issue 923 allow search results paging on single page

[jamesmisson]

Fixes a bug identified in Issue 923 that disabled the next/previous search results buttons when the manifest had a single canvas. #923

[vercel]

The latest updates on your projects. Learn more about Vercel for Git ↗︎

Name	Status	Preview	Comments	Updated (UTC)
universalviewer	✅ Ready (Inspect)	Visit Preview	💬 Add feedback	Jul 8, 2024 4:12pm

[demiankatz]

@jamesmisson, I re-targeted this from the main branch to the dev branch, since new code should get merged to dev first.

[demiankatz]

See below for some thoughts and questions; this is based on a fairly shallow look at the code, so apologies if I'm missing any important details! :-)

[demiankatz]

A couple of comments on this:

1.) I'm wondering if this special case is masking a different/larger bug. I'm not sure why we need to account for having just one canvas differently. If there's only one canvas, then shouldn't currentCanvasIndex equal lastSearchResultCanvasIndex? And if that's the case, how is it any different from being on the last page of a multi-page result? If that's not the case, then maybe there's a bug related to determining the current index, or in one of the other comparisons, and we'd be better off fixing that underlying problem instead of adding a workaround at the end.

2.) If we do indeed need to keep this code as-is, some minor stylistic changes are needed -- return true is missing a semi-colon, and the closing brace needs to be dropped to the next line. I'm guessing that perhaps a couple of characters were accidentally deleted here, but Javascript is tolerant enough that it works anyway. :-)

[jamesmisson]

Thanks for taking a look Demian. This did cross my mind, but currently the only logic that activates the next button in the first instance is the final return statement of this method (now line 388). The block above it is relevant only when an annotation is 'current' (after 'next' is pressed for the first time), and then it's checking that it is neither the last page or the last annotation. So my change is essentially just adding another condition to check for along with line 388. This is hard to explain! But we can discuss on the next UV call.

[demiankatz]

I see -- but I still wonder if there's a more general solution to this problem. Is it possible that there's another edge case involving (<OpenSeadragonExtension>this.extension).currentAnnotationRect that isn't being properly taken into account? I would think that if there is no current rectangle, but there are highlighted search results, we would want to allow the next button to be active. Additionally, if there is only one canvas but the user has navigated to the last highlighted search result, we would NOT want the next button to be active, but I think with this current code it will never get disabled.

We should also double check that any changes we make to the "next" logic do not have matching bugs in the "previous" logic.

As you say, we can discuss further on the call -- this logic is definitely a bit hard to wrap one's head around!

[jamesmisson]

The more streamlined solution worked so I've edited the PR. Thanks Demian!

[demiankatz]

Thanks, @jamesmisson!

[demiankatz]

I've approved this and added it to the community board. I think it can probably be merged, but I'll wait until we have clarity out of steering group before I do anything.