`matcher` attr fuzzy matching

I’ve got a list of countries in JSON format like so:

  {
    'name': 'United States',
    'number': '+1',
    'abbr': 'US'
  },
  {
    'name': 'United Kingdom',
    'number': '+44',
    'abbr': 'GB'
  },
  {
    'name': 'Japan',
    'number': '+81',
    'abbr': 'JP'
  },
  ...

At the moment I’m using power-select to allow the user to search by either the number or name properties by using the matcher option. My matcher code is:

    let concatenatedFields = get(country, 'name') + get(country, 'number');
    let lowerCasedFields = (concatenatedFields).toLowerCase();

    return (concatenatedFields + lowerCasedFields).indexOf(term);

I need to concatenate a lower case version of the fields onto the matcher because if I don’t then the matcher is taken literally and I only get exact matches. For example united states would not trigger a match for United States, which is annoying.

Basically, I want to be able to make use of the power-select’s fuzzy matching with the matcher attribute that you get when you use the searchField option, and I feel like this is how it should work out of the box.

This would mean that my matcher code just be:

    let concatenatedFields = get(country, 'name') + get(country, 'number');
    return (concatenatedFields).indexOf(term);

and I’d get all of the fuzzy matching options like fuzzy case matching as well as all the other great fuzzy matching features power-select has!

I’m not sure if this is already possible somehow. If it is I apologise.

I think another good way of achieving this behaviour would be to enable you to pass an array of the fields that you want to search by into the searchField attribute or create a searchFields option that would take multiple fields as described.

[cibernox]

@BennyAlex I don’t think it would be wise as there is a lot of possible strategies. Imagine you have first_name and last_name. If the user types jo it will find “John Williams” but also “Scarlet Johanson”. So far so good. But what about n a. Should it find “Anton Allaire” or “Akira Ronin” (because properties can be concatenated as “Ronin Akira”)? Or both? The space is meaninful or should “Navidad” also match?

I gave it some thought and I think that it will be better to let people do their own matchers. To avoid repetition you can extract your matchers to util helpers that you just import.

import { matchNameAndSurname } from 'my-app/utils/custom-matchers';
export default Component.extend({
  matchNameAndSurname
});

{{#power-select ... matcher=matchNameAndSurname}}

If you want to skip the javascript file, a helper to generate matchers is also extremely simple:

{{#power-select ... matcher=(match-by "firstName" "lastName")}}

export default Helper.helper(function(fields, { separator = ' ' } = {}) {
  return function(item, term) {
    let concatenation = fields.map((f) => get(item, f)).join(separator);
    return concatenation.indexOf(term);
  }
});

[cibernox]

You can. I made the default matches function an exported util: https://github.com/cibernox/ember-power-select/blob/master/addon/utils/group-utils.js#L970

It doesn’t use fuzzy matching, but it normalizes diacritics like áàøçĉ and such. If you invoke it with defaultMatcher(concatenatedFields, searchTerm) you will get the default behaviour searching on both fields. But it’s not fuzzy.