You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@tika.apache.org by ju...@apache.org on 2009/09/27 17:48:35 UTC
svn commit: r819328 -
/lucene/tika/trunk/tika-core/src/main/resources/org/apache/tika/mime/tika-mimetypes.xml
Author: jukka
Date: Sun Sep 27 15:48:35 2009
New Revision: 819328
URL: http://svn.apache.org/viewvc?rev=819328&view=rev
Log:
TIKA-285: Update media type registry to the latest httpd mime type database
More merging of the mime.types information. Work in progress...
Modified:
lucene/tika/trunk/tika-core/src/main/resources/org/apache/tika/mime/tika-mimetypes.xml
Modified: lucene/tika/trunk/tika-core/src/main/resources/org/apache/tika/mime/tika-mimetypes.xml
URL: http://svn.apache.org/viewvc/lucene/tika/trunk/tika-core/src/main/resources/org/apache/tika/mime/tika-mimetypes.xml?rev=819328&r1=819327&r2=819328&view=diff
==============================================================================
--- lucene/tika/trunk/tika-core/src/main/resources/org/apache/tika/mime/tika-mimetypes.xml (original)
+++ lucene/tika/trunk/tika-core/src/main/resources/org/apache/tika/mime/tika-mimetypes.xml Sun Sep 27 15:48:35 2009
@@ -801,103 +801,6 @@
</magic>
</mime-type>
- <mime-type type="image/vnd.wap.wbmp">
- <_comment>Wireless Bitmap File Format</_comment>
- <glob pattern="*.wbmp" />
- </mime-type>
-
- <mime-type type="image/x-psd">
- <alias type="image/photoshop" />
- </mime-type>
-
- <mime-type type="image/x-xcf">
- <alias type="image/xcf" />
- <magic priority="50">
- <match type="string" value="gimp xcf " offset="0" />
- </magic>
- </mime-type>
-
- <mime-type type="model/iges">
- <_comment>Initial Graphics Exchange Specification Format</_comment>
- <glob pattern="*.igs" />
- <glob pattern="*.iges" />
- </mime-type>
-
- <mime-type type="model/mesh">
- <glob pattern="*.msh" />
- <glob pattern="*.mesh" />
- <glob pattern="*.silo" />
- </mime-type>
-
- <mime-type type="model/vrml">
- <glob pattern="*.vrml" />
- </mime-type>
-
- <mime-type type="text/x-tcl">
- <alias type="application/x-tcl" />
- </mime-type>
-
- <mime-type type="text/x-tex">
- <magic priority="50">
- <match value="\\input" type="string" offset="0" />
- <match value="\\section" type="string" offset="0" />
- <match value="\\setlength" type="string" offset="0" />
- <match value="\\documentstyle" type="string" offset="0" />
- <match value="\\chapter" type="string" offset="0" />
- <match value="\\documentclass" type="string" offset="0" />
- <match value="\\relax" type="string" offset="0" />
- <match value="\\contentsline" type="string" offset="0" />
- </magic>
- <alias type="application/x-tex" />
- </mime-type>
-
- <mime-type type="text/x-texinfo">
- <magic priority="50">
- <match value="\\input\ texinfo" type="string" offset="0" />
- </magic>
- <alias type="application/x-texinfo" />
- </mime-type>
-
- <mime-type type="text/x-troff-me">
- <alias type="application/x-troff-me" />
- </mime-type>
-
- <mime-type type="video/vnd.mpegurl">
- <glob pattern="*.mxu" />
- </mime-type>
-
- <mime-type type="x-conference/x-cooltalk">
- <_comment>Cooltalk Audio</_comment>
- <glob pattern="*.ice" />
- </mime-type>
-
- <mime-type type="audio/mpeg">
- <acronym>MP3</acronym>
- <comment>MPEG-1 Audio Layer 3</comment>
- <magic priority="20">
- <!-- http://mpgedit.org/mpgedit/mpeg_format/MP3Format.html -->
- <!-- Bit pattern for first two bytes: 11111111 111VVLLC -->
