Refactor the get_sharedir function, and move datas to a proper data folder

7 files changed, 52 insertions, 27 deletions

diff --git a/MAT/FORMATS b/MAT/FORMATS
index c497524..3f0751c 100644
--- a/MAT/FORMATS
+++ b/MAT/FORMATS
@@ -2,6 +2,7 @@
     <format>
         <name>Portable Network Graphics</name>
         <extension>.png</extension>
+        <mimetype>image/png</mimetype>
         <support>full</support>
         <metadata>textual metadata + date</metadata>
         <method>removal of harmful fields is done with hachoir</method>
@@ -10,14 +11,19 @@
     <format>
         <name>Jpeg</name>
         <extension>.jpeg, .jpg</extension>
-        <support>full</support>
+        <mimetype>image/jpeg</mimetype>
+        <support>partial</support>
         <metadata>comment + exif/photoshop/adobe</metadata>
         <method>removal of harmful fields is done with hachoir</method>
+        <remaining>Canon Raw tags :
+            http://www.sno.phy.queensu.ca/~phil/exiftool/TagNames/CanonRaw.html
+        </remaining>
     </format>
 
     <format>
         <name>Open Document</name>
         <extension>.odt, .odx, .ods, ...</extension>
+        <mimetype>application/opendocument</mimetype>
         <support>full</support>
         <metadata>a meta.xml file</metadata>
         <method>removal of the meta.xml file</method>
@@ -26,6 +32,7 @@
     <format>
         <name>Office Openxml</name>
         <extension>.docx, .pptx, .xlsx, ...</extension>
+        <mimetype>application/officeopenxml</mimetype>
         <support>full</support>
         <metadata>a docProps folder containings xml metadata files</metadata>
         <method>removal of the docProps folder</method>
@@ -34,18 +41,21 @@
     <format>
         <name>Portable Document Fileformat</name>
         <extension>.pdf</extension>
+        <mimetype>application/pdf</mimetype>
         <support>full</support>
         <metadata>a lot</metadata>
-        <method>rendering of the pdf file on a cairo surface with the help of
-            poppler in order to remove all the internal metadata,
-            then removal of the remaining metadata fields of the pdf itself with
-            pdfrw (the next version of python-cairo will support metadata,
-            so we should get rid of pdfrw)</method>
+        <method> rendering of the PDF file on a cairo surface with the help of
+                poppler in order to remove all the internal metadata.
+                For now, cairo create some metadata.
+                They can be remove if you install either exiftool, or python-pdfrw.
+                The next version of python-cairo will support PDF metadata.
+        </method>
     </format>
 
     <format>
         <name>Tape ARchive</name>
         <extension>.tar, .tar.bz2, .tar.gz</extension>
+        <mimetype>application/x-tar, application/x-gzip, application/x-bzip2</mimetype>
         <support>full</support>
         <metadata>metadata from the file itself, metadata from the file contained
             into the archive, and metadata added by tar to the file at then
@@ -58,11 +68,11 @@
     <format>
         <name>Zip</name>
         <extension>.zip</extension>
-        <support>.partial</support>
+        <mimetype>application/zip</mimetype>
+        <support>partial</support>
         <metadata>metadata from the file itself, metadata from the file contained
             into the archive, and metadata added by zip to the file when added to
-            the archive.
-        </metadata>
+            the archive.</metadata>
         <method>extraction of each file, treatement of the file, add treated file
         to a new archive. When the new archive is complete, remove all his metadata</method>
         <remaining>metadata added by zip itself to internal files</remaining>
@@ -70,7 +80,8 @@
 
     <format>
         <name>MPEG Audio</name>
-        <extension>.mp3, .mp2, .mp1</extension>
+        <extension>.mp3, .mp2, .mp1, .mpa</extension>
+        <mimetype>audio/mpeg</mimetype>
         <support>full</support>
         <metadata>id3</metadata>
         <method>removal of harmful fields is done with hachoir</method>
@@ -79,6 +90,7 @@
     <format>
         <name>Ogg Vorbis</name>
         <extension>.ogg</extension>
+        <mimetype>audio/vorbis</mimetype>
         <support>full</support>
         <metadata>Vorbis</metadata>
         <method>removal of harmful fields is done with mutagen</method>
@@ -87,8 +99,19 @@
     <format>
         <name>Free Lossless Audio Codec</name>
         <extension>.flac</extension>
+        <mimetype>audio/x-flac</mimetype>
         <support>full</support>
         <metadata>Flac, Vorbis</metadata>
         <method>removal of harmful fields is done with mutagen</method>
     </format>
+
+    <format>
+        <name>Torrent</name>
+        <extension>.torrent</extension>
+    <mimetype>application/torrent</mimetype>
+        <support>full</support>
+        <metadata>torrent</metadata>
+        <method>remove all the compromizing metadata with a heavily tuned version
+        of the bencode lib by Petru Paled</method>
+    </format>
 </xml>
diff --git a/MAT/mat.py b/MAT/mat.py
index 7b5149f..b18cc63 100644
--- a/MAT/mat.py
+++ b/MAT/mat.py
@@ -30,19 +30,21 @@ LOGGING_LEVEL = logging.DEBUG
 
