xref: /freebsd/contrib/file/src/file.h (revision 898496ee)
1b6cee71dSXin LI /*
2b6cee71dSXin LI  * Copyright (c) Ian F. Darwin 1986-1995.
3b6cee71dSXin LI  * Software written by Ian F. Darwin and others;
4b6cee71dSXin LI  * maintained 1995-present by Christos Zoulas and others.
5b6cee71dSXin LI  *
6b6cee71dSXin LI  * Redistribution and use in source and binary forms, with or without
7b6cee71dSXin LI  * modification, are permitted provided that the following conditions
8b6cee71dSXin LI  * are met:
9b6cee71dSXin LI  * 1. Redistributions of source code must retain the above copyright
10b6cee71dSXin LI  *    notice immediately at the beginning of the file, without modification,
11b6cee71dSXin LI  *    this list of conditions, and the following disclaimer.
12b6cee71dSXin LI  * 2. Redistributions in binary form must reproduce the above copyright
13b6cee71dSXin LI  *    notice, this list of conditions and the following disclaimer in the
14b6cee71dSXin LI  *    documentation and/or other materials provided with the distribution.
15b6cee71dSXin LI  *
16b6cee71dSXin LI  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
17b6cee71dSXin LI  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
18b6cee71dSXin LI  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
19b6cee71dSXin LI  * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE FOR
20b6cee71dSXin LI  * ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
21b6cee71dSXin LI  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
22b6cee71dSXin LI  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
23b6cee71dSXin LI  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
24b6cee71dSXin LI  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
25b6cee71dSXin LI  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
26b6cee71dSXin LI  * SUCH DAMAGE.
27b6cee71dSXin LI  */
28b6cee71dSXin LI /*
29b6cee71dSXin LI  * file.h - definitions for file(1) program
30898496eeSXin LI  * @(#)$File: file.h,v 1.247 2023/07/27 19:40:22 christos Exp $
31b6cee71dSXin LI  */
32b6cee71dSXin LI 
33b6cee71dSXin LI #ifndef __file_h__
34b6cee71dSXin LI #define __file_h__
35b6cee71dSXin LI 
36b6cee71dSXin LI #ifdef HAVE_CONFIG_H
37b6cee71dSXin LI #include <config.h>
38b6cee71dSXin LI #endif
392726a701SXin LI 
4040427ccaSGordon Tetlow #ifdef HAVE_STDINT_H
412726a701SXin LI #include <stdint.h>
422726a701SXin LI #endif
432726a701SXin LI 
442726a701SXin LI #ifdef HAVE_INTTYPES_H
452726a701SXin LI #include <inttypes.h>
462726a701SXin LI #endif
472726a701SXin LI 
4840427ccaSGordon Tetlow #ifndef __STDC_LIMIT_MACROS
4940427ccaSGordon Tetlow #define __STDC_LIMIT_MACROS
5040427ccaSGordon Tetlow #endif
5148c779cdSXin LI #ifndef __STDC_FORMAT_MACROS
5248c779cdSXin LI #define __STDC_FORMAT_MACROS
5348c779cdSXin LI #endif
54b6cee71dSXin LI 
552726a701SXin LI #ifdef _WIN32
562726a701SXin LI # ifdef PRIu32
572726a701SXin LI #  ifdef _WIN64
582726a701SXin LI #   define SIZE_T_FORMAT PRIu64
592726a701SXin LI #  else
602726a701SXin LI #   define SIZE_T_FORMAT PRIu32
612726a701SXin LI #  endif
622726a701SXin LI #  define INT64_T_FORMAT PRIi64
632726a701SXin LI #  define INTMAX_T_FORMAT PRIiMAX
642726a701SXin LI # else
65b6cee71dSXin LI #  ifdef _WIN64
66b6cee71dSXin LI #   define SIZE_T_FORMAT "I64"
67b6cee71dSXin LI #  else
68b6cee71dSXin LI #   define SIZE_T_FORMAT ""
69b6cee71dSXin LI #  endif
70b6cee71dSXin LI #  define INT64_T_FORMAT "I64"
719ce06829SXin LI #  define INTMAX_T_FORMAT "I64"
722726a701SXin LI # endif
73b6cee71dSXin LI #else
74b6cee71dSXin LI # define SIZE_T_FORMAT "z"
75b6cee71dSXin LI # define INT64_T_FORMAT "ll"
769ce06829SXin LI # define INTMAX_T_FORMAT "j"
77b6cee71dSXin LI #endif
78b6cee71dSXin LI 
79b6cee71dSXin LI #include <stdio.h>	/* Include that here, to make sure __P gets defined */
80b6cee71dSXin LI #include <errno.h>
81b6cee71dSXin LI #include <fcntl.h>	/* For open and flags */
82b6cee71dSXin LI #include <regex.h>
83b6cee71dSXin LI #include <time.h>
84b6cee71dSXin LI #include <sys/types.h>
85c2931133SXin LI #ifndef WIN32
86b6cee71dSXin LI #include <sys/param.h>
87c2931133SXin LI #endif
88b6cee71dSXin LI /* Do this here and now, because struct stat gets re-defined on solaris */
89b6cee71dSXin LI #include <sys/stat.h>
90b6cee71dSXin LI #include <stdarg.h>
91a4d6d3b8SXin LI #include <locale.h>
92a4d6d3b8SXin LI #if defined(HAVE_XLOCALE_H)
93a4d6d3b8SXin LI #include <xlocale.h>
94a4d6d3b8SXin LI #endif
