![[BACK]](/icons/back.gif){kind=icon}
Return to magic-load.c CVS log ![[TXT]](/icons/text.gif){kind=icon}
![[DIR]](/icons/dir.gif){kind=icon}
Up to [local] / src / usr.bin / file|
Return to magic-load.c CVS log | Up to [local] / src / usr.bin / file |
1.19 ! tobias 1: /* $OpenBSD: magic-load.c,v 1.18 2015/10/05 20:05:52 nicm Exp $ */
1.1 nicm 2:
3: /*
4: * Copyright (c) 2015 Nicholas Marriott <nicm@openbsd.org>
5: *
6: * Permission to use, copy, modify, and distribute this software for any
7: * purpose with or without fee is hereby granted, provided that the above
8: * copyright notice and this permission notice appear in all copies.
9: *
10: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
11: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
12: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
13: * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
14: * WHATSOEVER RESULTING FROM LOSS OF MIND, USE, DATA OR PROFITS, WHETHER
15: * IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING
16: * OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
17: */
18:
19: #include <sys/types.h>
20:
21: #include <ctype.h>
22: #include <errno.h>
23: #include <limits.h>
24: #include <regex.h>
25: #include <stdarg.h>
26: #include <stdio.h>
27: #include <stdlib.h>
28: #include <string.h>
29:
30: #include "magic.h"
31: #include "xmalloc.h"
32:
33: static int
34: magic_odigit(u_char c)
35: {
36: if (c >= '0' && c <= '7')
37: return (c - '0');
38: return (-1);
39: }
40:
41: static int
42: magic_xdigit(u_char c)
43: {
44: if (c >= '0' && c <= '9')
45: return (c - '0');
46: if (c >= 'a' && c <= 'f')
47: return (10 + c - 'a');
48: if (c >= 'A' && c <= 'F')
49: return (10 + c - 'A');
50: return (-1);
51: }
52:
53: static void
54: magic_mark_text(struct magic_line *ml, int text)
55: {
56: do {
57: ml->text = text;
58: ml = ml->parent;
59: } while (ml != NULL);
60: }
61:
62: static int
63: magic_make_pattern(struct magic_line *ml, const char *name, regex_t *re,
64: const char *p)
65: {
66: int error;
67: char errbuf[256];
68:
69: error = regcomp(re, p, REG_EXTENDED|REG_NOSUB);
70: if (error != 0) {
71: regerror(error, re, errbuf, sizeof errbuf);
72: magic_warn(ml, "bad %s pattern: %s", name, errbuf);
73: return (-1);
74: }
75: return (0);
76: }
77:
78: static int
79: magic_set_result(struct magic_line *ml, const char *s)
80: {
81: const char *fmt;
82: const char *endfmt;
83: const char *cp;
84: regex_t *re = NULL;
85: regmatch_t pmatch;
86: size_t fmtlen;
87:
88: while (isspace((u_char)*s))
89: s++;
90: if (*s == '\0') {
91: ml->result = NULL;
92: return (0);
93: }
94: ml->result = xstrdup(s);
95:
96: fmt = NULL;
97: for (cp = s; *cp != '\0'; cp++) {
98: if (cp[0] == '%' && cp[1] != '%') {
99: if (fmt != NULL) {
100: magic_warn(ml, "multiple formats");
101: return (-1);
102: }
103: fmt = cp;
104: }
105: }
106: if (fmt == NULL)
107: return (0);
108: fmt++;
109:
110: for (endfmt = fmt; *endfmt != '\0'; endfmt++) {
111: if (strchr("diouxXeEfFgGsc", *endfmt) != NULL)
112: break;
113: }
114: if (*endfmt == '\0') {
115: magic_warn(ml, "unterminated format");
116: return (-1);
117: }
118: fmtlen = endfmt + 1 - fmt;
119: if (fmtlen > 32) {
120: magic_warn(ml, "format too long");
121: return (-1);
122: }
123:
124: if (*endfmt == 's') {
125: switch (ml->type) {
126: case MAGIC_TYPE_DATE:
127: case MAGIC_TYPE_LDATE:
128: case MAGIC_TYPE_UDATE:
129: case MAGIC_TYPE_ULDATE:
130: case MAGIC_TYPE_BEDATE:
131: case MAGIC_TYPE_BELDATE:
132: case MAGIC_TYPE_UBEDATE:
133: case MAGIC_TYPE_UBELDATE:
134: case MAGIC_TYPE_QDATE:
135: case MAGIC_TYPE_QLDATE:
136: case MAGIC_TYPE_UQDATE:
137: case MAGIC_TYPE_UQLDATE:
138: case MAGIC_TYPE_BEQDATE:
139: case MAGIC_TYPE_BEQLDATE:
140: case MAGIC_TYPE_UBEQDATE:
141: case MAGIC_TYPE_UBEQLDATE:
142: case MAGIC_TYPE_LEQDATE:
143: case MAGIC_TYPE_LEQLDATE:
144: case MAGIC_TYPE_ULEQDATE:
145: case MAGIC_TYPE_ULEQLDATE:
146: case MAGIC_TYPE_LEDATE:
147: case MAGIC_TYPE_LELDATE:
148: case MAGIC_TYPE_ULEDATE:
149: case MAGIC_TYPE_ULELDATE:
150: case MAGIC_TYPE_MEDATE:
151: case MAGIC_TYPE_MELDATE:
152: case MAGIC_TYPE_STRING:
153: case MAGIC_TYPE_PSTRING:
154: case MAGIC_TYPE_BESTRING16:
155: case MAGIC_TYPE_LESTRING16:
156: case MAGIC_TYPE_REGEX:
157: case MAGIC_TYPE_SEARCH:
158: break;
159: default:
160: ml->stringify = 1;
161: break;
162: }
163: }
164:
165: if (!ml->root->compiled) {
166: /*
167: * XXX %ld (and %lu and so on) is invalid on 64-bit platforms
168: * with byte, short, long. We get lucky because our first and
169: * only argument ends up in a register. Accept it for now.
