Improve Swiper minibuffer highlighting

[basil-conto]

Commentary

ivy-highlight-functions-alist does (and probably should) not include an entry for swiper--re-builder, so highlighting always falls back to ivy--highlight-default, even when ivy--regex-function is set to, say, ivy--regex-ignore-order.

Temporarily updating ivy-highlight-functions-alist to include an entry for swiper--re-builder is one suboptimal way to adapt highlighting accordingly. @abo-abo Can you think of a better way to achieve this, short of redesigning the whole regexp-builder/highlight-function system?

Re: #1550

Example

1. make plain

2. (setf (alist-get t ivy-re-builders-alist) #'ivy--regex-ignore-order)

3. C-j
4. C-xC-fswiper.elRET
5. C-sswiper isearch with an overview

Before

After

[basil-conto]

Temporarily updating ivy-highlight-functions-alist to include an entry for swiper--re-builder is one suboptimal way to adapt highlighting accordingly.

It is actually not suboptimal, but rather entirely wrong - the improved substring highlighting in the After screenshot is only applied the first time around; any repeated invocation of swiper reverts to the highlighting in the Before screenshot.

[muirmanders]

Coincidentally, I started trying to fix regex-ignore-order highlighting in swiper recently. I was looking at the buffer highlighting first instead of the minibuffer highlighting. For highlighting, swiper seems to only support regex builders that return a single regex. In particular swiper--update-input-ivy takes the caar of the regex parts, discarding all but the first regex part in the case of ignore-order. swiper--add-overlays similarly only works with a single regex. I was reworking these to operate on a list of regex parts.

Edit: just saw your comment in #1550 (comment)

Should I pause my effort and see what happens with this PR, or do you think they are unrelated? I'd like to contribute to making regex-ignore-order a first class regex option. I've noticed some other worrisome things like ignore-order does not make use of ivy--regex which seems to set some global state variables, so maybe other things don't work as well.

[basil-conto]

@muirmanders

Should I pause my effort and see what happens with this PR, or do you think they are unrelated?

I opened this PR because I thought I had stumbled across a quick fix to minibuffer highlighting at least, but I was wrong. I'm not actually familiar with how everything interoperates and I don't currently have much free time, so please don't pause your efforts because of me.

I'd like to contribute to making regex-ignore-order a first class regex option.

As a fellow user, I look forward to seeing the fruit of your efforts. Thanks for working on this.

[abo-abo]

Thanks for working on this.

Same from me. I would love to get this fixed, but I don't have much free time in the coming week or two.

[basil-conto]

Seeing as I won't be working on this again in the immediately foreseeable future, should I close this PR, or would you rather it be kept open as a reminder?

[muirmanders]

If you are done with it, please close it.