commit 2006edc5ee1878bd5c98e99b3f00c987838ba297 Author: Damian Johnson atagar@torproject.org Date: Sun Jan 5 12:54:42 2020 -0800

Differentiate bytes/unicode in hashes

Huh. To avoid bulk refactoring our tests we included a small hack that hashed byte and unicode types to the same value. Said another way, a unicde and byte string for "hello" would be equal.

I expected that dropping this hack would be invasive but our tests pass. Kinda confused, but... neat? --- stem/util/__init__.py | 17 ++++------------- 1 file changed, 4 insertions(+), 13 deletions(-)

diff --git a/stem/util/__init__.py b/stem/util/__init__.py index c6b9ce34..2bcc8a07 100644 --- a/stem/util/__init__.py +++ b/stem/util/__init__.py @@ -49,19 +49,10 @@ def _hash_value(val): if not HASH_TYPES: my_hash = 0 else: - # TODO: I hate doing this but until Python 2.x support is dropped we - # can't readily be strict about bytes vs unicode for attributes. This - # is because test assertions often use strings, and normalizing this - # would require wrapping most with to_unicode() calls. - # - # This hack will go away when we drop Python 2.x support. - - if isinstance(val, (bytes, str)): - my_hash = hash('str') - else: - # Hashing common builtins (ints, bools, etc) provide consistant values but many others vary their value on interpreter invokation. - - my_hash = hash(str(type(val))) + # Hashing common builtins (ints, bools, etc) provide consistant values but + # many others vary their value on interpreter invokation. + + my_hash = hash(str(type(val)))

if isinstance(val, (tuple, list)): for v in val: