Merge pull request #11 from cswegger/tablib

---

This change applies the same unicode CSV fix for TSV files, since all its done in the exporter is changing a few parameters of the CSV module.

All unit tests are still passing after this change.

HISTORY.rst

@@ -1,6 +1,12 @@
 History
 -------
 
+0.9.8
++++++
+
+* Full Unicode TSV support
+
+
 0.9.7 (2011-05-12)
 ++++++++++++++++++
 

tablib/formats/_tsv.py

@@ -5,11 +5,15 @@
 
 import sys
 if sys.version_info[0] > 2:
+    is_py3 = True
     from io import StringIO
+    import csv
 else:
+    is_py3 = False
     from cStringIO import StringIO
-    
-import csv
+    import tablib.packages.unicodecsv as csv
+
+
 import os
 
 import tablib
@@ -18,12 +22,16 @@ import tablib
 title = 'tsv'
 extentions = ('tsv',)
 
-
+DEFAULT_ENCODING = 'utf-8'
 
 def export_set(dataset):
     """Returns a TSV representation of Dataset."""
     stream = StringIO()
-    _tsv = csv.writer(stream, delimiter='\t')
+
+    if is_py3:
+        _tsv = csv.writer(stream, delimiter="\t")
+    else:
+        _tsv = csv.writer(stream, encoding=DEFAULT_ENCODING, delimiter="\t")
 
     for row in dataset._package(dicts=False):
         _tsv.writerow(row)
@@ -35,7 +43,12 @@ def import_set(dset, in_stream, headers=True):
     """Returns dataset from TSV stream."""
     dset.wipe()
 
-    rows = csv.reader(in_stream.split('\r\n'), delimiter='\t')
+    if is_py3:
+        rows = csv.reader(in_stream.split('\r\n'), delimiter='\t')
+    else:
+        rows = csv.reader(in_stream.split('\r\n'), delimiter='\t',
+                          encoding=DEFAULT_ENCODING)
+
     for i, row in enumerate(rows):
         # Skip empty rows
         if not row: