Ir-regular columns

[chfw]

What's the impact of https://github.com/pyexcel/pyexcel-text/blob/master/tests/test_io.py#L85?

[jayvdb]

Well, the rst and html formats handle this differently, having a cell for the missing data, but most importantly the column should be recognised as having numerical alignment. i.e. this bug is impacting the other rows.

If it is an upstream bug, it should be reported (but given the lack of response on the PR I sent them...) and a workaround included in pyexcel-text so that our users can rely on numbers being aligned correctly, otherwise we need to document this oddity.

Also the JSON uses "Column 3": "" instead of omitting the cell or using "Column 3": null.

[jayvdb]

There are multiple problematic results here, which may have a similar fix.

When pyexcel is loading a dict, it replaces empty cells with "".

Empty cells in CVS files loaded with pyexcel-io are removed , due to pyexcel/pyexcel-io@00973b4da9 , and then pyexcel replaces the missing with "".

So the pyexcel-text column formatting problem is because the column contains numbers and empty strings, so it renders the column as strings.

Issues when loading a dict isnt a serious problem -- the caller can fix the data. However when loading a csv file, using pyexcel-io, the caller shouldnt be expected to know that they need to restore the missing cells.

So the serious problem here is that -text is not compatible with -io's csv loader.

One possible workaround is for pyexcel-text to replace "" cells with None, and see if tabulate ignores the None when determining the data type for the column.

[chfw]

regarding the workaround, looks like placing " "(one space) in https://github.com/pyexcel/pyexcel-text/blob/master/tests/test_io.py#L217 will get the position filled. '' or None does not help.

[jayvdb]

A single space may fill the column but it will not fix the alignment of a column containing numbers.

[chfw]

What would be the work around?