DataFrame.from_records should be less strict about dict like types

[tebeka]

Currently if you try DataFrame.from_records(emails, columns=['Subject', 'From', 'Date']) it'll fail. Emails are of type email.message.Message. However if you convert the email to dictionaries - the above code works.

The problem seems to stem from _to_arrays function in frame.py where the check is elif instance(data[0], dict). This fails even though email.message.Message has the full dictionary interface.

email.message.Message uses the Mapping abstract base class so we can easily
put in support for that. If you provide a self-contained test-case I'll try to get a fix in for
the upcoming 0.11.0 release.

correction: should derive from Mapping, but it looks like the module goes all the
way beck to old-style classes, so I guess that won't happen.
There's a pending PR, and you'll just need to register the Message class with the
Mapping ABC, see test case.

closed via 15847d6.