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 1.173 2011/12/08 12:38:24 rrt 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 #if _MSC_VER <= 1300
55 # include "win32/php_strtoi64.h"
56 #endif
57 #define strtoull _strtoui64
58 #else
59 #include <unistd.h>
60 #endif
61
62 #include <string.h>
63 #include <assert.h>
64 #include <ctype.h>
65 #include <fcntl.h>
66
67 #define EATAB {while (isascii((unsigned char) *l) && \
68 isspace((unsigned char) *l)) ++l;}
69 #define LOWCASE(l) (isupper((unsigned char) (l)) ? \
70 tolower((unsigned char) (l)) : (l))
71 /*
72 * Work around a bug in headers on Digital Unix.
73 * At least confirmed for: OSF1 V4.0 878
74 */
75 #if defined(__osf__) && defined(__DECC)
76 #ifdef MAP_FAILED
77 #undef MAP_FAILED
78 #endif
79 #endif
80
81 #ifndef MAP_FAILED
82 #define MAP_FAILED (void *) -1
83 #endif
84
85 #ifndef MAP_FILE
86 #define MAP_FILE 0
87 #endif
88
89 struct magic_entry {
90 struct magic *mp;
91 uint32_t cont_count;
92 uint32_t max_count;
93 };
94
95 int file_formats[FILE_NAMES_SIZE];
96 const size_t file_nformats = FILE_NAMES_SIZE;
97 const char *file_names[FILE_NAMES_SIZE];
98 const size_t file_nnames = FILE_NAMES_SIZE;
99
100 private int getvalue(struct magic_set *ms, struct magic *, const char **, int);
101 private int hextoint(int);
102 private const char *getstr(struct magic_set *, struct magic *, const char *,
103 int);
104 private int parse(struct magic_set *, struct magic_entry **, uint32_t *,
105 const char *, size_t, int);
106 private void eatsize(const char **);
107 private int apprentice_1(struct magic_set *, const char *, int, struct mlist *);
108 private size_t apprentice_magic_strength(const struct magic *);
109 private int apprentice_sort(const void *, const void *);
110 private void apprentice_list(struct mlist *, int );
111 private int apprentice_load(struct magic_set *, struct magic **, uint32_t *,
112 const char *, int);
113 private void byteswap(struct magic *, uint32_t);
114 private void bs1(struct magic *);
115 private uint16_t swap2(uint16_t);
116 private uint32_t swap4(uint32_t);
117 private uint64_t swap8(uint64_t);
118 private char *mkdbname(struct magic_set *, const char *, int);
119 private int apprentice_map(struct magic_set *, struct magic **, uint32_t *,
120 const char *);
121 private int apprentice_compile(struct magic_set *, struct magic **, uint32_t *,
122 const char *);
123 private int check_format_type(const char *, int);
124 private int check_format(struct magic_set *, struct magic *);
125 private int get_op(char);
126 private int parse_mime(struct magic_set *, struct magic_entry *, const char *);
127 private int parse_strength(struct magic_set *, struct magic_entry *, const char *);
128 private int parse_apple(struct magic_set *, struct magic_entry *, const char *);
129
130 private size_t maxmagic = 0;
131 private size_t magicsize = sizeof(struct magic);
132
133 private const char usg_hdr[] = "cont\toffset\ttype\topcode\tmask\tvalue\tdesc";
134 private struct {
135 const char *name;
136 size_t len;
137 int (*fun)(struct magic_set *, struct magic_entry *, const char *);
138 } bang[] = {
139 #define DECLARE_FIELD(name) { # name, sizeof(# name) - 1, parse_ ## name }
140 DECLARE_FIELD(mime),
141 DECLARE_FIELD(apple),
142 DECLARE_FIELD(strength),
143 #undef DECLARE_FIELD
144 { NULL, 0, NULL }
145 };
146
147 #include "../data_file.c"
148
149 static const struct type_tbl_s {
150 const char name[16];
151 const size_t len;
152 const int type;
153 const int format;
154 } type_tbl[] = {
155 # define XX(s) s, (sizeof(s) - 1)
156 # define XX_NULL "", 0
157 { XX("byte"), FILE_BYTE, FILE_FMT_NUM },
158 { XX("short"), FILE_SHORT, FILE_FMT_NUM },
159 { XX("default"), FILE_DEFAULT, FILE_FMT_STR },
160 { XX("long"), FILE_LONG, FILE_FMT_NUM },
161 { XX("string"), FILE_STRING, FILE_FMT_STR },
162 { XX("date"), FILE_DATE, FILE_FMT_STR },
163 { XX("beshort"), FILE_BESHORT, FILE_FMT_NUM },
164 { XX("belong"), FILE_BELONG, FILE_FMT_NUM },
165 { XX("bedate"), FILE_BEDATE, FILE_FMT_STR },
166 { XX("leshort"), FILE_LESHORT, FILE_FMT_NUM },
167 { XX("lelong"), FILE_LELONG, FILE_FMT_NUM },
168 { XX("ledate"), FILE_LEDATE, FILE_FMT_STR },
169 { XX("pstring"), FILE_PSTRING, FILE_FMT_STR },
170 { XX("ldate"), FILE_LDATE, FILE_FMT_STR },
171 { XX("beldate"), FILE_BELDATE, FILE_FMT_STR },
172 { XX("leldate"), FILE_LELDATE, FILE_FMT_STR },
173 { XX("regex"), FILE_REGEX, FILE_FMT_STR },
174 { XX("bestring16"), FILE_BESTRING16, FILE_FMT_STR },
175 { XX("lestring16"), FILE_LESTRING16, FILE_FMT_STR },
176 { XX("search"), FILE_SEARCH, FILE_FMT_STR },
177 { XX("medate"), FILE_MEDATE, FILE_FMT_STR },
178 { XX("meldate"), FILE_MELDATE, FILE_FMT_STR },
179 { XX("melong"), FILE_MELONG, FILE_FMT_NUM },
180 { XX("quad"), FILE_QUAD, FILE_FMT_QUAD },
181 { XX("lequad"), FILE_LEQUAD, FILE_FMT_QUAD },
182 { XX("bequad"), FILE_BEQUAD, FILE_FMT_QUAD },
183 { XX("qdate"), FILE_QDATE, FILE_FMT_STR },
184 { XX("leqdate"), FILE_LEQDATE, FILE_FMT_STR },
185 { XX("beqdate"), FILE_BEQDATE, FILE_FMT_STR },
186 { XX("qldate"), FILE_QLDATE, FILE_FMT_STR },
187 { XX("leqldate"), FILE_LEQLDATE, FILE_FMT_STR },
188 { XX("beqldate"), FILE_BEQLDATE, FILE_FMT_STR },
189 { XX("float"), FILE_FLOAT, FILE_FMT_FLOAT },
190 { XX("befloat"), FILE_BEFLOAT, FILE_FMT_FLOAT },
191 { XX("lefloat"), FILE_LEFLOAT, FILE_FMT_FLOAT },
192 { XX("double"), FILE_DOUBLE, FILE_FMT_DOUBLE },
193 { XX("bedouble"), FILE_BEDOUBLE, FILE_FMT_DOUBLE },
194 { XX("ledouble"), FILE_LEDOUBLE, FILE_FMT_DOUBLE },
195 { XX("leid3"), FILE_LEID3, FILE_FMT_NUM },
196 { XX("beid3"), FILE_BEID3, FILE_FMT_NUM },
197 { XX("indirect"), FILE_INDIRECT, FILE_FMT_NONE },
198 { XX_NULL, FILE_INVALID, FILE_FMT_NONE },
199 # undef XX
200 # undef XX_NULL
201 };
202
203 #ifndef S_ISDIR
204 #define S_ISDIR(mode) ((mode) & _S_IFDIR)
205 #endif
206
207 private int
get_type(const char * l,const char ** t)208 get_type(const char *l, const char **t)
209 {
210 const struct type_tbl_s *p;
211
212 for (p = type_tbl; p->len; p++) {
213 if (strncmp(l, p->name, p->len) == 0) {
214 if (t)
215 *t = l + p->len;
216 break;
217 }
218 }
219 return p->type;
220 }
221
222 private void
init_file_tables(void)223 init_file_tables(void)
224 {
225 static int done = 0;
226 const struct type_tbl_s *p;
227
228 if (done)
229 return;
230 done++;
231
232 for (p = type_tbl; p->len; p++) {
233 assert(p->type < FILE_NAMES_SIZE);
234 file_names[p->type] = p->name;
235 file_formats[p->type] = p->format;
236 }
237 }
238
239 /*
240 * Handle one file or directory.
