Use multiple backreferences in a function to produce the replacement value in stringr functions in R

Question

How can I use multiple backreferences in a function to produce the replacement in stringr functions, for example, in stringr::str_replace()?

An example: suppose I want the replacement to be rounded to a whole number and concatenated into one string (this particular function is just an example, the important thing is that it accepts > 1 backreference)

I have tried some variations on the following without success

round_concat <- function(x, y) { paste(round(as.numeric(x),0), round(as.numeric(y, 0)))}

library(stringr)
"ABC 23.3 text 105.43 more text" %>% str_replace_all(., "(\\d+)(\\.)(\\d+)", round_concat("\\1", "\\2"))

Note: I have looked for similar functionality in functions like base::gsub (see here) but without luck

Answer 1

If you want to apply a function to the replacement backreference you could do:

prices %>% str_replace_all(., "(\\d+\\.\\d+)", function(x){round(as.numeric(x))})

Example:

prices = c("tomato: 12.23","potato: 9.53")
prices %>% str_replace_all(., "(\\d+\\.\\d+)", function(x){round(as.numeric(x))})

[1] "tomato: 12" "potato: 10"

In this case, the replacement is first converted to a number and the rounded to the nearest integer.

Or even:

str_replace_all(prices, "(\\d+\\.\\d+)", function(x){ nchar(x) })
[1] "tomato: 5" "potato: 4"

Answer 2

This is what I ended up using (although I'd still love to know if multiple backrefs can be used in stringr::str_replace()).

Importantly, the solution below allows multiple backreferences to be provided to the replacement function

library(gsubfn)
"This string 24.45,32 contains numbers 67.0.5,150 lots of them" %>% 
  gsubfn("(\\d+)\\.(\\d+),(\\d+)",  ~ { paste(as.numeric(x) * 2,  as.numeric(y) * 0.5,  as.numeric(z) + 7 ) }, . , backref = -3)

# [1] "This string 48 22.5 39 contains numbers"

There are a few of things to note here:

• x, y and z are provided to replacement, you can call them whatever you want
• x, y and z simply represent each of the regex capturing groups
• backref = -3 tells gsubfn() to expect 3 backreferences but not the match itself (see here)
• Changing the -3 to 3 would mean gsubfn() would expect you to do something with the match as well, otherwise it throws an unused argument error
• The above example uses 3 arguments, but you can use as many as you want
• You can name the arguments whatever you want; they'll be available to the function in whatever order they appear in capturing groupings (i.e. ()) in the regular expression
• don't forget the ~