Author: jukka
Date: Tue Feb 3 22:13:23 2009
New Revision: 740478
URL: http://svn.apache.org/viewvc?rev=740478&view=rev
Log:
TIKA-192: Add glob and magic patterns for image types
Added glob patterns for numerous camera raw file formats.
Modified:
lucene/tika/trunk/src/main/resources/mime/tika-mimetypes.xml
lucene/tika/trunk/src/test/java/org/apache/tika/mime/TestMimeTypes.java
Modified: lucene/tika/trunk/src/main/resources/mime/tika-mimetypes.xml
URL:
http://svn.apache.org/viewvc/lucene/tika/trunk/src/main/resources/mime/tika-mimetypes.xml?rev=740478&r1=740477&r2=740478&view=diff
==============================================================================
--- lucene/tika/trunk/src/main/resources/mime/tika-mimetypes.xml (original)
+++ lucene/tika/trunk/src/main/resources/mime/tika-mimetypes.xml Tue Feb 3
22:13:23 2009
@@ -1079,6 +1079,122 @@
<glob pattern="*.ppm" />
</mime-type>
+ <mime-type type="image/x-tika-dng">
+ <acronym>DNG</acronym>
+ <comment>Adobe Digital Negative</comment>
+ <glob pattern="*.dng" />
+ </mime-type>
+
+ <mime-type type="image/x-tika-hasselblad">
+ <comment>Hasselblad raw image</comment>
+ <glob pattern="*.3fr" />
+ </mime-type>
+
+ <mime-type type="image/x-tika-fuji">
+ <comment>Fuji raw image</comment>
+ <glob pattern="*.raf" />
+ </mime-type>
+
+ <mime-type type="image/x-tika-canon">
+ <comment>Canon raw image</comment>
+ <glob pattern="*.crw" />
+ <glob pattern="*.cr2" />
+ </mime-type>
+
+ <mime-type type="image/x-tika-kodak">
+ <comment>Kodak raw image</comment>
+ <glob pattern="*.k25" />
+ <glob pattern="*.kdc" />
+ <glob pattern="*.dcs" />
+ <glob pattern="*.drf" />
+ </mime-type>
+
+ <mime-type type="image/x-tika-minolta">
+ <comment>Minolta raw image</comment>
+ <glob pattern="*.mrw" />
+ </mime-type>
+
+ <mime-type type="image/x-tika-nikon">
+ <comment>Nikon raw image</comment>
+ <glob pattern="*.nef" />
+ <glob pattern="*.nrw" />
+ </mime-type>
+
+ <mime-type type="image/x-tika-olympus">
+ <comment>Olympus raw image</comment>
+ <glob pattern="*.orf" />
+ </mime-type>
+
+ <mime-type type="image/x-tika-pentax">
+ <comment>Pentax raw image</comment>
+ <glob pattern="*.ptx" />
+ <glob pattern="*.pef" />
+ </mime-type>
+
+ <mime-type type="image/x-tika-sony">
+ <comment>Sony raw image</comment>
+ <glob pattern="*.arw" />
+ <glob pattern="*.srf" />
+ <glob pattern="*.sr2" />
+ </mime-type>
+
+ <mime-type type="image/x-tika-sigma">
+ <comment>Sigma raw image</comment>
+ <glob pattern="*.x3f" />
+ </mime-type>
+
+ <mime-type type="image/x-tika-epson">
+ <comment>Epson raw image</comment>
+ <glob pattern="*.erf" />
+ </mime-type>
+
+ <mime-type type="image/x-tika-mamiya">
+ <comment>Mamiya raw image</comment>
+ <glob pattern="*.mef" />
+ </mime-type>
+
+ <mime-type type="image/x-tika-leaf">
+ <comment>Leaf raw image</comment>
+ <glob pattern="*.mos" />
+ </mime-type>
+
+ <mime-type type="image/x-tika-panasonic">
+ <comment>Panasonic raw image</comment>
+ <glob pattern="*.raw" />
+ <glob pattern="*.rw2" />
+ </mime-type>
+
+ <mime-type type="image/x-tika-phaseone">
+ <comment>Phase One raw image</comment>
+ <glob pattern="*.cap" />
+ <glob pattern="*.iiq" />
+ </mime-type>
+
