ENH: Support for PyExcelerate as an Excel writer engine.

[jmcnamara]

This is an initial patch to add PyExcelerate as an Excel writer engine.

It isn't suitable for merge yet since there is issue with date handling in PyExcelerate that I am working to resolve with the authors. I just want this to appear on the pandas timeline for now so I can refer to it in other PRs.

1. The requirements files are missing since the PR will require a PyExcelerate update and release. Also, the print_versions.py entry.
2. There is also a failing test due to versioning.
3. There is currently no merged cell support.

I'll fix all these before the final PR request.

John

[jtratner]

@jmcnamara awesome - really interested in this!

[cpcloud]

put this in one of the requirements files in pandas/ci

[jmcnamara]

This is just a a temporary measure while the PyExcelerate devs merge my PR and look at the date issue. Once the patched PyExcelerate is on PyPi I'll update this and the requirement files.

[jreback]

can you give a use case where PyExcelerate would be used rather than XlswWriter (undertsanding that you are the author of XlsxWriter!)

[jmcnamara]

@jreback Do you mean to put a use case in the docs?

The general use case for PyExcelerate would be that it is faster. In the optimised case it is 50% faster than XlsxWriter. (I don't know what the associated memory usage is, I'll have to benchmark it.)

The use case for XlsxWriter is that it is fast enough (as fast as xlwt) and that it supports a lot of features such as charts, conditional formatting, autofilters and more.

I can add a line or two about that to install.rst or release.rst if you think that would be useful.

[jreback]

@jmcnamara I think a brief explanation for all of the excel engines would be in order (in install.rst (or io.rst) with a link on the other side would be good). A user would be like, hmm, let me specify an engine if ..... (aside from the fact that the default engine order will do a good job)

[jmcnamara]

@jreback Yes. Good point. I'll add that.

[ghost]

This is advocating that we incorporate a fifth optional-dependency for excel files support?!
That's sheer insanity. we should strive for less deps to do a given job, not more.

Given that it's also been gathering dust for 3 months, I'm just closing this.

[jmcnamara]

This is advocating that we incorporate a fifth optional-dependency for excel files support?!
That's sheer insanity. we should strive for less deps to do a given job, not more.

I'm fine with closing it. Pyexcelerate isn't really stable enough or doesn't have enough formatting support for proper integration with Pandas.

That is also why the PR has stagnated for several months. I was waiting for the Pyexcelerate feature set to catch up.

All in all, it probably isn't worth adding at this stage.