95b6cee71dSXin LI 
96b6cee71dSXin LI #define ENABLE_CONDITIONALS
97b6cee71dSXin LI 
98b6cee71dSXin LI #ifndef MAGIC
99b6cee71dSXin LI #define MAGIC "/etc/magic"
100b6cee71dSXin LI #endif
101b6cee71dSXin LI 
102b6cee71dSXin LI #if defined(__EMX__) || defined (WIN32)
103b6cee71dSXin LI #define PATHSEP	';'
104b6cee71dSXin LI #else
105b6cee71dSXin LI #define PATHSEP	':'
106b6cee71dSXin LI #endif
107b6cee71dSXin LI 
108898496eeSXin LI #define file_private static
109b6cee71dSXin LI 
110b6cee71dSXin LI #if HAVE_VISIBILITY && !defined(WIN32)
111898496eeSXin LI #define file_public  __attribute__ ((__visibility__("default")))
112898496eeSXin LI #ifndef file_protected
113898496eeSXin LI #define file_protected __attribute__ ((__visibility__("hidden")))
114b6cee71dSXin LI #endif
115b6cee71dSXin LI #else
116898496eeSXin LI #define file_public
117898496eeSXin LI #ifndef file_protected
118898496eeSXin LI #define file_protected
119b6cee71dSXin LI #endif
120b6cee71dSXin LI #endif
121b6cee71dSXin LI 
122b6cee71dSXin LI #ifndef __arraycount
123b6cee71dSXin LI #define __arraycount(a) (sizeof(a) / sizeof(a[0]))
124b6cee71dSXin LI #endif
125b6cee71dSXin LI 
126b6cee71dSXin LI #ifndef __GNUC_PREREQ__
127b6cee71dSXin LI #ifdef __GNUC__
128b6cee71dSXin LI #define	__GNUC_PREREQ__(x, y)						\
129b6cee71dSXin LI 	((__GNUC__ == (x) && __GNUC_MINOR__ >= (y)) ||			\
130b6cee71dSXin LI 	 (__GNUC__ > (x)))
131b6cee71dSXin LI #else
132b6cee71dSXin LI #define	__GNUC_PREREQ__(x, y)	0
133b6cee71dSXin LI #endif
134b6cee71dSXin LI #endif
135b6cee71dSXin LI 
136b6cee71dSXin LI #ifndef __GNUC__
137b6cee71dSXin LI #ifndef __attribute__
138b6cee71dSXin LI #define __attribute__(a)
139b6cee71dSXin LI #endif
140b6cee71dSXin LI #endif
141b6cee71dSXin LI 
142b6cee71dSXin LI #ifndef MIN
143b6cee71dSXin LI #define	MIN(a,b)	(((a) < (b)) ? (a) : (b))
144b6cee71dSXin LI #endif
145b6cee71dSXin LI 
146b6cee71dSXin LI #ifndef MAX
147b6cee71dSXin LI #define	MAX(a,b)	(((a) > (b)) ? (a) : (b))
148b6cee71dSXin LI #endif
149b6cee71dSXin LI 
15043a5ec4eSXin LI #ifndef O_CLOEXEC
15143a5ec4eSXin LI # define O_CLOEXEC 0
15243a5ec4eSXin LI #endif
15343a5ec4eSXin LI 
15443a5ec4eSXin LI #ifndef FD_CLOEXEC
15543a5ec4eSXin LI # define FD_CLOEXEC 1
15643a5ec4eSXin LI #endif
15743a5ec4eSXin LI 
158898496eeSXin LI 
159898496eeSXin LI /*
160898496eeSXin LI  * Dec 31, 23:59:59 9999
161898496eeSXin LI  * we need to make sure that we don't exceed 9999 because some libc
162898496eeSXin LI  * implementations like muslc crash otherwise
163898496eeSXin LI  */
164898496eeSXin LI #define	MAX_CTIME	CAST(time_t, 0x3afff487cfULL)
165898496eeSXin LI 
1662726a701SXin LI #define FILE_BADSIZE CAST(size_t, ~0ul)
167b6cee71dSXin LI #define MAXDESC	64		/* max len of text description/MIME type */
168b6cee71dSXin LI #define MAXMIME	80		/* max len of text MIME type */
1692726a701SXin LI #define MAXstring 128		/* max len of "string" types */
170b6cee71dSXin LI 
171b6cee71dSXin LI #define MAGICNO		0xF11E041C
172a2dfb722SXin LI #define VERSIONNO	18
1732726a701SXin LI #define FILE_MAGICSIZE	376
1742726a701SXin LI 
1752726a701SXin LI #define FILE_GUID_SIZE	sizeof("XXXXXXXX-XXXX-XXXX-XXXX-XXXXXXXXXXXX")
176b6cee71dSXin LI 
177b6cee71dSXin LI #define	FILE_LOAD	0
178b6cee71dSXin LI #define FILE_CHECK	1
179b6cee71dSXin LI #define FILE_COMPILE	2
180b6cee71dSXin LI #define FILE_LIST	3
181b6cee71dSXin LI 
182a4d6d3b8SXin LI typedef regex_t file_regex_t;
183a4d6d3b8SXin LI 
18458a0f0d0SEitan Adler struct buffer {
18558a0f0d0SEitan Adler 	int fd;
18658a0f0d0SEitan Adler 	struct stat st;
18758a0f0d0SEitan Adler 	const void *fbuf;
18858a0f0d0SEitan Adler 	size_t flen;
18958a0f0d0SEitan Adler 	off_t eoff;
19058a0f0d0SEitan Adler 	void *ebuf;
19158a0f0d0SEitan Adler 	size_t elen;
19258a0f0d0SEitan Adler };
19358a0f0d0SEitan Adler 
194b6cee71dSXin LI union VALUETYPE {
195b6cee71dSXin LI 	uint8_t b;
196b6cee71dSXin LI 	uint16_t h;
197b6cee71dSXin LI 	uint32_t l;
198b6cee71dSXin LI 	uint64_t q;
199b6cee71dSXin LI 	uint8_t hs[2];	/* 2 bytes of a fixed-endian "short" */
200b6cee71dSXin LI 	uint8_t hl[4];	/* 4 bytes of a fixed-endian "long" */
201b6cee71dSXin LI 	uint8_t hq[8];	/* 8 bytes of a fixed-endian "quad" */
202b6cee71dSXin LI 	char s[MAXstring];	/* the search string or regex pattern */
203b6cee71dSXin LI 	unsigned char us[MAXstring];
2042726a701SXin LI 	uint64_t guid[2];
205b6cee71dSXin LI 	float f;
206b6cee71dSXin LI 	double d;
207b6cee71dSXin LI };
