[checkmk-commits] Improved Unicode robustness.

Module: check_mk Branch: master Commit: e4db79755a9feeb90e203fabfc5e8a0ab7aa4e30 URL: http://git.mathias-kettner.de/git/?p=check_mk.git;a=commit;h=e4db79755a9fee… Author: Sven Panne <sp(a)mathias-kettner.de> Date: Wed May 11 10:26:58 2016 +0200 Improved Unicode robustness. --- bin/mkeventd | 13 +++++++++++++ 1 file changed, 13 insertions(+) diff --git a/bin/mkeventd b/bin/mkeventd index 134e33e..f060feb 100755 --- a/bin/mkeventd +++ b/bin/mkeventd @@ -97,6 +97,19 @@ history_columns = [ # Alas, we often have no clue about the actual encoding, so we have to guess: # Initially we assume UTF-8, but fall back to latin-1 if it didn't work. def decode_from_bytes(string_as_bytes): + # This is just a safeguard if we are inadvertedly called with a Unicode + # string. In theory this should never happen, but given the typing chaos in + # this script, one never knows. In the Unicode case, Python tries to be + # "helpful", but this fails miserably: Calling 'decode' on a Unicode string + # implicitly converts it via 'encode("ascii")' to a byte string first, but + # this can of course fail and doesn't make sense at all when we immediately + # call 'decode' on this byte string again. In a nutshell: The implicit + # conversions between str and unicode are a fundamentally broken idea, just + # like all implicit things and "helpful" ideas in general. :-P For further + # info see e.g. http://nedbatchelder.com/text/unipain.html + if type(string_as_bytes) == unicode: + return string_as_bytes + try: return string_as_bytes.decode("utf-8") except: