Trying to support lite.duckduckgo.com

[king0947]

Question

How to target a result that is multiple <tr> as a group with no wrapping element or <div>?

Initial attempt at lite.duckduckgo.com filtering.

I had submitted a PR (ublacklist/builtin#67) but realized I should have started in Q&A first.

Each set of results on lite.duckduckgo.com is a quadruplet of <tr>.

I was able to target the link and title for each result within the first <tr> using result-link class.

But am not sure how to target additional <tr> containing result-snippet and link-text classes.

I searched issues and found some old code here: #392 & eb9c742

But it seems the code is injecting a <p> tag around the results list (sorry if I am misunderstanding the code), and I am not sure how to do similar within a YAML file.

Thank you for any help or advice!

And thank you for making this super useful extension!

Here is an example of a DuckDuckGo Lite result:

<tr>
  <td valign="top"> 2.&nbsp; </td>
  <td>
    <a rel="nofollow" href="//duckduckgo.com/l/?uddg=http%3A%2F%2Flite.duckduckgo.com%2F" class='result-link'>DuckDuckGo</a>
  </td>
</tr>
<!-- Only show abstract separately if there's a click URL (not EOF) -->
<tr>
  <td>&nbsp;&nbsp;&nbsp;</td>
  <td class='result-snippet'> We would like to show you a description here but the site won&#x27;t allow us. </td>
</tr>
<tr>
  <td>&nbsp;&nbsp;&nbsp;</td>
  <td>
    <span class='link-text'>lite.duckduckgo.com</span>
  </td>
</tr>
<tr>
  <td>&nbsp;</td>
  <td>&nbsp;</td>
</tr>

And here is my SERPINFO so far:

name: DuckDuckGo Lite
version: 1.0.1

pages:
  - name: Lite
    matches:
      - https://lite.duckduckgo.com/lite/
      - https://lite.duckduckgo.com/lite/?*
    commonProps:
      $site: duckduckgo
      $category: web
    results:
      - name: Default
        root: tr
        url: .result-link
        props:
          title: .result-link

Browser

Chrome 144.0.7559.254

uBlacklist Version

9.8.0

Checklist

• I have searched existing discussions and issues to ensure this question hasn't been asked before

[iorate]

Thanks for the detailed write-up — the example HTML and your SERPINFO made the problem clear.

You're right that this isn't currently expressible. SERPINFO assumes one result maps to one root element, and there's no way to say "also hide these sibling <tr>s." The .result-link, .result-snippet, and .link-text rows are separate siblings with no wrapper, so a single root selector can't cover them.

I'm considering adding an extraSelector field to a result definition to address exactly this. It would take a selector (resolved relative to the matched root via the CSS nesting selector &) for additional elements that should be blocked/highlighted along with the root. For your case it would look roughly like:

results:
  - name: Default
    root: tr:has(.result-link)
    extraSelector: "& + tr, & + tr + tr"   # snippet and link-text rows
    url: .result-link
    props:
      title: .result-link

This is still under consideration and the exact syntax may change, but the use case is understood. I'll follow up here when there's progress.

[king0947]

Awesome, thank you!

Does it makes sense to make root a list like so? Might be easier to understand visually that the & CSS nesting selector is relative to the first object of root. Conceptually root is the first item of the list, and any additional items are relative to root and form a root grouping:

results:
  - name: Default
    root:
      - tr:has(.result-link)
      - & + tr
      - & + tr + tr
    url: .result-link
    props:
      title: .result-link

[iorate]

Implemented in #850 — a result definition can now take an extraSelector, a selector list resolved relative to the matched root via the CSS nesting selector &, whose matched elements inherit the root's block/highlight state.

(I kept this separate from root rather than turning root into a list, since root is what identifies a single result — overloading it would blur which element the count, block button, and url/props resolve against.)

This will ship in the next release.

[king0947]

Excellent! Thank you so much, I look forward to using it for lite.duckduckgo.com -- thanks again.

EDIT:

Native support for lite.duckduckgo.com was added in ublacklist/builtin#69

[king0947]

Thanks to version 9.9.0 DuckDuckGo Lite is working great now!

EDIT: native support for lite.duckduckgo.com was added in ublacklist/builtin#69