X-Git-Url: https://git.arvados.org/arvados.git/blobdiff_plain/949d940b5f75114f86d381c347f0ba8f26b3e9b9..23f0fc06dbb6d7e82d820a8c65997f32c760f34e:/sdk/python/arvados/_normalize_stream.py

diff --git a/sdk/python/arvados/_normalize_stream.py b/sdk/python/arvados/_normalize_stream.py
index 9caef764ed..485c757e7f 100644
--- a/sdk/python/arvados/_normalize_stream.py
+++ b/sdk/python/arvados/_normalize_stream.py
@@ -8,7 +8,9 @@ from . import config
 import re
 
 def escape(path):
-    return re.sub('\\\\([0-3][0-7][0-7])', lambda m: '\\134'+m.group(1), path).replace(' ', '\\040')
+    path = re.sub('\\\\', lambda m: '\\134', path)
+    path = re.sub('[:\000-\040]', lambda m: "\\%03o" % ord(m.group(0)), path)
+    return path
 
 def normalize_stream(stream_name, stream):
     """Take manifest stream and return a list of tokens in normalized format.
@@ -53,13 +55,13 @@ def normalize_stream(stream_name, stream):
                 if streamoffset == current_span[1]:
                     current_span[1] += segment.segment_size
                 else:
-                    stream_tokens.append("{0}:{1}:{2}".format(current_span[0], current_span[1] - current_span[0], fout))
+                    stream_tokens.append(u"{0}:{1}:{2}".format(current_span[0], current_span[1] - current_span[0], fout))
                     current_span = [streamoffset, streamoffset + segment.segment_size]
 
         if current_span is not None:
-            stream_tokens.append("{0}:{1}:{2}".format(current_span[0], current_span[1] - current_span[0], fout))
+            stream_tokens.append(u"{0}:{1}:{2}".format(current_span[0], current_span[1] - current_span[0], fout))
 
         if not stream[streamfile]:
-            stream_tokens.append("0:0:{0}".format(fout))
+            stream_tokens.append(u"0:0:{0}".format(fout))
 
     return stream_tokens