[RFC] more fuzzy matcher

As per https://github.com/ncm2/ncm2/issues/21

However, pure fuzzy match may not be a good idea.

Here’s a proposal:

For candidate like HelloWorld, split it into Hello and World,

We only allow substr match in Hello, and World,

So ellold should match ell in Hello, rld in World.

But with eoworld , eo shouldn’t match Hello. So it should not match HelloWorld

I’m not sure what sort priority is better. It seems the abbrfuzzy still works for this proposal.

// cc @balta2ar

[roxma]

@fcying

You could try this config for now:

    let g:ncm2#matcher = {
                \ 'name': 'combine',
                \ 'matchers': ['abbrfuzzy', 'substr']
                \ }

Read https://github.com/ncm2/ncm2/pull/23 for more info

[balta2ar]

Thank you! Because I have only one plugin that is likely to require fuzzy matching, I enabled it only there and I’ll see how it goes.

let g:ncm2_jira#source = get(g:, 'ncm2_jira#jira_source', {
            \ 'name': 'jira',
            \ 'priority': 9,
            \ 'mark': 'jira',
            \ 'word_pattern': 'JI\w*',
            \ 'complete_pattern': 'JI\w*',
            \ 'matcher': {'key': 'custom', 'name': 'substrfuzzy'},
            \ 'on_complete': 'ncm2_jira#on_complete',
            \ 'on_warmup': 'ncm2_jira#on_warmup'
            \ })