Don't Understand Error Message: ... but found '"liga" 0'

[enuddleyarbl]

I have almost no knowledge of CSS. I'm using the Editor in Calibre on an epub I own and ran the bugcheck function on it. It's given me 8 errors that are basically the same thing for different places of CSS:

ERROR: Expected (<feature-tag-value> | normal) but found '"liga" 0'.****[stylesheet.css]

That one points to the 6th line of the following (the second occurence of "liga"):

.char-of-normal {
  -webkit-font-feature-settings: "liga" 0;
  -webkit-font-kerning: Normal;
  -webkit-font-variant-ligatures: no-common-ligatures;
  font-family: "Times New Roman", "TimesNewRomanPSMT", "Times", "serif";
  font-feature-settings: "liga" 0;
  font-kerning: Normal;
  font-size: 0.75rem;
  text-decoration: none;
}

Here's the specific line pulled out of the above:

font-feature-settings: "liga" 0;

Following the help link doesn't really tell me much. But, if I search for that line, I come up with pages showing it as valid CSS. Also, 4 lines above it is almost the identical line (except it's prefixed with "-webkit-") and it doesn't have a problem with that.

Can anyone explain why the Editor doesn't like that line? What's it supposed to be doing?

[JSWolf]

What version of the ePub is it? Is it ePub 2 or ePub 3?

[DNSB]

It may be valid CSS for a web browser but is it valid CSS in an ePub2 or ePub3 which use subsets of CSS2 and CSS3 respectively. I ran your snippet through the W3C online validator. For Epub3/CSS 3, it commented about the multiple vendor extensions (the -webkit-font-whatver lines), for ePub2/CSS2.1, it not only complained about the vendor extensions but added 3 errors (2 about settings that don't exist in CSS2.1, one that complained about the use of rem which not valid in CSS2.1. You may want to start using ePub3 if you need to make use of those features. OTOH, anyone who uses Times New Roman in an ebook is a hopeless romantic.

[enuddleyarbl]

Supposedly, the book is in ePub2. It's Roger Zelazny's 1965 book "This Immortal" which I bought in Kindle format from Amazon something like 8 years ago. I'm working with the Calibre converted epub. Again, I'm almost entirely clueless about CSS. I'm trying to improve the way it looks as a learning experiment, but the CSS/HTML just looks awful to me. For instance, it looks like almost every paragraph is part of a <div id=...> statement. Which, in my ignorance, seems to imply it wants to make every paragraph a unique object to, perhaps, be manipulated with Javascript (?). And, I can't make hide nor hair of what it's doing with all the paragraph and character style statements that it strews everywhere.

I'm going to attach a link to a Scrambled version of the book. If anyone has the time, ability and/or inclination to look at it, perhaps you can tell me what's going on with the CSS. For instance, part0005.html has one line of text on it and dozens of commands all down the page. I have no idea why. And, later, in the actual text, you can see how bad the formatting looks because of all the styles stuck in the middle of everything. I'm wondering if the only way to make it "better" would be to delete all the stylesheet stuff and try making it all up myself.

https://drive.google.com/file/d/1Z1e...ew?usp=sharing

[DNSB]

Taking a quick look at the file, it's a mess. If you have the original Amazon copy and it is KF8/azw3, I would suggest using KindleUnpack to recreate the epub file instead of using calibre to convert to epub. If it was mobi, you are kinda stuck with that mess.

Looking at items such as:

      <div id="box1_1201" class="para-normal">
<span class="char-of-normal"><br class="calibre1"/>
</span>
</div>

generating a blank line where a simple <p>& # 160;</p> would have pretty much the same effect.

Not to mention the <br class="calibre1"> being used to break paragraphs which are not present in my copy of the ebook.

[JSWolf]

I don't know who made This Immortal into an eBook, but just looking at the preview on Amazon, it's terrible. It's inconsistent. The line spacing is all over the place.'

It's no wonder the code is rather poor. But yes, I agree to shift to ePub using KindleUnpack and see what sort of code you get.

[enuddleyarbl]

Since I bought this about 8 years ago, I'm guessing the original was mobi. But, I downloaded a fresh copy from Amazon about 10 months ago when I switched over to Kobo and that's an azw3. I'll head on back, get an even fresher version and see what I can find about/with KindleUnpack. Thanks for the help. I'll report back.

[enuddleyarbl]

OK. It took me a while to get myself connected back to Amazon. But, I got a fresh copy of the azw3 and used KindleUnpack to unpack it. Opening that epub (and the azw3) in the Calibre editor shows it to still be the same kind of mess.

I guess I'll do something like noting the styles used in the stylesheet.css, removing most of them, seeing what I can do about those <div id=...> things and then creating and applying a couple of reasonable styles to the text. I wonder if I could find/generate just the text and start with that. It might be easier.

Also, on a semi-related note, if I add the KindleUnpacked version of the epub into Calibre, will it replace the ORIGINAL_EPUB or the EPUB format?

[DNSB]

Originally Posted by DaveLessnau

Also, on a semi-related note, if I add the KindleUnpacked version of the epub into Calibre, will it replace the ORIGINAL_EPUB or the EPUB format?

The EPUB format.

[enuddleyarbl]

Thanks, again.

Maybe I'll play around with Calibre's Convert Books function and see if it can do something programmatically with the azw3. That might give me a better starting point.

[JSWolf]

Originally Posted by DaveLessnau

Thanks, again.

Maybe I'll play around with Calibre's Convert Books function and see if it can do something programmatically with the azw3. That might give me a better starting point.

I don't think there's much you can do. The best solution is to hand edit the code in the Calibre editor. Start off by removing excess CSS (tools menu) and go from there. Also install the epubcheck editor plubin and use both the Calibre ePub checker and epubcheck to find any errors that need to be fixed.

[retiredbiker]

It was some books like this that got me started learning the editor and how these ebooks really work. And how simple they should be. It's a mystery how people even manage to build some ebooks, I can't imagine the tools that create such a mess, or why.

One like this, I'd take the dynamite approach and edit the html and css down to one or two very simple styles, remove all ids and <span>s, and change all those <div>s to <p>s. Great learning exercise if you want it. The other dynamite way is to convert the thing to txt, clean it up using paragraph styles in Writer or Word and convert from there. It's just a plain simple novel, so not much styling is needed.

[JSWolf]

Originally Posted by retiredbiker

It was some books like this that got me started learning the editor and how these ebooks really work. And how simple they should be. It's a mystery how people even manage to build some ebooks, I can't imagine the tools that create such a mess, or why.

One like this, I'd take the dynamite approach and edit the html and css down to one or two very simple styles, remove all ids and <span>s, and change all those <div>s to <p>s. Great learning exercise if you want it. The other dynamite way is to convert the thing to txt, clean it up using paragraph styles in Writer or Word and convert from there. It's just a plain simple novel, so not much styling is needed.

I find it easier to start with the eBook and edit it from there. It's a lot easier then trying to figure out where to put things like bold, italic, small caps, offset text, etc. So converting to a text file should be right out.

[enuddleyarbl]

I've got most of the weird character styles merged and/or deleted. Now, I'm trying to figure out how to get rid of the <div id=...></div> things. I guess if I assume there aren't nested divs within them, I won't have to worry about the </div> tag being for the wrong starting <div> and a regex search/replace would work.

And, after looking more closely at things, along with every paragraph being set as a block with <div id=...>, it looks like they're additionally set as a block with "display: block;" inside the paragraph styles.

[JSWolf]

@DaveLessnau Would you mind shifting from KF8 to ePub using KindleUnpack and then scrambling it and posting so we can see the original code? Thanks.