241 */
242 private int
apprentice_1(struct magic_set * ms,const char * fn,int action,struct mlist * mlist)243 apprentice_1(struct magic_set *ms, const char *fn, int action,
244 struct mlist *mlist)
245 {
246 struct magic *magic = NULL;
247 uint32_t nmagic = 0;
248 struct mlist *ml;
249 int rv = -1;
250 int mapped;
251
252 if (magicsize != FILE_MAGICSIZE) {
253 file_error(ms, 0, "magic element size %lu != %lu",
254 (unsigned long)sizeof(*magic),
255 (unsigned long)FILE_MAGICSIZE);
256 return -1;
257 }
258
259 if (action == FILE_COMPILE) {
260 rv = apprentice_load(ms, &magic, &nmagic, fn, action);
261 if (rv != 0)
262 return -1;
263 rv = apprentice_compile(ms, &magic, &nmagic, fn);
264 efree(magic);
265 return rv;
266 }
267
268 if ((rv = apprentice_map(ms, &magic, &nmagic, fn)) == -1) {
269 if (fn) {
270 if (ms->flags & MAGIC_CHECK)
271 file_magwarn(ms, "using regular magic file `%s'", fn);
272 rv = apprentice_load(ms, &magic, &nmagic, fn, action);
273 }
274
275 if (rv != 0)
276 return -1;
277 }
278
279 mapped = rv;
280
281 if (magic == NULL) {
282 file_delmagic(magic, mapped, nmagic);
283 return -1;
284 }
285
286 ml = emalloc(sizeof(*ml));
287
288 ml->magic = magic;
289 ml->nmagic = nmagic;
290 ml->mapped = mapped;
291
292 mlist->prev->next = ml;
293 ml->prev = mlist->prev;
294 ml->next = mlist;
295 mlist->prev = ml;
296
297 if (action == FILE_LIST) {
298 printf("Binary patterns:\n");
299 apprentice_list(mlist, BINTEST);
300 printf("Text patterns:\n");
301 apprentice_list(mlist, TEXTTEST);
302 }
303
304 return 0;
305 }
306
307 protected void
file_delmagic(struct magic * p,int type,size_t entries)308 file_delmagic(struct magic *p, int type, size_t entries)
309 {
310 if (p == NULL)
311 return;
312 switch (type) {
313 case 3:
314 /* Do nothing, it's part of the code segment */
315 break;
316
317 case 1:
318 p--;
319 /*FALLTHROUGH*/
320
321 case 0:
322 efree(p);
323 break;
324
325 default:
326 abort();
327 }
328 }
329
330 /* const char *fn: list of magic files and directories */
331 protected struct mlist *
file_apprentice(struct magic_set * ms,const char * fn,int action)332 file_apprentice(struct magic_set *ms, const char *fn, int action)
333 {
334 char *p, *mfn;
335 int file_err, errs = -1;
336 struct mlist *mlist;
337 /* XXX disabling default magic loading so the compiled in data is used */
338 #if 0
339 if ((fn = magic_getpath(fn, action)) == NULL)
340 return NULL;
341 #endif
342
343 init_file_tables();
344
345 if (fn == NULL)
346 fn = getenv("MAGIC");
347 if (fn == NULL) {
348 mlist = emalloc(sizeof(*mlist));
349 mlist->next = mlist->prev = mlist;
350 apprentice_1(ms, fn, action, mlist);
351 return mlist;
352 }
353
354 mfn = estrdup(fn);
355 fn = mfn;
356
357 mlist = emalloc(sizeof(*mlist));
358 mlist->next = mlist->prev = mlist;
359
360 while (fn) {
361 p = strchr(fn, PATHSEP);
362 if (p)
363 *p++ = '\0';
364 if (*fn == '\0')
365 break;
366 file_err = apprentice_1(ms, fn, action, mlist);
367 errs = MAX(errs, file_err);
368 fn = p;
369 }
370 if (errs == -1) {
371 efree(mfn);
372 efree(mlist);
373 mlist = NULL;
374 file_error(ms, 0, "could not find any magic files!");
375 return NULL;
376 }
377 efree(mfn);
378 return mlist;
379 }
380
381 /*
382 * Get weight of this magic entry, for sorting purposes.
383 */
384 private size_t
apprentice_magic_strength(const struct magic * m)385 apprentice_magic_strength(const struct magic *m)
386 {
387 #define MULT 10
388 size_t val = 2 * MULT; /* baseline strength */
389
390 switch (m->type) {
391 case FILE_DEFAULT: /* make sure this sorts last */
392 if (m->factor_op != FILE_FACTOR_OP_NONE)
393 abort();
394 return 0;
395
396 case FILE_BYTE:
397 val += 1 * MULT;
398 break;
399
400 case FILE_SHORT:
401 case FILE_LESHORT:
402 case FILE_BESHORT:
403 val += 2 * MULT;
404 break;
405
406 case FILE_LONG:
407 case FILE_LELONG:
408 case FILE_BELONG:
409 case FILE_MELONG:
410 val += 4 * MULT;
411 break;
412
413 case FILE_PSTRING:
414 case FILE_STRING:
415 val += m->vallen * MULT;
416 break;
417
418 case FILE_BESTRING16:
419 case FILE_LESTRING16:
420 val += m->vallen * MULT / 2;
421 break;
422
423 case FILE_SEARCH:
424 case FILE_REGEX:
425 val += m->vallen * MAX(MULT / m->vallen, 1);
426 break;
427
428 case FILE_DATE:
429 case FILE_LEDATE:
430 case FILE_BEDATE:
431 case FILE_MEDATE:
432 case FILE_LDATE:
433 case FILE_LELDATE:
434 case FILE_BELDATE:
435 case FILE_MELDATE:
436 case FILE_FLOAT:
437 case FILE_BEFLOAT:
438 case FILE_LEFLOAT:
439 val += 4 * MULT;
440 break;
441
442 case FILE_QUAD:
443 case FILE_BEQUAD:
444 case FILE_LEQUAD:
445 case FILE_QDATE:
446 case FILE_LEQDATE:
447 case FILE_BEQDATE:
448 case FILE_QLDATE:
449 case FILE_LEQLDATE:
450 case FILE_BEQLDATE:
451 case FILE_DOUBLE:
452 case FILE_BEDOUBLE:
453 case FILE_LEDOUBLE:
454 val += 8 * MULT;
455 break;
456
457 default:
458 val = 0;
459 (void)fprintf(stderr, "Bad type %d\n", m->type);
460 abort();
461 }
462
463 switch (m->reln) {
464 case 'x': /* matches anything penalize */
465 case '!': /* matches almost anything penalize */
466 val = 0;
467 break;
468
469 case '=': /* Exact match, prefer */
470 val += MULT;
471 break;
472
473 case '>':
474 case '<': /* comparison match reduce strength */
475 val -= 2 * MULT;
476 break;
477
478 case '^':
479 case '&': /* masking bits, we could count them too */
480 val -= MULT;
481 break;
482
483 default:
484 (void)fprintf(stderr, "Bad relation %c\n", m->reln);
485 abort();
486 }
487
488 if (val == 0) /* ensure we only return 0 for FILE_DEFAULT */
489 val = 1;
490
491 switch (m->factor_op) {
492 case FILE_FACTOR_OP_NONE:
493 break;
494 case FILE_FACTOR_OP_PLUS:
495 val += m->factor;
496 break;
497 case FILE_FACTOR_OP_MINUS:
498 val -= m->factor;
499 break;
500 case FILE_FACTOR_OP_TIMES:
501 val *= m->factor;
502 break;
503 case FILE_FACTOR_OP_DIV:
504 val /= m->factor;
505 break;
506 default:
507 abort();
508 }
509
510
511 /*
512 * Magic entries with no description get a bonus because they depend
513 * on subsequent magic entries to print something.
514 */
515 if (m->desc[0] == '\0')
516 val++;
517 return val;
518 }
519
520 /*
521 * Sort callback for sorting entries by "strength" (basically length)
522 */
523 private int
apprentice_sort(const void * a,const void * b)524 apprentice_sort(const void *a, const void *b)
525 {
526 const struct magic_entry *ma = a;
527 const struct magic_entry *mb = b;
528 size_t sa = apprentice_magic_strength(ma->mp);
529 size_t sb = apprentice_magic_strength(mb->mp);
530 if (sa == sb)
531 return 0;
532 else if (sa > sb)
533 return -1;
534 else
535 return 1;
536 }
537
538 /*
539 * Shows sorted patterns list in the order which is used for the matching
540 */
541 private void
apprentice_list(struct mlist * mlist,int mode)542 apprentice_list(struct mlist *mlist, int mode)
543 {
544 uint32_t magindex = 0;
545 struct mlist *ml;
546 for (ml = mlist->next; ml != mlist; ml = ml->next) {
547 for (magindex = 0; magindex < ml->nmagic; magindex++) {
548 struct magic *m = &ml->magic[magindex];
549 if ((m->flag & mode) != mode) {
550 /* Skip sub-tests */
551 while (magindex + 1 < ml->nmagic &&
552 ml->magic[magindex + 1].cont_level != 0)
553 ++magindex;
554 continue; /* Skip to next top-level test*/
555 }
556
557 /*
558 * Try to iterate over the tree until we find item with
559 * description/mimetype.
560 */
561 while (magindex + 1 < ml->nmagic &&
562 ml->magic[magindex + 1].cont_level != 0 &&
563 *ml->magic[magindex].desc == '\0' &&
564 *ml->magic[magindex].mimetype == '\0')
565 magindex++;
566
567 printf("Strength = %3" SIZE_T_FORMAT "u : %s [%s]\n",
568 apprentice_magic_strength(m),
569 ml->magic[magindex].desc,
570 ml->magic[magindex].mimetype);
571 }
572 }
573 }
574
575 private void
set_test_type(struct magic * mstart,struct magic * m)576 set_test_type(struct magic *mstart, struct magic *m)
577 {
578 switch (m->type) {
579 case FILE_BYTE:
580 case FILE_SHORT:
581 case FILE_LONG:
582 case FILE_DATE:
583 case FILE_BESHORT:
584 case FILE_BELONG:
585 case FILE_BEDATE:
586 case FILE_LESHORT:
587 case FILE_LELONG:
588 case FILE_LEDATE:
589 case FILE_LDATE:
590 case FILE_BELDATE:
591 case FILE_LELDATE:
592 case FILE_MEDATE:
593 case FILE_MELDATE:
594 case FILE_MELONG:
595 case FILE_QUAD:
596 case FILE_LEQUAD:
597 case FILE_BEQUAD:
598 case FILE_QDATE:
599 case FILE_LEQDATE:
600 case FILE_BEQDATE:
601 case FILE_QLDATE:
602 case FILE_LEQLDATE:
603 case FILE_BEQLDATE:
604 case FILE_FLOAT:
605 case FILE_BEFLOAT:
606 case FILE_LEFLOAT:
607 case FILE_DOUBLE:
608 case FILE_BEDOUBLE:
609 case FILE_LEDOUBLE:
610 mstart->flag |= BINTEST;
611 break;
612 case FILE_STRING:
613 case FILE_PSTRING:
614 case FILE_BESTRING16:
615 case FILE_LESTRING16:
616 /* Allow text overrides */
617 if (mstart->str_flags & STRING_TEXTTEST)
618 mstart->flag |= TEXTTEST;
619 else
620 mstart->flag |= BINTEST;
621 break;
622 case FILE_REGEX:
623 case FILE_SEARCH:
624 /* Check for override */
625 if (mstart->str_flags & STRING_BINTEST)
626 mstart->flag |= BINTEST;
627 if (mstart->str_flags & STRING_TEXTTEST)
628 mstart->flag |= TEXTTEST;
629
630 if (mstart->flag & (TEXTTEST|BINTEST))
631 break;
632
633 /* binary test if pattern is not text */
634 if (file_looks_utf8(m->value.us, (size_t)m->vallen, NULL,
635 NULL) <= 0)
636 mstart->flag |= BINTEST;
637 else
638 mstart->flag |= TEXTTEST;
639 break;
640 case FILE_DEFAULT:
641 /* can't deduce anything; we shouldn't see this at the
642 top level anyway */
643 break;
644 case FILE_INVALID:
645 default:
646 /* invalid search type, but no need to complain here */
647 break;
648 }
649 }
650
651 /*
652 * Load and parse one file.
