Fixed io types to make tests work on PY2 once again.

Now it's about going through PY3 issues

Author: Byron

git/compat.py

@@ -16,7 +16,8 @@
 
 from gitdb.utils.encoding import (
     string_types,
-    text_type
+    text_type,
+    force_bytes
 )
 
 if PY3:

git/index/base.py

@@ -8,7 +8,7 @@
 import sys
 import subprocess
 import glob
-from io import StringIO
+from io import BytesIO
 
 from stat import S_ISLNK
 
@@ -43,6 +43,7 @@
     izip,
     xrange,
     string_types,
+    force_bytes
 )
 
 from git.util import (
@@ -562,7 +563,8 @@ def _store_path(self, filepath, fprogress):
         st = os.lstat(filepath)     # handles non-symlinks as well
         stream = None
         if S_ISLNK(st.st_mode):
-            stream = StringIO(os.readlink(filepath))
+            # in PY3, readlink is string, but we need bytes. In PY2, it's just OS encoded bytes, we assume UTF-8
+            stream = BytesIO(force_bytes(os.readlink(filepath), encoding='utf-8'))
         else:
             stream = open(filepath, 'rb')
         # END handle stream

git/index/fun.py

@@ -12,7 +12,7 @@
 
 S_IFGITLINK = S_IFLNK | S_IFDIR     # a submodule
 
-from io import StringIO
+from io import BytesIO
 
 from git.util import IndexFileSHA1Writer
 from git.exc import UnmergedEntriesError
@@ -218,7 +218,7 @@ def write_tree_from_cache(entries, odb, sl, si=0):
     # END for each entry
 
     # finally create the tree
-    sio = StringIO()
+    sio = BytesIO()
     tree_to_stream(tree_items, sio.write)
     sio.seek(0)
 

git/objects/commit.py

@@ -30,7 +30,7 @@
     altzone
 )
 import os
-from io import StringIO
+from io import BytesIO
 import logging
 
 log = logging.getLogger('git.objects.commit')
@@ -133,7 +133,7 @@ def _set_cache_(self, attr):
         if attr in Commit.__slots__:
             # read the data in a chunk, its faster - then provide a file wrapper
             binsha, typename, self.size, stream = self.repo.odb.stream(self.binsha)
-            self._deserialize(StringIO(stream.read()))
+            self._deserialize(BytesIO(stream.read()))
         else:
             super(Commit, self)._set_cache_(attr)
         # END handle attrs
@@ -345,7 +345,7 @@ def create_from_tree(cls, repo, tree, message, parent_commits=None, head=False,
                          committer, committer_time, committer_offset,
                          message, parent_commits, conf_encoding)
 
-        stream = StringIO()
+        stream = BytesIO()
         new_commit._serialize(stream)
         streamlen = stream.tell()
         stream.seek(0)

git/objects/submodule/base.py

@@ -8,7 +8,7 @@
     find_first_remote_branch
 )
 from git.objects.util import Traversable
-from io import StringIO                   # need a dict to set bloody .name field
+from io import BytesIO                   # need a dict to set bloody .name field
 from git.util import (
     Iterable,
     join_path_native,
@@ -187,8 +187,8 @@ def _clear_cache(self):
 
     @classmethod
     def _sio_modules(cls, parent_commit):
-        """:return: Configuration file as StringIO - we only access it through the respective blob's data"""
-        sio = StringIO(parent_commit.tree[cls.k_modules_file].data_stream.read())
+        """:return: Configuration file as BytesIO - we only access it through the respective blob's data"""
+        sio = BytesIO(parent_commit.tree[cls.k_modules_file].data_stream.read())
         sio.name = cls.k_modules_file
         return sio
 

git/objects/submodule/util.py

@@ -1,7 +1,7 @@
 import git
 from git.exc import InvalidGitRepositoryError
 from git.config import GitConfigParser
-from io import StringIO
+from io import BytesIO
 import weakref
 
 __all__ = ('sm_section', 'sm_name', 'mkhead', 'unbare_repo', 'find_first_remote_branch',
@@ -83,7 +83,7 @@ def flush_to_index(self):
         """Flush changes in our configuration file to the index"""
         assert self._smref is not None
         # should always have a file here
-        assert not isinstance(self._file_or_files, StringIO)
+        assert not isinstance(self._file_or_files, BytesIO)
 
         sm = self._smref()
         if sm is not None:

git/refs/log.py

@@ -85,7 +85,7 @@ def from_line(cls, line):
         :param line: line without trailing newline
         :raise ValueError: If line could not be parsed"""
         try:
-            info, msg = line.split('\t', 2)
+            info, msg = line.split('\t', 1)
         except ValueError:
             raise ValueError("line is missing tab separator")
         # END handle first plit

git/test/fixtures/git_config_global

@@ -1,3 +1,4 @@
+# just a comment
 [alias]
 	st = status
 	ci = commit

git/test/lib/helper.py

@@ -48,8 +48,8 @@ class StringProcessAdapter(object):
     Its tailored to work with the test system only"""
 
     def __init__(self, input_string):
-        self.stdout = io.StringIO(input_string)
-        self.stderr = io.StringIO()
+        self.stdout = io.BytesIO(input_string)
+        self.stderr = io.BytesIO()
 
     def wait(self):
         return 0

git/test/performance/test_commit.py

@@ -4,7 +4,7 @@
 # This module is part of GitPython and is released under
 # the BSD License: http://www.opensource.org/licenses/bsd-license.php
 from __future__ import print_function
-from io import StringIO
+from io import BytesIO
 from time import time
 import sys
 
@@ -93,7 +93,7 @@ def test_commit_serialization(self):
                         hc.committer, hc.committed_date, hc.committer_tz_offset,
                         str(i), parents=hc.parents, encoding=hc.encoding)
 
-            stream = StringIO()
+            stream = BytesIO()
             cm._serialize(stream)
             slen = stream.tell()
             stream.seek(0)