic_set * ms,const char * fn,int action)408 apprentice_1(struct magic_set *ms, const char *fn, int action)
409 {
410 struct magic_map *map;
411 struct mlist *ml;
412 size_t i;
413
414 if (magicsize != FILE_MAGICSIZE) {
415 file_error(ms, 0, "magic element size %lu != %lu",
416 (unsigned long)sizeof(*map->magic[0]),
417 (unsigned long)FILE_MAGICSIZE);
418 return -1;
419 }
420
421 if (action == FILE_COMPILE) {
422 map = apprentice_load(ms, fn, action);
423 if (map == NULL)
424 return -1;
425 return apprentice_compile(ms, map, fn);
426 }
427
428 map = apprentice_map(ms, fn);
429 if (map == NULL) {
430 if (fn) {
431 if (ms->flags & MAGIC_CHECK)
432 file_magwarn(ms, "using regular magic file `%s'", fn);
433 map = apprentice_load(ms, fn, action);
434 }
435 if (map == NULL)
436 return -1;
437 }
438
439 for (i = 0; i < MAGIC_SETS; i++) {
440 if (add_mlist(ms->mlist[i], map, i) == -1) {
441 file_oomem(ms, sizeof(*ml));
442 apprentice_unmap(map);
443 return -1;
444 }
445 }
446
447 if (action == FILE_LIST) {
448 for (i = 0; i < MAGIC_SETS; i++) {
449 printf("Set %" SIZE_T_FORMAT "u:\nBinary patterns:\n",
450 i);
451 apprentice_list(ms->mlist[i], BINTEST);
452 printf("Text patterns:\n");
453 apprentice_list(ms->mlist[i], TEXTTEST);
454 }
455 }
456 return 0;
457 }
458
459 protected void
file_ms_free(struct magic_set * ms)460 file_ms_free(struct magic_set *ms)
461 {
462 size_t i;
463 if (ms == NULL)
464 return;
465 for (i = 0; i < MAGIC_SETS; i++)
466 mlist_free(ms->mlist[i]);
467 if (ms->o.pbuf) {
468 efree(ms->o.pbuf);
469 }
470 if (ms->o.buf) {
471 efree(ms->o.buf);
472 }
473 if (ms->c.li) {
474 efree(ms->c.li);
475 }
476 efree(ms);
477 }
478
479 protected struct magic_set *
file_ms_alloc(int flags)480 file_ms_alloc(int flags)
481 {
482 struct magic_set *ms;
483 size_t i, len;
484
485 if ((ms = CAST(struct magic_set *, ecalloc((size_t)1,
486 sizeof(struct magic_set)))) == NULL)
487 return NULL;
488
489 if (magic_setflags(ms, flags) == -1) {
490 errno = EINVAL;
491 goto free;
492 }
493
494 ms->o.buf = ms->o.pbuf = NULL;
495 len = (ms->c.len = 10) * sizeof(*ms->c.li);
496
497 if ((ms->c.li = CAST(struct level_info *, emalloc(len))) == NULL)
498 goto free;
499
500 ms->event_flags = 0;
501 ms->error = -1;
502 for (i = 0; i < MAGIC_SETS; i++)
503 ms->mlist[i] = NULL;
504 ms->file = "unknown";
505 ms->line = 0;
506 ms->indir_max = FILE_INDIR_MAX;
507 ms->name_max = FILE_NAME_MAX;
508 ms->elf_shnum_max = FILE_ELF_SHNUM_MAX;
509 ms->elf_phnum_max = FILE_ELF_PHNUM_MAX;
510 ms->elf_notes_max = FILE_ELF_NOTES_MAX;
511 return ms;
512 free:
513 efree(ms);
514 return NULL;
515 }
516
517 private void
apprentice_unmap(struct magic_map * map)518 apprentice_unmap(struct magic_map *map)
519 {
520 if (map == NULL)
521 return;
522 if (map->p != php_magic_database) {
523 if (map->p == NULL) {
524 int j;
525 for (j = 0; j < MAGIC_SETS; j++) {
526 if (map->magic[j]) {
527 efree(map->magic[j]);
528 }
529 }
530 } else {
531 efree(map->p);
532 }
533 }
534 efree(map);
535 }
536
537 private struct mlist *
mlist_alloc(void)538 mlist_alloc(void)
539 {
540 struct mlist *mlist;
541 if ((mlist = CAST(struct mlist *, ecalloc(1, sizeof(*mlist)))) == NULL) {
542 return NULL;
543 }
544 mlist->next = mlist->prev = mlist;
545 return mlist;
546 }
547
548 private void
mlist_free(struct mlist * mlist)549 mlist_free(struct mlist *mlist)
550 {
551 struct mlist *ml, *next;
552
553 if (mlist == NULL)
554 return;
555
556 ml = mlist->next;
557 for (ml = mlist->next; (next = ml->next) != NULL; ml = next) {
558 if (ml->map)
559 apprentice_unmap(ml->map);
560 efree(ml);
561 if (ml == mlist)
562 break;
563 }
564 }
565
566 /* const char *fn: list of magic files and directories */
567 protected int
file_apprentice(struct magic_set * ms,const char * fn,int action)568 file_apprentice(struct magic_set *ms, const char *fn, int action)
569 {
570 char *p, *mfn;
571 int file_err, errs = -1;
572 size_t i;
573
574 if (ms->mlist[0] != NULL)
575 file_reset(ms);
576
577 /* XXX disabling default magic loading so the compiled in data is used */
578 #if 0
579 if ((fn = magic_getpath(fn, action)) == NULL)
580 return -1;
581 #endif
582
583 init_file_tables();
584
585 if (fn == NULL)
586 fn = getenv("MAGIC");
587 if (fn == NULL) {
588 for (i = 0; i < MAGIC_SETS; i++) {
589 mlist_free(ms->mlist[i]);
590 if ((ms->mlist[i] = mlist_alloc()) == NULL) {
591 file_oomem(ms, sizeof(*ms->mlist[i]));
592 return -1;
593 }
594 }
595 return apprentice_1(ms, fn, action);
596 }
597
598 if ((mfn = estrdup(fn)) == NULL) {
599 file_oomem(ms, strlen(fn));
600 return -1;
601 }
602
603 for (i = 0; i < MAGIC_SETS; i++) {
604 mlist_free(ms->mlist[i]);
605 if ((ms->mlist[i] = mlist_alloc()) == NULL) {
606 file_oomem(ms, sizeof(*ms->mlist[i]));
607 while (i-- > 0) {
608 mlist_free(ms->mlist[i]);
609 ms->mlist[i] = NULL;
610 }
611 efree(mfn);
612 return -1;
613 }
614 }
615 fn = mfn;
616
617 while (fn) {
618 p = strchr(fn, PATHSEP);
619 if (p)
620 *p++ = '\0';
621 if (*fn == '\0')
622 break;
623 file_err = apprentice_1(ms, fn, action);
624 errs = MAX(errs, file_err);
625 fn = p;
626 }
627
628 efree(mfn);
629
630 if (errs == -1) {
631 for (i = 0; i < MAGIC_SETS; i++) {
632 mlist_free(ms->mlist[i]);
633 ms->mlist[i] = NULL;
634 }
635 file_error(ms, 0, "could not find any valid magic files!");
636 return -1;
637 }
638
639 #if 0
640 /*
641 * Always leave the database loaded
642 */
643 if (action == FILE_LOAD)
644 return 0;
645
646 for (i = 0; i < MAGIC_SETS; i++) {
647 mlist_free(ms->mlist[i]);
648 ms->mlist[i] = NULL;
649 }
650 #endif
651
652 switch (action) {
653 case FILE_LOAD:
654 case FILE_COMPILE:
655 case FILE_CHECK:
656 case FILE_LIST:
657 return 0;
658 default:
659 file_error(ms, 0, "Invalid action %d", action);
660 return -1;
661 }
662 }
663
664 /*
665 * Compute the real length of a magic expression, for the purposes
666 * of determining how "strong" a magic expression is (approximating
667 * how specific its matches are):
668 * - magic characters count 0 unless escaped.
669 * - [] expressions count 1
670 * - {} expressions count 0
671 * - regular characters or escaped magic characters count 1
672 * - 0 length expressions count as one
673 */
674 private size_t
nonmagic(const char * str)675 nonmagic(const char *str)
676 {
677 const char *p;
678 size_t rv = 0;
679
680 for (p = str; *p; p++)
681 switch (*p) {
682 case '\\': /* Escaped anything counts 1 */
683 if (!*++p)
684 p--;
685 rv++;
686 continue;
687 case '?': /* Magic characters count 0 */
688 case '*':
689 case '.':
690 case '+':
691 case '^':
692 case '$':
693 continue;
694 case '[': /* Bracketed expressions count 1 the ']' */
695 while (*p && *p != ']')
696 p++;
697 p--;
698 continue;
699 case '{': /* Braced expressions count 0 */
700 while (*p && *p != '}')
701 p++;
702 if (!*p)
703 p--;
704 continue;
705 default: /* Anything else counts 1 */
706 rv++;
707 continue;
708 }
709
710 return rv == 0 ? 1 : rv; /* Return at least 1 */
711 }
712
713 /*
714 * Get weight of this magic entry, for sorting purposes.
715 */
716 private size_t
apprentice_magic_strength(const struct magic * m)717 apprentice_magic_strength(const struct magic *m)
718 {
719 #define MULT 10
720 size_t v, val = 2 * MULT; /* baseline strength */
721
722 switch (m->type) {
723 case FILE_DEFAULT: /* make sure this sorts last */
724 if (m->factor_op != FILE_FACTOR_OP_NONE)
725 abort();
726 return 0;
727
728 case FILE_BYTE:
729 val += 1 * MULT;
730 break;
731
732 case FILE_SHORT:
733 case FILE_LESHORT:
734 case FILE_BESHORT:
735 val += 2 * MULT;
736 break;
737
738 case FILE_LONG:
739 case FILE_LELONG:
740 case FILE_BELONG:
741 case FILE_MELONG:
742 val += 4 * MULT;
743 break;
744
745 case FILE_PSTRING:
746 case FILE_STRING:
747 val += m->vallen * MULT;
748 break;
749
750 case FILE_BESTRING16:
751 case FILE_LESTRING16:
752 val += m->vallen * MULT / 2;
753 break;
754
755 case FILE_SEARCH:
756 val += m->vallen * MAX(MULT / m->vallen, 1);
757 break;
758
759 case FILE_REGEX:
760 v = nonmagic(m->value.s);
761 val += v * MAX(MULT / v, 1);
762 break;
763
764 case FILE_DATE:
765 case FILE_LEDATE:
766 case FILE_BEDATE:
767 case FILE_MEDATE:
768 case FILE_LDATE:
769 case FILE_LELDATE:
770 case FILE_BELDATE:
771 case FILE_MELDATE:
772 case FILE_FLOAT:
773 case FILE_BEFLOAT:
774 case FILE_LEFLOAT:
775 val += 4 * MULT;
776 break;
777
778 case FILE_QUAD:
779 case FILE_BEQUAD:
780 case FILE_LEQUAD:
781 case FILE_QDATE:
782 case FILE_LEQDATE:
783 case FILE_BEQDATE:
784 case FILE_QLDATE:
785 case FILE_LEQLDATE:
786 case FILE_BEQLDATE:
787 case FILE_QWDATE:
788 case FILE_LEQWDATE:
789 case FILE_BEQWDATE:
790 case FILE_DOUBLE:
791 case FILE_BEDOUBLE:
792 case FILE_LEDOUBLE:
793 val += 8 * MULT;
794 break;
795
796 case FILE_INDIRECT:
797 case FILE_NAME:
798 case FILE_USE:
799 break;
800
801 default:
802 (void)fprintf(stderr, "Bad type %d\n", m->type);
803 abort();
804 }
805
806 switch (m->reln) {
807 case 'x': /* matches anything penalize */
808 case '!': /* matches almost anything penalize */
809 val = 0;
810 break;
811
812 case '=': /* Exact match, prefer */
813 val += MULT;
814 break;
815
816 case '>':
817 case '<': /* comparison match reduce strength */
818 val -= 2 * MULT;
819 break;
820
821 case '^':
822 case '&': /* masking bits, we could count them too */
823 val -= MULT;
824 break;
825
826 default:
827 (void)fprintf(stderr, "Bad relation %c\n", m->reln);
828 abort();
829 }
830
831 if (val == 0) /* ensure we only return 0 for FILE_DEFAULT */
832 val = 1;
833
834 switch (m->factor_op) {
835 case FILE_FACTOR_OP_NONE:
836 break;
837 case FILE_FACTOR_OP_PLUS:
838 val += m->factor;
839 break;
840 case FILE_FACTOR_OP_MINUS:
841 val -= m->factor;
842 break;
843 case FILE_FACTOR_OP_TIMES:
844 val *= m->factor;
845 break;
846 case FILE_FACTOR_OP_DIV:
847 val /= m->factor;
848 break;
849 default:
850 abort();
851 }
852
853 /*
854 * Magic entries with no description get a bonus because they depend
855 * on subsequent magic entries to print something.
856 */
857 if (m->desc[0] == '\0')
858 val++;
859 return val;
860 }
861
862 /*
863 * Sort callback for sorting entries by "strength" (basically length)
864 */
865 private int
apprentice_sort(const void * a,const void * b)866 apprentice_sort(const void *a, const void *b)
867 {
868 const struct magic_entry *ma = CAST(const struct magic_entry *, a);
869 const struct magic_entry *mb = CAST(const struct magic_entry *, b);
870 size_t sa = apprentice_magic_strength(ma->mp);
871 size_t sb = apprentice_magic_strength(mb->mp);
872 if (sa == sb)
873 return 0;
874 else if (sa > sb)
875 return -1;
876 else
877 return 1;
878 }
879
880 /*
881 * Shows sorted patterns list in the order which is used for the matching
882 */
883 private void
apprentice_list(struct mlist * mlist,int mode)884 apprentice_list(struct mlist *mlist, int mode)
885 {
886 uint32_t magindex = 0;
887 struct mlist *ml;
888 for (ml = mlist->next; ml != mlist; ml = ml->next) {
889 for (magindex = 0; magindex < ml->nmagic; magindex++) {
890 struct magic *m = &ml->magic[magindex];
891 if ((m->flag & mode) != mode) {
892 /* Skip sub-tests */
893 while (magindex + 1 < ml->nmagic &&
894 ml->magic[magindex + 1].cont_level != 0)
895 ++magindex;
896 continue; /* Skip to next top-level test*/
897 }
898
899 /*
900 * Try to iterate over the tree until we find item with
901 * description/mimetype.
902 */
903 while (magindex + 1 < ml->nmagic &&
904 ml->magic[magindex + 1].cont_level != 0 &&
905 *ml->magic[magindex].desc == '\0' &&
906 *ml->magic[magindex].mimetype == '\0')
907 magindex++;
908
909 printf("Strength = %3" SIZE_T_FORMAT "u@%u: %s [%s]\n",
910 apprentice_magic_strength(m),
911 ml->magic[magindex].lineno,
912 ml->magic[magindex].desc,
913 ml->magic[magindex].mimetype);
914 }
915 }
916 }
917
918 private void
set_test_type(struct magic * mstart,struct magic * m)919 set_test_type(struct magic *mstart, struct magic *m)
920 {
921 switch (m->type) {
922 case FILE_BYTE:
923 case FILE_SHORT:
924 case FILE_LONG:
925 case FILE_DATE:
926 case FILE_BESHORT:
927 case FILE_BELONG:
928 case FILE_BEDATE:
929 case FILE_LESHORT:
930 case FILE_LELONG:
931 case FILE_LEDATE:
932 case FILE_LDATE:
933 case FILE_BELDATE:
934 case FILE_LELDATE:
935 case FILE_MEDATE:
936 case FILE_MELDATE:
937 case FILE_MELONG:
938 case FILE_QUAD:
939 case FILE_LEQUAD:
940 case FILE_BEQUAD:
941 case FILE_QDATE:
942 case FILE_LEQDATE:
943 case FILE_BEQDATE:
944 case FILE_QLDATE:
945 case FILE_LEQLDATE:
946 case FILE_BEQLDATE:
947 case FILE_QWDATE:
948 case FILE_LEQWDATE:
949 case FILE_BEQWDATE:
950 case FILE_FLOAT:
951 case FILE_BEFLOAT:
952 case FILE_LEFLOAT:
953 case FILE_DOUBLE:
954 case FILE_BEDOUBLE:
955 case FILE_LEDOUBLE:
956 mstart->flag |= BINTEST;
957 break;
958 case FILE_STRING:
959 case FILE_PSTRING:
960 case FILE_BESTRING16:
961 case FILE_LESTRING16:
962 /* Allow text overrides */
963 if (mstart->str_flags & STRING_TEXTTEST)
964 mstart->flag |= TEXTTEST;
965 else
966 mstart->flag |= BINTEST;
967 break;
968 case FILE_REGEX:
969 case FILE_SEARCH:
970 /* Check for override */
971 if (mstart->str_flags & STRING_BINTEST)
972 mstart->flag |= BINTEST;
973 if (mstart->str_flags & STRING_TEXTTEST)
974 mstart->flag |= TEXTTEST;
975
976 if (mstart->flag & (TEXTTEST|BINTEST))
977 break;
978
979 /* binary test if pattern is not text */
980 if (file_looks_utf8(m->value.us, (size_t)m->vallen, NULL,
981 NULL) <= 0)
982 mstart->flag |= BINTEST;
983 else
984 mstart->flag |= TEXTTEST;
985 break;
986 case FILE_DEFAULT:
987 /* can't deduce anything; we shouldn't see this at the
988 top level anyway */
989 break;
990 case FILE_INVALID:
991 default:
992 /* invalid search type, but no need to complain here */
993 break;
994 }
995 }
996
997 private int
addentry(struct magic_set * ms,struct magic_entry * me,struct magic_entry_set * mset)998 addentry(struct magic_set *ms, struct magic_entry *me,
999 struct magic_entry_set *mset)
1000 {
1001 size_t i = me->mp->type == FILE_NAME ? 1 : 0;
1002 if (mset[i].count == mset[i].max) {
1003 struct magic_entry *mp;
1004
1005 mset[i].max += ALLOC_INCR;
1006 if ((mp = CAST(struct magic_entry *,
1007 erealloc(mset[i].me, sizeof(*mp) * mset[i].max))) ==
1008 NULL) {
1009 file_oomem(ms, sizeof(*mp) * mset[i].max);
1010 return -1;
1011 }
1012 (void)memset(&mp[mset[i].count], 0, sizeof(*mp) *
1013 ALLOC_INCR);
1014 mset[i].me = mp;
1015 }
1016 mset[i].me[mset[i].count++] = *me;
1017 memset(me, 0, sizeof(*me));
1018 return 0;
1019 }
1020
1021 /*
1022 * Load and parse one file.
1023 */
1024 private void
load_1(struct magic_set * ms,int action,const char * fn,int * errs,struct magic_entry_set * mset)1025 load_1(struct magic_set *ms, int action, const char *fn, int *errs,
1026 struct magic_entry_set *mset)
1027 {
1028 char buffer[BUFSIZ + 1];
1029 char *line = NULL;
1030 size_t len;
1031 size_t lineno = 0;
1032 struct magic_entry me;
1033
1034 php_stream *stream;
1035
1036
1037 ms->file = fn;
1038 stream = php_stream_open_wrapper((char *)fn, "rb", REPORT_ERRORS, NULL);
1039
1040 if (stream == NULL) {
1041 if (errno != ENOENT)
1042 file_error(ms, errno, "cannot read magic file `%s'",
1043 fn);
1044 (*errs)++;
1045 return;
1046 }
1047
1048 memset(&me, 0, sizeof(me));
1049 /* read and parse this file */
1050 for (ms->line = 1; (line = php_stream_get_line(stream, buffer , BUFSIZ, &len)) != NULL; ms->line++) {
1051 if (len == 0) /* null line, garbage, etc */
1052 continue;
1053 if (line[len - 1] == '\n') {
1054 lineno++;
1055 line[len - 1] = '\0'; /* delete newline */
1056 }
1057 switch (line[0]) {
1058 case '\0': /* empty, do not parse */
1059 case '#': /* comment, do not parse */
1060 continue;
1061 case '!':
1062 if (line[1] == ':') {
1063 size_t i;
1064
1065 for (i = 0; bang[i].name != NULL; i++) {
1066 if ((size_t)(len - 2) > bang[i].len &&
1067 memcmp(bang[i].name, line + 2,
1068 bang[i].len) == 0)
1069 break;
1070 }
1071 if (bang[i].name == NULL) {
1072 file_error(ms, 0,
1073 "Unknown !: entry `%s'", line);
1074 (*errs)++;
1075 continue;
1076 }
1077 if (me.mp == NULL) {
1078 file_error(ms, 0,
1079 "No current entry for :!%s type",
1080 bang[i].name);
1081 (*errs)++;
1082 continue;
1083 }
1084 if ((*bang[i].fun)(ms, &me,
1085 line + bang[i].len + 2) != 0) {
1086 (*errs)++;
1087 continue;
1088 }
1089 continue;
1090 }
1091 /*FALLTHROUGH*/
1092 default:
1093 again:
1094 switch (parse(ms, &me, line, lineno, action)) {
1095 case 0:
1096 continue;
1097 case 1:
1098 (void)addentry(ms, &me, mset);
1099 goto again;
1100 default:
1101 (*errs)++;
1102 break;
1103 }
1104 }
1105 }
1106 if (me.mp)
1107 (void)addentry(ms, &me, mset);
1108 efree(line);
1109 php_stream_close(stream);
1110 }
1111
1112 /*
1113 * parse a file or directory of files
1114 * const char *fn: name of magic file or directory
1115 */
1116 private int
cmpstrp(const void * p1,const void * p2)1117 cmpstrp(const void *p1, const void *p2)
1118 {
1119 return strcmp(*(char *const *)p1, *(char *const *)p2);
1120 }
1121
1122
1123 private uint32_t
set_text_binary(struct magic_set * ms,struct magic_entry * me,uint32_t nme,uint32_t starttest)1124 set_text_binary(struct magic_set *ms, struct magic_entry *me, uint32_t nme,
1125 uint32_t starttest)
1126 {
1127 static const char text[] = "text";
1128 static const char binary[] = "binary";
1129 static const size_t len = sizeof(text);
1130
1131 uint32_t i = starttest;
1132
1133 do {
1134 set_test_type(me[starttest].mp, me[i].mp);
1135 if ((ms->flags & MAGIC_DEBUG) == 0)
1136 continue;
1137 (void)fprintf(stderr, "%s%s%s: %s\n",
1138 me[i].mp->mimetype,
1139 me[i].mp->mimetype[0] == '\0' ? "" : "; ",
1140 me[i].mp->desc[0] ? me[i].mp->desc : "(no description)",
1141 me[i].mp->flag & BINTEST ? binary : text);
1142 if (me[i].mp->flag & BINTEST) {
1143 char *p = strstr(me[i].mp->desc, text);
1144 if (p && (p == me[i].mp->desc ||
1145 isspace((unsigned char)p[-1])) &&
1146 (p + len - me[i].mp->desc == MAXstring
1147 || (p[len] == '\0' ||
1148 isspace((unsigned char)p[len]))))
1149 (void)fprintf(stderr, "*** Possible "
1150 "binary test for text type\n");
1151 }
1152 } while (++i < nme && me[i].mp->cont_level != 0);
1153 return i;
1154 }
1155
1156 private void
set_last_default(struct magic_set * ms,struct magic_entry * me,uint32_t nme)1157 set_last_default(struct magic_set *ms, struct magic_entry *me, uint32_t nme)
1158 {
1159 uint32_t i;
1160 for (i = 0; i < nme; i++) {
1161 if (me[i].mp->cont_level == 0 &&
1162 me[i].mp->type == FILE_DEFAULT) {
1163 while (++i < nme)
1164 if (me[i].mp->cont_level == 0)
1165 break;
1166 if (i != nme) {
1167 /* XXX - Ugh! */
1168 ms->line = me[i].mp->lineno;
1169 file_magwarn(ms,
1170 "level 0 \"default\" did not sort last");
1171 }
1172 return;
1173 }
1174 }
1175 }
1176
1177 private int
coalesce_entries(struct magic_set * ms,struct magic_entry * me,uint32_t nme,struct magic ** ma,uint32_t * nma)1178 coalesce_entries(struct magic_set *ms, struct magic_entry *me, uint32_t nme,
1179 struct magic **ma, uint32_t *nma)
1180 {
1181 uint32_t i, mentrycount = 0;
1182 size_t slen;
1183
1184 for (i = 0; i < nme; i++)
1185 mentrycount += me[i].cont_count;
1186
1187 slen = sizeof(**ma) * mentrycount;
1188 if ((*ma = CAST(struct magic *, emalloc(slen))) == NULL) {
1189 file_oomem(ms, slen);
1190 return -1;
1191 }
1192
1193 mentrycount = 0;
1194 for (i = 0; i < nme; i++) {
1195 (void)memcpy(*ma + mentrycount, me[i].mp,
1196 me[i].cont_count * sizeof(**ma));
1197 mentrycount += me[i].cont_count;
1198 }
1199 *nma = mentrycount;
1200 return 0;
1201 }
1202
1203 private void
magic_entry_free(struct magic_entry * me,uint32_t nme)1204 magic_entry_free(struct magic_entry *me, uint32_t nme)
1205 {
1206 uint32_t i;
1207 if (me == NULL)
1208 return;
1209 for (i = 0; i < nme; i++)
1210 efree(me[i].mp);
1211 efree(me);
1212 }
1213
1214 private struct magic_map *
apprentice_load(struct magic_set * ms,const char * fn,int action)1215 apprentice_load(struct magic_set *ms, const char *fn, int action)
1216 {
1217 int errs = 0;
1218 uint32_t i, j;
1219 size_t files = 0, maxfiles = 0;
1220 char **filearr = NULL;
1221 zend_stat_t st;
1222 struct magic_map *map;
1223 struct magic_entry_set mset[MAGIC_SETS];
1224 php_stream *dir;
1225 php_stream_dirent d;
1226
1227
1228 memset(mset, 0, sizeof(mset));
1229 ms->flags |= MAGIC_CHECK; /* Enable checks for parsed files */
1230
1231
1232 if ((map = CAST(struct magic_map *, ecalloc(1, sizeof(*map)))) == NULL)
1233 {
1234 file_oomem(ms, sizeof(*map));
1235 return NULL;
1236 }
1237
1238 /* print silly verbose header for USG compat. */
1239 if (action == FILE_CHECK)
1240 (void)fprintf(stderr, "%s\n", usg_hdr);
1241
1242 /* load directory or file */
1243 /* FIXME: Read file names and sort them to prevent
1244 non-determinism. See Debian bug #488562. */
1245 if (php_sys_stat(fn, &st) == 0 && S_ISDIR(st.st_mode)) {
1246 int mflen;
1247 char mfn[MAXPATHLEN];
1248
1249 dir = php_stream_opendir((char *)fn, REPORT_ERRORS, NULL);
1250 if (!dir) {
1251 errs++;
1252 goto out;
1253 }
1254 while (php_stream_readdir(dir, &d)) {
1255 if ((mflen = snprintf(mfn, sizeof(mfn), "%s/%s", fn, d.d_name)) < 0) {
1256 file_oomem(ms,
1257 strlen(fn) + strlen(d.d_name) + 2);
1258 errs++;
1259 php_stream_closedir(dir);
1260 goto out;
1261 }
1262 if (zend_stat(mfn, &st) == -1 || !S_ISREG(st.st_mode)) {
1263 continue;
1264 }
1265 if (files >= maxfiles) {
1266 size_t mlen;
1267 maxfiles = (maxfiles + 1) * 2;
1268 mlen = maxfiles * sizeof(*filearr);
1269 if ((filearr = CAST(char **,
1270 erealloc(filearr, mlen))) == NULL) {
1271 file_oomem(ms, mlen);
1272 php_stream_closedir(dir);
1273 errs++;
1274 goto out;
1275 }
1276 }
1277 filearr[files++] = estrndup(mfn, (mflen > sizeof(mfn) - 1)? sizeof(mfn) - 1: mflen);
1278 }
1279 php_stream_closedir(dir);
1280 qsort(filearr, files, sizeof(*filearr), cmpstrp);
1281 for (i = 0; i < files; i++) {
1282 load_1(ms, action, filearr[i], &errs, mset);
1283 efree(filearr[i]);
1284 }
1285 efree(filearr);
1286 } else
1287 load_1(ms, action, fn, &errs, mset);
1288 if (errs)
1289 goto out;
1290
1291 for (j = 0; j < MAGIC_SETS; j++) {
1292 /* Set types of tests */
1293 for (i = 0; i < mset[j].count; ) {
1294 if (mset[j].me[i].mp->cont_level != 0) {
1295 i++;
1296 continue;
1297 }
1298 i = set_text_binary(ms, mset[j].me, mset[j].count, i);
1299 }
1300 qsort(mset[j].me, mset[j].count, sizeof(*mset[j].me),
1301 apprentice_sort);
1302
1303 /*
1304 * Make sure that any level 0 "default" line is last
1305 * (if one exists).
1306 */
1307 set_last_default(ms, mset[j].me, mset[j].count);
1308
1309 /* coalesce per file arrays into a single one */
1310 if (coalesce_entries(ms, mset[j].me, mset[j].count,
1311 &map->magic[j], &map->nmagic[j]) == -1) {
1312 errs++;
1313 goto out;
1314 }
1315 }
1316
1317 out:
1318 for (j = 0; j < MAGIC_SETS; j++)
1319 magic_entry_free(mset[j].me, mset[j].count);
1320
1321 if (errs) {
1322 apprentice_unmap(map);
1323 return NULL;
1324 }
1325 return map;
1326 }
1327
1328 /*
1329 * extend the sign bit if the comparison is to be signed
1330 */
1331 protected uint64_t
file_signextend(struct magic_set * ms,struct magic * m,uint64_t v)1332 file_signextend(struct magic_set *ms, struct magic *m, uint64_t v)
1333 {
1334 if (!(m->flag & UNSIGNED)) {
1335 switch(m->type) {
1336 /*
1337 * Do not remove the casts below. They are
1338 * vital. When later compared with the data,
1339 * the sign extension must have happened.
1340 */
1341 case FILE_BYTE:
1342 v = (signed char) v;
1343 break;
1344 case FILE_SHORT:
1345 case FILE_BESHORT:
1346 case FILE_LESHORT:
1347 v = (short) v;
1348 break;
1349 case FILE_DATE:
1350 case FILE_BEDATE:
1351 case FILE_LEDATE:
1352 case FILE_MEDATE:
1353 case FILE_LDATE:
1354 case FILE_BELDATE:
1355 case FILE_LELDATE:
1356 case FILE_MELDATE:
1357 case FILE_LONG:
1358 case FILE_BELONG:
1359 case FILE_LELONG:
1360 case FILE_MELONG:
1361 case FILE_FLOAT:
1362 case FILE_BEFLOAT:
1363 case FILE_LEFLOAT:
1364 v = (int32_t) v;
1365 break;
1366 case FILE_QUAD:
1367 case FILE_BEQUAD:
1368 case FILE_LEQUAD:
1369 case FILE_QDATE:
1370 case FILE_QLDATE:
1371 case FILE_QWDATE:
1372 case FILE_BEQDATE:
1373 case FILE_BEQLDATE:
1374 case FILE_BEQWDATE:
1375 case FILE_LEQDATE:
1376 case FILE_LEQLDATE:
1377 case FILE_LEQWDATE:
1378 case FILE_DOUBLE:
1379 case FILE_BEDOUBLE:
1380 case FILE_LEDOUBLE:
1381 v = (int64_t) v;
1382 break;
1383 case FILE_STRING:
1384 case FILE_PSTRING:
1385 case FILE_BESTRING16:
1386 case FILE_LESTRING16:
1387 case FILE_REGEX:
1388 case FILE_SEARCH:
1389 case FILE_DEFAULT:
1390 case FILE_INDIRECT:
1391 case FILE_NAME:
1392 case FILE_USE:
1393 case FILE_CLEAR:
1394 break;
1395 default:
1396 if (ms->flags & MAGIC_CHECK)
1397 file_magwarn(ms, "cannot happen: m->type=%d\n",
1398 m->type);
1399 return ~0U;
1400 }
1401 }
1402 return v;
1403 }
1404
1405 private int
string_modifier_check(struct magic_set * ms,struct magic * m)1406 string_modifier_check(struct magic_set *ms, struct magic *m)
1407 {
1408 if ((ms->flags & MAGIC_CHECK) == 0)
1409 return 0;
1410
1411 if ((m->type != FILE_REGEX || (m->str_flags & REGEX_LINE_COUNT) == 0) &&
1412 (m->type != FILE_PSTRING && (m->str_flags & PSTRING_LEN) != 0)) {
1413 file_magwarn(ms,
1414 "'/BHhLl' modifiers are only allowed for pascal strings\n");
1415 return -1;
1416 }
1417 switch (m->type) {
1418 case FILE_BESTRING16:
1419 case FILE_LESTRING16:
1420 if (m->str_flags != 0) {
1421 file_magwarn(ms,
1422 "no modifiers allowed for 16-bit strings\n");
1423 return -1;
1424 }
1425 break;
1426 case FILE_STRING:
1427 case FILE_PSTRING:
1428 if ((m->str_flags & REGEX_OFFSET_START) != 0) {
1429 file_magwarn(ms,
1430 "'/%c' only allowed on regex and search\n",
1431 CHAR_REGEX_OFFSET_START);
1432 return -1;
1433 }
1434 break;
1435 case FILE_SEARCH:
1436 if (m->str_range == 0) {
1437 file_magwarn(ms,
1438 "missing range; defaulting to %d\n",
1439 STRING_DEFAULT_RANGE);
1440 m->str_range = STRING_DEFAULT_RANGE;
1441 return -1;
1442 }
1443 break;
1444 case FILE_REGEX:
1445 if ((m->str_flags & STRING_COMPACT_WHITESPACE) != 0) {
1446 file_magwarn(ms, "'/%c' not allowed on regex\n",
1447 CHAR_COMPACT_WHITESPACE);
1448 return -1;
1449 }
1450 if ((m->str_flags & STRING_COMPACT_OPTIONAL_WHITESPACE) != 0) {
1451 file_magwarn(ms, "'/%c' not allowed on regex\n",
1452 CHAR_COMPACT_OPTIONAL_WHITESPACE);
1453 return -1;
1454 }
1455 break;
1456 default:
1457 file_magwarn(ms, "coding error: m->type=%d\n",
1458 m->type);
1459 return -1;
1460 }
1461 return 0;
1462 }
1463
1464 private int
get_op(char c)1465 get_op(char c)
1466 {
1467 switch (c) {
1468 case '&':
1469 return FILE_OPAND;
1470 case '|':
1471 return FILE_OPOR;
1472 case '^':
1473 return FILE_OPXOR;
1474 case '+':
1475 return FILE_OPADD;
1476 case '-':
1477 return FILE_OPMINUS;
1478 case '*':
1479 return FILE_OPMULTIPLY;
1480 case '/':
1481 return FILE_OPDIVIDE;
1482 case '%':
1483 return FILE_OPMODULO;
1484 default:
1485 return -1;
1486 }
1487 }
1488
1489 #ifdef ENABLE_CONDITIONALS
1490 private int
get_cond(const char * l,const char ** t)1491 get_cond(const char *l, const char **t)
1492 {
1493 static const struct cond_tbl_s {
1494 char name[8];
1495 size_t len;
1496 int cond;
1497 } cond_tbl[] = {
1498 { "if", 2, COND_IF },
1499 { "elif", 4, COND_ELIF },
1500 { "else", 4, COND_ELSE },
1501 { "", 0, COND_NONE },
1502 };
1503 const struct cond_tbl_s *p;
1504
1505 for (p = cond_tbl; p->len; p++) {
1506 if (strncmp(l, p->name, p->len) == 0 &&
1507 isspace((unsigned char)l[p->len])) {
1508 if (t)
1509 *t = l + p->len;
1510 break;
1511 }
1512 }
1513 return p->cond;
1514 }
1515
1516 private int
check_cond(struct magic_set * ms,int cond,uint32_t cont_level)1517 check_cond(struct magic_set *ms, int cond, uint32_t cont_level)
1518 {
1519 int last_cond;
1520 last_cond = ms->c.li[cont_level].last_cond;
1521
1522 switch (cond) {
1523 case COND_IF:
1524 if (last_cond != COND_NONE && last_cond != COND_ELIF) {
1525 if (ms->flags & MAGIC_CHECK)
1526 file_magwarn(ms, "syntax error: `if'");
1527 return -1;
1528 }
1529 last_cond = COND_IF;
1530 break;
1531
1532 case COND_ELIF:
1533 if (last_cond != COND_IF && last_cond != COND_ELIF) {
1534 if (ms->flags & MAGIC_CHECK)
1535 file_magwarn(ms, "syntax error: `elif'");
1536 return -1;
1537 }
1538 last_cond = COND_ELIF;
1539 break;
1540
1541 case COND_ELSE:
1542 if (last_cond != COND_IF && last_cond != COND_ELIF) {
1543 if (ms->flags & MAGIC_CHECK)
1544 file_magwarn(ms, "syntax error: `else'");
1545 return -1;
1546 }
1547 last_cond = COND_NONE;
1548 break;
1549
1550 case COND_NONE:
1551 last_cond = COND_NONE;
1552 break;
1553 }
1554
1555 ms->c.li[cont_level].last_cond = last_cond;
1556 return 0;
1557 }
1558 #endif /* ENABLE_CONDITIONALS */
1559
1560 private int
parse_indirect_modifier(struct magic_set * ms,struct magic * m,const char ** lp)1561 parse_indirect_modifier(struct magic_set *ms, struct magic *m, const char **lp)
1562 {
1563 const char *l = *lp;
1564
1565 while (!isspace((unsigned char)*++l))
1566 switch (*l) {
1567 case CHAR_INDIRECT_RELATIVE:
1568 m->str_flags |= INDIRECT_RELATIVE;
1569 break;
1570 default:
1571 if (ms->flags & MAGIC_CHECK)
1572 file_magwarn(ms, "indirect modifier `%c' "
1573 "invalid", *l);
1574 *lp = l;
1575 return -1;
1576 }
1577 *lp = l;
1578 return 0;
1579 }
1580
1581 private void
parse_op_modifier(struct magic_set * ms,struct magic * m,const char ** lp,int op)1582 parse_op_modifier(struct magic_set *ms, struct magic *m, const char **lp,
1583 int op)
1584 {
1585 const char *l = *lp;
1586 char *t;
1587 uint64_t val;
1588
1589 ++l;
1590 m->mask_op |= op;
1591 val = (uint64_t)strtoull(l, &t, 0);
1592 l = t;
1593 m->num_mask = file_signextend(ms, m, val);
1594 eatsize(&l);
1595 *lp = l;
1596 }
1597
1598 private int
parse_string_modifier(struct magic_set * ms,struct magic * m,const char ** lp)1599 parse_string_modifier(struct magic_set *ms, struct magic *m, const char **lp)
1600 {
1601 const char *l = *lp;
1602 char *t;
1603 int have_range = 0;
1604
1605 while (!isspace((unsigned char)*++l)) {
1606 switch (*l) {
1607 case '0': case '1': case '2':
1608 case '3': case '4': case '5':
1609 case '6': case '7': case '8':
1610 case '9':
1611 if (have_range && (ms->flags & MAGIC_CHECK))
1612 file_magwarn(ms, "multiple ranges");
1613 have_range = 1;
1614 m->str_range = CAST(uint32_t, strtoul(l, &t, 0));
1615 if (m->str_range == 0)
1616 file_magwarn(ms, "zero range");
1617 l = t - 1;
1618 break;
1619 case CHAR_COMPACT_WHITESPACE:
1620 m->str_flags |= STRING_COMPACT_WHITESPACE;
1621 break;
1622 case CHAR_COMPACT_OPTIONAL_WHITESPACE:
1623 m->str_flags |= STRING_COMPACT_OPTIONAL_WHITESPACE;
1624 break;
1625 case CHAR_IGNORE_LOWERCASE:
1626 m->str_flags |= STRING_IGNORE_LOWERCASE;
1627 break;
1628 case CHAR_IGNORE_UPPERCASE:
1629 m->str_flags |= STRING_IGNORE_UPPERCASE;
1630 break;
1631 case CHAR_REGEX_OFFSET_START:
1632 m->str_flags |= REGEX_OFFSET_START;
1633 break;
1634 case CHAR_BINTEST:
1635 m->str_flags |= STRING_BINTEST;
1636 break;
1637 case CHAR_TEXTTEST:
1638 m->str_flags |= STRING_TEXTTEST;
1639 break;
1640 case CHAR_TRIM:
1641 m->str_flags |= STRING_TRIM;
1642 break;
1643 case CHAR_PSTRING_1_LE:
1644 #define SET_LENGTH(a) m->str_flags = (m->str_flags & ~PSTRING_LEN) | (a)
1645 if (m->type != FILE_PSTRING)
1646 goto bad;
1647 SET_LENGTH(PSTRING_1_LE);
1648 break;
1649 case CHAR_PSTRING_2_BE:
1650 if (m->type != FILE_PSTRING)
1651 goto bad;
1652 SET_LENGTH(PSTRING_2_BE);
1653 break;
1654 case CHAR_PSTRING_2_LE:
1655 if (m->type != FILE_PSTRING)
1656 goto bad;
1657 SET_LENGTH(PSTRING_2_LE);
1658 break;
1659 case CHAR_PSTRING_4_BE:
1660 if (m->type != FILE_PSTRING)
1661 goto bad;
1662 SET_LENGTH(PSTRING_4_BE);
1663 break;
1664 case CHAR_PSTRING_4_LE:
1665 switch (m->type) {
1666 case FILE_PSTRING:
1667 case FILE_REGEX:
1668 break;
1669 default:
1670 goto bad;
1671 }
1672 SET_LENGTH(PSTRING_4_LE);
1673 break;
1674 case CHAR_PSTRING_LENGTH_INCLUDES_ITSELF:
1675 if (m->type != FILE_PSTRING)
1676 goto bad;
1677 m->str_flags |= PSTRING_LENGTH_INCLUDES_ITSELF;
1678 break;
1679 default:
1680 bad:
1681 if (ms->flags & MAGIC_CHECK)
1682 file_magwarn(ms, "string modifier `%c' "
1683 "invalid", *l);
1684 goto out;
1685 }
1686 /* allow multiple '/' for readability */
1687 if (l[1] == '/' && !isspace((unsigned char)l[2]))
1688 l++;
1689 }
1690 if (string_modifier_check(ms, m) == -1)
1691 goto out;
1692 *lp = l;
1693 return 0;
1694 out:
1695 *lp = l;
1696 return -1;
1697 }
1698
1699 /*
1700 * parse one line from magic file, put into magic[index++] if valid
1701 */
1702 private int
parse(struct magic_set * ms,struct magic_entry * me,const char * line,size_t lineno,int action)1703 parse(struct magic_set *ms, struct magic_entry *me, const char *line,
1704 size_t lineno, int action)
1705 {
1706 #ifdef ENABLE_CONDITIONALS
1707 static uint32_t last_cont_level = 0;
1708 #endif
1709 size_t i;
1710 struct magic *m;
1711 const char *l = line;
1712 char *t;
1713 int op;
1714 uint32_t cont_level;
1715 int32_t diff;
1716
1717 cont_level = 0;
1718
1719 /*
1720 * Parse the offset.
1721 */
1722 while (*l == '>') {
1723 ++l; /* step over */
1724 cont_level++;
1725 }
1726 #ifdef ENABLE_CONDITIONALS
1727 if (cont_level == 0 || cont_level > last_cont_level)
1728 if (file_check_mem(ms, cont_level) == -1)
1729 return -1;
1730 last_cont_level = cont_level;
1731 #endif
1732 if (cont_level != 0) {
1733 if (me->mp == NULL) {
1734 file_magerror(ms, "No current entry for continuation");
1735 return -1;
1736 }
1737 if (me->cont_count == 0) {
1738 file_magerror(ms, "Continuations present with 0 count");
1739 return -1;
1740 }
1741 m = &me->mp[me->cont_count - 1];
1742 diff = (int32_t)cont_level - (int32_t)m->cont_level;
1743 if (diff > 1)
1744 file_magwarn(ms, "New continuation level %u is more "
1745 "than one larger than current level %u", cont_level,
1746 m->cont_level);
1747 if (me->cont_count == me->max_count) {
1748 struct magic *nm;
1749 size_t cnt = me->max_count + ALLOC_CHUNK;
1750 if ((nm = CAST(struct magic *, erealloc(me->mp,
1751 sizeof(*nm) * cnt))) == NULL) {
1752 file_oomem(ms, sizeof(*nm) * cnt);
1753 return -1;
1754 }
1755 me->mp = m = nm;
1756 me->max_count = CAST(uint32_t, cnt);
1757 }
1758 m = &me->mp[me->cont_count++];
1759 (void)memset(m, 0, sizeof(*m));
1760 m->cont_level = cont_level;
1761 } else {
1762 static const size_t len = sizeof(*m) * ALLOC_CHUNK;
1763 if (me->mp != NULL)
1764 return 1;
1765 if ((m = CAST(struct magic *, emalloc(len))) == NULL) {
1766 file_oomem(ms, len);
1767 return -1;
1768 }
1769 me->mp = m;
1770 me->max_count = ALLOC_CHUNK;
1771 (void)memset(m, 0, sizeof(*m));
1772 m->factor_op = FILE_FACTOR_OP_NONE;
1773 m->cont_level = 0;
1774 me->cont_count = 1;
1775 }
1776 m->lineno = CAST(uint32_t, lineno);
1777
1778 if (*l == '&') { /* m->cont_level == 0 checked below. */
1779 ++l; /* step over */
1780 m->flag |= OFFADD;
1781 }
1782 if (*l == '(') {
1783 ++l; /* step over */
1784 m->flag |= INDIR;
1785 if (m->flag & OFFADD)
1786 m->flag = (m->flag & ~OFFADD) | INDIROFFADD;
1787
1788 if (*l == '&') { /* m->cont_level == 0 checked below */
1789 ++l; /* step over */
1790 m->flag |= OFFADD;
1791 }
1792 }
1793 /* Indirect offsets are not valid at level 0. */
1794 if (m->cont_level == 0 && (m->flag & (OFFADD | INDIROFFADD)))
1795 if (ms->flags & MAGIC_CHECK)
1796 file_magwarn(ms, "relative offset at level 0");
1797
1798 /* get offset, then skip over it */
1799 m->offset = (uint32_t)strtoul(l, &t, 0);
1800 if (l == t)
1801 if (ms->flags & MAGIC_CHECK)
1802 file_magwarn(ms, "offset `%s' invalid", l);
1803 l = t;
1804
1805 if (m->flag & INDIR) {
1806 m->in_type = FILE_LONG;
1807 m->in_offset = 0;
1808 /*
1809 * read [.lbs][+-]nnnnn)
1810 */
1811 if (*l == '.') {
1812 l++;
1813 switch (*l) {
1814 case 'l':
1815 m->in_type = FILE_LELONG;
1816 break;
1817 case 'L':
1818 m->in_type = FILE_BELONG;
1819 break;
1820 case 'm':
1821 m->in_type = FILE_MELONG;
1822 break;
1823 case 'h':
1824 case 's':
1825 m->in_type = FILE_LESHORT;
1826 break;
1827 case 'H':
1828 case 'S':
1829 m->in_type = FILE_BESHORT;
1830 break;
1831 case 'c':
1832 case 'b':
1833 case 'C':
1834 case 'B':
1835 m->in_type = FILE_BYTE;
1836 break;
1837 case 'e':
1838 case 'f':
1839 case 'g':
1840 m->in_type = FILE_LEDOUBLE;
1841 break;
1842 case 'E':
1843 case 'F':
1844 case 'G':
1845 m->in_type = FILE_BEDOUBLE;
1846 break;
1847 case 'i':
1848 m->in_type = FILE_LEID3;
1849 break;
1850 case 'I':
1851 m->in_type = FILE_BEID3;
1852 break;
1853 default:
1854 if (ms->flags & MAGIC_CHECK)
1855 file_magwarn(ms,
1856 "indirect offset type `%c' invalid",
1857 *l);
1858 break;
1859 }
1860 l++;
1861 }
1862
1863 m->in_op = 0;
1864 if (*l == '~') {
1865 m->in_op |= FILE_OPINVERSE;
1866 l++;
1867 }
1868 if ((op = get_op(*l)) != -1) {
1869 m->in_op |= op;
1870 l++;
1871 }
1872 if (*l == '(') {
1873 m->in_op |= FILE_OPINDIRECT;
1874 l++;
1875 }
1876 if (isdigit((unsigned char)*l) || *l == '-') {
1877 m->in_offset = (int32_t)strtol(l, &t, 0);
1878 if (l == t)
1879 if (ms->flags & MAGIC_CHECK)
1880 file_magwarn(ms,
1881 "in_offset `%s' invalid", l);
1882 l = t;
1883 }
1884 if (*l++ != ')' ||
1885 ((m->in_op & FILE_OPINDIRECT) && *l++ != ')'))
1886 if (ms->flags & MAGIC_CHECK)
1887 file_magwarn(ms,
1888 "missing ')' in indirect offset");
1889 }
1890 EATAB;
1891
1892 #ifdef ENABLE_CONDITIONALS
1893 m->cond = get_cond(l, &l);
1894 if (check_cond(ms, m->cond, cont_level) == -1)
1895 return -1;
1896
1897 EATAB;
1898 #endif
1899
1900 /*
1901 * Parse the type.
