[gnue] r10110 - trunk/gnue-common/src/datasources/drivers/file

[reinhard]

Author: reinhard
Date: 2009-12-09 16:01:12 -0600 (Wed, 09 Dec 2009)
New Revision: 10110

Modified:
   trunk/gnue-common/src/datasources/drivers/file/csvfile.py
Log:
Allow for non ASCII content in CSV files.


Modified: trunk/gnue-common/src/datasources/drivers/file/csvfile.py
===================================================================
--- trunk/gnue-common/src/datasources/drivers/file/csvfile.py   2009-12-09 
21:36:38 UTC (rev 10109)
+++ trunk/gnue-common/src/datasources/drivers/file/csvfile.py   2009-12-09 
22:01:12 UTC (rev 10110)
@@ -28,6 +28,7 @@
 __all__ = ['Connection']
 
 import csv
+import locale
 
 from gnue.common.datasources import GSchema
 from gnue.common.datasources.drivers.file import Base
@@ -82,6 +83,22 @@
   """
 
   # ---------------------------------------------------------------------------
+  # Constructor
+  # ---------------------------------------------------------------------------
+
+  def __init__ (self, connections, name, parameters):
+
+    Base.Connection.__init__ (self, connections, name, parameters)
+
+    if parameters.has_key ("encoding"):
+      self.__encoding = parameters ['encoding']
+    else:
+      self.__encoding = locale.getlocale()[1]
+      if self.__encoding is None:
+          self.__encoding = 'ascii'
+
+
+  # ---------------------------------------------------------------------------
   # Iterate through the list of field names
   # ---------------------------------------------------------------------------
 
@@ -110,8 +127,9 @@
 
     reader = csv.DictReader (f, fieldnames, dialect = dialect)
 
-    # Make a real list of dictionaries
-    return [row for row in reader]
+    # Make a real list of dictionaries and convert to unicode.
+    return [dict( [(k, unicode(v, self.__encoding, errors='replace')) \
+                for (k, v) in row.iteritems()]) for row in reader]
 
 
   # ---------------------------------------------------------------------------