passing array=pyexcel.Sheet to pyexcel.save_as is an un-intended use

[chfw]

Hi all,

I just realize that passing array=pyexcel.Sheet is an un-intended use. And these code changes are against the interface between pyexcel-io and pyexcel is a dictionary of two dimensional array. I am sorry that this interface is not documented publicly. The reason for such an interface is to break the tight coupling between pyexcel and pyexcel-io.

https://github.com/pyexcel/pyexcel-text/blob/master/pyexcel_text/__init__.py#L125
https://github.com/pyexcel/pyexcel-text/blob/master/pyexcel_text/__init__.py#L185

I guess this line has played a mis-leading role in it:

https://github.com/pyexcel/pyexcel-text/blob/master/tests/test_io.py#L208

With that said, pyexcel-text is an exception among the pyexcel-io plugins because it also serves as a presentation plugin to pyexcel, in other words, pyexcel-text is a tightly coupled with pyexcel.

So, I will need to find a way to in-corporate the new use cases, either making this un-intended use as intended or implementing them within the interface set out.

[jayvdb]

Hmm. If it helps, I am happy to pin my repo dependency to the current revision where my use works correctly, so you can break it in master

[chfw]

Please hang on the current commit in master. I found a way to incorporate your use cases and will update this thread in a few days

[chfw]

Please evaluate current pyexcel-text with pyexcel v0.2.1.

I change its extension mechanism. It used to extend pyexcel-io but now it extends from pyexcel.source.FileSource, where structure context is available.

[chfw]

pyexcel v0.2.1 will try to load pyexcel-text if it is installed. hence, you won't need to import it. The same will happen to pyexcel-io plugins when pyexcel-io v0.2.0 will be released.

[jayvdb]

I've retested using the latest pyexcel-text and pyexcel, and it works for my use after the following change:

-python -c "import pyexcel, pyexcel.ext.text, pyexcel.ext.xlsx; sheet = pyexcel.sheets.NominableSheet(pyexcel.get_sheet(file_name='downloads/ERA2012JournalList.xlsx'), name_columns_by_row=0); pyexcel.save_as(array=sheet, dest_file_name='downloads/ERA2012JournalList.json')"
+python -c "import pyexcel, pyexcel.ext.xlsx; pyexcel.save_as(file_name='downloads/ERA2012JournalList.xlsx', dest_file_name='downloads/ERA2012JournalList.json', name_columns_by_row=0)"

The new json file is identical to the old one.

The old command now produces exceptions (AttributeError if I recall correctly), so the new version isnt backwards compatible, but my new command is obviously much better, and is roughly what I expected I should write when I first tried to use pyexcel. So I'm happy it now works and is readable.
Thanks.

[chfw]

Please be aware that this commit 4308dec will affect your downstream projects. Mainly it is because of pyexcel/pyexcel#32, pyexcel/pyexcel#35

[chfw]

pyexcel-text v0.2.2 is released. Will have an impact on your projects