1902 */
1903 if (*l == 'u') {
1904 /*
1905 * Try it as a keyword type prefixed by "u"; match what
1906 * follows the "u". If that fails, try it as an SUS
1907 * integer type.
1908 */
1909 m->type = get_type(type_tbl, l + 1, &l);
1910 if (m->type == FILE_INVALID) {
1911 /*
1912 * Not a keyword type; parse it as an SUS type,
1913 * 'u' possibly followed by a number or C/S/L.
1914 */
1915 m->type = get_standard_integer_type(l, &l);
1916 }
1917 /* It's unsigned. */
1918 if (m->type != FILE_INVALID)
1919 m->flag |= UNSIGNED;
1920 } else {
1921 /*
1922 * Try it as a keyword type. If that fails, try it as
1923 * an SUS integer type if it begins with "d" or as an
1924 * SUS string type if it begins with "s". In any case,
1925 * it's not unsigned.
1926 */
1927 m->type = get_type(type_tbl, l, &l);
1928 if (m->type == FILE_INVALID) {
1929 /*
1930 * Not a keyword type; parse it as an SUS type,
1931 * either 'd' possibly followed by a number or
1932 * C/S/L, or just 's'.
1933 */
1934 if (*l == 'd')
1935 m->type = get_standard_integer_type(l, &l);
1936 else if (*l == 's' && !isalpha((unsigned char)l[1])) {
1937 m->type = FILE_STRING;
1938 ++l;
1939 }
1940 }
1941 }
1942
1943 if (m->type == FILE_INVALID) {
1944 /* Not found - try it as a special keyword. */
1945 m->type = get_type(special_tbl, l, &l);
1946 }
1947
1948 if (m->type == FILE_INVALID) {
1949 if (ms->flags & MAGIC_CHECK)
1950 file_magwarn(ms, "type `%s' invalid", l);
1951 return -1;
1952 }
1953
1954 /* New-style anding: "0 byte&0x80 =0x80 dynamically linked" */
1955 /* New and improved: ~ & | ^ + - * / % -- exciting, isn't it? */
1956
1957 m->mask_op = 0;
1958 if (*l == '~') {
1959 if (!IS_LIBMAGIC_STRING(m->type))
1960 m->mask_op |= FILE_OPINVERSE;
1961 else if (ms->flags & MAGIC_CHECK)
1962 file_magwarn(ms, "'~' invalid for string types");
1963 ++l;
1964 }
1965 m->str_range = 0;
1966 m->str_flags = m->type == FILE_PSTRING ? PSTRING_1_LE : 0;
1967 if ((op = get_op(*l)) != -1) {
1968 if (IS_LIBMAGIC_STRING(m->type)) {
1969 int r;
1970
1971 if (op != FILE_OPDIVIDE) {
1972 if (ms->flags & MAGIC_CHECK)
1973 file_magwarn(ms,
1974 "invalid string/indirect op: "
1975 "`%c'", *t);
1976 return -1;
1977 }
1978
1979 if (m->type == FILE_INDIRECT)
1980 r = parse_indirect_modifier(ms, m, &l);
1981 else
1982 r = parse_string_modifier(ms, m, &l);
1983 if (r == -1)
1984 return -1;
1985 } else
1986 parse_op_modifier(ms, m, &l, op);
1987 }
1988
1989 /*
1990 * We used to set mask to all 1's here, instead let's just not do
1991 * anything if mask = 0 (unless you have a better idea)
1992 */
1993 EATAB;
1994
1995 switch (*l) {
1996 case '>':
1997 case '<':
1998 m->reln = *l;
1999 ++l;
2000 if (*l == '=') {
2001 if (ms->flags & MAGIC_CHECK) {
2002 file_magwarn(ms, "%c= not supported",
2003 m->reln);
2004 return -1;
2005 }
2006 ++l;
2007 }
2008 break;
2009 /* Old-style anding: "0 byte &0x80 dynamically linked" */
2010 case '&':
2011 case '^':
2012 case '=':
2013 m->reln = *l;
2014 ++l;
2015 if (*l == '=') {
2016 /* HP compat: ignore &= etc. */
2017 ++l;
2018 }
2019 break;
2020 case '!':
2021 m->reln = *l;
2022 ++l;
2023 break;
2024 default:
2025 m->reln = '='; /* the default relation */
2026 if (*l == 'x' && ((isascii((unsigned char)l[1]) &&
2027 isspace((unsigned char)l[1])) || !l[1])) {
2028 m->reln = *l;
2029 ++l;
2030 }
2031 break;
2032 }
2033 /*
2034 * Grab the value part, except for an 'x' reln.
2035 */
2036 if (m->reln != 'x' && getvalue(ms, m, &l, action))
2037 return -1;
2038
2039 /*
2040 * TODO finish this macro and start using it!
2041 * #define offsetcheck {if (offset > HOWMANY-1)
2042 * magwarn("offset too big"); }
2043 */
2044
2045 /*
2046 * Now get last part - the description
2047 */
2048 EATAB;
2049 if (l[0] == '\b') {
2050 ++l;
2051 m->flag |= NOSPACE;
2052 } else if ((l[0] == '\\') && (l[1] == 'b')) {
2053 ++l;
2054 ++l;
2055 m->flag |= NOSPACE;
2056 }
2057 for (i = 0; (m->desc[i++] = *l++) != '\0' && i < sizeof(m->desc); )
2058 continue;
2059 if (i == sizeof(m->desc)) {
2060 m->desc[sizeof(m->desc) - 1] = '\0';
2061 if (ms->flags & MAGIC_CHECK)
2062 file_magwarn(ms, "description `%s' truncated", m->desc);
2063 }
2064
2065 /*
2066 * We only do this check while compiling, or if any of the magic
2067 * files were not compiled.
2068 */
2069 if (ms->flags & MAGIC_CHECK) {
2070 if (check_format(ms, m) == -1)
2071 return -1;
2072 }
2073 m->mimetype[0] = '\0'; /* initialise MIME type to none */
2074 return 0;
2075 }
2076
2077 /*
2078 * parse a STRENGTH annotation line from magic file, put into magic[index - 1]
2079 * if valid
2080 */
2081 private int
parse_strength(struct magic_set * ms,struct magic_entry * me,const char * line)2082 parse_strength(struct magic_set *ms, struct magic_entry *me, const char *line)
2083 {
2084 const char *l = line;
2085 char *el;
2086 unsigned long factor;
2087 struct magic *m = &me->mp[0];
2088
2089 if (m->factor_op != FILE_FACTOR_OP_NONE) {
2090 file_magwarn(ms,
2091 "Current entry already has a strength type: %c %d",
2092 m->factor_op, m->factor);
2093 return -1;
2094 }
2095 if (m->type == FILE_NAME) {
2096 file_magwarn(ms, "%s: Strength setting is not supported in "
2097 "\"name\" magic entries", m->value.s);
2098 return -1;
2099 }
2100 EATAB;
2101 switch (*l) {
2102 case FILE_FACTOR_OP_NONE:
2103 case FILE_FACTOR_OP_PLUS:
2104 case FILE_FACTOR_OP_MINUS:
2105 case FILE_FACTOR_OP_TIMES:
2106 case FILE_FACTOR_OP_DIV:
2107 m->factor_op = *l++;
2108 break;
2109 default:
2110 file_magwarn(ms, "Unknown factor op `%c'", *l);
2111 return -1;
2112 }
2113 EATAB;
2114 factor = strtoul(l, &el, 0);
2115 if (factor > 255) {
2116 file_magwarn(ms, "Too large factor `%lu'", factor);
2117 goto out;
2118 }
2119 if (*el && !isspace((unsigned char)*el)) {
2120 file_magwarn(ms, "Bad factor `%s'", l);
2121 goto out;
2122 }
2123 m->factor = (uint8_t)factor;
2124 if (m->factor == 0 && m->factor_op == FILE_FACTOR_OP_DIV) {
2125 file_magwarn(ms, "Cannot have factor op `%c' and factor %u",
2126 m->factor_op, m->factor);
2127 goto out;
2128 }
2129 return 0;
2130 out:
2131 m->factor_op = FILE_FACTOR_OP_NONE;
2132 m->factor = 0;
2133 return -1;
2134 }
2135
2136 private int
goodchar(unsigned char x,const char * extra)2137 goodchar(unsigned char x, const char *extra)
2138 {
2139 return (isascii(x) && isalnum(x)) || strchr(extra, x);
2140 }
2141
2142 private int
parse_extra(struct magic_set * ms,struct magic_entry * me,const char * line,zend_off_t off,size_t len,const char * name,const char * extra,int nt)2143 parse_extra(struct magic_set *ms, struct magic_entry *me, const char *line,
2144 zend_off_t off, size_t len, const char *name, const char *extra, int nt)
2145 {
2146 size_t i;
2147 const char *l = line;
2148 struct magic *m = &me->mp[me->cont_count == 0 ? 0 : me->cont_count - 1];
2149 char *buf = CAST(char *, CAST(void *, m)) + off;
2150
2151 if (buf[0] != '\0') {
2152 len = nt ? strlen(buf) : len;
2153 file_magwarn(ms, "Current entry already has a %s type "
2154 "`%.*s', new type `%s'", name, (int)len, buf, l);
2155 return -1;
2156 }
2157
2158 if (*m->desc == '\0') {
2159 file_magwarn(ms, "Current entry does not yet have a "
2160 "description for adding a %s type", name);
2161 return -1;
2162 }
2163
2164 EATAB;
2165 for (i = 0; *l && i < len && goodchar(*l, extra); buf[i++] = *l++)
2166 continue;
2167
2168 if (i == len && *l) {
2169 if (nt)
2170 buf[len - 1] = '\0';
2171 if (ms->flags & MAGIC_CHECK)
2172 file_magwarn(ms, "%s type `%s' truncated %"
2173 SIZE_T_FORMAT "u", name, line, i);
2174 } else {
2175 if (!isspace((unsigned char)*l) && !goodchar(*l, extra))
2176 file_magwarn(ms, "%s type `%s' has bad char '%c'",
2177 name, line, *l);
2178 if (nt)
2179 buf[i] = '\0';
2180 }
2181
2182 if (i > 0)
2183 return 0;
2184
2185 file_magerror(ms, "Bad magic entry '%s'", line);
2186 return -1;
2187 }
2188
2189 /*
2190 * Parse an Apple CREATOR/TYPE annotation from magic file and put it into
2191 * magic[index - 1]
2192 */
2193 private int
parse_apple(struct magic_set * ms,struct magic_entry * me,const char * line)2194 parse_apple(struct magic_set *ms, struct magic_entry *me, const char *line)
2195 {
2196 struct magic *m = &me->mp[0];
2197
2198 return parse_extra(ms, me, line,
2199 CAST(off_t, offsetof(struct magic, apple)),
2200 sizeof(m->apple), "APPLE", "!+-./", 0);
2201 }
2202
2203 /*
2204 * parse a MIME annotation line from magic file, put into magic[index - 1]
2205 * if valid
2206 */
2207 private int
parse_mime(struct magic_set * ms,struct magic_entry * me,const char * line)2208 parse_mime(struct magic_set *ms, struct magic_entry *me, const char *line)
2209 {
2210 struct magic *m = &me->mp[0];
2211
2212 return parse_extra(ms, me, line,
2213 CAST(zend_off_t, offsetof(struct magic, mimetype)),
2214 sizeof(m->mimetype), "MIME", "+-/.", 1);
2215 }
2216
2217 private int
check_format_type(const char * ptr,int type)2218 check_format_type(const char *ptr, int type)
2219 {
2220 int quad = 0, h;
2221 if (*ptr == '\0') {
2222 /* Missing format string; bad */
2223 return -1;
2224 }
2225
2226 switch (file_formats[type]) {
2227 case FILE_FMT_QUAD:
2228 quad = 1;
2229 /*FALLTHROUGH*/
2230 case FILE_FMT_NUM:
2231 if (quad == 0) {
2232 switch (type) {
2233 case FILE_BYTE:
2234 h = 2;
2235 break;
2236 case FILE_SHORT:
2237 case FILE_BESHORT:
2238 case FILE_LESHORT:
2239 h = 1;
2240 break;
2241 case FILE_LONG:
2242 case FILE_BELONG:
2243 case FILE_LELONG:
2244 case FILE_MELONG:
2245 case FILE_LEID3:
2246 case FILE_BEID3:
2247 case FILE_INDIRECT:
2248 h = 0;
2249 break;
2250 default:
2251 abort();
2252 }
2253 } else
2254 h = 0;
2255 if (*ptr == '-')
2256 ptr++;
2257 if (*ptr == '.')
