Search code examples
rnlpevalgsubstring-substitution

How to convert string rule to an expression which has a range in R

x1=3 
x2=3
x3=4
x4=7
df_new = data.frame(x1,x2,x3,x4)

a<-"-3>x1>=1.45 & -3<=x2<3 & x4==7"
df_new <- data.frame(x1=3,x2=3,x3=4,x4=7)
eval(parse(text=a),df_new)

I need to convert the above rule to "-3>x1 & x1>=-1.45 & -3<=x2 & x2<3 & x4==7" using regular expressions. Currently it will prompt an error. Please help me to solve this.

Solution

  • I'd do it in several steps:

    1. Split on logical operators into separate inequalities.
    2. Change double inequalities like -3>x1>=-1.45 into two inequalities.
    3. Change "=" to "==", and put it all together.

    For example:

    a1 <- strsplit(a, "&", fixed = TRUE)[[1]]
a1a <- gsub(" ", "", a1) # get rid of spaces
a2 <- gsub("([-0-9.]+[<>=]+)([[:alpha:]]+[[:alnum:]]*)([<>=]+.+)",
           "\\1\\2 & \\2\\3", a1a)
a3 <- paste(gsub("[^<>]={1}", "==", a2), collapse = " & ")

    which gives a3 equal to

    [1] "-3>x1 & x1>=-1.45 & -3<=x2 & x2<3 & x==7"

    This won't work if you have other logical operations besides &, or if your variables have non-alphanumeric characters like dots or underscores. It will also fail on inequalities like x1 <= x2 <= x3. I'll leave you to handle those cases if they come up.