xref: /openbsd/usr.bin/file/magic.5 (revision 55cc5ba3)
1.\" $OpenBSD: magic.5,v 1.19 2020/11/14 10:35:58 jmc Exp $
2.\"
3.\" @(#)$FreeBSD: src/usr.bin/file/magic.5,v 1.11 2000/03/01 12:19:39 sheldonh Exp $
4.\"
5.\" install as magic.4 on USG, magic.5 on V7 or Berkeley systems.
6.\"
7.\" Copyright (c) Ian F. Darwin 1986-1995.
8.\" Software written by Ian F. Darwin and others;
9.\" maintained 1995-present by Christos Zoulas and others.
10.\"
11.\" Redistribution and use in source and binary forms, with or without
12.\" modification, are permitted provided that the following conditions
13.\" are met:
14.\" 1. Redistributions of source code must retain the above copyright
15.\"    notice immediately at the beginning of the file, without modification,
16.\"    this list of conditions, and the following disclaimer.
17.\" 2. Redistributions in binary form must reproduce the above copyright
18.\"    notice, this list of conditions and the following disclaimer in the
19.\"    documentation and/or other materials provided with the distribution.
20.\"
21.\" THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
22.\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
23.\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
24.\" ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE FOR
25.\" ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
26.\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
27.\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
28.\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
29.\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
30.\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
31.\" SUCH DAMAGE.
32.\"
33.Dd $Mdocdate: November 14 2020 $
34.Dt MAGIC 5
35.Os
36.\" install as magic.4 on USG, magic.5 on V7, Berkeley and Linux systems.
37.Sh NAME
38.Nm magic
39.Nd file command's magic pattern file
40.Sh DESCRIPTION
41This manual page documents the format of the magic file as
42used by the
43.Xr file 1
44command.
45.Xr file 1
46identifies the type of a file using,
47among other tests,
48a test for whether the file contains certain
49.Dq "magic patterns" .
50The file
51.Pa /etc/magic
52specifies what magic numbers are to be tested for,
53what message to print if a particular magic number is found,
54and additional information to extract from the file.
55.Pp
56Each line of the file specifies a test to be performed.
57A test compares the data starting at a particular offset
58in the file with a byte value, a string or a numeric value.
59If the test succeeds, a message is printed.
60The line consists of the following fields:
61.Bl -tag -width "message"
62.It Dv offset
63A number specifying the offset, in bytes, into the file of the data
64which is to be tested.
65.It Dv type
66The type of the data to be tested.
67The possible values are:
68.Bl -tag -width "lestring16"
69.It Dv byte
70A one-byte value.
71.It Dv short
72A two-byte value in this machine's native byte order.
73.It Dv long
74A four-byte value in this machine's native byte order.
75.It Dv quad
76An eight-byte value in this machine's native byte order.
77.It Dv float
78A 32-bit single precision IEEE floating point number in this machine's native byte order.
79.It Dv double
80A 64-bit double precision IEEE floating point number in this machine's native byte order.
81.It Dv string
82A string of bytes.
83The string type specification can be optionally followed
84by /[Bbc]*.
85The
86.Dq B
87flag compacts whitespace in the target, which must
88contain at least one whitespace character.
89If the magic has
90.Dv n
91consecutive blanks, the target needs at least
92.Dv n
93consecutive blanks to match.
94The
95.Dq b
96flag treats every blank in the target as an optional blank.
97Finally the
98.Dq c
99flag, specifies case insensitive matching: lowercase
100characters in the magic match both lower and upper case characters in the
101target, whereas upper case characters in the magic only match uppercase
102characters in the target.
103.It Dv pstring
104A Pascal-style string where the first byte is interpreted as the an
105unsigned length.
106The string is not NUL terminated.
107.It Dv date
108A four-byte value interpreted as a UNIX date.
109.It Dv qdate
110An eight-byte value interpreted as a UNIX date.
111.It Dv ldate
112A four-byte value interpreted as a UNIX-style date, but interpreted as
113local time rather than UTC.
114.It Dv qldate
115An eight-byte value interpreted as a UNIX-style date, but interpreted as
116local time rather than UTC.
117.It Dv beshort
118A two-byte value in big-endian byte order.
119.It Dv belong
120A four-byte value in big-endian byte order.
121.It Dv bequad
122An eight-byte value in big-endian byte order.
123.It Dv befloat
124A 32-bit single precision IEEE floating point number in big-endian byte order.
125.It Dv bedouble
126A 64-bit double precision IEEE floating point number in big-endian byte order.
127.It Dv bedate
128A four-byte value in big-endian byte order,
129interpreted as a Unix date.
130.It Dv beqdate
131An eight-byte value in big-endian byte order,
132interpreted as a Unix date.
133.It Dv beldate
134A four-byte value in big-endian byte order,
135interpreted as a UNIX-style date, but interpreted as local time rather
136than UTC.
137.It Dv beqldate
138An eight-byte value in big-endian byte order,
139interpreted as a UNIX-style date, but interpreted as local time rather
140than UTC.
141.It Dv bestring16
142A two-byte unicode (UCS16) string in big-endian byte order.
143.It Dv leshort
144A two-byte value in little-endian byte order.
145.It Dv lelong
146A four-byte value in little-endian byte order.
147.It Dv lequad
148An eight-byte value in little-endian byte order.
149.It Dv lefloat
150A 32-bit single precision IEEE floating point number in little-endian byte order.
151.It Dv ledouble
152A 64-bit double precision IEEE floating point number in little-endian byte order.
153.It Dv ledate
154A four-byte value in little-endian byte order,
155interpreted as a UNIX date.
156.It Dv leqdate
157An eight-byte value in little-endian byte order,
158interpreted as a UNIX date.
159.It Dv leldate
160A four-byte value in little-endian byte order,
161interpreted as a UNIX-style date, but interpreted as local time rather
162than UTC.
163.It Dv leqldate
164An eight-byte value in little-endian byte order,
165interpreted as a UNIX-style date, but interpreted as local time rather
166than UTC.
167.It Dv lestring16
168A two-byte unicode (UCS16) string in little-endian byte order.
169.It Dv melong
170A four-byte value in middle-endian (PDP-11) byte order.
171.It Dv medate
172A four-byte value in middle-endian (PDP-11) byte order,
173interpreted as a UNIX date.
174.It Dv meldate
175A four-byte value in middle-endian (PDP-11) byte order,
176interpreted as a UNIX-style date, but interpreted as local time rather
177than UTC.
178.It Dv regex
179A regular expression match in extended POSIX regular expression syntax
180(like egrep).
181Regular expressions can take exponential time to process,
182and their performance is hard to predict, so their use is discouraged.
183When used in production environments,
184their performance should be carefully checked.
185The type specification can be optionally followed by
186.Dv /[c][s] .
187The
188.Dq c
189flag makes the match case insensitive, while the
190.Dq s
191flag update the offset to the start offset of the match, rather than the end.
192The regular expression is tested against line
193.Dv N + 1
194onwards, where
195.Dv N
196is the given offset.
197Line endings are assumed to be in the machine's native format.
198.Dv ^
199and
200.Dv $
201match the beginning and end of individual lines, respectively,
202not beginning and end of file.
203.It Dv search
204A literal string search starting at the given offset.
205The same modifier flags can be used as for string patterns.
206The modifier flags (if any) must be followed by
207.Dv /number
208the range, that is, the number of positions at which the match will be
209attempted, starting from the start offset.
210This is suitable for searching larger binary expressions
211with variable offsets, using
212.Dv \e
213escapes for special characters.
214The offset works as for regex.
215.It Dv default
216This is intended to be used with the test
217.Em x
218(which is always true) and a message that is to be used if there are
219no other matches.
220.It Dv clear
221This test is always true and clears the match flag for that level.
222It is intended to be used with the default test.
