Add (in xml) the supported fileformat list, and a parser

2 files changed, 126 insertions, 2 deletions

diff --git a/FORMATS b/FORMATS
new file mode 100644
index 0000000..cc38bae
--- /dev/null
+++ b/FORMATS
@@ -0,0 +1,86 @@
+<xml>
+    <format>
+        <name>Portable Network Graphics</name>
+        <extension>.png</extension>
+        <support>full</support>
+        <metadata>textual metadata + date</metadata>
+        <method>removal of harmful fields is done with hachoir</method>
+    </format>
+
+    <format>
+        <name>Jpeg</name>
+        <extension>.jpeg, .jpg</extension>
+        <support>full</support>
+        <metadata>comment + exif/photoshop/adobe</metadata>
+        <method>removal of harmful fields is done with hachoir</method>
+    </format>
+
+    <format>
+        <name>Open Document</name>
+        <extension>.odt, .odx, .ods, ...</extension>
+        <support>full</support>
+        <metadata>a meta.xml file</metadata>
+        <method>removal of the meta.xml file</method>
+    </format>
+
+    <format>
+        <name>Portable Document Fileformat</name>
+        <extension>.pdf</extension>
+        <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>
+    </format>
+
+    <format>
+        <name>Tape ARchive</name>
+        <extension>.tar, .tar.bz2, .tar.gz</extension>
+        <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
+            creation of the archive</metadata>
+        <method>extraction of each file, treatement of the file, add treated file
+        to a new archive, right before the add, remove the metadata added by tar
+        itself. When the new archive is complete, remove all his metadata.</method>
+    </format>
+
+    <format>
+        <name>Zip</name>
+        <extension>.zip</extension>
+        <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>
+        <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>
+    </format>
+
+    <format>
+        <name>MPEG Audio</name>
+        <extension>.mp3, .mp2, .mp1</extension>
+        <support>full</support>
+        <metadata>id3</metadata>
+        <method>removal of harmful fields is done with hachoir</method>
+    </format>
+
+    <format>
+        <name>Ogg Vorbis</name>
+        <extension>.ogg</extension>
+        <support>full</support>
+        <metadata>Vorbis</metadata>
+        <method>removal of harmful fields is done with mutagen</method>
+    </format>
+
+    <format>
+        <name>Free Lossless Audio Codec</name>
+        <extension>.flac</extension>
+        <support>full</support>
+        <metadata>Flac, Vorbis</metadata>
+        <method>removal of harmful fields is done with mutagen</method>
+    </format>
+</xml>
diff --git a/lib/mat.py b/lib/mat.py
index 8226c7e..8fe6fb4 100644
--- a/lib/mat.py
+++ b/lib/mat.py
@@ -7,6 +7,7 @@
 import os
 import subprocess
 import logging
+import xml.sax
 
 import hachoir_core.cmd_line
 import hachoir_parser
@@ -45,13 +46,50 @@ except ImportError:
 try:
     import mutagen
     STRIPPERS['audio/x-flac'] = audio.FlacStripper
-    STRIPPERS['audio/x-ape'] = audio.Apev2Stripper
-    STRIPPERS['audio/x-wavpack'] = audio.Apev2Stripper
     STRIPPERS['audio/vorbis'] = audio.OggStripper
 except ImportError:
     print('unable to import python-mutagen : limited audio format support')
 
 
+class XMLParser(xml.sax.handler.ContentHandler):
+    '''
+        Parse the supported format xml, and return a corresponding
+        list of dict
+    '''
+    def __init__(self):
+        self.dict = {}
+        self.list = []
+        self.content, self.key = '', ''
+        self.between= False
+
+    def startElement(self, name, attrs):
+        '''
+            Called when entering into xml balise
+        '''
+        self.between = True
+        self.key = name
+        self.content = ''
+
+    def endElement(self, name):
+        '''
+            Called when exiting a xml balise
+        '''
+        if name == 'format':  # exiting a fileformat section
+            self.list.append(self.dict.copy())
+            self.dict.clear()
+        else:
+            content = self.content.replace('\n', ' ')
+            self.dict[self.key] = content
+            self.between = False
+
+    def characters(self, characters):
+        '''
+            Concatenate the content between opening and closing balises
+        '''
+        if self.between is True:
+            self.content += characters
+
+
 def secure_remove(filename):
     '''
         securely remove the file