Infinity np.inf behavior is not consistent with not a number np.nan behavior.

[lendres]

It seems that np.inf behaves differently than np.nan when using Pint Quanties in a Series/DataFrame. Finding and replacing seem to work transparently for np.nan, but np.inf can only be found when searching on the magnitudes of Pint Quantities. It would be convent if np.inf search/replaced worked like np.nan.

import numpy                  as     np
import pandas                 as     pd
import pint_pandas

series = pd.Series([np.nan, 6, np.inf])
pintSeries = pd.Series(series, dtype="pint[meter]")

print("\nOriginal Series:\n{0}".format(series))
print("Not a number check:", series.isna().sum())
print("Infinity check:", series.isin([np.inf]).sum())
series = series.replace(np.nan, 0)
series = series.replace(np.inf, 0)
print("\nSeries After Replacement:\n{0}".format(series))


print("\nOriginal Pint Series:\n{0}".format(pintSeries))
print("Not a number check succeeds:", pintSeries.isna().sum())
print("Standard infinity check fails:", pintSeries.isin([np.inf]).sum())
print("Infinity check on magnitudes succeeds:", pintSeries.pint.magnitude.isin([np.inf]).sum())
pintSeries = pintSeries.replace(np.nan, 0)
pintSeries = pintSeries.replace(np.inf, 0)
print("\nPint Series After Replacement:\n{0}".format(pintSeries))

Output:

Original Series:
0    NaN
1    6.0
2    inf
dtype: float64
Not a number check: 1
Infinity check: 1

Series After Replacement:
0    0.0
1    6.0
2    0.0
dtype: float64

Original Pint Series:
0    nan
1    6.0
2    inf
dtype: pint[meter]
Not a number check succeeds: 1
Standard infinity check fails: 0
Infinity check on magnitudes succeeds: 1

Pint Series After Replacement:
0    0.0
1    6.0
2    inf
dtype: pint[meter]