653 */
654 private void
load_1(struct magic_set * ms,int action,const char * fn,int * errs,struct magic_entry ** marray,uint32_t * marraycount)655 load_1(struct magic_set *ms, int action, const char *fn, int *errs,
656 struct magic_entry **marray, uint32_t *marraycount)
657 {
658 char buffer[BUFSIZ + 1];
659 char *line = NULL;
660 size_t len;
661 size_t lineno = 0;
662
663 php_stream *stream;
664
665 TSRMLS_FETCH();
666
667 #if PHP_API_VERSION < 20100412
668 stream = php_stream_open_wrapper((char *)fn, "rb", REPORT_ERRORS|ENFORCE_SAFE_MODE, NULL);
669 #else
670 stream = php_stream_open_wrapper((char *)fn, "rb", REPORT_ERRORS, NULL);
671 #endif
672
673 if (stream == NULL) {
674 if (errno != ENOENT)
675 file_error(ms, errno, "cannot read magic file `%s'",
676 fn);
677 (*errs)++;
678 return;
679 }
680
681 /* read and parse this file */
682 #if (PHP_MAJOR_VERSION < 6)
683 for (ms->line = 1; (line = php_stream_get_line(stream, buffer , BUFSIZ, &len)) != NULL; ms->line++) {
684 #else
685 for (ms->line = 1; (line = php_stream_get_line(stream, ZSTR(buffer), BUFSIZ, &len)) != NULL; ms->line++) {
686 #endif
687 if (len == 0) /* null line, garbage, etc */
688 continue;
689 if (line[len - 1] == '\n') {
690 lineno++;
691 line[len - 1] = '\0'; /* delete newline */
692 }
693 switch (line[0]) {
694 case '\0': /* empty, do not parse */
695 case '#': /* comment, do not parse */
696 continue;
697 case '!':
698 if (line[1] == ':') {
699 size_t i;
700
701 for (i = 0; bang[i].name != NULL; i++) {
702 if ((size_t)(len - 2) > bang[i].len &&
703 memcmp(bang[i].name, line + 2,
704 bang[i].len) == 0)
705 break;
706 }
707 if (bang[i].name == NULL) {
708 file_error(ms, 0,
709 "Unknown !: entry `%s'", line);
710 (*errs)++;
711 continue;
712 }
713 if (*marraycount == 0) {
714 file_error(ms, 0,
715 "No current entry for :!%s type",
716 bang[i].name);
717 (*errs)++;
718 continue;
719 }
720 if ((*bang[i].fun)(ms,
721 &(*marray)[*marraycount - 1],
722 line + bang[i].len + 2) != 0) {
723 (*errs)++;
724 continue;
725 }
726 continue;
727 }
728 /*FALLTHROUGH*/
729 default:
730 if (parse(ms, marray, marraycount, line, lineno,
731 action) != 0)
732 (*errs)++;
733 break;
734 }
735 }
736 php_stream_close(stream);
737 }
738
739 /*
740 * parse a file or directory of files
741 * const char *fn: name of magic file or directory
742 */
743 private int
744 cmpstrp(const void *p1, const void *p2)
745 {
746 return strcmp(*(char *const *)p1, *(char *const *)p2);
747 }
748
749 private int
750 apprentice_load(struct magic_set *ms, struct magic **magicp, uint32_t *nmagicp,
751 const char *fn, int action)
752 {
753 int errs = 0, mflen = 0;
754 struct magic_entry *marray;
755 uint32_t marraycount, i, mentrycount = 0, starttest;
756 size_t files = 0, maxfiles = 0;
757 char **filearr = NULL, mfn[MAXPATHLEN];
758 struct stat st;
759 php_stream *dir;
760 php_stream_dirent d;
761
762 TSRMLS_FETCH();
763
764 ms->flags |= MAGIC_CHECK; /* Enable checks for parsed files */
765
766 maxmagic = MAXMAGIS;
767 marray = ecalloc(maxmagic, sizeof(*marray));
768 marraycount = 0;
769
770 /* print silly verbose header for USG compat. */
771 if (action == FILE_CHECK)
772 (void)fprintf(stderr, "%s\n", usg_hdr);
773
774 /* load directory or file */
775 /* FIXME: Read file names and sort them to prevent
776 non-determinism. See Debian bug #488562. */
777 if (php_sys_stat(fn, &st) == 0 && S_ISDIR(st.st_mode)) {
778 int mflen;
779 char mfn[MAXPATHLEN];
780
781 dir = php_stream_opendir(fn, REPORT_ERRORS, NULL);
782 if (!dir) {
783 errs++;
784 goto out;
785 }
786 while (php_stream_readdir(dir, &d)) {
787 if ((mflen = snprintf(mfn, sizeof(mfn), "%s/%s", fn, d.d_name)) < 0) {
788 file_oomem(ms, strlen(fn) + strlen(d.d_name) + 2);
789 errs++;
790 php_stream_closedir(dir);
791 goto out;
792 }
793 if (stat(mfn, &st) == -1 || !S_ISREG(st.st_mode)) {
794 continue;
795 }
796 if (files >= maxfiles) {
797 size_t mlen;
798 maxfiles = (maxfiles + 1) * 2;
799 mlen = maxfiles * sizeof(*filearr);
800 if ((filearr = CAST(char **,
801 realloc(filearr, mlen))) == NULL) {
802 file_oomem(ms, mlen);
803 php_stream_closedir(dir);
804 errs++;
805 goto out;
806 }
807 }
808 filearr[files++] = estrndup(mfn, mflen);
809 }
810 php_stream_closedir(dir);
811 qsort(filearr, files, sizeof(*filearr), cmpstrp);
812 for (i = 0; i < files; i++) {
813 load_1(ms, action, filearr[i], &errs, &marray,
814 &marraycount);
815 efree(filearr[i]);
816 }
817 free(filearr);
818 } else
819 load_1(ms, action, fn, &errs, &marray, &marraycount);
820 if (errs)
821 goto out;
822
823 /* Set types of tests */
824 for (i = 0; i < marraycount; ) {
825 if (marray[i].mp->cont_level != 0) {
826 i++;
827 continue;
828 }
829
830 starttest = i;
831 do {
832 static const char text[] = "text";
833 static const char binary[] = "binary";
834 static const size_t len = sizeof(text);
835 set_test_type(marray[starttest].mp, marray[i].mp);
836 if ((ms->flags & MAGIC_DEBUG) == 0)
837 continue;
838 (void)fprintf(stderr, "%s%s%s: %s\n",
839 marray[i].mp->mimetype,
840 marray[i].mp->mimetype[0] == '\0' ? "" : "; ",
841 marray[i].mp->desc[0] ? marray[i].mp->desc :
842 "(no description)",
843 marray[i].mp->flag & BINTEST ? binary : text);
844 if (marray[i].mp->flag & BINTEST) {
845 char *p = strstr(marray[i].mp->desc, text);
846 if (p && (p == marray[i].mp->desc ||
847 isspace((unsigned char)p[-1])) &&
848 (p + len - marray[i].mp->desc ==
849 MAXstring || (p[len] == '\0' ||
850 isspace((unsigned char)p[len]))))
851 (void)fprintf(stderr, "*** Possible "
852 "binary test for text type\n");
853 }
854 } while (++i < marraycount && marray[i].mp->cont_level != 0);
855 }
856
857 qsort(marray, marraycount, sizeof(*marray), apprentice_sort);
858
859 /*
860 * Make sure that any level 0 "default" line is last (if one exists).
861 */
862 for (i = 0; i < marraycount; i++) {
863 if (marray[i].mp->cont_level == 0 &&
864 marray[i].mp->type == FILE_DEFAULT) {
865 while (++i < marraycount)
866 if (marray[i].mp->cont_level == 0)
867 break;
868 if (i != marraycount) {
869 /* XXX - Ugh! */
870 ms->line = marray[i].mp->lineno;
871 file_magwarn(ms,
872 "level 0 \"default\" did not sort last");
873 }
874 break;
875 }
876 }
877
878 for (i = 0; i < marraycount; i++)
879 mentrycount += marray[i].cont_count;
880
881 *magicp = emalloc(sizeof(**magicp) * mentrycount);
882
883 mentrycount = 0;
884 for (i = 0; i < marraycount; i++) {
885 (void)memcpy(*magicp + mentrycount, marray[i].mp,
886 marray[i].cont_count * sizeof(**magicp));
887 mentrycount += marray[i].cont_count;
888 }
889 out:
890 for (i = 0; i < marraycount; i++) {
891 if (marray[i].mp) {
892 efree(marray[i].mp);
893 }
894 }
895 if (marray) {
896 efree(marray);
897 }
898 if (errs) {
899 *magicp = NULL;
900 *nmagicp = 0;
901 return errs;
902 } else {
903 *nmagicp = mentrycount;
904 return 0;
905 }
906
907 }
908
909 /*
910 * extend the sign bit if the comparison is to be signed
911 */
912 protected uint64_t
913 file_signextend(struct magic_set *ms, struct magic *m, uint64_t v)
914 {
915 if (!(m->flag & UNSIGNED)) {
916 switch(m->type) {
917 /*
918 * Do not remove the casts below. They are
919 * vital. When later compared with the data,
920 * the sign extension must have happened.
921 */
922 case FILE_BYTE:
923 v = (char) v;
924 break;
925 case FILE_SHORT:
926 case FILE_BESHORT:
927 case FILE_LESHORT:
928 v = (short) v;
929 break;
930 case FILE_DATE:
931 case FILE_BEDATE:
932 case FILE_LEDATE:
933 case FILE_MEDATE:
934 case FILE_LDATE:
935 case FILE_BELDATE:
936 case FILE_LELDATE:
937 case FILE_MELDATE:
938 case FILE_LONG:
939 case FILE_BELONG:
940 case FILE_LELONG:
941 case FILE_MELONG:
942 case FILE_FLOAT:
943 case FILE_BEFLOAT:
944 case FILE_LEFLOAT:
945 v = (int32_t) v;
946 break;
947 case FILE_QUAD:
948 case FILE_BEQUAD:
949 case FILE_LEQUAD:
950 case FILE_QDATE:
951 case FILE_QLDATE:
952 case FILE_BEQDATE:
953 case FILE_BEQLDATE:
954 case FILE_LEQDATE:
955 case FILE_LEQLDATE:
956 case FILE_DOUBLE:
957 case FILE_BEDOUBLE:
958 case FILE_LEDOUBLE:
959 v = (int64_t) v;
960 break;
961 case FILE_STRING:
962 case FILE_PSTRING:
963 case FILE_BESTRING16:
964 case FILE_LESTRING16:
965 case FILE_REGEX:
966 case FILE_SEARCH:
967 case FILE_DEFAULT:
968 case FILE_INDIRECT:
969 break;
970 default:
971 if (ms->flags & MAGIC_CHECK)
972 file_magwarn(ms, "cannot happen: m->type=%d\n",
973 m->type);
974 return ~0U;
975 }
976 }
977 return v;
978 }
979
980 private int
981 string_modifier_check(struct magic_set *ms, struct magic *m)
982 {
983 if ((ms->flags & MAGIC_CHECK) == 0)
984 return 0;
985
986 if (m->type != FILE_PSTRING && (m->str_flags & PSTRING_LEN) != 0) {
987 file_magwarn(ms,
988 "'/BHhLl' modifiers are only allowed for pascal strings\n");
989 return -1;
990 }
991 switch (m->type) {
992 case FILE_BESTRING16:
993 case FILE_LESTRING16:
994 if (m->str_flags != 0) {
995 file_magwarn(ms,
996 "no modifiers allowed for 16-bit strings\n");
997 return -1;
998 }
999 break;
1000 case FILE_STRING:
1001 case FILE_PSTRING:
1002 if ((m->str_flags & REGEX_OFFSET_START) != 0) {
1003 file_magwarn(ms,
1004 "'/%c' only allowed on regex and search\n",
1005 CHAR_REGEX_OFFSET_START);
1006 return -1;
1007 }
1008 break;
1009 case FILE_SEARCH:
1010 if (m->str_range == 0) {
1011 file_magwarn(ms,
1012 "missing range; defaulting to %d\n",
1013 STRING_DEFAULT_RANGE);
1014 m->str_range = STRING_DEFAULT_RANGE;
1015 return -1;
1016 }
1017 break;
1018 case FILE_REGEX:
1019 if ((m->str_flags & STRING_COMPACT_WHITESPACE) != 0) {
1020 file_magwarn(ms, "'/%c' not allowed on regex\n",
1021 CHAR_COMPACT_WHITESPACE);
1022 return -1;
1023 }
1024 if ((m->str_flags & STRING_COMPACT_OPTIONAL_WHITESPACE) != 0) {
1025 file_magwarn(ms, "'/%c' not allowed on regex\n",
1026 CHAR_COMPACT_OPTIONAL_WHITESPACE);
1027 return -1;
1028 }
1029 break;
1030 default:
1031 file_magwarn(ms, "coding error: m->type=%d\n",
1032 m->type);
1033 return -1;
1034 }
1035 return 0;
1036 }
1037
1038 private int
1039 get_op(char c)
1040 {
1041 switch (c) {
1042 case '&':
1043 return FILE_OPAND;
1044 case '|':
1045 return FILE_OPOR;
1046 case '^':
1047 return FILE_OPXOR;
1048 case '+':
1049 return FILE_OPADD;
1050 case '-':
1051 return FILE_OPMINUS;
1052 case '*':
1053 return FILE_OPMULTIPLY;
1054 case '/':
1055 return FILE_OPDIVIDE;
1056 case '%':
1057 return FILE_OPMODULO;
1058 default:
1059 return -1;
1060 }
1061 }
1062
1063 #ifdef ENABLE_CONDITIONALS
1064 private int
1065 get_cond(const char *l, const char **t)
1066 {
1067 static const struct cond_tbl_s {
1068 char name[8];
1069 size_t len;
1070 int cond;
1071 } cond_tbl[] = {
1072 { "if", 2, COND_IF },
1073 { "elif", 4, COND_ELIF },
1074 { "else", 4, COND_ELSE },
1075 { "", 0, COND_NONE },
1076 };
1077 const struct cond_tbl_s *p;
1078
1079 for (p = cond_tbl; p->len; p++) {
1080 if (strncmp(l, p->name, p->len) == 0 &&
1081 isspace((unsigned char)l[p->len])) {
1082 if (t)
1083 *t = l + p->len;
1084 break;
1085 }
1086 }
1087 return p->cond;
1088 }
1089
1090 private int
1091 check_cond(struct magic_set *ms, int cond, uint32_t cont_level)
1092 {
1093 int last_cond;
1094 last_cond = ms->c.li[cont_level].last_cond;
1095
1096 switch (cond) {
1097 case COND_IF:
1098 if (last_cond != COND_NONE && last_cond != COND_ELIF) {
1099 if (ms->flags & MAGIC_CHECK)
1100 file_magwarn(ms, "syntax error: `if'");
1101 return -1;
1102 }
1103 last_cond = COND_IF;
1104 break;
1105
1106 case COND_ELIF:
1107 if (last_cond != COND_IF && last_cond != COND_ELIF) {
1108 if (ms->flags & MAGIC_CHECK)
1109 file_magwarn(ms, "syntax error: `elif'");
1110 return -1;
1111 }
1112 last_cond = COND_ELIF;
1113 break;
1114
1115 case COND_ELSE:
1116 if (last_cond != COND_IF && last_cond != COND_ELIF) {
1117 if (ms->flags & MAGIC_CHECK)
1118 file_magwarn(ms, "syntax error: `else'");
1119 return -1;
1120 }
1121 last_cond = COND_NONE;
1122 break;
1123
1124 case COND_NONE:
1125 last_cond = COND_NONE;
1126 break;
1127 }
1128
1129 ms->c.li[cont_level].last_cond = last_cond;
1130 return 0;
1131 }
1132 #endif /* ENABLE_CONDITIONALS */
1133
1134 /*
1135 * parse one line from magic file, put into magic[index++] if valid
1136 */
1137 private int
1138 parse(struct magic_set *ms, struct magic_entry **mentryp, uint32_t *nmentryp,
1139 const char *line, size_t lineno, int action)
1140 {
1141 #ifdef ENABLE_CONDITIONALS
1142 static uint32_t last_cont_level = 0;
1143 #endif
1144 size_t i;
1145 struct magic_entry *me;
1146 struct magic *m;
1147 const char *l = line;
1148 char *t;
1149 int op;
1150 uint32_t cont_level;
1151
1152 cont_level = 0;
1153
1154 while (*l == '>') {
1155 ++l; /* step over */
1156 cont_level++;
1157 }
1158 #ifdef ENABLE_CONDITIONALS
1159 if (cont_level == 0 || cont_level > last_cont_level)
1160 if (file_check_mem(ms, cont_level) == -1)
1161 return -1;
1162 last_cont_level = cont_level;
1163 #endif
1164
1165 #define ALLOC_CHUNK (size_t)10
1166 #define ALLOC_INCR (size_t)200
1167
1168 if (cont_level != 0) {
1169 if (*nmentryp == 0) {
1170 file_error(ms, 0, "No current entry for continuation");
1171 return -1;
1172 }
1173 me = &(*mentryp)[*nmentryp - 1];
1174 if (me->mp == NULL) {
1175 return -1;
1176 }
1177 if (me->cont_count == me->max_count) {
1178 struct magic *nm;
1179 size_t cnt = me->max_count + ALLOC_CHUNK;
1180 nm = erealloc(me->mp, sizeof(*nm) * cnt);
1181 me->mp = m = nm;
1182 me->max_count = CAST(uint32_t, cnt);
1183 }
1184 m = &me->mp[me->cont_count++];
1185 (void)memset(m, 0, sizeof(*m));
1186 m->cont_level = cont_level;
1187 } else {
1188 if (*nmentryp == maxmagic) {
1189 struct magic_entry *mp;
1190
1191 maxmagic += ALLOC_INCR;
1192 mp = erealloc(*mentryp, sizeof(*mp) * maxmagic);
1193 (void)memset(&mp[*nmentryp], 0, sizeof(*mp) * ALLOC_INCR);
1194 *mentryp = mp;
1195 }
1196 me = &(*mentryp)[*nmentryp];
1197 if (me->mp == NULL) {
1198 m = safe_emalloc(sizeof(*m), ALLOC_CHUNK, 0);
1199 me->mp = m;
1200 me->max_count = ALLOC_CHUNK;
1201 } else
1202 m = me->mp;
1203 (void)memset(m, 0, sizeof(*m));
1204 m->factor_op = FILE_FACTOR_OP_NONE;
1205 m->cont_level = 0;
1206 me->cont_count = 1;
1207 }
1208 m->lineno = CAST(uint32_t, lineno);
1209
1210 if (*l == '&') { /* m->cont_level == 0 checked below. */
1211 ++l; /* step over */
1212 m->flag |= OFFADD;
1213 }
1214 if (*l == '(') {
1215 ++l; /* step over */
1216 m->flag |= INDIR;
1217 if (m->flag & OFFADD)
1218 m->flag = (m->flag & ~OFFADD) | INDIROFFADD;
1219
1220 if (*l == '&') { /* m->cont_level == 0 checked below */
1221 ++l; /* step over */
1222 m->flag |= OFFADD;
1223 }
1224 }
1225 /* Indirect offsets are not valid at level 0. */
1226 if (m->cont_level == 0 && (m->flag & (OFFADD | INDIROFFADD)))
1227 if (ms->flags & MAGIC_CHECK)
1228 file_magwarn(ms, "relative offset at level 0");
1229
1230 /* get offset, then skip over it */
