Increase pwcrypto unit test coverage #343
Conversation
jonasbardino
added
the
test-only
|
The unit tests failing here |
jonasbardino
force-pushed
the
adjust/pwcrypto-unit-test-to-expand-coverage-and-clean-pickle-heritage
branch
from
7a5fb13 to
d4dbb39
Compare
jonasbardino
added
the
follow-up pending
|
Marked follow-up to cover remaining polish of a few minor corner cases at outlined in TODO comments in the code. Still, better to get wide coverage in to support the related #347 and polish last bits later. |
…unit test leftovers.
simple hash, hash and digest. Mixes in various policy configurations.
* make_safe_hash (sha256) * valid_login_password * make_scramble+check_scramble * make_encrypt+make_decrypt for fernet, aesgcm and aesgcm_static * make_encrypt output variation for fernet, aesgcm and aesgcm_static * make_encrypt+check_encrypt for fernet, aesgcm and aesgcm_static * generate/parse/verify_reset_token
…. The conversion is already explicitly handled in all `make_encrypt` backends where it's needed and the bytes break the initial `assure_password_strength` call. Including the fix here as only the new unit test triggers it.
* clarify that make_hash generates value depending on random seed * assure check_hash works despite this variation in hashes depending on seed * make sure otherwise valid password reset tokens fail when expired
jonasbardino
force-pushed
the
adjust/pwcrypto-unit-test-to-expand-coverage-and-clean-pickle-heritage
branch
from
78c1429 to
22b25d9
Compare
Better to get improved coverage as you say. Other suggestions can wait :) |
jonasbardino
deleted the
adjust/pwcrypto-unit-test-to-expand-coverage-and-clean-pickle-heritage
branch
Rereading your original reply I noticed that I missed one important point 👍 because I got caught up in the first bit about the basic test of make_hash itself being removed, which was not in fact the case 8-D I do generally agree that it makes sense to test dependent functions on static output of their dependecy functions (like from make_hash in this example) to differentiate and pinpoint where if/when errors creep in. In many cases it just also has value to test the combination IMO, so I'd rather that we aim for both. |
2 participants
Add more test coverage and clean up some copy/pasted pickle unit test leftovers.
NB: changes one single line of (currently unused) actual code as explained in
d4dbb39
despite the test-only tag.