The repo contains three libraries for reading data from Microsoft's document formats ("Office Open XML").
open_packagingparses Office Open XML's "Open Packaging Conventions" (the container format for all Microsoft Office documents)
spreadsheetmlparses the XML data in SpreadsheetML (i.e. Excel's XLSX format)
easy_xlsxreads XLSX documents, applies the formatting in the document, and returns the result as a list of sheets (with a sheet name and then a
string list listof data). The goal of this library is to give the same results as if you exported the XLSX file to CSV and then read it with a standard CSV reader.
spreadsheetml are relatively safe to use but incomplete
(it should be obvious what they're mising -- if a field doesn't exist, I
haven't got to it yet). Everything that does exist should be parsed properly.
easy_xlsx is in very early stages. It should properly give read XLSX files
and output correct types, but the SpreadsheetML spec doesn't list all of the
built-in format strings, so some types may not be handled correctly. At the
easy_xlsx will bail out in any case where it can't understand the
formatting, although I'd be open to patches to make this optional.
opam pin add -n easy_xlsx . opam depext easy_xlsx opam install --deps-only easy_xlsx
You can run the tests if you want:
Makefile is just a thin wrapper around jbuilder, so you can use
jbuilder commands too if you prefer.
If you want to help with this, create an issue with what you'd like to work on, and mention it if you need me to help with anything (point you at the relevant spec, give my opinion on approaches, etc.).
Some things I could use help with:
- Implement more of the specs. See ECMA 376. The editions only contain changes, so most of what's interesting is in the 1st edition. Part 2 is the most interesting for the Open Packaging Conventions and Part 4 has specifics for SpreadsheetML (or the other office formats if you want to start a library for them).
- Add more tests. Right now there's a set of extremely high level tests that we get the sound output as OpenOffice's CSV export, but being so high level means a lot of things are completely untested until we're 100% finished, which isn't a good situation to be in. It's easy to have a typo when implementing this spec, so I'd like to aim for 100% test coverage.
- The cell formatting needs a lot of work. It's what I'm likely to work on next, but if you think you can do it first I'd be happy to let someone else do it (let me know if you start working on this though, so we don't duplicate effort).
- Make CamlZip support a string or bigstring interface. Right now we can only read actual files on the filesystem since that's the interface CamlZip gives us. I'd like to be able to read data from memory.
- Make a pure-OCaml ZIP library. Then we won't need any system dependencies
and should work with