We have plenty of data buried in a lot of Excel files (.xls, .xlsx). The data comprises mainly statistics (first row comtains colum headers, first column contains row headers, rest of data are decimal numbers).

We want to provide this data in various file formats that better follow the open data best practices than Excel files. Excel has just 2 stars, we want more (see http://5stardata.info).

Since we live in an Linux world, we are looking for batch data conversion tools (libraries, scripts, tools, etc.) - PHP prefered, but this is no must - that convert these Excel files to

  • CSV (decimal point, decimal comma)
  • TSV (tab separated values)
  • JSON
  • XML (schema file and data file)
  • RDF
  • ODF
  • Maybe there are other open data formats, too.

...or at least some of them.

The Excel files are updated periodically (yearly, monthly,...), so the conversion process should be repeatable.

Some Excels have headers in multiple rows. How can we deal with that?

12 Answers 12


I don't believe that you will find a "ready" library that will be the case for all your excel files. My advice is to create a script by yourself in base of your file structures. Especially if the updates will still have the same structure.

I cannot help you in PHP, but in Python you can find an answer for what you need here: A Python guide for open data file formats. If you decide to do it, you can update your question with an example of your files and let users help you with the code.

  • Thanks. I edited the question and I am already happy with some of the above mentioned data formats.
    – WeSee
    Commented May 25, 2014 at 16:56

For example with Python, using Pandas:

import pandas as pd

data = pd.from_excel("path_to.xls", sheetname="sheet1")
data.to_csv("path_to.csv", encoding="utf8")

(As simple as that.)

Pandas can read and write from/to more formats.

However, for typical tabular data CSV is the best option for open data.


You may want to look at this previous stackoverflow question on converting Excel to CSV in batch mode on Linux. Once it is in CSV, you have lots of options of converting to other formats - but CSV is the most common format for open datasets


  • 2
    Yes, I would especially recommend Libreoffice (in headless mode), which I have used myself for such tasks. It is a more portable solution than Gnumeric. Just write a PHP function that calls libreoffice --headless --convert-to csv $filename --outdir $outdir on command line.
    – Nicolas Raoul
    Commented May 26, 2014 at 3:48

unoconv can convert CSV to OpenDocument spreadsheet:

exec('unoconv -f ods -o output.ods input.xlsx');

unoconv is based on LibreOffice, and has features for batch processing.


That sounds like you'll do fine with a CSV format. (Assuming the datasets are not too large.)

The problem, as pointed out before, is that you may have to design a process for each format/version etc of Excel sheet. They are optimised for human-readability, not machines.

You can use your tool of your choice to convert them, but there will be some manual work to codify the conversion.

If you're thinking of publishing data, also consider aspects that go beyond technical aspects; more e.g. in our Open Data Certificates.

  • Cool, I didn't know about Open Data Certificates - nice idea. Coming to CSV: We want to provide multiple data formats for open data consumers. They should not have to convert the data but simply use it as we provide it. So we strive to provide multiple formats in parallel.
    – WeSee
    Commented May 27, 2014 at 20:01

Talend Data Integration can help you go convert Excel files, including those with more than 1 sheet, to any format or your choice for free. Have à look at www.talend.com download the tool and enjoy.

  • I thought about talend too. This is an heavy framework that will need some learning curve but will save you hours if you have dozen of files ...
    – magdmartin
    Commented Sep 12, 2014 at 15:21

I would look at csvkit


in particular the in2csv tool documented here



Have you tried the xlrd package in python? you can use it to import your file. exporting requires the library specific to your output filetype.

Example links.


On the understanding that I do work for this company, I'll mention a product called FME that is capable of converting Excel into any of several hundred formats. It's a commercial (i.e. non-open) product, but does support many open formats of data.

There is also a special website section on FME and Open Data Initiatives

Since I work for them I won't go into detail. But it is a definite solution to this problem, so I think (hope) it's acceptable to mention it here.


simpleexcelphplibrary is the answer you seek:
JSON!!!!!!!!!, HTML, CSV, and XML

seriously just github search this if you want/need more. there are so many conversion scripts its ridiculous. you could do the same with npm, but that's not php...i know codeplex (ms) has a php/excel library. also google code, bit bucket, and sourceforge are going to have a similar findings, although perhaps not as up to date.

saving them in google drive allows exportation in ods, and csv, as well as offers a pretty big space of free storage for your data.
Note: this option is only optimal when doing one or two conversions at a time, but keeping that in mind, its a free, easy, cloud hosted/converter.


If you are trying to convert Excel to JSON there is a free tool that will do it



I'd use Perl, I know there is a CPAN lib for that ant your webserver may also be able to use Perl out of the box.

Search for it and you're done. Easier is Spreadsheet::Read + JSON::XS.

It will end up in a code looking like

use Spreadsheet::Read;
use JSON::XS;

open dest,">>$1.json";
$json = JSON::XS->new->ascii;

print $dest,$json->encode(ReadData($1));

If you need it prettyified, just add ->pretty after the ->new->ascii

  • "Search for it and you're done": This is Stack Exchange, answers have to be fully researched.
    – Nicolas Raoul
    Commented Nov 18, 2015 at 7:26
  • Sorry hard to type on mobile+ switching app clears the edits not saved yet. Lost my edits 3 times now I'm sad Commented Nov 18, 2015 at 8:15
  • Sorry to hear that! In such cases I guess a comment would be the best until you have a way to edit comfortably :-) The rules of Stack Exchange are unusually severe against short answers that don't include all of the information. Thanks for participating and looking forward to your full answer! I remove my downvote for now.
    – Nicolas Raoul
    Commented Nov 18, 2015 at 8:48

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.