xpath selector doesn't work on `text()` nodes #1777

DetachHead · 2021-10-24T08:10:23Z

Prerequisites

I verified that this is not a filter issue (MUST be reported at filter issue tracker)
This is not a support issue or a question
I performed a cursory search of the issue tracker to avoid opening a duplicate issue
The issue is not present after wholly disabling uBlock Origin ("uBO") in the browser
I checked the documentation to understand that the issue I report is not a normal behavior

I tried to reproduce the issue when...

uBO is the only extension
uBO with default lists/settings
using a new, unmodified browser profile

Description

xpath selectors don't seem to work on individual text nodes, for example when an element has multiple text nodes and you're trying to match one of them

A specific URL where the issue occurs

n/a but see steps to reproduce for a minimal html file to reproduce

Steps to Reproduce

start a webserver with the following html:

<body>
  <p/> 
</body>
<script>
  document.querySelector('p').appendChild(document.createTextNode('hello'));
  document.querySelector('p').appendChild(document.createTextNode('hello2'));
</script>

search for the following xpath in devtools to verify that it's correct: //p/text()[.='hello2']
attempt to create a filter rule using the same xpath: ##body:has(:xpath(//p/text()[.='hello2']))

Expected behavior

body element is blocked as the //p/text()[.='hello2'] xpath should match a node within it

Actual behavior

no match

uBlock Origin version

1.38.6

Browser name and version

edge 94.0.992.50

Operating System and version

windows 10

The text was updated successfully, but these errors were encountered:

uBlock-user · 2021-10-24T08:22:56Z

Text nodes cannot be queried in uBO.

DetachHead · 2021-10-24T08:24:29Z

why not?

uBlock-user · 2021-10-24T08:32:42Z

Only HTML Elements are queried by uBO, non-HTML Elements are discarded, so not possible. Support for querying text nodes doesn't exist in Chromium/Firefox either.

@DetachHead - w3c/csswg-drafts#2208

uBlock-user · 2021-10-24T08:49:31Z

Duplicate of #1654

gorhill · 2021-10-24T14:59:40Z

why not?

The question should be the other way around, you need to provide cases where that would be genuinely useful in the real world.

DetachHead · 2021-10-24T15:04:11Z

@gorhill facebook ads

gorhill · 2021-10-24T15:10:56Z

Use :has-text()? Looks to me a typical case of XY problem.

DetachHead · 2021-10-24T15:24:11Z

That's what I went with, but I don't like that approach because it seems less accurate, as in it seems more likely that some completely unrelated div that has the word sponsored in it might get filtered. that's why I try to avoid non-exact matching where I can

I also tried it with regex to get an exact match (:has-text(^Sponsored$)) but that didn't work either, presumably for the same reason

I was simply pointing out a feature that wasn't working as expected and provided a minimal example to reproduce it

gorhill · 2021-10-24T15:32:14Z

I also tried it with regex to get an exact match (:has-text(^Sponsored$)) but that didn't work either

You are not reading the documentation properly. Read carefully, especially before opening invalid issues. If you read carefully -- and it's very clearly explained in the documentation -- you would have understood that what you want is :has-text(/^Sponsored$/).

DetachHead · 2021-10-24T15:53:45Z

that was a typo

i get that you probably have to deal with dozens of invalid issues every day, i apologize if i didn't make my use case clear in the op but i'm just wondering if there's a way around situations like this, because from what i can tell exact matching with regex just doesn't work here

gorhill · 2021-10-24T16:05:59Z

because from what i can tell exact matching with regex just doesn't work here

It appears you purposefully designed your regex to not match, while you avoided to reveal what is under that first p tag in the inspector so that we can point out to you why specifically your regex does not match.

DetachHead · 2021-10-24T16:14:18Z

here is the full html

<body>
    <p/> 
    <p>this is not a Sponsored post. don't block me</p>
</body>
<script>
document.querySelector('p').appendChild(document.createTextNode('Sponsored'));
document.querySelector('p').appendChild(document.createTextNode('.'));
document.querySelector('p').appendChild(document.createTextNode('some other text'));
</script>

i didn't mean for the p to be collapsed in that screenshot but it's the same as what i had at the start i just changed the text.

