[BACK]Return to magic.5 CVS log [TXT][DIR] Up to [local] / src / usr.bin / file

Diff for /src/usr.bin/file/magic.5 between version 1.11 and 1.12

version 1.11, 2007/05/31 19:20:10 version 1.12, 2009/10/26 21:03:03
Line 33 
Line 33 
 .Dd $Mdocdate$  .Dd $Mdocdate$
 .Dt MAGIC 5  .Dt MAGIC 5
 .Os  .Os
   .\" install as magic.4 on USG, magic.5 on V7, Berkeley and Linux systems.
 .Sh NAME  .Sh NAME
 .Nm magic  .Nm magic
 .Nd file command's magic number file  .Nd file command's magic pattern file
 .Sh DESCRIPTION  .Sh DESCRIPTION
 This manual page documents the format of the magic file as  This manual page documents the format of the magic file as
 used by the  used by the
 .Xr file 1  .Xr file 1
 command, version 3.22.  command, version 4.24.
 The  The
 .Nm file  .Xr file 1
 command identifies the type of a file using,  command identifies the type of a file using,
 among other tests,  among other tests,
 a test for whether the file begins with a certain  a test for whether the file contains certain
 .Dq magic number .  .Dq "magic patterns" .
 .Pp  
 The file  The file
 .Pa /etc/magic  .Pa /etc/magic
 specifies what magic numbers are to be tested for,  specifies what magic numbers are to be tested for,
Line 56 
Line 56 
 .Pp  .Pp
 Each line of the file specifies a test to be performed.  Each line of the file specifies a test to be performed.
 A test compares the data starting at a particular offset  A test compares the data starting at a particular offset
 in the file with a 1-byte, 2-byte, or 4-byte numeric value or  in the file with a byte value, a string or a numeric value.
 a string.  
 If the test succeeds, a message is printed.  If the test succeeds, a message is printed.
 The line consists of the following fields:  The line consists of the following fields:
 .Bl -tag -width indent  .Bl -tag -width ".Dv message"
 .It Sy offset  .It Dv offset
 A number specifying the offset, in bytes, into the file of the data  A number specifying the offset, in bytes, into the file of the data
 which is to be tested.  which is to be tested.
 .It Sy type  .It Dv type
 The type of the data to be tested.  The type of the data to be tested.
 The possible values are:  The possible values are:
 .Bl -tag -width beshort  .Bl -tag -width ".Dv lestring16"
 .It Sy byte  .It Dv byte
 A one-byte value.  A one-byte value.
 .It Sy short  .It Dv short
 A two-byte value (on most systems) in this machine's native byte order.  A two-byte value in this machine's native byte order.
 .It Sy long  .It Dv long
 A four-byte value (on most systems) in this machine's native byte order.  A four-byte value in this machine's native byte order.
 .It Sy string  .It Dv quad
   An eight-byte value in this machine's native byte order.
   .It Dv float
   A 32-bit single precision IEEE floating point number in this machine's native byte order.
   .It Dv double
   A 64-bit double precision IEEE floating point number in this machine's native byte order.
   .It Dv string
 A string of bytes.  A string of bytes.
 .It Sy date  The string type specification can be optionally followed
 A four-byte value interpreted as a  by /[Bbc]*.
 .Ux  The
 date.  .Dq B
 .It Sy beshort  flag compacts whitespace in the target, which must
 A two-byte value (on most systems) in big-endian byte order.  contain at least one whitespace character.
 .It Sy belong  If the magic has
 A four-byte value (on most systems) in big-endian byte order.  .Dv n
 .It Sy bedate  consecutive blanks, the target needs at least
 A four-byte value (on most systems) in big-endian byte order,  .Dv n
 interpreted as a  consecutive blanks to match.
 .Ux  The
 date.  .Dq b
 .It Sy leshort  flag treats every blank in the target as an optional blank.
 A two-byte value (on most systems) in little-endian byte order.  Finally the
 .It Sy lelong  .Dq c
 A four-byte value (on most systems) in little-endian byte order.  flag, specifies case insensitive matching: lowercase
 .It Sy ledate  characters in the magic match both lower and upper case characters in the
 A four-byte value (on most systems) in little-endian byte order,  target, whereas upper case characters in the magic only match uppercase
 interpreted as a  characters in the target.
 .Ux  .It Dv pstring
 date.  A Pascal-style string where the first byte is interpreted as the an
   unsigned length.
   The string is not NUL terminated.
   .It Dv date
   A four-byte value interpreted as a UNIX date.
   .It Dv qdate
   A eight-byte value interpreted as a UNIX date.
   .It Dv ldate
   A four-byte value interpreted as a UNIX-style date, but interpreted as
   local time rather than UTC.
   .It Dv qldate
   An eight-byte value interpreted as a UNIX-style date, but interpreted as
   local time rather than UTC.
   .It Dv beshort
   A two-byte value in big-endian byte order.
   .It Dv belong
   A four-byte value in big-endian byte order.
   .It Dv bequad
   An eight-byte value in big-endian byte order.
   .It Dv befloat
   A 32-bit single precision IEEE floating point number in big-endian byte order.
   .It Dv bedouble
   A 64-bit double precision IEEE floating point number in big-endian byte order.
   .It Dv bedate
   A four-byte value in big-endian byte order,
   interpreted as a Unix date.
   .It Dv beqdate
   An eight-byte value in big-endian byte order,
   interpreted as a Unix date.
   .It Dv beldate
   A four-byte value in big-endian byte order,
   interpreted as a UNIX-style date, but interpreted as local time rather
   than UTC.
   .It Dv beqldate
   An eight-byte value in big-endian byte order,
   interpreted as a UNIX-style date, but interpreted as local time rather
   than UTC.
   .It Dv bestring16
   A two-byte unicode (UCS16) string in big-endian byte order.
   .It Dv leshort
   A two-byte value in little-endian byte order.
   .It Dv lelong
   A four-byte value in little-endian byte order.
   .It Dv lequad
   An eight-byte value in little-endian byte order.
   .It Dv lefloat
   A 32-bit single precision IEEE floating point number in little-endian byte order.
   .It Dv ledouble
   A 64-bit double precision IEEE floating point number in little-endian byte order.
   .It Dv ledate
   A four-byte value in little-endian byte order,
   interpreted as a UNIX date.
   .It Dv leqdate
   An eight-byte value in little-endian byte order,
   interpreted as a UNIX date.
   .It Dv leldate
   A four-byte value in little-endian byte order,
   interpreted as a UNIX-style date, but interpreted as local time rather
   than UTC.
   .It Dv leqldate
   An eight-byte value in little-endian byte order,
   interpreted as a UNIX-style date, but interpreted as local time rather
   than UTC.
   .It Dv lestring16
   A two-byte unicode (UCS16) string in little-endian byte order.
   .It Dv melong
   A four-byte value in middle-endian (PDP-11) byte order.
   .It Dv medate
   A four-byte value in middle-endian (PDP-11) byte order,
   interpreted as a UNIX date.
   .It Dv meldate
   A four-byte value in middle-endian (PDP-11) byte order,
   interpreted as a UNIX-style date, but interpreted as local time rather
   than UTC.
   .It Dv regex
   A regular expression match in extended POSIX regular expression syntax
   (like egrep).
   Regular expressions can take exponential time to process,
   and their performance is hard to predict, so their use is discouraged.
   When used in production environments,
   their performance should be carefully checked.
   The type specification can be optionally followed by
   .Dv /[c][s] .
   The
   .Dq c
   flag makes the match case insensitive, while the
   .Dq s
   flag update the offset to the start offset of the match, rather than the end.
   The regular expression is tested against line
   .Dv N + 1
   onwards, where
   .Dv N
   is the given offset.
   Line endings are assumed to be in the machine's native format.
   .Dv ^
   and
   .Dv $
   match the beginning and end of individual lines, respectively,
   not beginning and end of file.
   .It Dv search
   A literal string search starting at the given offset.
   The same modifier flags can be used as for string patterns.
   The modifier flags (if any) must be followed by
   .Dv /number
   the range, that is, the number of positions at which the match will be
   attempted, starting from the start offset.
   This is suitable for searching larger binary expressions
   with variable offsets, using
   .Dv \e
   escapes for special characters.
   The offset works as for regex.
   .It Dv default
   This is intended to be used with the test
   .Em x
   (which is always true) and a message that is to be used if there are
   no other matches.
 .El  .El
 .El  
 .Pp  .Pp
   Each top-level magic pattern (see below for an explanation of levels)
   is classified as text or binary according to the types used.
   Types
   .Dq regex
   and
   .Dq search
   are classified as text tests, unless non-printable characters are used
   in the pattern.
   All other tests are classified as binary.
   A top-level pattern is considered to be a test text
   when all its patterns are text
   patterns; otherwise, it is considered to be a binary pattern.
   When matching a file, binary patterns are tried first; if no match is
   found, and the file looks like text, then its encoding is determined
   and the text patterns are tried.
   .Pp
 The numeric types may optionally be followed by  The numeric types may optionally be followed by
 .Ql &  .Dv &
 and a numeric value,  and a numeric value,
 to specify that the value is to be AND'ed with the  to specify that the value is to be AND'ed with the
 numeric value before any comparisons are done.  numeric value before any comparisons are done.
 Prepending a  Prepending a
 .Sq u  .Dv u
 to the type indicates that ordered comparisons should be unsigned.  to the type indicates that ordered comparisons should be unsigned.
 .Bl -tag -width indent  .It Dv test
 .It Sy test  
 The value to be compared with the value from the file.  The value to be compared with the value from the file.
 If the type is  If the type is
 numeric, this value  numeric, this value
 is specified in C form; if it is a string, it is specified as a C string  is specified in C form; if it is a string, it is specified as a C string
 with the usual escapes permitted (e.g.,  with the usual escapes permitted (e.g. \en for new-line).
 .Ql \en  .Pp
 for newline).  
 .It Sy ""  
 Numeric values  Numeric values
 may be preceded by a character indicating the operation to be performed.  may be preceded by a character indicating the operation to be performed.
 It may be  It may be
 .Ql =  .Dv = ,
 to specify that the value from the file must equal the specified value,  to specify that the value from the file must equal the specified value,
 .Ql <  .Dv \*(Lt ,
 to specify that the value from the file must be less than the specified  to specify that the value from the file must be less than the specified
 value,  value,
 .Ql >  .Dv \*(Gt ,
 to specify that the value from the file must be greater than the specified  to specify that the value from the file must be greater than the specified
 value,  value,
 .Ql &  .Dv & ,
 to specify that the value from the file must have set all of the bits  to specify that the value from the file must have set all of the bits
 that are set in the specified value,  that are set in the specified value,
 .Ql ^  .Dv ^ ,
 to specify that the value from the file must have clear any of the bits  to specify that the value from the file must have clear any of the bits
 that are set in the specified value, or  that are set in the specified value, or
 .Sq x  .Dv ~ ,
   the value specified after is negated before tested.
   .Dv x ,
 to specify that any value will match.  to specify that any value will match.
 If the character is omitted,  If the character is omitted, it is assumed to be
 it is assumed to be  .Dv = .
 .Ql = .  Operators
 .It Sy ""  .Dv & ,
 Numeric values are specified in C form; e.g.,  .Dv ^ ,
 .Dq 13  and
   .Dv ~
   don't work with floats and doubles.
   The operator
   .Dv !\&
   specifies that the line matches if the test does
   .Em not
   succeed.
   .Pp
   Numeric values are specified in C form; e.g.
