mime magic for extensionless files

[apolkosnik-old]

This method actually uses the mime type to generate a list of possible extensions for the given mimetype, and tries them all. Obviously if one works, we get the results. I did run make all in tests, and it ran all the tests fine, I guess I'm missing a few test cases in travis-ci

[apolkosnik-old]

Looks like I need to put some more thought into this, I like your table approach, so perhaps using a table might be a better approach here

[deanmalmgren]

Sorry I didn't get back to you sooner on this. What do you mean by "my table approach"?

[apolkosnik-old]

IN the current incarnation of this, I'm getting back a list of typical extensions for the filetype, and I'm going through all of these, instead of doing that, perhaps it would be much better to just just fill in a lookup table and use it for the supported formats.

[deanmalmgren]

I just took a closer look at your branch and I see what you're talking about. I agree that it would be better to have a lookup table of supported formats that maps mimetype to file extension.

I'm going to add a couple of comments to your PR with a few things that I noticed but, assuming we can get this to play nicely with the rest of the package (meaning: the tests either pass or fail for known reasons, in which case we should fix the tests), this seems like a great thing to incorporate.

[cahya-wirawan]

Hi,
any progress with this functionality? I would like to convert a document directly from memory to text, otherwise I have first to save the document in memory to a file, and then convert it.
Thanks.