1231 m->offset = (uint32_t)strtoul(l, &t, 0);
1232 if (l == t)
1233 if (ms->flags & MAGIC_CHECK)
1234 file_magwarn(ms, "offset `%s' invalid", l);
1235 l = t;
1236
1237 if (m->flag & INDIR) {
1238 m->in_type = FILE_LONG;
1239 m->in_offset = 0;
1240 /*
1241 * read [.lbs][+-]nnnnn)
1242 */
1243 if (*l == '.') {
1244 l++;
1245 switch (*l) {
1246 case 'l':
1247 m->in_type = FILE_LELONG;
1248 break;
1249 case 'L':
1250 m->in_type = FILE_BELONG;
1251 break;
1252 case 'm':
1253 m->in_type = FILE_MELONG;
1254 break;
1255 case 'h':
1256 case 's':
1257 m->in_type = FILE_LESHORT;
1258 break;
1259 case 'H':
1260 case 'S':
1261 m->in_type = FILE_BESHORT;
1262 break;
1263 case 'c':
1264 case 'b':
1265 case 'C':
1266 case 'B':
1267 m->in_type = FILE_BYTE;
1268 break;
1269 case 'e':
1270 case 'f':
1271 case 'g':
1272 m->in_type = FILE_LEDOUBLE;
1273 break;
1274 case 'E':
1275 case 'F':
1276 case 'G':
1277 m->in_type = FILE_BEDOUBLE;
1278 break;
1279 case 'i':
1280 m->in_type = FILE_LEID3;
1281 break;
1282 case 'I':
1283 m->in_type = FILE_BEID3;
1284 break;
1285 default:
1286 if (ms->flags & MAGIC_CHECK)
1287 file_magwarn(ms,
1288 "indirect offset type `%c' invalid",
1289 *l);
1290 break;
1291 }
1292 l++;
1293 }
1294
1295 m->in_op = 0;
1296 if (*l == '~') {
1297 m->in_op |= FILE_OPINVERSE;
1298 l++;
1299 }
1300 if ((op = get_op(*l)) != -1) {
1301 m->in_op |= op;
1302 l++;
1303 }
1304 if (*l == '(') {
1305 m->in_op |= FILE_OPINDIRECT;
1306 l++;
1307 }
1308 if (isdigit((unsigned char)*l) || *l == '-') {
1309 m->in_offset = (int32_t)strtol(l, &t, 0);
1310 if (l == t)
1311 if (ms->flags & MAGIC_CHECK)
1312 file_magwarn(ms,
1313 "in_offset `%s' invalid", l);
1314 l = t;
1315 }
1316 if (*l++ != ')' ||
1317 ((m->in_op & FILE_OPINDIRECT) && *l++ != ')'))
1318 if (ms->flags & MAGIC_CHECK)
1319 file_magwarn(ms,
1320 "missing ')' in indirect offset");
1321 }
1322 EATAB;
1323
1324 #ifdef ENABLE_CONDITIONALS
1325 m->cond = get_cond(l, &l);
1326 if (check_cond(ms, m->cond, cont_level) == -1)
1327 return -1;
1328
1329 EATAB;
1330 #endif
1331
1332 if (*l == 'u') {
1333 ++l;
1334 m->flag |= UNSIGNED;
1335 }
1336
1337 m->type = get_type(l, &l);
1338 if (m->type == FILE_INVALID) {
1339 if (ms->flags & MAGIC_CHECK)
1340 file_magwarn(ms, "type `%s' invalid", l);
1341 if (me->mp) {
1342 efree(me->mp);
1343 me->mp = NULL;
1344 }
1345 return -1;
1346 }
1347
1348 /* New-style anding: "0 byte&0x80 =0x80 dynamically linked" */
1349 /* New and improved: ~ & | ^ + - * / % -- exciting, isn't it? */
1350
1351 m->mask_op = 0;
1352 if (*l == '~') {
1353 if (!IS_LIBMAGIC_STRING(m->type))
1354 m->mask_op |= FILE_OPINVERSE;
1355 else if (ms->flags & MAGIC_CHECK)
1356 file_magwarn(ms, "'~' invalid for string types");
1357 ++l;
1358 }
1359 m->str_range = 0;
1360 m->str_flags = m->type == FILE_PSTRING ? PSTRING_1_LE : 0;
1361 if ((op = get_op(*l)) != -1) {
1362 if (!IS_LIBMAGIC_STRING(m->type)) {
1363 uint64_t val;
1364 ++l;
1365 m->mask_op |= op;
1366 val = (uint64_t)strtoull(l, &t, 0);
1367 l = t;
1368 m->num_mask = file_signextend(ms, m, val);
1369 eatsize(&l);
1370 }
1371 else if (op == FILE_OPDIVIDE) {
1372 int have_range = 0;
1373 while (!isspace((unsigned char)*++l)) {
1374 switch (*l) {
1375 case '0': case '1': case '2':
1376 case '3': case '4': case '5':
1377 case '6': case '7': case '8':
1378 case '9':
1379 if (have_range &&
1380 (ms->flags & MAGIC_CHECK))
1381 file_magwarn(ms,
1382 "multiple ranges");
1383 have_range = 1;
1384 m->str_range = CAST(uint32_t,
1385 strtoul(l, &t, 0));
1386 if (m->str_range == 0)
1387 file_magwarn(ms,
1388 "zero range");
1389 l = t - 1;
1390 break;
1391 case CHAR_COMPACT_WHITESPACE:
1392 m->str_flags |=
1393 STRING_COMPACT_WHITESPACE;
1394 break;
1395 case CHAR_COMPACT_OPTIONAL_WHITESPACE:
1396 m->str_flags |=
1397 STRING_COMPACT_OPTIONAL_WHITESPACE;
1398 break;
1399 case CHAR_IGNORE_LOWERCASE:
1400 m->str_flags |= STRING_IGNORE_LOWERCASE;
1401 break;
1402 case CHAR_IGNORE_UPPERCASE:
1403 m->str_flags |= STRING_IGNORE_UPPERCASE;
1404 break;
1405 case CHAR_REGEX_OFFSET_START:
1406 m->str_flags |= REGEX_OFFSET_START;
1407 break;
1408 case CHAR_BINTEST:
1409 m->str_flags |= STRING_BINTEST;
1410 break;
1411 case CHAR_TEXTTEST:
1412 m->str_flags |= STRING_TEXTTEST;
1413 break;
1414 case CHAR_PSTRING_1_LE:
1415 if (m->type != FILE_PSTRING)
1416 goto bad;
1417 m->str_flags = (m->str_flags & ~PSTRING_LEN) | PSTRING_1_LE;
1418 break;
1419 case CHAR_PSTRING_2_BE:
1420 if (m->type != FILE_PSTRING)
1421 goto bad;
1422 m->str_flags = (m->str_flags & ~PSTRING_LEN) | PSTRING_2_BE;
1423 break;
1424 case CHAR_PSTRING_2_LE:
1425 if (m->type != FILE_PSTRING)
1426 goto bad;
1427 m->str_flags = (m->str_flags & ~PSTRING_LEN) | PSTRING_2_LE;
1428 break;
1429 case CHAR_PSTRING_4_BE:
1430 if (m->type != FILE_PSTRING)
1431 goto bad;
1432 m->str_flags = (m->str_flags & ~PSTRING_LEN) | PSTRING_4_BE;
1433 break;
1434 case CHAR_PSTRING_4_LE:
1435 if (m->type != FILE_PSTRING)
1436 goto bad;
1437 m->str_flags = (m->str_flags & ~PSTRING_LEN) | PSTRING_4_LE;
1438 break;
1439 case CHAR_PSTRING_LENGTH_INCLUDES_ITSELF:
1440 if (m->type != FILE_PSTRING)
1441 goto bad;
1442 m->str_flags |= PSTRING_LENGTH_INCLUDES_ITSELF;
1443 break;
1444 default:
1445 bad:
1446 if (ms->flags & MAGIC_CHECK)
1447 file_magwarn(ms,
1448 "string extension `%c' "
1449 "invalid", *l);
1450 return -1;
1451 }
1452 /* allow multiple '/' for readability */
1453 if (l[1] == '/' &&
1454 !isspace((unsigned char)l[2]))
1455 l++;
1456 }
1457 if (string_modifier_check(ms, m) == -1)
1458 return -1;
1459 }
1460 else {
1461 if (ms->flags & MAGIC_CHECK)
1462 file_magwarn(ms, "invalid string op: %c", *t);
1463 return -1;
1464 }
1465 }
1466 /*
1467 * We used to set mask to all 1's here, instead let's just not do
1468 * anything if mask = 0 (unless you have a better idea)
1469 */
1470 EATAB;
1471
1472 switch (*l) {
1473 case '>':
1474 case '<':
1475 m->reln = *l;
1476 ++l;
1477 if (*l == '=') {
1478 if (ms->flags & MAGIC_CHECK) {
1479 file_magwarn(ms, "%c= not supported",
1480 m->reln);
1481 return -1;
1482 }
1483 ++l;
1484 }
1485 break;
1486 /* Old-style anding: "0 byte &0x80 dynamically linked" */
1487 case '&':
1488 case '^':
1489 case '=':
1490 m->reln = *l;
1491 ++l;
1492 if (*l == '=') {
1493 /* HP compat: ignore &= etc. */
1494 ++l;
1495 }
1496 break;
1497 case '!':
1498 m->reln = *l;
1499 ++l;
1500 break;
1501 default:
1502 m->reln = '='; /* the default relation */
1503 if (*l == 'x' && ((isascii((unsigned char)l[1]) &&
1504 isspace((unsigned char)l[1])) || !l[1])) {
1505 m->reln = *l;
1506 ++l;
1507 }
1508 break;
1509 }
1510 /*
1511 * Grab the value part, except for an 'x' reln.
1512 */
1513 if (m->reln != 'x' && getvalue(ms, m, &l, action))
1514 return -1;
1515
1516 /*
1517 * TODO finish this macro and start using it!
1518 * #define offsetcheck {if (offset > HOWMANY-1)
1519 * magwarn("offset too big"); }
1520 */
1521
1522 /*
1523 * Now get last part - the description
1524 */
1525 EATAB;
1526 if (l[0] == '\b') {
1527 ++l;
1528 m->flag |= NOSPACE;
1529 } else if ((l[0] == '\\') && (l[1] == 'b')) {
1530 ++l;
1531 ++l;
1532 m->flag |= NOSPACE;
1533 }
1534 for (i = 0; (m->desc[i++] = *l++) != '\0' && i < sizeof(m->desc); )
1535 continue;
1536 if (i == sizeof(m->desc)) {
1537 m->desc[sizeof(m->desc) - 1] = '\0';
1538 if (ms->flags & MAGIC_CHECK)
1539 file_magwarn(ms, "description `%s' truncated", m->desc);
1540 }
1541
1542 /*
1543 * We only do this check while compiling, or if any of the magic
1544 * files were not compiled.
