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 1.283 2019/02/20 02:35:27 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
70 #define EATAB {while (isascii(CAST(unsigned char, *l)) && \
71 isspace(CAST(unsigned char, *l))) ++l;}
72 #define LOWCASE(l) (isupper(CAST(unsigned char, l)) ? \
73 tolower(CAST(unsigned char, l)) : (l))
74 /*
75 * Work around a bug in headers on Digital Unix.
76 * At least confirmed for: OSF1 V4.0 878
77 */
78 #if defined(__osf__) && defined(__DECC)
79 #ifdef MAP_FAILED
80 #undef MAP_FAILED
81 #endif
82 #endif
83
84 #ifndef offsetof
85 #define offsetof(STRUCTURE,FIELD) ((int)((char*)&((STRUCTURE*)0)->FIELD))
86 #endif
87
88 #ifndef MAP_FAILED
89 #define MAP_FAILED (void *) -1
90 #endif
91
92 #ifndef MAP_FILE
93 #define MAP_FILE 0
94 #endif
95
96 #define ALLOC_CHUNK CAST(size_t, 10)
97 #define ALLOC_INCR CAST(size_t, 200)
98
99 #define MAP_TYPE_USER 0
100 #define MAP_TYPE_MALLOC 1
101 #define MAP_TYPE_MMAP 2
102
103 struct magic_entry {
104 struct magic *mp;
105 uint32_t cont_count;
106 uint32_t max_count;
107 };
108
109 struct magic_entry_set {
110 struct magic_entry *me;
111 uint32_t count;
112 uint32_t max;
113 };
114
115 struct magic_map {
116 void *p;
117 size_t len;
118 int type;
119 struct magic *magic[MAGIC_SETS];
120 uint32_t nmagic[MAGIC_SETS];
121 };
122
123 int file_formats[FILE_NAMES_SIZE];
124 const size_t file_nformats = FILE_NAMES_SIZE;
125 const char *file_names[FILE_NAMES_SIZE];
126 const size_t file_nnames = FILE_NAMES_SIZE;
127
128 private int getvalue(struct magic_set *ms, struct magic *, const char **, int);
129 private int hextoint(int);
130 private const char *getstr(struct magic_set *, struct magic *, const char *,
131 int);
132 private int parse(struct magic_set *, struct magic_entry *, const char *,
133 size_t, int);
134 private void eatsize(const char **);
135 private int apprentice_1(struct magic_set *, const char *, int);
136 private size_t apprentice_magic_strength(const struct magic *);
137 private int apprentice_sort(const void *, const void *);
138 private void apprentice_list(struct mlist *, int );
139 private struct magic_map *apprentice_load(struct magic_set *,
140 const char *, int);
141 private struct mlist *mlist_alloc(void);
142 private void mlist_free(struct mlist *);
143 private void byteswap(struct magic *, uint32_t);
144 private void bs1(struct magic *);
145 private uint16_t swap2(uint16_t);
146 private uint32_t swap4(uint32_t);
147 private uint64_t swap8(uint64_t);
148 private char *mkdbname(struct magic_set *, const char *, int);
149 private struct magic_map *apprentice_map(struct magic_set *, const char *);
150 private void apprentice_unmap(struct magic_map *);
151 private int apprentice_compile(struct magic_set *, struct magic_map *,
152 const char *);
153 private int check_format_type(const char *, int, const char **);
154 private int check_format(struct magic_set *, struct magic *);
155 private int get_op(char);
156 private int parse_mime(struct magic_set *, struct magic_entry *, const char *);
157 private int parse_strength(struct magic_set *, struct magic_entry *, const char *);
158 private int parse_apple(struct magic_set *, struct magic_entry *, const char *);
159 private int parse_ext(struct magic_set *, struct magic_entry *, const char *);
160
161
162 private size_t magicsize = sizeof(struct magic);
163
164 private const char usg_hdr[] = "cont\toffset\ttype\topcode\tmask\tvalue\tdesc";
165
166 private struct {
167 const char *name;
168 size_t len;
169 int (*fun)(struct magic_set *, struct magic_entry *, const char *);
170 } bang[] = {
171 #define DECLARE_FIELD(name) { # name, sizeof(# name) - 1, parse_ ## name }
172 DECLARE_FIELD(mime),
173 DECLARE_FIELD(apple),
174 DECLARE_FIELD(ext),
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("der"), FILE_DER, FILE_FMT_STR },
248 { XX_NULL, FILE_INVALID, FILE_FMT_NONE },
249 };
250
251 /*
252 * These are not types, and cannot be preceded by "u" to make them
253 * unsigned.
254 */
255 static const struct type_tbl_s special_tbl[] = {
256 { XX("der"), FILE_DER, FILE_FMT_STR },
257 { XX("name"), FILE_NAME, FILE_FMT_STR },
258 { XX("use"), FILE_USE, FILE_FMT_STR },
259 { XX_NULL, FILE_INVALID, FILE_FMT_NONE },
260 };
261 # undef XX
262 # undef XX_NULL
263
264 private int
get_type(const struct type_tbl_s * tbl,const char * l,const char ** t)265 get_type(const struct type_tbl_s *tbl, const char *l, const char **t)
266 {
267 const struct type_tbl_s *p;
268
269 for (p = tbl; p->len; p++) {
270 if (strncmp(l, p->name, p->len) == 0) {
271 if (t)
272 *t = l + p->len;
273 break;
274 }
275 }
276 return p->type;
277 }
278
279 private off_t
maxoff_t(void)280 maxoff_t(void) {
281 if (/*CONSTCOND*/sizeof(off_t) == sizeof(int))
282 return CAST(off_t, INT_MAX);
283 if (/*CONSTCOND*/sizeof(off_t) == sizeof(long))
284 return CAST(off_t, LONG_MAX);
285 return 0x7fffffff;
286 }
287
288 private int
get_standard_integer_type(const char * l,const char ** t)289 get_standard_integer_type(const char *l, const char **t)
290 {
291 int type;
292
293 if (isalpha(CAST(unsigned char, l[1]))) {
294 switch (l[1]) {
295 case 'C':
296 /* "dC" and "uC" */
297 type = FILE_BYTE;
298 break;
299 case 'S':
300 /* "dS" and "uS" */
301 type = FILE_SHORT;
302 break;
303 case 'I':
304 case 'L':
305 /*
306 * "dI", "dL", "uI", and "uL".
307 *
308 * XXX - the actual Single UNIX Specification says
309 * that "L" means "long", as in the C data type,
310 * but we treat it as meaning "4-byte integer".
311 * Given that the OS X version of file 5.04 did
312 * the same, I guess that passes the actual SUS
313 * validation suite; having "dL" be dependent on
314 * how big a "long" is on the machine running
315 * "file" is silly.
316 */
317 type = FILE_LONG;
318 break;
319 case 'Q':
320 /* "dQ" and "uQ" */
321 type = FILE_QUAD;
322 break;
323 default:
324 /* "d{anything else}", "u{anything else}" */
325 return FILE_INVALID;
326 }
327 l += 2;
328 } else if (isdigit(CAST(unsigned char, l[1]))) {
329 /*
330 * "d{num}" and "u{num}"; we only support {num} values
331 * of 1, 2, 4, and 8 - the Single UNIX Specification
332 * doesn't say anything about whether arbitrary
333 * values should be supported, but both the Solaris 10
334 * and OS X Mountain Lion versions of file passed the
335 * Single UNIX Specification validation suite, and
336 * neither of them support values bigger than 8 or
337 * non-power-of-2 values.
338 */
339 if (isdigit(CAST(unsigned char, l[2]))) {
340 /* Multi-digit, so > 9 */
341 return FILE_INVALID;
342 }
343 switch (l[1]) {
344 case '1':
345 type = FILE_BYTE;
346 break;
347 case '2':
348 type = FILE_SHORT;
349 break;
350 case '4':
351 type = FILE_LONG;
352 break;
353 case '8':
354 type = FILE_QUAD;
355 break;
356 default:
357 /* XXX - what about 3, 5, 6, or 7? */
358 return FILE_INVALID;
359 }
360 l += 2;
361 } else {
362 /*
363 * "d" or "u" by itself.
364 */
365 type = FILE_LONG;
366 ++l;
367 }
368 if (t)
369 *t = l;
370 return type;
371 }
372
373 private void
init_file_tables(void)374 init_file_tables(void)
375 {
376 static int done = 0;
377 const struct type_tbl_s *p;
378
379 if (done)
380 return;
381 done++;
382
383 for (p = type_tbl; p->len; p++) {
384 assert(p->type < FILE_NAMES_SIZE);
385 file_names[p->type] = p->name;
386 file_formats[p->type] = p->format;
387 }
388 assert(p - type_tbl == FILE_NAMES_SIZE);
389 }
390
391 private int
add_mlist(struct mlist * mlp,struct magic_map * map,size_t idx)392 add_mlist(struct mlist *mlp, struct magic_map *map, size_t idx)
393 {
394 struct mlist *ml;
395
396 mlp->map = NULL;
397 if ((ml = CAST(struct mlist *, emalloc(sizeof(*ml)))) == NULL)
398 return -1;
399
400 ml->map = idx == 0 ? map : NULL;
401 ml->magic = map->magic[idx];
402 ml->nmagic = map->nmagic[idx];
403
404 mlp->prev->next = ml;
405 ml->prev = mlp->prev;
406 ml->next = mlp;
407 mlp->prev = ml;
408 return 0;
409 }
410
411 /*
412 * Handle one file or directory.
413 */
414 private int
apprentice_1(struct magic_set * ms,const char * fn,int action)415 apprentice_1(struct magic_set *ms, const char *fn, int action)
416 {
417 struct magic_map *map;
418 struct mlist *ml;
419 size_t i;
420
421 if (magicsize != FILE_MAGICSIZE) {
422 file_error(ms, 0, "magic element size %lu != %lu",
423 CAST(unsigned long, sizeof(*map->magic[0])),
424 CAST(unsigned long, FILE_MAGICSIZE));
425 return -1;
426 }
427
428 if (action == FILE_COMPILE) {
429 map = apprentice_load(ms, fn, action);
430 if (map == NULL)
431 return -1;
432 return apprentice_compile(ms, map, fn);
433 }
434
435 map = apprentice_map(ms, fn);
436 if (map == RCAST(struct magic_map *, -1))
437 return -1;
438 if (map == NULL) {
439 if (fn) {
440 if (ms->flags & MAGIC_CHECK)
441 file_magwarn(ms, "using regular magic file `%s'", fn);
442 map = apprentice_load(ms, fn, action);
443 }
444 if (map == NULL)
445 return -1;
446 }
447
448 for (i = 0; i < MAGIC_SETS; i++) {
449 if (add_mlist(ms->mlist[i], map, i) == -1) {
450 file_oomem(ms, sizeof(*ml));
451 return -1;
452 }
453 }
454
455 if (action == FILE_LIST) {
456 for (i = 0; i < MAGIC_SETS; i++) {
457 printf("Set %" SIZE_T_FORMAT "u:\nBinary patterns:\n",
458 i);
459 apprentice_list(ms->mlist[i], BINTEST);
460 printf("Text patterns:\n");
461 apprentice_list(ms->mlist[i], TEXTTEST);
462 }
463 }
464 return 0;
465 }
466
467 protected void
file_ms_free(struct magic_set * ms)468 file_ms_free(struct magic_set *ms)
469 {
470 size_t i;
471 if (ms == NULL)
472 return;
473 for (i = 0; i < MAGIC_SETS; i++)
474 mlist_free(ms->mlist[i]);
475 if (ms->o.pbuf) {
476 efree(ms->o.pbuf);
477 }
478 if (ms->o.buf) {
479 efree(ms->o.buf);
480 }
481 if (ms->c.li) {
482 efree(ms->c.li);
483 }
484 efree(ms);
485 }
486
487 protected struct magic_set *
file_ms_alloc(int flags)488 file_ms_alloc(int flags)
489 {
490 struct magic_set *ms;
491 size_t i, len;
492
493 if ((ms = CAST(struct magic_set *, ecalloc(CAST(size_t, 1u),
494 sizeof(struct magic_set)))) == NULL)
495 return NULL;
496
497 if (magic_setflags(ms, flags) == -1) {
498 errno = EINVAL;
499 goto free;
500 }
501
502 ms->o.buf = ms->o.pbuf = NULL;
503 len = (ms->c.len = 10) * sizeof(*ms->c.li);
504
505 if ((ms->c.li = CAST(struct level_info *, emalloc(len))) == NULL)
506 goto free;
507
508 ms->event_flags = 0;
509 ms->error = -1;
510 for (i = 0; i < MAGIC_SETS; i++)
511 ms->mlist[i] = NULL;
512 ms->file = "unknown";
513 ms->line = 0;
514 ms->indir_max = FILE_INDIR_MAX;
515 ms->name_max = FILE_NAME_MAX;
516 ms->elf_shnum_max = FILE_ELF_SHNUM_MAX;
517 ms->elf_phnum_max = FILE_ELF_PHNUM_MAX;
518 ms->elf_notes_max = FILE_ELF_NOTES_MAX;
519 ms->regex_max = FILE_REGEX_MAX;
520 ms->bytes_max = FILE_BYTES_MAX;
521 return ms;
522 free:
523 efree(ms);
524 return NULL;
525 }
526
527 private void
apprentice_unmap(struct magic_map * map)528 apprentice_unmap(struct magic_map *map)
529 {
530 if (map == NULL)
531 return;
532 if (map->p != php_magic_database) {
533 if (map->p == NULL) {
534 int j;
535 for (j = 0; j < MAGIC_SETS; j++) {
536 if (map->magic[j]) {
537 efree(map->magic[j]);
538 }
539 }
540 } else {
541 efree(map->p);
542 }
543 }
544 efree(map);
545 }
546
547 private struct mlist *
mlist_alloc(void)548 mlist_alloc(void)
549 {
550 struct mlist *mlist;
551 if ((mlist = CAST(struct mlist *, ecalloc(1, sizeof(*mlist)))) == NULL) {
552 return NULL;
553 }
554 mlist->next = mlist->prev = mlist;
555 return mlist;
556 }
557
558 private void
mlist_free_one(struct mlist * ml)559 mlist_free_one(struct mlist *ml)
560 {
561 if (ml->map)
562 apprentice_unmap(CAST(struct magic_map *, ml->map));
563 efree(ml);
564 }
565
566 private void
mlist_free(struct mlist * mlist)567 mlist_free(struct mlist *mlist)
568 {
569 struct mlist *ml, *next;
570
571 if (mlist == NULL)
572 return;
573
574 for (ml = mlist->next; ml != mlist; ml = next) {
575 next = ml->next;
576 mlist_free_one(ml);
577 }
578 mlist_free_one(mlist);
579 }
580
581 /* const char *fn: list of magic files and directories */
582 protected int
file_apprentice(struct magic_set * ms,const char * fn,int action)583 file_apprentice(struct magic_set *ms, const char *fn, int action)
584 {
585 char *p, *mfn;
586 int fileerr, errs = -1;
587 size_t i;
588
589 (void)file_reset(ms, 0);
590
591 /* XXX disabling default magic loading so the compiled in data is used */
592 #if 0
593 if ((fn = magic_getpath(fn, action)) == NULL)
594 return -1;
595 #endif
596
597 init_file_tables();
598
599 if (fn == NULL)
600 fn = getenv("MAGIC");
601 if (fn == NULL) {
602 for (i = 0; i < MAGIC_SETS; i++) {
603 mlist_free(ms->mlist[i]);
604 if ((ms->mlist[i] = mlist_alloc()) == NULL) {
605 file_oomem(ms, sizeof(*ms->mlist[i]));
606 return -1;
607 }
608 }
609 return apprentice_1(ms, fn, action);
610 }
611
612 if ((mfn = estrdup(fn)) == NULL) {
613 file_oomem(ms, strlen(fn));
614 return -1;
615 }
616
617 for (i = 0; i < MAGIC_SETS; i++) {
618 mlist_free(ms->mlist[i]);
619 if ((ms->mlist[i] = mlist_alloc()) == NULL) {
620 file_oomem(ms, sizeof(*ms->mlist[i]));
621 while (i-- > 0) {
622 mlist_free(ms->mlist[i]);
623 ms->mlist[i] = NULL;
624 }
625 efree(mfn);
626 return -1;
627 }
628 }
629 fn = mfn;
630
631 while (fn) {
632 p = strchr(fn, PATHSEP);
633 if (p)
634 *p++ = '\0';
635 if (*fn == '\0')
636 break;
637 fileerr = apprentice_1(ms, fn, action);
638 errs = MAX(errs, fileerr);
639 fn = p;
640 }
641
642 efree(mfn);
643
644 if (errs == -1) {
645 for (i = 0; i < MAGIC_SETS; i++) {
646 mlist_free(ms->mlist[i]);
647 ms->mlist[i] = NULL;
648 }
649 file_error(ms, 0, "could not find any valid magic files!");
650 return -1;
651 }
652
653 #if 0
654 /*
655 * Always leave the database loaded
656 */
657 if (action == FILE_LOAD)
658 return 0;
659
660 for (i = 0; i < MAGIC_SETS; i++) {
661 mlist_free(ms->mlist[i]);
662 ms->mlist[i] = NULL;
663 }
664 #endif
665
666 switch (action) {
667 case FILE_LOAD:
668 case FILE_COMPILE:
669 case FILE_CHECK:
670 case FILE_LIST:
671 return 0;
672 default:
673 file_error(ms, 0, "Invalid action %d", action);
674 return -1;
675 }
676 }
677
678 /*
679 * Compute the real length of a magic expression, for the purposes
680 * of determining how "strong" a magic expression is (approximating
681 * how specific its matches are):
682 * - magic characters count 0 unless escaped.
