indices for more than one needle in Perl 6

Question

I want to find indices for more than one letter in a word. I don't want to use Regexes, because they will slow down the program (which is already slower than I wanted).

> "banana".indices(("a", "b").any)
any((1 3 5), (0))

How can I instead get 0, 1, 3, 5?

score 9 · Accepted Answer · answered May 11 '18 at 21:32

9

I would go for something like this (in the REPL):

> gather "banana".indices("a"|"b").deepmap: *.take
(1 3 5 0)

answered May 11 '18 at 21:32

nxadm

2,079
12
17

chenyf · Answer 2 · 2018-05-12T05:20:48.823

8

> "banana".comb.grep: 'a' | 'b',:k
(0 1 3 5)

I don't know if comb use regex in this case:

gather for 'banana'.comb.antipairs  {.value.take if .key ∈ ['a','b'] } 

# or
gather 'banana'.comb.antipairs».&{.value.take if .key ∈ ['a','b'] }

edited May 12 '18 at 05:20

answered May 12 '18 at 04:05

chenyf

5,048
1
12
35

Nice, but the OP said he didn't want to use regex. – Håkon Hægland May 12 '18 at 04:24
@Håkon Hægland is `"banana".comb` equals to `"banana".comb(/./)` – chenyf May 12 '18 at 05:20
Sorry I wasn't thinking. Although the name `grep` comes from "global regexp" it is not using regex here. `'a' | 'b'` is just a junction of two characters. I am not sure about `comb`, but I see no reason why it should use regex. – Håkon Hægland May 12 '18 at 05:31
1

Actuallly, `.comb` does *not* use any regexes internally. It just creates an iterator on all graphemes in the string. – Elizabeth Mattijsen May 12 '18 at 07:13

score 7 · Answer 3 · answered May 11 '18 at 22:40

7

My solution would be:

> <a b>.map( { |"banana".indices($_) } ).sort
(0 1 3 5)

Basically, loop over all of the letters you want to look for (<a b>.map) and map those letters to their indices ("banana".indices($_)), then slip the indices found (|) and sort the result (.sort).

answered May 11 '18 at 22:40

Elizabeth Mattijsen

25,654
3
75
105

Which version, do you think, will be the fastest? – Eugene Barsky May 12 '18 at 08:16
3

@EugeneBarsky What is fastest today may not be the fastest in the future. It may also be dependent on your particular data. If speed is important to you, I recommend trying the different approaches on your machine with your data (or a representative sample of your data). – Christopher Bottoms May 12 '18 at 11:08
@ChristopherBottoms Thanks, I'll try it. – Eugene Barsky May 12 '18 at 12:06

indices for more than one needle in Perl 6

3 Answers3