2258 ptr++;
2259 while (isdigit((unsigned char)*ptr)) ptr++;
2260 if (*ptr == '.')
2261 ptr++;
2262 while (isdigit((unsigned char)*ptr)) ptr++;
2263 if (quad) {
2264 if (*ptr++ != 'l')
2265 return -1;
2266 if (*ptr++ != 'l')
2267 return -1;
2268 }
2269
2270 switch (*ptr++) {
2271 #ifdef STRICT_FORMAT /* "long" formats are int formats for us */
2272 /* so don't accept the 'l' modifier */
2273 case 'l':
2274 switch (*ptr++) {
2275 case 'i':
2276 case 'd':
2277 case 'u':
2278 case 'o':
2279 case 'x':
2280 case 'X':
2281 return h != 0 ? -1 : 0;
2282 default:
2283 return -1;
2284 }
2285
2286 /*
2287 * Don't accept h and hh modifiers. They make writing
2288 * magic entries more complicated, for very little benefit
2289 */
2290 case 'h':
2291 if (h-- <= 0)
2292 return -1;
2293 switch (*ptr++) {
2294 case 'h':
2295 if (h-- <= 0)
2296 return -1;
2297 switch (*ptr++) {
2298 case 'i':
2299 case 'd':
2300 case 'u':
2301 case 'o':
2302 case 'x':
2303 case 'X':
2304 return 0;
2305 default:
2306 return -1;
2307 }
2308 case 'i':
2309 case 'd':
2310 case 'u':
2311 case 'o':
2312 case 'x':
2313 case 'X':
2314 return h != 0 ? -1 : 0;
2315 default:
2316 return -1;
2317 }
2318 #endif
2319 case 'c':
2320 return h != 2 ? -1 : 0;
2321 case 'i':
2322 case 'd':
2323 case 'u':
2324 case 'o':
2325 case 'x':
2326 case 'X':
2327 #ifdef STRICT_FORMAT
2328 return h != 0 ? -1 : 0;
2329 #else
2330 return 0;
2331 #endif
2332 default:
2333 return -1;
2334 }
2335
2336 case FILE_FMT_FLOAT:
2337 case FILE_FMT_DOUBLE:
2338 if (*ptr == '-')
2339 ptr++;
2340 if (*ptr == '.')
2341 ptr++;
2342 while (isdigit((unsigned char)*ptr)) ptr++;
2343 if (*ptr == '.')
2344 ptr++;
2345 while (isdigit((unsigned char)*ptr)) ptr++;
2346
2347 switch (*ptr++) {
2348 case 'e':
2349 case 'E':
2350 case 'f':
2351 case 'F':
2352 case 'g':
2353 case 'G':
2354 return 0;
2355
2356 default:
2357 return -1;
2358 }
2359
2360
2361 case FILE_FMT_STR:
2362 if (*ptr == '-')
2363 ptr++;
2364 while (isdigit((unsigned char )*ptr))
2365 ptr++;
2366 if (*ptr == '.') {
2367 ptr++;
2368 while (isdigit((unsigned char )*ptr))
2369 ptr++;
2370 }
2371
2372 switch (*ptr++) {
2373 case 's':
2374 return 0;
2375 default:
2376 return -1;
2377 }
2378
2379 default:
2380 /* internal error */
2381 abort();
2382 }
2383 /*NOTREACHED*/
2384 return -1;
2385 }
2386
2387 /*
2388 * Check that the optional printf format in description matches
2389 * the type of the magic.
2390 */
2391 private int
check_format(struct magic_set * ms,struct magic * m)2392 check_format(struct magic_set *ms, struct magic *m)
2393 {
2394 char *ptr;
2395
2396 for (ptr = m->desc; *ptr; ptr++)
2397 if (*ptr == '%')
2398 break;
2399 if (*ptr == '\0') {
2400 /* No format string; ok */
2401 return 1;
2402 }
2403
2404 assert(file_nformats == file_nnames);
2405
2406 if (m->type >= file_nformats) {
2407 file_magwarn(ms, "Internal error inconsistency between "
2408 "m->type and format strings");
2409 return -1;
2410 }
2411 if (file_formats[m->type] == FILE_FMT_NONE) {
2412 file_magwarn(ms, "No format string for `%s' with description "
2413 "`%s'", m->desc, file_names[m->type]);
2414 return -1;
2415 }
2416
2417 ptr++;
2418 if (check_format_type(ptr, m->type) == -1) {
2419 /*
2420 * TODO: this error message is unhelpful if the format
2421 * string is not one character long
2422 */
2423 file_magwarn(ms, "Printf format `%c' is not valid for type "
2424 "`%s' in description `%s'", *ptr ? *ptr : '?',
2425 file_names[m->type], m->desc);
2426 return -1;
2427 }
2428
2429 for (; *ptr; ptr++) {
2430 if (*ptr == '%') {
2431 file_magwarn(ms,
2432 "Too many format strings (should have at most one) "
2433 "for `%s' with description `%s'",
2434 file_names[m->type], m->desc);
2435 return -1;
2436 }
2437 }
2438 return 0;
2439 }
2440
2441 /*
2442 * Read a numeric value from a pointer, into the value union of a magic
2443 * pointer, according to the magic type. Update the string pointer to point
2444 * just after the number read. Return 0 for success, non-zero for failure.
2445 */
2446 private int
getvalue(struct magic_set * ms,struct magic * m,const char ** p,int action)2447 getvalue(struct magic_set *ms, struct magic *m, const char **p, int action)
2448 {
2449 switch (m->type) {
2450 case FILE_BESTRING16:
2451 case FILE_LESTRING16:
2452 case FILE_STRING:
2453 case FILE_PSTRING:
2454 case FILE_REGEX:
2455 case FILE_SEARCH:
2456 case FILE_NAME:
2457 case FILE_USE:
2458 *p = getstr(ms, m, *p, action == FILE_COMPILE);
2459 if (*p == NULL) {
2460 if (ms->flags & MAGIC_CHECK)
2461 file_magwarn(ms, "cannot get string from `%s'",
2462 m->value.s);
2463 return -1;
2464 }
2465 if (m->type == FILE_REGEX) {
2466 /* XXX do we need this? */
2467 /*zval pattern;
2468 int options = 0;
2469 pcre_cache_entry *pce;
2470
2471 convert_libmagic_pattern(&pattern, m->value.s, strlen(m->value.s), options);
2472
2473 if ((pce = pcre_get_compiled_regex_cache(Z_STR(pattern))) == NULL) {
2474 return -1;
2475 }
2476
2477 return 0;*/
2478 }
2479 return 0;
2480 case FILE_FLOAT:
2481 case FILE_BEFLOAT:
2482 case FILE_LEFLOAT:
2483 if (m->reln != 'x') {
2484 char *ep;
2485 #ifdef HAVE_STRTOF
2486 m->value.f = strtof(*p, &ep);
2487 #else
2488 m->value.f = (float)strtod(*p, &ep);
2489 #endif
2490 *p = ep;
2491 }
2492 return 0;
2493 case FILE_DOUBLE:
2494 case FILE_BEDOUBLE:
2495 case FILE_LEDOUBLE:
2496 if (m->reln != 'x') {
2497 char *ep;
2498 m->value.d = strtod(*p, &ep);
2499 *p = ep;
2500 }
2501 return 0;
2502 default:
2503 if (m->reln != 'x') {
2504 char *ep;
2505 m->value.q = file_signextend(ms, m,
2506 (uint64_t)strtoull(*p, &ep, 0));
2507 *p = ep;
2508 eatsize(p);
2509 }
2510 return 0;
2511 }
2512 }
2513
2514 /*
2515 * Convert a string containing C character escapes. Stop at an unescaped
2516 * space or tab.
2517 * Copy the converted version to "m->value.s", and the length in m->vallen.
2518 * Return updated scan pointer as function result. Warn if set.
