Org-roam: Normalized completion

[oatmealm]

Brief Abstract

Looking at org-roam-completion-ignore-case I was wondering if it'd be possible to allow searching for the normalized forms of Unicode strings.

Long Description

When text is saved with accents/diacritics matching text for retrieval requires entering the exact forms... which is a nightmare for Hebrew/Arabic.

Proposed Implementation (if any)

I'm not sure if it was brought up here or elsewhere, but there seem to be normalization functions available for regex searches, as mentioned here: https://nullprogram.com/blog/2014/06/13/

Please check the following:

• [ ?] No similar feature requests

[d12frosted]

I also encountered this and you actually need to configure your completion system to support char folding. In many cases it's enough to set search-default-mode to #'char-fold-to-regexp somewhere in your init.el file:

(setq search-default-mode #'char-fold-to-regexp)

(or via any other prefered way). You can verify if that's working by executing following code and typing buna as input:

(completing-read
 "greeting: "
 '("Bună dimineața"
   "Buna dimineata"))

If you are using built-in completion, this should be enough. As well as for ivy, it respects this setting. If you are also using prescient.el, char folding is supported for literal filter. Now, for helm I don't know. 🤷

Here is a screenshot of me using org-roam-find-file (using ivy + prescient):

Hope that helps.

[oatmealm]

Wow! This is so helpful and seems to work both in Hebrew and Arabic. Some false positives, but this really solving the most pressing problem!
Thanks

[jethrokuan]

Closing as mini-buffer completion is not really in the realm of Org-roam.

[d12frosted]

@jethrokuan you forgot to close 😸