From 0e0c1400b57d5de8aa8c18dd4897527f905a4b42 Mon Sep 17 00:00:00 2001
From: Tom Clegg <tom@curoverse.com>
Date: Sat, 1 Apr 2017 02:36:46 -0400
Subject: [PATCH] 11308: Fix bytes vs. str problems.

---
 sdk/python/arvados/keep.py           | 20 ++++++++++++--------
 sdk/python/tests/test_arv_put.py     |  5 ++++-
 sdk/python/tests/test_arvfile.py     |  2 +-
 sdk/python/tests/test_cache.py       |  4 ++--
 sdk/python/tests/test_keep_client.py |  4 ++--
 5 files changed, 21 insertions(+), 14 deletions(-)

diff --git a/sdk/python/arvados/keep.py b/sdk/python/arvados/keep.py
index f680d6ce10..ee91491efa 100644
--- a/sdk/python/arvados/keep.py
+++ b/sdk/python/arvados/keep.py
@@ -22,6 +22,11 @@ import threading
 from . import timer
 import urllib.parse
 
+if sys.version_info >= (3, 0):
+    from io import BytesIO
+else:
+    from cStringIO import StringIO as BytesIO
+
 import arvados
 import arvados.config as config
 import arvados.errors
@@ -335,7 +340,7 @@ class KeepClient(object):
             try:
                 with timer.Timer() as t:
                     self._headers = {}
-                    response_body = io.StringIO()
+                    response_body = BytesIO()
                     curl.setopt(pycurl.NOSIGNAL, 1)
                     curl.setopt(pycurl.OPENSOCKETFUNCTION, self._socket_open)
                     curl.setopt(pycurl.URL, url.encode('utf-8'))
@@ -412,8 +417,8 @@ class KeepClient(object):
             try:
                 with timer.Timer() as t:
                     self._headers = {}
-                    body_reader = io.StringIO(body)
-                    response_body = io.StringIO()
+                    body_reader = BytesIO(body)
+                    response_body = BytesIO()
                     curl.setopt(pycurl.NOSIGNAL, 1)
                     curl.setopt(pycurl.OPENSOCKETFUNCTION, self._socket_open)
                     curl.setopt(pycurl.URL, url.encode('utf-8'))
@@ -486,7 +491,8 @@ class KeepClient(object):
             curl.setopt(pycurl.LOW_SPEED_LIMIT, int(math.ceil(bandwidth_bps)))
 
         def _headerfunction(self, header_line):
-            header_line = header_line.decode('iso-8859-1')
+            if isinstance(header_line, bytes):
+                header_line = header_line.decode('iso-8859-1')
             if ':' in header_line:
                 name, value = header_line.split(':', 1)
                 name = name.strip().lower()
@@ -1066,10 +1072,8 @@ class KeepClient(object):
           KeepClient is initialized.
         """
 
-        if isinstance(data, str):
-            data = data.encode("ascii")
-        elif not isinstance(data, str):
-            raise arvados.errors.ArgumentError("Argument 'data' to KeepClient.put is not type 'str'")
+        if not isinstance(data, bytes):
+            data = data.encode()
 
         self.put_counter.add(1)
 
diff --git a/sdk/python/tests/test_arv_put.py b/sdk/python/tests/test_arv_put.py
index 667abbe63e..05e5e06b65 100644
--- a/sdk/python/tests/test_arv_put.py
+++ b/sdk/python/tests/test_arv_put.py
@@ -24,7 +24,10 @@ import threading
 import hashlib
 import random
 
-from io import StringIO
+if sys.version_info >= (3, 0):
+    from io import StringIO
+else:
+    from cStringIO import StringIO
 
 import arvados
 import arvados.commands.put as arv_put
diff --git a/sdk/python/tests/test_arvfile.py b/sdk/python/tests/test_arvfile.py
index f4bcd8a497..0b2b7412a5 100644
--- a/sdk/python/tests/test_arvfile.py
+++ b/sdk/python/tests/test_arvfile.py
@@ -36,7 +36,7 @@ class ArvadosFileWriterTestCase(unittest.TestCase):
             return self.blocks.get(locator)
         def put(self, data, num_retries=None, copies=None):
             pdh = tutil.str_keep_locator(data)
-            self.blocks[pdh] = str(data)
+            self.blocks[pdh] = bytes(data)
             return pdh
 
     class MockApi(object):
diff --git a/sdk/python/tests/test_cache.py b/sdk/python/tests/test_cache.py
index ea34c8051d..a1fbc27e35 100644
--- a/sdk/python/tests/test_cache.py
+++ b/sdk/python/tests/test_cache.py
@@ -34,10 +34,10 @@ class CacheTestThread(threading.Thread):
         for x in range(16):
             try:
                 data_in = _random(128)
-                data_in = md5.new(data_in).hexdigest() + "\n" + str(data_in)
+                data_in = md5.new(data_in).hexdigest() + bytes("\n") + bytes(data_in)
                 c.set(url, data_in)
                 data_out = c.get(url)
-                digest, content = data_out.split("\n", 1)
+                digest, _, content = data_out.partition("\n")
                 if digest != md5.new(content).hexdigest():
                     self.ok = False
             except Exception as err:
diff --git a/sdk/python/tests/test_keep_client.py b/sdk/python/tests/test_keep_client.py
index 8c2a67130a..1b0b627a65 100644
--- a/sdk/python/tests/test_keep_client.py
+++ b/sdk/python/tests/test_keep_client.py
@@ -107,8 +107,8 @@ class KeepTestCase(run_test_server.TestCaseWithServers):
             # Error if it is not ASCII
             self.keep_client.put(u'\xe2')
 
-        with self.assertRaises(arvados.errors.ArgumentError):
-            # Must be a string type
+        with self.assertRaises(AttributeError):
+            # Must be bytes or have an encode() method
             self.keep_client.put({})
 
     def test_KeepHeadTest(self):
-- 
2.30.2