contrib/file/magic.mime

   1 # Magic data for KMimeMagic (originally for file(1) command)
   2 #
   3 # The format is 4-5 columns:
   4 #    Column #1: byte number to begin checking from, ">" indicates continuation
   5 #    Column #2: type of data to match
   6 #    Column #3: contents of data to match
   7 #    Column #4: MIME type of result
   8 #    Column #5: MIME encoding of result (optional)
   9
  10 #------------------------------------------------------------------------------
  11 # Localstuff:  file(1) magic for locally observed files
  12 # Add any locally observed files here.
  13
  14 #------------------------------------------------------------------------------
  15 # end local stuff
  16 #------------------------------------------------------------------------------
  17
  18 #------------------------------------------------------------------------------
  19 # Java
  20
  21 0       short           0xcafe
  22 >2      short           0xbabe          application/java
  23
  24 #------------------------------------------------------------------------------
  25 # audio:  file(1) magic for sound formats
  26 #
  27 # from Jan Nicolai Langfeldt <janl@ifi.uio.no>,
  28 #
  29
  30 # Sun/NeXT audio data
  31 0       string          .snd
  32 >12     belong          1               audio/basic
  33 >12     belong          2               audio/basic
  34 >12     belong          3               audio/basic
  35 >12     belong          4               audio/basic
  36 >12     belong          5               audio/basic
  37 >12     belong          6               audio/basic
  38 >12     belong          7               audio/basic
  39
  40 >12     belong          23              audio/x-adpcm
  41
  42 # DEC systems (e.g. DECstation 5000) use a variant of the Sun/NeXT format
  43 # that uses little-endian encoding and has a different magic number
  44 # (0x0064732E in little-endian encoding).
  45 0       lelong          0x0064732E
  46 >12     lelong          1               audio/x-dec-basic
  47 >12     lelong          2               audio/x-dec-basic
  48 >12     lelong          3               audio/x-dec-basic
  49 >12     lelong          4               audio/x-dec-basic
  50 >12     lelong          5               audio/x-dec-basic
  51 >12     lelong          6               audio/x-dec-basic
  52 >12     lelong          7               audio/x-dec-basic
  53 #                                       compressed (G.721 ADPCM)
  54 >12     lelong          23              audio/x-dec-adpcm
  55
  56 # Bytes 0-3 of AIFF, AIFF-C, & 8SVX audio files are "FORM"
  57 #                                       AIFF audio data
  58 8       string          AIFF            audio/x-aiff
  59 #                                       AIFF-C audio data
  60 8       string          AIFC            audio/x-aiff
  61 #                                       IFF/8SVX audio data
  62 8       string          8SVX            audio/x-aiff
  63
  64 # Creative Labs AUDIO stuff
  65 #                                       Standard MIDI data
  66 0       string  MThd                    audio/unknown
  67 #>9     byte    >0                      (format %d)
  68 #>11    byte    >1                      using %d channels
  69 #                                       Creative Music (CMF) data
  70 0       string  CTMF                    audio/unknown
  71 #                                       SoundBlaster instrument data
  72 0       string  SBI                     audio/unknown
  73 #                                       Creative Labs voice data
  74 0       string  Creative\ Voice\ File   audio/unknown
  75 ## is this next line right?  it came this way...
  76 #>19    byte    0x1A
  77 #>23    byte    >0                      - version %d
  78 #>22    byte    >0                      \b.%d
  79
  80 # [GRR 950115:  is this also Creative Labs?  Guessing that first line
  81 #  should be string instead of unknown-endian long...]
