Improve detection of application/x-ole-storage subtypes

Most OLE file formats include a 128-bit root class identifier somewhere
after the 592nd byte.

StarImpress files are essentially bundles of StarDraw files, hence the
need to give the StarImpress magic higher priority.

http://fileformats.archiveteam.org/wiki/OLE

24 files changed, 130 insertions, 8 deletions

diff --git a/data/freedesktop.org.xml.in b/data/freedesktop.org.xml.in
index 1fa11f9..bd09d6c 100644
--- a/data/freedesktop.org.xml.in
+++ b/data/freedesktop.org.xml.in
@@ -704,7 +704,11 @@ command to generate the output files.
     <comment>Excel spreadsheet</comment>
     <sub-class-of type="application/x-ole-storage"/>
     <generic-icon name="x-office-spreadsheet"/>
-    <magic>
+    <magic priority="60">
+      <match type="string" value="\xd0\xcf\x11\xe0\xa1\xb1\x1a\xe1" offset="0">
+        <match type="string" value="\x10\x08\x02\x00\x00\x00\x00\x00\xc0\x00\x00\x00\x00\x00\x00\x46" offset="592:8192"/>
+        <match type="string" value="\x20\x08\x02\x00\x00\x00\x00\x00\xc0\x00\x00\x00\x00\x00\x00\x46" offset="592:8192"/>
+      </match>
       <match type="string" value="Microsoft Excel 5.0 Worksheet" offset="2080"/>
     </magic>
     <glob pattern="*.xls"/>
@@ -747,6 +751,12 @@ command to generate the output files.
     <comment>PowerPoint presentation</comment>
     <sub-class-of type="application/x-ole-storage"/>
     <generic-icon name="x-office-presentation"/>
+    <magic priority="60">
+      <match type="string" value="\xd0\xcf\x11\xe0\xa1\xb1\x1a\xe1" offset="0">
+        <match type="string" value="\x10\x8d\x81\x64\x9b\x4f\xcf\x11\x86\xea\x00\xaa\x00\xb9\x29\xe8" offset="592:8192"/>
+        <match type="string" value="\x70\xae\x7b\xea\x3b\xfb\xcd\x11\xa9\x03\x00\xaa\x00\x51\x0e\xa3" offset="592:8192"/>
+      </match>
+    </magic>
     <glob pattern="*.ppz"/>
     <glob pattern="*.ppt"/>
     <glob pattern="*.pps"/>
@@ -853,6 +863,16 @@ command to generate the output files.
     <comment>Microsoft Works document</comment>
     <sub-class-of type="application/x-ole-storage"/>
     <generic-icon name="x-office-document"/>
+    <magic priority="60">
+      <match type="string" value="\xd0\xcf\x11\xe0\xa1\xb1\x1a\xe1" offset="0">
+        <match type="string" value="\x02\x13\x02\x00\x00\x00\x00\x00\xc0\x00\x00\x00\x00\x00\x00\x46" offset="592:8192"/>
+        <match type="string" value="\x03\x13\x02\x00\x00\x00\x00\x00\xc0\x00\x00\x00\x00\x00\x00\x46" offset="592:8192"/>
+        <match type="string" value="\xb2\x5a\xa4\x0e\x0a\x9e\xd1\x11\xa4\x07\x00\xc0\x4f\xb9\x32\xba" offset="592:8192"/>
+        <match type="string" value="\xc0\xc7\x26\x6e\xb9\x8c\xd3\x11\xa1\xc8\x00\xc0\x4f\x61\x24\x52" offset="592:8192"/>
+        <match type="string" value="\xc2\xdb\xcd\x28\xe2\x0a\xce\x11\xa2\x9a\x00\xaa\x00\x4a\x1a\x72" offset="592:8192"/>
+        <match type="string" value="\xc3\xdb\xcd\x28\xe2\x0a\xce\x11\xa2\x9a\x00\xaa\x00\x4a\x1a\x72" offset="592:8192"/>
+      </match>
+    </magic>
     <glob pattern="*.wcm"/>
     <glob pattern="*.wdb"/>
     <glob pattern="*.wks"/>
@@ -863,6 +883,12 @@ command to generate the output files.
