I'm working with stringr, but the solution doesn't have to be. In the following string I want to replace all the spaces that have a letter to their left and right (i.e. first 2) with underscores.
str = 'ab cd ef 29 17 2 3'
stringr's replace_all don't seem to support substring matching (like str_match_all do by enclosing with (...)). And str_match_all doesn't replace. Grateful for suggestions.
Use gsub function with lookaround based regex.
> str = 'ab cd ef 29 17 2 3'
> gsub("(?<=[A-Za-z])\\s+(?=[A-Za-z])", "_",str, perl=T)
[1] "ab_cd_ef 29 17 2 3"
OR
> gsub("[A-Za-z]\\K\\s+(?=[A-Za-z])", "_",str, perl=T)
[1] "ab_cd_ef 29 17 2 3"
OR
> library(stringr)
> str_replace_all('ab cd ef 29 17 2 3', perl('(?<=[A-Za-z])\\s+(?=[A-Za-z])'), '_')
[1] "ab_cd_ef 29 17 2 3"
stringr can't handle it and returns an invalid expression error for some reason.
(?<=[a-zA-Z])[ ](?=[a-zA-Z])
This should do it for you.Lookbehind will make sure you have letter behind space.
And lookahead will make sure you have a letter ahead of it.See demo.
Here's another possibility:
gsub("([[:alpha:]])(\\s+)([[:alpha:]])", "\\1_\\3", x)
## "ab_cd_ef 29 17 2 3"
