Explicitly use utf-8 when decoding bytestrings

While Python 3 defaults to utf-8 in `bytes.decode()`, Python 2's
equivalent (`str.decode()`) will use the default encoding as set by
site.py (which is almost always ascii).

From looking at the code, it seems that these decodes have just sort of
been fixed piecemeal (likely when someone realized that pygit2 was
failing to handle unicode properly, but any decodes which run on Python
2 that don't specify utf-8 as the encoding are a ticking time bomb. I
personally noticed this was a problem when I encountered a traceback in
the RemoteCallbacks while fetching a new branch which contained utf-8
characters. During the fetch, when `pygit2.remote.maybe_string()` was
invoked by `_update_tips_cb()` with a pointer to a bytestring containing
unicode, the decode fails because the default encoding is ascii. As it
turns out, this was fixed in master, but there are a number which still
have no explicit encoding.

This commit explicitly uses utf-8 for all remaining bytestring decodes
which do not have an encoding specified, aside from one in PY3-specific
code where doing so would be redundant.

pygit2/blame.py

@@ -97,7 +97,7 @@ def orig_path(self):
         if not path:
             return None
 
-        return ffi.string(path).decode()
+        return ffi.string(path).decode('utf-8')
 
 
 class Blame(object):

pygit2/config.py

@@ -263,7 +263,7 @@ def _from_found_config(fn):
         buf = ffi.new('git_buf *', (ffi.NULL, 0))
         err = fn(buf)
         check_error(err, True)
-        cpath = ffi.string(buf.ptr).decode()
+        cpath = ffi.string(buf.ptr).decode('utf-8')
         C.git_buf_free(buf)
 
         return Config(cpath)

pygit2/refspec.py

@@ -43,12 +43,12 @@ def __init__(self, owner, ptr):
     @property
     def src(self):
         """Source or lhs of the refspec"""
-        return ffi.string(C.git_refspec_src(self._refspec)).decode()
+        return ffi.string(C.git_refspec_src(self._refspec)).decode('utf-8')
 
     @property
     def dst(self):
         """Destinaton or rhs of the refspec"""
-        return ffi.string(C.git_refspec_dst(self._refspec)).decode()
+        return ffi.string(C.git_refspec_dst(self._refspec)).decode('utf-8')
 
     @property
     def force(self):
@@ -58,7 +58,7 @@ def force(self):
     @property
     def string(self):
         """String which was used to create this refspec"""
-        return ffi.string(C.git_refspec_string(self._refspec)).decode()
+        return ffi.string(C.git_refspec_string(self._refspec)).decode('utf-8')
 
     @property
     def direction(self):
@@ -82,7 +82,7 @@ def _transform(self, ref, fn):
         check_error(err)
 
         try:
-            return ffi.string(buf.ptr).decode()
+            return ffi.string(buf.ptr).decode('utf-8')
         finally:
             C.git_buf_free(buf)
 

pygit2/remote.py

@@ -238,7 +238,7 @@ def _sideband_progress_cb(string, length, data):
             return 0
 
         try:
-            s = ffi.string(string, length).decode()
+            s = ffi.string(string, length).decode('utf-8')
             progress(s)
         except Exception as e:
             self._stored_exception = e

pygit2/utils.py

@@ -45,7 +45,7 @@
 def strarray_to_strings(arr):
     l = [None] * arr.count
     for i in range(arr.count):
-        l[i] = ffi.string(arr.strings[i]).decode()
+        l[i] = ffi.string(arr.strings[i]).decode('utf-8')
 
     return l