  82 #0      long            0x4e54524b      MultiTrack sound data
  83 #0      string          NTRK            MultiTrack sound data
  84 #>4     long            x               - version %ld
  85
  86 # Microsoft WAVE format (*.wav)
  87 # [GRR 950115:  probably all of the shorts and longs should be leshort/lelong]
  88 #                                       Microsoft RIFF
  89 0       string          RIFF            audio/unknown
  90 #                                       - WAVE format
  91 >8      string          WAVE            audio/x-wav
  92 >8      string          AVI             video/x-msvideo
  93 #
  94 0       belong          0x2e7261fd      application/x-realaudio
  95
  96 # MPEG Layer 3 sound files
  97 # Modified the 11/20/97 at 15:59:04 by Christophe Prud'homme <christophe.prudhomme@asci.fr>
  98 0       belong          0xfffb          audio/x-mp3
  99 #MP3 with ID3 tag
 100 0       string          ID3             audio/x-mp3
 101 # Ogg/Vorbis
 102 0       string          OggS            audio/x-ogg
 103
 104 #------------------------------------------------------------------------------
 105 # c-lang:  file(1) magic for C programs or various scripts
 106 #
 107
 108 # XPM icons (Greg Roelofs, newt@uchicago.edu)
 109 # ideally should go into "images", but entries below would tag XPM as C source
 110 0       string          /*\ XPM         image/x-xpm     7bit
 111
 112 # 3DS (3d Studio files)
 113 16      beshort         0x3d3d          image/x-3ds
 114
 115 # this first will upset you if you're a PL/1 shop... (are there any left?)
 116 # in which case rm it; ascmagic will catch real C programs
 117 #                                       C or REXX program text
 118 #0      string          /*              text/x-c
 119 #                                       C++ program text
 120 #0      string          //              text/x-c++
 121
 122 #------------------------------------------------------------------------------
 123 # commands:  file(1) magic for various shells and interpreters
 124 #
 125 #0       string          :\ shell archive or commands for antique kernel text
 126 0       string          #!/bin/sh               application/x-shellscript
 127 0       string          #!\ /bin/sh             application/x-shellscript
 128 0       string          #!/bin/csh              application/x-shellscript
 129 0       string          #!\ /bin/csh            application/x-shellscript
 130 # korn shell magic, sent by George Wu, gwu@clyde.att.com
 131 0       string          #!/bin/ksh              application/x-shellscript
 132 0       string          #!\ /bin/ksh            application/x-shellscript
 133 0       string          #!/bin/tcsh             application/x-shellscript
 134 0       string          #!\ /bin/tcsh           application/x-shellscript
 135 0       string          #!/usr/local/tcsh       application/x-shellscript
 136 0       string          #!\ /usr/local/tcsh     application/x-shellscript
 137 0       string          #!/usr/local/bin/tcsh   application/x-shellscript
 138 0       string          #!\ /usr/local/bin/tcsh application/x-shellscript
 139 # bash shell magic, from Peter Tobias (tobias@server.et-inf.fho-emden.de)
 140 0       string          #!/bin/bash                     application/x-shellscript
 141 0       string          #!\ /bin/bash           application/x-shellscript
 142 0       string          #!/usr/local/bin/bash   application/x-shellscript
 143 0       string          #!\ /usr/local/bin/bash application/x-shellscript
 144
 145 #
 146 # zsh/ash/ae/nawk/gawk magic from cameron@cs.unsw.oz.au (Cameron Simpson)
 147 0       string          #!/usr/local/bin/zsh    application/x-shellscript
 148 0       string          #!\ /usr/local/bin/zsh  application/x-shellscript
 149 0       string          #!/usr/local/bin/ash    application/x-shellscript
 150 0       string          #!\ /usr/local/bin/ash  application/x-shellscript
 151 #0       string          #!/usr/local/bin/ae     Neil Brown's ae
 152 #0       string          #!\ /usr/local/bin/ae   Neil Brown's ae
 153 0       string          #!/bin/nawk             application/x-nawk
 154 0       string          #!\ /bin/nawk           application/x-nawk
 155 0       string          #!/usr/bin/nawk         application/x-nawk
 156 0       string          #!\ /usr/bin/nawk       application/x-nawk
 157 0       string          #!/usr/local/bin/nawk   application/x-nawk
 158 0       string          #!\ /usr/local/bin/nawk application/x-nawk
 159 0       string          #!/bin/gawk             application/x-gawk
 160 0       string          #!\ /bin/gawk           application/x-gawk
 161 0       string          #!/usr/bin/gawk         application/x-gawk
 162 0       string          #!\ /usr/bin/gawk       application/x-gawk
 163 0       string          #!/usr/local/bin/gawk   application/x-gawk
 164 0       string          #!\ /usr/local/bin/gawk application/x-gawk
 165 #
 166 0       string          #!/bin/awk              application/x-awk
 167 0       string          #!\ /bin/awk            application/x-awk
 168 0       string          #!/usr/bin/awk          application/x-awk
 169 0       string          #!\ /usr/bin/awk        application/x-awk
 170 0       string          BEGIN                   application/x-awk
 171
 172 # For Larry Wall's perl language.  The ``eval'' line recognizes an
 173 # outrageously clever hack for USG systems.
