Instead of using wgTidy we could be less naive when constructing the html extract.. E.g. Iterate through dom elements. Using wgtidy to me feels like a last resort.

@phuedx it was stalled because we hadn't decided how to do this not if we should do this. In TextExtracts or in RESTBASE? wgTidy or DOM iteration.

In T166272#3364566, @Jdlrobson wrote:

@phuedx it was stalled because we hadn't decided how to do this not if we should do this. In TextExtracts or in RESTBASE? wgTidy or DOM iteration.

TextExtracts as this is where the buggy behaviour is.

Let's see if Tidy can actually fix these errors first – if nothing else, we shouldn't be relying on deprecated APIs. If not, then we'll have to move to DOM iteration. Let's loop in someone from Parsing-Team--ARCHIVED?

Hopefully this can be fixed by T168329. Right now sentences are determined by a regex. Using HTMLFormatter will make the end result much tidier.

In T166272#3437125, @Jdlrobson wrote:

For those who want well formed HTML we will be providing a new service on RESTBase which will guarantee that (please follow along with T113094)

That thread is a bit confusing on this particular point. Would it be correct to assume that T165017 is the task related to providing well-formed HTML in the summary extracts? In which case, it being resolved then means that the functionality is now available and ready for consumption by third-parties?

Hi @waldyrious sorry for the confusion. The API will be hosted here:
https://en.wikipedia.org/api/rest_v1/page/summary/Spain

We are currently in the process of switching out the old version to the new version which will have well formed HTML. Please track T177431 to keep up to date with updates there!