regex - return all before the second occurrence

Question

Given this string:

DNS000001320_309.0/121.0_t0

How would I return everything before the second occurrence of "_"?

DNS000001320_309.0/121.0

I am using R.

Thanks.

Solution 1

I think this might do the task (regex to match everything befor the last occurence of _):

_([^_]*)$

E.g.:

> sub('_([^_]*)$', '', "DNS000001320_309.0/121.0_t0")
[1] "DNS000001320_309.0/121.0"

Solution 2

The following script:

s <- "DNS000001320_309.0/121.0_t0"
t <- gsub("^([^_]*_[^_]*)_.*$", "\\1", s)
t

will print:

DNS000001320_309.0/121.0

A quick explanation of the regex:

^         # the start of the input
(         # start group 1
  [^_]*   #   zero or more chars other than `_`
  _       #   a literal `_`
  [^_]*   #   zero or more chars other than `_`
)         # end group 1
_         # a literal `_`
.*        # consume the rest of the string
$         # the end of the input

which is replaced with:

\\1       # whatever is matched in group 1

And if there are less than 2 underscores, the string is not changed.

Solution 3

Personally, I hate regex, so luckily there's a way to do this without them, just by splitting the string:

> s <- "DNS000001320_309.0/121.0_t0"      
> paste(strsplit(s,"_")[[1]][1:2],collapse = "_")
[1] "DNS000001320_309.0/121.0"

Although of course this assumes that there will always be at least 2 underscores in your string, so be careful if you vectorize this and that isn't the case.

Solution 4

not pretty but this will do the trick

mystr <- "DNS000001320_309.0/121.0_t0"

mytok <- paste(strsplit(mystr,"_")[[1]][1:2],collapse="_")

View more solutions

Comments

[Bart Kiers]

Yeah, if the regex-path is walked, sub would be more appropriate than gsub.

[joran]

If there are more than two underscores this will select beyond the second underscore, although apparently that doesn't matter to the OP, so I point it out only for posterity.

[Bart Kiers]

Yeah, good point @joran. It may matter to the OP, but he might not be aware of it.

[daroczig]

Wow, really nice, detailed answer (+1). I definitely like your solution better than mine :)

[Connor Murray]

What if there are more than 2 underscores?? Is there a way to further generalize this function?