223.It Dv name
224Define a named magic instance that can be called from another
225.Dv use
226magic entry, like a subroutine call.
227Named instance direct magic offsets are relative to the offset of the
228previous matched entry, but indirect offsets are relative to the
229beginning of the file as usual.
230Named magic entries always match.
231.It Dv use
232Recursively call the named magic starting from the current offset.
233If the name of the referenced instance begins with a
234.Dv ^
235then the endianness of the magic is switched; if the magic mentioned
236.Dv leshort
237for example,
238it is treated as
239.Dv beshort
240and vice versa.
241This is useful to avoid duplicating the rules for different endianness.
242.El
243.Pp
244Each top-level magic pattern (see below for an explanation of levels)
245is classified as text or binary according to the types used.
246Types
247.Dq regex
248and
249.Dq search
250are classified as text tests, unless non-printable characters are used
251in the pattern.
252All other tests are classified as binary.
253A top-level pattern is considered to be a test text
254when all its patterns are text
255patterns; otherwise, it is considered to be a binary pattern.
256When matching a file, binary patterns are tried first; if no match is
257found, and the file looks like text, then its encoding is determined
258and the text patterns are tried.
259.Pp
260The numeric types may optionally be followed by
261.Dv &
262and a numeric value,
263to specify that the value is to be AND'ed with the
264numeric value before any comparisons are done.
265Prepending a
266.Dv u
267to the type indicates that ordered comparisons should be unsigned.
268.It Dv test
269The value to be compared with the value from the file.
270If the type is
271numeric, this value
272is specified in C form; if it is a string, it is specified as a C string
273with the usual escapes permitted (e.g. \en for new-line).
274.Pp
275Numeric values
276may be preceded by a character indicating the operation to be performed.
277It may be
278.Dv = ,
279to specify that the value from the file must equal the specified value,
280.Dv \*(Lt ,
281to specify that the value from the file must be less than the specified
282value,
283.Dv \*(Gt ,
284to specify that the value from the file must be greater than the specified
285value,
286.Dv & ,
287to specify that the value from the file must have set all of the bits
288that are set in the specified value,
289.Dv ^ ,
290to specify that the value from the file must have clear any of the bits
291that are set in the specified value, or
292.Dv ~ ,
293the value specified after is negated before tested.
294.Dv x ,
295to specify that any value will match.
296If the character is omitted, it is assumed to be
297.Dv = .
298Operators
299.Dv & ,
300.Dv ^ ,
301and
302.Dv ~
303don't work with floats and doubles.
304The operator
305.Dv !\&
306specifies that the line matches if the test does
307.Em not
308succeed.
309.Pp
310Numeric values are specified in C form; e.g.
311.Dv 13
312is decimal,
313.Dv 013
314is octal, and
315.Dv 0x13
316is hexadecimal.
317.Pp
318For string values, the string from the
319file must match the specified string.
320The operators
321.Dv = ,
322.Dv \*(Lt
323and
324.Dv \*(Gt
325(but not
326.Dv & )
327can be applied to strings.
328The length used for matching is that of the string argument
329in the magic file.
330This means that a line can match any non-empty string (usually used to
331then print the string), with
332.Em \*(Gt\e0
333(because all non-empty strings are greater than the empty string).
334.Pp
335The special test
336.Em x
337always evaluates to true.
338.It Dv message
339The message to be printed if the comparison succeeds.
340If the string contains a
341.Xr printf 3
342format specification, the value from the file (with any specified masking
343performed) is printed using the message as the format string.
344If the string begins with
345.Dq \eb ,
346the message printed is the remainder of the string with no whitespace
347added before it: multiple matches are normally separated by a single
348space.
349.El
350.Pp
351A MIME type is given on a separate line, which must be the next
352non-blank or comment line after the magic line that identifies the
353file type, and has the following format:
354.Bd -literal -offset indent
355!:mime	MIMETYPE
356.Ed
357.Pp
358i.e. the literal string
359.Dq !:mime
360followed by the MIME type.
