how to calculate mean value of each column within range of quantile?

I’d like to calculate mean value of each column but only consider values are within range of quantile, like 20%-80% only.

Here is why I have done, but not complete

df=pd.DataFrame({"A":[1,1,20,2,2,3,50,7,8,15,20,35,50,70],"B":[10,100,20,20,200,30,50,70,80,150,200,350,500,700]})
df
    A   B
0   1   10
1   1   100
2   20  20
3   2   20
4   2   200
5   3   30
6   50  50
7   7   70
8   8   80
9   15  150
10  20  200
11  35  350
12  50  500
13  70  70

then find q20 and 180 for each column using np.quantile()

q20=np.quantile(df,0.2,axis=0)
q20

array([ 2., 26.])

q80=np.quantile(df,0.8,axis=0)
q80

array([ 41., 260.])

Now How do I filter the values between q20 and q80 for each column

I am doing below then I got an error

mask=(a>q20)&(a<q80)

TypeError: Cannot compare a Categorical for op __gt__ with type <class 'numpy.ndarray'>.
If you want to compare values, use 'np.asarray(cat) <op> other'.

Thanks for your help

>Solution :

use rank with pct

out = df[df.rank(pct=True).ge(0.2) & df.rank(pct=True).le(0.8)].mean()

out:

A     12.444444
B    110.000000
dtype: float64