170: */
171: if (magic_make_pattern(ml, "short", &ml->root->format_short,
172: "^-?[0-9]*(\\.[0-9]*)?(c|(l|h|hh)?[iduxX])$") != 0)
173: return (-1);
174: if (magic_make_pattern(ml, "long", &ml->root->format_long,
175: "^-?[0-9]*(\\.[0-9]*)?(c|(l|h|hh)?[iduxX])$") != 0)
176: return (-1);
177: if (magic_make_pattern(ml, "quad", &ml->root->format_quad,
178: "^-?[0-9]*(\\.[0-9]*)?ll[iduxX]$") != 0)
179: return (-1);
180: if (magic_make_pattern(ml, "float", &ml->root->format_float,
181: "^-?[0-9]*(\\.[0-9]*)?[eEfFgG]$") != 0)
182: return (-1);
183: if (magic_make_pattern(ml, "string", &ml->root->format_string,
184: "^-?[0-9]*(\\.[0-9]*)?s$") != 0)
185: return (-1);
186: ml->root->compiled = 1;
187: }
188:
189: if (ml->stringify)
190: re = &ml->root->format_string;
191: else {
192: switch (ml->type) {
193: case MAGIC_TYPE_NONE:
194: case MAGIC_TYPE_DEFAULT:
1.17 nicm 195: case MAGIC_TYPE_BESTRING16:
196: case MAGIC_TYPE_LESTRING16:
1.1 nicm 197: return (0); /* don't use result */
198: case MAGIC_TYPE_BYTE:
199: case MAGIC_TYPE_UBYTE:
200: case MAGIC_TYPE_SHORT:
201: case MAGIC_TYPE_USHORT:
202: case MAGIC_TYPE_BESHORT:
203: case MAGIC_TYPE_UBESHORT:
204: case MAGIC_TYPE_LESHORT:
205: case MAGIC_TYPE_ULESHORT:
206: re = &ml->root->format_short;
207: break;
208: case MAGIC_TYPE_LONG:
209: case MAGIC_TYPE_ULONG:
210: case MAGIC_TYPE_BELONG:
211: case MAGIC_TYPE_UBELONG:
212: case MAGIC_TYPE_LELONG:
213: case MAGIC_TYPE_ULELONG:
214: case MAGIC_TYPE_MELONG:
215: re = &ml->root->format_long;
216: break;
217: case MAGIC_TYPE_QUAD:
218: case MAGIC_TYPE_UQUAD:
219: case MAGIC_TYPE_BEQUAD:
220: case MAGIC_TYPE_UBEQUAD:
221: case MAGIC_TYPE_LEQUAD:
222: case MAGIC_TYPE_ULEQUAD:
223: re = &ml->root->format_quad;
224: break;
225: case MAGIC_TYPE_FLOAT:
226: case MAGIC_TYPE_BEFLOAT:
227: case MAGIC_TYPE_LEFLOAT:
228: case MAGIC_TYPE_DOUBLE:
229: case MAGIC_TYPE_BEDOUBLE:
230: case MAGIC_TYPE_LEDOUBLE:
231: re = &ml->root->format_float;
232: break;
233: case MAGIC_TYPE_DATE:
234: case MAGIC_TYPE_LDATE:
235: case MAGIC_TYPE_UDATE:
236: case MAGIC_TYPE_ULDATE:
237: case MAGIC_TYPE_BEDATE:
238: case MAGIC_TYPE_BELDATE:
239: case MAGIC_TYPE_UBEDATE:
240: case MAGIC_TYPE_UBELDATE:
241: case MAGIC_TYPE_QDATE:
242: case MAGIC_TYPE_QLDATE:
243: case MAGIC_TYPE_UQDATE:
244: case MAGIC_TYPE_UQLDATE:
245: case MAGIC_TYPE_BEQDATE:
246: case MAGIC_TYPE_BEQLDATE:
247: case MAGIC_TYPE_UBEQDATE:
248: case MAGIC_TYPE_UBEQLDATE:
249: case MAGIC_TYPE_LEQDATE:
250: case MAGIC_TYPE_LEQLDATE:
251: case MAGIC_TYPE_ULEQDATE:
252: case MAGIC_TYPE_ULEQLDATE:
253: case MAGIC_TYPE_LEDATE:
254: case MAGIC_TYPE_LELDATE:
255: case MAGIC_TYPE_ULEDATE:
256: case MAGIC_TYPE_ULELDATE:
257: case MAGIC_TYPE_MEDATE:
258: case MAGIC_TYPE_MELDATE:
259: case MAGIC_TYPE_STRING:
260: case MAGIC_TYPE_PSTRING:
261: case MAGIC_TYPE_REGEX:
262: case MAGIC_TYPE_SEARCH:
263: re = &ml->root->format_string;
264: break;
265: }
266: }
267:
268: pmatch.rm_so = 0;
269: pmatch.rm_eo = fmtlen;
270: if (regexec(re, fmt, 1, &pmatch, REG_STARTEND) != 0) {
271: magic_warn(ml, "bad format for %s: %%%.*s", ml->type_string,
272: (int)fmtlen, fmt);
273: return (-1);
274: }
275:
276: return (0);
277: }
278:
279: static u_int
280: magic_get_strength(struct magic_line *ml)
281: {
282: int n;
283: size_t size;
284:
1.13 nicm 285: if (ml->type == MAGIC_TYPE_NONE)
286: return (0);
287:
1.18 nicm 288: if (ml->test_not || ml->test_operator == 'x') {
289: n = 1;
290: goto skip;
291: }
1.1 nicm 292:
1.5 nicm 293: n = 2 * MAGIC_STRENGTH_MULTIPLIER;
1.1 nicm 294: switch (ml->type) {
295: case MAGIC_TYPE_NONE:
296: case MAGIC_TYPE_DEFAULT:
297: return (0);
298: case MAGIC_TYPE_BYTE:
299: case MAGIC_TYPE_UBYTE:
300: n += 1 * MAGIC_STRENGTH_MULTIPLIER;
301: break;
302: case MAGIC_TYPE_SHORT:
303: case MAGIC_TYPE_USHORT:
304: case MAGIC_TYPE_BESHORT:
305: case MAGIC_TYPE_UBESHORT:
306: case MAGIC_TYPE_LESHORT:
307: case MAGIC_TYPE_ULESHORT:
308: n += 2 * MAGIC_STRENGTH_MULTIPLIER;
309: break;
310: case MAGIC_TYPE_LONG:
311: case MAGIC_TYPE_ULONG:
312: case MAGIC_TYPE_FLOAT:
313: case MAGIC_TYPE_DATE:
314: case MAGIC_TYPE_LDATE:
315: case MAGIC_TYPE_UDATE:
316: case MAGIC_TYPE_ULDATE:
317: case MAGIC_TYPE_BELONG:
318: case MAGIC_TYPE_UBELONG:
319: case MAGIC_TYPE_BEFLOAT:
320: case MAGIC_TYPE_BEDATE:
321: case MAGIC_TYPE_BELDATE:
322: case MAGIC_TYPE_UBEDATE:
323: case MAGIC_TYPE_UBELDATE:
324: n += 4 * MAGIC_STRENGTH_MULTIPLIER;
325: break;
326: case MAGIC_TYPE_QUAD:
327: case MAGIC_TYPE_UQUAD:
328: case MAGIC_TYPE_DOUBLE:
329: case MAGIC_TYPE_QDATE:
330: case MAGIC_TYPE_QLDATE:
331: case MAGIC_TYPE_UQDATE:
332: case MAGIC_TYPE_UQLDATE:
333: case MAGIC_TYPE_BEQUAD:
334: case MAGIC_TYPE_UBEQUAD:
335: case MAGIC_TYPE_BEDOUBLE:
336: case MAGIC_TYPE_BEQDATE:
337: case MAGIC_TYPE_BEQLDATE:
338: case MAGIC_TYPE_UBEQDATE:
339: case MAGIC_TYPE_UBEQLDATE:
340: case MAGIC_TYPE_LEQUAD:
341: case MAGIC_TYPE_ULEQUAD:
342: case MAGIC_TYPE_LEDOUBLE:
343: case MAGIC_TYPE_LEQDATE:
344: case MAGIC_TYPE_LEQLDATE:
345: case MAGIC_TYPE_ULEQDATE:
346: case MAGIC_TYPE_ULEQLDATE:
347: case MAGIC_TYPE_LELONG:
348: case MAGIC_TYPE_ULELONG:
349: case MAGIC_TYPE_LEFLOAT:
350: case MAGIC_TYPE_LEDATE:
351: case MAGIC_TYPE_LELDATE:
352: case MAGIC_TYPE_ULEDATE:
353: case MAGIC_TYPE_ULELDATE:
354: case MAGIC_TYPE_MELONG:
355: case MAGIC_TYPE_MEDATE:
356: case MAGIC_TYPE_MELDATE:
357: n += 8 * MAGIC_STRENGTH_MULTIPLIER;
358: break;
359: case MAGIC_TYPE_STRING:
360: case MAGIC_TYPE_PSTRING:
361: n += ml->test_string_size * MAGIC_STRENGTH_MULTIPLIER;
362: break;
363: case MAGIC_TYPE_BESTRING16:
364: case MAGIC_TYPE_LESTRING16:
365: n += ml->test_string_size * MAGIC_STRENGTH_MULTIPLIER / 2;
366: break;
367: case MAGIC_TYPE_REGEX:
368: case MAGIC_TYPE_SEARCH:
369: size = MAGIC_STRENGTH_MULTIPLIER / ml->test_string_size;
370: if (size < 1)
371: size = 1;
372: n += ml->test_string_size * size;
373: break;
374: }
375: switch (ml->test_operator) {
376: case '=':
377: n += MAGIC_STRENGTH_MULTIPLIER;
378: break;
379: case '<':
380: case '>':
381: case '[':
382: case ']':
383: n -= 2 * MAGIC_STRENGTH_MULTIPLIER;
384: break;
385: case '^':
386: case '&':
387: n -= MAGIC_STRENGTH_MULTIPLIER;
388: break;
389: }
1.18 nicm 390:
391: skip:
392: switch (ml->strength_operator) {
393: case '+':
394: n += ml->strength_value;
395: break;
396: case '-':
397: n -= ml->strength_value;
398: break;
399: case '*':
400: n *= ml->strength_value;
401: break;
402: case '/':
403: n /= ml->strength_value;
404: break;
405: }
1.1 nicm 406: return (n <= 0 ? 1 : n);
407: }
408:
409: static int
410: magic_get_string(char **line, char *out, size_t *outlen)
411: {
412: char *start, *cp, c;
413: int d0, d1, d2;
414:
415: start = out;
416: for (cp = *line; *cp != '\0' && !isspace((u_char)*cp); cp++) {
417: if (*cp != '\\') {
418: *out++ = *cp;
419: continue;
420: }
421:
422: switch (c = *++cp) {
1.3 nicm 423: case '\0': /* end of line */
424: return (-1);
1.1 nicm 425: case ' ':
426: *out++ = ' ';
427: break;
428: case '0':
429: case '1':
430: case '2':
431: case '3':
432: case '4':
433: case '5':
434: case '6':
435: case '7':
436: d0 = magic_odigit(cp[0]);
437: if (cp[0] != '\0')
438: d1 = magic_odigit(cp[1]);
439: else
440: d1 = -1;
441: if (cp[0] != '\0' && cp[1] != '\0')
442: d2 = magic_odigit(cp[2]);
443: else
444: d2 = -1;
445:
446: if (d0 != -1 && d1 != -1 && d2 != -1) {
447: *out = d2 | (d1 << 3) | (d0 << 6);
448: cp += 2;
449: } else if (d0 != -1 && d1 != -1) {
450: *out = d1 | (d0 << 3);
451: cp++;
452: } else if (d0 != -1)
453: *out = d0;
454: else
455: return (-1);
456: out++;
457: break;
458: case 'x':
459: d0 = magic_xdigit(cp[1]);
460: if (cp[1] != '\0')
461: d1 = magic_xdigit(cp[2]);
462: else
463: d1 = -1;
464:
465: if (d0 != -1 && d1 != -1) {
466: *out = d1 | (d0 << 4);
467: cp += 2;
468: } else if (d0 != -1) {
469: *out = d0;
470: cp++;
471: } else
472: return (-1);
473: out++;
474:
475: break;
476: case 'a':
477: *out++ = '\a';
478: break;
479: case 'b':
480: *out++ = '\b';
481: break;
482: case 't':
483: *out++ = '\t';
484: break;
485: case 'f':
486: *out++ = '\f';
487: break;
488: case 'n':
489: *out++ = '\n';
490: break;
491: case 'r':
492: *out++ = '\r';
493: break;
494: case '\\':
495: *out++ = '\\';
496: break;
497: case '\'':
498: *out++ = '\'';
499: break;
500: case '\"':
501: *out++ = '\"';
502: break;
503: default:
504: *out++ = c;
505: break;
506: }
507: }
508: *out = '\0';
509: *outlen = out - start;
510:
511: *line = cp;
512: return (0);
513: }
514:
515: static int
516: magic_parse_offset(struct magic_line *ml, char **line)
517: {
518: char *copy, *s, *cp, *endptr;
519:
520: while (isspace((u_char)**line))
521: (*line)++;
522: copy = s = cp = xmalloc(strlen(*line) + 1);
523: while (**line != '\0' && !isspace((u_char)**line))
524: *cp++ = *(*line)++;
525: *cp = '\0';
526:
527: ml->offset = 0;
528: ml->offset_relative = 0;
529:
530: ml->indirect_type = ' ';
531: ml->indirect_relative = 0;
532: ml->indirect_offset = 0;
533: ml->indirect_operator = ' ';
534: ml->indirect_operand = 0;
535:
536: if (*s == '&') {
537: ml->offset_relative = 1;
538: s++;
539: }
540:
541: if (*s != '(') {
542: endptr = magic_strtoll(s, &ml->offset);
543: if (endptr == NULL || *endptr != '\0') {
544: magic_warn(ml, "missing closing bracket");
545: goto fail;
546: }
547: if (ml->offset < 0 && !ml->offset_relative) {
548: magic_warn(ml, "negative absolute offset");
549: goto fail;
550: }
551: goto done;
552: }
553: s++;
554:
555: if (*s == '&') {
556: ml->indirect_relative = 1;
557: s++;
558: }
559:
560: endptr = magic_strtoll(s, &ml->indirect_offset);
561: if (endptr == NULL) {
1.8 nicm 562: magic_warn(ml, "can't parse offset: %s", s);
1.1 nicm 563: goto fail;
564: }
565: s = endptr;
566: if (*s == ')')
567: goto done;
568:
569: if (*s == '.') {
570: s++;
1.6 tobias 571: if (*s == '\0' || strchr("bslBSL", *s) == NULL) {
1.8 nicm 572: magic_warn(ml, "unknown offset type: %c", *s);
1.1 nicm 573: goto fail;
574: }
575: ml->indirect_type = *s;
576: s++;
577: if (*s == ')')
578: goto done;
579: }
580:
1.6 tobias 581: if (*s == '\0' || strchr("+-*", *s) == NULL) {
1.8 nicm 582: magic_warn(ml, "unknown offset operator: %c", *s);
1.1 nicm 583: goto fail;
584: }
585: ml->indirect_operator = *s;
586: s++;
587: if (*s == ')')
588: goto done;
589:
590: if (*s == '(') {
591: s++;
592: endptr = magic_strtoll(s, &ml->indirect_operand);
593: if (endptr == NULL || *endptr != ')') {
594: magic_warn(ml, "missing closing bracket");
595: goto fail;
596: }
597: if (*++endptr != ')') {
598: magic_warn(ml, "missing closing bracket");
599: goto fail;
600: }
601: } else {
602: endptr = magic_strtoll(s, &ml->indirect_operand);
603: if (endptr == NULL || *endptr != ')') {
604: magic_warn(ml, "missing closing bracket");
605: goto fail;
606: }
607: }
608:
609: done:
610: free(copy);
611: return (0);
612:
613: fail:
614: free(copy);
615: return (-1);
616: }
617:
618: static int
619: magic_parse_type(struct magic_line *ml, char **line)
620: {
621: char *copy, *s, *cp, *endptr;
622:
623: while (isspace((u_char)**line))
624: (*line)++;
625: copy = s = cp = xmalloc(strlen(*line) + 1);
626: while (**line != '\0' && !isspace((u_char)**line))
627: *cp++ = *(*line)++;
628: *cp = '\0';
629:
630: ml->type = MAGIC_TYPE_NONE;
631: ml->type_operator = ' ';
632: ml->type_operand = 0;
633:
1.16 nicm 634: if (strncmp(s, "string", (sizeof "string") - 1) == 0 ||
635: strncmp(s, "ustring", (sizeof "ustring") - 1) == 0) {
636: if (*s == 'u')
637: ml->type_string = xstrdup(s + 1);
638: else
639: ml->type_string = xstrdup(s);
1.1 nicm 640: ml->type = MAGIC_TYPE_STRING;
641: magic_mark_text(ml, 0);
642: goto done;
643: }
1.16 nicm 644: if (strncmp(s, "pstring", (sizeof "pstring") - 1) == 0 ||
645: strncmp(s, "upstring", (sizeof "upstring") - 1) == 0) {
646: if (*s == 'u')
647: ml->type_string = xstrdup(s + 1);
648: else
649: ml->type_string = xstrdup(s);
650: ml->type = MAGIC_TYPE_PSTRING;
651: magic_mark_text(ml, 0);
652: goto done;
653: }
654: if (strncmp(s, "search", (sizeof "search") - 1) == 0 ||
655: strncmp(s, "usearch", (sizeof "usearch") - 1) == 0) {
656: if (*s == 'u')
657: ml->type_string = xstrdup(s + 1);
658: else
659: ml->type_string = xstrdup(s);
1.1 nicm 660: ml->type = MAGIC_TYPE_SEARCH;
661: goto done;
662: }
1.16 nicm 663: if (strncmp(s, "regex", (sizeof "regex") - 1) == 0 ||
664: strncmp(s, "uregex", (sizeof "uregex") - 1) == 0) {
665: if (*s == 'u')
666: ml->type_string = xstrdup(s + 1);
667: else
668: ml->type_string = xstrdup(s);
1.1 nicm 669: ml->type = MAGIC_TYPE_REGEX;
670: goto done;
671: }
1.16 nicm 672: ml->type_string = xstrdup(s);
1.1 nicm 673:
1.12 nicm 674: cp = &s[strcspn(s, "+-&/%*")];
1.1 nicm 675: if (*cp != '\0') {
676: ml->type_operator = *cp;
677: endptr = magic_strtoull(cp + 1, &ml->type_operand);
678: if (endptr == NULL || *endptr != '\0') {
1.8 nicm 679: magic_warn(ml, "can't parse operand: %s", cp + 1);
1.1 nicm 680: goto fail;
681: }
682: *cp = '\0';
683: }
684:
685: if (strcmp(s, "byte") == 0)
686: ml->type = MAGIC_TYPE_BYTE;
687: else if (strcmp(s, "short") == 0)
688: ml->type = MAGIC_TYPE_SHORT;
689: else if (strcmp(s, "long") == 0)
690: ml->type = MAGIC_TYPE_LONG;
691: else if (strcmp(s, "quad") == 0)
692: ml->type = MAGIC_TYPE_QUAD;
693: else if (strcmp(s, "ubyte") == 0)
694: ml->type = MAGIC_TYPE_UBYTE;
695: else if (strcmp(s, "ushort") == 0)
696: ml->type = MAGIC_TYPE_USHORT;
697: else if (strcmp(s, "ulong") == 0)
698: ml->type = MAGIC_TYPE_ULONG;
699: else if (strcmp(s, "uquad") == 0)
700: ml->type = MAGIC_TYPE_UQUAD;
1.16 nicm 701: else if (strcmp(s, "float") == 0 || strcmp(s, "ufloat") == 0)
1.1 nicm 702: ml->type = MAGIC_TYPE_FLOAT;
1.16 nicm 703: else if (strcmp(s, "double") == 0 || strcmp(s, "udouble") == 0)
1.1 nicm 704: ml->type = MAGIC_TYPE_DOUBLE;
705: else if (strcmp(s, "date") == 0)
706: ml->type = MAGIC_TYPE_DATE;
707: else if (strcmp(s, "qdate") == 0)
708: ml->type = MAGIC_TYPE_QDATE;
709: else if (strcmp(s, "ldate") == 0)
710: ml->type = MAGIC_TYPE_LDATE;
711: else if (strcmp(s, "qldate") == 0)
712: ml->type = MAGIC_TYPE_QLDATE;
713: else if (strcmp(s, "udate") == 0)
714: ml->type = MAGIC_TYPE_UDATE;
715: else if (strcmp(s, "uqdate") == 0)
716: ml->type = MAGIC_TYPE_UQDATE;
717: else if (strcmp(s, "uldate") == 0)
718: ml->type = MAGIC_TYPE_ULDATE;
719: else if (strcmp(s, "uqldate") == 0)
720: ml->type = MAGIC_TYPE_UQLDATE;
721: else if (strcmp(s, "beshort") == 0)
722: ml->type = MAGIC_TYPE_BESHORT;
723: else if (strcmp(s, "belong") == 0)
724: ml->type = MAGIC_TYPE_BELONG;
725: else if (strcmp(s, "bequad") == 0)
726: ml->type = MAGIC_TYPE_BEQUAD;
727: else if (strcmp(s, "ubeshort") == 0)
728: ml->type = MAGIC_TYPE_UBESHORT;
729: else if (strcmp(s, "ubelong") == 0)
730: ml->type = MAGIC_TYPE_UBELONG;
731: else if (strcmp(s, "ubequad") == 0)
732: ml->type = MAGIC_TYPE_UBEQUAD;
1.16 nicm 733: else if (strcmp(s, "befloat") == 0 || strcmp(s, "ubefloat") == 0)
1.1 nicm 734: ml->type = MAGIC_TYPE_BEFLOAT;
1.16 nicm 735: else if (strcmp(s, "bedouble") == 0 || strcmp(s, "ubedouble") == 0)
1.1 nicm 736: ml->type = MAGIC_TYPE_BEDOUBLE;
737: else if (strcmp(s, "bedate") == 0)
738: ml->type = MAGIC_TYPE_BEDATE;
739: else if (strcmp(s, "beqdate") == 0)
740: ml->type = MAGIC_TYPE_BEQDATE;
741: else if (strcmp(s, "beldate") == 0)
742: ml->type = MAGIC_TYPE_BELDATE;
743: else if (strcmp(s, "beqldate") == 0)
744: ml->type = MAGIC_TYPE_BEQLDATE;
745: else if (strcmp(s, "ubedate") == 0)
746: ml->type = MAGIC_TYPE_UBEDATE;
747: else if (strcmp(s, "ubeqdate") == 0)
748: ml->type = MAGIC_TYPE_UBEQDATE;
749: else if (strcmp(s, "ubeldate") == 0)
750: ml->type = MAGIC_TYPE_UBELDATE;
751: else if (strcmp(s, "ubeqldate") == 0)
752: ml->type = MAGIC_TYPE_UBEQLDATE;
1.16 nicm 753: else if (strcmp(s, "bestring16") == 0 || strcmp(s, "ubestring16") == 0)
1.1 nicm 754: ml->type = MAGIC_TYPE_BESTRING16;
755: else if (strcmp(s, "leshort") == 0)
756: ml->type = MAGIC_TYPE_LESHORT;
757: else if (strcmp(s, "lelong") == 0)
758: ml->type = MAGIC_TYPE_LELONG;
759: else if (strcmp(s, "lequad") == 0)
760: ml->type = MAGIC_TYPE_LEQUAD;
761: else if (strcmp(s, "uleshort") == 0)
762: ml->type = MAGIC_TYPE_ULESHORT;
763: else if (strcmp(s, "ulelong") == 0)
764: ml->type = MAGIC_TYPE_ULELONG;
765: else if (strcmp(s, "ulequad") == 0)
766: ml->type = MAGIC_TYPE_ULEQUAD;
1.16 nicm 767: else if (strcmp(s, "lefloat") == 0 || strcmp(s, "ulefloat") == 0)
1.1 nicm 768: ml->type = MAGIC_TYPE_LEFLOAT;
1.16 nicm 769: else if (strcmp(s, "ledouble") == 0 || strcmp(s, "uledouble") == 0)
1.1 nicm 770: ml->type = MAGIC_TYPE_LEDOUBLE;
771: else if (strcmp(s, "ledate") == 0)
772: ml->type = MAGIC_TYPE_LEDATE;
773: else if (strcmp(s, "leqdate") == 0)
774: ml->type = MAGIC_TYPE_LEQDATE;
775: else if (strcmp(s, "leldate") == 0)
776: ml->type = MAGIC_TYPE_LELDATE;
777: else if (strcmp(s, "leqldate") == 0)
778: ml->type = MAGIC_TYPE_LEQLDATE;
779: else if (strcmp(s, "uledate") == 0)
780: ml->type = MAGIC_TYPE_ULEDATE;
781: else if (strcmp(s, "uleqdate") == 0)
782: ml->type = MAGIC_TYPE_ULEQDATE;
783: else if (strcmp(s, "uleldate") == 0)
784: ml->type = MAGIC_TYPE_ULELDATE;
785: else if (strcmp(s, "uleqldate") == 0)
786: ml->type = MAGIC_TYPE_ULEQLDATE;
1.16 nicm 787: else if (strcmp(s, "lestring16") == 0 || strcmp(s, "ulestring16") == 0)
1.1 nicm 788: ml->type = MAGIC_TYPE_LESTRING16;
1.16 nicm 789: else if (strcmp(s, "melong") == 0 || strcmp(s, "umelong") == 0)
1.1 nicm 790: ml->type = MAGIC_TYPE_MELONG;
1.16 nicm 791: else if (strcmp(s, "medate") == 0 || strcmp(s, "umedate") == 0)
1.1 nicm 792: ml->type = MAGIC_TYPE_MEDATE;
1.16 nicm 793: else if (strcmp(s, "meldate") == 0 || strcmp(s, "umeldate") == 0)
1.1 nicm 794: ml->type = MAGIC_TYPE_MELDATE;
1.16 nicm 795: else if (strcmp(s, "default") == 0 || strcmp(s, "udefault") == 0)
1.1 nicm 796: ml->type = MAGIC_TYPE_DEFAULT;
797: else {
1.8 nicm 798: magic_warn(ml, "unknown type: %s", s);
1.1 nicm 799: goto fail;
800: }
801: magic_mark_text(ml, 0);
802:
803: done:
804: free(copy);
805: return (0);
806:
807: fail:
808: free(copy);
809: return (-1);
810: }
811:
812: static int
813: magic_parse_value(struct magic_line *ml, char **line)
814: {
815: char *copy, *s, *cp, *endptr;
816: size_t slen;
1.10 nicm 817: uint64_t u;
1.1 nicm 818:
819: while (isspace((u_char)**line))
820: (*line)++;
821:
822: ml->test_operator = '=';
823: ml->test_not = 0;
824: ml->test_string = NULL;
825: ml->test_string_size = 0;
826: ml->test_unsigned = 0;
827: ml->test_signed = 0;
828:
1.9 nicm 829: if (**line == '\0')
830: return (0);
831:
1.1 nicm 832: s = *line;
833: if (s[0] == 'x' && (s[1] == '\0' || isspace((u_char)s[1]))) {
834: (*line)++;
835: ml->test_operator = 'x';
836: return (0);
837: }
838:
839: if (**line == '!') {
840: ml->test_not = 1;
841: (*line)++;
842: }
843:
844: switch (ml->type) {
845: case MAGIC_TYPE_STRING:
846: case MAGIC_TYPE_PSTRING:
847: case MAGIC_TYPE_SEARCH:
848: if (**line == '>' || **line == '<' || **line == '=') {
849: ml->test_operator = **line;
850: (*line)++;
851: }
852: /* FALLTHROUGH */
853: case MAGIC_TYPE_REGEX:
854: copy = s = xmalloc(strlen(*line) + 1);
855: if (magic_get_string(line, s, &slen) != 0) {
856: magic_warn(ml, "can't parse string");
857: goto fail;
858: }
859: ml->test_string_size = slen;
860: ml->test_string = s;
861: return (0); /* do not free */
862: default:
863: break;
864: }
865:
1.9 nicm 866: while (isspace((u_char)**line))
867: (*line)++;
868: if ((*line)[0] == '<' && (*line)[1] == '=') {
869: ml->test_operator = '[';
870: (*line) += 2;
871: } else if ((*line)[0] == '>' && (*line)[1] == '=') {
872: ml->test_operator = ']';
1.1 nicm 873: (*line) += 2;
1.19 ! tobias 874: } else if (**line != '\0' && strchr("=<>&^", **line) != NULL) {
1.9 nicm 875: ml->test_operator = **line;
876: (*line)++;
1.1 nicm 877: }
1.9 nicm 878:
879: while (isspace((u_char)**line))
880: (*line)++;
881: copy = cp = xmalloc(strlen(*line) + 1);
1.1 nicm 882: while (**line != '\0' && !isspace((u_char)**line))
883: *cp++ = *(*line)++;
884: *cp = '\0';
885:
1.11 nicm 886: switch (ml->type) {
887: case MAGIC_TYPE_FLOAT:
888: case MAGIC_TYPE_DOUBLE:
889: case MAGIC_TYPE_BEFLOAT:
890: case MAGIC_TYPE_BEDOUBLE:
891: case MAGIC_TYPE_LEFLOAT:
892: case MAGIC_TYPE_LEDOUBLE:
893: errno = 0;
894: ml->test_double = strtod(copy, &endptr);
895: if (errno == ERANGE)
896: endptr = NULL;
897: break;
898: default:
899: if (*ml->type_string == 'u')
900: endptr = magic_strtoull(copy, &ml->test_unsigned);
901: else {
902: endptr = magic_strtoll(copy, &ml->test_signed);
903: if (endptr == NULL || *endptr != '\0') {
904: /*
905: * If we can't parse this as a signed number,
906: * try as unsigned instead.
907: */
908: endptr = magic_strtoull(copy, &u);
909: if (endptr != NULL && *endptr == '\0')
910: ml->test_signed = (int64_t)u;
911: }
1.10 nicm 912: }
1.11 nicm 913: break;
1.10 nicm 914: }
1.1 nicm 915: if (endptr == NULL || *endptr != '\0') {
1.9 nicm 916: magic_warn(ml, "can't parse number: %s", copy);
1.1 nicm 917: goto fail;
918: }
919:
920: free(copy);
921: return (0);
922:
923: fail:
924: free(copy);
925: return (-1);
926: }
927:
928: int
929: magic_compare(struct magic_line *ml1, struct magic_line *ml2)
930: {
931: if (ml1->strength < ml2->strength)
932: return (1);
933: if (ml1->strength > ml2->strength)
934: return (-1);
935:
936: /*
937: * The original file depends on the (undefined!) qsort(3) behaviour
938: * when the strength is equal. This is impossible to reproduce with an
939: * RB tree so just use the line number and hope for the best.