2519 */
2520 private const char *
getstr(struct magic_set * ms,struct magic * m,const char * s,int warn)2521 getstr(struct magic_set *ms, struct magic *m, const char *s, int warn)
2522 {
2523 const char *origs = s;
2524 char *p = m->value.s;
2525 size_t plen = sizeof(m->value.s);
2526 char *origp = p;
2527 char *pmax = p + plen - 1;
2528 int c;
2529 int val;
2530
2531 while ((c = *s++) != '\0') {
2532 if (isspace((unsigned char) c))
2533 break;
2534 if (p >= pmax) {
2535 file_error(ms, 0, "string too long: `%s'", origs);
2536 return NULL;
2537 }
2538 if (c == '\\') {
2539 switch(c = *s++) {
2540
2541 case '\0':
2542 if (warn)
2543 file_magwarn(ms, "incomplete escape");
2544 goto out;
2545
2546 case '\t':
2547 if (warn) {
2548 file_magwarn(ms,
2549 "escaped tab found, use \\t instead");
2550 warn = 0; /* already did */
2551 }
2552 /*FALLTHROUGH*/
2553 default:
2554 if (warn) {
2555 if (isprint((unsigned char)c)) {
2556 /* Allow escaping of
2557 * ``relations'' */
2558 if (strchr("<>&^=!", c) == NULL
2559 && (m->type != FILE_REGEX ||
2560 strchr("[]().*?^$|{}", c)
2561 == NULL)) {
2562 file_magwarn(ms, "no "
2563 "need to escape "
2564 "`%c'", c);
2565 }
2566 } else {
2567 file_magwarn(ms,
2568 "unknown escape sequence: "
2569 "\\%03o", c);
2570 }
2571 }
2572 /*FALLTHROUGH*/
2573 /* space, perhaps force people to use \040? */
2574 case ' ':
2575 #if 0
2576 /*
2577 * Other things people escape, but shouldn't need to,
2578 * so we disallow them
2579 */
2580 case '\'':
2581 case '"':
2582 case '?':
2583 #endif
2584 /* Relations */
2585 case '>':
2586 case '<':
2587 case '&':
2588 case '^':
2589 case '=':
2590 case '!':
2591 /* and baskslash itself */
2592 case '\\':
2593 *p++ = (char) c;
2594 break;
2595
2596 case 'a':
2597 *p++ = '\a';
2598 break;
2599
2600 case 'b':
2601 *p++ = '\b';
2602 break;
2603
2604 case 'f':
2605 *p++ = '\f';
2606 break;
2607
2608 case 'n':
2609 *p++ = '\n';
2610 break;
2611
2612 case 'r':
2613 *p++ = '\r';
2614 break;
2615
2616 case 't':
2617 *p++ = '\t';
2618 break;
2619
2620 case 'v':
2621 *p++ = '\v';
2622 break;
2623
2624 /* \ and up to 3 octal digits */
2625 case '0':
2626 case '1':
2627 case '2':
2628 case '3':
2629 case '4':
2630 case '5':
2631 case '6':
2632 case '7':
2633 val = c - '0';
2634 c = *s++; /* try for 2 */
2635 if (c >= '0' && c <= '7') {
2636 val = (val << 3) | (c - '0');
2637 c = *s++; /* try for 3 */
2638 if (c >= '0' && c <= '7')
2639 val = (val << 3) | (c-'0');
2640 else
2641 --s;
2642 }
2643 else
2644 --s;
2645 *p++ = (char)val;
2646 break;
2647
2648 /* \x and up to 2 hex digits */
2649 case 'x':
2650 val = 'x'; /* Default if no digits */
2651 c = hextoint(*s++); /* Get next char */
2652 if (c >= 0) {
2653 val = c;
2654 c = hextoint(*s++);
2655 if (c >= 0)
2656 val = (val << 4) + c;
2657 else
2658 --s;
2659 } else
2660 --s;
2661 *p++ = (char)val;
2662 break;
2663 }
2664 } else
2665 *p++ = (char)c;
2666 }
2667 out:
2668 *p = '\0';
2669 m->vallen = CAST(unsigned char, (p - origp));
2670 if (m->type == FILE_PSTRING)
2671 m->vallen += (unsigned char)file_pstring_length_size(m);
2672 return s;
2673 }
2674
2675
2676 /* Single hex char to int; -1 if not a hex char. */
2677 private int
hextoint(int c)2678 hextoint(int c)
2679 {
2680 if (!isascii((unsigned char) c))
2681 return -1;
2682 if (isdigit((unsigned char) c))
2683 return c - '0';
2684 if ((c >= 'a') && (c <= 'f'))
2685 return c + 10 - 'a';
2686 if (( c>= 'A') && (c <= 'F'))
2687 return c + 10 - 'A';
2688 return -1;
2689 }
2690
2691
2692 /*
2693 * Print a string containing C character escapes.
2694 */
2695 protected void
file_showstr(FILE * fp,const char * s,size_t len)2696 file_showstr(FILE *fp, const char *s, size_t len)
2697 {
2698 char c;
2699
2700 for (;;) {
2701 if (len == ~0U) {
2702 c = *s++;
2703 if (c == '\0')
2704 break;
2705 }
2706 else {
2707 if (len-- == 0)
2708 break;
2709 c = *s++;
2710 }
2711 if (c >= 040 && c <= 0176) /* TODO isprint && !iscntrl */
2712 (void) fputc(c, fp);
2713 else {
2714 (void) fputc('\\', fp);
2715 switch (c) {
2716 case '\a':
2717 (void) fputc('a', fp);
2718 break;
2719
2720 case '\b':
2721 (void) fputc('b', fp);
2722 break;
2723
2724 case '\f':
2725 (void) fputc('f', fp);
2726 break;
2727
2728 case '\n':
2729 (void) fputc('n', fp);
2730 break;
2731
2732 case '\r':
2733 (void) fputc('r', fp);
2734 break;
2735
2736 case '\t':
2737 (void) fputc('t', fp);
2738 break;
2739
2740 case '\v':
2741 (void) fputc('v', fp);
2742 break;
2743
2744 default:
2745 (void) fprintf(fp, "%.3o", c & 0377);
2746 break;
2747 }
2748 }
2749 }
2750 }
2751
2752 /*
2753 * eatsize(): Eat the size spec from a number [eg. 10UL]
2754 */
2755 private void
eatsize(const char ** p)2756 eatsize(const char **p)
2757 {
2758 const char *l = *p;
2759
2760 if (LOWCASE(*l) == 'u')
2761 l++;
2762
2763 switch (LOWCASE(*l)) {
2764 case 'l': /* long */
2765 case 's': /* short */
2766 case 'h': /* short */
2767 case 'b': /* char/byte */
2768 case 'c': /* char/byte */
2769 l++;
2770 /*FALLTHROUGH*/
2771 default:
2772 break;
2773 }
2774
2775 *p = l;
2776 }
2777
2778 /*
2779 * handle a buffer containing a compiled file.
2780 */
2781 private struct magic_map *
apprentice_buf(struct magic_set * ms,struct magic * buf,size_t len)2782 apprentice_buf(struct magic_set *ms, struct magic *buf, size_t len)
2783 {
2784 struct magic_map *map;
2785
2786 if ((map = CAST(struct magic_map *, calloc(1, sizeof(*map)))) == NULL) {
2787 file_oomem(ms, sizeof(*map));
2788 return NULL;
2789 }
2790 map->len = len;
2791 map->p = buf;
2792 map->type = MAP_TYPE_USER;
2793 if (check_buffer(ms, map, "buffer") != 0) {
2794 apprentice_unmap(map);
2795 return NULL;
2796 }
2797 return map;
2798 }
2799
2800 /*
2801 * handle a compiled file.
2802 */
2803
2804 private struct magic_map *
apprentice_map(struct magic_set * ms,const char * fn)2805 apprentice_map(struct magic_set *ms, const char *fn)
2806 {
2807 uint32_t *ptr;
2808 uint32_t version, entries, nentries;
2809 int needsbyteswap;
2810 char *dbname = NULL;
2811 struct magic_map *map;
2812 size_t i;
2813 php_stream *stream = NULL;
2814 php_stream_statbuf st;
2815
2816
2817
2818 if ((map = CAST(struct magic_map *, ecalloc(1, sizeof(*map)))) == NULL) {
2819 file_oomem(ms, sizeof(*map));
2820 return NULL;
2821 }
2822
2823 if (fn == NULL) {
2824 map->p = (void *)&php_magic_database;
2825 goto internal_loaded;
2826 }
2827
2828 #ifdef PHP_WIN32
2829 /* Don't bother on windows with php_stream_open_wrapper,
2830 return to give apprentice_load() a chance. */
2831 if (php_stream_stat_path_ex((char *)fn, 0, &st, NULL) == SUCCESS) {
2832 if (st.sb.st_mode & S_IFDIR) {
2833 return NULL;
2834 }
2835 }
2836 #endif
2837
2838 dbname = mkdbname(ms, fn, 0);
2839 if (dbname == NULL)
2840 goto error;
2841
2842 stream = php_stream_open_wrapper((char *)fn, "rb", REPORT_ERRORS, NULL);
2843
2844 if (!stream) {
2845 goto error;
2846 }
2847
2848 if (php_stream_stat(stream, &st) < 0) {
2849 file_error(ms, errno, "cannot stat `%s'", dbname);
2850 goto error;
2851 }
2852 if (st.sb.st_size < 8 || st.sb.st_size > MAXMAGIC_SIZE) {
2853 file_error(ms, 0, "file `%s' is too %s", dbname,
2854 st.sb.st_size < 8 ? "small" : "large");
2855 goto error;
2856 }
2857
2858 map->len = (size_t)st.sb.st_size;
2859 if ((map->p = CAST(void *, emalloc(map->len))) == NULL) {
2860 file_oomem(ms, map->len);
2861 goto error;
2862 }
2863 if (php_stream_read(stream, map->p, (size_t)st.sb.st_size) != (size_t)st.sb.st_size) {
2864 file_badread(ms);
2865 goto error;
2866 }
2867 map->len = 0;
2868 #define RET 1
2869
2870 php_stream_close(stream);
2871 stream = NULL;
2872
2873 internal_loaded:
2874 ptr = (uint32_t *)(void *)map->p;
2875 if (*ptr != MAGICNO) {
2876 if (swap4(*ptr) != MAGICNO) {
2877 file_error(ms, 0, "bad magic in `%s'", dbname);
2878 goto error;
2879 }
2880 needsbyteswap = 1;
2881 } else
2882 needsbyteswap = 0;
2883 if (needsbyteswap)
2884 version = swap4(ptr[1]);
2885 else
2886 version = ptr[1];
2887 if (version != VERSIONNO) {
2888 file_error(ms, 0, "File %d.%d supports only version %d magic "
2889 "files. `%s' is version %d", FILE_VERSION_MAJOR, patchlevel,
2890 VERSIONNO, dbname, version);
2891 goto error;
2892 }
2893
2894 /* php_magic_database is a const, performing writes will segfault. This is for big-endian
2895 machines only, PPC and Sparc specifically. Consider static variable or MINIT in
2896 future. */
2897 if (needsbyteswap && fn == NULL) {
2898 map->p = emalloc(sizeof(php_magic_database));
2899 map->p = memcpy(map->p, php_magic_database, sizeof(php_magic_database));
2900 }
2901
2902 if (NULL != fn) {
2903 nentries = (uint32_t)(st.sb.st_size / sizeof(struct magic));
2904 entries = (uint32_t)(st.sb.st_size / sizeof(struct magic));
2905 if ((zend_off_t)(entries * sizeof(struct magic)) != st.sb.st_size) {
2906 file_error(ms, 0, "Size of `%s' %llu is not a multiple of %zu",
2907 dbname, (unsigned long long)st.sb.st_size,
2908 sizeof(struct magic));
2909 goto error;
2910 }
2911 }
2912 map->magic[0] = CAST(struct magic *, map->p) + 1;
2913 nentries = 0;
2914 for (i = 0; i < MAGIC_SETS; i++) {
2915 if (needsbyteswap)
2916 map->nmagic[i] = swap4(ptr[i + 2]);
2917 else
2918 map->nmagic[i] = ptr[i + 2];
2919 if (i != MAGIC_SETS - 1)
2920 map->magic[i + 1] = map->magic[i] + map->nmagic[i];
2921 nentries += map->nmagic[i];
2922 }
2923 if (NULL != fn && entries != nentries + 1) {
2924 file_error(ms, 0, "Inconsistent entries in `%s' %u != %u",
2925 dbname, entries, nentries + 1);
2926 goto error;
2927 }
2928 if (needsbyteswap)
2929 for (i = 0; i < MAGIC_SETS; i++)
2930 byteswap(map->magic[i], map->nmagic[i]);
2931
2932 if (dbname) {
2933 efree(dbname);
2934 }
2935 return map;
2936
2937 error:
2938 if (stream) {
2939 php_stream_close(stream);
2940 }
2941 apprentice_unmap(map);
2942 if (dbname) {
2943 efree(dbname);
2944 }
2945 return NULL;
2946 }
2947
2948 private int
check_buffer(struct magic_set * ms,struct magic_map * map,const char * dbname)2949 check_buffer(struct magic_set *ms, struct magic_map *map, const char *dbname)
2950 {
2951 uint32_t *ptr;
2952 uint32_t entries, nentries;
2953 uint32_t version;
2954 int i, needsbyteswap;
2955
2956 ptr = CAST(uint32_t *, map->p);
2957 if (*ptr != MAGICNO) {
2958 if (swap4(*ptr) != MAGICNO) {
2959 file_error(ms, 0, "bad magic in `%s'", dbname);
2960 return -1;
2961 }
2962 needsbyteswap = 1;
2963 } else
2964 needsbyteswap = 0;
2965 if (needsbyteswap)
2966 version = swap4(ptr[1]);
2967 else
2968 version = ptr[1];
2969 if (version != VERSIONNO) {
2970 file_error(ms, 0, "File %s supports only version %d magic "
2971 "files. `%s' is version %d", FILE_VERSION_MAJOR,
2972 VERSIONNO, dbname, version);
2973 return -1;
2974 }
2975 entries = (uint32_t)(map->len / sizeof(struct magic));
2976 if ((entries * sizeof(struct magic)) != map->len) {
2977 file_error(ms, 0, "Size of `%s' %" SIZE_T_FORMAT "u is not "
2978 "a multiple of %" SIZE_T_FORMAT "u",
2979 dbname, map->len, sizeof(struct magic));
2980 return -1;
2981 }
2982 map->magic[0] = CAST(struct magic *, map->p) + 1;
2983 nentries = 0;
2984 for (i = 0; i < MAGIC_SETS; i++) {
2985 if (needsbyteswap)
2986 map->nmagic[i] = swap4(ptr[i + 2]);
2987 else
2988 map->nmagic[i] = ptr[i + 2];
2989 if (i != MAGIC_SETS - 1)
2990 map->magic[i + 1] = map->magic[i] + map->nmagic[i];
2991 nentries += map->nmagic[i];
2992 }
2993 if (entries != nentries + 1) {
2994 file_error(ms, 0, "Inconsistent entries in `%s' %u != %u",
2995 dbname, entries, nentries + 1);
2996 return -1;
2997 }
2998 if (needsbyteswap)
2999 for (i = 0; i < MAGIC_SETS; i++)
3000 byteswap(map->magic[i], map->nmagic[i]);
3001 return 0;
3002 }
3003
3004 /*
3005 * handle an mmaped file.