   .Dv 13
 is decimal,  is decimal,
 .Dq 013  .Dv 013
 is octal, and  is octal, and
 .Dq 0x13  .Dv 0x13
 is hexadecimal.  is hexadecimal.
 .It Sy ""  .Pp
 For string values, the byte string from the  For string values, the string from the
 file must match the specified byte string.  file must match the specified string.
 The operators  The operators
 .Ql = ,  .Dv = ,
 .Ql < ,  .Dv \*(Lt
 and  and
 .Ql >  .Dv \*(Gt
 (but not  (but not
 .Ql & )  .Dv & )
 can be applied to strings.  can be applied to strings.
 The length used for matching is that of the string argument  The length used for matching is that of the string argument
 in the magic file.  in the magic file.
 This means that a line can match any string, and  This means that a line can match any non-empty string (usually used to
 then presumably print that string, by doing  then print the string), with
 .Ql >\e0  .Em \*(Gt\e0
 (because all strings are greater than the null string).  (because all non-empty strings are greater than the empty string).
 .It Sy message  .Pp
   The special test
   .Em x
   always evaluates to true.
   .Dv message
 The message to be printed if the comparison succeeds.  The message to be printed if the comparison succeeds.
 If the string  If the string contains a
 contains a  
 .Xr printf 3  .Xr printf 3
 format specification, the value from the file (with any specified masking  format specification, the value from the file (with any specified masking
 performed) is printed using the message as the format string.  performed) is printed using the message as the format string.
   If the string begins with
   .Dq \eb ,
   the message printed is the remainder of the string with no whitespace
   added before it: multiple matches are normally separated by a single
   space.
 .El  .El
 .Pp  .Pp
   A MIME type is given on a separate line, which must be the next
   non-blank or comment line after the magic line that identifies the
   file type, and has the following format:
   .Bd -literal -offset indent
   !:mime  MIMETYPE
   .Ed
   .Pp
   i.e. the literal string
   .Dq !:mime
   followed by the MIME type.
   .Pp
 Some file formats contain additional information which is to be printed  Some file formats contain additional information which is to be printed
 along with the file type.  along with the file type or need additional tests to determine the true
 A line which begins with the character  file type.
 .Ql >  These additional tests are introduced by one or more
 indicates additional tests and messages to be printed.  .Em \*(Gt
   characters preceding the offset.
 The number of  The number of
 .Ql >  .Em \*(Gt
 on the line indicates the level of the test; a line with no  on the line indicates the level of the test; a line with no
 .Ql >  .Em \*(Gt
 at the beginning is considered to be at level 0.  at the beginning is considered to be at level 0.
 .Pp  Tests are arranged in a tree-like hierarchy:
 Each line at level  If a the test on a line at level
 .Em n+1  
 is under the control of the line at level  
 .Em n  .Em n
 most closely preceding it in the magic file.  succeeds, all following tests at level
 If the test on a line at level  
 .Em n  
 succeeds, the tests specified in all the subsequent lines at level  
 .Em n+1  .Em n+1
 are performed, and the messages printed if the tests succeed.  are performed, and the messages printed if the tests succeed, untile a line
 The next  with level
 line at level  
 .Em n  .Em n
 terminates this.  (or less) appears.
   For more complex files, one can use empty messages to get just the
   "if/then" effect, in the following way:
   .Bd -literal -offset indent
   0      string   MZ
   \*(Gt0x18  leshort  \*(Lt0x40   MS-DOS executable
   \*(Gt0x18  leshort  \*(Gt0x3f   extended PC executable (e.g., MS Windows)
   .Ed
 .Pp  .Pp
   Offsets do not need to be constant, but can also be read from the file
   being examined.
 If the first character following the last  If the first character following the last
 .Ql >  .Em \*(Gt
 is a  is a
 .Ql (  .Em (
 then the string after the parenthesis is interpreted as an indirect offset.  then the string after the parenthesis is interpreted as an indirect offset.