683 * - [] expressions count 1
684 * - {} expressions count 0
685 * - regular characters or escaped magic characters count 1
686 * - 0 length expressions count as one
687 */
688 private size_t
nonmagic(const char * str)689 nonmagic(const char *str)
690 {
691 const char *p;
692 size_t rv = 0;
693
694 for (p = str; *p; p++)
695 switch (*p) {
696 case '\\': /* Escaped anything counts 1 */
697 if (!*++p)
698 p--;
699 rv++;
700 continue;
701 case '?': /* Magic characters count 0 */
702 case '*':
703 case '.':
704 case '+':
705 case '^':
706 case '$':
707 continue;
708 case '[': /* Bracketed expressions count 1 the ']' */
709 while (*p && *p != ']')
710 p++;
711 p--;
712 continue;
713 case '{': /* Braced expressions count 0 */
714 while (*p && *p != '}')
715 p++;
716 if (!*p)
717 p--;
718 continue;
719 default: /* Anything else counts 1 */
720 rv++;
721 continue;
722 }
723
724 return rv == 0 ? 1 : rv; /* Return at least 1 */
725 }
726
727
728 private size_t
typesize(int type)729 typesize(int type)
730 {
731 switch (type) {
732 case FILE_BYTE:
733 return 1;
734
735 case FILE_SHORT:
736 case FILE_LESHORT:
737 case FILE_BESHORT:
738 return 2;
739
740 case FILE_LONG:
741 case FILE_LELONG:
742 case FILE_BELONG:
743 case FILE_MELONG:
744 return 4;
745
746 case FILE_DATE:
747 case FILE_LEDATE:
748 case FILE_BEDATE:
749 case FILE_MEDATE:
750 case FILE_LDATE:
751 case FILE_LELDATE:
752 case FILE_BELDATE:
753 case FILE_MELDATE:
754 case FILE_FLOAT:
755 case FILE_BEFLOAT:
756 case FILE_LEFLOAT:
757 return 4;
758
759 case FILE_QUAD:
760 case FILE_BEQUAD:
761 case FILE_LEQUAD:
762 case FILE_QDATE:
763 case FILE_LEQDATE:
764 case FILE_BEQDATE:
765 case FILE_QLDATE:
766 case FILE_LEQLDATE:
767 case FILE_BEQLDATE:
768 case FILE_QWDATE:
769 case FILE_LEQWDATE:
770 case FILE_BEQWDATE:
771 case FILE_DOUBLE:
772 case FILE_BEDOUBLE:
773 case FILE_LEDOUBLE:
774 return 8;
775 default:
776 return CAST(size_t, ~0);
777 }
778 }
779
780 /*
781 * Get weight of this magic entry, for sorting purposes.
782 */
783 private size_t
apprentice_magic_strength(const struct magic * m)784 apprentice_magic_strength(const struct magic *m)
785 {
786 #define MULT 10U
787 size_t ts, v;
788 ssize_t val = 2 * MULT; /* baseline strength */
789
790 switch (m->type) {
791 case FILE_DEFAULT: /* make sure this sorts last */
792 if (m->factor_op != FILE_FACTOR_OP_NONE)
793 abort();
794 return 0;
795
796 case FILE_BYTE:
797 case FILE_SHORT:
798 case FILE_LESHORT:
799 case FILE_BESHORT:
800 case FILE_LONG:
801 case FILE_LELONG:
802 case FILE_BELONG:
803 case FILE_MELONG:
804 case FILE_DATE:
805 case FILE_LEDATE:
806 case FILE_BEDATE:
807 case FILE_MEDATE:
808 case FILE_LDATE:
809 case FILE_LELDATE:
810 case FILE_BELDATE:
811 case FILE_MELDATE:
812 case FILE_FLOAT:
813 case FILE_BEFLOAT:
814 case FILE_LEFLOAT:
815 case FILE_QUAD:
816 case FILE_BEQUAD:
817 case FILE_LEQUAD:
818 case FILE_QDATE:
819 case FILE_LEQDATE:
820 case FILE_BEQDATE:
821 case FILE_QLDATE:
822 case FILE_LEQLDATE:
823 case FILE_BEQLDATE:
824 case FILE_QWDATE:
825 case FILE_LEQWDATE:
826 case FILE_BEQWDATE:
827 case FILE_DOUBLE:
828 case FILE_BEDOUBLE:
829 case FILE_LEDOUBLE:
830 ts = typesize(m->type);
831 if (ts == CAST(size_t, ~0))
832 abort();
833 val += ts * MULT;
834 break;
835
836 case FILE_PSTRING:
837 case FILE_STRING:
838 val += m->vallen * MULT;
839 break;
840
841 case FILE_BESTRING16:
842 case FILE_LESTRING16:
843 val += m->vallen * MULT / 2;
844 break;
845
846 case FILE_SEARCH:
847 if (m->vallen == 0)
848 break;
849 val += m->vallen * MAX(MULT / m->vallen, 1);
850 break;
851
852 case FILE_REGEX:
853 v = nonmagic(m->value.s);
854 val += v * MAX(MULT / v, 1);
855 break;
856
857 case FILE_INDIRECT:
858 case FILE_NAME:
859 case FILE_USE:
860 break;
861
862 case FILE_DER:
863 val += MULT;
864 break;
865
866 default:
867 (void)fprintf(stderr, "Bad type %d\n", m->type);
868 abort();
869 }
870
871 switch (m->reln) {
872 case 'x': /* matches anything penalize */
873 case '!': /* matches almost anything penalize */
874 val = 0;
875 break;
876
877 case '=': /* Exact match, prefer */
878 val += MULT;
879 break;
880
881 case '>':
882 case '<': /* comparison match reduce strength */
883 val -= 2 * MULT;
884 break;
885
886 case '^':
887 case '&': /* masking bits, we could count them too */
888 val -= MULT;
889 break;
890
891 default:
892 (void)fprintf(stderr, "Bad relation %c\n", m->reln);
893 abort();
894 }
895
896 switch (m->factor_op) {
897 case FILE_FACTOR_OP_NONE:
898 break;
899 case FILE_FACTOR_OP_PLUS:
900 val += m->factor;
901 break;
902 case FILE_FACTOR_OP_MINUS:
903 val -= m->factor;
904 break;
905 case FILE_FACTOR_OP_TIMES:
906 val *= m->factor;
907 break;
908 case FILE_FACTOR_OP_DIV:
909 val /= m->factor;
910 break;
911 default:
912 abort();
913 }
914
915 if (val <= 0) /* ensure we only return 0 for FILE_DEFAULT */
916 val = 1;
917
918 /*
919 * Magic entries with no description get a bonus because they depend
920 * on subsequent magic entries to print something.
921 */
922 if (m->desc[0] == '\0')
923 val++;
924 return val;
925 }
926
927 /*
928 * Sort callback for sorting entries by "strength" (basically length)
929 */
930 private int
apprentice_sort(const void * a,const void * b)931 apprentice_sort(const void *a, const void *b)
932 {
933 const struct magic_entry *ma = CAST(const struct magic_entry *, a);
934 const struct magic_entry *mb = CAST(const struct magic_entry *, b);
935 size_t sa = apprentice_magic_strength(ma->mp);
936 size_t sb = apprentice_magic_strength(mb->mp);
937 if (sa == sb)
938 return 0;
939 else if (sa > sb)
940 return -1;
941 else
942 return 1;
943 }
944
945 /*
946 * Shows sorted patterns list in the order which is used for the matching
947 */
948 private void
apprentice_list(struct mlist * mlist,int mode)949 apprentice_list(struct mlist *mlist, int mode)
950 {
951 uint32_t magindex = 0;
952 struct mlist *ml;
953 for (ml = mlist->next; ml != mlist; ml = ml->next) {
954 for (magindex = 0; magindex < ml->nmagic; magindex++) {
955 struct magic *m = &ml->magic[magindex];
956 if ((m->flag & mode) != mode) {
957 /* Skip sub-tests */
958 while (magindex + 1 < ml->nmagic &&
959 ml->magic[magindex + 1].cont_level != 0)
960 ++magindex;
961 continue; /* Skip to next top-level test*/
962 }
963
964 /*
965 * Try to iterate over the tree until we find item with
966 * description/mimetype.