208b6cee71dSXin LI 
209b6cee71dSXin LI struct magic {
210b6cee71dSXin LI 	/* Word 1 */
211b6cee71dSXin LI 	uint16_t cont_level;	/* level of ">" */
212b6cee71dSXin LI 	uint8_t flag;
213b6cee71dSXin LI #define INDIR		0x01	/* if '(...)' appears */
214b6cee71dSXin LI #define OFFADD		0x02	/* if '>&' or '>...(&' appears */
215b6cee71dSXin LI #define INDIROFFADD	0x04	/* if '>&(' appears */
216b6cee71dSXin LI #define UNSIGNED	0x08	/* comparison is unsigned */
217b6cee71dSXin LI #define NOSPACE		0x10	/* suppress space character before output */
218b6cee71dSXin LI #define BINTEST		0x20	/* test is for a binary type (set only
219b6cee71dSXin LI 				   for top-level tests) */
220b6cee71dSXin LI #define TEXTTEST	0x40	/* for passing to file_softmagic */
2212726a701SXin LI #define OFFNEGATIVE	0x80	/* relative to the end of file */
222b6cee71dSXin LI 
223b6cee71dSXin LI 	uint8_t factor;
224b6cee71dSXin LI 
225b6cee71dSXin LI 	/* Word 2 */
226b6cee71dSXin LI 	uint8_t reln;		/* relation (0=eq, '>'=gt, etc) */
227b6cee71dSXin LI 	uint8_t vallen;		/* length of string value, if any */
228b6cee71dSXin LI 	uint8_t type;		/* comparison type (FILE_*) */
229b6cee71dSXin LI 	uint8_t in_type;	/* type of indirection */
230b6cee71dSXin LI #define 			FILE_INVALID		0
231b6cee71dSXin LI #define 			FILE_BYTE		1
232b6cee71dSXin LI #define				FILE_SHORT		2
233b6cee71dSXin LI #define				FILE_DEFAULT		3
234b6cee71dSXin LI #define				FILE_LONG		4
235b6cee71dSXin LI #define				FILE_STRING		5
236b6cee71dSXin LI #define				FILE_DATE		6
237b6cee71dSXin LI #define				FILE_BESHORT		7
238b6cee71dSXin LI #define				FILE_BELONG		8
239b6cee71dSXin LI #define				FILE_BEDATE		9
240b6cee71dSXin LI #define				FILE_LESHORT		10
241b6cee71dSXin LI #define				FILE_LELONG		11
242b6cee71dSXin LI #define				FILE_LEDATE		12
243b6cee71dSXin LI #define				FILE_PSTRING		13
244b6cee71dSXin LI #define				FILE_LDATE		14
245b6cee71dSXin LI #define				FILE_BELDATE		15
246b6cee71dSXin LI #define				FILE_LELDATE		16
247b6cee71dSXin LI #define				FILE_REGEX		17
248b6cee71dSXin LI #define				FILE_BESTRING16		18
249b6cee71dSXin LI #define				FILE_LESTRING16		19
250b6cee71dSXin LI #define				FILE_SEARCH		20
251b6cee71dSXin LI #define				FILE_MEDATE		21
252b6cee71dSXin LI #define				FILE_MELDATE		22
253b6cee71dSXin LI #define				FILE_MELONG		23
254b6cee71dSXin LI #define				FILE_QUAD		24
255b6cee71dSXin LI #define				FILE_LEQUAD		25
256b6cee71dSXin LI #define				FILE_BEQUAD		26
257b6cee71dSXin LI #define				FILE_QDATE		27
258b6cee71dSXin LI #define				FILE_LEQDATE		28
259b6cee71dSXin LI #define				FILE_BEQDATE		29
260b6cee71dSXin LI #define				FILE_QLDATE		30
261b6cee71dSXin LI #define				FILE_LEQLDATE		31
262b6cee71dSXin LI #define				FILE_BEQLDATE		32
263b6cee71dSXin LI #define				FILE_FLOAT		33
264b6cee71dSXin LI #define				FILE_BEFLOAT		34
265b6cee71dSXin LI #define				FILE_LEFLOAT		35
266b6cee71dSXin LI #define				FILE_DOUBLE		36
267b6cee71dSXin LI #define				FILE_BEDOUBLE		37
268b6cee71dSXin LI #define				FILE_LEDOUBLE		38
269b6cee71dSXin LI #define				FILE_BEID3		39
270b6cee71dSXin LI #define				FILE_LEID3		40
271b6cee71dSXin LI #define				FILE_INDIRECT		41
272b6cee71dSXin LI #define				FILE_QWDATE		42
273b6cee71dSXin LI #define				FILE_LEQWDATE		43
274b6cee71dSXin LI #define				FILE_BEQWDATE		44
275b6cee71dSXin LI #define				FILE_NAME		45
276b6cee71dSXin LI #define				FILE_USE		46
277b6cee71dSXin LI #define				FILE_CLEAR		47
2783e41d09dSXin LI #define				FILE_DER		48
2792726a701SXin LI #define				FILE_GUID		49
2802726a701SXin LI #define				FILE_OFFSET		50
28143a5ec4eSXin LI #define				FILE_BEVARINT		51
28243a5ec4eSXin LI #define				FILE_LEVARINT		52
283a4d6d3b8SXin LI #define				FILE_MSDOSDATE		53
284a4d6d3b8SXin LI #define				FILE_LEMSDOSDATE	54
285a4d6d3b8SXin LI #define				FILE_BEMSDOSDATE	55
286a4d6d3b8SXin LI #define				FILE_MSDOSTIME		56
287a4d6d3b8SXin LI #define				FILE_LEMSDOSTIME	57
288a4d6d3b8SXin LI #define				FILE_BEMSDOSTIME	58
289a2dfb722SXin LI #define				FILE_OCTAL		59
290a2dfb722SXin LI #define				FILE_NAMES_SIZE		60 /* size of array to contain all names */
291b6cee71dSXin LI 
292b6cee71dSXin LI #define IS_STRING(t) \
293b6cee71dSXin LI 	((t) == FILE_STRING || \
294b6cee71dSXin LI 	 (t) == FILE_PSTRING || \
295b6cee71dSXin LI 	 (t) == FILE_BESTRING16 || \
296b6cee71dSXin LI 	 (t) == FILE_LESTRING16 || \
297b6cee71dSXin LI 	 (t) == FILE_REGEX || \
298b6cee71dSXin LI 	 (t) == FILE_SEARCH || \
