Replace multiple letters with accents with gsub

Question

of course I could replace specific arguments like this:

    mydata=c("á","é","ó")
    mydata=gsub("á","a",mydata)
    mydata=gsub("é","e",mydata)
    mydata=gsub("ó","o",mydata)
    mydata

but surely there is a easier way to do this all in onle line, right? I dont find the gsub help to be very comprehensive on this.

Labiodental

If you wanted to replace different patterns with the same thing, it should be possible with lapply, but as you want to replace different patterns with different strings, I think you will still have to specified these one way or another... – Wriggler

You might be able to use chartr to do this. – Odontograph

The gsubfn function in the gsubfn package is a generalization of gsub that can do that in one call: gsubfn(".", list("á"="a", "é"="e", "ó"="o"), c("á","é","ó")) – Emend

@G.Grothendieck. Thats great and also working for all type of characters. Very valuable comment. Thank you! – Labiodental

For people searching for a more general solution to this question, here is a more helpful answer: https://mcmap.net/q/136101/-grep-using-a-character-vector-with-multiple-patterns – Feudatory

@G.Grothendieck would you also post this as an answer so that future visitors see it as such? – Baily

Answer 1

Use the character translation function

chartr("áéó", "aeo", mydata)

Ayotte

Thats cool for characters... But does this also work with special characaters e.g. underscores, points, etc... It's not within the question, still would be interesting to know something for this case too... – Labiodental

@Joschi, your question doesn't talk about it. I think you'll have to escape them because they are special characters... – Davina

Answer 2

An interesting question! I think the simplest option is to devise a special function, something like a "multi" gsub():

mgsub <- function(pattern, replacement, x, ...) {
  if (length(pattern)!=length(replacement)) {
    stop("pattern and replacement do not have the same length.")
  }
  result <- x
  for (i in 1:length(pattern)) {
    result <- gsub(pattern[i], replacement[i], result, ...)
  }
  result
}

Which gives me:

> mydata <- c("á","é","ó")
> mgsub(c("á","é","ó"), c("a","e","o"), mydata)
[1] "a" "e" "o"

Melnick

Answer 3

Maybe this can be usefull:

iconv('áéóÁÉÓçã', to="ASCII//TRANSLIT")
[1] "aeoAEOca"

Saddlebag

On the most current version of R that I'm using the call iconv('áéóÁÉÓçã', to="ASCII//TRANSLIT") returns "'a'e'o'A'E'Oc~a". Did the behavior change across R versions, or does this have to do with my default encoding? – Incarnadine

@Aaron: Don't know if is an encoding problem. I tried here at R 3.3.1 and worked as expected. – Saddlebag

Answer 4

You can use stringi package to replace these characters.

> stri_trans_general(c("á","é","ó"), "latin-ascii")

[1] "a" "e" "o"

Thompkins

Answer 5

This is very similar to @kith, but in function form, and with the most common diacritcs cases:

removeDiscritics <- function(string) {
  chartr(
     "ŠŽšžŸÀÁÂÃÄÅÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖÙÚÛÜÝàáâãäåçèéêëìíîïðñòóôõöùúûüýÿ"
    ,"SZszYAAAAAACEEEEIIIIDNOOOOOUUUUYaaaaaaceeeeiiiidnooooouuuuyy"
    , string
  )
}


removeDiscritics("test áéíóú")

"test aeiou"

Slue

Answer 6

Another mgsub implementation using Reduce

mystring = 'This is good'
myrepl = list(c('o', 'a'), c('i', 'n'))

mgsub2 <- function(myrepl, mystring){
  gsub2 <- function(l, x){
   do.call('gsub', list(x = x, pattern = l[1], replacement = l[2]))
  }
  Reduce(gsub2, myrepl, init = mystring, right = T) 
}

Verbena

Answer 7

A problem with some of the implementations above (e.g., Theodore Lytras's) is that if the patterns are multiple characters, they may conflict in the case that one pattern is a substring of another. A way to solve this is to create a copy of the object and perform the pattern replacement in that copy. This is implemented in my package bayesbio, available on CRAN.

mgsub <- function(pattern, replacement, x, ...) {
  n = length(pattern)
  if (n != length(replacement)) {
    stop("pattern and replacement do not have the same length.")
  }
  result = x
  for (i in 1:n) {
    result[grep(pattern[i], x, ...)] = replacement[i]
  }
  return(result)
}

Here is a test case:

  asdf = c(4, 0, 1, 1, 3, 0, 2, 0, 1, 1)

  res = mgsub(c("0", "1", "2"), c("10", "11", "12"), asdf)

Kallick

Answer 8

Not so elegant, but it works and does what you want

> diag(sapply(1:length(mydata), function(i, x, y) {
+   gsub(x[i],y[i], x=x)
+ }, x=mydata, y=c('a', 'b', 'c')))
[1] "a" "b" "c"

Yuan

Answer 9

Related to Justin's answer:

> m <- c("á"="a", "é"="e", "ó"="o")
> m[mydata]
  á   é   ó 
"a" "e" "o" 

And you can get rid of the names with names(*) <- NULL if you want.

Georgetta

Answer 10

You can use the match function. Here match(x, y) returns the index of y where the element of x is matched. Then you can use the returned indices, to subset another vector (say z) that contains the replacements for the values of x, appropriately matched with y. In your case:

mydata <- c("á","é","ó")
desired <- c('a', 'e', 'o')

desired[match(mydata, mydata)]

In a simpler example, consider the situation below, where I was trying to substitute a for 'alpha', 'b' for 'beta' and so forth.

x <- c('a', 'a', 'b', 'c', 'b', 'c', 'e', 'e', 'd')

y <- c('a', 'b', 'c', 'd', 'e')
z <- c('alpha', 'beta', 'gamma', 'delta', 'epsilon')

z[match(x, y)]

Dink

Answer 11

You can also combine them with gsub:

mydata <- gsub("á","a", gsub("é","e", gsub("í","i", gsub("ó","o", gsub ("ú", "u", mydata)))))

Wehner