I wrote a couple days about about importing Excel files into R. There are lots of ways to do this, but all the ways that use only R have drawbacks (as I outlined in my last post), and all the other ways require installation of programs other than R. I’m not opposed to using programs other than R – it’s easy enough to weave, for example, Python and R code into each other. But I’d become curious about the possibility of solving this problem without the need for added programs, so I did some more searching. Turns out you can import an .xlsx document into pretty much anything that can parse XML, because that’s all an .xlsx document is.
The following gist uses R’s XML package, as well as a little help from the `plyr` package. It will import all sheets from an .xlsx document into a list of data frames and doesn’t require any additional installations of converters, and as far as I can tell, it should work on any platform. They key is to rename the document to have a .zip extension. Then you can unzip the document, pull out the XML files, and match them up to recreate a view similar to what you would see in Excel. The issue of how to deal with XML files in R is an issue that comes up regularly on Stack Overflow, so I thought this gist could also be a nice little example of how to handle that sort of problem (long story short: convert the XML to a list and manipulate the list – it’s generally much easier than trying to manipulate the XML directly).
EDIT 1: You may notice that this converts all times and date-times into dates. That’s because the particular data I was working with didn’t have times or date-times, and because I’m still trying to figure out exactly how Excel handles date/time origins. I’ll edit the function when I find out more.
EDIT 2: Ok, times and date-times are now appropriate, but only if you’re opening an xlsx document that was created on Windows. The Mac version of Excel apparently uses a different origin for dates, which is stupid. I’m trying to figure out how to tell from the XML which platform an xlsx document was created on, but so far I haven’t had any luck.
EDIT 3: I’ve now updated the function. Interestingly, it seems Microsoft might have standardized its handling of dates across operating systems. I created an .xlsx file on my Mac and only the Windows origin (“1899-12-30″) resulted in the correct dates. I’ve left a note in the code to indicate the origin that is supposed to be the right origin for Mac Excel (“1904-01-01″). I’ve also included some options in the function to select only specific sheets from a workbook, and to indicate whether the function should expect the sheets to have headers.