- <!-- VV = MPEG Audio Version ID; 10 = V2, 11 = V1 -->
- <!-- LL = Layer description; 01 = L3, 10 = L2, 11 = L1 -->
- <!-- C = Protection bit; 0 = CRC, 1 = no CRC -->
- <match value="0xfff2" type="string" offset="0" /> <!-- V2, L3, CRC -->
- <match value="0xfff3" type="string" offset="0" /> <!-- V2, L3 -->
- <match value="0xfff4" type="string" offset="0" /> <!-- V2, L2, CRC -->
- <match value="0xfff5" type="string" offset="0" /> <!-- V2, L2 -->
- <match value="0xfff6" type="string" offset="0" /> <!-- V2, L1, CRC -->
- <match value="0xfff7" type="string" offset="0" /> <!-- V2, L1 -->
- <match value="0xfffa" type="string" offset="0" /> <!-- V1, L3, CRC -->
- <match value="0xfffb" type="string" offset="0" /> <!-- V1, L3 -->
- <match value="0xfffc" type="string" offset="0" /> <!-- V1, L2, CRC -->
- <match value="0xfffd" type="string" offset="0" /> <!-- V1, L2 -->
- <match value="0xfffe" type="string" offset="0" /> <!-- V1, L1, CRC -->
- <match value="0xffff" type="string" offset="0" /> <!-- V1, L1 -->
- <match value="ID3" type="string" offset="0" />
- </magic>
- <glob pattern="*.mp3" />
- </mime-type>
-
-
<mime-type type="application/activemessage"/>
<mime-type type="application/andrew-inset">
<glob pattern="*.ez"/>
@@ -2936,19 +2839,41 @@
<mime-type type="application/x-tar">
<glob pattern="*.tar"/>
</mime-type>
+
<mime-type type="application/x-tcl">
+ <alias type="text/x-tcl"/>
<glob pattern="*.tcl"/>
+ <sub-class-of type="text/plain"/>
</mime-type>
+
<mime-type type="application/x-tex">
+ <alias type="text/x-tex"/>
+ <magic priority="50">
+ <match value="\\input" type="string" offset="0"/>
+ <match value="\\section" type="string" offset="0"/>
+ <match value="\\setlength" type="string" offset="0"/>
+ <match value="\\documentstyle" type="string" offset="0"/>
+ <match value="\\chapter" type="string" offset="0"/>
+ <match value="\\documentclass" type="string" offset="0"/>
+ <match value="\\relax" type="string" offset="0"/>
+ <match value="\\contentsline" type="string" offset="0"/>
+ </magic>
<glob pattern="*.tex"/>
</mime-type>
+
<mime-type type="application/x-tex-tfm">
<glob pattern="*.tfm"/>
</mime-type>
+
<mime-type type="application/x-texinfo">
+ <alias type="text/x-texinfo" />
+ <magic priority="50">
+ <match value="\\input\ texinfo" type="string" offset="0"/>
+ </magic>
<glob pattern="*.texinfo"/>
<glob pattern="*.texi"/>
</mime-type>
+
<mime-type type="application/x-ustar">
<glob pattern="*.ustar"/>
</mime-type>
@@ -3132,7 +3057,30 @@
<mime-type type="audio/mp4a-latm"/>
<mime-type type="audio/mpa"/>
<mime-type type="audio/mpa-robust"/>
+
<mime-type type="audio/mpeg">
+ <acronym>MP3</acronym>
+ <comment>MPEG-1 Audio Layer 3</comment>
+ <magic priority="20">
+ <!-- http://mpgedit.org/mpgedit/mpeg_format/MP3Format.html -->
+ <!-- Bit pattern for first two bytes: 11111111 111VVLLC -->
+ <!-- VV = MPEG Audio Version ID; 10 = V2, 11 = V1 -->
+ <!-- LL = Layer description; 01 = L3, 10 = L2, 11 = L1 -->
+ <!-- C = Protection bit; 0 = CRC, 1 = no CRC -->
+ <match value="0xfff2" type="string" offset="0"/> <!-- V2, L3, CRC -->
+ <match value="0xfff3" type="string" offset="0"/> <!-- V2, L3 -->
+ <match value="0xfff4" type="string" offset="0"/> <!-- V2, L2, CRC -->
+ <match value="0xfff5" type="string" offset="0"/> <!-- V2, L2 -->
+ <match value="0xfff6" type="string" offset="0"/> <!-- V2, L1, CRC -->
+ <match value="0xfff7" type="string" offset="0"/> <!-- V2, L1 -->
