passing multiple "on" conditions in merge_asof

Is there any way to pass multiple on conditions to merge_asof with either similar or different tolerances?

Let's consider df1 as:

by1	by2	on1	on2
a	1	1.1	1.2
b	2	1.3	1.4
a	3	1.5	1.6

and df2 as:

by1	by2	on1	on2
a	1	2.1	2.2
x	9	1.3	1.4
a	3	2.5	6.6

by a tolerance of 1 for on1 and 2 for on2, I want such merge:

by1	by2	on1	on2
a	1	2.1	2.2

I have tried other methods besides merge_asof, such as nested loops, which are definitely not time efficient. currently I'm using this code which only works for only one of my "on: conditions, in this case "on1":

df1= df1.sort_values(["on1"])
df2= df2.sort_values(["on1"])

df3 = pd.merge_asof(df1, CaDdDfChanges, on=["on1"], by=["by1", "by2"], tolerance=ToleranceFor1, direction="nearest")

from Recent Questions - Stack Overflow https://ift.tt/3DXJaZv
https://ift.tt/eA8V8J