3006 */
3007 private int
apprentice_compile(struct magic_set * ms,struct magic_map * map,const char * fn)3008 apprentice_compile(struct magic_set *ms, struct magic_map *map, const char *fn)
3009 {
3010 static const size_t nm = sizeof(*map->nmagic) * MAGIC_SETS;
3011 static const size_t m = sizeof(**map->magic);
3012 size_t len;
3013 char *dbname;
3014 int rv = -1;
3015 uint32_t i;
3016 union {
3017 struct magic m;
3018 uint32_t h[2 + MAGIC_SETS];
3019 } hdr;
3020 php_stream *stream;
3021
3022
3023 dbname = mkdbname(ms, fn, 0);
3024
3025 if (dbname == NULL)
3026 goto out;
3027
3028 /* wb+ == O_WRONLY|O_CREAT|O_TRUNC|O_BINARY */
3029 stream = php_stream_open_wrapper((char *)fn, "wb+", REPORT_ERRORS, NULL);
3030
3031 if (!stream) {
3032 file_error(ms, errno, "cannot open `%s'", dbname);
3033 goto out;
3034 }
3035 memset(&hdr, 0, sizeof(hdr));
3036 hdr.h[0] = MAGICNO;
3037 hdr.h[1] = VERSIONNO;
3038 memcpy(hdr.h + 2, map->nmagic, nm);
3039
3040 if (php_stream_write(stream,(const char *)&hdr, sizeof(hdr)) != (ssize_t)sizeof(hdr)) {
3041 file_error(ms, errno, "error writing `%s'", dbname);
3042 goto out;
3043 }
3044
3045 for (i = 0; i < MAGIC_SETS; i++) {
3046 len = m * map->nmagic[i];
3047 if (php_stream_write(stream, (const char *)map->magic[i], len) != (ssize_t)len) {
3048 file_error(ms, errno, "error writing `%s'", dbname);
3049 goto out;
3050 }
3051 }
3052
3053 if (stream) {
3054 php_stream_close(stream);
3055 }
3056 rv = 0;
3057 out:
3058 efree(dbname);
3059 return rv;
3060 }
3061
3062 private const char ext[] = ".mgc";
3063 /*
3064 * make a dbname
3065 */
3066 private char *
mkdbname(struct magic_set * ms,const char * fn,int strip)3067 mkdbname(struct magic_set *ms, const char *fn, int strip)
3068 {
3069 const char *p, *q;
3070 char *buf;
3071
3072 if (strip) {
3073 if ((p = strrchr(fn, '/')) != NULL)
3074 fn = ++p;
3075 }
3076
3077 for (q = fn; *q; q++)
3078 continue;
3079 /* Look for .mgc */
3080 for (p = ext + sizeof(ext) - 1; p >= ext && q >= fn; p--, q--)
3081 if (*p != *q)
3082 break;
3083
3084 /* Did not find .mgc, restore q */
3085 if (p >= ext)
3086 while (*q)
3087 q++;
3088
3089 q++;
3090 /* Compatibility with old code that looked in .mime */
3091 if (ms->flags & MAGIC_MIME) {
3092 spprintf(&buf, MAXPATHLEN, "%.*s.mime%s", (int)(q - fn), fn, ext);
3093 #ifdef PHP_WIN32
3094 if (VCWD_ACCESS(buf, R_OK) == 0) {
3095 #else
3096 if (VCWD_ACCESS(buf, R_OK) != -1) {
3097 #endif
3098 ms->flags &= MAGIC_MIME_TYPE;
3099 return buf;
3100 }
3101 efree(buf);
3102 }
3103 spprintf(&buf, MAXPATHLEN, "%.*s%s", (int)(q - fn), fn, ext);
3104
3105 /* Compatibility with old code that looked in .mime */
3106 if (strstr(p, ".mime") != NULL)
3107 ms->flags &= MAGIC_MIME_TYPE;
3108 return buf;
3109 }
3110
3111 /*
3112 * Byteswap an mmap'ed file if needed
3113 */
3114 private void
3115 byteswap(struct magic *magic, uint32_t nmagic)
3116 {
3117 uint32_t i;
3118 for (i = 0; i < nmagic; i++)
3119 bs1(&magic[i]);
3120 }
3121
3122 /*
3123 * swap a short
3124 */
3125 private uint16_t
3126 swap2(uint16_t sv)
3127 {
3128 uint16_t rv;
3129 uint8_t *s = (uint8_t *)(void *)&sv;
3130 uint8_t *d = (uint8_t *)(void *)&rv;
3131 d[0] = s[1];
3132 d[1] = s[0];
3133 return rv;
3134 }
3135
3136 /*
3137 * swap an int
3138 */
3139 private uint32_t
3140 swap4(uint32_t sv)
3141 {
3142 uint32_t rv;
3143 uint8_t *s = (uint8_t *)(void *)&sv;
3144 uint8_t *d = (uint8_t *)(void *)&rv;
3145 d[0] = s[3];
3146 d[1] = s[2];
3147 d[2] = s[1];
3148 d[3] = s[0];
3149 return rv;
3150 }
3151
3152 /*
3153 * swap a quad
3154 */
3155 private uint64_t
3156 swap8(uint64_t sv)
3157 {
3158 uint64_t rv;
3159 uint8_t *s = (uint8_t *)(void *)&sv;
3160 uint8_t *d = (uint8_t *)(void *)&rv;
3161 #if 0
3162 d[0] = s[3];
3163 d[1] = s[2];
3164 d[2] = s[1];
3165 d[3] = s[0];
3166 d[4] = s[7];
3167 d[5] = s[6];
3168 d[6] = s[5];
3169 d[7] = s[4];
3170 #else
3171 d[0] = s[7];
3172 d[1] = s[6];
3173 d[2] = s[5];
3174 d[3] = s[4];
3175 d[4] = s[3];
3176 d[5] = s[2];
3177 d[6] = s[1];
3178 d[7] = s[0];
3179 #endif
3180 return rv;
3181 }
3182
3183 /*
3184 * byteswap a single magic entry
3185 */
3186 private void
3187 bs1(struct magic *m)
3188 {
3189 m->cont_level = swap2(m->cont_level);
3190 m->offset = swap4((uint32_t)m->offset);
3191 m->in_offset = swap4((uint32_t)m->in_offset);
3192 m->lineno = swap4((uint32_t)m->lineno);
3193 if (IS_LIBMAGIC_STRING(m->type)) {
3194 m->str_range = swap4(m->str_range);
3195 m->str_flags = swap4(m->str_flags);
3196 }
3197 else {
3198 m->value.q = swap8(m->value.q);
3199 m->num_mask = swap8(m->num_mask);
3200 }
3201 }
3202
3203 protected size_t
3204 file_pstring_length_size(const struct magic *m)
3205 {
3206 switch (m->str_flags & PSTRING_LEN) {
3207 case PSTRING_1_LE:
3208 return 1;
3209 case PSTRING_2_LE:
3210 case PSTRING_2_BE:
3211 return 2;
3212 case PSTRING_4_LE:
3213 case PSTRING_4_BE:
3214 return 4;
3215 default:
3216 abort(); /* Impossible */
3217 return 1;
3218 }
3219 }
3220 protected size_t
3221 file_pstring_get_length(const struct magic *m, const char *s)
3222 {
3223 size_t len = 0;
3224
3225 switch (m->str_flags & PSTRING_LEN) {
3226 case PSTRING_1_LE:
3227 len = *s;
3228 break;
3229 case PSTRING_2_LE:
3230 len = (s[1] << 8) | s[0];
3231 break;
3232 case PSTRING_2_BE:
3233 len = (s[0] << 8) | s[1];
3234 break;
3235 case PSTRING_4_LE:
3236 len = (s[3] << 24) | (s[2] << 16) | (s[1] << 8) | s[0];
3237 break;
3238 case PSTRING_4_BE:
3239 len = (s[0] << 24) | (s[1] << 16) | (s[2] << 8) | s[3];
3240 break;
3241 default:
3242 abort(); /* Impossible */
3243 }
3244
3245 if (m->str_flags & PSTRING_LENGTH_INCLUDES_ITSELF)
3246 len -= file_pstring_length_size(m);
3247
3248 return len;
3249 }
3250
3251 protected int
3252 file_magicfind(struct magic_set *ms, const char *name, struct mlist *v)
3253 {
3254 uint32_t i, j;
3255 struct mlist *mlist, *ml;
3256
3257 mlist = ms->mlist[1];
3258
3259 for (ml = mlist->next; ml != mlist; ml = ml->next) {
3260 struct magic *ma = ml->magic;
3261 uint32_t nma = ml->nmagic;
3262 for (i = 0; i < nma; i++) {
3263 if (ma[i].type != FILE_NAME)
3264 continue;
3265 if (strcmp(ma[i].value.s, name) == 0) {
3266 v->magic = &ma[i];
3267 for (j = i + 1; j < nma; j++)
3268 if (ma[j].cont_level == 0)
3269 break;
3270 v->nmagic = j - i;
3271 return 0;
3272 }
3273 }
3274 }
3275 return -1;
3276 }
3277
