Counting number of vertical bars | in pandas string has strange behaviour?-CodePudding

I want to count the number of instances of vertical bars "|" in a each row of a particular column in a pandas dataframe. But using str.count("|") yields some strange behaviour:

import pandas as pd
df = pd.DataFrame(['some text', None, 'a|few|vertical|bars', 'one|'])
df[0].str.count("|")

outputs

0    10.0
1     NaN
2    20.0
3     5.0
Name: 0, dtype: float64

What's going on here? If I use apply instead, I get the expected answer:

df.apply(lambda x: str(x[0]).count("|"),axis=1)

yields

0    0
1    0
2    3
3    1
dtype: int64

CodePudding user response：

Try this, pat is a regex string and | is a regex operator, OR, so escape with '\', blackslash:

df[0].str.count('\|')

Output:

0    0.0
1    NaN
2    3.0
3    1.0
Name: 0, dtype: float64

Note: str.count in the standard library is different from pd.Series.str.count where the former doesn't use regex, but the method from pandas does per docs linked above.

CodePudding user response：

Looks like str(x[0]) converts the None into the actual string 'None' of type str - losing its NoneType. So when count() encounters a string is simply does a count - when it encounters a NoneType it returns NaN.

df[0].apply(lambda x: type(x))

0         <class 'str'>
1    <class 'NoneType'>
2         <class 'str'>
3         <class 'str'>
Name: 0, dtype: object

df.apply(lambda x: type(str(x[0])), axis=1)

0    <class 'str'>
1    <class 'str'>
2    <class 'str'>
3    <class 'str'>
dtype: object