 174 #                               Keith Waclena <keith@cerberus.uchicago.edu>
 175 0       string          #!/bin/perl                     application/x-perl
 176 0       string          #!\ /bin/perl                   application/x-perl
 177 0       string          eval\ "exec\ /bin/perl          application/x-perl
 178 0       string          #!/usr/bin/perl                 application/x-perl
 179 0       string          #!\ /usr/bin/perl               application/x-perl
 180 0       string          eval\ "exec\ /usr/bin/perl      application/x-perl
 181 0       string          #!/usr/local/bin/perl           application/x-perl
 182 0       string          #!\ /usr/local/bin/perl         application/x-perl
 183 0       string          eval\ "exec\ /usr/local/bin/perl application/x-perl
 184
 185 #------------------------------------------------------------------------------
 186 # compress:  file(1) magic for pure-compression formats (no archives)
 187 #
 188 # compress, gzip, pack, compact, huf, squeeze, crunch, freeze, yabba, whap, etc.
 189 #
 190 # Formats for various forms of compressed data
 191 # Formats for "compress" proper have been moved into "compress.c",
 192 # because it tries to uncompress it to figure out what's inside.
 193
 194 # standard unix compress
 195 0       string          \037\235        application/x-compress
 196
 197 # gzip (GNU zip, not to be confused with [Info-ZIP/PKWARE] zip archiver)
 198 0       string          \037\213        application/x-gzip
 199
 200 0               string                  PK\003\004              application/x-zip
 201
 202 # According to gzip.h, this is the correct byte order for packed data.
 203 0       string          \037\036        application/octet-stream
 204 #
 205 # This magic number is byte-order-independent.
 206 #
 207 0       short           017437          application/octet-stream
 208
 209 # XXX - why *two* entries for "compacted data", one of which is
 210 # byte-order independent, and one of which is byte-order dependent?
 211 #
 212 # compacted data
 213 0       short           0x1fff          application/octet-stream
 214 0       string          \377\037        application/octet-stream
 215 # huf output
 216 0       short           0145405         application/octet-stream
 217
 218 # Squeeze and Crunch...
 219 # These numbers were gleaned from the Unix versions of the programs to
 220 # handle these formats.  Note that I can only uncrunch, not crunch, and
 221 # I didn't have a crunched file handy, so the crunch number is untested.
 222 #                               Keith Waclena <keith@cerberus.uchicago.edu>
 223 #0      leshort         0x76FF          squeezed data (CP/M, DOS)
 224 #0      leshort         0x76FE          crunched data (CP/M, DOS)
 225
 226 # Freeze
 227 #0      string          \037\237        Frozen file 2.1
 228 #0      string          \037\236        Frozen file 1.0 (or gzip 0.5)
 229
 230 # lzh?
