From 2e407c0b21730dc5c768cc16e8844cacc2190c28 Mon Sep 17 00:00:00 2001
From: Nate Coraor <nate@bx.psu.edu>
Date: Tue, 25 Aug 2009 14:58:37 -0400
Subject: [PATCH] Fix for multi-byte file uploads

---
 lib/galaxy/tools/actions/upload.py      | 1 +
 lib/galaxy/tools/parameters/grouping.py | 2 ++
 tools/data_source/upload.py             | 6 +++++-
 3 files changed, 8 insertions(+), 1 deletion(-)

diff --git a/lib/galaxy/tools/actions/upload.py b/lib/galaxy/tools/actions/upload.py
index ec5338d579e..f2177eeecf4 100644
--- a/lib/galaxy/tools/actions/upload.py
+++ b/lib/galaxy/tools/actions/upload.py
@@ -104,6 +104,7 @@ class UploadToolAction( ToolAction ):
                     except:
                         is_binary = None
                     json = dict( file_type = uploaded_dataset.file_type,
+                                 ext = uploaded_dataset.ext,
                                  name = uploaded_dataset.name,
                                  dataset_id = data.dataset.id,
                                  dbkey = uploaded_dataset.dbkey,
diff --git a/lib/galaxy/tools/parameters/grouping.py b/lib/galaxy/tools/parameters/grouping.py
index adda7f80db7..e05f8f0213a 100644
--- a/lib/galaxy/tools/parameters/grouping.py
+++ b/lib/galaxy/tools/parameters/grouping.py
@@ -327,6 +327,8 @@ class UploadDataset( Group ):
             rval = []
             for dataset in datasets:
                 dataset.file_type = file_type
+                dataset.datatype = d_type
+                dataset.ext = self.get_datatype_ext( trans, context )
                 dataset.dbkey = dbkey
                 rval.append( dataset )
             return rval
diff --git a/tools/data_source/upload.py b/tools/data_source/upload.py
index 5b40b28ddd9..97500008804 100644
--- a/tools/data_source/upload.py
+++ b/tools/data_source/upload.py
@@ -136,7 +136,7 @@ def add_file( dataset, json_file ):
         file_err( 'The uploaded file is empty', dataset, json_file )
         return
     if 'is_multi_byte' not in dir( dataset ):
-        dataset.is_multi_byte = util.is_multi_byte( open( dataset.path, 'r' ).read( 1024 )[:100] )
+        dataset.is_multi_byte = util.is_multi_byte( open( dataset.path, 'r' ).read( 1024 ) )
     if dataset.is_multi_byte:
         ext = sniff.guess_ext( dataset.path, is_multi_byte=True )
         data_type = ext
@@ -225,6 +225,10 @@ def add_file( dataset, json_file ):
                 ext = dataset.file_type
             data_type = ext
     # Save job info for the framework
+    if ext == 'auto' and dataset.ext:
+        ext = dataset.ext
+    if ext == 'auto':
+        ext = 'data'
     info = dict( type = 'dataset',
                  dataset_id = dataset.dataset_id,
                  path = dataset.path,