     <comment>Microsoft Visio document</comment>
     <sub-class-of type="application/x-ole-storage"/>
     <generic-icon name="x-office-document"/>
+    <magic priority="60">
+      <match type="string" value="\xd0\xcf\x11\xe0\xa1\xb1\x1a\xe1" offset="0">
+        <match type="string" value="\x13\x1a\x02\x00\x00\x00\x00\x00\xc0\x00\x00\x00\x00\x00\x00\x46" offset="592:8192"/>
+        <match type="string" value="\x14\x1a\x02\x00\x00\x00\x00\x00\xc0\x00\x00\x00\x00\x00\x00\x46" offset="592:8192"/>
+      </match>
+    </magic>
     <glob pattern="*.vsd"/>
     <glob pattern="*.vst"/>
     <glob pattern="*.vsw"/>
@@ -873,6 +899,10 @@ command to generate the output files.
     <sub-class-of type="application/x-ole-storage"/>
     <generic-icon name="x-office-document"/>
     <magic priority="60">
+      <match type="string" value="\xd0\xcf\x11\xe0\xa1\xb1\x1a\xe1" offset="0">
+        <match type="string" value="\x00\x09\x02\x00\x00\x00\x00\x00\x00\xc0\x46\x00\x00\x00\x00\x00" offset="592:8192"/>
+        <match type="string" value="\x06\x09\x02\x00\x00\x00\x00\x00\x00\xc0\x46\x00\x00\x00\x00\x00" offset="592:8192"/>
+      </match>
       <match type="string" value="\x31\xbe\x00\x00" offset="0"/>
       <match type="string" value="PO^Q`" offset="0"/>
       <match type="string" value="\376\067\0\043" offset="0"/>
@@ -923,22 +953,53 @@ command to generate the output files.
   </mime-type>
   <mime-type type="application/vnd.stardivision.calc">
     <comment>StarCalc spreadsheet</comment>
+    <sub-class-of type="application/x-ole-storage"/>
     <generic-icon name="x-office-spreadsheet"/>
+    <magic priority="60">
+      <match type="string" value="\xd0\xcf\x11\xe0\xa1\xb1\x1a\xe1" offset="0">
+        <match type="string" value="\x41\xd4\x61\x63\x35\x42\xd0\x11\x89\xcb\x00\x80\x29\xe4\xb0\xb1" offset="592:8192"/>
+        <match type="string" value="\x61\xb8\xa5\xc6\xd6\x85\xd1\x11\x89\xcb\x00\x80\x29\xe4\xb0\xb1" offset="592:8192"/>
+        <match type="string" value="\xa0\x3f\x54\x3f\xa6\xb6\x1b\x10\x99\x61\x04\x02\x1c\x00\x70\x02" offset="592:8192"/>
+      </match>
+    </magic>
     <glob pattern="*.sdc"/>
   </mime-type>
   <mime-type type="application/vnd.stardivision.chart">
     <comment>StarChart chart</comment>
+    <sub-class-of type="application/x-ole-storage"/>
     <generic-icon name="x-office-spreadsheet"/>
+    <magic priority="60">
+      <match type="string" value="\xd0\xcf\x11\xe0\xa1\xb1\x1a\xe1" offset="0">
+        <match type="string" value="\x21\x43\x88\xbf\xdd\x85\xd1\x11\x89\xd0\x00\x80\x29\xe4\xb0\xb1" offset="592:8192"/>
+        <match type="string" value="\xe0\x99\x9c\xfb\x6d\x2c\x1c\x10\x8e\x2c\x00\x00\x1b\x4c\xc7\x11" offset="592:8192"/>
+        <match type="string" value="\xe0\xb7\xb3\x02\x25\x42\xd0\x11\x89\xca\x00\x80\x29\xe4\xb0\xb1" offset="592:8192"/>
+      </match>
+    </magic>
     <glob pattern="*.sds"/>
   </mime-type>
   <mime-type type="application/vnd.stardivision.draw">
     <comment>StarDraw drawing</comment>
+    <sub-class-of type="application/x-ole-storage"/>
     <generic-icon name="image-x-generic"/>
+    <magic priority="60">
+      <match type="string" value="\xd0\xcf\x11\xe0\xa1\xb1\x1a\xe1" offset="0">
+        <match type="string" value="\xa0\x05\x89\x2e\xbd\x85\xd1\x11\x89\xd0\x00\x80\x29\xe4\xb0\xb1" offset="592:8192"/>
+        <match type="string" value="\xe0\xaa\x10\xaf\x6d\xb3\x1b\x10\x99\x61\x04\x02\x1c\x00\x70\x02" offset="592:8192"/>
+      </match>
+    </magic>
     <glob pattern="*.sda"/>
   </mime-type>
   <mime-type type="application/vnd.stardivision.impress">
     <comment>StarImpress presentation</comment>
+    <sub-class-of type="application/x-ole-storage"/>
     <generic-icon name="x-office-presentation"/>
+    <magic priority="70">
+      <match type="string" value="\xd0\xcf\x11\xe0\xa1\xb1\x1a\xe1" offset="0">
+        <match type="string" value="\x21\x72\x5c\x56\xbc\x85\xd1\x11\x89\xd0\x00\x80\x29\xe4\xb0\xb1" offset="592:8192"/>
+        <match type="string" value="\xc0\x3c\x2d\x01\x16\x42\xd0\x11\x89\xcb\x00\x80\x29\xe4\xb0\xb1" offset="592:8192"/>
+        <match type="string" value="\xe0\xaa\x10\xaf\x6d\xb3\x1b\x10\x99\x61\x04\x02\x1c\x00\x70\x02" offset="592:8192"/>
+      </match>
+    </magic>
     <glob pattern="*.sdd"/>
     <glob pattern="*.sdp"/>
   </mime-type>
@@ -948,16 +1009,30 @@ command to generate the output files.
   </mime-type>
   <mime-type type="application/vnd.stardivision.math">
     <comment>StarMath formula</comment>
+    <sub-class-of type="application/x-ole-storage"/>
     <generic-icon name="x-office-document"/>
+    <magic priority="60">
+      <match type="string" value="\xd0\xcf\x11\xe0\xa1\xb1\x1a\xe1" offset="0">
+        <match type="string" value="\x40\xe6\xb5\xff\xde\x85\xd1\x11\x89\xd0\x00\x80\x29\xe4\xb0\xb1" offset="592:8192"/>
+        <match type="string" value="\x60\x04\x59\xd4\xfd\x35\x1c\x10\xb1\x2a\x04\x02\x1c\x00\x70\x02" offset="592:8192"/>
+        <match type="string" value="\xe1\xb7\xb3\x02\x25\x42\xd0\x11\x89\xca\x00\x80\x29\xe4\xb0\xb1" offset="592:8192"/>
+      </match>
+    </magic>
     <glob pattern="*.smf"/>
   </mime-type>
   <mime-type type="application/vnd.stardivision.writer">
     <comment>StarWriter document</comment>
+    <sub-class-of type="application/x-ole-storage"/>
     <generic-icon name="x-office-document"/>
     <glob pattern="*.sdw"/>
     <glob pattern="*.vor"/>
     <glob pattern="*.sgl"/>
     <magic priority="90">
+      <match type="string" value="\xd0\xcf\x11\xe0\xa1\xb1\x1a\xe1" offset="0">
+        <match type="string" value="\x40\x7e\x5c\xdc\x5c\xb3\x1b\x10\x99\x61\x04\x02\x1c\x00\x70\x02" offset="592:8192"/>
+        <match type="string" value="\xb0\xe9\x04\x8b\x0e\x42\xd0\x11\xa4\x5e\x00\xa0\x24\x9d\x57\xb1" offset="592:8192"/>
+        <match type="string" value="\xd1\xf9\x0c\xc2\xae\x85\xd1\x11\xaa\xb4\x00\x06\x09\x7d\x56\x1a" offset="592:8192"/>
+      </match>
       <match type="string" value="StarWriter" offset="2089" />
     </magic>
     <alias type="application/vnd.stardivision.writer-global"/>
@@ -1413,9 +1488,12 @@ command to generate the output files.
