Unexpected join_left

[flexatone]

Thanks, @michaeljpeters , for isolating this.

>>> co = sf.Frame.from_dict_records(
...     [
...         {"gvkey": "123456", "company_name": "A0"},
...         {"gvkey": "567890", "company_name": "A1"},
...         {"gvkey": "234567", "company_name": "B"},
...         {"gvkey": "345678", "company_name": "C"},
...         {"gvkey": "456789", "company_name": "D"},
...     ]
... )
>>> gvk_to_cik = sf.Frame.from_dict_records(
...     [
...         {"gvkey-b": "123456", "cik": 1111},
...         {"gvkey-b": "567890", "cik": 1111},
...         {"gvkey-b": "345678", "cik": 3333},
...         {"gvkey-b": "456789", "cik": 4444},
...     ]
... )
>>> co_w_cik = co.join_left(
...     gvk_to_cik,
...     left_columns="gvkey",
...     right_columns="gvkey-b",
...     fill_value="-1",  # intentional string
... )

>>> co_w_cik
<Frame>
<Index> gvkey  company_name gvkey-b cik     <<U12>
<Index>
0       123456 A0           123456  1111
1       567890 A1           567890  1111
2       234567 B            345678  3333 # not expected
3       345678 C            345678  3333
4       456789 D            456789  4444
<int64> <<U6>  <<U2>        <<U6>   <int64>

[flexatone]

Thanks again for this issue. This has been corrected and bundled with numerous performance enhancements.

>>> co = sf.Frame.from_dict_records(
... ...     [
... ...         {"gvkey": "123456", "company_name": "A0"},
... ...         {"gvkey": "567890", "company_name": "A1"},
... ...         {"gvkey": "234567", "company_name": "B"},
... ...         {"gvkey": "345678", "company_name": "C"},
... ...         {"gvkey": "456789", "company_name": "D"},
... ...     ]
... ... )
>>> gvk_to_cik = sf.Frame.from_dict_records(
... ...     [
... ...         {"gvkey-b": "123456", "cik": 1111},
... ...         {"gvkey-b": "567890", "cik": 1111},
... ...         {"gvkey-b": "345678", "cik": 3333},
... ...         {"gvkey-b": "456789", "cik": 4444},
... ...     ]
... ... )
>>> co.join_left(
... ...     gvk_to_cik,
... ...     left_columns="gvkey",
... ...     right_columns="gvkey-b",
... ...     fill_value="-1",  # intentional string
... ... )
<Frame>
<Index> gvkey  company_name gvkey-b cik      <<U12>
<Index>
0       123456 A0           123456  1111
1       567890 A1           567890  1111
2       234567 B            -1      -1
3       345678 C            345678  3333
4       456789 D            456789  4444
<int64> <<U6>  <<U2>        <<U6>   <object>