 231 #0      string          \037\240        LZH compressed data
 232
 233 257     string          ustar\0         application/x-tar       posix
 234 257     string          ustar\040\040\0         application/x-tar       gnu
 235
 236 0       short           070707          application/x-cpio
 237 0       short           0143561         application/x-cpio      swapped
 238
 239 0       string          =<ar>           application/x-archive
 240 0       string          !<arch>         application/x-archive
 241 >8      string          debian          application/x-debian-package
 242
 243 #------------------------------------------------------------------------------
 244 #
 245 # RPM: file(1) magic for Red Hat Packages   Erik Troan (ewt@redhat.com)
 246 #
 247 0       beshort         0xedab
 248 >2      beshort         0xeedb          application/x-rpm
 249
 250 0       lelong&0x8080ffff       0x0000081a      application/x-arc       lzw
 251 0       lelong&0x8080ffff       0x0000091a      application/x-arc       squashed
 252 0       lelong&0x8080ffff       0x0000021a      application/x-arc       uncompressed
 253 0       lelong&0x8080ffff       0x0000031a      application/x-arc       packed
 254 0       lelong&0x8080ffff       0x0000041a      application/x-arc       squeezed
 255 0       lelong&0x8080ffff       0x0000061a      application/x-arc       crunched
 256
 257 0       leshort 0xea60  application/octet-stream        x-arj
 258
 259 # LHARC/LHA archiver (Greg Roelofs, newt@uchicago.edu)
 260 2       string  -lh0-   application/x-lharc     lh0
 261 2       string  -lh1-   application/x-lharc     lh1
 262 2       string  -lz4-   application/x-lharc     lz4
 263 2       string  -lz5-   application/x-lharc     lz5
 264 #       [never seen any but the last; -lh4- reported in comp.compression:]
 265 2       string  -lzs-   application/x-lha       lzs
 266 2       string  -lh\ -  application/x-lha       lh
 267 2       string  -lhd-   application/x-lha       lhd
 268 2       string  -lh2-   application/x-lha       lh2
 269 2       string  -lh3-   application/x-lha       lh3
 270 2       string  -lh4-   application/x-lha       lh4
 271 2       string  -lh5-   application/x-lha       lh5
 272 2       string  -lh6-   application/x-lha       lh6
 273 2       string  -lh7-   application/x-lha       lh7
 274 # Shell archives
 275 10      string  #\ This\ is\ a\ shell\ archive  application/octet-stream        x-shell
 276
 277 #------------------------------------------------------------------------------
 278 # frame:  file(1) magic for FrameMaker files
 279 #
 280 # This stuff came on a FrameMaker demo tape, most of which is
 281 # copyright, but this file is "published" as witness the following:
 282 #
 283 0       string          \<MakerFile     application/x-frame
 284 0       string          \<MIFFile       application/x-frame
 285 0       string          \<MakerDictionary       application/x-frame
 286 0       string          \<MakerScreenFon        application/x-frame
 287 0       string          \<MML           application/x-frame
 288 0       string          \<Book          application/x-frame
 289 0       string          \<Maker         application/x-frame
 290
 291 #------------------------------------------------------------------------------
 292 # html:  file(1) magic for HTML (HyperText Markup Language) docs
 293 #
 294 # from Daniel Quinlan <quinlan@yggdrasil.com>
 295 #
 296 0       string          \<HEAD  text/html
 297 0       string          \<head  text/html
 298 0       string          \<TITLE text/html
 299 0       string          \<title text/html
 300 0       string          \<html  text/html
 301 0       string          \<HTML  text/html
 302 0       string          \<!--   text/html
 303 0       string          \<h1    text/html
 304 0       string          \<H1    text/html
 305 0       string          \<!doctype\ HTML        text/html
 306 0       string          \<!DOCTYPE\ HTML        text/html
 307 0       string          \<!doctype\ html        text/html
 308
 309 #------------------------------------------------------------------------------
 310 # images:  file(1) magic for image formats (see also "c-lang" for XPM bitmaps)
 311 #
 312 # originally from jef@helios.ee.lbl.gov (Jef Poskanzer),
 313 # additions by janl@ifi.uio.no as well as others. Jan also suggested
 314 # merging several one- and two-line files into here.
 315 #
 316 # XXX - byte order for GIF and TIFF fields?
 317 # [GRR:  TIFF allows both byte orders; GIF is probably little-endian]
 318 #
 319
 320 # [GRR:  what the hell is this doing in here?]
 321 #0      string          xbtoa           btoa'd file
 322
 323 # PBMPLUS
 324 #                                       PBM file
 325 0       string          P1              image/x-portable-bitmap 7bit
 326 #                                       PGM file
 327 0       string          P2              image/x-portable-greymap        7bit
 328 #                                       PPM file
 329 0       string          P3              image/x-portable-pixmap 7bit
 330 #                                       PBM "rawbits" file
 331 0       string          P4              image/x-portable-bitmap
 332 #                                       PGM "rawbits" file
 333 0       string          P5              image/x-portable-greymap
 334 #                                       PPM "rawbits" file
 335 0       string          P6              image/x-portable-pixmap
 336
 337 # NIFF (Navy Interchange File Format, a modification of TIFF)
 338 # [GRR:  this *must* go before TIFF]
 339 0       string          IIN1            image/x-niff
 340
 341 # TIFF and friends
 342 #                                       TIFF file, big-endian
 343 0       string          MM              image/tiff
 344 #                                       TIFF file, little-endian
 345 0       string          II              image/tiff
 346
 347 # possible GIF replacements; none yet released!
 348 # (Greg Roelofs, newt@uchicago.edu)
 349 #
 350 # GRR 950115:  this was mine ("Zip GIF"):
 351 #                                       ZIF image (GIF+deflate alpha)
 352 0       string          GIF94z          image/unknown
 353 #
 354 # GRR 950115:  this is Jeremy Wohl's Free Graphics Format (better):
 355 #                                       FGF image (GIF+deflate beta)
 356 0       string          FGF95a          image/unknown
 357 #
 358 # GRR 950115:  this is Thomas Boutell's Portable Bitmap Format proposal
 359 # (best; not yet implemented):
 360 #                                       PBF image (deflate compression)
 361 0       string          PBF             image/unknown
 362
 363 # GIF
 364 0       string          GIF             image/gif
 365
 366 # JPEG images
 367 0       beshort         0xffd8          image/jpeg
 368
 369 # PC bitmaps (OS/2, Windoze BMP files)  (Greg Roelofs, newt@uchicago.edu)
 370 0       string          BM              image/bmp
 371 #>14    byte            12              (OS/2 1.x format)
 372 #>14    byte            64              (OS/2 2.x format)
 373 #>14    byte            40              (Windows 3.x format)
 374 #0      string          IC              icon
 375 #0      string          PI              pointer
 376 #0      string          CI              color icon
 377 #0      string          CP              color pointer
 378 #0      string          BA              bitmap array
 379
 380
 381 #------------------------------------------------------------------------------
 382 # lisp:  file(1) magic for lisp programs
 383 #
 384 # various lisp types, from Daniel Quinlan (quinlan@yggdrasil.com)
 385 0       string  ;;                      text/plain      8bit
 386 # Emacs 18 - this is always correct, but not very magical.
 387 0       string  \012(                   application/x-elc
 388 # Emacs 19
 389 0       string  ;ELC\023\000\000\000    application/x-elc
 390
 391 #------------------------------------------------------------------------------
 392 # mail.news:  file(1) magic for mail and news
 393 #
 394 # There are tests to ascmagic.c to cope with mail and news.
 395 0       string          Relay-Version:  message/rfc822  7bit
 396 0       string          #!\ rnews       message/rfc822  7bit
 397 0       string          N#!\ rnews      message/rfc822  7bit
 398 0       string          Forward\ to     message/rfc822  7bit
 399 0       string          Pipe\ to        message/rfc822  7bit
 400 0       string          Return-Path:    message/rfc822  7bit
 401 0       string          Path:           message/news    8bit
 402 0       string          Xref:           message/news    8bit
 403 0       string          From:           message/rfc822  7bit
 404 0       string          Article         message/news    8bit
 405 #------------------------------------------------------------------------------
 406 # msword: file(1) magic for MS Word files
 407 #
 408 # Contributor claims:
 409 # Reversed-engineered MS Word magic numbers
 410 #
 411
 412 0       string          \376\067\0\043                  application/msword
 413 0       string          \320\317\021\340\241\261        application/msword
 414 0       string          \333\245-\0\0\0                 application/msword
 415
 416
 417
 418 #------------------------------------------------------------------------------
 419 # printer:  file(1) magic for printer-formatted files
 420 #
 421
 422 # PostScript
 423 0       string          %!              application/postscript
 424 0       string          \004%!          application/postscript
 425
 426 # Acrobat
 427 # (due to clamen@cs.cmu.edu)
 428 0       string          %PDF-           application/pdf
 429
 430 #------------------------------------------------------------------------------
 431 # sc:  file(1) magic for "sc" spreadsheet
 432 #
 433 38      string          Spreadsheet     application/x-sc
 434
 435 #------------------------------------------------------------------------------
 436 # tex:  file(1) magic for TeX files
 437 #
 438 # XXX - needs byte-endian stuff (big-endian and little-endian DVI?)
 439 #
 440 # From <conklin@talisman.kaleida.com>
 441
 442 # Although we may know the offset of certain text fields in TeX DVI
 443 # and font files, we can't use them reliably because they are not
 444 # zero terminated. [but we do anyway, christos]
 445 0       string          \367\002        application/x-dvi
 446 #0      string          \367\203        TeX generic font data
 447 #0      string          \367\131        TeX packed font data
 448 #0      string          \367\312        TeX virtual font data
 449 #0      string          This\ is\ TeX,  TeX transcript text
 450 #0      string          This\ is\ METAFONT,     METAFONT transcript text
 451
 452 # There is no way to detect TeX Font Metric (*.tfm) files without
 453 # breaking them apart and reading the data.  The following patterns
 454 # match most *.tfm files generated by METAFONT or afm2tfm.
 455 2       string          \000\021        application/x-tex-tfm
 456 2       string          \000\022        application/x-tex-tfm
 457 #>34    string          >\0             (%s)
 458
 459 # Texinfo and GNU Info, from Daniel Quinlan (quinlan@yggdrasil.com)
 460 #0      string          \\input\ texinfo        Texinfo source text
 461 #0      string          This\ is\ Info\ file    GNU Info text
 462
 463 # correct TeX magic for Linux (and maybe more)
 464 # from Peter Tobias (tobias@server.et-inf.fho-emden.de)
 465 #
 466 0       leshort         0x02f7          application/x-dvi
 467
 468 # RTF - Rich Text Format
 469 0       string          {\\rtf          text/rtf
 470
 471 #------------------------------------------------------------------------------
 472 # animation:  file(1) magic for animation/movie formats
 473 #
 474 # animation formats, originally from vax@ccwf.cc.utexas.edu (VaX#n8)
 475 #                                               MPEG file
 476 0       belong          0x000001b3                      video/mpeg
 477 0       belong          0x000001ba                      video/mpeg
 478 # FLI animation format
 479 0       leshort         0xAF11                          video/fli
 480 # FLC animation format
 481 0       leshort         0xAF12                          video/flc
 482 # AVI
 483 >8      string          AVI\                            video/avi
 484 #
 485 # SGI and Apple formats
 486 #
 487 0       string          MOVI                            video/sgi
 488 4       string          moov                            video/quicktime moov
 489 4       string          mdat                            video/quicktime mdat
 490 # The contributor claims:
 491 #   I couldn't find a real magic number for these, however, this
 492 #   -appears- to work.  Note that it might catch other files, too,
 493 #   so BE CAREFUL!
 494 #
 495 # Note that title and author appear in the two 20-byte chunks
 496 # at decimal offsets 2 and 22, respectively, but they are XOR'ed with
 497 # 255 (hex FF)! DL format SUCKS BIG ROCKS.
 498 #
 499 #                                               DL file version 1 , medium format (160x100, 4 images/screen)
 500 0       byte            1                       video/unknown
 501 0       byte            2                       video/unknown
 502 #
 503 # Databases
 504 #
 505 # GDBM magic numbers
 506 #  Will be maintained as part of the GDBM distribution in the future.
 507 #  <downsj@teeny.org>
 508 0       belong  0x13579ace      application/x-gdbm
 509 0       lelong  0x13579ace      application/x-gdbm
 510 0       string  GDBM            application/x-gdbm
 511 #
 512 0       belong  0x061561        application/x-dbm
 513 #
 514 # Executables
 515 #
 516 0       string          \177ELF
 517 >4      byte            0
 518 >4      byte            1
 519 >4      byte            2
 520 >5      byte            0
 521 >5      byte            1
 522 >>16    leshort         0
 523 >>16    leshort         1               application/x-object
 524 >>16    leshort         2               application/x-executable
 525 >>16    leshort         3               application/x-sharedlib
 526 >>16    leshort         4               application/x-coredump
 527 #
 528 # DOS
 529 0               string                  MZ                              application/x-dosexec
 530 #
 531 # KDE
 532 0               string  [KDE\ Desktop\ Entry]   application/x-kdelnk
 533 0               string  \#\ KDE\ Config\ File   application/x-kdelnk
 534 # xmcd database file for kscd
 535 0               string  \#\ xmcd                text/xmcd
 536
 537 #------------------------------------------------------------------------------
 538 # pkgadd:  file(1) magic for SysV R4 PKG Datastreams
 539 #
 540 0       string          #\ PaCkAgE\ DaTaStReAm  application/x-svr4-package
 541
 542 #PNG Image Format
 543 0       string          \x89PNG                 image/png