16 KB

  1. .\" $File:,v 1.73 2008/02/19 17:58:00 rrt Exp $
  2. .Dd February 19, 2008
  3. .Dt FILE __CSECTION__
  4. .Os
  5. .Sh NAME
  6. .Nm file
  7. .Nd determine file type
  9. .Nm
  10. .Op Fl bchikLnNprsvz
  11. .Op Fl -mime-type
  12. .Op Fl -mime-encoding
  13. .Op Fl f Ar namefile
  14. .Op Fl F Ar separator
  15. .Op Fl m Ar magicfiles
  16. .Ar file
  17. .Nm
  18. .Fl C
  19. .Op Fl m Ar magicfile
  20. .Nm
  21. .Op Fl -help
  23. This manual page documents version __VERSION__ of the
  24. .Nm
  25. command.
  26. .Pp
  27. .Nm
  28. tests each argument in an attempt to classify it.
  29. There are three sets of tests, performed in this order:
  30. filesystem tests, magic tests, and language tests.
  31. The
  32. .Em first
  33. test that succeeds causes the file type to be printed.
  34. .Pp
  35. The type printed will usually contain one of the words
  36. .Em text
  37. (the file contains only
  38. printing characters and a few common control
  39. characters and is probably safe to read on an
  40. .Dv ASCII
  41. terminal),
  42. .Em executable
  43. (the file contains the result of compiling a program
  44. in a form understandable to some
  45. .Dv UNIX
  46. kernel or another),
  47. or
  48. .Em data
  49. meaning anything else (data is usually
  50. .Sq binary
  51. or non-printable).
  52. Exceptions are well-known file formats (core files, tar archives)
  53. that are known to contain binary data.
  54. When modifying magic files or the program itself, make sure to
  55. .Em "preserve these keywords" .
  56. Users depend on knowing that all the readable files in a directory
  57. have the word
  58. .Dq text
  59. printed.
  60. Don't do as Berkeley did and change
  61. .Dq shell commands text
  62. to
  63. .Dq shell script .
  64. .Pp
  65. The filesystem tests are based on examining the return from a
  66. .Xr stat 2
  67. system call.
  68. The program checks to see if the file is empty,
  69. or if it's some sort of special file.
  70. Any known file types appropriate to the system you are running on
  71. (sockets, symbolic links, or named pipes (FIFOs) on those systems that
  72. implement them)
  73. are intuited if they are defined in
  74. the system header file
  75. .In sys/stat.h .
  76. .Pp
  77. The magic tests are used to check for files with data in
  78. particular fixed formats.
  79. The canonical example of this is a binary executable (compiled program)
  80. .Dv a.out
  81. file, whose format is defined in
  82. .In elf.h ,
  83. .In a.out.h
  84. and possibly
  85. .In exec.h
  86. in the standard include directory.
  87. These files have a
  88. .Sq "magic number"
  89. stored in a particular place
  90. near the beginning of the file that tells the
  91. .Dv UNIX operating system
  92. that the file is a binary executable, and which of several types thereof.
  93. The concept of a
  94. .Sq "magic"
  95. has been applied by extension to data files.
  96. Any file with some invariant identifier at a small fixed
  97. offset into the file can usually be described in this way.
  98. The information identifying these files is read from the compiled
  99. magic file
  100. .Pa __MAGIC__.mgc ,
  101. or the files in the directory
  102. .Pa __MAGIC__
  103. if the compiled file does not exist. In addition, if
  104. .Pa $HOME/.magic.mgc
  105. or
  106. .Pa $HOME/.magic
  107. exists, it will be used in preference to the system magic files.
  108. .Pp
  109. If a file does not match any of the entries in the magic file,
  110. it is examined to see if it seems to be a text file.
  111. ASCII, ISO-8859-x, non-ISO 8-bit extended-ASCII character sets
  112. (such as those used on Macintosh and IBM PC systems),
  113. UTF-8-encoded Unicode, UTF-16-encoded Unicode, and EBCDIC
  114. character sets can be distinguished by the different
  115. ranges and sequences of bytes that constitute printable text
  116. in each set.
  117. If a file passes any of these tests, its character set is reported.
  118. ASCII, ISO-8859-x, UTF-8, and extended-ASCII files are identified
  119. as
  120. .Dq text
  121. because they will be mostly readable on nearly any terminal;
  122. UTF-16 and EBCDIC are only
  123. .Dq character data
  124. because, while
  125. they contain text, it is text that will require translation
  126. before it can be read.
  127. In addition,
  128. .Nm
  129. will attempt to determine other characteristics of text-type files.
  130. If the lines of a file are terminated by CR, CRLF, or NEL, instead
  131. of the Unix-standard LF, this will be reported.
  132. Files that contain embedded escape sequences or overstriking
  133. will also be identified.
  134. .Pp
  135. Once
  136. .Nm
  137. has determined the character set used in a text-type file,
  138. it will
  139. attempt to determine in what language the file is written.
  140. The language tests look for particular strings (cf.
  141. .In names.h
  142. ) that can appear anywhere in the first few blocks of a file.
  143. For example, the keyword
  144. .Em .br
  145. indicates that the file is most likely a
  146. .Xr troff 1
  147. input file, just as the keyword
  148. .Em struct
  149. indicates a C program.
  150. These tests are less reliable than the previous
  151. two groups, so they are performed last.
  152. The language test routines also test for some miscellany
  153. (such as
  154. .Xr tar 1
  155. archives).
  156. .Pp
  157. Any file that cannot be identified as having been written
  158. in any of the character sets listed above is simply said to be
  159. .Dq data .
  160. .Sh OPTIONS
  161. .Bl -tag -width indent
  162. .It Fl b , -brief
  163. Do not prepend filenames to output lines (brief mode).
  164. .It Fl c , -checking-printout
  165. Cause a checking printout of the parsed form of the magic file.
  166. This is usually used in conjunction with the
  167. .Fl m
  168. flag to debug a new magic file before installing it.
  169. .It Fl C , -compile
  170. Write a
  171. .Pa magic.mgc
  172. output file that contains a pre-parsed version of the magic file or directory.
  173. .It Fl e , -exclude Ar testname
  174. Exclude the test named in
  175. .Ar testname
  176. from the list of tests made to determine the file type. Valid test names
  177. are:
  178. .Bl -tag -width
  179. .It apptype
  180. Check for
  181. .Dv EMX
  182. application type (only on EMX).
  183. .It ascii
  184. Check for various types of ascii files.
  185. .It compress
  186. Don't look for, or inside compressed files.
  187. .It elf
  188. Don't print elf details.
  189. .It fortran
  190. Don't look for fortran sequences inside ascii files.
  191. .It soft
  192. Don't consult magic files.
  193. .It tar
  194. Don't examine tar files.
  195. .It token
  196. Don't look for known tokens inside ascii files.
  197. .It troff
  198. Don't look for troff sequences inside ascii files.
  199. .El
  200. .It Fl f , -files-from Ar namefile
  201. Read the names of the files to be examined from
  202. .Ar namefile
  203. (one per line)
  204. before the argument list.
  205. Either
  206. .Ar namefile
  207. or at least one filename argument must be present;
  208. to test the standard input, use
  209. .Sq -
  210. as a filename argument.
  211. .It Fl F , -separator Ar separator
  212. Use the specified string as the separator between the filename and the
  213. file result returned. Defaults to
  214. .Sq \&: .
  215. .It Fl h , -no-dereference
  216. option causes symlinks not to be followed
  217. (on systems that support symbolic links). This is the default if the
  218. environment variable
  220. is not defined.
  221. .It Fl i , -mime
  222. Causes the file command to output mime type strings rather than the more
  223. traditional human readable ones. Thus it may say
  224. .Dq text/plain charset=us-ascii
  225. rather than
  226. .Dq ASCII text .
