Let checks.has_dtypes accept functions as values of the items dict

[topper-123]

Today is is a bit difficult to check on pandas dtypes directly. For example see below:

import numpy as np
import pandas as pd
import engarde.checks as ck

# this works because it has dtype int32
df1 = pd.DataFrame({'A':[1,2,3]}, dtype=np.int32)
df1.pipe(ck.has_dtypes, items={'A':int})

# this fails because it has dtype int64
df2 = pd.DataFrame({'A':[1,2,3]})
df2.pipe(ck.has_dtypes, items={'A':int})

To make dtype-checking more robust, I propose that the parameter items in has_dtypes can accept functions as values. This will allow us to use the functions in pandas.api.types to check dtypes which will ease dtype. checking

For example:

df2 = pd.DataFrame({'A':[1,2,3]})
df2.pipe(ck.has_dtypes, items={'A':pd.api.types.is_integer_dtype})

I have already written a proposal on this (but it needs tests). I will just push it straight away so for your review.

[abhijeetmote]

I was trying the above examples and it was not working this is the way it is working.

import numpy as np
import pandas as pd
import engarde.checks as ck

# this works because it has dtype int32
df1 = pd.DataFrame({'A':[1,2,3]}, dtype=np.int32)
df1.pipe(ck.has_dtypes, items={'A':np.int32})

# this fails because it has dtype int64
df2 = pd.DataFrame({'A':[1,2,3]})
df2.pipe(ck.has_dtypes, items={'A':np.int64})

print(df1.dtypes)
print(df2.dtypes)

[TomAugspurger]

Just FYI, engarde isn't maintained. You might want to look a bulwark.

…

On Fri, Oct 4, 2019 at 4:34 AM abhijeet mote ***@***.***> wrote: I was trying the above examples and it was not working this is the way it is working. import numpy as np import pandas as pd import engarde.checks as ck # this works because it has dtype int32 df1 = pd.DataFrame({'A':[1,2,3]}, dtype=np.int32) df1.pipe(ck.has_dtypes, items={'A':np.int32}) # this fails because it has dtype int64 df2 = pd.DataFrame({'A':[1,2,3]}) df2.pipe(ck.has_dtypes, items={'A':np.int64}) print(df1.dtypes) print(df2.dtypes) — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#36?email_source=notifications&email_token=AAKAOIT5MJR4BZDH7VWM663QM4E33A5CNFSM4DNBZA52YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEALDJFA#issuecomment-538326164>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAKAOIX6EG4PUMTSCXGEM3LQM4E33ANCNFSM4DNBZA5Q> .