2994460e5b0SXin LI 	 (t) == FILE_INDIRECT || \
300b6cee71dSXin LI 	 (t) == FILE_NAME || \
301a2dfb722SXin LI 	 (t) == FILE_USE || \
302a2dfb722SXin LI 	 (t) == FILE_OCTAL)
303b6cee71dSXin LI 
304b6cee71dSXin LI #define FILE_FMT_NONE 0
305b6cee71dSXin LI #define FILE_FMT_NUM  1 /* "cduxXi" */
306b6cee71dSXin LI #define FILE_FMT_STR  2 /* "s" */
307b6cee71dSXin LI #define FILE_FMT_QUAD 3 /* "ll" */
308b6cee71dSXin LI #define FILE_FMT_FLOAT 4 /* "eEfFgG" */
309b6cee71dSXin LI #define FILE_FMT_DOUBLE 5 /* "eEfFgG" */
310b6cee71dSXin LI 
311b6cee71dSXin LI 	/* Word 3 */
312b6cee71dSXin LI 	uint8_t in_op;		/* operator for indirection */
313b6cee71dSXin LI 	uint8_t mask_op;	/* operator for mask */
314b6cee71dSXin LI #ifdef ENABLE_CONDITIONALS
315b6cee71dSXin LI 	uint8_t cond;		/* conditional type */
316b6cee71dSXin LI #else
317b6cee71dSXin LI 	uint8_t dummy;
318b6cee71dSXin LI #endif
319b6cee71dSXin LI 	uint8_t factor_op;
320b6cee71dSXin LI #define		FILE_FACTOR_OP_PLUS	'+'
321b6cee71dSXin LI #define		FILE_FACTOR_OP_MINUS	'-'
322b6cee71dSXin LI #define		FILE_FACTOR_OP_TIMES	'*'
323b6cee71dSXin LI #define		FILE_FACTOR_OP_DIV	'/'
324b6cee71dSXin LI #define		FILE_FACTOR_OP_NONE	'\0'
325b6cee71dSXin LI 
326b6cee71dSXin LI #define				FILE_OPS	"&|^+-*/%"
327b6cee71dSXin LI #define				FILE_OPAND	0
328b6cee71dSXin LI #define				FILE_OPOR	1
329b6cee71dSXin LI #define				FILE_OPXOR	2
330b6cee71dSXin LI #define				FILE_OPADD	3
331b6cee71dSXin LI #define				FILE_OPMINUS	4
332b6cee71dSXin LI #define				FILE_OPMULTIPLY	5
333b6cee71dSXin LI #define				FILE_OPDIVIDE	6
334b6cee71dSXin LI #define				FILE_OPMODULO	7
335b6cee71dSXin LI #define				FILE_OPS_MASK	0x07 /* mask for above ops */
336b6cee71dSXin LI #define				FILE_UNUSED_1	0x08
337b6cee71dSXin LI #define				FILE_UNUSED_2	0x10
338a5d223e6SXin LI #define				FILE_OPSIGNED	0x20
339b6cee71dSXin LI #define				FILE_OPINVERSE	0x40
340b6cee71dSXin LI #define				FILE_OPINDIRECT	0x80
341b6cee71dSXin LI 
342b6cee71dSXin LI #ifdef ENABLE_CONDITIONALS
343b6cee71dSXin LI #define				COND_NONE	0
344b6cee71dSXin LI #define				COND_IF		1
345b6cee71dSXin LI #define				COND_ELIF	2
346b6cee71dSXin LI #define				COND_ELSE	3
347b6cee71dSXin LI #endif /* ENABLE_CONDITIONALS */
348b6cee71dSXin LI 
349b6cee71dSXin LI 	/* Word 4 */
35058a0f0d0SEitan Adler 	int32_t offset;		/* offset to magic number */
351b6cee71dSXin LI 	/* Word 5 */
352b6cee71dSXin LI 	int32_t in_offset;	/* offset from indirection */
353b6cee71dSXin LI 	/* Word 6 */
354b6cee71dSXin LI 	uint32_t lineno;	/* line number in magic file */
355b6cee71dSXin LI 	/* Word 7,8 */
356b6cee71dSXin LI 	union {
357b6cee71dSXin LI 		uint64_t _mask;	/* for use with numeric and date types */
358b6cee71dSXin LI 		struct {
359b6cee71dSXin LI 			uint32_t _count;	/* repeat/line count */
360b6cee71dSXin LI 			uint32_t _flags;	/* modifier flags */
361b6cee71dSXin LI 		} _s;		/* for use with string types */
362b6cee71dSXin LI 	} _u;
363b6cee71dSXin LI #define num_mask _u._mask
364b6cee71dSXin LI #define str_range _u._s._count
365b6cee71dSXin LI #define str_flags _u._s._flags
3669ce06829SXin LI 	/* Words 9-24 */
367b6cee71dSXin LI 	union VALUETYPE value;	/* either number or string */
3689ce06829SXin LI 	/* Words 25-40 */
369b6cee71dSXin LI 	char desc[MAXDESC];	/* description */
3709ce06829SXin LI 	/* Words 41-60 */
371b6cee71dSXin LI 	char mimetype[MAXMIME]; /* MIME type */
3729ce06829SXin LI 	/* Words 61-62 */
3735f0216bdSXin LI 	char apple[8];		/* APPLE CREATOR/TYPE */
3749ce06829SXin LI 	/* Words 63-78 */
3755f0216bdSXin LI 	char ext[64];		/* Popular extensions */
376b6cee71dSXin LI };
377b6cee71dSXin LI 
378b6cee71dSXin LI #define BIT(A)   (1 << (A))
379b6cee71dSXin LI #define STRING_COMPACT_WHITESPACE		BIT(0)
380b6cee71dSXin LI #define STRING_COMPACT_OPTIONAL_WHITESPACE	BIT(1)
381b6cee71dSXin LI #define STRING_IGNORE_LOWERCASE			BIT(2)
382b6cee71dSXin LI #define STRING_IGNORE_UPPERCASE			BIT(3)
383b6cee71dSXin LI #define REGEX_OFFSET_START			BIT(4)
384b6cee71dSXin LI #define STRING_TEXTTEST				BIT(5)
385b6cee71dSXin LI #define STRING_BINTEST				BIT(6)
386b6cee71dSXin LI #define PSTRING_1_BE				BIT(7)
387b6cee71dSXin LI #define PSTRING_1_LE				BIT(7)
388b6cee71dSXin LI #define PSTRING_2_BE				BIT(8)
389b6cee71dSXin LI #define PSTRING_2_LE				BIT(9)
390b6cee71dSXin LI #define PSTRING_4_BE				BIT(10)
391b6cee71dSXin LI #define PSTRING_4_LE				BIT(11)
392b6cee71dSXin LI #define REGEX_LINE_COUNT			BIT(11)
