Why is this regex using lookbehinds invalid in R?

Question

I'm trying to do a lookbehind regex in R to find a pattern. I expect this would pull the 'b' in 'bob', but instead I get an error.

> regexpr("(?<=a)b","thingamabob")
Error in regexpr("(?<=a)b", "thingamabob") : 
invalid regular expression '(?<=a)b', reason 'Invalid regexp'

This does not throw an error, but it also doesn't find anything.

> regexpr("(.<=a)b","thingamabob")
[1] -1
attr(,"match.length")
[1] -1
attr(,"useBytes")
[1] TRUE

I'm confused because the help page for regexpr specifically indicates that lookbehind should work: http://stat.ethz.ch/R-manual/R-patched/library/base/html/regex.html

Any ideas?

Oh fine, but getting rep for questions like these makes me feel so dirty. — joran, Nov 16 '12 at 16:31
@joran If it makes you feel better we could all downvote you? ;) — Dason, Nov 16 '12 at 16:34
@Dason If you're willing to spend the rep to assuage my guilt, far be it from me to stand in your way! :) — joran, Nov 16 '12 at 16:42
If you trying to get everything from the first b after an a onwards then this will do it without lookbehind: `library(gsubfn); strapplyc("thingamabob", "a(b.*)")` — G. Grothendieck, Nov 16 '12 at 17:47
@joran, I had the same issue and someone referred me to this page and setting Perl=TRUE to solve my issue as well but honestly, I am not sure how it's working. Tried to search but it's vague for me. Can you please explain that in more detail? — Ross_you, Nov 07 '20 at 03:11

score 25 · Accepted Answer · answered Nov 16 '12 at 16:30

25

You just need to switch to PERL regular expressions by setting perl = TRUE.

answered Nov 16 '12 at 16:30

joran

1

Regardless of the simplicity of the issue. I was still missing that small part. Thanks. – Jesse Nov 16 '12 at 16:35
@joran, can you please explain what exactly `Perl` is doing? I searched and found this `perl: logical. Should perl-compatible regexps be used? Has priority over extended` here , http://www.endmemo.com/r/sub.php. but it's still not clear for me – Ross_you Nov 07 '20 at 03:09

1 Answers1