Add support for EPUB files (#1439)

* Add support for EPUB files

Also generalize open document zip mime detection

* Fix exception handling in MimeFromOpenDocument

hydrus/client/caches/ClientCaches.py

@@ -746,7 +746,7 @@ def Clear( self ):
 
             self._special_thumbs = {}
 
-            names = [ 'hydrus', 'pdf', 'psd', 'clip', 'sai', 'krita', 'xcf', 'svg', 'audio', 'video', 'zip' ]
+            names = [ 'hydrus', 'pdf', 'psd', 'clip', 'sai', 'krita', 'xcf', 'svg', 'audio', 'video', 'zip', 'epub' ]
 
             bounding_dimensions = self._controller.options[ 'thumbnail_dimensions' ]
             thumbnail_scale_type = self._controller.new_options.GetInteger( 'thumbnail_scale_type' )
@@ -861,6 +861,7 @@ def GetThumbnail( self, media ):
             elif mime in HC.AUDIO: return self._special_thumbs[ 'audio' ]
             elif mime in HC.VIDEO: return self._special_thumbs[ 'video' ]
             elif mime == HC.APPLICATION_PDF: return self._special_thumbs[ 'pdf' ]
+            elif mime == HC.APPLICATION_EPUB: return self._special_thumbs[ 'epub' ]
             elif mime == HC.APPLICATION_PSD: return self._special_thumbs[ 'psd' ]
             elif mime == HC.APPLICATION_SAI2: return self._special_thumbs[ 'sai' ]
             elif mime == HC.APPLICATION_KRITA: return self._special_thumbs[ 'krita' ]

hydrus/core/HydrusArchiveHandling.py

@@ -1,5 +1,7 @@
 import zipfile
 
+from hydrus.core import HydrusConstants as HC
+
 def ExtractSingleFileFromZip( path_to_zip, filename_to_extract, extract_into_file_path ):
 
     with zipfile.ZipFile( path_to_zip ) as zip_handle:
@@ -26,3 +28,18 @@ def ReadSingleFileFromZip( path_to_zip, filename_to_extract ):
 def GetZipAsPath( path_to_zip, path_in_zip="" ):
 
     return zipfile.Path( path_to_zip, at=path_in_zip )
+
+
+def MimeFromOpenDocument( path ):
+
+    try:
+
+        mimetype_data = GetZipAsPath( path, 'mimetype' ).read_text()
+
+        filetype = HC.mime_enum_lookup.get(mimetype_data, None)
+
+        return filetype if filetype in HC.OPEN_DOCUMENT_ZIPS else None
+
+    except:
+
+        return None

hydrus/core/HydrusConstants.py

@@ -730,6 +730,7 @@
 IMAGE_GIF = 68
 APPLICATION_PROCREATE = 69
 IMAGE_QOI = 70
+APPLICATION_EPUB = 71
 APPLICATION_OCTET_STREAM = 100
 APPLICATION_UNKNOWN = 101
 
@@ -778,6 +779,7 @@
     APPLICATION_XCF,
     APPLICATION_PROCREATE,
     APPLICATION_PDF,
+    APPLICATION_EPUB,
     APPLICATION_ZIP,
     APPLICATION_RAR,
     APPLICATION_7Z,
@@ -861,7 +863,8 @@
 
 APPLICATIONS = [
     APPLICATION_FLASH,
-    APPLICATION_PDF
+    APPLICATION_PDF,
+    APPLICATION_EPUB
 ]
 
 IMAGE_PROJECT_FILES = [
@@ -883,6 +886,9 @@
 
 VIEWABLE_IMAGE_PROJECT_FILES = { APPLICATION_PSD }
 
+# zip files that have a `mimetype` file inside
+OPEN_DOCUMENT_ZIPS = { APPLICATION_KRITA, APPLICATION_EPUB }
+
 general_mimetypes_to_mime_groups = {
     GENERAL_APPLICATION : APPLICATIONS,
     GENERAL_APPLICATION_ARCHIVE : ARCHIVES,
@@ -968,6 +974,7 @@
     'application/x-yaml' : APPLICATION_YAML,
     'PDF document' : APPLICATION_PDF,
     'application/pdf' : APPLICATION_PDF,
+    'application/epub+zip' : APPLICATION_EPUB,
     'application/zip' : APPLICATION_ZIP,
     'application/vnd.rar' : APPLICATION_RAR,
     'application/x-7z-compressed' : APPLICATION_7Z,
@@ -1031,6 +1038,7 @@
     APPLICATION_JSON : 'json',
     APPLICATION_CBOR : 'cbor',
     APPLICATION_PDF : 'pdf',
+    APPLICATION_EPUB : 'epub',
     APPLICATION_PSD : 'psd',
     APPLICATION_CLIP : 'clip',
     APPLICATION_SAI2 : 'sai2',
@@ -1108,6 +1116,7 @@
     APPLICATION_JSON : 'application/json',
     APPLICATION_CBOR : 'application/cbor',
     APPLICATION_PDF : 'application/pdf',
+    APPLICATION_EPUB : 'application/epub+zip',
     APPLICATION_PSD : 'image/vnd.adobe.photoshop',
     APPLICATION_CLIP : 'application/clip', # made up
     APPLICATION_SAI2: 'application/sai2', # made up
@@ -1183,6 +1192,7 @@
     APPLICATION_YAML : '.yaml',
     APPLICATION_JSON : '.json',
     APPLICATION_PDF : '.pdf',
+    APPLICATION_EPUB : '.epub',
     APPLICATION_PSD : '.psd',
     APPLICATION_CLIP : '.clip',
     APPLICATION_SAI2: '.sai2',

hydrus/core/HydrusFileHandling.py

@@ -5,6 +5,7 @@
 from hydrus.core import HydrusAudioHandling
 from hydrus.core import HydrusPSDHandling
 from hydrus.core import HydrusClipHandling
+from hydrus.core import HydrusArchiveHandling
 from hydrus.core import HydrusConstants as HC
 from hydrus.core import HydrusData
 from hydrus.core import HydrusDocumentHandling
@@ -75,6 +76,8 @@
     ( ( ( 42, b'application/x-krita' ), ), HC.APPLICATION_KRITA ), # https://gitlab.freedesktop.org/xdg/shared-mime-info/-/blob/master/data/freedesktop.org.xml.in#L2829
     ( ( ( 58, b'application/x-krita' ), ), HC.APPLICATION_KRITA ), 
     ( ( ( 63, b'application/x-krita' ), ), HC.APPLICATION_KRITA ), 
+    ( ( ( 38, b'application/epub+zip' ), ), HC.APPLICATION_EPUB ), 
+    ( ( ( 43, b'application/epub+zip' ), ), HC.APPLICATION_EPUB ),
     ( ( ( 0, b'PK\x03\x04' ), ), HC.APPLICATION_ZIP ),
     ( ( ( 0, b'PK\x05\x06' ), ), HC.APPLICATION_ZIP ),
     ( ( ( 0, b'PK\x07\x08' ), ), HC.APPLICATION_ZIP ),
@@ -620,19 +623,18 @@ def GetMime( path, ok_to_look_for_hydrus_updates = False ):
         if it_passes:
 
             if mime == HC.APPLICATION_ZIP:
+
+                opendoc_mime = HydrusArchiveHandling.MimeFromOpenDocument( path )
+
+                if opendoc_mime is not None:
+
+                    return opendoc_mime
 
-                # TODO: since we'll be expanding this to other zip-likes, we should make the zipfile object up here and pass that to various checkers downstream
-                if HydrusKritaHandling.ZipLooksLikeAKrita( path ):
-
-                    return HC.APPLICATION_KRITA
-
-                elif HydrusProcreateHandling.ZipLooksLikeProcreate( path ):
+                if HydrusProcreateHandling.ZipLooksLikeProcreate( path ):
 
                     return HC.APPLICATION_PROCREATE
 
-                else:
-
-                    return HC.APPLICATION_ZIP
+                return HC.APPLICATION_ZIP
 
 
             if mime in ( HC.UNDETERMINED_WM, HC.UNDETERMINED_MP4 ):

hydrus/core/HydrusKritaHandling.py

@@ -6,7 +6,6 @@
 
 KRITA_FILE_THUMB = "preview.png"
 KRITA_FILE_MERGED = "mergedimage.png"
-KRITA_MIMETYPE = 'mimetype'
 
 def ExtractZippedImageToPath( path_to_zip, temp_path_file ):
 
@@ -63,29 +62,11 @@ def GetKraProperties( path ):
                     if width is not None and height is not None:
 
                         break
-                        
+
 
-
-
-
     except KeyError:
 
         raise HydrusExceptions.NoResolutionFileException( f'This krita file had no {DOCUMENT_INFO_FILE}!' )
 
 
     return ( width, height )
-
-
-def ZipLooksLikeAKrita( path ):
-
-    try:
-
-        mimetype_data = HydrusArchiveHandling.ReadSingleFileFromZip( path, KRITA_MIMETYPE )
-
-        return b'application/x-krita' in mimetype_data
-
-    except KeyError:
-
-        return False
-
-

hydrus/core/HydrusPaths.py

@@ -22,6 +22,7 @@
 mimes_to_default_thumbnail_paths = collections.defaultdict( lambda: os.path.join( HC.STATIC_DIR, 'hydrus.png' ) )
 
 mimes_to_default_thumbnail_paths[ HC.APPLICATION_PDF ] = os.path.join( HC.STATIC_DIR, 'pdf.png' )
+mimes_to_default_thumbnail_paths[ HC.APPLICATION_EPUB ] = os.path.join( HC.STATIC_DIR, 'epub.png' )
 mimes_to_default_thumbnail_paths[ HC.APPLICATION_PSD ] = os.path.join( HC.STATIC_DIR, 'psd.png' )
 mimes_to_default_thumbnail_paths[ HC.APPLICATION_CLIP ] = os.path.join( HC.STATIC_DIR, 'clip.png' )
 mimes_to_default_thumbnail_paths[ HC.APPLICATION_SAI2 ] = os.path.join( HC.STATIC_DIR, 'sai.png' )