 That means that the number after the parenthesis is used as an offset in  That means that the number after the parenthesis is used as an offset in
 the file.  the file.
 The value at that offset is read, and is used again as an offset  The value at that offset is read, and is used again as an offset
 in the file.  in the file.
 .Pp  
 Indirect offsets are of the form:  Indirect offsets are of the form:
 .Dq (x[.[bsl]][+-][y]) .  .Em (( x [.[bslBSL]][+\-][ y ]) .
 The value of  The value of
 .Sq x  .Em x
 is used as an offset in the file.  is used as an offset in the file.
 A byte, short or long is read at that offset  A byte, short or long is read at that offset depending on the
 depending on the  .Op bslBSLm
 .Dq [bsl]  
 type specifier.  type specifier.
   The capitalized types interpret the number as a big endian
   value, whereas the small letter versions interpret the number as a little
   endian value;
   the
   .Em m
   type interprets the number as a middle endian (PDP-11) value.
 To that number the value of  To that number the value of
 .Sq y  .Em y
 is added and the result is used as an offset in the file.  is added and the result is used as an offset in the file.
 The default type  The default type if one is not specified is long.
 if one is not specified is long.  
 .Pp  .Pp
 Sometimes you do not know the exact offset as this depends on the length of  That way variable length structures can be examined:
 preceding fields.  .Bd -literal -offset indent
 You can specify an offset relative to the end of the  # MS Windows executables are also valid MS-DOS executables
 last uplevel field (of course this may only be done for sublevel tests, i.e.,  0           string  MZ
 test beginning with  \*(Gt0x18       leshort \*(Lt0x40   MZ executable (MS-DOS)
 .Ql > ) .  # skip the whole block below if it is not an extended executable
 Such a relative offset is specified using  \*(Gt0x18       leshort \*(Gt0x3f
 .Ql &  \*(Gt\*(Gt(0x3c.l)  string  PE\e0\e0  PE executable (MS-Windows)
 as a prefix to the offset.  \*(Gt\*(Gt(0x3c.l)  string  LX\e0\e0  LX executable (OS/2)
 .Sh FILES  .Ed
 .Bl -tag -width /etc/magic  .Pp
 .It Pa /etc/magic  This strategy of examining has a drawback: You must make sure that
 .El  you eventually print something, or users may get empty output (like, when
   there is neither PE\e0\e0 nor LE\e0\e0 in the above example)
   .Pp
   If this indirect offset cannot be used directly, simple calculations are
   possible: appending
   .Em [+-*/%&|^]number
   inside parentheses allows one to modify
   the value read from the file before it is used as an offset:
   .Bd -literal -offset indent
   # MS Windows executables are also valid MS-DOS executables
   0           string  MZ
   # sometimes, the value at 0x18 is less that 0x40 but there's still an
   # extended executable, simply appended to the file
   \*(Gt0x18       leshort \*(Lt0x40
   \*(Gt\*(Gt(4.s*512) leshort 0x014c  COFF executable (MS-DOS, DJGPP)
   \*(Gt\*(Gt(4.s*512) leshort !0x014c MZ executable (MS-DOS)
   .Ed
   .Pp
   Sometimes you do not know the exact offset as this depends on the length or
   position (when indirection was used before) of preceding fields.
   You can specify an offset relative to the end of the last up-level
   field using
   .Sq &
   as a prefix to the offset:
   .Bd -literal -offset indent
   0           string  MZ
   \*(Gt0x18       leshort \*(Gt0x3f
   \*(Gt\*(Gt(0x3c.l)  string  PE\e0\e0    PE executable (MS-Windows)
   # immediately following the PE signature is the CPU type
   \*(Gt\*(Gt\*(Gt&0       leshort 0x14c     for Intel 80386
   \*(Gt\*(Gt\*(Gt&0       leshort 0x184     for DEC Alpha
   .Ed
   .Pp
   Indirect and relative offsets can be combined:
   .Bd -literal -offset indent
   0             string  MZ
   \*(Gt0x18         leshort \*(Lt0x40
   \*(Gt\*(Gt(4.s*512)   leshort !0x014c MZ executable (MS-DOS)
   # if it's not COFF, go back 512 bytes and add the offset taken
   # from byte 2/3, which is yet another way of finding the start
   # of the extended executable
   \*(Gt\*(Gt\*(Gt&(2.s-514) string  LE      LE executable (MS Windows VxD driver)
   .Ed
   .Pp
   Or the other way around:
   .Bd -literal -offset indent
   0                 string  MZ
   \*(Gt0x18             leshort \*(Gt0x3f
   \*(Gt\*(Gt(0x3c.l)        string  LE\e0\e0  LE executable (MS-Windows)
   # at offset 0x80 (-4, since relative offsets start at the end
   # of the up-level match) inside the LE header, we find the absolute
   # offset to the code area, where we look for a specific signature
   \*(Gt\*(Gt\*(Gt(&0x7c.l+0x26) string  UPX     \eb, UPX compressed
   .Ed
   .Pp
   Or even both!
   .Bd -literal -offset indent
   0                string  MZ
   \*(Gt0x18            leshort \*(Gt0x3f
   \*(Gt\*(Gt(0x3c.l)       string  LE\e0\e0 LE executable (MS-Windows)
   # at offset 0x58 inside the LE header, we find the relative offset
   # to a data area where we look for a specific signature
   \*(Gt\*(Gt\*(Gt&(&0x54.l-3)  string  UNACE  \eb, ACE self-extracting archive
   .Ed
   .Pp
   Finally, if you have to deal with offset/length pairs in your file, even the
   second value in a parenthesized expression can be taken from the file itself,
   using another set of parentheses.
   Note that this additional indirect offset is always relative to the
   start of the main indirect offset.
   .Bd -literal -offset indent
   0                 string       MZ
   \*(Gt0x18             leshort      \*(Gt0x3f
   \*(Gt\*(Gt(0x3c.l)        string       PE\e0\e0 PE executable (MS-Windows)
   # search for the PE section called ".idata"...
   \*(Gt\*(Gt\*(Gt&0xf4          search/0x140 .idata
   # ...and go to the end of it, calculated from start+length;
   # these are located 14 and 10 bytes after the section name
   \*(Gt\*(Gt\*(Gt\*(Gt(&0xe.l+(-4)) string       PK\e3\e4 \eb, ZIP self-extracting archive
   .Ed
 .Sh SEE ALSO  .Sh SEE ALSO
 .Xr file 1  .Xr file 1
   \- the command that reads this file.
 .Sh BUGS  .Sh BUGS
 The formats  The formats
 .Li long ,  .Dv long ,
 .Li belong ,  .Dv belong ,
 .Li lelong ,  .Dv lelong ,
 .Li short ,  .Dv melong ,
 .Li beshort ,  .Dv short ,
 .Li leshort ,  .Dv beshort ,
 .Li date ,  .Dv leshort ,
 .Li bedate ,  .Dv date ,
   .Dv bedate ,
   .Dv medate ,
   .Dv ledate ,
   .Dv beldate ,
   .Dv leldate ,
 and  and
 .Li ledate  .Dv meldate
 are system-dependent; perhaps they should be specified as a number  are system-dependent; perhaps they should be specified as a number
 of bytes (2B, 4B, etc),  of bytes (2B, 4B, etc),
 since the files being recognized typically come from  since the files being recognized typically come from
 a system on which the lengths are invariant.  a system on which the lengths are invariant.
 .Pp  
 There is (currently) no support for specified-endian data to be used in  
 indirect offsets.  
 .\"  .\"
 .\" From: guy@sun.uucp (Guy Harris)  .\" From: guy@sun.uucp (Guy Harris)
 .\" Newsgroups: net.bugs.usg  .\" Newsgroups: net.bugs.usg
Legend:
Removed from v.1.11  
changed lines
  Added in v.1.12