Replacing text in gsub by evaluating a backreference

user_14770546 0
Viewed 80

Let's say i have some text :

myF <- "lag.variable.1+1"

I would like to get for all similar expressions the following result : lag.variable.2 (that is replacing 1+1 by the actual sum

The following doesn't seem to work, it appears that the backreference doesnt carry through in the eval(parse() bit ):

myF<-gsub("(\\.\\w+)\\.([0-9]+\\+[0-9]+)", 
            paste0( "\\1." ,eval(parse(text ="\\2"))) ,
            myF )

Any tips on how to achieve the desired result ? Thanks!

backreference gsub r regex
2 Answers

We can use gsubfn

library(gsubfn)
gsubfn("(\\d+\\+\\d+)", ~ eval(parse(text = x)), myF)
#[1] "lag.variable.2"

 gsubfn("\\.([0-9]+\\+[0-9]+)", ~ paste0(".", eval(parse(text = x))), myF2)
#[1] "lag.variable0.3 * lag.variable1.2 + 9892"

Or with str_replace

library(stringr)
str_replace(myF, "(\\d+\\+\\d+)", function(x) eval(parse(text = x)))
#[1] "lag.variable.2"

Or an option with strsplit and paste

v1 <- strsplit(myF, "\\.(?=\\d)", perl = TRUE)[[1]]
paste(v1[1], eval(parse(text = v1[2])), sep=".")
#[1] "lag.variable.2"

data

myF <- "lag.variable.1+1"
myF2 <- "lag.variable0.3 * lag.variable1.1+1 + 9892"
user_37322710

Here is how you can use your current pattern with gsubfn:

library(gsubfn)
x <- " lag.variable0.3 * lag.variable1.1+1 + 9892"
p <- "(\\.\\w+)\\.([0-9]+\\+[0-9]+)"
gsubfn(p, function(n,m) paste0(n, ".", eval(parse(text = m))), x)
# => [1] " lag.variable0.3 * lag.variable1.2 + 9892"

Note the match is passed to the callable in this case where Group 1 is assigned to n variable and Group 2 is assigned to m. The return is a concatenation of Group 1, . and evaled Group 2 contents.

Note you may simplify the callable part using a PCRE regex (add perl=TRUE argument) \K, match reset operator that discards all text matched so far:

p <- "\\.\\w+\\.\\K(\\d+\\+\\d+)"
gsubfn(p, ~ eval(parse(text = z)), x, perl=TRUE)
[1] " lag.variable0.3 * lag.variable1.2 + 9892"

You may further enhance the pattern to support other operands by replacing \\+ with [-+/*] and if you need to support numbers with fractional parts, replace [0-9]+ with \\d*\\.?\\d+:

p <- "(\\.\\w+)\\.(\\d*\\.?\\d+[-+/*]\\d*\\.?\\d+)"
## or a PCRE regex:
p <- "\\.\\w+\\.\\K(\\d*\\.?\\d+[-+/*]\\d*\\.?\\d+)"
user_38329700
Related