1545 */
1546 if (ms->flags & MAGIC_CHECK) {
1547 if (check_format(ms, m) == -1)
1548 return -1;
1549 }
1550 m->mimetype[0] = '\0'; /* initialise MIME type to none */
1551 if (m->cont_level == 0)
1552 ++(*nmentryp); /* make room for next */
1553 return 0;
1554 }
1555
1556 /*
1557 * parse a STRENGTH annotation line from magic file, put into magic[index - 1]
1558 * if valid
1559 */
1560 private int
1561 parse_strength(struct magic_set *ms, struct magic_entry *me, const char *line)
1562 {
1563 const char *l = line;
1564 char *el;
1565 unsigned long factor;
1566 struct magic *m = &me->mp[0];
1567
1568 if (m->factor_op != FILE_FACTOR_OP_NONE) {
1569 file_magwarn(ms,
1570 "Current entry already has a strength type: %c %d",
1571 m->factor_op, m->factor);
1572 return -1;
1573 }
1574 EATAB;
1575 switch (*l) {
1576 case FILE_FACTOR_OP_NONE:
1577 case FILE_FACTOR_OP_PLUS:
1578 case FILE_FACTOR_OP_MINUS:
1579 case FILE_FACTOR_OP_TIMES:
1580 case FILE_FACTOR_OP_DIV:
1581 m->factor_op = *l++;
1582 break;
1583 default:
1584 file_magwarn(ms, "Unknown factor op `%c'", *l);
1585 return -1;
1586 }
1587 EATAB;
1588 factor = strtoul(l, &el, 0);
1589 if (factor > 255) {
1590 file_magwarn(ms, "Too large factor `%lu'", factor);
1591 goto out;
1592 }
1593 if (*el && !isspace((unsigned char)*el)) {
1594 file_magwarn(ms, "Bad factor `%s'", l);
1595 goto out;
1596 }
1597 m->factor = (uint8_t)factor;
1598 if (m->factor == 0 && m->factor_op == FILE_FACTOR_OP_DIV) {
1599 file_magwarn(ms, "Cannot have factor op `%c' and factor %u",
1600 m->factor_op, m->factor);
1601 goto out;
1602 }
1603 return 0;
1604 out:
1605 m->factor_op = FILE_FACTOR_OP_NONE;
1606 m->factor = 0;
1607 return -1;
1608 }
1609
1610 /*
1611 * Parse an Apple CREATOR/TYPE annotation from magic file and put it into
1612 * magic[index - 1]
1613 */
1614 private int
1615 parse_apple(struct magic_set *ms, struct magic_entry *me, const char *line)
1616 {
1617 size_t i;
1618 const char *l = line;
1619 struct magic *m = &me->mp[me->cont_count == 0 ? 0 : me->cont_count - 1];
1620
1621 if (m->apple[0] != '\0') {
1622 file_magwarn(ms, "Current entry already has a APPLE type "
1623 "`%.8s', new type `%s'", m->mimetype, l);
1624 return -1;
1625 }
1626
1627 EATAB;
1628 for (i = 0; *l && ((isascii((unsigned char)*l) &&
1629 isalnum((unsigned char)*l)) || strchr("-+/.", *l)) &&
1630 i < sizeof(m->apple); m->apple[i++] = *l++)
1631 continue;
1632 if (i == sizeof(m->apple) && *l) {
1633 /* We don't need to NUL terminate here, printing handles it */
1634 if (ms->flags & MAGIC_CHECK)
1635 file_magwarn(ms, "APPLE type `%s' truncated %"
1636 SIZE_T_FORMAT "u", line, i);
1637 }
1638
1639 if (i > 0)
1640 return 0;
1641 else
1642 return -1;
1643 }
1644
1645 /*
1646 * parse a MIME annotation line from magic file, put into magic[index - 1]
1647 * if valid
1648 */
1649 private int
1650 parse_mime(struct magic_set *ms, struct magic_entry *me, const char *line)
1651 {
1652 size_t i;
1653 const char *l = line;
1654 struct magic *m = &me->mp[me->cont_count == 0 ? 0 : me->cont_count - 1];
1655
1656 if (m->mimetype[0] != '\0') {
1657 file_magwarn(ms, "Current entry already has a MIME type `%s',"
1658 " new type `%s'", m->mimetype, l);
1659 return -1;
1660 }
1661
1662 EATAB;
1663 for (i = 0; *l && ((isascii((unsigned char)*l) &&
1664 isalnum((unsigned char)*l)) || strchr("-+/.", *l)) &&
1665 i < sizeof(m->mimetype); m->mimetype[i++] = *l++)
1666 continue;
1667 if (i == sizeof(m->mimetype)) {
1668 m->mimetype[sizeof(m->mimetype) - 1] = '\0';
1669 if (ms->flags & MAGIC_CHECK)
1670 file_magwarn(ms, "MIME type `%s' truncated %"
1671 SIZE_T_FORMAT "u", m->mimetype, i);
1672 } else
1673 m->mimetype[i] = '\0';
1674
1675 if (i > 0)
1676 return 0;
1677 else
1678 return -1;
1679 }
1680
1681 private int
1682 check_format_type(const char *ptr, int type)
1683 {
1684 int quad = 0;
1685 if (*ptr == '\0') {
1686 /* Missing format string; bad */
1687 return -1;
1688 }
1689
1690 switch (type) {
1691 case FILE_FMT_QUAD:
1692 quad = 1;
1693 /*FALLTHROUGH*/
1694 case FILE_FMT_NUM:
1695 if (*ptr == '-')
1696 ptr++;
1697 if (*ptr == '.')
1698 ptr++;
1699 while (isdigit((unsigned char)*ptr)) ptr++;
1700 if (*ptr == '.')
1701 ptr++;
1702 while (isdigit((unsigned char)*ptr)) ptr++;
1703 if (quad) {
1704 if (*ptr++ != 'l')
1705 return -1;
1706 if (*ptr++ != 'l')
1707 return -1;
1708 }
1709
1710 switch (*ptr++) {
1711 case 'l':
1712 switch (*ptr++) {
1713 case 'i':
1714 case 'd':
1715 case 'u':
1716 case 'x':
1717 case 'X':
1718 return 0;
1719 default:
1720 return -1;
1721 }
1722
1723 case 'h':
1724 switch (*ptr++) {
1725 case 'h':
1726 switch (*ptr++) {
1727 case 'i':
1728 case 'd':
1729 case 'u':
1730 case 'x':
1731 case 'X':
1732 return 0;
1733 default:
1734 return -1;
1735 }
1736 case 'd':
1737 return 0;
1738 default:
1739 return -1;
1740 }
1741
1742 case 'i':
1743 case 'c':
1744 case 'd':
1745 case 'u':
1746 case 'x':
1747 case 'X':
1748 return 0;
1749
1750 default:
1751 return -1;
1752 }
1753
1754 case FILE_FMT_FLOAT:
1755 case FILE_FMT_DOUBLE:
1756 if (*ptr == '-')
1757 ptr++;
1758 if (*ptr == '.')
1759 ptr++;
1760 while (isdigit((unsigned char)*ptr)) ptr++;
1761 if (*ptr == '.')
1762 ptr++;
1763 while (isdigit((unsigned char)*ptr)) ptr++;
1764
1765 switch (*ptr++) {
1766 case 'e':
1767 case 'E':
1768 case 'f':
1769 case 'F':
1770 case 'g':
1771 case 'G':
1772 return 0;
1773
1774 default:
1775 return -1;
1776 }
1777
1778
1779 case FILE_FMT_STR:
1780 if (*ptr == '-')
1781 ptr++;
1782 while (isdigit((unsigned char )*ptr))
1783 ptr++;
1784 if (*ptr == '.') {
1785 ptr++;
1786 while (isdigit((unsigned char )*ptr))
1787 ptr++;
1788 }
1789
1790 switch (*ptr++) {
1791 case 's':
1792 return 0;
1793 default:
1794 return -1;
1795 }
1796
1797 default:
1798 /* internal error */
1799 abort();
1800 }
1801 /*NOTREACHED*/
1802 return -1;
1803 }
1804
1805 /*
1806 * Check that the optional printf format in description matches
1807 * the type of the magic.
1808 */
1809 private int
1810 check_format(struct magic_set *ms, struct magic *m)
1811 {
1812 char *ptr;
1813
1814 for (ptr = m->desc; *ptr; ptr++)
1815 if (*ptr == '%')
1816 break;
1817 if (*ptr == '\0') {
1818 /* No format string; ok */
1819 return 1;
1820 }
1821
1822 assert(file_nformats == file_nnames);
1823
1824 if (m->type >= file_nformats) {
1825 file_magwarn(ms, "Internal error inconsistency between "
1826 "m->type and format strings");
1827 return -1;
1828 }
1829 if (file_formats[m->type] == FILE_FMT_NONE) {
1830 file_magwarn(ms, "No format string for `%s' with description "
1831 "`%s'", m->desc, file_names[m->type]);
1832 return -1;
1833 }
1834
1835 ptr++;
1836 if (check_format_type(ptr, file_formats[m->type]) == -1) {
1837 /*
1838 * TODO: this error message is unhelpful if the format
1839 * string is not one character long
1840 */
1841 file_magwarn(ms, "Printf format `%c' is not valid for type "
1842 "`%s' in description `%s'", *ptr ? *ptr : '?',
1843 file_names[m->type], m->desc);
1844 return -1;
1845 }
1846
1847 for (; *ptr; ptr++) {
1848 if (*ptr == '%') {
1849 file_magwarn(ms,
1850 "Too many format strings (should have at most one) "
1851 "for `%s' with description `%s'",
1852 file_names[m->type], m->desc);
1853 return -1;
1854 }
1855 }
1856 return 0;
1857 }
1858
1859 /*
1860 * Read a numeric value from a pointer, into the value union of a magic
1861 * pointer, according to the magic type. Update the string pointer to point
1862 * just after the number read. Return 0 for success, non-zero for failure.
