Rearrange a string using regex

I have an expression as a string, like the following:

orig <- "mean(Sepal.Length, na.rm = TRUE)"
orig
#> [1] "mean(Sepal.Length, na.rm = TRUE)"

I would like to rearrange this string so that I get the following output:

"Sepal.Length$mean(na.rm = TRUE)"
#> [1] "Sepal.Length$mean(na.rm = TRUE)"

I know that I can use capture groups like this:

gsub("(Sepal.Length)", "\\\$", orig)
#> [1] "mean(Sepal.Length$, na.rm = TRUE)"

but this doesn’t work to move text in the string:

gsub("(Sepal.Length)(.*)", "\\\$\", orig)
#> [1] "mean(Sepal.Length$, na.rm = TRUE)"

This question was helpful but the solution there is hardcoded, whereas here I don't know the expression I will have at all, just that it will contain Sepal.Length. The expression above could be "sum(Sepal.Length)" for example.

I’m looking for a solution in base R only.

You can use the following pattern:

gsub("(.+)\\(Sepal\\.Length,? *(.*)\\)", "Sepal.Length$\(\)", orig)

• (.+) matches anything before the first parenthesis \\(;
• Then, we always have "Sepal.Length". Note that . is a special character, so to use a literal dot you need \\.;
• We then might have a comma, and space(s) ,? * (? means "0 or 1 times", and * means "0 or more times");
• After that we might have the other arguments (.*), followed by the closing parenthesis \\).

Edit: Thanks for @rps1227 for the suggested improvements.

Parse the expression giving p in which case p[[2]] is Sepal.Length in the first example but could be something else as in the second exxample. Then convert p to a list and NULL out the 2nd element which is Sepal.Length in the first example and then convert that to a call object and from that to a string. Finally paste p[[2]] onto the front of it using $ as the separator. No packages or regular expressions are used and it is independent of whether the first argument name is Sepal.Length or something else.

f <- function(orig) {
  p <- str2lang(orig)
  paste(p[[2]], format(as.call(replace(as.list(p), 2, NULL))), sep = "$")
}

orig <- "mean(Sepal.Length, na.rm = TRUE)"
f(orig)
## [1] "Sepal.Length$mean(na.rm = TRUE)"

orig2 <- "sum(Sepal.Width)"
f(orig2)
## [1] "Sepal.Width$sum()"

If you know "Sepal.Length" in advance then you don't need regex to stick it on the front, you can paste it there:

pattern = "Sepal.Length"

result = sub(pattern = paste0(pattern, ", "), replacement = "", x = orig, fixed = TRUE)
result = paste0(pattern, "$", result)
result
# [1] "Sepal.Length$mean(na.rm = TRUE)"