False positive on "Pass"

[nscherer]

Description

This text: "I’ve been waiting for the first decent snow at Snoqualmie Pass"
was rendered as "I’ve been waiting for the first decent snow at Snoqualmie P***"

Later, when "pass" was not capitalized, it was handled correctly.

Details (Bugs Only)

Firefox Version: 63.0.3 64 bit
Extension Version: don't know, I don't see where this is displayed
Operating System: Windows 10
Site(s) Affected: I assume any but found here in the comments: https://jalopnik.com/reminder-yes-you-absolutely-need-winter-tires-even-if-1830590607

Steps to reproduce (Bugs only)

Other details

Screenshots

[richardfrost]

Thanks for submitting this issue, I'm sorry that its filtering a word that it shouldn't be. Unfortunately this is happening due to how the page is structured, as shown below:

When the filter analyzes the text, it is getting it in pieces. So, the first piece that it gets is: "I live in the PNW too (25", then it gets the next piece, " minutes from Seattle)...", and so on. By the time it gets to the "False positive", you can see that the word Pass was split between two of these pieces.

Unfortunately, I am not sure the best way to address this, but at least we can explain why it is happening. There shouldn't be a false positive with the word "pass", which you can verify in other places, but this instance makes it tough. I will need some time to see if I can find a good way to address this.

[duckbrain]

Sounds like problem is that the text is being split up into separate text nodes.

• Here's a blog post about handling the topic: https://j11y.io/javascript/replacing-text-in-the-dom-solved/
• And a GitHub project that implements it: https://github.com/padolsey/findAndReplaceDOMText

I haven't tested to see if that solves the problem, but a simpler solution could be to just call Node#normalize before the search.

[richardfrost]

Thanks for the suggestions! I will take a look at the implementation when I get some time, but that blog post and accompanying demo look promising.

Unfortunately normalize doesn't seem to work in this instance, and it has the drawback of usually formatting.

[richardfrost]

Thanks to @duckbrain for pointing me to that wonderful library, I have now released version 2.0.0 which should address the issue when running in Advanced mode. I am hesitant to use the library everywhere because I did observe some slowdown, though not significant. Please let me know if it works for you in advanced mode, or if you have any more concerns or comments.