从 HDF5 获取表索引的最有效方法

Question

I have an HDF5 file containing pandas Series/DataFrame tables. I need to get (pandas) index of a table stored under a key in HDF, but not necessarily the whole table:

I can think of two (effectively the same) methods of getting the index:

import pandas as pd

hdfPath = 'c:/example.h5'
hdfKey = 'dfkey'
# way 1:
with pd.HDFStore(hdfPath) as hdf:
    index = hdf[hdfKey].index

# way 2:
index = pd.read_hdf(hdfPath, hdfKey)

However for a pandas Series of ~2000 rows this takes 0.6 sec:

%timeit pd.read_hdf(hdfPath, hdfKey).index
1 loops, best of 3: 605 ms per loop

Is there a way to get only index of a table in HDF?

Answer 1

The HDFStore object has a select_column method that will allow you to get the index. Note that it will return a Series with the index as the values.

with pd.HDFStore(hdfPath) as hdf:
    index = hdf.select_column(hdfKey, 'index').values