R help - Nov 2008 - Efficient passing through big data.frame and modifying select fields

Hi all,

I have relatively big data frames (> 10000 rows by 80 columns) that need to
be exposed to "merge". Works marvelously well in general, but some
fields of the data frames actually contain multiple ";"-separated
values encoded as a character string without defined order, which makes the
fields not match each other.

Example:
> frame1[1,1]
[1] "some;thing"
>frame2[2,1]
[2] "thing;some"

In order to enable merging/duplicate identification of columns containing these
strings, I wrote the following function, which passes through the rows one by
one, identifies ";"-containing cells, splits and resorts them.

ResortCombinedFields <- function(dframe){
  if(!is.data.frame(dframe)){
    stop("\"ResortCombinedFields\" input needs to be a data
frame.")
  }
  for(row in seq(nrow(dframe))){
    for(mef in grep(";",dframe[row,])){
      dframe[row,mef] <-
paste(sort(unlist(strsplit(dframe[row,mef],";"))),collapse=";")
    }
  }
  return(dframe)
}

works fine, but is horribly inefficient. How might this be tackled more
elegantly?

Thanks for any input, Joh