940: */
941: if (ml1->line < ml2->line)
942: return (-1);
943: if (ml1->line > ml2->line)
944: return (1);
945:
946: return (0);
947: }
948: RB_GENERATE(magic_tree, magic_line, node, magic_compare);
949:
950: static void
1.18 nicm 951: magic_adjust_strength(struct magic *m, u_int at, struct magic_line *ml,
952: char *line)
953: {
954: char *cp, *s;
955: int64_t value;
956:
957: cp = line + (sizeof "!:strength") - 1;
958: while (isspace((u_char)*cp))
959: cp++;
960: s = cp;
961:
962: cp = strchr(s, '#');
963: if (cp != NULL)
964: *cp = '\0';
965: cp = s;
966:
1.19 ! tobias 967: if (*s == '\0' || strchr("+-*/", *s) == NULL) {
1.18 nicm 968: magic_warnm(m, at, "invalid strength operator: %s", s);
969: return;
970: }
971: ml->strength_operator = *cp++;
972:
973: while (isspace((u_char)*cp))
974: cp++;
975: cp = magic_strtoll(cp, &value);
976: while (cp != NULL && isspace((u_char)*cp))
977: cp++;
978: if (cp == NULL || *cp != '\0' || value < 0 || value > 255) {
979: magic_warnm(m, at, "invalid strength value: %s", s);
980: return;
981: }
982: ml->strength_value = value;
983: }
984:
985: static void
1.1 nicm 986: magic_set_mimetype(struct magic *m, u_int at, struct magic_line *ml, char *line)
987: {
988: char *mimetype, *cp;
989:
990: mimetype = line + (sizeof "!:mime") - 1;
991: while (isspace((u_char)*mimetype))
992: mimetype++;
993:
994: cp = strchr(mimetype, '#');
995: if (cp != NULL)
996: *cp = '\0';
997:
998: if (*mimetype != '\0') {
999: cp = mimetype + strlen(mimetype) - 1;
1000: while (cp != mimetype && isspace((u_char)*cp))
1001: *cp-- = '\0';
1002: }
1003:
1004: cp = mimetype;
1005: while (*cp != '\0') {
1006: if (!isalnum((u_char)*cp) && strchr("/-.+", *cp) == NULL)
1007: break;
1008: cp++;
1009: }
1010: if (*mimetype == '\0' || *cp != '\0') {
1.7 nicm 1011: magic_warnm(m, at, "invalid MIME type: %s", mimetype);
1.1 nicm 1012: return;
1013: }
1014: if (ml == NULL) {
1.7 nicm 1015: magic_warnm(m, at, "stray MIME type: %s", mimetype);
1.1 nicm 1016: return;
1017: }
1018: ml->mimetype = xstrdup(mimetype);
1019: }
1020:
1021: struct magic *
1022: magic_load(FILE *f, const char *path, int warnings)
1023: {
1024: struct magic *m;
1025: struct magic_line *ml = NULL, *parent, *parent0;
1026: char *line, *tmp;
1027: size_t size;
1028: u_int at, level, n, i;
1029:
1030: m = xcalloc(1, sizeof *m);
1031: m->path = xstrdup(path);
1032: m->warnings = warnings;
1033: RB_INIT(&m->tree);
1034:
1035: parent = NULL;
1036: parent0 = NULL;
1037: level = 0;
1038:
1039: at = 0;
1040: tmp = NULL;
1041: while ((line = fgetln(f, &size))) {
1042: if (line[size - 1] == '\n')
1043: line[size - 1] = '\0';
1044: else {
1045: tmp = xmalloc(size + 1);
1046: memcpy(tmp, line, size);
1047: tmp[size] = '\0';
1048: line = tmp;
1049: }
1050: at++;
1051:
1052: while (isspace((u_char)*line))
1053: line++;
1054: if (*line == '\0' || *line == '#')
1055: continue;
1056:
1.15 nicm 1057: if (strncmp (line, "!:mime", 6) == 0) {
1.1 nicm 1058: magic_set_mimetype(m, at, ml, line);
1.18 nicm 1059: continue;
1060: }
1061: if (strncmp (line, "!:strength", 10) == 0) {
1062: magic_adjust_strength(m, at, ml, line);
1.15 nicm 1063: continue;
1064: }
1065: if (strncmp (line, "!:", 2) == 0) {
1066: for (i = 0; i < 64 && line[i] != '\0'; i++) {
1067: if (isspace((u_char)line[i]))
1068: break;
1069: }
1070: magic_warnm(m, at, "%.*s not supported", i, line);
1.1 nicm 1071: continue;
1072: }
1073:
1074: n = 0;
1075: for (; *line == '>'; line++)
1076: n++;
1077:
1078: ml = xcalloc(1, sizeof *ml);
1079: ml->root = m;
1080: ml->line = at;
1081: ml->type = MAGIC_TYPE_NONE;
1082: TAILQ_INIT(&ml->children);
1083: ml->text = 1;
1084:
1.13 nicm 1085: /*
1086: * At this point n is the level we want, level is the current
1087: * level. parent0 is the last line at the same level and parent
1088: * is the last line at the previous level.
1089: */
1.1 nicm 1090: if (n == level + 1) {
1091: parent = parent0;
1092: } else if (n < level) {
1093: for (i = n; i < level && parent != NULL; i++)
1094: parent = parent->parent;
1095: } else if (n != level) {
1096: magic_warn(ml, "level skipped (%u->%u)", level, n);
1097: free(ml);
1098: continue;
1099: }
1100: ml->parent = parent;
1101: level = n;
1102:
1103: if (magic_parse_offset(ml, &line) != 0 ||
1104: magic_parse_type(ml, &line) != 0 ||
1105: magic_parse_value(ml, &line) != 0 ||
1106: magic_set_result(ml, line) != 0) {
1.13 nicm 1107: /*
1108: * An invalid line still needs to appear in the tree in
1109: * case it has any children.
1110: */
1111: ml->type = MAGIC_TYPE_NONE;
1.1 nicm 1112: }
1113:
1114: ml->strength = magic_get_strength(ml);
1115: if (ml->parent == NULL)
1116: RB_INSERT(magic_tree, &m->tree, ml);
1117: else
1118: TAILQ_INSERT_TAIL(&ml->parent->children, ml, entry);
1119: parent0 = ml;
1120: }
1121: free(tmp);
1122:
1123: fclose(f);
1124: return (m);
1125: }