help with a Find regex

[Ilcaa]

here is my html

<p class="c8">
    <span class="c3">It is a long established fact that a reader will be distracted by the readable content of a page when looking at its layout. The point of using Lorem Ipsum is .</span>
  </p>

my regex is

/<span[^>]*>(.*?)<\/span>/g

if i use a regex tester online it work, https://regex101.com/r/cqkkT1/1

I am trying to find the text within the span, i will eventually remove the span and place the text in the parent p-tag.

in calibre it doesnt find the text, "no matches were found". any help? thanks

[theducks]

K I S S
Use diaps toolbag plugin. It does proper matching of nested opening and closing tags
Preferences: Getplugins: Diaps Editing Toolbag

[Karellen]

Ensure you have set the search Mode to Regex.

This might work better...

Find...
<p class="c8">\s+<span class="c3">(.*?)</span>\s+</p>

Replace...
<p class="c8">\1</p>

Assuming the classes are static.

[JSWolf]

Originally Posted by theducks

K I S S
Use diaps toolbag plugin. It does proper matching of nested opening and closing tags
Preferences: Getplugins: Diaps Editing Toolbag

Better to now use Otter's Editing Toolbag as it does the same thing and more. It's the new version.

[lomkiri]

Originally Posted by Ilcaa

my regex is

/<span[^>]*>(.*?)<\/span>/g

The correct regex for calibre is :

<span[^>]*>(.*?)<\/span>

But it will rip out ALL the spans in your epub, you'll have to pass it in manual, or to be more specific about the class of the span (as Karellen emphasized)

The Theducks' suggestion to use the diaps toolbag plugin is excellent, it's very flexible and will take care of the embedded spans.

Originally Posted by JSWolf

Better to now use Otter's Editing Toolbag as it does the same thing and more. It's the new version.

To be noticed that this plugin is vibe-coded. The author is honest and advises in the 1st line of the post. But for my concern, I wont use LLM-generated code. I know this wave will submerge most of the beach, but the implications of those techniques are frightening enough (in various aspects : human, intellectual, social, ecological and educational, without speaking of the quality of the production) for me to avoid the most possible what LLMs have produced.

For the same reason (and to avoid to overshadow the contents of the web), I don't use google and Cie, not even bare ddg, but ddg with its noAI option : https://noai.duckduckgo.com

[JSWolf]

If there a reason not to use Otter's Editing Toolbag just because it was created using AI? It does work.

[theducks]

Originally Posted by JSWolf

If there a reason not to use Otter's Editing Toolbag just because it was created using AI? It does work.

Yes!
We recommended tools we had USED regularly for the task and our recommendations are still VALID.

[DNSB]

Originally Posted by Ilcaa

if i use a regex tester online it work, https://regex101.com/r/cqkkT1/1

I am trying to find the text within the span, i will eventually remove the span and place the text in the parent p-tag.

in calibre it doesnt find the text, "no matches were found". any help? thanks

When you tested using Regex101, which flavour of Regex did you test with? calibre uses the Python re and regex modules which Regex101 supports in it's flavour menu under Python. However the default flavour (for me at least) on Regex101 was PCRE2 (PHP) flavour which I kept since that is what Sigil uses. When I want to test for calibre, I change the flavour to Python.

Simpler regex searches will work under most implementations however even with those, you may find yourself being bitten on the butt.

[lomkiri]

Originally Posted by theducks

Yes!
We recommended tools we had USED regularly for the task and our recommendations are still VALID.

Exactly. Even putting apart all the very serious and crucial problems that LLMs are generating and that I summarily spoke about, there is the fundamental problem of the quality of the code.

Speaking of this precise case : Diap is a good and experienced dev, he knows what is doing –and why– when he writes a group of instructions. He knows which part of the text he will (or should) reach with the code. And afterwards, he knows where may be the flaws and which situations to test.

A machine doesn't think (the word AI is a lie, it's just a slogan for those things so people think it's magic, the real name is LLM, Large Language Model), it just makes hundreds of thousands iterations with all the code it has stolen in the internet (mainly in Github and Cie). We have no guaranty that the code is correct, even if it has been tested by humans.

An LLM-generated plugin may silently corrupt an epub, or may eat part of sentences or even some paragraphs without any warning, so in theory I should display and check all the changes, at each time I use it, which is not acceptable.

When you vibe-codes, you have no control over the code.

LLM may be used to create some small parts of simple code that the dev can understand and check and that he will integrate in hiis code. In other words, it may be used as an assistant for small parts of code. The dev is still the master of the code. Even if I'm not in favor of this use for ethical reasons.

LLM SHALL NOT be used to create entire softs or plugins, and even more if the author is not a dev.(I don't know if the creator of the Otter plugin is a dev or not). And mainly if there is a risk of data corruption during the use of this soft or plugin.