     <alias type="application/x-wordperfect"/>
     <alias type="application/wordperfect"/>
     <generic-icon name="x-office-document"/>
-    <magic>
+    <magic priority="60">
       <match type="string" value="WPC" offset="1"/>
       <!-- <match type="big32" value="0xff575053c405" offset="0"/> -->
+      <match type="string" value="\xd0\xcf\x11\xe0\xa1\xb1\x1a\xe1" offset="0">
+        <match type="string" value="\xff\x73\x98\x51\xad\x2d\x20\x02\x19\x37\x00\x00\x92\x96\x79\xcd" offset="592:8192"/>
+      </match>
     </magic>
     <glob pattern="*.wp"/>
     <glob pattern="*.wp4"/>
@@ -3657,8 +3735,7 @@ command to generate the output files.
     <comment>OLE2 compound document storage</comment>
     <generic-icon name="x-office-document"/>
     <magic>
-      <match type="string" value="\320\317\021\340\241\261\032\341" offset="0"/>
-      <match type="little32" value="0xe011cfd0" offset="0"/>
+      <match type="string" value="\xd0\xcf\x11\xe0\xa1\xb1\x1a\xe1" offset="0"/>
     </magic>
   </mime-type>
   <mime-type type="application/vnd.microsoft.windows.thumbnail-cache">
@@ -3676,12 +3753,22 @@ command to generate the output files.
   <mime-type type="application/vnd.ms-publisher">
     <comment>Microsoft Publisher document</comment>
     <sub-class-of type="application/x-ole-storage"/>
+    <magic priority="60">
+      <match type="string" value="\xd0\xcf\x11\xe0\xa1\xb1\x1a\xe1" offset="0">
+        <match type="string" value="\x01\x12\x02\x00\x00\x00\x00\x00\x00\xc0\x00\x00\x00\x00\x00\x46" offset="592:8192"/>
+      </match>
+    </magic>
     <glob pattern="*.pub"/>
   </mime-type>
   <mime-type type="application/x-msi">
     <comment>Windows Installer package</comment>
     <sub-class-of type="application/x-ole-storage"/>
     <generic-icon name="package-x-generic"/>
+    <magic priority="60">
+      <match type="string" value="\xd0\xcf\x11\xe0\xa1\xb1\x1a\xe1" offset="0">
+        <match type="string" value="\x84\x10\x0c\x00\x00\x00\x00\x00\xc0\x00\x00\x00\x00\x00\x00\x46" offset="592:8192"/>
+      </match>
+    </magic>
     <glob pattern="*.msi"/>
   </mime-type>
   <mime-type type="application/appx">
@@ -3860,9 +3947,15 @@ command to generate the output files.
   <mime-type type="application/x-quattropro">
     <comment>Quattro Pro spreadsheet</comment>
     <generic-icon name="x-office-spreadsheet"/>
+    <magic priority="60">
+      <match type="string" value="\xd0\xcf\x11\xe0\xa1\xb1\x1a\xe1" offset="0">
+        <match type="string" value="\x00\xb4\x02\x00\x00\x00\x00\x00\xc0\x00\x00\x00\x00\x00\x00\x46" offset="592:8192"/>
+      </match>
+    </magic>
     <glob pattern="*.wb1"/>
     <glob pattern="*.wb2"/>
     <glob pattern="*.wb3"/>
+    <glob pattern="*.qpw"/>
   </mime-type>
   <mime-type type="application/x-quicktime-media-link">
     <comment>QuickTime playlist</comment>
diff --git a/tests/mime-detection/compound.wpd b/tests/mime-detection/compound.wpd
new file mode 100644
index 0000000..72b6fa2
--- /dev/null
+++ b/tests/mime-detection/compound.wpd
diff --git a/tests/mime-detection/list b/tests/mime-detection/list
index 5338ded..3149c0e 100644
--- a/tests/mime-detection/list
+++ b/tests/mime-detection/list
@@ -162,6 +162,35 @@ ooo-test.fodg application/vnd.oasis.opendocument.graphics-flat-xml ox
 ooo-test.fodp application/vnd.oasis.opendocument.presentation-flat-xml ox
 ooo-test.fods application/vnd.oasis.opendocument.spreadsheet-flat-xml ox
