Fixes unicode support on gtfs

gtfs/entity/unmapped_entities.py

@@ -7,7 +7,9 @@ def __init__(self, **kwargs):
         attrvaluestr = kwargs[attrname]
 	if attrvaluestr == '':
 	  attrvalue = None
-	else:
+	elif attrtype == str:
+          attrvalue = unicode(attrvaluestr)
+        else:
 	  attrvalue = attrtype( attrvaluestr )
 
         setattr( self, attrname, attrvalue )

gtfs/feed.py

@@ -11,7 +11,8 @@ def __init__(self,header,row):
     self.row = row 
 
   def to_dict(self):
-    return dict([(fieldname,self.row[fieldindex] if fieldindex<len(self.row) else None) for fieldname,fieldindex in self.header.items()])
+    return dict([(fieldname,self.row[fieldindex] if fieldindex<len(self.row) 
+                  else None) for fieldname,fieldindex in self.header.items()])
 
   def __repr__(self):
     return repr(self.to_dict())
@@ -41,8 +42,8 @@ def next(self):
     return Record( self.header, self.rows.next() )
 
 class Feed(object):
-  """A Feed is a collection of CSV files with headers, either zipped into an archive
-     or loose in a folder"""
+  """A Feed is a collection of CSV files with headers, either zipped into an
+  archive or loose in a folder"""
 
   def __init__(self, filename):
     self.filename = filename 
@@ -51,15 +52,23 @@ def __init__(self, filename):
     if not os.path.isdir( filename ):
       self.zf = ZipFile( filename )
 
+  def unicode_csv_reader(self, content_iter):
+    reader = csv.reader( [ x.encode("utf-8") for x in iterdecode(content_iter,
+                                                                 "utf-8") ] )
+    for row in reader:
+      yield [ unicode(x, "utf-8") for x in row ]
+    return
+
   def get_rows(self, filename):
     if self.zf:
       try:
         contents = self.zf.read(filename)
       except KeyError:
         raise KeyError( "%s is not present feed"%filename )
-      return csv.reader( iterdecode( contents.split("\n"), "utf-8" ) )
+      return self.unicode_csv_reader( contents.split("\n") )
     else:
-      return csv.reader( iterdecode( open( os.path.join( self.filename, filename ) ), "utf-8" ) )
+      return self.unicode_csv_reader( open( os.path.join( self.filename,
+                                                          filename ) ) )
 
   def get_table(self, filename):
     rows = self.get_rows( filename )