1863 */
1864 private int
1865 getvalue(struct magic_set *ms, struct magic *m, const char **p, int action)
1866 {
1867 switch (m->type) {
1868 case FILE_BESTRING16:
1869 case FILE_LESTRING16:
1870 case FILE_STRING:
1871 case FILE_PSTRING:
1872 case FILE_REGEX:
1873 case FILE_SEARCH:
1874 *p = getstr(ms, m, *p, action == FILE_COMPILE);
1875 if (*p == NULL) {
1876 if (ms->flags & MAGIC_CHECK)
1877 file_magwarn(ms, "cannot get string from `%s'",
1878 m->value.s);
1879 return -1;
1880 }
1881 return 0;
1882 case FILE_FLOAT:
1883 case FILE_BEFLOAT:
1884 case FILE_LEFLOAT:
1885 if (m->reln != 'x') {
1886 char *ep;
1887 #ifdef HAVE_STRTOF
1888 m->value.f = strtof(*p, &ep);
1889 #else
1890 m->value.f = (float)strtod(*p, &ep);
1891 #endif
1892 *p = ep;
1893 }
1894 return 0;
1895 case FILE_DOUBLE:
1896 case FILE_BEDOUBLE:
1897 case FILE_LEDOUBLE:
1898 if (m->reln != 'x') {
1899 char *ep;
1900 m->value.d = strtod(*p, &ep);
1901 *p = ep;
1902 }
1903 return 0;
1904 default:
1905 if (m->reln != 'x') {
1906 char *ep;
1907 m->value.q = file_signextend(ms, m,
1908 (uint64_t)strtoull(*p, &ep, 0));
1909 *p = ep;
1910 eatsize(p);
1911 }
1912 return 0;
1913 }
1914 }
1915
1916 /*
1917 * Convert a string containing C character escapes. Stop at an unescaped
1918 * space or tab.
1919 * Copy the converted version to "m->value.s", and the length in m->vallen.
1920 * Return updated scan pointer as function result. Warn if set.
1921 */
1922 private const char *
1923 getstr(struct magic_set *ms, struct magic *m, const char *s, int warn)
1924 {
1925 const char *origs = s;
1926 char *p = m->value.s;
1927 size_t plen = sizeof(m->value.s);
1928 char *origp = p;
1929 char *pmax = p + plen - 1;
1930 int c;
1931 int val;
1932
1933 while ((c = *s++) != '\0') {
1934 if (isspace((unsigned char) c))
1935 break;
1936 if (p >= pmax) {
1937 file_error(ms, 0, "string too long: `%s'", origs);
1938 return NULL;
1939 }
1940 if (c == '\\') {
1941 switch(c = *s++) {
1942
1943 case '\0':
1944 if (warn)
1945 file_magwarn(ms, "incomplete escape");
1946 goto out;
1947
1948 case '\t':
1949 if (warn) {
1950 file_magwarn(ms,
1951 "escaped tab found, use \\t instead");
1952 warn = 0; /* already did */
1953 }
1954 /*FALLTHROUGH*/
1955 default:
1956 if (warn) {
1957 if (isprint((unsigned char)c)) {
1958 /* Allow escaping of
1959 * ``relations'' */
1960 if (strchr("<>&^=!", c) == NULL
1961 && (m->type != FILE_REGEX ||
1962 strchr("[]().*?^$|{}", c)
1963 == NULL)) {
1964 file_magwarn(ms, "no "
1965 "need to escape "
1966 "`%c'", c);
1967 }
1968 } else {
1969 file_magwarn(ms,
1970 "unknown escape sequence: "
1971 "\\%03o", c);
1972 }
1973 }
1974 /*FALLTHROUGH*/
1975 /* space, perhaps force people to use \040? */
1976 case ' ':
1977 #if 0
1978 /*
1979 * Other things people escape, but shouldn't need to,
1980 * so we disallow them
1981 */
1982 case '\'':
1983 case '"':
1984 case '?':
1985 #endif
1986 /* Relations */
1987 case '>':
1988 case '<':
1989 case '&':
1990 case '^':
1991 case '=':
1992 case '!':
1993 /* and baskslash itself */
1994 case '\\':
1995 *p++ = (char) c;
1996 break;
1997
1998 case 'a':
1999 *p++ = '\a';
2000 break;
2001
2002 case 'b':
2003 *p++ = '\b';
2004 break;
2005
2006 case 'f':
2007 *p++ = '\f';
2008 break;
2009
2010 case 'n':
2011 *p++ = '\n';
2012 break;
2013
2014 case 'r':
2015 *p++ = '\r';
2016 break;
2017
2018 case 't':
2019 *p++ = '\t';
2020 break;
2021
2022 case 'v':
2023 *p++ = '\v';
2024 break;
2025
2026 /* \ and up to 3 octal digits */
2027 case '0':
2028 case '1':
2029 case '2':
2030 case '3':
2031 case '4':
2032 case '5':
2033 case '6':
2034 case '7':
2035 val = c - '0';
2036 c = *s++; /* try for 2 */
2037 if (c >= '0' && c <= '7') {
2038 val = (val << 3) | (c - '0');
2039 c = *s++; /* try for 3 */
2040 if (c >= '0' && c <= '7')
2041 val = (val << 3) | (c-'0');
2042 else
2043 --s;
2044 }
2045 else
2046 --s;
2047 *p++ = (char)val;
2048 break;
2049
2050 /* \x and up to 2 hex digits */
2051 case 'x':
2052 val = 'x'; /* Default if no digits */
2053 c = hextoint(*s++); /* Get next char */
2054 if (c >= 0) {
2055 val = c;
2056 c = hextoint(*s++);
2057 if (c >= 0)
2058 val = (val << 4) + c;
2059 else
2060 --s;
2061 } else
2062 --s;
2063 *p++ = (char)val;
2064 break;
2065 }
2066 } else
2067 *p++ = (char)c;
2068 }
2069 out:
2070 *p = '\0';
2071 m->vallen = CAST(unsigned char, (p - origp));
2072 if (m->type == FILE_PSTRING)
2073 m->vallen += (unsigned char)file_pstring_length_size(m);
2074 return s;
2075 }
2076
2077
2078 /* Single hex char to int; -1 if not a hex char. */
2079 private int
2080 hextoint(int c)
2081 {
2082 if (!isascii((unsigned char) c))
2083 return -1;
2084 if (isdigit((unsigned char) c))
2085 return c - '0';
2086 if ((c >= 'a') && (c <= 'f'))
2087 return c + 10 - 'a';
2088 if (( c>= 'A') && (c <= 'F'))
2089 return c + 10 - 'A';
2090 return -1;
2091 }
2092
2093
2094 /*
2095 * Print a string containing C character escapes.
2096 */
2097 protected void
2098 file_showstr(FILE *fp, const char *s, size_t len)
2099 {
2100 char c;
2101
2102 for (;;) {
2103 if (len == ~0U) {
2104 c = *s++;
2105 if (c == '\0')
2106 break;
2107 }
2108 else {
2109 if (len-- == 0)
2110 break;
2111 c = *s++;
2112 }
2113 if (c >= 040 && c <= 0176) /* TODO isprint && !iscntrl */
2114 (void) fputc(c, fp);
2115 else {
2116 (void) fputc('\\', fp);
2117 switch (c) {
2118 case '\a':
2119 (void) fputc('a', fp);
2120 break;
2121
2122 case '\b':
2123 (void) fputc('b', fp);
2124 break;
2125
2126 case '\f':
2127 (void) fputc('f', fp);
2128 break;
2129
2130 case '\n':
2131 (void) fputc('n', fp);
2132 break;
2133
2134 case '\r':
2135 (void) fputc('r', fp);
2136 break;
2137
2138 case '\t':
2139 (void) fputc('t', fp);
2140 break;
2141
2142 case '\v':
2143 (void) fputc('v', fp);
2144 break;
2145
2146 default:
2147 (void) fprintf(fp, "%.3o", c & 0377);
2148 break;
2149 }
2150 }
2151 }
2152 }
2153
2154 /*
2155 * eatsize(): Eat the size spec from a number [eg. 10UL]
2156 */
2157 private void
2158 eatsize(const char **p)
2159 {
2160 const char *l = *p;
2161
2162 if (LOWCASE(*l) == 'u')
2163 l++;
2164
2165 switch (LOWCASE(*l)) {
2166 case 'l': /* long */
2167 case 's': /* short */
2168 case 'h': /* short */
2169 case 'b': /* char/byte */
2170 case 'c': /* char/byte */
2171 l++;
2172 /*FALLTHROUGH*/
2173 default:
2174 break;
2175 }
2176
2177 *p = l;
2178 }
2179
2180 /*
2181 * handle a compiled file.
2182 * return -1 = error
2183 * return 1 = memory structure you can free
2184 * return 3 = bundled library from PHP
2185 */
2186 private int
2187 apprentice_map(struct magic_set *ms, struct magic **magicp, uint32_t *nmagicp,
2188 const char *fn)
2189 {
2190 uint32_t *ptr;
2191 uint32_t version;
2192 int needsbyteswap;
2193 char *dbname = NULL;
2194 void *mm = NULL;
2195 int ret = 0;
2196 php_stream *stream = NULL;
2197 php_stream_statbuf st;
2198
2199
2200 TSRMLS_FETCH();
2201
2202 if (fn == NULL) {
2203 mm = (void *)&php_magic_database;
2204 ret = 3;
2205 goto internal_loaded;
2206 }
2207
2208 #ifdef PHP_WIN32
2209 /* Don't bother on windows with php_stream_open_wrapper,
2210 return to give apprentice_load() a chance. */
2211 if (php_stream_stat_path_ex(fn, 0, &st, NULL) == SUCCESS) {
2212 if (st.sb.st_mode & S_IFDIR) {
2213 goto error2;
2214 }
2215 }
2216 #endif
2217
2218 dbname = mkdbname(ms, fn, 0);
2219 if (dbname == NULL)
2220 goto error2;
2221
2222 #if PHP_API_VERSION < 20100412
2223 stream = php_stream_open_wrapper((char *)fn, "rb", REPORT_ERRORS|ENFORCE_SAFE_MODE, NULL);
2224 #else
2225 stream = php_stream_open_wrapper((char *)fn, "rb", REPORT_ERRORS, NULL);
2226 #endif
2227
2228 if (!stream) {
2229 goto error2;
2230 }
2231
2232 if (php_stream_stat(stream, &st) < 0) {
2233 file_error(ms, errno, "cannot stat `%s'", dbname);
2234 goto error1;
2235 }
2236
2237 if (st.sb.st_size < 8) {
2238 file_error(ms, 0, "file `%s' is too small", dbname);
2239 goto error1;
2240 }
2241
2242 mm = emalloc((size_t)st.sb.st_size);
2243 if (php_stream_read(stream, mm, (size_t)st.sb.st_size) != (size_t)st.sb.st_size) {
2244 file_badread(ms);
2245 ret = 1;
2246 goto error1;
2247 }
2248 ret = 1;
2249
2250 php_stream_close(stream);
2251 stream = NULL;
2252
2253 internal_loaded:
2254 *magicp = mm;
2255 ptr = (uint32_t *)(void *)*magicp;
2256 if (*ptr != MAGICNO) {
2257 if (swap4(*ptr) != MAGICNO) {
2258 file_error(ms, 0, "bad magic in `%s'", dbname);
2259 goto error1;
2260 }
2261 needsbyteswap = 1;
2262 } else
2263 needsbyteswap = 0;
2264 if (needsbyteswap)
2265 version = swap4(ptr[1]);
2266 else
2267 version = ptr[1];
2268 if (version != VERSIONNO) {
2269 file_error(ms, 0, "File %d.%d supports only version %d magic "
2270 "files. `%s' is version %d", FILE_VERSION_MAJOR, patchlevel,
2271 VERSIONNO, dbname, version);
2272 goto error1;
2273 }
2274
2275 /* php_magic_database is a const, performing writes will segfault. This is for big-endian
2276 machines only, PPC and Sparc specifically. Consider static variable or MINIT in
2277 future. */
2278 if (needsbyteswap && fn == NULL) {
2279 mm = emalloc(sizeof(php_magic_database));
2280 mm = memcpy(mm, php_magic_database, sizeof(php_magic_database));
2281 *magicp = mm;
2282 ret = 1;
2283 }
2284
2285 if (fn == NULL) {
2286 *nmagicp = (sizeof(php_magic_database) / sizeof(struct magic));
2287 } else {
2288 *nmagicp = (uint32_t)(st.sb.st_size / sizeof(struct magic));
2289 }
2290 if (*nmagicp > 0) {
2291 (*nmagicp)--;
2292 }
2293 (*magicp)++;
2294 if (needsbyteswap) {
2295 byteswap(*magicp, *nmagicp);
2296 }
2297
2298 if (dbname) {
2299 efree(dbname);
2300 }
2301 return ret;
2302
2303 error1:
2304 if (stream) {
2305 php_stream_close(stream);
2306 }
2307
2308 if (mm && ret == 1) {
2309 efree(mm);
2310 } else {
2311 *magicp = NULL;
2312 *nmagicp = 0;
2313 }
2314 error2:
2315 if (dbname) {
2316 efree(dbname);
2317 }
2318 return -1;
2319 }
2320
2321 private const uint32_t ar[] = {
2322 MAGICNO, VERSIONNO
2323 };
2324 /*
2325 * handle an mmaped file.