  227. In order for this option to work, file changes the way
  228. it handles files recognized by the command itself (such as many of the
  229. text file types, directories etc), and makes use of an alternative
  230. .Dq magic
  231. file.
  232. (See
  233. .Dq FILES
  234. section, below).
  235. .It Fl -mime-type , -mime-encoding
  236. Like
  237. .Fl i ,
  238. but print only the specified element(s).
  239. .It Fl k , -keep-going
  240. Don't stop at the first match, keep going. Subsequent matches will be
  241. have the string
  242. .Dq "\[rs]012\- "
  243. prepended.
  244. (If you want a newline, see the
  245. .Dq "\-r"
  246. option.)
  247. .It Fl L , -dereference
  248. option causes symlinks to be followed, as the like-named option in
  249. .Xr ls 1
  250. (on systems that support symbolic links).
  251. This is the default if the environment variable
  253. is defined.
  254. .It Fl m , -magic-file Ar list
  255. Specify an alternate list of files and directories containing magic.
  256. This can be a single item, or a colon-separated list.
  257. If a compiled magic file is found alongside a file or directory, it will be used instead.
  258. .It Fl n , -no-buffer
  259. Force stdout to be flushed after checking each file.
  260. This is only useful if checking a list of files.
  261. It is intended to be used by programs that want filetype output from a pipe.
  262. .It Fl N , -no-pad
  263. Don't pad filenames so that they align in the output.
  264. .It Fl p , -preserve-date
  265. On systems that support
  266. .Xr utime 2
  267. or
  268. .Xr utimes 2 ,
  269. attempt to preserve the access time of files analyzed, to pretend that
  270. .Nm
  271. never read them.
  272. .It Fl r , -raw
  273. Don't translate unprintable characters to \eooo.
  274. Normally
  275. .Nm
  276. translates unprintable characters to their octal representation.
  277. .It Fl s , -special-files
  278. Normally,
  279. .Nm
  280. only attempts to read and determine the type of argument files which
  281. .Xr stat 2
  282. reports are ordinary files.
  283. This prevents problems, because reading special files may have peculiar
  284. consequences.
  285. Specifying the
  286. .Fl s
  287. option causes
  288. .Nm
  289. to also read argument files which are block or character special files.
  290. This is useful for determining the filesystem types of the data in raw
  291. disk partitions, which are block special files.
  292. This option also causes
  293. .Nm
  294. to disregard the file size as reported by
  295. .Xr stat 2
  296. since on some systems it reports a zero size for raw disk partitions.
  297. .It Fl v , -version
  298. Print the version of the program and exit.
  299. .It Fl z , -uncompress
  300. Try to look inside compressed files.
  301. .It Fl 0 , -print0
  302. Output a null character
  303. .Sq \e0
  304. after the end of the filename. Nice to
  305. .Xr cut 1
  306. the output. This does not affect the separator which is still printed.
  307. .It Fl -help
  308. Print a help message and exit.
  309. .El
  310. .Sh FILES
  311. .Bl -tag -width __MAGIC__.mgc -compact
  312. .It Pa __MAGIC__.mgc
  313. Default compiled list of magic.
  314. .It Pa __MAGIC__
  315. Directory containing default magic files.
  316. .El
  318. The environment variable
  319. .Dv MAGIC
  320. can be used to set the default magic file name.
  321. If that variable is set, then
  322. .Nm
  323. will not attempt to open
  324. .Pa $HOME/.magic .
  325. .Nm
  326. adds
  327. .Dq .mgc
  328. to the value of this variable as appropriate.
  329. The environment variable
  331. controls (on systems that support symbolic links), whether
  332. .Nm
  333. will attempt to follow symlinks or not. If set, then
  334. .Nm
  335. follows symlink, otherwise it does not. This is also controlled
  336. by the
  337. .Fl L
  338. and
  339. .Fl h
  340. options.