393b6cee71dSXin LI #define PSTRING_LEN	\
394b6cee71dSXin LI     (PSTRING_1_BE|PSTRING_2_LE|PSTRING_2_BE|PSTRING_4_LE|PSTRING_4_BE)
395b6cee71dSXin LI #define PSTRING_LENGTH_INCLUDES_ITSELF		BIT(12)
396b6cee71dSXin LI #define	STRING_TRIM				BIT(13)
39743a5ec4eSXin LI #define	STRING_FULL_WORD			BIT(14)
398b6cee71dSXin LI #define CHAR_COMPACT_WHITESPACE			'W'
399b6cee71dSXin LI #define CHAR_COMPACT_OPTIONAL_WHITESPACE	'w'
400b6cee71dSXin LI #define CHAR_IGNORE_LOWERCASE			'c'
401b6cee71dSXin LI #define CHAR_IGNORE_UPPERCASE			'C'
402b6cee71dSXin LI #define CHAR_REGEX_OFFSET_START			's'
403b6cee71dSXin LI #define CHAR_TEXTTEST				't'
404b6cee71dSXin LI #define	CHAR_TRIM				'T'
40543a5ec4eSXin LI #define	CHAR_FULL_WORD				'f'
406b6cee71dSXin LI #define CHAR_BINTEST				'b'
407b6cee71dSXin LI #define CHAR_PSTRING_1_BE			'B'
408b6cee71dSXin LI #define CHAR_PSTRING_1_LE			'B'
409b6cee71dSXin LI #define CHAR_PSTRING_2_BE			'H'
410b6cee71dSXin LI #define CHAR_PSTRING_2_LE			'h'
411b6cee71dSXin LI #define CHAR_PSTRING_4_BE			'L'
412b6cee71dSXin LI #define CHAR_PSTRING_4_LE			'l'
413b6cee71dSXin LI #define CHAR_PSTRING_LENGTH_INCLUDES_ITSELF     'J'
414b6cee71dSXin LI #define STRING_IGNORE_CASE		(STRING_IGNORE_LOWERCASE|STRING_IGNORE_UPPERCASE)
415b6cee71dSXin LI #define STRING_DEFAULT_RANGE		100
416b6cee71dSXin LI 
4174460e5b0SXin LI #define	INDIRECT_RELATIVE			BIT(0)
4184460e5b0SXin LI #define	CHAR_INDIRECT_RELATIVE			'r'
419b6cee71dSXin LI 
420b6cee71dSXin LI /* list of magic entries */
421b6cee71dSXin LI struct mlist {
422b6cee71dSXin LI 	struct magic *magic;		/* array of magic entries */
423a4d6d3b8SXin LI 	file_regex_t **magic_rxcomp;	/* array of compiled regexps */
424a4d6d3b8SXin LI 	size_t nmagic;			/* number of entries in array */
425b6cee71dSXin LI 	void *map;			/* internal resources used by entry */
426b6cee71dSXin LI 	struct mlist *next, *prev;
427b6cee71dSXin LI };
428b6cee71dSXin LI 
429b6cee71dSXin LI #ifdef __cplusplus
430b6cee71dSXin LI #define CAST(T, b)	static_cast<T>(b)
431b6cee71dSXin LI #define RCAST(T, b)	reinterpret_cast<T>(b)
4323e41d09dSXin LI #define CCAST(T, b)	const_cast<T>(b)
433b6cee71dSXin LI #else
4343e41d09dSXin LI #define CAST(T, b)	((T)(b))
43548c779cdSXin LI #define RCAST(T, b)	((T)(uintptr_t)(b))
4363e41d09dSXin LI #define CCAST(T, b)	((T)(uintptr_t)(b))
437b6cee71dSXin LI #endif
438b6cee71dSXin LI 
439b6cee71dSXin LI struct level_info {
440b6cee71dSXin LI 	int32_t off;
441b6cee71dSXin LI 	int got_match;
442b6cee71dSXin LI #ifdef ENABLE_CONDITIONALS
443b6cee71dSXin LI 	int last_match;
444b6cee71dSXin LI 	int last_cond;	/* used for error checking by parse() */
445b6cee71dSXin LI #endif
446b6cee71dSXin LI };
447b6cee71dSXin LI 
44843a5ec4eSXin LI struct cont {
44943a5ec4eSXin LI 	size_t len;
45043a5ec4eSXin LI 	struct level_info *li;
45143a5ec4eSXin LI };
45243a5ec4eSXin LI 
453b6cee71dSXin LI #define MAGIC_SETS	2
454b6cee71dSXin LI 
455b6cee71dSXin LI struct magic_set {
456b6cee71dSXin LI 	struct mlist *mlist[MAGIC_SETS];	/* list of regular entries */
45743a5ec4eSXin LI 	struct cont c;
458b6cee71dSXin LI 	struct out {
459b6cee71dSXin LI 		char *buf;		/* Accumulation buffer */
4602726a701SXin LI 		size_t blen;		/* Length of buffer */
461b6cee71dSXin LI 		char *pbuf;		/* Printable buffer */
462b6cee71dSXin LI 	} o;
46358a0f0d0SEitan Adler 	uint32_t offset;			/* a copy of m->offset while we */
46458a0f0d0SEitan Adler 					/* are working on the magic entry */
46558a0f0d0SEitan Adler 	uint32_t eoffset;		/* offset from end of file */
466b6cee71dSXin LI 	int error;
467b6cee71dSXin LI 	int flags;			/* Control magic tests. */
468b6cee71dSXin LI 	int event_flags;		/* Note things that happened. */
469b6cee71dSXin LI #define 		EVENT_HAD_ERR		0x01
470b6cee71dSXin LI 	const char *file;
471b6cee71dSXin LI 	size_t line;			/* current magic line number */
4722dc4dbb9SEitan Adler 	mode_t mode;			/* copy of current stat mode */
473b6cee71dSXin LI 
474b6cee71dSXin LI 	/* data for searches */
475b6cee71dSXin LI 	struct {
476b6cee71dSXin LI 		const char *s;		/* start of search in original source */
477b6cee71dSXin LI 		size_t s_len;		/* length of search region */
478b6cee71dSXin LI 		size_t offset;		/* starting offset in source: XXX - should this be off_t? */
479b6cee71dSXin LI 		size_t rm_len;		/* match length */
480b6cee71dSXin LI 	} search;
481b6cee71dSXin LI 
482b6cee71dSXin LI 	/* FIXME: Make the string dynamically allocated so that e.g.
483b6cee71dSXin LI 	   strings matched in files can be longer than MAXstring */
484b6cee71dSXin LI 	union VALUETYPE ms_value;	/* either number or string */
485c2931133SXin LI 	uint16_t indir_max;
486c2931133SXin LI 	uint16_t name_max;
487c2931133SXin LI 	uint16_t elf_shnum_max;
488c2931133SXin LI 	uint16_t elf_phnum_max;
4894460e5b0SXin LI 	uint16_t elf_notes_max;
4909ce06829SXin LI 	uint16_t regex_max;
4913e41d09dSXin LI 	size_t bytes_max;		/* number of bytes to read from file */
49243a5ec4eSXin LI 	size_t encoding_max;		/* bytes to look for encoding */
493898496eeSXin LI 	size_t	elf_shsize_max;
4942726a701SXin LI #ifndef FILE_BYTES_MAX
495898496eeSXin LI # define FILE_BYTES_MAX (7 * 1024 * 1024)/* how much of the file to look at */
496898496eeSXin LI #endif /* above 0x6ab0f4 map offset for HelveticaNeue.dfont */
4974460e5b0SXin LI #define	FILE_ELF_NOTES_MAX		256
4982726a701SXin LI #define	FILE_ELF_PHNUM_MAX		2048
4992726a701SXin LI #define	FILE_ELF_SHNUM_MAX		32768
500898496eeSXin LI #define	FILE_ELF_SHSIZE_MAX		(128 * 1024 * 1024)
5012726a701SXin LI #define	FILE_INDIR_MAX			50
5022726a701SXin LI #define	FILE_NAME_MAX			50
5039ce06829SXin LI #define	FILE_REGEX_MAX			8192
50443a5ec4eSXin LI #define	FILE_ENCODING_MAX		(64 * 1024)
505a4d6d3b8SXin LI #if defined(HAVE_NEWLOCALE) && defined(HAVE_USELOCALE) && defined(HAVE_FREELOCALE)
506a4d6d3b8SXin LI #define USE_C_LOCALE
507a4d6d3b8SXin LI 	locale_t c_lc_ctype;
508a4d6d3b8SXin LI #define file_locale_used
509a4d6d3b8SXin LI #else
510a4d6d3b8SXin LI #define file_locale_used __attribute__((__unused__))
511a4d6d3b8SXin LI #endif
512b6cee71dSXin LI };
513b6cee71dSXin LI 
514b6cee71dSXin LI /* Type for Unicode characters */
51543a5ec4eSXin LI typedef unsigned long file_unichar_t;
516b6cee71dSXin LI 
517b6cee71dSXin LI struct stat;
518b6cee71dSXin LI #define FILE_T_LOCAL	1
519b6cee71dSXin LI #define FILE_T_WINDOWS	2
520898496eeSXin LI file_protected const char *file_fmtdatetime(char *, size_t, uint64_t, int);
521898496eeSXin LI file_protected const char *file_fmtdate(char *, size_t, uint16_t);
522898496eeSXin LI file_protected const char *file_fmttime(char *, size_t, uint16_t);
523898496eeSXin LI file_protected const char *file_fmtvarint(char *, size_t, const unsigned char *,
524a2dfb722SXin LI     int);
525898496eeSXin LI file_protected const char *file_fmtnum(char *, size_t, const char *, int);
526898496eeSXin LI file_protected struct magic_set *file_ms_alloc(int);
527898496eeSXin LI file_protected void file_ms_free(struct magic_set *);
528898496eeSXin LI file_protected int file_default(struct magic_set *, size_t);
529898496eeSXin LI file_protected int file_buffer(struct magic_set *, int, struct stat *,
530898496eeSXin LI     const char *, const void *, size_t);
531898496eeSXin LI file_protected int file_fsmagic(struct magic_set *, const char *,
532898496eeSXin LI     struct stat *);
533898496eeSXin LI file_protected int file_pipe2file(struct magic_set *, int, const void *,
534898496eeSXin LI     size_t);
535898496eeSXin LI file_protected int file_vprintf(struct magic_set *, const char *, va_list)
536b6cee71dSXin LI     __attribute__((__format__(__printf__, 2, 0)));
537898496eeSXin LI file_protected int file_separator(struct magic_set *);
538898496eeSXin LI file_protected char *file_copystr(char *, size_t, size_t, const char *);
539898496eeSXin LI file_protected int file_checkfmt(char *, size_t, const char *);
540898496eeSXin LI file_protected size_t file_printedlen(const struct magic_set *);
541898496eeSXin LI file_protected int file_print_guid(char *, size_t, const uint64_t *);
542898496eeSXin LI file_protected int file_parse_guid(const char *, uint64_t *);
543898496eeSXin LI file_protected int file_replace(struct magic_set *, const char *, const char *);
544898496eeSXin LI file_protected int file_printf(struct magic_set *, const char *, ...)
