[pandas-vet] autofixes for PD003, PD004, PD008, PD009 and PD011
#2741
Conversation
| let replacement = match *diagnostic.kind.rule() { | ||
| Rule::UseOfDotAt => Some("loc"), | ||
| Rule::UseOfDotIat => Some("iloc"), | ||
| Rule::UseOfDotValues => Some("to_numpy()"), |
There was a problem hiding this comment.
I consider this fix dangerous as there is no check if .values is actually a pandas object.
There was a problem hiding this comment.
The fix is only applied when the use-of-dot-values is detected. (If it's not, then the rule logic should be updated, not that of the fix)
There was a problem hiding this comment.
yes, and use-of-dot-values detects every .values access instead of only pandas related ones: #2229
There was a problem hiding this comment.
Yeah, thats partially limitation of not having type information. We could try to make the detection more conservative, but if the user enables it consciously, and the detection is correct, then the fix is in principle correct.. I think.
There was a problem hiding this comment.
My concern here is just that we still have known false-positives for this rule, and so we'll be knowingly breaking code in certain places by applying these autofixes. I'm tempted to say that these should be off-by-default, but enableable by users, which we haven't done before but I'd like to start doing in some cases.
There was a problem hiding this comment.
This one should benefit from the autofix safety levels (#1992). Is there anything we could do in the meantime to merge this, like make the tests more conservative? Implemented this one for autofixing https://github.com/ing-bank/popmon
sbrugman
force-pushed
the
pd-vet-autofixes
branch
from
af2853e
to
a0dc90c
Compare
sbrugman
changed the title
pandas-vet] autofixes for PD003, PD004, PD008, PD009 and PD011
|
This PR probably needs autofix aggressiveness levels (#1997) |
|
Ok, sounds good! Anyone is free to pick up this where it was left off |
Autofixes for PD003, PD004, PD008, PD009 and PD011