  341. .Sh SEE ALSO
  342. .Xr magic __FSECTION__ ,
  343. .Xr strings 1 ,
  344. .Xr od 1 ,
  345. .Xr hexdump 1,
  346. .Xr file 1posix
  348. This program is believed to exceed the System V Interface Definition
  349. of FILE(CMD), as near as one can determine from the vague language
  350. contained therein.
  351. Its behavior is mostly compatible with the System V program of the same name.
  352. This version knows more magic, however, so it will produce
  353. different (albeit more accurate) output in many cases.
  354. .\" URL:
  355. .Pp
  356. The one significant difference
  357. between this version and System V
  358. is that this version treats any white space
  359. as a delimiter, so that spaces in pattern strings must be escaped.
  360. For example,
  361. .Bd -literal -offset indent
  362. >10 string language impress\ (imPRESS data)
  363. .Ed
  364. .Pp
  365. in an existing magic file would have to be changed to
  366. .Bd -literal -offset indent
  367. >10 string language\e impress (imPRESS data)
  368. .Ed
  369. .Pp
  370. In addition, in this version, if a pattern string contains a backslash,
  371. it must be escaped.
  372. For example
  373. .Bd -literal -offset indent
  374. 0 string \ebegindata Andrew Toolkit document
  375. .Ed
  376. .Pp
  377. in an existing magic file would have to be changed to
  378. .Bd -literal -offset indent
  379. 0 string \e\ebegindata Andrew Toolkit document
  380. .Ed
  381. .Pp
  382. SunOS releases 3.2 and later from Sun Microsystems include a
  383. .Nm
  384. command derived from the System V one, but with some extensions.
  385. My version differs from Sun's only in minor ways.
  386. It includes the extension of the
  387. .Sq &
  388. operator, used as,
  389. for example,
  390. .Bd -literal -offset indent
  391. >16 long&0x7fffffff >0 not stripped
  392. .Ed
  394. The magic file entries have been collected from various sources,
  395. mainly USENET, and contributed by various authors.
  396. Christos Zoulas (address below) will collect additional
  397. or corrected magic file entries.
  398. A consolidation of magic file entries
  399. will be distributed periodically.
  400. .Pp
  401. The order of entries in the magic file is significant.
  402. Depending on what system you are using, the order that
  403. they are put together may be incorrect.
  404. If your old
  405. .Nm
  406. command uses a magic file,
  407. keep the old magic file around for comparison purposes
  408. (rename it to
  409. .Pa __MAGIC__.orig ).
  410. .Sh EXAMPLES
  411. .Bd -literal -offset indent
  412. $ file file.c file /dev/{wd0a,hda}
  413. file.c: C program text
  414. file: ELF 32-bit LSB executable, Intel 80386, version 1 (SYSV),
  415. dynamically linked (uses shared libs), stripped
  416. /dev/wd0a: block special (0/0)
  417. /dev/hda: block special (3/0)
  418. $ file -s /dev/wd0{b,d}
  419. /dev/wd0b: data
  420. /dev/wd0d: x86 boot sector
  421. $ file -s /dev/hda{,1,2,3,4,5,6,7,8,9,10}
  422. /dev/hda: x86 boot sector
  423. /dev/hda1: Linux/i386 ext2 filesystem
  424. /dev/hda2: x86 boot sector
  425. /dev/hda3: x86 boot sector, extended partition table
  426. /dev/hda4: Linux/i386 ext2 filesystem
  427. /dev/hda5: Linux/i386 swap file
  428. /dev/hda6: Linux/i386 swap file
  429. /dev/hda7: Linux/i386 swap file
  430. /dev/hda8: Linux/i386 swap file
  431. /dev/hda9: empty
  432. /dev/hda10: empty
  433. $ file -i file.c file /dev/{wd0a,hda}
  434. file.c: text/x-c
  435. file: application/x-executable
  436. /dev/hda: application/x-not-regular-file
  437. /dev/wd0a: application/x-not-regular-file
  438. .Ed
  439. .Sh HISTORY
  440. There has been a
  441. .Nm
  442. command in every
  443. .Dv UNIX since at least Research Version 4
  444. (man page dated November, 1973).
  445. The System V version introduced one significant major change:
  446. the external list of magic types.
  447. This slowed the program down slightly but made it a lot more flexible.
  448. .Pp
  449. This program, based on the System V version,
  450. was written by Ian Darwin <>
  451. without looking at anybody else's source code.
  452. .Pp
  453. John Gilmore revised the code extensively, making it better than
  454. the first version.
  455. Geoff Collyer found several inadequacies
  456. and provided some magic file entries.
  457. Contributions by the `&' operator by Rob McMahon,, 1989.
  458. .Pp
  459. Guy Harris,, made many changes from 1993 to the present.
  460. .Pp
  461. Primary development and maintenance from 1990 to the present by
  462. Christos Zoulas (
  463. .Pp
  464. Altered by Chris Lowth,, 2000:
  465. Handle the
  466. .Fl i
  467. option to output mime type strings, using an alternative
  468. magic file and internal logic.
  469. .Pp
  470. Altered by Eric Fischer (, July, 2000,
  471. to identify character codes and attempt to identify the languages
  472. of non-ASCII files.
  473. .Pp
  474. Altered by Reuben Thomas (, 2007 to 2008, to improve MIME
  475. support and merge MIME and non-MIME magic, support directories as well
  476. as files of magic, apply many bug fixes and improve the build system.
  477. .Pp
  478. The list of contributors to the
  479. .Dq magic
  480. directory (magic files)
  481. is too long to include here.
  482. You know who you are; thank you.
  483. Many contributors are listed in the source files.
  485. Copyright (c) Ian F. Darwin, Toronto, Canada, 1986-1999.
  486. Covered by the standard Berkeley Software Distribution copyright; see the file
  487. LEGAL.NOTICE in the source distribution.
  488. .Pp
  489. The files
  490. .Dv tar.h
  491. and
  492. .Dv is_tar.c
  493. were written by John Gilmore from his public-domain
  494. .Xr tar 1
  495. program, and are not covered by the above license.
  496. .Sh BUGS
  497. .Pp
  498. There must be a better way to automate the construction of the Magic
  499. file from all the glop in Magdir.
  500. What is it?
  501. .Pp
  502. .Nm
  503. uses several algorithms that favor speed over accuracy,
  504. thus it can be misled about the contents of
  505. text
  506. files.
  507. .Pp
  508. The support for text files (primarily for programming languages)
  509. is simplistic, inefficient and requires recompilation to update.
  510. .Pp
  511. The list of keywords in
  512. .Dv ascmagic
  513. probably belongs in the Magic file.
  514. This could be done by using some keyword like
  515. .Sq *
  516. for the offset value.
  517. .Pp
  518. Complain about conflicts in the magic file entries.
  519. Make a rule that the magic entries sort based on file offset rather
  520. than position within the magic file?
  521. .Pp
  522. The program should provide a way to give an estimate
  523. of
  524. .Dq how good
  525. a guess is.
  526. We end up removing guesses (e.g.
  527. .Dq From\
  528. as first 5 chars of file) because
  529. they are not as good as other guesses (e.g.
  530. .Dq Newsgroups:
  531. versus
  532. .Dq Return-Path:
  533. ).
  534. Still, if the others don't pan out, it should be possible to use the
  535. first guess.
  536. .Pp
  537. This manual page, and particularly this section, is too long.
  538. .Sh RETURN CODE
  539. .Nm
  540. returns 0 on success, and non-zero on error.
  542. You can obtain the original author's latest version by anonymous FTP
  543. on
  544. .Dv
  545. in the directory
  546. .Dv /pub/file/file-X.YZ.tar.gz