Using shorthand character classes inside character classes in R regex

Question

I have defined

vec <- "5f 110y, Fast"

and

gsub("[\\s0-9a-z]+,", "", vec)

gives "5f Fast"

I would have expected it to give "Fast" since everything before the comma should get matched by the regex.

Can anyone explain to me why this is not the case?

Answer 1

You should keep in mind that, in TRE regex patterns, you cannot use regex escapes like \s, \d, \w inside bracket expressions.

So, the regex in your case, "[\\s0-9a-z]+,", matches 1 or more \, s, digits and lowercase ASCII letters, and then a single ,.

You may use POSIX character classes instead, like [:space:] (any whitespaces) or [:blank:] (horizontal whitespaces):

> gsub("[[:space:]0-9a-z]+,", "", vec)
[1] " Fast"

Or, use a PCRE regex with \s and perl=TRUE argument:

> gsub("[\\s0-9a-z]+,", "", vec, perl=TRUE)
[1] " Fast"

To make \s match all Unicode whitespaces, add (*UCP) PCRE verb at the pattern start: gsub("(*UCP)[\\s0-9a-z]+,", "", vec, perl=TRUE).

Answer 2

Could you please try folllowing and let me know if this helps you.

vec <- c("5f 110y, Fast")
gsub(".*,","",vec)

OR

gsub("[[:alnum:]]+ [[:alnum:]]+,","",vec)