 logging.basicConfig(filename=fname, level=LOGGING_LEVEL)
 
+def get_logo():
+    if os.path.isfile('./data/mat.png'):
+        return './data/mat.png'
+    elif os.path.isfile('/usr/share/pixmaps/mat.png'):
+        return '/usr/share/pixmaps/mat.png'
+    elif os.path.isfile('/usr/local/share/pixmaps/mat.png'):
+        return '/usr/local/share/pixmaps/mat.png'
 
-def get_sharedir(filename):
-    '''
-        An ugly hack to find various files
-    '''
-    if os.path.isfile(filename):
-        return filename
-    elif os.path.exists(os.path.join('/usr/local/share/mat/', filename)):
-        return os.path.join('/usr/local/share/mat/', filename)
-    elif os.path.exists(os.path.join('/usr/share/mat/', filename)):
-        return os.path.join('/usr/share/mat', filename)
-    elif os.path.exists(os.path.join('/usr/local/share/pixmaps/', filename)):
-        return os.path.join('/usr/local/share/pixmaps/', filename)
+def get_formats():
+    if os.path.isfile('./data/FORMATS'):
+        return './data/FORMATS'
+    elif os.path.isfile('/usr/share/mat/FORMATS'):
+        return '/usr/share/mat/FORMATS'
+    elif os.path.isfile('/usr/local/share/mat/FORMATS'):
+        return '/usr/local/share/mat/FORMATS'
 
 
 class XMLParser(xml.sax.handler.ContentHandler):
diff --git a/FORMATS b/data/FORMATS
index 3f0751c..3f0751c 100644
--- a/FORMATS
+++ b/data/FORMATS
diff --git a/mat.png b/data/mat.png
index 4198c42..4198c42 100644
--- a/mat.png
+++ b/data/mat.png
diff --git a/mat b/mat
index 5cd61ee..da43198 100755
--- a/mat
+++ b/mat
@@ -104,7 +104,7 @@ def list_supported():
     handler = mat.XMLParser()
     parser = xml.sax.make_parser()
     parser.setContentHandler(handler)
-    path = mat.get_sharedir('FORMATS')
+    path = mat.get_formats()
     with open(path, 'r') as xmlfile:
         parser.parse(xmlfile)
 
diff --git a/mat-gui b/mat-gui
index 121db56..be966b0 100755
--- a/mat-gui
+++ b/mat-gui
@@ -53,7 +53,7 @@ class GUI:
         self.window.set_title('Metadata Anonymisation Toolkit')
         self.window.connect('destroy', gtk.main_quit)
         self.window.set_default_size(800, 600)
-        self.logo = mat.get_sharedir('mat.png')
+        self.logo = mat.get_logo()
         icon = gtk.gdk.pixbuf_new_from_file_at_size(self.logo, 50, 50)
         self.window.set_icon(icon)
 
@@ -405,7 +405,7 @@ class GUI:
         handler = mat.XMLParser()
         parser = xml.sax.make_parser()
         parser.setContentHandler(handler)
-        path = mat.get_sharedir('FORMATS')
+        path = mat.get_formats()
         with open(path, 'r') as xmlfile:
             parser.parse(xmlfile)
 
diff --git a/setup.py b/setup.py
index f69a326..5852466 100755
--- a/setup.py
+++ b/setup.py
@@ -27,8 +27,8 @@ setup(
     scripts           = ['mat', 'mat-gui'],
     data_files        = [
         ( 'share/applications', ['mat.desktop'] ),
-        ( 'share/mat', ['FORMATS'] ),
-        ( 'share/pixmaps', ['mat.png'] ),
+        ( 'share/mat', ['data/FORMATS'] ),
+        ( 'share/pixmaps', ['data/mat.png'] ),
         ( 'share/doc/mat', ['README', 'TODO'] ),
         ( 'share/man/man1', ['mat.1', 'mat-gui.1'] ),
     ],