DetachHead · 2021-10-24T16:22:35Z

so it turns out the innerHTML has whitespace at the start which seems to be the cause. i can go ##p:has-text(/^\s+Sponsored/) which works

gorhill · 2021-10-24T16:28:55Z

In any case, you can use ##body:xpath(//p[text()="Sponsored"]), this works already.

u-RraaLL · 2021-10-24T17:08:47Z

That's what I went with, but I don't like that approach because it seems less accurate, as in it seems more likely that some completely unrelated div that has the word sponsored in it might get filtered. that's why I try to avoid non-exact matching where I can

Not if you narrow down the matches with with other attributes and ancestor nodes.

facebook.com##[role="feed"] span[id]>[role=button]:has-text(/^Sponsored|Paid for by/):upward([role="feed"]>div)

https://www.reddit.com/r/uBlockOrigin/wiki/solutions#wiki_facebook

DetachHead · 2021-10-24T17:23:01Z

DetachHead :

Prerequisites

[x] I performed a cursory search of the issue tracker to avoid opening a duplicate issue

Really? Simply typing xpath in the tracker search field, and it returns the thread you duplicated (#1654) as the first result.

as it says i performed a cursory search. the word "xpath" wasn't in either the title or description of that issue

uBlock-user added the something to address something to address label Oct 24, 2021

uBlock-user marked this as a duplicate of #1654 Oct 24, 2021

uBlock-user closed this as completed Oct 24, 2021

uBlock-user added duplicate This issue or pull request already exists and removed something to address something to address labels Oct 24, 2021

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

xpath selector doesn't work on `text()` nodes #1777

xpath selector doesn't work on `text()` nodes #1777

DetachHead commented Oct 24, 2021

uBlock-user commented Oct 24, 2021

DetachHead commented Oct 24, 2021

uBlock-user commented Oct 24, 2021 •

edited

Loading

uBlock-user commented Oct 24, 2021

gorhill commented Oct 24, 2021

DetachHead commented Oct 24, 2021

gorhill commented Oct 24, 2021 •

edited

Loading

DetachHead commented Oct 24, 2021

gorhill commented Oct 24, 2021

DetachHead commented Oct 24, 2021

gorhill commented Oct 24, 2021

DetachHead commented Oct 24, 2021

DetachHead commented Oct 24, 2021

gorhill commented Oct 24, 2021

u-RraaLL commented Oct 24, 2021

DetachHead commented Oct 24, 2021

Prerequisites

xpath selector doesn't work on text() nodes #1777

xpath selector doesn't work on text() nodes #1777

Comments

DetachHead commented Oct 24, 2021

Prerequisites

I tried to reproduce the issue when...

Description

A specific URL where the issue occurs

Steps to Reproduce

Expected behavior

Actual behavior

uBlock Origin version

Browser name and version

Operating System and version

uBlock-user commented Oct 24, 2021

DetachHead commented Oct 24, 2021

uBlock-user commented Oct 24, 2021 • edited Loading

uBlock-user commented Oct 24, 2021

gorhill commented Oct 24, 2021

DetachHead commented Oct 24, 2021

gorhill commented Oct 24, 2021 • edited Loading

DetachHead commented Oct 24, 2021

gorhill commented Oct 24, 2021

DetachHead commented Oct 24, 2021

gorhill commented Oct 24, 2021

DetachHead commented Oct 24, 2021

DetachHead commented Oct 24, 2021

gorhill commented Oct 24, 2021

u-RraaLL commented Oct 24, 2021

DetachHead commented Oct 24, 2021

Prerequisites

xpath selector doesn't work on `text()` nodes #1777

xpath selector doesn't work on `text()` nodes #1777

uBlock-user commented Oct 24, 2021 •

edited

Loading

gorhill commented Oct 24, 2021 •

edited

Loading