361.Pp
362Some file formats contain additional information which is to be printed
363along with the file type or need additional tests to determine the true
364file type.
365These additional tests are introduced by one or more
366.Em \*(Gt
367characters preceding the offset.
368The number of
369.Em \*(Gt
370on the line indicates the level of the test; a line with no
371.Em \*(Gt
372at the beginning is considered to be at level 0.
373Tests are arranged in a tree-like hierarchy:
374If a test on a line at level
375.Em n
376succeeds, all following tests at level
377.Em n+1
378are performed, and the messages printed if the tests succeed, until a line
379with level
380.Em n
381(or less) appears.
382For more complex files, one can use empty messages to get just the
383"if/then" effect, in the following way:
384.Bd -literal -offset indent
3850      string   MZ
386\*(Gt0x18  leshort  \*(Lt0x40   MS-DOS executable
387\*(Gt0x18  leshort  \*(Gt0x3f   extended PC executable (e.g., MS Windows)
388.Ed
389.Pp
390Offsets do not need to be constant, but can also be read from the file
391being examined.
392If the first character following the last
393.Em \*(Gt
394is a
395.Em \&(
396then the string after the parenthesis is interpreted as an indirect offset.
397That means that the number after the parenthesis is used as an offset in
398the file.
399The value at that offset is read, and is used again as an offset
400in the file.
401Indirect offsets are of the form:
402.Em (( x [.[bslBSL]][+\-][ y ]) .
403The value of
404.Em x
405is used as an offset in the file.
406A byte, short or long is read at that offset depending on the
407.Op bslBSLm
408type specifier.
409The capitalized types interpret the number as a big endian
410value, whereas the small letter versions interpret the number as a little
411endian value;
412the
413.Em m
414type interprets the number as a middle endian (PDP-11) value.
415To that number the value of
416.Em y
417is added and the result is used as an offset in the file.
418The default type if one is not specified is long.
419.Pp
420That way variable length structures can be examined:
421.Bd -literal -offset indent
422# MS Windows executables are also valid MS-DOS executables
4230           string  MZ
424\*(Gt0x18       leshort \*(Lt0x40   MZ executable (MS-DOS)
425# skip the whole block below if it is not an extended executable
426\*(Gt0x18       leshort \*(Gt0x3f
427\*(Gt\*(Gt(0x3c.l)  string  PE\e0\e0  PE executable (MS-Windows)
428\*(Gt\*(Gt(0x3c.l)  string  LX\e0\e0  LX executable (OS/2)
429.Ed
430.Pp
431This strategy of examining has a drawback: You must make sure that
432you eventually print something, or users may get empty output (like, when
433there is neither PE\e0\e0 nor LE\e0\e0 in the above example)
434.Pp
435If this indirect offset cannot be used directly, simple calculations are
436possible: appending
437.Em [+-*/%&|^]number
438inside parentheses allows one to modify
439the value read from the file before it is used as an offset:
440.Bd -literal -offset indent
441# MS Windows executables are also valid MS-DOS executables
4420           string  MZ
443# sometimes, the value at 0x18 is less that 0x40 but there's still an
444# extended executable, simply appended to the file
445\*(Gt0x18       leshort \*(Lt0x40
446\*(Gt\*(Gt(4.s*512) leshort 0x014c  COFF executable (MS-DOS, DJGPP)
447\*(Gt\*(Gt(4.s*512) leshort !0x014c MZ executable (MS-DOS)
448.Ed
449.Pp
450Sometimes you do not know the exact offset as this depends on the length or
451position (when indirection was used before) of preceding fields.
