1

I currently have a dataset which has two columns that I'd like to compare. In one column, I have a string that I'd like to search for (let's call it column A). In a second column (let's call it column B) are some more strings.

The problem is that both columns have varying contents, so the pattern being searched for in the regular expression is likely to change from one row to another. Normally, when I'm searching a column for a particular string, I use something like this:

df$output <- NA
df$output[grep("TARGET_STRING", df$column_B)] <- "STRING_FOUND"

However, now that I'm trying to do this:

df$output[grep(df$column_A, df$column_B)] <- "STRING_FOUND"

Unfortunately, this gives an error:

argument 'pattern' has length > 1 and only the first element will be used

I've tried various methods to fix this, and can't seem to find a simple solution, and I'm sure there must be one. I can see why it's throwing an error (I think), but I'm not sure how to solve it. What do I need to do to get the regular expression working?

Edit: Here's the testing data.frame I've been using to explore it:

column_A <- c("A", "A", "B", "B")
column_B <- c("A", "zzz", "B", "zzz")

greptest <- data.frame(column_A, column_B)

greptest$output<-NA
greptest$output[grep(greptest$column_A, greptest$column_B)] <- "STRING_FOUND"
vize
  • 281
  • 1
  • 2
  • 10

2 Answers2

4

You can write a function that wraps grepl and then use apply:

grepFun <- function(rw){
    grepl(rw[1],rw[2],fixed=TRUE)
}

xx <- apply(greptest,1,grepFun)
greptest$output[xx] <- "STRING_FOUND"

You've already excepted my answer, but I thought I'd provide another, somewhat more efficient version using ddply:

grepFun1 <- function(x){
    ind <- grepl(x$column_A[1],x$column_B,fixed=TRUE)
    x$output <- NA
    x$output[ind] <- "STRING_FOUND"
    x
}

ddply(greptest,.(column_A),.fun=grepFun1)

This version will likely be faster if you have lots of repetition in the values for column_A.

joran
  • 169,992
  • 32
  • 429
  • 468
  • 1
    @vize - No problem. You have to act fast if you're going to get your answer in before some of the speedy R guru's around here. ;) – joran Jun 16 '11 at 22:20
  • Good point - it must be all the caffeine that drives them to be so quick! Thanks for the improved function as well, I'll give that a go too. I'm still learning, so it's good for a newbie like me to see different ways to approach various problems! – vize Jun 16 '11 at 22:36
2

I'm not sure what your expected result is, but here's my code:

    > grep(greptest[1,"column_A"], greptest$column_B)
    [1] 1 2
    > grep(greptest[2,"column_A"], greptest$column_B)
    integer(0)
    > grep(greptest[3,"column_A"], greptest$column_B)
    [1] 3 4
    > grep(greptest[4,"column_A"], greptest$column_B)
    integer(0)

    > cbind(column_A,column_B,column_A==column_B)
         column_A column_B        
    [1,] "A"      "A"      "TRUE" 
    [2,] "A"      "zzz"    "FALSE"
    [3,] "B"      "B"      "TRUE" 
    [4,] "B"      "zzz"    "FALSE"

I switched A and B in the grep code, because otherwise you only get one hit per grep. You have to loop through elements, if you'd like to search for all of them (or use a loop equivalent).

If you'd like just to compare row by row, then a simple == suffices.

JSawyer
  • 110
  • 3