tags:

views:

254

answers:

1

I'm trying to learn R's XML package. I'm trying to create a data.frame from books.xml sample xml data file. Here's what I get:

library(XML)
books <- "http://www.w3schools.com/XQuery/books.xml"
doc <- xmlTreeParse(books, useInternalNodes = TRUE)
doc
xpathApply(doc, "//book", function(x) do.call(paste, as.list(xmlValue(x))))
xpathSApply(doc, "//book", function(x) strsplit(xmlValue(x), " "))
xpathSApply(doc, "//book/child::*", xmlValue)

Each of these xpathSApply's don't get me even close to my intention. How should one proceed toward a well formed data.frame?

+6  A: 

Ordinarily, I would suggest trying the xmlToDataFrame() function, but I believe that this will actually be fairly tricky because it isn't well structured to begin with.

I would recommend working with this function:

xmlToList(books)

One problem is that there are multiple authors per book, so you will need to decide how to handle that when you're structuring your data frame.

Once you have decided what to do with the multiple authors issue, then it's fairly straight forward to turn your book list into a data frame with the ldply() function in plyr (or just use lapply and convert the return value into a data.frame by using do.call("rbind"...).

Here's a complete example (excluding author):

library(XML)
books <- "http://www.w3schools.com/XQuery/books.xml"
library(plyr)
ldply(xmlToList(books), function(x) { data.frame(x[!names(x)=="author"]) } )

   .id        title.text title..attrs year price   .attrs
 1 book  Everyday Italian           en 2005 30.00  COOKING
 2 book      Harry Potter           en 2005 29.99 CHILDREN
 3 book XQuery Kick Start           en 2003 49.99      WEB
 4 book      Learning XML           en 2003 39.95      WEB

Here's what it looks like with author included. You need to use ldply in this instance since the list is "jagged"...lapply can't handle that properly. [Otherwise you can use lapply with rbind.fill (also courtesy of Hadley), but why bother when plyr automatically does it for you?]:

ldply(xmlToList(books), data.frame)

   .id        title.text title..attrs              author year price   .attrs
1 book  Everyday Italian           en Giada De Laurentiis 2005 30.00  COOKING
2 book      Harry Potter           en        J K. Rowling 2005 29.99 CHILDREN
3 book XQuery Kick Start           en      James McGovern 2003 49.99      WEB
4 book      Learning XML           en         Erik T. Ray 2003 39.95      WEB
     author.1   author.2   author.3               author.4
1        <NA>       <NA>       <NA>                   <NA>
2        <NA>       <NA>       <NA>                   <NA>
3 Per Bothner Kurt Cagle James Linn Vaidyanathan Nagarajan
4        <NA>       <NA>       <NA>                   <NA>
Shane
"but why bother when plyr automatically does it for you?" I find myself saying that an awful lot.
JD Long
Thanks, that's great!
larus