967 */
968 while (magindex + 1 < ml->nmagic &&
969 ml->magic[magindex + 1].cont_level != 0 &&
970 *ml->magic[magindex].desc == '\0' &&
971 *ml->magic[magindex].mimetype == '\0')
972 magindex++;
973
974 printf("Strength = %3" SIZE_T_FORMAT "u@%u: %s [%s]\n",
975 apprentice_magic_strength(m),
976 ml->magic[magindex].lineno,
977 ml->magic[magindex].desc,
978 ml->magic[magindex].mimetype);
979 }
980 }
981 }
982
983 private void
set_test_type(struct magic * mstart,struct magic * m)984 set_test_type(struct magic *mstart, struct magic *m)
985 {
986 switch (m->type) {
987 case FILE_BYTE:
988 case FILE_SHORT:
989 case FILE_LONG:
990 case FILE_DATE:
991 case FILE_BESHORT:
992 case FILE_BELONG:
993 case FILE_BEDATE:
994 case FILE_LESHORT:
995 case FILE_LELONG:
996 case FILE_LEDATE:
997 case FILE_LDATE:
998 case FILE_BELDATE:
999 case FILE_LELDATE:
1000 case FILE_MEDATE:
1001 case FILE_MELDATE:
1002 case FILE_MELONG:
1003 case FILE_QUAD:
1004 case FILE_LEQUAD:
1005 case FILE_BEQUAD:
1006 case FILE_QDATE:
1007 case FILE_LEQDATE:
1008 case FILE_BEQDATE:
1009 case FILE_QLDATE:
1010 case FILE_LEQLDATE:
1011 case FILE_BEQLDATE:
1012 case FILE_QWDATE:
1013 case FILE_LEQWDATE:
1014 case FILE_BEQWDATE:
1015 case FILE_FLOAT:
1016 case FILE_BEFLOAT:
1017 case FILE_LEFLOAT:
1018 case FILE_DOUBLE:
1019 case FILE_BEDOUBLE:
1020 case FILE_LEDOUBLE:
1021 case FILE_DER:
1022 mstart->flag |= BINTEST;
1023 break;
1024 case FILE_STRING:
1025 case FILE_PSTRING:
1026 case FILE_BESTRING16:
1027 case FILE_LESTRING16:
1028 /* Allow text overrides */
1029 if (mstart->str_flags & STRING_TEXTTEST)
1030 mstart->flag |= TEXTTEST;
1031 else
1032 mstart->flag |= BINTEST;
1033 break;
1034 case FILE_REGEX:
1035 case FILE_SEARCH:
1036 /* Check for override */
1037 if (mstart->str_flags & STRING_BINTEST)
1038 mstart->flag |= BINTEST;
1039 if (mstart->str_flags & STRING_TEXTTEST)
1040 mstart->flag |= TEXTTEST;
1041
1042 if (mstart->flag & (TEXTTEST|BINTEST))
1043 break;
1044
1045 /* binary test if pattern is not text */
1046 if (file_looks_utf8(m->value.us, CAST(size_t, m->vallen), NULL,
1047 NULL) <= 0)
1048 mstart->flag |= BINTEST;
1049 else
1050 mstart->flag |= TEXTTEST;
1051 break;
1052 case FILE_DEFAULT:
1053 /* can't deduce anything; we shouldn't see this at the
1054 top level anyway */
1055 break;
1056 case FILE_INVALID:
1057 default:
1058 /* invalid search type, but no need to complain here */
1059 break;
1060 }
1061 }
1062
1063 private int
addentry(struct magic_set * ms,struct magic_entry * me,struct magic_entry_set * mset)1064 addentry(struct magic_set *ms, struct magic_entry *me,
1065 struct magic_entry_set *mset)
1066 {
1067 size_t i = me->mp->type == FILE_NAME ? 1 : 0;
1068 if (mset[i].count == mset[i].max) {
1069 struct magic_entry *mp;
1070
1071 mset[i].max += ALLOC_INCR;
1072 if ((mp = CAST(struct magic_entry *,
1073 erealloc(mset[i].me, sizeof(*mp) * mset[i].max))) ==
1074 NULL) {
1075 file_oomem(ms, sizeof(*mp) * mset[i].max);
1076 return -1;
1077 }
1078 (void)memset(&mp[mset[i].count], 0, sizeof(*mp) *
1079 ALLOC_INCR);
1080 mset[i].me = mp;
1081 }
1082 mset[i].me[mset[i].count++] = *me;
1083 memset(me, 0, sizeof(*me));
1084 return 0;
1085 }
1086
1087 /*
1088 * Load and parse one file.
1089 */
1090 private void
load_1(struct magic_set * ms,int action,const char * fn,int * errs,struct magic_entry_set * mset)1091 load_1(struct magic_set *ms, int action, const char *fn, int *errs,
1092 struct magic_entry_set *mset)
1093 {
1094 char buffer[BUFSIZ + 1];
1095 char *line = NULL;
1096 size_t len;
1097 size_t lineno = 0;
1098 struct magic_entry me;
1099
1100 php_stream *stream;
1101
1102
1103 ms->file = fn;
1104 stream = php_stream_open_wrapper((char *)fn, "rb", REPORT_ERRORS, NULL);
1105
1106 if (stream == NULL) {
1107 if (errno != ENOENT)
1108 file_error(ms, errno, "cannot read magic file `%s'",
1109 fn);
1110 (*errs)++;
1111 return;
1112 }
1113
1114 memset(&me, 0, sizeof(me));
1115 /* read and parse this file */
1116 for (ms->line = 1; (line = php_stream_get_line(stream, buffer , BUFSIZ, &len)) != NULL; ms->line++) {
1117 if (len == 0) /* null line, garbage, etc */
1118 continue;
1119 if (line[len - 1] == '\n') {
1120 lineno++;
1121 line[len - 1] = '\0'; /* delete newline */
1122 }
1123 switch (line[0]) {
1124 case '\0': /* empty, do not parse */
1125 case '#': /* comment, do not parse */
1126 continue;
1127 case '!':
1128 if (line[1] == ':') {
1129 size_t i;
1130
1131 for (i = 0; bang[i].name != NULL; i++) {
1132 if (CAST(size_t, len - 2) > bang[i].len &&
1133 memcmp(bang[i].name, line + 2,
1134 bang[i].len) == 0)
1135 break;
1136 }
1137 if (bang[i].name == NULL) {
1138 file_error(ms, 0,
1139 "Unknown !: entry `%s'", line);
1140 (*errs)++;
1141 continue;
1142 }
1143 if (me.mp == NULL) {
1144 file_error(ms, 0,
1145 "No current entry for :!%s type",
1146 bang[i].name);
1147 (*errs)++;
1148 continue;
1149 }
1150 if ((*bang[i].fun)(ms, &me,
1151 line + bang[i].len + 2) != 0) {
1152 (*errs)++;
1153 continue;
1154 }
1155 continue;
1156 }
1157 /*FALLTHROUGH*/
1158 default:
1159 again:
1160 switch (parse(ms, &me, line, lineno, action)) {
1161 case 0:
1162 continue;
1163 case 1:
1164 (void)addentry(ms, &me, mset);
1165 goto again;
1166 default:
1167 (*errs)++;
1168 break;
1169 }
1170 }
1171 }
1172 if (me.mp)
1173 (void)addentry(ms, &me, mset);
1174 efree(line);
1175 php_stream_close(stream);
1176 }
1177
1178 /*
1179 * parse a file or directory of files
1180 * const char *fn: name of magic file or directory
1181 */
1182 private int
cmpstrp(const void * p1,const void * p2)1183 cmpstrp(const void *p1, const void *p2)
1184 {
1185 return strcmp(*RCAST(char *const *, p1), *RCAST(char *const *, p2));
1186 }
1187
1188
1189 private uint32_t
set_text_binary(struct magic_set * ms,struct magic_entry * me,uint32_t nme,uint32_t starttest)1190 set_text_binary(struct magic_set *ms, struct magic_entry *me, uint32_t nme,
1191 uint32_t starttest)
1192 {
1193 static const char text[] = "text";
1194 static const char binary[] = "binary";
1195 static const size_t len = sizeof(text);
1196
1197 uint32_t i = starttest;
1198
1199 do {
1200 set_test_type(me[starttest].mp, me[i].mp);
1201 if ((ms->flags & MAGIC_DEBUG) == 0)
1202 continue;
1203 (void)fprintf(stderr, "%s%s%s: %s\n",
1204 me[i].mp->mimetype,
1205 me[i].mp->mimetype[0] == '\0' ? "" : "; ",
1206 me[i].mp->desc[0] ? me[i].mp->desc : "(no description)",
1207 me[i].mp->flag & BINTEST ? binary : text);
1208 if (me[i].mp->flag & BINTEST) {
1209 char *p = strstr(me[i].mp->desc, text);
1210 if (p && (p == me[i].mp->desc ||
1211 isspace(CAST(unsigned char, p[-1]))) &&
1212 (p + len - me[i].mp->desc == MAXstring
1213 || (p[len] == '\0' ||
1214 isspace(CAST(unsigned char, p[len])))))
1215 (void)fprintf(stderr, "*** Possible "
1216 "binary test for text type\n");
1217 }
1218 } while (++i < nme && me[i].mp->cont_level != 0);
1219 return i;
1220 }
1221
1222 private void
set_last_default(struct magic_set * ms,struct magic_entry * me,uint32_t nme)1223 set_last_default(struct magic_set *ms, struct magic_entry *me, uint32_t nme)
1224 {
1225 uint32_t i;
1226 for (i = 0; i < nme; i++) {
1227 if (me[i].mp->cont_level == 0 &&
1228 me[i].mp->type == FILE_DEFAULT) {
1229 while (++i < nme)
1230 if (me[i].mp->cont_level == 0)
1231 break;
1232 if (i != nme) {
1233 /* XXX - Ugh! */
1234 ms->line = me[i].mp->lineno;
1235 file_magwarn(ms,
1236 "level 0 \"default\" did not sort last");
1237 }
1238 return;
1239 }
1240 }
1241 }
1242
1243 private int
coalesce_entries(struct magic_set * ms,struct magic_entry * me,uint32_t nme,struct magic ** ma,uint32_t * nma)1244 coalesce_entries(struct magic_set *ms, struct magic_entry *me, uint32_t nme,
1245 struct magic **ma, uint32_t *nma)
1246 {
1247 uint32_t i, mentrycount = 0;
1248 size_t slen;
1249
1250 for (i = 0; i < nme; i++)
1251 mentrycount += me[i].cont_count;
1252
1253 slen = sizeof(**ma) * mentrycount;
1254 if ((*ma = CAST(struct magic *, emalloc(slen))) == NULL) {
1255 file_oomem(ms, slen);
1256 return -1;
1257 }
1258
1259 mentrycount = 0;
1260 for (i = 0; i < nme; i++) {
1261 (void)memcpy(*ma + mentrycount, me[i].mp,
1262 me[i].cont_count * sizeof(**ma));
1263 mentrycount += me[i].cont_count;
1264 }
1265 *nma = mentrycount;
1266 return 0;
1267 }
1268
1269 private void
magic_entry_free(struct magic_entry * me,uint32_t nme)1270 magic_entry_free(struct magic_entry *me, uint32_t nme)
1271 {
1272 uint32_t i;
1273 if (me == NULL)
1274 return;
1275 for (i = 0; i < nme; i++)
1276 efree(me[i].mp);
1277 efree(me);
1278 }
1279
1280 private struct magic_map *
apprentice_load(struct magic_set * ms,const char * fn,int action)1281 apprentice_load(struct magic_set *ms, const char *fn, int action)
1282 {
1283 int errs = 0;
1284 uint32_t i, j;
1285 size_t files = 0, maxfiles = 0;
1286 char **filearr = NULL;
1287 zend_stat_t st;
1288 struct magic_map *map;
1289 struct magic_entry_set mset[MAGIC_SETS];
1290 php_stream *dir;
1291 php_stream_dirent d;
1292
1293
1294 memset(mset, 0, sizeof(mset));
1295 ms->flags |= MAGIC_CHECK; /* Enable checks for parsed files */
1296
1297
1298 if ((map = CAST(struct magic_map *, ecalloc(1, sizeof(*map)))) == NULL)
1299 {
1300 file_oomem(ms, sizeof(*map));
1301 return NULL;
1302 }
1303 map->type = MAP_TYPE_MALLOC;
1304
1305 /* print silly verbose header for USG compat. */
1306 if (action == FILE_CHECK)
1307 (void)fprintf(stderr, "%s\n", usg_hdr);
1308
1309 /* load directory or file */
1310 /* FIXME: Read file names and sort them to prevent
1311 non-determinism. See Debian bug #488562. */
1312 if (php_sys_stat(fn, &st) == 0 && S_ISDIR(st.st_mode)) {
1313 int mflen;
1314 char mfn[MAXPATHLEN];
1315
1316 dir = php_stream_opendir((char *)fn, REPORT_ERRORS, NULL);
1317 if (!dir) {
1318 errs++;
1319 goto out;
1320 }
1321 while (php_stream_readdir(dir, &d)) {
1322 if ((mflen = snprintf(mfn, sizeof(mfn), "%s/%s", fn, d.d_name)) < 0) {
1323 file_oomem(ms,
1324 strlen(fn) + strlen(d.d_name) + 2);
1325 errs++;
1326 php_stream_closedir(dir);
1327 goto out;
1328 }
1329 if (zend_stat(mfn, &st) == -1 || !S_ISREG(st.st_mode)) {
1330 continue;
1331 }
1332 if (files >= maxfiles) {
1333 size_t mlen;
1334 maxfiles = (maxfiles + 1) * 2;
1335 mlen = maxfiles * sizeof(*filearr);
1336 if ((filearr = CAST(char **,
1337 erealloc(filearr, mlen))) == NULL) {
1338 file_oomem(ms, mlen);
1339 php_stream_closedir(dir);
1340 errs++;
1341 goto out;
1342 }
1343 }
1344 filearr[files++] = estrndup(mfn, (mflen > sizeof(mfn) - 1)? sizeof(mfn) - 1: mflen);
1345 }
1346 php_stream_closedir(dir);
1347 if (filearr) {
1348 qsort(filearr, files, sizeof(*filearr), cmpstrp);
1349 for (i = 0; i < files; i++) {
1350 load_1(ms, action, filearr[i], &errs, mset);
1351 efree(filearr[i]);
1352 }
1353 efree(filearr);
1354 }
1355 } else
1356 load_1(ms, action, fn, &errs, mset);
1357 if (errs)
1358 goto out;
1359
1360 for (j = 0; j < MAGIC_SETS; j++) {
1361 /* Set types of tests */
1362 for (i = 0; i < mset[j].count; ) {
1363 if (mset[j].me[i].mp->cont_level != 0) {
1364 i++;
1365 continue;
1366 }
1367 i = set_text_binary(ms, mset[j].me, mset[j].count, i);
1368 }
1369 if (mset[j].me)
1370 qsort(mset[j].me, mset[j].count, sizeof(*mset[j].me),
1371 apprentice_sort);
1372
1373 /*
1374 * Make sure that any level 0 "default" line is last
1375 * (if one exists).
1376 */
1377 set_last_default(ms, mset[j].me, mset[j].count);
1378
1379 /* coalesce per file arrays into a single one */
1380 if (coalesce_entries(ms, mset[j].me, mset[j].count,
1381 &map->magic[j], &map->nmagic[j]) == -1) {
1382 errs++;
1383 goto out;
1384 }
1385 }
1386
1387 out:
1388 for (j = 0; j < MAGIC_SETS; j++)
1389 magic_entry_free(mset[j].me, mset[j].count);
1390
1391 if (errs) {
1392 apprentice_unmap(map);
1393 return NULL;
1394 }
1395 return map;
1396 }
1397
1398 /*
1399 * extend the sign bit if the comparison is to be signed
1400 */
1401 protected uint64_t
file_signextend(struct magic_set * ms,struct magic * m,uint64_t v)1402 file_signextend(struct magic_set *ms, struct magic *m, uint64_t v)
1403 {
1404 if (!(m->flag & UNSIGNED)) {
1405 switch(m->type) {
1406 /*
1407 * Do not remove the casts below. They are
1408 * vital. When later compared with the data,
1409 * the sign extension must have happened.
1410 */
1411 case FILE_BYTE:
1412 v = CAST(signed char, v);
1413 break;
1414 case FILE_SHORT:
1415 case FILE_BESHORT:
1416 case FILE_LESHORT:
1417 v = CAST(short, v);
1418 break;
1419 case FILE_DATE:
1420 case FILE_BEDATE:
1421 case FILE_LEDATE:
1422 case FILE_MEDATE:
1423 case FILE_LDATE:
1424 case FILE_BELDATE:
1425 case FILE_LELDATE:
1426 case FILE_MELDATE:
1427 case FILE_LONG:
1428 case FILE_BELONG:
1429 case FILE_LELONG:
1430 case FILE_MELONG:
1431 case FILE_FLOAT:
1432 case FILE_BEFLOAT:
1433 case FILE_LEFLOAT:
1434 v = CAST(int32_t, v);
1435 break;
1436 case FILE_QUAD:
1437 case FILE_BEQUAD:
1438 case FILE_LEQUAD:
1439 case FILE_QDATE:
1440 case FILE_QLDATE:
1441 case FILE_QWDATE:
1442 case FILE_BEQDATE:
1443 case FILE_BEQLDATE:
1444 case FILE_BEQWDATE:
1445 case FILE_LEQDATE:
1446 case FILE_LEQLDATE:
1447 case FILE_LEQWDATE:
1448 case FILE_DOUBLE:
1449 case FILE_BEDOUBLE:
1450 case FILE_LEDOUBLE:
1451 v = CAST(int64_t, v);
1452 break;
1453 case FILE_STRING:
1454 case FILE_PSTRING:
1455 case FILE_BESTRING16:
1456 case FILE_LESTRING16:
1457 case FILE_REGEX:
1458 case FILE_SEARCH:
1459 case FILE_DEFAULT:
1460 case FILE_INDIRECT:
1461 case FILE_NAME:
1462 case FILE_USE:
1463 case FILE_CLEAR:
1464 case FILE_DER:
1465 break;
1466 default:
1467 if (ms->flags & MAGIC_CHECK)
1468 file_magwarn(ms, "cannot happen: m->type=%d\n",
1469 m->type);
1470 return ~0U;
1471 }
1472 }
1473 return v;
1474 }
1475
1476 private int
string_modifier_check(struct magic_set * ms,struct magic * m)1477 string_modifier_check(struct magic_set *ms, struct magic *m)
1478 {
1479 if ((ms->flags & MAGIC_CHECK) == 0)
1480 return 0;
1481
1482 if ((m->type != FILE_REGEX || (m->str_flags & REGEX_LINE_COUNT) == 0) &&
1483 (m->type != FILE_PSTRING && (m->str_flags & PSTRING_LEN) != 0)) {
1484 file_magwarn(ms,
1485 "'/BHhLl' modifiers are only allowed for pascal strings\n");
1486 return -1;
1487 }
1488 switch (m->type) {
1489 case FILE_BESTRING16:
1490 case FILE_LESTRING16:
1491 if (m->str_flags != 0) {
1492 file_magwarn(ms,
1493 "no modifiers allowed for 16-bit strings\n");
1494 return -1;
1495 }
1496 break;
1497 case FILE_STRING:
1498 case FILE_PSTRING:
1499 if ((m->str_flags & REGEX_OFFSET_START) != 0) {
1500 file_magwarn(ms,
1501 "'/%c' only allowed on regex and search\n",
1502 CHAR_REGEX_OFFSET_START);
1503 return -1;
1504 }
1505 break;
1506 case FILE_SEARCH:
1507 if (m->str_range == 0) {
1508 file_magwarn(ms,
1509 "missing range; defaulting to %d\n",
1510 STRING_DEFAULT_RANGE);
1511 m->str_range = STRING_DEFAULT_RANGE;
1512 return -1;
1513 }
1514 break;
1515 case FILE_REGEX:
1516 if ((m->str_flags & STRING_COMPACT_WHITESPACE) != 0) {
1517 file_magwarn(ms, "'/%c' not allowed on regex\n",
1518 CHAR_COMPACT_WHITESPACE);
1519 return -1;
1520 }
1521 if ((m->str_flags & STRING_COMPACT_OPTIONAL_WHITESPACE) != 0) {
1522 file_magwarn(ms, "'/%c' not allowed on regex\n",
1523 CHAR_COMPACT_OPTIONAL_WHITESPACE);
1524 return -1;
1525 }
1526 break;
1527 default:
1528 file_magwarn(ms, "coding error: m->type=%d\n",
1529 m->type);
1530 return -1;
1531 }
1532 return 0;
1533 }
1534
1535 private int
get_op(char c)1536 get_op(char c)
1537 {
1538 switch (c) {
1539 case '&':
1540 return FILE_OPAND;
1541 case '|':
1542 return FILE_OPOR;
1543 case '^':
1544 return FILE_OPXOR;
1545 case '+':
1546 return FILE_OPADD;
1547 case '-':
1548 return FILE_OPMINUS;
1549 case '*':
1550 return FILE_OPMULTIPLY;
1551 case '/':
1552 return FILE_OPDIVIDE;
1553 case '%':
1554 return FILE_OPMODULO;
1555 default:
1556 return -1;
1557 }
1558 }
1559
1560 #ifdef ENABLE_CONDITIONALS
1561 private int
get_cond(const char * l,const char ** t)1562 get_cond(const char *l, const char **t)
1563 {
1564 static const struct cond_tbl_s {
1565 char name[8];
1566 size_t len;
1567 int cond;
1568 } cond_tbl[] = {
1569 { "if", 2, COND_IF },
1570 { "elif", 4, COND_ELIF },
1571 { "else", 4, COND_ELSE },
1572 { "", 0, COND_NONE },
1573 };
1574 const struct cond_tbl_s *p;
1575
1576 for (p = cond_tbl; p->len; p++) {
1577 if (strncmp(l, p->name, p->len) == 0 &&
1578 isspace(CAST(unsigned char, l[p->len]))) {
1579 if (t)
1580 *t = l + p->len;
1581 break;
1582 }
1583 }
1584 return p->cond;
1585 }
1586
1587 private int
check_cond(struct magic_set * ms,int cond,uint32_t cont_level)1588 check_cond(struct magic_set *ms, int cond, uint32_t cont_level)
1589 {
1590 int last_cond;
1591 last_cond = ms->c.li[cont_level].last_cond;
1592
1593 switch (cond) {
1594 case COND_IF:
1595 if (last_cond != COND_NONE && last_cond != COND_ELIF) {
1596 if (ms->flags & MAGIC_CHECK)
1597 file_magwarn(ms, "syntax error: `if'");
1598 return -1;
1599 }
1600 last_cond = COND_IF;
1601 break;
1602
1603 case COND_ELIF:
1604 if (last_cond != COND_IF && last_cond != COND_ELIF) {
1605 if (ms->flags & MAGIC_CHECK)
1606 file_magwarn(ms, "syntax error: `elif'");
1607 return -1;
1608 }
1609 last_cond = COND_ELIF;
1610 break;
1611
1612 case COND_ELSE:
1613 if (last_cond != COND_IF && last_cond != COND_ELIF) {
1614 if (ms->flags & MAGIC_CHECK)
1615 file_magwarn(ms, "syntax error: `else'");
1616 return -1;
1617 }
1618 last_cond = COND_NONE;
1619 break;
1620
1621 case COND_NONE:
1622 last_cond = COND_NONE;
1623 break;
1624 }
1625
1626 ms->c.li[cont_level].last_cond = last_cond;
1627 return 0;
1628 }
1629 #endif /* ENABLE_CONDITIONALS */
1630
1631 private int
parse_indirect_modifier(struct magic_set * ms,struct magic * m,const char ** lp)1632 parse_indirect_modifier(struct magic_set *ms, struct magic *m, const char **lp)
1633 {
1634 const char *l = *lp;
1635
1636 while (!isspace(CAST(unsigned char, *++l)))
1637 switch (*l) {
1638 case CHAR_INDIRECT_RELATIVE:
1639 m->str_flags |= INDIRECT_RELATIVE;
1640 break;
1641 default:
1642 if (ms->flags & MAGIC_CHECK)
1643 file_magwarn(ms, "indirect modifier `%c' "
1644 "invalid", *l);
1645 *lp = l;
1646 return -1;
1647 }
1648 *lp = l;
1649 return 0;
1650 }
1651
1652 private void
parse_op_modifier(struct magic_set * ms,struct magic * m,const char ** lp,int op)1653 parse_op_modifier(struct magic_set *ms, struct magic *m, const char **lp,
1654 int op)
1655 {
1656 const char *l = *lp;
1657 char *t;
1658 uint64_t val;
1659
1660 ++l;
1661 m->mask_op |= op;
1662 val = CAST(uint64_t, strtoull(l, &t, 0));
1663 l = t;
1664 m->num_mask = file_signextend(ms, m, val);
1665 eatsize(&l);
1666 *lp = l;
1667 }
1668
1669 private int
parse_string_modifier(struct magic_set * ms,struct magic * m,const char ** lp)1670 parse_string_modifier(struct magic_set *ms, struct magic *m, const char **lp)
1671 {
1672 const char *l = *lp;
1673 char *t;
1674 int have_range = 0;
1675
1676 while (!isspace(CAST(unsigned char, *++l))) {
1677 switch (*l) {
1678 case '0': case '1': case '2':
1679 case '3': case '4': case '5':
1680 case '6': case '7': case '8':
1681 case '9':
1682 if (have_range && (ms->flags & MAGIC_CHECK))
1683 file_magwarn(ms, "multiple ranges");
1684 have_range = 1;
1685 m->str_range = CAST(uint32_t, strtoul(l, &t, 0));
1686 if (m->str_range == 0)
1687 file_magwarn(ms, "zero range");
1688 l = t - 1;
1689 break;
1690 case CHAR_COMPACT_WHITESPACE:
1691 m->str_flags |= STRING_COMPACT_WHITESPACE;
1692 break;
1693 case CHAR_COMPACT_OPTIONAL_WHITESPACE:
1694 m->str_flags |= STRING_COMPACT_OPTIONAL_WHITESPACE;
1695 break;
1696 case CHAR_IGNORE_LOWERCASE:
1697 m->str_flags |= STRING_IGNORE_LOWERCASE;
1698 break;
1699 case CHAR_IGNORE_UPPERCASE:
1700 m->str_flags |= STRING_IGNORE_UPPERCASE;
1701 break;
1702 case CHAR_REGEX_OFFSET_START:
1703 m->str_flags |= REGEX_OFFSET_START;
1704 break;
1705 case CHAR_BINTEST:
1706 m->str_flags |= STRING_BINTEST;
1707 break;
1708 case CHAR_TEXTTEST:
1709 m->str_flags |= STRING_TEXTTEST;
1710 break;
1711 case CHAR_TRIM:
1712 m->str_flags |= STRING_TRIM;
1713 break;
1714 case CHAR_PSTRING_1_LE:
1715 #define SET_LENGTH(a) m->str_flags = (m->str_flags & ~PSTRING_LEN) | (a)
1716 if (m->type != FILE_PSTRING)
1717 goto bad;
1718 SET_LENGTH(PSTRING_1_LE);
1719 break;
1720 case CHAR_PSTRING_2_BE:
1721 if (m->type != FILE_PSTRING)
1722 goto bad;
1723 SET_LENGTH(PSTRING_2_BE);
1724 break;
1725 case CHAR_PSTRING_2_LE:
1726 if (m->type != FILE_PSTRING)
1727 goto bad;
1728 SET_LENGTH(PSTRING_2_LE);
1729 break;
1730 case CHAR_PSTRING_4_BE:
1731 if (m->type != FILE_PSTRING)
1732 goto bad;
1733 SET_LENGTH(PSTRING_4_BE);
1734 break;
1735 case CHAR_PSTRING_4_LE:
1736 switch (m->type) {
1737 case FILE_PSTRING:
1738 case FILE_REGEX:
1739 break;
1740 default:
1741 goto bad;
1742 }
1743 SET_LENGTH(PSTRING_4_LE);
1744 break;
1745 case CHAR_PSTRING_LENGTH_INCLUDES_ITSELF:
1746 if (m->type != FILE_PSTRING)
1747 goto bad;
1748 m->str_flags |= PSTRING_LENGTH_INCLUDES_ITSELF;
1749 break;
1750 default:
1751 bad:
1752 if (ms->flags & MAGIC_CHECK)
1753 file_magwarn(ms, "string modifier `%c' "
1754 "invalid", *l);
1755 goto out;
1756 }
1757 /* allow multiple '/' for readability */
1758 if (l[1] == '/' && !isspace(CAST(unsigned char, l[2])))
1759 l++;
1760 }
1761 if (string_modifier_check(ms, m) == -1)
1762 goto out;
1763 *lp = l;
1764 return 0;
1765 out:
1766 *lp = l;
1767 return -1;
1768 }
1769
1770 /*
1771 * parse one line from magic file, put into magic[index++] if valid
1772 */
1773 private int
parse(struct magic_set * ms,struct magic_entry * me,const char * line,size_t lineno,int action)1774 parse(struct magic_set *ms, struct magic_entry *me, const char *line,
1775 size_t lineno, int action)
1776 {
1777 #ifdef ENABLE_CONDITIONALS
1778 static uint32_t last_cont_level = 0;
1779 #endif
1780 size_t i;
1781 struct magic *m;
1782 const char *l = line;
1783 char *t;
1784 int op;
1785 uint32_t cont_level;
1786 int32_t diff;
1787
1788 cont_level = 0;
1789
1790 /*
1791 * Parse the offset.
1792 */
1793 while (*l == '>') {
1794 ++l; /* step over */
1795 cont_level++;
1796 }
1797 #ifdef ENABLE_CONDITIONALS
1798 if (cont_level == 0 || cont_level > last_cont_level)
1799 if (file_check_mem(ms, cont_level) == -1)
1800 return -1;
1801 last_cont_level = cont_level;
1802 #endif
1803 if (cont_level != 0) {
1804 if (me->mp == NULL) {
1805 file_magerror(ms, "No current entry for continuation");
1806 return -1;
1807 }
1808 if (me->cont_count == 0) {
1809 file_magerror(ms, "Continuations present with 0 count");
1810 return -1;
1811 }
1812 m = &me->mp[me->cont_count - 1];
1813 diff = CAST(int32_t, cont_level) - CAST(int32_t, m->cont_level);
1814 if (diff > 1)
1815 file_magwarn(ms, "New continuation level %u is more "
1816 "than one larger than current level %u", cont_level,
1817 m->cont_level);
1818 if (me->cont_count == me->max_count) {
1819 struct magic *nm;
1820 size_t cnt = me->max_count + ALLOC_CHUNK;
1821 if ((nm = CAST(struct magic *, erealloc(me->mp,
1822 sizeof(*nm) * cnt))) == NULL) {
1823 file_oomem(ms, sizeof(*nm) * cnt);
1824 return -1;
1825 }
1826 me->mp = nm;
1827 me->max_count = CAST(uint32_t, cnt);
1828 }
1829 m = &me->mp[me->cont_count++];
1830 (void)memset(m, 0, sizeof(*m));
1831 m->cont_level = cont_level;
1832 } else {
1833 static const size_t len = sizeof(*m) * ALLOC_CHUNK;
1834 if (me->mp != NULL)
1835 return 1;
1836 if ((m = CAST(struct magic *, emalloc(len))) == NULL) {
1837 file_oomem(ms, len);
1838 return -1;
1839 }
1840 me->mp = m;
1841 me->max_count = ALLOC_CHUNK;
1842 (void)memset(m, 0, sizeof(*m));
1843 m->factor_op = FILE_FACTOR_OP_NONE;
1844 m->cont_level = 0;
1845 me->cont_count = 1;
1846 }
1847 m->lineno = CAST(uint32_t, lineno);
1848
1849 if (*l == '&') { /* m->cont_level == 0 checked below. */
1850 ++l; /* step over */
1851 m->flag |= OFFADD;
1852 }
1853 if (*l == '(') {
1854 ++l; /* step over */
1855 m->flag |= INDIR;
1856 if (m->flag & OFFADD)
1857 m->flag = (m->flag & ~OFFADD) | INDIROFFADD;
1858
1859 if (*l == '&') { /* m->cont_level == 0 checked below */
1860 ++l; /* step over */
1861 m->flag |= OFFADD;
1862 }
1863 }
1864 /* Indirect offsets are not valid at level 0. */
1865 if (m->cont_level == 0 && (m->flag & (OFFADD | INDIROFFADD))) {
1866 if (ms->flags & MAGIC_CHECK)
1867 file_magwarn(ms, "relative offset at level 0");
1868 return -1;
1869 }
1870
1871 /* get offset, then skip over it */
1872 m->offset = CAST(int32_t, strtol(l, &t, 0));
1873 if (l == t) {
1874 if (ms->flags & MAGIC_CHECK)
1875 file_magwarn(ms, "offset `%s' invalid", l);
1876 return -1;
1877 }
1878 l = t;
1879
1880 if (m->flag & INDIR) {
1881 m->in_type = FILE_LONG;
1882 m->in_offset = 0;
1883 m->in_op = 0;
1884 /*
1885 * read [.,lbs][+-]nnnnn)
1886 */
1887 if (*l == '.' || *l == ',') {
1888 if (*l == ',')
1889 m->in_op |= FILE_OPSIGNED;
1890 l++;
1891 switch (*l) {
1892 case 'l':
1893 m->in_type = FILE_LELONG;
1894 break;
1895 case 'L':
1896 m->in_type = FILE_BELONG;
1897 break;
1898 case 'm':
1899 m->in_type = FILE_MELONG;
1900 break;
1901 case 'h':
1902 case 's':
1903 m->in_type = FILE_LESHORT;
1904 break;
1905 case 'H':
1906 case 'S':
1907 m->in_type = FILE_BESHORT;
1908 break;
1909 case 'c':
1910 case 'b':
1911 case 'C':
1912 case 'B':
1913 m->in_type = FILE_BYTE;
1914 break;
1915 case 'e':
1916 case 'f':
1917 case 'g':
1918 m->in_type = FILE_LEDOUBLE;
1919 break;
1920 case 'E':
1921 case 'F':
1922 case 'G':
1923 m->in_type = FILE_BEDOUBLE;
1924 break;
1925 case 'i':
1926 m->in_type = FILE_LEID3;
1927 break;
1928 case 'I':
1929 m->in_type = FILE_BEID3;
1930 break;
1931 case 'q':
1932 m->in_type = FILE_LEQUAD;
1933 break;
1934 case 'Q':
1935 m->in_type = FILE_BEQUAD;
1936 break;
1937 default:
1938 if (ms->flags & MAGIC_CHECK)
1939 file_magwarn(ms,
1940 "indirect offset type `%c' invalid",
1941 *l);
1942 return -1;
1943 }
1944 l++;
1945 }
1946
1947 if (*l == '~') {
1948 m->in_op |= FILE_OPINVERSE;
1949 l++;
1950 }
1951 if ((op = get_op(*l)) != -1) {
1952 m->in_op |= op;
1953 l++;
1954 }
1955 if (*l == '(') {
1956 m->in_op |= FILE_OPINDIRECT;
1957 l++;
1958 }
1959 if (isdigit(CAST(unsigned char, *l)) || *l == '-') {
1960 m->in_offset = CAST(int32_t, strtol(l, &t, 0));
1961 if (l == t) {
1962 if (ms->flags & MAGIC_CHECK)
1963 file_magwarn(ms,
1964 "in_offset `%s' invalid", l);
1965 return -1;
1966 }
1967 l = t;
1968 }
1969 if (*l++ != ')' ||
1970 ((m->in_op & FILE_OPINDIRECT) && *l++ != ')')) {
1971 if (ms->flags & MAGIC_CHECK)
1972 file_magwarn(ms,
1973 "missing ')' in indirect offset");
1974 return -1;
1975 }
1976 }
1977 EATAB;
1978
1979 #ifdef ENABLE_CONDITIONALS
1980 m->cond = get_cond(l, &l);
1981 if (check_cond(ms, m->cond, cont_level) == -1)
1982 return -1;
1983
1984 EATAB;
1985 #endif
1986
1987 /*
1988 * Parse the type.
1989 */
1990 if (*l == 'u') {
1991 /*
1992 * Try it as a keyword type prefixed by "u"; match what
1993 * follows the "u". If that fails, try it as an SUS
1994 * integer type.
1995 */
1996 m->type = get_type(type_tbl, l + 1, &l);
1997 if (m->type == FILE_INVALID) {
1998 /*
1999 * Not a keyword type; parse it as an SUS type,
2000 * 'u' possibly followed by a number or C/S/L.
2001 */
2002 m->type = get_standard_integer_type(l, &l);
2003 }
2004 /* It's unsigned. */
2005 if (m->type != FILE_INVALID)
2006 m->flag |= UNSIGNED;
2007 } else {
2008 /*
2009 * Try it as a keyword type. If that fails, try it as
2010 * an SUS integer type if it begins with "d" or as an
2011 * SUS string type if it begins with "s". In any case,
2012 * it's not unsigned.
2013 */
2014 m->type = get_type(type_tbl, l, &l);
2015 if (m->type == FILE_INVALID) {
2016 /*
2017 * Not a keyword type; parse it as an SUS type,
2018 * either 'd' possibly followed by a number or
2019 * C/S/L, or just 's'.
2020 */
2021 if (*l == 'd')
2022 m->type = get_standard_integer_type(l, &l);
2023 else if (*l == 's'
2024 && !isalpha(CAST(unsigned char, l[1]))) {
2025 m->type = FILE_STRING;
2026 ++l;
2027 }
2028 }
2029 }
2030
2031 if (m->type == FILE_INVALID) {
2032 /* Not found - try it as a special keyword. */
2033 m->type = get_type(special_tbl, l, &l);
2034 }
2035
2036 if (m->type == FILE_INVALID) {
2037 if (ms->flags & MAGIC_CHECK)
2038 file_magwarn(ms, "type `%s' invalid", l);
2039 return -1;
2040 }
2041
2042 /* New-style anding: "0 byte&0x80 =0x80 dynamically linked" */
2043 /* New and improved: ~ & | ^ + - * / % -- exciting, isn't it? */
2044
2045 m->mask_op = 0;
2046 if (*l == '~') {
2047 if (!IS_LIBMAGIC_STRING(m->type))
2048 m->mask_op |= FILE_OPINVERSE;
2049 else if (ms->flags & MAGIC_CHECK)
2050 file_magwarn(ms, "'~' invalid for string types");
2051 ++l;
2052 }
2053 m->str_range = 0;
2054 m->str_flags = m->type == FILE_PSTRING ? PSTRING_1_LE : 0;
2055 if ((op = get_op(*l)) != -1) {
2056 if (IS_LIBMAGIC_STRING(m->type)) {
2057 int r;
2058
2059 if (op != FILE_OPDIVIDE) {
2060 if (ms->flags & MAGIC_CHECK)
2061 file_magwarn(ms,
2062 "invalid string/indirect op: "
2063 "`%c'", *t);
2064 return -1;
2065 }
2066
2067 if (m->type == FILE_INDIRECT)
2068 r = parse_indirect_modifier(ms, m, &l);
2069 else
2070 r = parse_string_modifier(ms, m, &l);
2071 if (r == -1)
2072 return -1;
2073 } else
2074 parse_op_modifier(ms, m, &l, op);
2075 }
2076
2077 /*
2078 * We used to set mask to all 1's here, instead let's just not do
2079 * anything if mask = 0 (unless you have a better idea)
2080 */
2081 EATAB;
2082
2083 switch (*l) {
2084 case '>':
2085 case '<':
2086 m->reln = *l;
2087 ++l;
2088 if (*l == '=') {
2089 if (ms->flags & MAGIC_CHECK) {
2090 file_magwarn(ms, "%c= not supported",
2091 m->reln);
2092 return -1;
2093 }
2094 ++l;
2095 }
2096 break;
2097 /* Old-style anding: "0 byte &0x80 dynamically linked" */
2098 case '&':
2099 case '^':
2100 case '=':
2101 m->reln = *l;
2102 ++l;
2103 if (*l == '=') {
2104 /* HP compat: ignore &= etc. */
2105 ++l;
2106 }
2107 break;
2108 case '!':
2109 m->reln = *l;
2110 ++l;
2111 break;
2112 default:
2113 m->reln = '='; /* the default relation */
2114 if (*l == 'x' && ((isascii(CAST(unsigned char, l[1])) &&
2115 isspace(CAST(unsigned char, l[1]))) || !l[1])) {
2116 m->reln = *l;
2117 ++l;
2118 }
2119 break;
2120 }
2121 /*
2122 * Grab the value part, except for an 'x' reln.
2123 */
2124 if (m->reln != 'x' && getvalue(ms, m, &l, action))
2125 return -1;
2126
2127 /*
2128 * TODO finish this macro and start using it!
2129 * #define offsetcheck {if (offset > ms->bytes_max -1)
2130 * magwarn("offset too big"); }
2131 */
2132
2133 /*
2134 * Now get last part - the description
2135 */
2136 EATAB;
2137 if (l[0] == '\b') {
2138 ++l;
2139 m->flag |= NOSPACE;
2140 } else if ((l[0] == '\\') && (l[1] == 'b')) {
2141 ++l;
2142 ++l;
2143 m->flag |= NOSPACE;
2144 }
2145 for (i = 0; (m->desc[i++] = *l++) != '\0' && i < sizeof(m->desc); )
2146 continue;
2147 if (i == sizeof(m->desc)) {
2148 m->desc[sizeof(m->desc) - 1] = '\0';
2149 if (ms->flags & MAGIC_CHECK)
2150 file_magwarn(ms, "description `%s' truncated", m->desc);
2151 }
2152
2153 /*
2154 * We only do this check while compiling, or if any of the magic
2155 * files were not compiled.
2156 */
2157 if (ms->flags & MAGIC_CHECK) {
2158 if (check_format(ms, m) == -1)
2159 return -1;
2160 }
2161 m->mimetype[0] = '\0'; /* initialise MIME type to none */
2162 return 0;
2163 }
2164
2165 /*
2166 * parse a STRENGTH annotation line from magic file, put into magic[index - 1]
2167 * if valid
2168 */
2169 private int
parse_strength(struct magic_set * ms,struct magic_entry * me,const char * line)2170 parse_strength(struct magic_set *ms, struct magic_entry *me, const char *line)
2171 {
2172 const char *l = line;
2173 char *el;
2174 unsigned long factor;
2175 struct magic *m = &me->mp[0];
2176
2177 if (m->factor_op != FILE_FACTOR_OP_NONE) {
2178 file_magwarn(ms,
2179 "Current entry already has a strength type: %c %d",
2180 m->factor_op, m->factor);
2181 return -1;
2182 }
2183 if (m->type == FILE_NAME) {
2184 file_magwarn(ms, "%s: Strength setting is not supported in "
2185 "\"name\" magic entries", m->value.s);
2186 return -1;
2187 }
2188 EATAB;
2189 switch (*l) {
2190 case FILE_FACTOR_OP_NONE:
2191 case FILE_FACTOR_OP_PLUS:
2192 case FILE_FACTOR_OP_MINUS:
2193 case FILE_FACTOR_OP_TIMES:
2194 case FILE_FACTOR_OP_DIV:
2195 m->factor_op = *l++;
2196 break;
2197 default:
2198 file_magwarn(ms, "Unknown factor op `%c'", *l);
2199 return -1;
2200 }
2201 EATAB;
2202 factor = strtoul(l, &el, 0);
2203 if (factor > 255) {
2204 file_magwarn(ms, "Too large factor `%lu'", factor);
2205 goto out;
2206 }
2207 if (*el && !isspace(CAST(unsigned char, *el))) {
2208 file_magwarn(ms, "Bad factor `%s'", l);
2209 goto out;
2210 }
2211 m->factor = CAST(uint8_t, factor);
2212 if (m->factor == 0 && m->factor_op == FILE_FACTOR_OP_DIV) {
2213 file_magwarn(ms, "Cannot have factor op `%c' and factor %u",
2214 m->factor_op, m->factor);
2215 goto out;
2216 }
2217 return 0;
2218 out:
2219 m->factor_op = FILE_FACTOR_OP_NONE;
2220 m->factor = 0;
2221 return -1;
2222 }
2223
2224 private int
goodchar(unsigned char x,const char * extra)2225 goodchar(unsigned char x, const char *extra)
2226 {
2227 return (isascii(x) && isalnum(x)) || strchr(extra, x);
2228 }
2229
2230 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)2231 parse_extra(struct magic_set *ms, struct magic_entry *me, const char *line,
2232 zend_off_t off, size_t len, const char *name, const char *extra, int nt)
2233 {
2234 size_t i;
2235 const char *l = line;
2236 struct magic *m = &me->mp[me->cont_count == 0 ? 0 : me->cont_count - 1];
2237 char *buf = CAST(char *, CAST(void *, m)) + off;
2238
2239 if (buf[0] != '\0') {
2240 len = nt ? strlen(buf) : len;
2241 file_magwarn(ms, "Current entry already has a %s type "
2242 "`%.*s', new type `%s'", name, CAST(int, len), buf, l);
2243 return -1;
2244 }
2245
2246 if (*m->desc == '\0') {
2247 file_magwarn(ms, "Current entry does not yet have a "
2248 "description for adding a %s type", name);
2249 return -1;
2250 }
2251
2252 EATAB;
2253 for (i = 0; *l && i < len && goodchar(*l, extra); buf[i++] = *l++)
2254 continue;
2255
2256 if (i == len && *l) {
2257 if (nt)
2258 buf[len - 1] = '\0';
2259 if (ms->flags & MAGIC_CHECK)
2260 file_magwarn(ms, "%s type `%s' truncated %"
2261 SIZE_T_FORMAT "u", name, line, i);
2262 } else {
2263 if (!isspace(CAST(unsigned char, *l)) && !goodchar(*l, extra))
2264 file_magwarn(ms, "%s type `%s' has bad char '%c'",
2265 name, line, *l);
2266 if (nt)
2267 buf[i] = '\0';
2268 }
2269
2270 if (i > 0)
2271 return 0;
2272
2273 file_magerror(ms, "Bad magic entry '%s'", line);
2274 return -1;
2275 }
2276
2277 /*
2278 * Parse an Apple CREATOR/TYPE annotation from magic file and put it into
2279 * magic[index - 1]
2280 */
2281 private int
parse_apple(struct magic_set * ms,struct magic_entry * me,const char * line)2282 parse_apple(struct magic_set *ms, struct magic_entry *me, const char *line)
2283 {
2284 struct magic *m = &me->mp[0];
2285
2286 return parse_extra(ms, me, line,
2287 CAST(off_t, offsetof(struct magic, apple)),
2288 sizeof(m->apple), "APPLE", "!+-./?", 0);
2289 }
2290
2291 /*
2292 * Parse a comma-separated list of extensions
2293 */
2294 private int
parse_ext(struct magic_set * ms,struct magic_entry * me,const char * line)2295 parse_ext(struct magic_set *ms, struct magic_entry *me, const char *line)
2296 {
2297 struct magic *m = &me->mp[0];
2298
2299 return parse_extra(ms, me, line,
2300 CAST(off_t, offsetof(struct magic, ext)),
2301 sizeof(m->ext), "EXTENSION", ",!+-/@?_$", 0);
2302 }
2303
2304 /*
2305 * parse a MIME annotation line from magic file, put into magic[index - 1]
2306 * if valid
2307 */
2308 private int
parse_mime(struct magic_set * ms,struct magic_entry * me,const char * line)2309 parse_mime(struct magic_set *ms, struct magic_entry *me, const char *line)
2310 {
2311 struct magic *m = &me->mp[0];
2312
2313 return parse_extra(ms, me, line,
2314 CAST(zend_off_t, offsetof(struct magic, mimetype)),
2315 sizeof(m->mimetype), "MIME", "+-/.$?:{}", 1);
2316 }
2317
2318 private int
check_format_type(const char * ptr,int type,const char ** estr)2319 check_format_type(const char *ptr, int type, const char **estr)
2320 {
2321 int quad = 0, h;
2322 size_t len, cnt;
2323 if (*ptr == '\0') {
2324 /* Missing format string; bad */
2325 *estr = "missing format spec";
2326 return -1;
2327 }
2328
2329 switch (file_formats[type]) {
2330 case FILE_FMT_QUAD:
2331 quad = 1;
2332 /*FALLTHROUGH*/
2333 case FILE_FMT_NUM:
2334 if (quad == 0) {
2335 switch (type) {
2336 case FILE_BYTE:
2337 h = 2;
2338 break;
2339 case FILE_SHORT:
2340 case FILE_BESHORT:
2341 case FILE_LESHORT:
2342 h = 1;
2343 break;
2344 case FILE_LONG:
2345 case FILE_BELONG:
2346 case FILE_LELONG:
2347 case FILE_MELONG:
2348 case FILE_LEID3:
2349 case FILE_BEID3:
2350 case FILE_INDIRECT:
2351 h = 0;
2352 break;
2353 default:
2354 abort();
2355 }
2356 } else
2357 h = 0;
2358 if (*ptr == '-')
2359 ptr++;
2360 if (*ptr == '.')
2361 ptr++;
2362 if (*ptr == '#')
2363 ptr++;
2364 #define CHECKLEN() do { \
2365 for (len = cnt = 0; isdigit(CAST(unsigned char, *ptr)); ptr++, cnt++) \
2366 len = len * 10 + (*ptr - '0'); \
2367 if (cnt > 5 || len > 1024) \
2368 goto toolong; \
2369 } while (/*CONSTCOND*/0)
2370
2371 CHECKLEN();
2372 if (*ptr == '.')
2373 ptr++;
2374 CHECKLEN();
2375 if (quad) {
2376 if (*ptr++ != 'l')
2377 goto invalid;
2378 if (*ptr++ != 'l')
2379 goto invalid;
2380 }
2381
2382 switch (*ptr++) {
2383 #ifdef STRICT_FORMAT /* "long" formats are int formats for us */
2384 /* so don't accept the 'l' modifier */
2385 case 'l':
2386 switch (*ptr++) {
2387 case 'i':
2388 case 'd':
2389 case 'u':
2390 case 'o':
2391 case 'x':
2392 case 'X':
2393 if (h == 0)
2394 return 0;
2395 /*FALLTHROUGH*/
2396 default:
2397 goto invalid;
2398 }
2399
2400 /*
2401 * Don't accept h and hh modifiers. They make writing
2402 * magic entries more complicated, for very little benefit
2403 */
2404 case 'h':
2405 if (h-- <= 0)
2406 goto invalid;
2407 switch (*ptr++) {
2408 case 'h':
2409 if (h-- <= 0)
2410 goto invalid;
2411 switch (*ptr++) {
2412 case 'i':
2413 case 'd':
2414 case 'u':
2415 case 'o':
2416 case 'x':
2417 case 'X':
2418 return 0;
2419 default:
2420 goto invalid;
2421 }
2422 case 'i':
2423 case 'd':
2424 case 'u':
2425 case 'o':
2426 case 'x':
2427 case 'X':
2428 if (h == 0)
2429 return 0;
2430 /*FALLTHROUGH*/
2431 default:
2432 goto invalid;
2433 }
2434 #endif
2435 case 'c':
2436 if (h == 2)
2437 return 0;
2438 goto invalid;
2439 case 'i':
2440 case 'd':
2441 case 'u':
2442 case 'o':
2443 case 'x':
2444 case 'X':
2445 #ifdef STRICT_FORMAT
2446 if (h == 0)
2447 return 0;
2448 /*FALLTHROUGH*/
2449 #else
2450 return 0;
2451 #endif
2452 default:
2453 goto invalid;
2454 }
2455
2456 case FILE_FMT_FLOAT:
2457 case FILE_FMT_DOUBLE:
2458 if (*ptr == '-')
2459 ptr++;
2460 if (*ptr == '.')
2461 ptr++;
2462 CHECKLEN();
2463 if (*ptr == '.')
2464 ptr++;
2465 CHECKLEN();
2466 switch (*ptr++) {
2467 case 'e':
2468 case 'E':
2469 case 'f':
2470 case 'F':
2471 case 'g':
2472 case 'G':
2473 return 0;
2474
2475 default:
2476 goto invalid;
2477 }
2478
2479
2480 case FILE_FMT_STR:
2481 if (*ptr == '-')
2482 ptr++;
2483 while (isdigit(CAST(unsigned char, *ptr)))
2484 ptr++;
2485 if (*ptr == '.') {
2486 ptr++;
2487 while (isdigit(CAST(unsigned char , *ptr)))
2488 ptr++;
2489 }
2490
2491 switch (*ptr++) {
2492 case 's':
2493 return 0;
2494 default:
2495 goto invalid;
2496 }
2497
2498 default:
2499 /* internal error */
2500 abort();
2501 }
2502 invalid:
2503 *estr = "not valid";
2504 toolong:
2505 *estr = "too long";
2506 return -1;
2507 }
2508
2509 /*
2510 * Check that the optional printf format in description matches
2511 * the type of the magic.
2512 */
2513 private int
check_format(struct magic_set * ms,struct magic * m)2514 check_format(struct magic_set *ms, struct magic *m)
2515 {
2516 char *ptr;
2517 const char *estr;
2518
2519 for (ptr = m->desc; *ptr; ptr++)
2520 if (*ptr == '%')
2521 break;
2522 if (*ptr == '\0') {
2523 /* No format string; ok */
2524 return 1;
2525 }
2526
2527 assert(file_nformats == file_nnames);
2528
2529 if (m->type >= file_nformats) {
2530 file_magwarn(ms, "Internal error inconsistency between "
2531 "m->type and format strings");
2532 return -1;
2533 }
2534 if (file_formats[m->type] == FILE_FMT_NONE) {
2535 file_magwarn(ms, "No format string for `%s' with description "
2536 "`%s'", m->desc, file_names[m->type]);
2537 return -1;
2538 }
2539
2540 ptr++;
2541 if (check_format_type(ptr, m->type, &estr) == -1) {
2542 /*
2543 * TODO: this error message is unhelpful if the format
2544 * string is not one character long
2545 */
2546 file_magwarn(ms, "Printf format is %s for type "
2547 "`%s' in description `%s'", estr,
2548 file_names[m->type], m->desc);
2549 return -1;
2550 }
2551
2552 for (; *ptr; ptr++) {
2553 if (*ptr == '%') {
2554 file_magwarn(ms,
2555 "Too many format strings (should have at most one) "
2556 "for `%s' with description `%s'",
2557 file_names[m->type], m->desc);
2558 return -1;
2559 }
2560 }
2561 return 0;
2562 }
2563
2564 /*
2565 * Read a numeric value from a pointer, into the value union of a magic
2566 * pointer, according to the magic type. Update the string pointer to point
2567 * just after the number read. Return 0 for success, non-zero for failure.
