SQL like join on Column id for 2D arrays

[miskolc]

I've been searching the docs and on google for a way to do and SQL JOIN between 2D numpy arrays. So far the best I've found was the join_by function in recfunctions but even this one seems to require the key to be a string:

key : {string, sequence}
        A string or a sequence of strings corresponding to the fields used
        for comparison.

The alternative is to use pandas's merge but this requieres a type conversion.between numpy array and pandas Dataframes. I would like to avoid doing that. I would like to have something like an:

numpy.merge(left_array, right_array, how='inner', on=None, left_on=None, right_on=None, left_index=False, right_index=False, sort=False, suffixes=('_x', '_y'), copy=True, indicator=False)

where left_on and right_on are either integers or lists of integers

[eric-wieser]

Why is requiring the key to be a string a problem? Attaching names to your columns via structured dtypes is essentially free.

For example:

d = np.arange(12).reshape(4, 3)  # sample data

# data with column names
df = d.view([('a', d.dtype), ('b', d.dtype), ('c', d.dtype)]).squeeze(axis=-1)

And then you can use join_by on df