545b6cee71dSXin LI     __attribute__((__format__(__printf__, 2, 3)));
546898496eeSXin LI file_protected int file_reset(struct magic_set *, int);
547898496eeSXin LI file_protected int file_tryelf(struct magic_set *, const struct buffer *);
548898496eeSXin LI file_protected int file_trycdf(struct magic_set *, const struct buffer *);
549b6cee71dSXin LI #if HAVE_FORK
550898496eeSXin LI file_protected int file_zmagic(struct magic_set *, const struct buffer *,
55158a0f0d0SEitan Adler     const char *);
552b6cee71dSXin LI #endif
553898496eeSXin LI file_protected int file_ascmagic(struct magic_set *, const struct buffer *,
554b6cee71dSXin LI     int);
555898496eeSXin LI file_protected int file_ascmagic_with_encoding(struct magic_set *,
55643a5ec4eSXin LI     const struct buffer *, file_unichar_t *, size_t, const char *, const char *, int);
557898496eeSXin LI file_protected int file_encoding(struct magic_set *, const struct buffer *,
55843a5ec4eSXin LI     file_unichar_t **, size_t *, const char **, const char **, const char **);
559898496eeSXin LI file_protected int file_is_json(struct magic_set *, const struct buffer *);
560898496eeSXin LI file_protected int file_is_csv(struct magic_set *, const struct buffer *, int,
561898496eeSXin LI     const char *);
562898496eeSXin LI file_protected int file_is_simh(struct magic_set *, const struct buffer *);
563898496eeSXin LI file_protected int file_is_tar(struct magic_set *, const struct buffer *);
564898496eeSXin LI file_protected int file_softmagic(struct magic_set *, const struct buffer *,
5653e41d09dSXin LI     uint16_t *, uint16_t *, int, int);
566898496eeSXin LI file_protected int file_apprentice(struct magic_set *, const char *, int);
567898496eeSXin LI file_protected size_t file_magic_strength(const struct magic *, size_t);
568898496eeSXin LI file_protected int buffer_apprentice(struct magic_set *, struct magic **,
569c2931133SXin LI     size_t *, size_t);
570898496eeSXin LI file_protected int file_magicfind(struct magic_set *, const char *,
571898496eeSXin LI     struct mlist *);
572898496eeSXin LI file_protected uint64_t file_signextend(struct magic_set *, struct magic *,
573b6cee71dSXin LI     uint64_t);
574898496eeSXin LI file_protected uintmax_t file_varint2uintmax_t(const unsigned char *, int,
575b6cee71dSXin LI     size_t *);
576898496eeSXin LI 
577898496eeSXin LI file_protected void file_badread(struct magic_set *);
578898496eeSXin LI file_protected void file_badseek(struct magic_set *);
579898496eeSXin LI file_protected void file_oomem(struct magic_set *, size_t);
580898496eeSXin LI file_protected void file_error(struct magic_set *, int, const char *, ...)
581898496eeSXin LI     __attribute__((__format__(__printf__, 3, 4)));
582898496eeSXin LI file_protected void file_magerror(struct magic_set *, const char *, ...)
583898496eeSXin LI     __attribute__((__format__(__printf__, 2, 3)));
584898496eeSXin LI file_protected void file_magwarn(struct magic_set *, const char *, ...)
585898496eeSXin LI     __attribute__((__format__(__printf__, 2, 3)));
586898496eeSXin LI file_protected void file_mdump(struct magic *);
587898496eeSXin LI file_protected void file_showstr(FILE *, const char *, size_t);
588898496eeSXin LI file_protected size_t file_mbswidth(struct magic_set *, const char *);
589898496eeSXin LI file_protected const char *file_getbuffer(struct magic_set *);
590898496eeSXin LI file_protected ssize_t sread(int, void *, size_t, int);
591898496eeSXin LI file_protected int file_check_mem(struct magic_set *, unsigned int);
592898496eeSXin LI file_protected int file_looks_utf8(const unsigned char *, size_t,
593898496eeSXin LI     file_unichar_t *, size_t *);
594898496eeSXin LI file_protected size_t file_pstring_length_size(struct magic_set *,
5952726a701SXin LI     const struct magic *);
596898496eeSXin LI file_protected size_t file_pstring_get_length(struct magic_set *,
5972726a701SXin LI     const struct magic *, const char *);
598898496eeSXin LI file_protected char * file_printable(struct magic_set *, char *, size_t,
59943a5ec4eSXin LI     const char *, size_t);
600b6cee71dSXin LI #ifdef __EMX__
601898496eeSXin LI file_protected int file_os2_apptype(struct magic_set *, const char *,
60248c779cdSXin LI     const void *, size_t);
603898496eeSXin LI #endif /* __EMX__ */
604898496eeSXin LI file_protected int file_pipe_closexec(int *);
605898496eeSXin LI file_protected int file_clear_closexec(int);
606898496eeSXin LI file_protected char *file_strtrim(char *);
607898496eeSXin LI 
608898496eeSXin LI file_protected void buffer_init(struct buffer *, int, const struct stat *,
609898496eeSXin LI     const void *, size_t);
610898496eeSXin LI file_protected void buffer_fini(struct buffer *);
611898496eeSXin LI file_protected int buffer_fill(const struct buffer *);
61258a0f0d0SEitan Adler 
613c2931133SXin LI 
614b6cee71dSXin LI 
615898496eeSXin LI file_protected int file_regcomp(struct magic_set *, file_regex_t *,
616898496eeSXin LI     const char *, int);
617898496eeSXin LI file_protected int file_regexec(struct magic_set *, file_regex_t *,