2568 */
2569 private int
getvalue(struct magic_set * ms,struct magic * m,const char ** p,int action)2570 getvalue(struct magic_set *ms, struct magic *m, const char **p, int action)
2571 {
2572 char *ep;
2573 uint64_t ull;
2574
2575 switch (m->type) {
2576 case FILE_BESTRING16:
2577 case FILE_LESTRING16:
2578 case FILE_STRING:
2579 case FILE_PSTRING:
2580 case FILE_REGEX:
2581 case FILE_SEARCH:
2582 case FILE_NAME:
2583 case FILE_USE:
2584 case FILE_DER:
2585 *p = getstr(ms, m, *p, action == FILE_COMPILE);
2586 if (*p == NULL) {
2587 if (ms->flags & MAGIC_CHECK)
2588 file_magwarn(ms, "cannot get string from `%s'",
2589 m->value.s);
2590 return -1;
2591 }
2592 if (m->type == FILE_REGEX) {
2593 zval pattern;
2594 int options = 0;
2595 pcre_cache_entry *pce;
2596
2597 convert_libmagic_pattern(&pattern, m->value.s, strlen(m->value.s), options);
2598
2599 if ((pce = pcre_get_compiled_regex_cache(Z_STR(pattern))) == NULL) {
2600 zval_dtor(&pattern);
2601 return -1;
2602 }
2603 zval_dtor(&pattern);
2604
2605 return 0;
2606 }
2607 return 0;
2608 default:
2609 if (m->reln == 'x')
2610 return 0;
2611 break;
2612 }
2613
2614 switch (m->type) {
2615 case FILE_FLOAT:
2616 case FILE_BEFLOAT:
2617 case FILE_LEFLOAT:
2618 errno = 0;
2619 #ifdef HAVE_STRTOF
2620 m->value.f = strtof(*p, &ep);
2621 #else
2622 m->value.f = (float)strtod(*p, &ep);
2623 #endif
2624 if (errno == 0)
2625 *p = ep;
2626 return 0;
2627 case FILE_DOUBLE:
2628 case FILE_BEDOUBLE:
2629 case FILE_LEDOUBLE:
2630 errno = 0;
2631 m->value.d = strtod(*p, &ep);
2632 if (errno == 0)
2633 *p = ep;
2634 return 0;
2635 default:
2636 errno = 0;
2637 ull = CAST(uint64_t, strtoull(*p, &ep, 0));
2638 m->value.q = file_signextend(ms, m, ull);
2639 if (*p == ep) {
2640 file_magwarn(ms, "Unparseable number `%s'", *p);
2641 } else {
2642 size_t ts = typesize(m->type);
2643 uint64_t x;
2644 const char *q;
2645
2646 if (ts == CAST(size_t, ~0)) {
2647 file_magwarn(ms,
2648 "Expected numeric type got `%s'",
2649 type_tbl[m->type].name);
2650 }
2651 for (q = *p; isspace(CAST(unsigned char, *q)); q++)
2652 continue;
2653 if (*q == '-')
2654 ull = -CAST(int64_t, ull);
2655 switch (ts) {
2656 case 1:
2657 x = CAST(uint64_t, ull & ~0xffULL);
2658 break;
2659 case 2:
2660 x = CAST(uint64_t, ull & ~0xffffULL);
2661 break;
2662 case 4:
2663 x = CAST(uint64_t, ull & ~0xffffffffULL);
2664 break;
2665 case 8:
2666 x = 0;
2667 break;
2668 default:
2669 abort();
2670 }
2671 if (x) {
2672 file_magwarn(ms, "Overflow for numeric"
2673 " type `%s' value %#" PRIx64,
2674 type_tbl[m->type].name, ull);
2675 }
2676 }
2677 if (errno == 0) {
2678 *p = ep;
2679 eatsize(p);
2680 }
2681 return 0;
2682 }
2683 }
2684
2685 /*
2686 * Convert a string containing C character escapes. Stop at an unescaped
2687 * space or tab.
2688 * Copy the converted version to "m->value.s", and the length in m->vallen.
2689 * Return updated scan pointer as function result. Warn if set.
2690 */
2691 private const char *
getstr(struct magic_set * ms,struct magic * m,const char * s,int warn)2692 getstr(struct magic_set *ms, struct magic *m, const char *s, int warn)
2693 {
2694 const char *origs = s;
2695 char *p = m->value.s;
2696 size_t plen = sizeof(m->value.s);
2697 char *origp = p;
2698 char *pmax = p + plen - 1;
2699 int c;
2700 int val;
2701
2702 while ((c = *s++) != '\0') {
2703 if (isspace(CAST(unsigned char, c)))
2704 break;
2705 if (p >= pmax) {
2706 file_error(ms, 0, "string too long: `%s'", origs);
2707 return NULL;
2708 }
2709 if (c == '\\') {
2710 switch(c = *s++) {
2711
2712 case '\0':
2713 if (warn)
2714 file_magwarn(ms, "incomplete escape");
2715 s--;
2716 goto out;
2717
2718 case '\t':
2719 if (warn) {
2720 file_magwarn(ms,
2721 "escaped tab found, use \\t instead");
2722 warn = 0; /* already did */
2723 }
2724 /*FALLTHROUGH*/
2725 default:
2726 if (warn) {
2727 if (isprint(CAST(unsigned char, c))) {
2728 /* Allow escaping of
2729 * ``relations'' */
2730 if (strchr("<>&^=!", c) == NULL
2731 && (m->type != FILE_REGEX ||
2732 strchr("[]().*?^$|{}", c)
2733 == NULL)) {
2734 file_magwarn(ms, "no "
2735 "need to escape "
2736 "`%c'", c);
2737 }
2738 } else {
2739 file_magwarn(ms,
2740 "unknown escape sequence: "
2741 "\\%03o", c);
2742 }
2743 }
2744 /*FALLTHROUGH*/
2745 /* space, perhaps force people to use \040? */
2746 case ' ':
2747 #if 0
2748 /*
2749 * Other things people escape, but shouldn't need to,
2750 * so we disallow them
2751 */
2752 case '\'':
2753 case '"':
2754 case '?':
2755 #endif
2756 /* Relations */
2757 case '>':
2758 case '<':
2759 case '&':
2760 case '^':
2761 case '=':
2762 case '!':
2763 /* and baskslash itself */
2764 case '\\':
2765 *p++ = CAST(char, c);
2766 break;
2767
2768 case 'a':
2769 *p++ = '\a';
2770 break;
2771
2772 case 'b':
2773 *p++ = '\b';
2774 break;
2775
2776 case 'f':
2777 *p++ = '\f';
2778 break;
2779
2780 case 'n':
2781 *p++ = '\n';
2782 break;
2783
2784 case 'r':
2785 *p++ = '\r';
2786 break;
2787
2788 case 't':
2789 *p++ = '\t';
2790 break;
2791
2792 case 'v':
2793 *p++ = '\v';
2794 break;
2795
2796 /* \ and up to 3 octal digits */
2797 case '0':
2798 case '1':
2799 case '2':
2800 case '3':
2801 case '4':
2802 case '5':
2803 case '6':
2804 case '7':
2805 val = c - '0';
2806 c = *s++; /* try for 2 */
2807 if (c >= '0' && c <= '7') {
2808 val = (val << 3) | (c - '0');
2809 c = *s++; /* try for 3 */
2810 if (c >= '0' && c <= '7')
2811 val = (val << 3) | (c-'0');
2812 else
2813 --s;
2814 }
2815 else
2816 --s;
2817 *p++ = CAST(char, val);
2818 break;
2819
2820 /* \x and up to 2 hex digits */
2821 case 'x':
2822 val = 'x'; /* Default if no digits */
2823 c = hextoint(*s++); /* Get next char */
2824 if (c >= 0) {
2825 val = c;
2826 c = hextoint(*s++);
2827 if (c >= 0)
2828 val = (val << 4) + c;
2829 else
2830 --s;
2831 } else
2832 --s;
2833 *p++ = CAST(char, val);
2834 break;
2835 }
2836 } else
2837 *p++ = CAST(char, c);
2838 }
2839 --s;
2840 out:
2841 *p = '\0';
2842 m->vallen = CAST(unsigned char, (p - origp));
2843 if (m->type == FILE_PSTRING)
2844 m->vallen += CAST(unsigned char, file_pstring_length_size(m));
2845 return s;
2846 }
2847
2848
2849 /* Single hex char to int; -1 if not a hex char. */
2850 private int
hextoint(int c)2851 hextoint(int c)
2852 {
2853 if (!isascii(CAST(unsigned char, c)))
2854 return -1;
2855 if (isdigit(CAST(unsigned char, c)))
2856 return c - '0';
2857 if ((c >= 'a') && (c <= 'f'))
2858 return c + 10 - 'a';
2859 if (( c>= 'A') && (c <= 'F'))
2860 return c + 10 - 'A';
2861 return -1;
2862 }
2863
2864
2865 /*
2866 * Print a string containing C character escapes.
2867 */
2868 protected void
file_showstr(FILE * fp,const char * s,size_t len)2869 file_showstr(FILE *fp, const char *s, size_t len)
2870 {
2871 char c;
2872
2873 for (;;) {
2874 if (len == ~0U) {
2875 c = *s++;
2876 if (c == '\0')
2877 break;
2878 }
2879 else {
2880 if (len-- == 0)
2881 break;
2882 c = *s++;
2883 }
2884 if (c >= 040 && c <= 0176) /* TODO isprint && !iscntrl */
2885 (void) fputc(c, fp);
2886 else {
2887 (void) fputc('\\', fp);
2888 switch (c) {
2889 case '\a':
2890 (void) fputc('a', fp);
2891 break;
2892
2893 case '\b':
2894 (void) fputc('b', fp);
2895 break;
2896
2897 case '\f':
2898 (void) fputc('f', fp);
2899 break;
2900
2901 case '\n':
2902 (void) fputc('n', fp);
2903 break;
2904
2905 case '\r':
2906 (void) fputc('r', fp);
2907 break;
2908
2909 case '\t':
2910 (void) fputc('t', fp);
2911 break;
2912
2913 case '\v':
2914 (void) fputc('v', fp);
2915 break;
2916
2917 default:
2918 (void) fprintf(fp, "%.3o", c & 0377);
2919 break;
2920 }
2921 }
2922 }
2923 }
2924
2925 /*
2926 * eatsize(): Eat the size spec from a number [eg. 10UL]
2927 */
2928 private void
eatsize(const char ** p)2929 eatsize(const char **p)
2930 {
2931 const char *l = *p;
2932
2933 if (LOWCASE(*l) == 'u')
2934 l++;
2935
2936 switch (LOWCASE(*l)) {
2937 case 'l': /* long */
2938 case 's': /* short */
2939 case 'h': /* short */
2940 case 'b': /* char/byte */
2941 case 'c': /* char/byte */
2942 l++;
2943 /*FALLTHROUGH*/
2944 default:
2945 break;
2946 }
2947
2948 *p = l;
2949 }
2950
2951 #if 0
2952 /*
2953 * handle a buffer containing a compiled file.
2954 */
2955 private struct magic_map *
2956 apprentice_buf(struct magic_set *ms, struct magic *buf, size_t len)
2957 {
2958 struct magic_map *map;
2959
2960 if ((map = CAST(struct magic_map *, calloc(1, sizeof(*map)))) == NULL) {
2961 file_oomem(ms, sizeof(*map));
2962 return NULL;
2963 }
2964 map->len = len;
2965 map->p = buf;
2966 map->type = MAP_TYPE_USER;
2967 if (check_buffer(ms, map, "buffer") != 0) {
2968 apprentice_unmap(map);
2969 return NULL;
2970 }
2971 return map;
2972 }
2973 #endif
2974
2975 /*
2976 * handle a compiled file.
2977 */
2978
2979 private struct magic_map *
apprentice_map(struct magic_set * ms,const char * fn)2980 apprentice_map(struct magic_set *ms, const char *fn)
2981 {
2982 uint32_t *ptr;
2983 uint32_t version, entries = 0, nentries;
2984 int needsbyteswap;
2985 char *dbname = NULL;
2986 struct magic_map *map;
2987 size_t i;
2988 php_stream *stream = NULL;
2989 php_stream_statbuf st;
2990
2991
2992
2993 if ((map = CAST(struct magic_map *, ecalloc(1, sizeof(*map)))) == NULL) {
2994 file_oomem(ms, sizeof(*map));
2995 return NULL;
2996 }
2997
2998 if (fn == NULL) {
2999 map->p = (void *)&php_magic_database;
3000 goto internal_loaded;
3001 }
3002
3003 #ifdef PHP_WIN32
3004 /* Don't bother on windows with php_stream_open_wrapper,
3005 return to give apprentice_load() a chance. */
3006 if (php_stream_stat_path_ex((char *)fn, 0, &st, NULL) == SUCCESS) {
3007 if (st.sb.st_mode & S_IFDIR) {
3008 goto error;
3009 }
3010 }
3011 #endif
3012
3013 dbname = mkdbname(ms, fn, 0);
3014 if (dbname == NULL)
3015 goto error;
3016
3017 stream = php_stream_open_wrapper((char *)fn, "rb", REPORT_ERRORS, NULL);
3018
3019 if (!stream) {
3020 goto error;
3021 }
3022
3023 #ifndef PHP_WIN32
3024 if (php_stream_stat(stream, &st) < 0) {
3025 file_error(ms, errno, "cannot stat `%s'", dbname);
3026 goto error;
3027 }
3028 #endif
3029 if (st.sb.st_size < 8 || st.sb.st_size > maxoff_t()) {
3030 file_error(ms, 0, "file `%s' is too %s", dbname,
3031 st.sb.st_size < 8 ? "small" : "large");
3032 goto error;
3033 }
3034
3035 map->type = MAP_TYPE_MALLOC;
3036 map->len = CAST(size_t, st.sb.st_size);
3037 map->p = CAST(void *, emalloc(map->len));
3038
3039 if (php_stream_read(stream, map->p, (size_t)st.sb.st_size) != (size_t)st.sb.st_size) {
3040 file_badread(ms);
3041 goto error;
3042 }
3043
3044 php_stream_close(stream);
3045 stream = NULL;
3046
3047 internal_loaded:
3048 ptr = (uint32_t *)(void *)map->p;
3049 if (*ptr != MAGICNO) {
3050 if (swap4(*ptr) != MAGICNO) {
3051 file_error(ms, 0, "bad magic in `%s'", dbname);
3052 goto error;
3053 }
3054 needsbyteswap = 1;
3055 } else
3056 needsbyteswap = 0;
3057 if (needsbyteswap)
3058 version = swap4(ptr[1]);
3059 else
3060 version = ptr[1];
3061 if (version != VERSIONNO) {
3062 file_error(ms, 0, "File %d.%d supports only version %d magic "
3063 "files. `%s' is version %d", FILE_VERSION_MAJOR, patchlevel,
3064 VERSIONNO, dbname, version);
3065 goto error;
3066 }
3067
3068 /* php_magic_database is a const, performing writes will segfault. This is for big-endian
3069 machines only, PPC and Sparc specifically. Consider static variable or MINIT in
3070 future. */
3071 if (needsbyteswap && fn == NULL) {
3072 map->p = emalloc(sizeof(php_magic_database));
3073 map->p = memcpy(map->p, php_magic_database, sizeof(php_magic_database));
3074 }
3075
3076 if (NULL != fn) {
3077 nentries = (uint32_t)(st.sb.st_size / sizeof(struct magic));
3078 entries = (uint32_t)(st.sb.st_size / sizeof(struct magic));
3079 if ((zend_off_t)(entries * sizeof(struct magic)) != st.sb.st_size) {
3080 file_error(ms, 0, "Size of `%s' %llu is not a multiple of %zu",
3081 dbname, (unsigned long long)st.sb.st_size,
3082 sizeof(struct magic));
3083 goto error;
3084 }
3085 }
3086 map->magic[0] = CAST(struct magic *, map->p) + 1;
3087 nentries = 0;
3088 for (i = 0; i < MAGIC_SETS; i++) {
3089 if (needsbyteswap)
3090 map->nmagic[i] = swap4(ptr[i + 2]);
3091 else
3092 map->nmagic[i] = ptr[i + 2];
3093 if (i != MAGIC_SETS - 1)
3094 map->magic[i + 1] = map->magic[i] + map->nmagic[i];
3095 nentries += map->nmagic[i];
3096 }
3097 if (NULL != fn && entries != nentries + 1) {
3098 file_error(ms, 0, "Inconsistent entries in `%s' %u != %u",
3099 dbname, entries, nentries + 1);
3100 goto error;
3101 }
3102 if (needsbyteswap)
3103 for (i = 0; i < MAGIC_SETS; i++)
3104 byteswap(map->magic[i], map->nmagic[i]);
3105
3106 if (dbname) {
3107 efree(dbname);
3108 }
3109 return map;
3110
3111 error:
3112 if (stream) {
3113 php_stream_close(stream);
3114 }
3115 apprentice_unmap(map);
3116 if (dbname) {
3117 efree(dbname);
3118 }
3119 return NULL;
3120 }
3121
3122 #if 0
3123 private int
3124 check_buffer(struct magic_set *ms, struct magic_map *map, const char *dbname)
3125 {
3126 uint32_t *ptr;
3127 uint32_t entries, nentries;
3128 uint32_t version;
3129 int i, needsbyteswap;
3130
3131 ptr = CAST(uint32_t *, map->p);
3132 if (*ptr != MAGICNO) {
3133 if (swap4(*ptr) != MAGICNO) {
3134 file_error(ms, 0, "bad magic in `%s'", dbname);
3135 return -1;
3136 }
3137 needsbyteswap = 1;
3138 } else
3139 needsbyteswap = 0;
3140 if (needsbyteswap)
3141 version = swap4(ptr[1]);
3142 else
3143 version = ptr[1];
3144 if (version != VERSIONNO) {
3145 file_error(ms, 0, "File %s supports only version %d magic "
3146 "files. `%s' is version %d", FILE_VERSION_MAJOR,
3147 VERSIONNO, dbname, version);
3148 return -1;
3149 }
3150 entries = CAST(uint32_t, map->len / sizeof(struct magic));
3151 if ((entries * sizeof(struct magic)) != map->len) {
3152 file_error(ms, 0, "Size of `%s' %" SIZE_T_FORMAT "u is not "
3153 "a multiple of %" SIZE_T_FORMAT "u",
3154 dbname, map->len, sizeof(struct magic));
3155 return -1;
3156 }
3157 map->magic[0] = CAST(struct magic *, map->p) + 1;
3158 nentries = 0;
3159 for (i = 0; i < MAGIC_SETS; i++) {
3160 if (needsbyteswap)
3161 map->nmagic[i] = swap4(ptr[i + 2]);
3162 else
3163 map->nmagic[i] = ptr[i + 2];
3164 if (i != MAGIC_SETS - 1)
3165 map->magic[i + 1] = map->magic[i] + map->nmagic[i];
3166 nentries += map->nmagic[i];
3167 }
3168 if (entries != nentries + 1) {
3169 file_error(ms, 0, "Inconsistent entries in `%s' %u != %u",
3170 dbname, entries, nentries + 1);
3171 return -1;
3172 }
3173 if (needsbyteswap)
3174 for (i = 0; i < MAGIC_SETS; i++)
3175 byteswap(map->magic[i], map->nmagic[i]);
3176 return 0;
3177 }
3178 #endif
3179
3180 /*
3181 * handle an mmaped file.
3182 */
3183 private int
apprentice_compile(struct magic_set * ms,struct magic_map * map,const char * fn)3184 apprentice_compile(struct magic_set *ms, struct magic_map *map, const char *fn)
3185 {
3186 static const size_t nm = sizeof(*map->nmagic) * MAGIC_SETS;
3187 static const size_t m = sizeof(**map->magic);
3188 size_t len;
3189 char *dbname;
3190 int rv = -1;
3191 uint32_t i;
3192 union {
3193 struct magic m;
3194 uint32_t h[2 + MAGIC_SETS];
3195 } hdr;
3196 php_stream *stream;
3197
3198 dbname = mkdbname(ms, fn, 1);
3199
3200 if (dbname == NULL)
3201 goto out;
3202
3203 /* wb+ == O_WRONLY|O_CREAT|O_TRUNC|O_BINARY */
3204 stream = php_stream_open_wrapper((char *)fn, "wb+", REPORT_ERRORS, NULL);
3205
3206 if (!stream) {
3207 file_error(ms, errno, "cannot open `%s'", dbname);
3208 goto out;
3209 }
3210 memset(&hdr, 0, sizeof(hdr));
3211 hdr.h[0] = MAGICNO;
3212 hdr.h[1] = VERSIONNO;
3213 memcpy(hdr.h + 2, map->nmagic, nm);
3214
3215 if (php_stream_write(stream,(const char *)&hdr, sizeof(hdr)) != (ssize_t)sizeof(hdr)) {
3216 file_error(ms, errno, "error writing `%s'", dbname);
3217 goto out;
3218 }
3219
3220 for (i = 0; i < MAGIC_SETS; i++) {
3221 len = m * map->nmagic[i];
3222 if (php_stream_write(stream, (const char *)map->magic[i], len) != (ssize_t)len) {
3223 file_error(ms, errno, "error writing `%s'", dbname);
3224 goto out;
3225 }
3226 }
3227
3228 if (stream) {
3229 php_stream_close(stream);
3230 }
3231 rv = 0;
3232 out:
3233 efree(dbname);
3234 return rv;
3235 }
3236
3237 private const char ext[] = ".mgc";
3238 /*
3239 * make a dbname
3240 */
3241 private char *
mkdbname(struct magic_set * ms,const char * fn,int strip)3242 mkdbname(struct magic_set *ms, const char *fn, int strip)
3243 {
3244 const char *p, *q;
3245 char *buf;
3246
3247 if (strip) {
3248 if ((p = strrchr(fn, '/')) != NULL)
3249 fn = ++p;
3250 }
3251
3252 for (q = fn; *q; q++)
3253 continue;
3254 /* Look for .mgc */
3255 for (p = ext + sizeof(ext) - 1; p >= ext && q >= fn; p--, q--)
3256 if (*p != *q)
3257 break;
3258
3259 /* Did not find .mgc, restore q */
3260 if (p >= ext)
3261 while (*q)
3262 q++;
3263
3264 q++;
3265 /* Compatibility with old code that looked in .mime */
3266 if (ms->flags & MAGIC_MIME) {
3267 spprintf(&buf, MAXPATHLEN, "%.*s.mime%s", CAST(int, q - fn), fn, ext);
3268 #ifdef PHP_WIN32
3269 if (VCWD_ACCESS(buf, R_OK) == 0) {
3270 #else
3271 if (VCWD_ACCESS(buf, R_OK) != -1) {
3272 #endif
3273 ms->flags &= MAGIC_MIME_TYPE;
3274 return buf;
3275 }
3276 efree(buf);
3277 }
3278 spprintf(&buf, MAXPATHLEN, "%.*s%s", CAST(int, q - fn), fn, ext);
3279
3280 /* Compatibility with old code that looked in .mime */
3281 if (strstr(fn, ".mime") != NULL)
3282 ms->flags &= MAGIC_MIME_TYPE;
3283 return buf;
3284 }
3285
3286 /*
3287 * Byteswap an mmap'ed file if needed
3288 */
3289 private void
3290 byteswap(struct magic *magic, uint32_t nmagic)
3291 {
3292 uint32_t i;
3293 for (i = 0; i < nmagic; i++)
3294 bs1(&magic[i]);
3295 }
3296
3297 /*
3298 * swap a short
3299 */
3300 private uint16_t
3301 swap2(uint16_t sv)
3302 {
3303 uint16_t rv;
3304 uint8_t *s = RCAST(uint8_t *, RCAST(void *, &sv));
3305 uint8_t *d = RCAST(uint8_t *, RCAST(void *, &rv));
3306 d[0] = s[1];
3307 d[1] = s[0];
3308 return rv;
3309 }
3310
3311 /*
3312 * swap an int
3313 */
3314 private uint32_t
3315 swap4(uint32_t sv)
3316 {
3317 uint32_t rv;
3318 uint8_t *s = RCAST(uint8_t *, RCAST(void *, &sv));
3319 uint8_t *d = RCAST(uint8_t *, RCAST(void *, &rv));
3320 d[0] = s[3];
3321 d[1] = s[2];
3322 d[2] = s[1];
3323 d[3] = s[0];
3324 return rv;
3325 }
3326
3327 /*
3328 * swap a quad
3329 */
3330 private uint64_t
3331 swap8(uint64_t sv)
3332 {
3333 uint64_t rv;
3334 uint8_t *s = RCAST(uint8_t *, RCAST(void *, &sv));
3335 uint8_t *d = RCAST(uint8_t *, RCAST(void *, &rv));
3336 #if 0
3337 d[0] = s[3];
3338 d[1] = s[2];
3339 d[2] = s[1];
3340 d[3] = s[0];
3341 d[4] = s[7];
3342 d[5] = s[6];
3343 d[6] = s[5];
3344 d[7] = s[4];
3345 #else
3346 d[0] = s[7];
3347 d[1] = s[6];
3348 d[2] = s[5];
3349 d[3] = s[4];
3350 d[4] = s[3];
3351 d[5] = s[2];
3352 d[6] = s[1];
3353 d[7] = s[0];
3354 #endif
3355 return rv;
3356 }
3357
3358 /*
3359 * byteswap a single magic entry
3360 */
3361 private void
3362 bs1(struct magic *m)
3363 {
3364 m->cont_level = swap2(m->cont_level);
3365 m->offset = swap4(CAST(uint32_t, m->offset));
3366 m->in_offset = swap4(CAST(uint32_t, m->in_offset));
3367 m->lineno = swap4(CAST(uint32_t, m->lineno));
3368 if (IS_LIBMAGIC_STRING(m->type)) {
3369 m->str_range = swap4(m->str_range);
3370 m->str_flags = swap4(m->str_flags);
3371 }
3372 else {
3373 m->value.q = swap8(m->value.q);
3374 m->num_mask = swap8(m->num_mask);
3375 }
3376 }
3377
3378 protected size_t
3379 file_pstring_length_size(const struct magic *m)
3380 {
3381 switch (m->str_flags & PSTRING_LEN) {
3382 case PSTRING_1_LE:
3383 return 1;
3384 case PSTRING_2_LE:
3385 case PSTRING_2_BE:
3386 return 2;
3387 case PSTRING_4_LE:
3388 case PSTRING_4_BE:
3389 return 4;
3390 default:
3391 abort(); /* Impossible */
3392 return 1;
3393 }
3394 }
3395 protected size_t
3396 file_pstring_get_length(const struct magic *m, const char *ss)
3397 {
3398 size_t len = 0;
3399 const unsigned char *s = RCAST(const unsigned char *, ss);
3400 unsigned int s3, s2, s1, s0;
3401
3402 switch (m->str_flags & PSTRING_LEN) {
3403 case PSTRING_1_LE:
3404 len = *s;
3405 break;
3406 case PSTRING_2_LE:
3407 s0 = s[0];
3408 s1 = s[1];
3409 len = (s1 << 8) | s0;
3410 break;
3411 case PSTRING_2_BE:
3412 s0 = s[0];
3413 s1 = s[1];
3414 len = (s0 << 8) | s1;
3415 break;
3416 case PSTRING_4_LE:
3417 s0 = s[0];
3418 s1 = s[1];
3419 s2 = s[2];
3420 s3 = s[3];
3421 len = (s3 << 24) | (s2 << 16) | (s1 << 8) | s0;
3422 break;
3423 case PSTRING_4_BE:
3424 s0 = s[0];
3425 s1 = s[1];
3426 s2 = s[2];
3427 s3 = s[3];
3428 len = (s0 << 24) | (s1 << 16) | (s2 << 8) | s3;
3429 break;
3430 default:
3431 abort(); /* Impossible */
3432 }
3433
3434 if (m->str_flags & PSTRING_LENGTH_INCLUDES_ITSELF)
3435 len -= file_pstring_length_size(m);
3436
3437 return len;
3438 }
3439
3440 protected int
3441 file_magicfind(struct magic_set *ms, const char *name, struct mlist *v)
3442 {
3443 uint32_t i, j;
3444 struct mlist *mlist, *ml;
3445
3446 mlist = ms->mlist[1];
3447
3448 for (ml = mlist->next; ml != mlist; ml = ml->next) {
3449 struct magic *ma = ml->magic;
3450 uint32_t nma = ml->nmagic;
3451 for (i = 0; i < nma; i++) {
3452 if (ma[i].type != FILE_NAME)
3453 continue;
3454 if (strcmp(ma[i].value.s, name) == 0) {
3455 v->magic = &ma[i];
3456 for (j = i + 1; j < nma; j++)
3457 if (ma[j].cont_level == 0)
3458 break;
3459 v->nmagic = j - i;
3460 return 0;
3461 }
3462 }
3463 }
3464 return -1;
3465 }
3466
