#1040: on py 3.6 use sys.getfilesystemencodeerrors() to determined the default error handler instead of guessing it

4 files changed, 31 insertions, 15 deletions

diff --git a/docs/index.rst b/docs/index.rst
index 02f93570..350af93f 100644
--- a/docs/index.rst
+++ b/docs/index.rst
@@ -2250,11 +2250,13 @@ The notes below apply to *any* API returning a string such as
 :meth:`Process.exe` or :meth:`Process.cwd`, including non-filesystem related
 methods such as :meth:`Process.username` or :meth:`WindowsService.description`:
 
-* all strings are encoded by using the OS filesystem encoding which varies
-  depending on the platform (e.g. UTF-8 on Linux, mbcs on Win)
+* all strings are encoded by using the OS filesystem encoding
+  (``sys.getfilesystemencoding()``) which varies depending on the platform
+  (e.g. "UTF-8" on OSX, "mbcs" on Win)
 * no API call is supposed to crash with ``UnicodeDecodeError``
 * instead, in case of badly encoded data returned by the OS, the following error handlers are used to replace the corrupted characters in the string:
-    * Python 3: ``"surrogatescape"`` on POSIX and ``"replace"`` on Windows
+    * Python 3: ``sys.getfilesystemencodeerrors()`` (PY 3.6+) or
+      ``"surrogatescape"`` on POSIX and ``"replace"`` on Windows
     * Python 2: ``"replace"``
 * on Python 2 all APIs return bytes (``str`` type), never ``unicode``
 * on Python 2, you can go back to ``unicode`` by doing:
diff --git a/psutil/_common.py b/psutil/_common.py
index 54cb1ff5..e6ba9dc3 100644
--- a/psutil/_common.py
+++ b/psutil/_common.py
@@ -27,6 +27,8 @@ try:
 except ImportError:
     AF_UNIX = None
 
+from psutil._compat import PY3
+
 if sys.version_info >= (3, 4):
     import enum
 else:
@@ -132,6 +134,17 @@ else:
 
     globals().update(BatteryTime.__members__)
 
+# --- others
+
+ENCODING = sys.getfilesystemencoding()
+if not PY3:
+    ENCODING_ERRS = "replace"
+else:
+    try:
+        ENCODING_ERRS = sys.getfilesystemencodeerrors()  # py 3.6
+    except AttributeError:
+        ENCODING_ERRS = "surrogateescape" if POSIX else "replace"
+
 
 # ===================================================================
 # --- namedtuples
diff --git a/psutil/_pslinux.py b/psutil/_pslinux.py
index 1a890a87..7a03940a 100644
--- a/psutil/_pslinux.py
+++ b/psutil/_pslinux.py
@@ -25,13 +25,15 @@ from . import _common
 from . import _psposix
 from . import _psutil_linux as cext
 from . import _psutil_posix as cext_posix
+from ._common import ENCODING
+from ._common import ENCODING_ERRS
 from ._common import isfile_strict
 from ._common import memoize
 from ._common import memoize_when_activated
-from ._common import parse_environ_block
 from ._common import NIC_DUPLEX_FULL
 from ._common import NIC_DUPLEX_HALF
 from ._common import NIC_DUPLEX_UNKNOWN
+from ._common import parse_environ_block
 from ._common import path_exists_strict
 from ._common import supports_ipv6
 from ._common import usage_percent
@@ -84,9 +86,6 @@ BOOT_TIME = None  # set later
 BIGGER_FILE_BUFFERING = -1 if PY3 else 8192
 LITTLE_ENDIAN = sys.byteorder == 'little'
 SECTOR_SIZE_FALLBACK = 512
-if PY3:
-    FS_ENCODING = sys.getfilesystemencoding()
-    ENCODING_ERRORS_HANDLER = 'surrogateescape'
 if enum is None:
     AF_LINK = socket.AF_PACKET
 else:
@@ -200,14 +199,14 @@ def open_text(fname, **kwargs):
         # See:
         # https://github.com/giampaolo/psutil/issues/675
         # https://github.com/giampaolo/psutil/pull/733
-        kwargs.setdefault('encoding', FS_ENCODING)
-        kwargs.setdefault('errors', ENCODING_ERRORS_HANDLER)
+        kwargs.setdefault('encoding', ENCODING)
+        kwargs.setdefault('errors', ENCODING_ERRS)
     return open(fname, "rt", **kwargs)
 
 
 if PY3:
     def decode(s):
-        return s.decode(encoding=FS_ENCODING, errors=ENCODING_ERRORS_HANDLER)
+        return s.decode(encoding=ENCODING, errors=ENCODING_ERRS)
 else:
     def decode(s):
         return s
diff --git a/psutil/_pswindows.py b/psutil/_pswindows.py
index dd7b07a1..9f55194f 100644
--- a/psutil/_pswindows.py
+++ b/psutil/_pswindows.py
@@ -32,11 +32,13 @@ except ImportError as err:
         raise
 
 from ._common import conn_tmap
+from ._common import ENCODING
+from ._common import ENCODING_ERRS
 from ._common import isfile_strict
+from ._common import memoize_when_activated
 from ._common import parse_environ_block
 from ._common import sockfam_to_enum
 from ._common import socktype_to_enum
-from ._common import memoize_when_activated
 from ._common import usage_percent
 from ._compat import long
 from ._compat import lru_cache
@@ -71,8 +73,6 @@ __extra__all__ = [
 # --- globals
 # =====================================================================
 
-FS_ENCODING = sys.getfilesystemencoding()
-PY2_ENCODING_ERRS = "replace"
 CONN_DELETE_TCB = "DELETE_TCB"
 WAIT_TIMEOUT = 0x00000102  # 258 in decimal
 ACCESS_DENIED_SET = frozenset([errno.EPERM, errno.EACCES,
@@ -198,7 +198,7 @@ def py2_strencode(s):
         if isinstance(s, str):
             return s
         else:
-            return s.encode(FS_ENCODING, errors=PY2_ENCODING_ERRS)
+            return s.encode(ENCODING, errors=ENCODING_ERRS)
 
 
 # =====================================================================
@@ -240,7 +240,9 @@ disk_io_counters = cext.disk_io_counters
 def disk_usage(path):
     """Return disk usage associated with path."""
     if PY3 and isinstance(path, bytes):
-        path = path.decode(FS_ENCODING)
+        # XXX: do we want to use "strict"? Probably yes, in order
+        # to fail immediately. After all we are accepting input here...
+        path = path.decode(ENCODING, errors="strict")
     total, free = cext.disk_usage(path)
     used = total - free
     percent = usage_percent(used, total, _round=1)