How to manipulate regex replacement strings in Elixir

Question

I found myself wanting to do this in Elixir:

re_sentence_frag = %r/(\w([^\.]|\.(?!\s|$))*)(?=\.(\s|$))/
Regex.replace(re_sentence_frag, " oh.  a DOG. woOf. ", String.capitalize("\\1"))

Of course, that has no effect. (It capitalizes the string "\\1" just once.) What I really meant is to apply String.capitalize/1 to every match found by the replace function. But the 3rd parameter can't take a function reference, so passing &(String.capitalize("\\1") also doesn't work.

This seems so fundamental that I'm surprised it's not possible. Is there another approach that would as neatly express this kind of manipulation? It looks like the underlying Erlang libraries would not immediately support passing a function reference as the 3rd parameter, so this may not be completely trivial to fix in Elixir.

How would you program manipulation of each matched string?

The "\\1" is meant for consumption by the regex engine, not the String class. — user557597
– user557597, Commented Jan 22, 2014 at 17:54
I would look if a function ref is a parameter option. Where the function recieves match results and returns the replacement string. If it can't do that, then you have to reconstruct a new string in a regex find loop. — user557597
– user557597, Commented Jan 22, 2014 at 17:57
Your best bet is use scan and use the information from the result to manually replace them. For code-reuse purpose, you can create a wrapper function that accepts a function as parameter. — nhahtdh
– nhahtdh, Commented Jan 22, 2014 at 17:58
@nhahtdh, I think you're correct, although I was leaning toward using split. One of the goals above was to avoid changing the in-between bits, and it's not clear how to get those (or include them in the final results) using scan. I'll post one possible answer based on split. — Daniel Ashton
– Daniel Ashton, Commented Jan 22, 2014 at 21:36
You are right. We can't pass a function to the Erlang side so it is non trivial to support this feature. :( Split seems to be the best way to go for this case. — José Valim
– José Valim, Commented Jan 22, 2014 at 23:34

Daniel Ashton · Accepted Answer · 2014-01-22 21:41:44Z

2

Here is one solution based on split:

" oh.  a DOG. woOf. pi is 3.14159. try version 7.a." |>
String.split(%r/(^|\.)(\s+|$)/)                      |>
Enum.map_join(&String.capitalize/1)

I guess it's not much more clumsy than my original attempt. The regex is considerably simpler, as it only needs to find the bits between sentences.

answered Jan 22, 2014 at 21:41

Daniel Ashton

1,44812 silver badges24 bronze badges

Sign up to request clarification or add additional context in comments.

1 Comment

user557597 Over a year ago

I left some comments on split under your original posted question. Don't know if it will help or not.

Collectives™ on Stack Overflow

How to manipulate regex replacement strings in Elixir

1 Answer 1

1 Comment

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

1 Comment

Your Answer

Sign up or log in

Post as a guest

Linked

Related