+ <mime-type type="image/x-tika-red">
+ <comment>Red raw image</comment>
+ <glob pattern="*.r3d" />
+ </mime-type>
+
+ <mime-type type="image/x-tika-imacon">
+ <comment>Imacon raw image</comment>
+ <glob pattern="*.fff" />
+ </mime-type>
+
+ <mime-type type="image/x-tika-logitech">
+ <comment>Logitech raw image</comment>
+ <glob pattern="*.pxn" />
+ </mime-type>
+
+ <mime-type type="image/x-tika-casio">
+ <comment>Casio raw image</comment>
+ <glob pattern="*.bay" />
+ </mime-type>
+
+ <mime-type type="image/x-tika-rawzor">
+ <comment>Rawzor raw image</comment>
+ <glob pattern="*.rwz" />
+ </mime-type>
+
<mime-type type="message/news">
<magic priority="50">
<match value="Path:" type="string" offset="0" />
Modified:
lucene/tika/trunk/src/test/java/org/apache/tika/mime/TestMimeTypes.java
URL:
http://svn.apache.org/viewvc/lucene/tika/trunk/src/test/java/org/apache/tika/mime/TestMimeTypes.java?rev=740478&r1=740477&r2=740478&view=diff
==============================================================================
--- lucene/tika/trunk/src/test/java/org/apache/tika/mime/TestMimeTypes.java
(original)
+++ lucene/tika/trunk/src/test/java/org/apache/tika/mime/TestMimeTypes.java Tue
Feb 3 22:13:23 2009
@@ -153,6 +153,42 @@
assertTypeByName("image/x-portable-pixmap", "x.PPM");
}
+ public void testRawDetection() throws Exception {
+ assertTypeByName("image/x-tika-dng", "x.dng");
+ assertTypeByName("image/x-tika-dng", "x.DNG");
+ assertTypeByName("image/x-tika-hasselblad", "x.3fr");
+ assertTypeByName("image/x-tika-fuji", "x.raf");
+ assertTypeByName("image/x-tika-canon", "x.crw");
+ assertTypeByName("image/x-tika-canon", "x.cr2");
+ assertTypeByName("image/x-tika-kodak", "x.k25");
+ assertTypeByName("image/x-tika-kodak", "x.kdc");
+ assertTypeByName("image/x-tika-kodak", "x.dcs");
+ assertTypeByName("image/x-tika-kodak", "x.drf");
+ assertTypeByName("image/x-tika-minolta", "x.mrw");
+ assertTypeByName("image/x-tika-nikon", "x.nef");
+ assertTypeByName("image/x-tika-nikon", "x.nrw");
+ assertTypeByName("image/x-tika-olympus", "x.orf");
+ assertTypeByName("image/x-tika-pentax", "x.ptx");
+ assertTypeByName("image/x-tika-pentax", "x.pef");
+ assertTypeByName("image/x-tika-sony", "x.arw");
+ assertTypeByName("image/x-tika-sony", "x.srf");
+ assertTypeByName("image/x-tika-sony", "x.sr2");
+ assertTypeByName("image/x-tika-sigma", "x.x3f");
+ assertTypeByName("image/x-tika-epson", "x.erf");
+ assertTypeByName("image/x-tika-mamiya", "x.mef");
+ assertTypeByName("image/x-tika-leaf", "x.mos");
+ assertTypeByName("image/x-tika-panasonic", "x.raw");
+ assertTypeByName("image/x-tika-panasonic", "x.rw2");
+ assertTypeByName("image/x-tika-phaseone", "x.cap");
+ assertTypeByName("image/x-tika-phaseone", "x.iiq");
+ assertTypeByName("image/x-tika-phaseone", "x.cap");
+ assertTypeByName("image/x-tika-red", "x.r3d");
+ assertTypeByName("image/x-tika-imacon", "x.fff");
+ assertTypeByName("image/x-tika-logitech", "x.pxn");
+ assertTypeByName("image/x-tika-casio", "x.bay");
+ assertTypeByName("image/x-tika-rawzor", "x.rwz");
+ }
+
/**
* Tests MimeTypes.getMimeType(URL), which examines both the byte header
* and, if necessary, the URL's extension.