618898496eeSXin LI     const char *, size_t, regmatch_t *, int);
619898496eeSXin LI file_protected void file_regfree(file_regex_t *);
620b6cee71dSXin LI 
6212c4f1647SXin LI typedef struct {
6222c4f1647SXin LI 	char *buf;
6232726a701SXin LI 	size_t blen;
6242c4f1647SXin LI 	uint32_t offset;
6252c4f1647SXin LI } file_pushbuf_t;
6262c4f1647SXin LI 
627898496eeSXin LI file_protected file_pushbuf_t *file_push_buffer(struct magic_set *);
628898496eeSXin LI file_protected char  *file_pop_buffer(struct magic_set *, file_pushbuf_t *);
6292c4f1647SXin LI 
630b6cee71dSXin LI #ifndef COMPILE_ONLY
631b6cee71dSXin LI extern const char *file_names[];
632b6cee71dSXin LI extern const size_t file_nnames;
633b6cee71dSXin LI #endif
634b6cee71dSXin LI 
635b6cee71dSXin LI #ifndef HAVE_PREAD
636b6cee71dSXin LI ssize_t pread(int, void *, size_t, off_t);
637b6cee71dSXin LI #endif
638b6cee71dSXin LI #ifndef HAVE_VASPRINTF
639b6cee71dSXin LI int vasprintf(char **, const char *, va_list);
640b6cee71dSXin LI #endif
641b6cee71dSXin LI #ifndef HAVE_ASPRINTF
642b6cee71dSXin LI int asprintf(char **, const char *, ...);
643b6cee71dSXin LI #endif
6443e41d09dSXin LI #ifndef HAVE_DPRINTF
6453e41d09dSXin LI int dprintf(int, const char *, ...);
6463e41d09dSXin LI #endif
647b6cee71dSXin LI 
648b6cee71dSXin LI #ifndef HAVE_STRLCPY
649b6cee71dSXin LI size_t strlcpy(char *, const char *, size_t);
650b6cee71dSXin LI #endif
651b6cee71dSXin LI #ifndef HAVE_STRLCAT
652b6cee71dSXin LI size_t strlcat(char *, const char *, size_t);
653b6cee71dSXin LI #endif
654b6cee71dSXin LI #ifndef HAVE_STRCASESTR
655b6cee71dSXin LI char *strcasestr(const char *, const char *);
656b6cee71dSXin LI #endif
657b6cee71dSXin LI #ifndef HAVE_GETLINE
658b6cee71dSXin LI ssize_t getline(char **, size_t *, FILE *);
659b6cee71dSXin LI ssize_t getdelim(char **, size_t *, int, FILE *);
660b6cee71dSXin LI #endif
661b6cee71dSXin LI #ifndef HAVE_CTIME_R
662b6cee71dSXin LI char   *ctime_r(const time_t *, char *);
663b6cee71dSXin LI #endif
664b6cee71dSXin LI #ifndef HAVE_ASCTIME_R
665b6cee71dSXin LI char   *asctime_r(const struct tm *, char *);
666b6cee71dSXin LI #endif
6675f0216bdSXin LI #ifndef HAVE_GMTIME_R
6685f0216bdSXin LI struct tm *gmtime_r(const time_t *, struct tm *);
6695f0216bdSXin LI #endif
6705f0216bdSXin LI #ifndef HAVE_LOCALTIME_R
6715f0216bdSXin LI struct tm *localtime_r(const time_t *, struct tm *);
6725f0216bdSXin LI #endif
673b6cee71dSXin LI #ifndef HAVE_FMTCHECK
674b6cee71dSXin LI const char *fmtcheck(const char *, const char *)
675b6cee71dSXin LI      __attribute__((__format_arg__(2)));
676b6cee71dSXin LI #endif
677b6cee71dSXin LI 
67858a0f0d0SEitan Adler #ifdef HAVE_LIBSECCOMP
67958a0f0d0SEitan Adler // basic filter
68058a0f0d0SEitan Adler // this mode should not interfere with normal operations
68158a0f0d0SEitan Adler // only some dangerous syscalls are blacklisted
68258a0f0d0SEitan Adler int enable_sandbox_basic(void);
68358a0f0d0SEitan Adler 
68458a0f0d0SEitan Adler // enhanced filter
68558a0f0d0SEitan Adler // this mode allows only the necessary syscalls used during normal operation
68658a0f0d0SEitan Adler // extensive testing required !!!
68758a0f0d0SEitan Adler int enable_sandbox_full(void);
68858a0f0d0SEitan Adler #endif
68958a0f0d0SEitan Adler 
690898496eeSXin LI file_protected const char *file_getprogname(void);
691898496eeSXin LI file_protected void file_setprogname(const char *);
692898496eeSXin LI file_protected void file_err(int, const char *, ...)
6932dc4dbb9SEitan Adler     __attribute__((__format__(__printf__, 2, 3), __noreturn__));
694898496eeSXin LI file_protected void file_errx(int, const char *, ...)
6952dc4dbb9SEitan Adler     __attribute__((__format__(__printf__, 2, 3), __noreturn__));
696898496eeSXin LI file_protected void file_warn(const char *, ...)
69758a0f0d0SEitan Adler     __attribute__((__format__(__printf__, 1, 2)));
698898496eeSXin LI file_protected void file_warnx(const char *, ...)
69958a0f0d0SEitan Adler     __attribute__((__format__(__printf__, 1, 2)));
70058a0f0d0SEitan Adler 
701b6cee71dSXin LI #if defined(HAVE_MMAP) && defined(HAVE_SYS_MMAN_H) && !defined(QUICK)
702b6cee71dSXin LI #define QUICK
703b6cee71dSXin LI #endif
704b6cee71dSXin LI 
705b6cee71dSXin LI #ifndef O_BINARY
706b6cee71dSXin LI #define O_BINARY	0
707b6cee71dSXin LI #endif
70848c779cdSXin LI #ifndef O_NONBLOCK
70948c779cdSXin LI #define O_NONBLOCK	0
71048c779cdSXin LI #endif
711b6cee71dSXin LI 
712b6cee71dSXin LI #ifndef __cplusplus
713b6cee71dSXin LI #if defined(__GNUC__) && (__GNUC__ >= 3)
714b6cee71dSXin LI #define FILE_RCSID(id) \
715b6cee71dSXin LI static const char rcsid[] __attribute__((__used__)) = id;
716b6cee71dSXin LI #else
717b6cee71dSXin LI #define FILE_RCSID(id) \
718b6cee71dSXin LI static const char *rcsid(const char *p) { \
719b6cee71dSXin LI 	return rcsid(p = id); \
720b6cee71dSXin LI }
721b6cee71dSXin LI #endif
722b6cee71dSXin LI #else
723b6cee71dSXin LI #define FILE_RCSID(id)
724b6cee71dSXin LI #endif
7255f0216bdSXin LI #ifndef __RCSID
7265f0216bdSXin LI #define __RCSID(a)
7275f0216bdSXin LI #endif
728b6cee71dSXin LI 
729b6cee71dSXin LI #endif /* __file_h__ */
730