Suggestion for Improving the "Insert Special Character" Feature

[KevinH]

Great! I will start trying to clean up the code, add version attribute, and integrate it into Sigil itself.

And yes in cvrt.py when label="" I should not have bothered to include that attribute to save space. Good point.

Nice work!

[KevinH]

Okay, I have done a bit of code cleanup and reintegrated the new SelectCharacter code and new FlowLayoutCode into Sigil. Tested it and it seems to work just fine. Pushed it to master in my personal repo along with all the other changes we have made so far.

https://github.com/kevinhendricks/Sigil.git

So please if you build your own give it a go and if okay, we can then start tweaking it for further improvements.

Once it looks good, I will push it to Sigil-Ebook Sigil master.

Thanks!

[BeckyEbook]

I built Sigil and… it just works!


I saw in the commit that if a version isn’t specified, we assume it’s version 1.0.0.
I think the new plugin will add the version when saving, even if it loads a file without one.

And honestly – I don’t see any bugs. It just works, maintaining backward compatibility, though with that one small difference I pointed out in my comment: now, even with the default character layout, subsequent lines fill the entire width of the window.


Of course, I recommend that other testers conduct additional testing.

[KevinH]

Glad to hear it! Yes please add the version="1.0.0" string to the parent tag in your plugin if possible.

I will commit it tomorrow to Sigil-Ebook master for others to begin playing with.

Thanks for your idea and code! Very nicely done!

[DiapDealer]

Good stuff you two!

[DNSB]

To simplify keeping my test builds links up to date, you can find the latest test build for Windows in the Downloading Sigil test builds for Windows thread.

I've attached the sample special_characters.xml which goes into sigil's preferences directory (C:\Users\<usernames>\AppData\Local\sigil-ebook\sigil is the default).

[icearch]

Sorting them is certainly a very uesful feature.

But what's these chinese characters doing in punctuation?

[BeckyEbook]

This is just a text file that showed that grouping works.

The whole idea is that – if you want – you define the groups and the characters that will be in them.
If you are happy with the default characters, simply do nothing.

[philja]

I've just tried DNSB's test compile on my Windows 11 box. I re-organised the sample special_characters.xml file and added a new section with the characters I'm interested in.

It works well - as did the original, of course.

It's all very well having an editable file, but the potential difficulty rests, as always, with the reader interface having fonts with the glyphs required to display the characters.

[KevinH]

Yes, just like any other character used in any ebook, if your current font does not support it, you must embed a font that does or use a different character. Nothing has changed in that regard.

But the default insert special character list is just the chars that make-up the set of html named entities so support for all of those should be good in most decent fonts.

AFWIW, the epub 3.4 spec does include a new change to allow remote fonts now. So using an external url in a font stylesheet font face is now allowed as long as it is properly listed in the manifest and the correct manifest property is set on the xhtml files that use it.

[BeckyEbook]

I just released the SpecialCharacterConfig plugin for early adopters.

[KevinH]

Wonderful. I will give it a test run tomorrow.

[un_pogaz]

That look promising.

I have a few questions:
What happens if a user creates two groups with the same name? And what if the user group is the same as one of the built-in groups?
I suggest that each groups are readed individualy, so in case of dupicate, it just append regardless of the presence of a other one.

By the way, the built-in characters will it use a "special_characters.xml" file in the software directory as source, or will it be hardcoded? The first option will be good for providing an example file that's easy to share.

Finally, I suggest that "special_characters.xml” allow the insertion of character combinations. A very simple use case is that any country flag emoji 🇦🇶 is actually a combination of 2 different Unicode characters. And many emoji can require more.
The simplest solution would be to allow the "hex" attribute to have multiple codepoints separated by spaces.

<cp hex="1F1E6 1F1E6" label="" desc="antarctica flag"/>
<cp hex="1F3F3 FE0F 200D 1F308" label="" desc="rainbow flag"/>

[BeckyEbook]

I checked, and you can have several groups with the same name – in fact, they don't even have to have a name and can look like this: []

As for the combination of characters – I’ll leave that decision up to Kevin. IMVHO, it’s a bit over the top, but maybe I’m just stuck in the insular world of books, not emojis I'm simply looking at the practical aspects of ensuring that emojis display correctly on standard e-readers. But the world is changing, so who knows…

[KevinH]

Many emojis fit in one 32 bit codepoint. Many do not. Some emojis require zero width joiner chars. So let me think about this.

Grapheme Clusters are one of those things that the unicode standard has way overthought and counting those as one codepoint (even though multiple) messes with huge amounts of installed software. Frankly they are graphics/images trying to masquerade as some form of text.

I think if I wrote an epub the uses emojis that are more than one 32 bit unicode codepoint long I would use svg scalable image versions as most platforms, even if the device supports them, they typically includes their own version of the emoji which can be significantly visually different.

Either way these are not true single special characters but instead strings, that clips might be better suited for.

Something to think about for the future, not this release.