Add tokens_keep and dfm_keep?

[koheiw]

We have tokens_remove(), which is a shortcut to tokens_select(x, selection = 'remove'), but not tokens_keep(). It makes our life easier, especially with regex patters.

For example, we have to write to keep capitalized words

tokens_select(toks, '^[A-Z]', 'keep', 'regex', FALSE)
# or 
tokens_select(toks, '^[A-Z]', valuetype = 'regex', case_insensitive = FALSE)

These become just

tokens_keep(toks, '^[A-Z]', 'regex', FALSE)

The same apply to dfm_select().

Am I too lazy?

[kbenoit]

You're right, it's consistent, but also a slippery slope: we would then need tokens_keep(), and fcm_keep(). But hey, why not. Look at how many functions stringi has!

[koheiw]

Short cut functions do not harm anyone as far as they are consistent.