@U02N27RK69K when you get your 2nd (3rd? nth?) wind, lemme know. Maybe I can help to make those sections easier to find for both adoc and md.

the problem itself isn't particularly difficult. we need to find all the header tags with ids and the text between the headers is paired with the header before it

something must have changed because it did work at one point

or maybe it only works on some fraction of documents

I’m sure we could make it less brittle. I think you were looking at resulting HTML to find sections? Are you just wanting to breakdown by toplevel sections or are you looking at nested sections too?

nested sections? and yeah I was using the rendered html because I need whatever the library generates for the header id

if there's a better way I'm here for it

also a ton of this could be cached but I've avoided doing that, too

By nested mean like: • header1 ◦ some content ◦ header1.1 (nested) ▪︎ moar content • header1.1.1 (nested) ◦ more stuff • header2 ◦ header2 content If I search for stuff, would I match header1.1.1 or header1?

Yeah, caching can come when we see a need for it, I guess. Cljdoc does what seems to me a ton of rework for repeated requests. But maybe we don’t care at this point.

header 1.1.1 would be a match

https://github.com/cljdoc/cljdoc/blob/master/src/cljdoc/render/api_searchset.clj#L10

I have it to match all headers

I don't see why people shouldn't get to as close as possible to what they searched

about caching, I do, at least, cache client-side for a day

we could cache longer without causing much issue now that we also have a version number

Hmmm… I just recently added https://github.com/cljdoc/cljdoc/pull/601… Just looking at how these currently render today: md:

<h2><a href="#formatting-marks" id="formatting-marks" class="md-anchor">Formatting marks</a></h2>

<h2 id="_formatting_marks"><a class="link" href="#_formatting_marks">Formatting marks</a></h2>

well then

https://github.com/cljdoc/cljdoc/blob/7fc7f513a94785d2247b673ff9db4cc6d667dffe/src/cljdoc/render/api_searchset.clj#L32

https://github.com/cljdoc/cljdoc/blob/master/src/cljdoc/render/api_searchset.clj#L54

ok yeah I get it off of the anchor not the header

I don't remember writing that code but that's pretty slick

that md-anchor for class md is a https://github.com/cljdoc/cljdoc/blob/7fc7f513a94785d2247b673ff9db4cc6d667dffe/src/cljdoc/render/rich_text.clj#L98…. If adding some markers would make it easier to do what you need to do, I think we could do that.

But maybe you have enough to go on? Header tag that contains an anchor tag…

I think that's enough to go on? I just don't know why it isn't working

also concerned about how to represent the name of the doc

like do you go "h1 text > h2 text > h3 text"

or do you just use h3 text

Yeah, good question. Next release could start with h3 text. Then refine if necessary?

that's what I was thinking

I glanced at your elegant code, but would need to fire up a REPL to see what’s what.

any time it's loop/recur it's like that 😂

ya, same!

could be a jsoup nuance tho too.. maybe..

totally possible

Want me to take a REPL run at it?

(if you need another set of eyes) (or are burnt out on search!) happy to do it if it would be helpful

if you'd like! I'm at family stuff all weekend so can't really do much with it

I'd love it

Coolio, will get to it sometime quite soon!

Oh I see… your elegant code is not being invoked https://github.com/cljdoc/cljdoc/blob/7fc7f513a94785d2247b673ff9db4cc6d667dffe/src/cljdoc/render/api_searchset.clj#L100-L103 (:cljdoc.doc/type doc) will always return nil because that key is down under a doc’s :attrs.

oh wow

well, it's an easy fix?

as much as I'm proud of this code I'm not proud of those function names

Ya, I expect! I’ll fiddle and see. Not sure why we have that :attrs subkey… but that’s a question for another day!

I'm sure there's a good historical reason 😊

Yeah… maybe! Maybe not! It is what is and that’s OK.

Yep, your code is breaking down .md by sections with a wee fix. And it would not take much more for it to also handle .adoc, (now that we have that recent change).

do you want to make the adoc fix, too?

otherwise I anticipate having more energy this next week (I hope)

Yeah I could take a crack at it, unless the fix would bring you some pleasure/fun/joy? (I’d not want to steal that from you, but am also at the same time, happy to help!) If if do this, should I bump some client version cache thingy, to force a client-side update?

oh, no, I mostly care that people benefit from it so go for it!!

you could bump this version if you want but the cache only is around for a day so it's nbd if not https://github.com/cljdoc/cljdoc/blob/master/js/single-docset-search.tsx#L11

Ok, cool, I’ll give it a go!

Ok started to take a peek. A little complication is that the html for adoc sections is nested. Ex. sect3 div is within sect2 div is within sect1 div. This is not the case for html for md which has a flat structure. I’m sure I can sort it out. I’ll take another peek tomorrow.

I figured it needed a whole different way for adoc

Ya, you were right! Taking another swing at it now. Getting there, I think!

At the test phase now. Very cool!

yesssss

I hope I have not trounced too badly on your beautiful code! I’m a little bleary eyed, so won’t merge just yet. https://github.com/cljdoc/cljdoc/pull/616?

looks awesome and I love the doc improvements 💜

Thanks! That was me just wrapping my head around stuff and jotting notes.

And thanks for the PR review, mucho appreciated! I'll merge tomorrow, I don't have enough pep left tonight to recover should anything go sideways!