2326 */
2327 private int
2328 apprentice_compile(struct magic_set *ms, struct magic **magicp,
2329 uint32_t *nmagicp, const char *fn)
2330 {
2331 char *dbname;
2332 int rv = -1;
2333 php_stream *stream;
2334
2335 TSRMLS_FETCH();
2336
2337 dbname = mkdbname(ms, fn, 0);
2338
2339 if (dbname == NULL)
2340 goto out;
2341
2342 /* wb+ == O_WRONLY|O_CREAT|O_TRUNC|O_BINARY */
2343 #if PHP_API_VERSION < 20100412
2344 stream = php_stream_open_wrapper((char *)fn, "wb+", REPORT_ERRORS|ENFORCE_SAFE_MODE, NULL);
2345 #else
2346 stream = php_stream_open_wrapper((char *)fn, "wb+", REPORT_ERRORS, NULL);
2347 #endif
2348
2349 if (!stream) {
2350 file_error(ms, errno, "cannot open `%s'", dbname);
2351 goto out;
2352 }
2353
2354 if (php_stream_write(stream, (char *)ar, sizeof(ar)) != (ssize_t)sizeof(ar)) {
2355 file_error(ms, errno, "error writing `%s'", dbname);
2356 goto out;
2357 }
2358
2359 if (php_stream_seek(stream,(off_t)sizeof(struct magic), SEEK_SET) != sizeof(struct magic)) {
2360 file_error(ms, errno, "error seeking `%s'", dbname);
2361 goto out;
2362 }
2363
2364 if (php_stream_write(stream, (char *)*magicp, (sizeof(struct magic) * *nmagicp) != (ssize_t)(sizeof(struct magic) * *nmagicp))) {
2365 file_error(ms, errno, "error writing `%s'", dbname);
2366 goto out;
2367 }
2368
2369 php_stream_close(stream);
2370
2371 rv = 0;
2372 out:
2373 efree(dbname);
2374 return rv;
2375 }
2376
2377 private const char ext[] = ".mgc";
2378 /*
2379 * make a dbname
2380 */
2381 private char *
2382 mkdbname(struct magic_set *ms, const char *fn, int strip)
2383 {
2384 const char *p, *q;
2385 char *buf;
2386 TSRMLS_FETCH();
2387
2388 if (strip) {
2389 if ((p = strrchr(fn, '/')) != NULL)
2390 fn = ++p;
2391 }
2392
2393 for (q = fn; *q; q++)
2394 continue;
2395 /* Look for .mgc */
2396 for (p = ext + sizeof(ext) - 1; p >= ext && q >= fn; p--, q--)
2397 if (*p != *q)
2398 break;
2399
2400 /* Did not find .mgc, restore q */
2401 if (p >= ext)
2402 while (*q)
2403 q++;
2404
2405 q++;
2406 /* Compatibility with old code that looked in .mime */
2407 if (ms->flags & MAGIC_MIME) {
2408 spprintf(&buf, MAXPATHLEN, "%.*s.mime%s", (int)(q - fn), fn, ext);
2409 #ifdef PHP_WIN32
2410 if (VCWD_ACCESS(buf, R_OK) == 0) {
2411 #else
2412 if (VCWD_ACCESS(buf, R_OK) != -1) {
2413 #endif
2414 ms->flags &= MAGIC_MIME_TYPE;
2415 return buf;
2416 }
2417 efree(buf);
2418 }
2419 spprintf(&buf, MAXPATHLEN, "%.*s%s", (int)(q - fn), fn, ext);
2420
2421 /* Compatibility with old code that looked in .mime */
2422 if (strstr(p, ".mime") != NULL)
2423 ms->flags &= MAGIC_MIME_TYPE;
2424 return buf;
2425 }
2426
2427 /*
2428 * Byteswap an mmap'ed file if needed
2429 */
2430 private void
2431 byteswap(struct magic *magic, uint32_t nmagic)
2432 {
2433 uint32_t i;
2434 for (i = 0; i < nmagic; i++)
2435 bs1(&magic[i]);
2436 }
2437
2438 /*
2439 * swap a short
2440 */
2441 private uint16_t
2442 swap2(uint16_t sv)
2443 {
2444 uint16_t rv;
2445 uint8_t *s = (uint8_t *)(void *)&sv;
2446 uint8_t *d = (uint8_t *)(void *)&rv;
2447 d[0] = s[1];
2448 d[1] = s[0];
2449 return rv;
2450 }
2451
2452 /*
2453 * swap an int
2454 */
2455 private uint32_t
2456 swap4(uint32_t sv)
2457 {
2458 uint32_t rv;
2459 uint8_t *s = (uint8_t *)(void *)&sv;
2460 uint8_t *d = (uint8_t *)(void *)&rv;
2461 d[0] = s[3];
2462 d[1] = s[2];
2463 d[2] = s[1];
2464 d[3] = s[0];
2465 return rv;
2466 }
2467
2468 /*
2469 * swap a quad
2470 */
2471 private uint64_t
2472 swap8(uint64_t sv)
2473 {
2474 uint64_t rv;
2475 uint8_t *s = (uint8_t *)(void *)&sv;
2476 uint8_t *d = (uint8_t *)(void *)&rv;
2477 #if 0
2478 d[0] = s[3];
2479 d[1] = s[2];
2480 d[2] = s[1];
2481 d[3] = s[0];
2482 d[4] = s[7];
2483 d[5] = s[6];
2484 d[6] = s[5];
2485 d[7] = s[4];
2486 #else
2487 d[0] = s[7];
2488 d[1] = s[6];
2489 d[2] = s[5];
2490 d[3] = s[4];
2491 d[4] = s[3];
2492 d[5] = s[2];
2493 d[6] = s[1];
2494 d[7] = s[0];
2495 #endif
2496 return rv;
2497 }
2498
2499 /*
2500 * byteswap a single magic entry
2501 */
2502 private void
2503 bs1(struct magic *m)
2504 {
2505 m->cont_level = swap2(m->cont_level);
2506 m->offset = swap4((uint32_t)m->offset);
2507 m->in_offset = swap4((uint32_t)m->in_offset);
2508 m->lineno = swap4((uint32_t)m->lineno);
2509 if (IS_LIBMAGIC_STRING(m->type)) {
2510 m->str_range = swap4(m->str_range);
2511 m->str_flags = swap4(m->str_flags);
2512 }
2513 else {
2514 m->value.q = swap8(m->value.q);
2515 m->num_mask = swap8(m->num_mask);
2516 }
2517 }
2518
2519 protected size_t
2520 file_pstring_length_size(const struct magic *m)
2521 {
2522 switch (m->str_flags & PSTRING_LEN) {
2523 case PSTRING_1_LE:
2524 return 1;
2525 case PSTRING_2_LE:
2526 case PSTRING_2_BE:
2527 return 2;
2528 case PSTRING_4_LE:
2529 case PSTRING_4_BE:
2530 return 4;
2531 default:
2532 abort(); /* Impossible */
2533 return 1;
2534 }
2535 }
2536 protected size_t
2537 file_pstring_get_length(const struct magic *m, const char *s)
2538 {
2539 size_t len = 0;
2540
2541 switch (m->str_flags & PSTRING_LEN) {
2542 case PSTRING_1_LE:
2543 len = *s;
2544 break;
2545 case PSTRING_2_LE:
2546 len = (s[1] << 8) | s[0];
2547 break;
2548 case PSTRING_2_BE:
2549 len = (s[0] << 8) | s[1];
2550 break;
2551 case PSTRING_4_LE:
2552 len = (s[3] << 24) | (s[2] << 16) | (s[1] << 8) | s[0];
2553 break;
2554 case PSTRING_4_BE:
2555 len = (s[0] << 24) | (s[1] << 16) | (s[2] << 8) | s[3];
2556 break;
2557 default:
2558 abort(); /* Impossible */
2559 }
2560
2561 if (m->str_flags & PSTRING_LENGTH_INCLUDES_ITSELF)
2562 len -= file_pstring_length_size(m);
2563
2564 return len;
2565 }
2566
