0

Building off this previous post: How to extract string after 2nd delimiter in R

Have a string like the following:

"dat1/set1/set1_covars.csv"

And want to extract all the values before the second / like:

"dat1/set1/"

I was using variations of:

sub("^([^/]+/){2}", "", "dat1/set1/set1_covars.csv")

With ^ and .* moved around in different places, but just can't seem to get the syntax right.

Any help would be appreciated.

TDeramus
  • 77
  • 5
  • 1
    You could use `dirname("dat1/set1/set1_covars.csv")` to get the path to the last directory, if that's what the problem really is. (And `paste0` a `"/"` on the end if needed.) – Gregor Thomas Jul 11 '23 at 20:15
  • I used my own built-in neural network to solve this problem, but I wonder whether ChatGPT would answer this correctly ... – Ben Bolker Jul 11 '23 at 20:20
  • 1
    It's deeper than that unfortunately. It's a list of strings but I didn't clarify that in the thread due to all the extra things that would involve writing. – TDeramus Jul 11 '23 at 20:20
  • @BenBolker No Copilot here unfortunately. – TDeramus Jul 11 '23 at 20:20
  • That occurred to me, but it would be good to give a deeper file path as an example, just to head off inappropriate answers like `dirname()` ... – Ben Bolker Jul 11 '23 at 20:21

1 Answers1

2

This seems to work:

sub("(^([^/]+/){2}).*$", "\\1", "dat1/set1/set1_covars.csv")
  • add () around the stuff that defines the stuff up to the second delimiter;
  • add .*$ to include the rest of the line;
  • replace the blank-string replacement with a replacement by the first capture group.

@GregorThomas points out that for this example dirname() would work, but not if your tree is deeper.

Alternatively:

stringr::str_extract("dat1/set1/set1_covars.csv", "^([^/]+/){2}")

It seemed as though you could also do this with a lookbehind expression (i.e., define a pattern "(?<=^([^/]+/){2}).*$" that says ".*$ preceded by two delimiters, but don't count delimiter stuff in the matched expression") and replacing with a blank, but we run into trouble:

  • repetition quantifiers (i.e. "{2}") aren't allowed in lookbehind expressions
  • if we spell out the repetition explicitly ("(?<=^[^/]+/[^/]+/).*$") and specify perl = TRUE then it notes we're only allowed to use fixed-length expressions
  • lookahead/lookbehind always hurts my brain anyway
Ben Bolker
  • 211,554
  • 25
  • 370
  • 453