 ooo-test.fodt application/vnd.oasis.opendocument.text-flat-xml ox
+# Created with StarOffice 3.1
+so3.sdw application/vnd.stardivision.writer
+so3.sdc application/vnd.stardivision.calc
+so3.sdd application/vnd.stardivision.impress
+so3.sds application/vnd.stardivision.chart
+# Created with StarOffice 4.0
+so4.sdw application/vnd.stardivision.writer
+so4.sdc application/vnd.stardivision.calc
+so4.sdd application/vnd.stardivision.impress
+so4.sds application/vnd.stardivision.chart
+so4.smf application/vnd.stardivision.math
+# Created with StarOffice 5.2
+so5.sdw application/vnd.stardivision.writer
+so5.sdc application/vnd.stardivision.calc
+so5.sdd application/vnd.stardivision.impress
+so5.sda application/vnd.stardivision.draw
+so5.sds application/vnd.stardivision.chart
+so5.smf application/vnd.stardivision.math
+# Created with LibreOffice 7.6
+lo7.doc application/msword
+lo7.xls application/vnd.ms-excel
+lo7.ppt application/vnd.ms-powerpoint
+# Created with WordPerfect 9
+regular.wpd application/vnd.wordperfect
+compound.wpd application/vnd.wordperfect
+# Created with Quattro Pro 6
+quattro6.wb2 application/x-quattropro ox
+# Created with Quattro Pro 9
+quattro9.qpw application/x-quattropro
 # Copied from https://bugs.freedesktop.org/show_bug.cgi?id=20604
 foo.doc application/msword
 # Copied from http://bugzilla.abisource.com/show_bug.cgi?id=6435
@@ -172,7 +201,7 @@ ocf10-20060911.epub application/epub+zip
 # Copied from https://standardebooks.org/ebooks/elizabeth-barrett-browning/sonnets-from-the-portuguese
 elizabeth-barrett-browning_sonnets-from-the-portuguese.azw3 application/vnd.amazon.mobi8-ebook ox
 # Copied from http://libwps.svn.sourceforge.net/viewvc/libwps/trunk/reference/
-test.wps application/vnd.ms-works ox
+test.wps application/vnd.ms-works
 # Copied from https://bugs.freedesktop.org/show_bug.cgi?id=45922
 test.xlr application/vnd.ms-works ox
 # Copied from http://openxmldeveloper.org/articles/OpenXMLsamples.aspx
@@ -203,7 +232,7 @@ white_640x480.kra application/x-krita
 test.mobi application/x-mobipocket-ebook
 test.lwp application/vnd.lotus-wordpro
 # Copied from https://bugs.freedesktop.org/show_bug.cgi?id=64946
-TS010082249.pub application/vnd.ms-publisher ox
+TS010082249.pub application/vnd.ms-publisher
 test.key application/vnd.apple.keynote
 # Created with Numbers 3.5
 test.numbers application/vnd.apple.numbers ox
@@ -260,7 +289,7 @@ live-streaming.m3u application/vnd.apple.mpegurl x
 # Copied from http://qthttp.apple.com.edgesuite.net/1006ad9g4hjk/all_w.m3u8
 all_w.m3u8 application/vnd.apple.mpegurl x
 # Copied from http://bugs.freedesktop.org/show_bug.cgi?id=42922
-test.vsd application/vnd.visio ox
+test.vsd application/vnd.visio
 # Copied from https://bugs.freedesktop.org/show_bug.cgi?id=40800
 test.dts audio/vnd.dts
 test.dtshd audio/vnd.dts.hd
@@ -505,7 +534,7 @@ test.jar application/java-archive ox
 # Copied from http://bugs.freedesktop.org/show_bug.cgi?id=12620
 hello.pack application/x-java-pack200
 # Copied from http://www.apache.org/dist/httpd/binaries/win32/apache_2.2.9-win32-x86-no_ssl-r2.msi
-test.msi application/x-msi ox
+test.msi application/x-msi
 # Created by building the default blank Universal Windows C++ app in Visual Studio 2022
 App1_1.0.0.0_x64.appx application/appx ox
 App1_1.0.0.0_x64.appxbundle application/appxbundle ox
diff --git a/tests/mime-detection/lo7.doc b/tests/mime-detection/lo7.doc
new file mode 100644