452You can specify an offset relative to the end of the last up-level
453field using
454.Sq &
455as a prefix to the offset:
456.Bd -literal -offset indent
4570           string  MZ
458\*(Gt0x18       leshort \*(Gt0x3f
459\*(Gt\*(Gt(0x3c.l)  string  PE\e0\e0    PE executable (MS-Windows)
460# immediately following the PE signature is the CPU type
461\*(Gt\*(Gt\*(Gt&0       leshort 0x14c     for Intel 80386
462\*(Gt\*(Gt\*(Gt&0       leshort 0x184     for DEC Alpha
463.Ed
464.Pp
465Indirect and relative offsets can be combined:
466.Bd -literal -offset indent
4670             string  MZ
468\*(Gt0x18         leshort \*(Lt0x40
469\*(Gt\*(Gt(4.s*512)   leshort !0x014c MZ executable (MS-DOS)
470# if it's not COFF, go back 512 bytes and add the offset taken
471# from byte 2/3, which is yet another way of finding the start
472# of the extended executable
473\*(Gt\*(Gt\*(Gt&(2.s-514) string  LE      LE executable (MS Windows VxD driver)
474.Ed
475.Pp
476Or the other way around:
477.Bd -literal -offset indent
4780                 string  MZ
479\*(Gt0x18             leshort \*(Gt0x3f
480\*(Gt\*(Gt(0x3c.l)        string  LE\e0\e0  LE executable (MS-Windows)
481# at offset 0x80 (-4, since relative offsets start at the end
482# of the up-level match) inside the LE header, we find the absolute
483# offset to the code area, where we look for a specific signature
484\*(Gt\*(Gt\*(Gt(&0x7c.l+0x26) string  UPX     \eb, UPX compressed
485.Ed
486.Pp
487Or even both!
488.Bd -literal -offset indent
4890                string  MZ
490\*(Gt0x18            leshort \*(Gt0x3f
491\*(Gt\*(Gt(0x3c.l)       string  LE\e0\e0 LE executable (MS-Windows)
492# at offset 0x58 inside the LE header, we find the relative offset
493# to a data area where we look for a specific signature
494\*(Gt\*(Gt\*(Gt&(&0x54.l-3)  string  UNACE  \eb, ACE self-extracting archive
495.Ed
496.Pp
497Finally, if you have to deal with offset/length pairs in your file, even the
498second value in a parenthesized expression can be taken from the file itself,
499using another set of parentheses.
500Note that this additional indirect offset is always relative to the
501start of the main indirect offset.
502.Bd -literal -offset indent
5030                 string       MZ
504\*(Gt0x18             leshort      \*(Gt0x3f
505\*(Gt\*(Gt(0x3c.l)        string       PE\e0\e0 PE executable (MS-Windows)
506# search for the PE section called ".idata"...
507\*(Gt\*(Gt\*(Gt&0xf4          search/0x140 .idata
508# ...and go to the end of it, calculated from start+length;
509# these are located 14 and 10 bytes after the section name
510\*(Gt\*(Gt\*(Gt\*(Gt(&0xe.l+(-4)) string       PK\e3\e4 \eb, ZIP self-extracting archive
511.Ed
512.Sh SEE ALSO
513.Xr file 1
514\- the command that reads this file.
515.Sh BUGS
516The formats
517.Dv long ,
518.Dv belong ,
519.Dv lelong ,
520.Dv melong ,
521.Dv short ,
522.Dv beshort ,
523.Dv leshort ,
524.Dv date ,
525.Dv bedate ,
526.Dv medate ,
527.Dv ledate ,
528.Dv beldate ,
529.Dv leldate ,
530and
531.Dv meldate
532are system-dependent; perhaps they should be specified as a number
533of bytes (2B, 4B, etc),
534since the files being recognized typically come from
535a system on which the lengths are invariant.
536.\"
537.\" From: guy@sun.uucp (Guy Harris)
538.\" Newsgroups: net.bugs.usg
539.\" Subject: /etc/magic's format isn't well documented
540.\" Message-ID: <2752@sun.uucp>
541.\" Date: 3 Sep 85 08:19:07 GMT
542.\" Organization: Sun Microsystems, Inc.
543.\" Lines: 136
544.\"
545.\" Here's a manual page for the format accepted by the "file" made by adding
546.\" the changes I posted to the S5R2 version.
547.\"
548.\" Modified for Ian Darwin's version of the file command.
549