* src/DisplayModels.py: try to handle encoding errors

* src/ImageSelect.py: call fix_encoding to handle path names
* src/Utils.py: provide a routine to attempt to fix encoding of
str values
* src/plugins/Check.py: repair bad filenames


svn: r5403

gramps2/ChangeLog

@@ -1,3 +1,10 @@
+2005-11-16  Don Allingham  <don@gramps-project.org>
+	* src/DisplayModels.py: try to handle encoding errors
+	* src/ImageSelect.py: call fix_encoding to handle path names
+	* src/Utils.py: provide a routine to attempt to fix encoding of
+	str values
+	* src/plugins/Check.py: repair bad filenames
+
 2005-11-16  Eero Tamminen  <eerot@sf>
 	* src/po/fi.po: Translation update to latest template,
 	  finetune keyboard shortcuts.

gramps2/src/DisplayModels.py

@@ -456,10 +456,16 @@ class MediaModel(BaseModel):
         return len(self.fmap)+1
 
     def column_description(self,data):
+        try:
             return unicode(data[4])
+        except:
+            return unicode(data[4],'latin1')
 
     def column_path(self,data):
+        try:
             return unicode(data[2])
+        except:
+            return unicode(data[2].encode('iso-8859-1'))
 
     def column_mime(self,data):
         if data[3]:

gramps2/src/ImageSelect.py

@@ -486,10 +486,10 @@ class Gallery(ImageSelect):
     def on_photolist_drag_data_received(self,w, context, x, y, data, info, time):
         if data and data.format == 8:
             icon_index = self.get_index(w,x,y)
-            d = data.data.replace('\0',' ').strip()
+            d = Utils.fix_encoding(data.data.replace('\0',' ').strip())
             protocol,site,mfile,j,k,l = urlparse.urlparse(d)
             if protocol == "file":
-                name = mfile
+                name = Utils.fix_encoding(mfile)
                 mime = GrampsMime.get_type(name)
                 photo = RelLib.MediaObject()
                 photo.set_path(name)
@@ -515,6 +515,7 @@ class Gallery(ImageSelect):
                     t = _("Could not import %s") % d
                     ErrorDialog(t,str(msg))
                     return
+                tfile = Utils.fix_encoding(tfile)
                 mime = GrampsMime.get_type(tfile)
                 photo = RelLib.MediaObject()
                 photo.set_mime_type(mime)

gramps2/src/Utils.py

@@ -63,6 +63,19 @@ def history_broken():
 data_recover_msg = _('The data can only be recovered by Undo operation '
             'or by quitting with abandoning changes.')
 
+def fix_encoding(value):
+    import locale
+    if type(value) != unicode:
+        try:
+            return unicode(value)
+        except:
+            codeset = locale.getpreferredencoding()
+            if codeset == 'UTF-8':
+                codeset = 'latin1'
+            return unicode(value,codeset)
+    else:
+        return value
+
 #-------------------------------------------------------------------------
 #
 # force_unicode

gramps2/src/plugins/Check.py

@@ -73,6 +73,7 @@ class Check(Tool.Tool):
             trans.set_batch(True)
             db.disable_signals()
             checker = CheckIntegrity(db,parent,trans)
+            checker.fix_encoding()
             checker.cleanup_missing_photos(cli)
             
             prev_total = -1
@@ -154,6 +155,31 @@ class CheckIntegrity:
             self.progress.step()
         cursor.close()
 
+
+    def fix_encoding(self):
+        import locale
+        codeset = locale.nl_langinfo(locale.CODESET)
+        if codeset == 'UTF-8':
+            codeset = 'latin1'
+        
+        self.progress.set_pass(_('Looking for character encoding errors'),
+                               self.db.get_number_of_media_objects())
+
+        cursor = self.db.get_media_cursor()
+        value = cursor.first()
+        while value:
+            (handle,data) = value
+            if type(data[2]) != unicode or type(data[4]) != unicode:
+                obj = self.db.get_object_from_handle(handle)
+                if type(obj.path) != unicode:
+                    obj.path = unicode(obj.path,codeset)
+                if type(obj.desc) != unicode:
+                    obj.desc = unicode(obj.desc,codeset)
+                self.db.commit_media_object(obj,self.trans)
+            self.progress.step()
+            value = cursor.next()
+        cursor.close()
+
     def check_for_broken_family_links(self):
         # Check persons referenced by the family objects