index 0000000..db71a1c
--- /dev/null
+++ b/tests/mime-detection/lo7.doc
diff --git a/tests/mime-detection/lo7.ppt b/tests/mime-detection/lo7.ppt
new file mode 100644
index 0000000..bedaacb
--- /dev/null
+++ b/tests/mime-detection/lo7.ppt
diff --git a/tests/mime-detection/lo7.xls b/tests/mime-detection/lo7.xls
new file mode 100644
index 0000000..ae90db5
--- /dev/null
+++ b/tests/mime-detection/lo7.xls
diff --git a/tests/mime-detection/quattro6.wb2 b/tests/mime-detection/quattro6.wb2
new file mode 100644
index 0000000..fb032ea
--- /dev/null
+++ b/tests/mime-detection/quattro6.wb2
diff --git a/tests/mime-detection/quattro9.qpw b/tests/mime-detection/quattro9.qpw
new file mode 100644
index 0000000..a1e2f19
--- /dev/null
+++ b/tests/mime-detection/quattro9.qpw
diff --git a/tests/mime-detection/regular.wpd b/tests/mime-detection/regular.wpd
new file mode 100644
index 0000000..4c4c6e5
--- /dev/null
+++ b/tests/mime-detection/regular.wpd
diff --git a/tests/mime-detection/so3.sdc b/tests/mime-detection/so3.sdc
new file mode 100644
index 0000000..c8f5991
--- /dev/null
+++ b/tests/mime-detection/so3.sdc
diff --git a/tests/mime-detection/so3.sdd b/tests/mime-detection/so3.sdd
new file mode 100644
index 0000000..89346fb
--- /dev/null
+++ b/tests/mime-detection/so3.sdd
diff --git a/tests/mime-detection/so3.sds b/tests/mime-detection/so3.sds
new file mode 100644
index 0000000..aba5cee
--- /dev/null
+++ b/tests/mime-detection/so3.sds
diff --git a/tests/mime-detection/so3.sdw b/tests/mime-detection/so3.sdw
new file mode 100644
index 0000000..4957951
--- /dev/null
+++ b/tests/mime-detection/so3.sdw
diff --git a/tests/mime-detection/so4.sdc b/tests/mime-detection/so4.sdc
new file mode 100644
index 0000000..fa84662
--- /dev/null
+++ b/tests/mime-detection/so4.sdc
diff --git a/tests/mime-detection/so4.sdd b/tests/mime-detection/so4.sdd
new file mode 100644
index 0000000..2280762
--- /dev/null
+++ b/tests/mime-detection/so4.sdd
diff --git a/tests/mime-detection/so4.sds b/tests/mime-detection/so4.sds
new file mode 100644
index 0000000..dc9d0eb
--- /dev/null
+++ b/tests/mime-detection/so4.sds
diff --git a/tests/mime-detection/so4.sdw b/tests/mime-detection/so4.sdw
new file mode 100644
index 0000000..f5d56d9
--- /dev/null
+++ b/tests/mime-detection/so4.sdw
diff --git a/tests/mime-detection/so4.smf b/tests/mime-detection/so4.smf
new file mode 100644
index 0000000..6ddb036
--- /dev/null
+++ b/tests/mime-detection/so4.smf
diff --git a/tests/mime-detection/so5.sda b/tests/mime-detection/so5.sda
new file mode 100644
index 0000000..e517d24
--- /dev/null
+++ b/tests/mime-detection/so5.sda
diff --git a/tests/mime-detection/so5.sdc b/tests/mime-detection/so5.sdc
new file mode 100644
index 0000000..cf5ffb8
--- /dev/null
+++ b/tests/mime-detection/so5.sdc
diff --git a/tests/mime-detection/so5.sdd b/tests/mime-detection/so5.sdd
new file mode 100644
index 0000000..b344024
--- /dev/null
+++ b/tests/mime-detection/so5.sdd
diff --git a/tests/mime-detection/so5.sds b/tests/mime-detection/so5.sds
new file mode 100644
index 0000000..ba3ee98
--- /dev/null
+++ b/tests/mime-detection/so5.sds
diff --git a/tests/mime-detection/so5.sdw b/tests/mime-detection/so5.sdw
new file mode 100644
index 0000000..5a45742
--- /dev/null
+++ b/tests/mime-detection/so5.sdw
diff --git a/tests/mime-detection/so5.smf b/tests/mime-detection/so5.smf
new file mode 100644
index 0000000..77b2c2c
--- /dev/null
+++ b/tests/mime-detection/so5.smf