BUG: merge_asof raising KeyError for extension dtypes

[lukemanley]

• closes ENH: Support pyarrow timestamps in merge_asof #52904
• Tests added and passed if fixing a bug or adding a new feature
• All code checks passed.
• Added type annotations to new arguments/methods/functions.
• Added an entry in the latest doc/source/whatsnew/v2.1.0.rst file if fixing a bug or adding a new feature.

[mroeschke]

IIRC we do something like the below in another area of this file correct? If so can we reuse?

[lukemanley]

yes, updated a few locations to reuse these methods.

[mroeschke]

Thanks @lukemanley

[0x26res]

@lukemanley thanks for fixing. I was wondering, do you expect it to work for timezone aware pyarrow timestamps eg pa.timestamp("ns", "UTC")

[lukemanley]

@0x26res - yes, tz-aware pyarrow timestamps should work in pandas 2.1 (to be released later this month):

In [1]: import pandas as pd

In [2]: import pyarrow as pa

In [3]: dtype = pd.ArrowDtype(pa.timestamp('ns', 'UTC'))

In [4]: df1 = pd.DataFrame({"a": [1, 3]}, index=pd.Index([1, 3], dtype=dtype))

In [5]: df2 = pd.DataFrame({"b": [0, 2]}, index=pd.Index([0, 2], dtype=dtype))

In [6]: pd.merge_asof(df1, df2, left_index=True, right_index=True)
Out[6]: 
                                     a  b
1970-01-01 00:00:00.000000001+00:00  1  0
1970-01-01 00:00:00.000000003+00:00  3  2