+ <match value="0xfffa" type="string" offset="0"/> <!-- V1, L3, CRC -->
+ <match value="0xfffb" type="string" offset="0"/> <!-- V1, L3 -->
+ <match value="0xfffc" type="string" offset="0"/> <!-- V1, L2, CRC -->
+ <match value="0xfffd" type="string" offset="0"/> <!-- V1, L2 -->
+ <match value="0xfffe" type="string" offset="0"/> <!-- V1, L1, CRC -->
+ <match value="0xffff" type="string" offset="0"/> <!-- V1, L1 -->
+ <match value="ID3" type="string" offset="0"/>
+ </magic>
<glob pattern="*.mpga"/>
<glob pattern="*.mp2"/>
<glob pattern="*.mp2a"/>
@@ -3140,6 +3088,7 @@
<glob pattern="*.m2a"/>
<glob pattern="*.m3a"/>
</mime-type>
+
<mime-type type="audio/mpeg4-generic"/>
<mime-type type="audio/ogg">
<glob pattern="*.oga"/>
@@ -3428,9 +3377,12 @@
</mime-type>
<mime-type type="image/tiff-fx"/>
+
<mime-type type="image/vnd.adobe.photoshop">
+ <alias type="image/x-psd"/>
<glob pattern="*.psd"/>
</mime-type>
+
<mime-type type="image/vnd.cns.inf2"/>
<mime-type type="image/vnd.djvu">
<glob pattern="*.djvu"/>
@@ -3471,9 +3423,12 @@
<mime-type type="image/vnd.sealedmedia.softseal.gif"/>
<mime-type type="image/vnd.sealedmedia.softseal.jpg"/>
<mime-type type="image/vnd.svf"/>
+
<mime-type type="image/vnd.wap.wbmp">
+ <_comment>Wireless Bitmap File Format</_comment>
<glob pattern="*.wbmp"/>
</mime-type>
+
<mime-type type="image/vnd.xiff">
<glob pattern="*.xif"/>
</mime-type>
@@ -3663,6 +3618,14 @@
<mime-type type="image/x-xbitmap">
<glob pattern="*.xbm"/>
</mime-type>
+
+ <mime-type type="image/x-xcf">
+ <alias type="image/xcf"/>
+ <magic priority="50">
+ <match type="string" value="gimp xcf " offset="0"/>
+ </magic>
+ </mime-type>
+
<mime-type type="image/x-xpixmap">
<glob pattern="*.xpm"/>
</mime-type>
@@ -3714,15 +3677,19 @@
<mime-type type="message/vnd.si.simp"/>
<mime-type type="model/example"/>
+
<mime-type type="model/iges">
+ <_comment>Initial Graphics Exchange Specification Format</_comment>
<glob pattern="*.igs"/>
<glob pattern="*.iges"/>
</mime-type>
+
<mime-type type="model/mesh">
<glob pattern="*.msh"/>
<glob pattern="*.mesh"/>
<glob pattern="*.silo"/>
</mime-type>
+
<mime-type type="model/vnd.dwf">
<glob pattern="*.dwf"/>
</mime-type>
@@ -3744,10 +3711,12 @@
<mime-type type="model/vnd.vtu">
<glob pattern="*.vtu"/>
</mime-type>
+
<mime-type type="model/vrml">
<glob pattern="*.wrl"/>
<glob pattern="*.vrml"/>
</mime-type>
+
<mime-type type="multipart/alternative"/>
<mime-type type="multipart/appledouble"/>
<mime-type type="multipart/byteranges"/>
@@ -3814,6 +3783,10 @@
</mime-type>
<mime-type type="text/troff">
+ <alias type="application/x-troff"/>
+ <alias type="application/x-troff-man"/>
+ <alias type="application/x-troff-me"/>
+ <alias type="application/x-troff-ms"/>
<magic priority="50">
<match value=".\\"" type="string" offset="0"/>
<match value="'\\"" type="string" offset="0"/>
@@ -3926,6 +3899,7 @@
<mime-type type="text/x-setext">
<glob pattern="*.etx"/>
</mime-type>
+
<mime-type type="text/x-uuencode">
<glob pattern="*.uu"/>
</mime-type>
@@ -3990,6 +3964,10 @@
<glob pattern="*.m2v"/>
</mime-type>
+ <mime-type type="video/vnd.mpegurl">
+ <glob pattern="*.mxu"/>
+ </mime-type>
+
<mime-type type="video/mpeg4-generic"/>
<mime-type type="video/mpv"/>
<mime-type type="video/nv"/>
@@ -4118,6 +4096,7 @@
</mime-type>
<mime-type type="x-conference/x-cooltalk">
+ <_comment>Cooltalk Audio</_comment>
<glob pattern="*.ice"/>
</mime-type>