How to create a clean ePub? (total beginner)

[Fstab]

Hello,

I want to create an ePub, and if possible, with a good method to have a well-made and clean ePub.
So I have started to look for a guide and have found a lot of tools and methods with Sigil, LibreOffice, writer2epub, writer2xhtml, Amanuensis, and others.

But I am lost in the middle of all this, because some tools are old (Amanuensis 2011, writer2epub 2014), some are abandoned (writer2xhtml will not have new version), some are maybe not good enough (apparently, LibreOffice did not made clean code in version 6.0, is it still the case with the 6.3 version?), and I don't know where to start.

What do you think about the LibreOffice export with the 6.3 version? Is it clean compared to 6.0 and 6.1?

Do you think I can use old tools or are they some improvements in the past few years?

Maybe Sigil is the best way, but I was reading on the website it was not updates for years.

My main computer is on Windows, but I use also OSX and several GNU-Linux, mainly Debian-based. I want to create mostly text only ebooks (plus cover), but also some with a few pictures in them.

All guides I have seen are years old, I don't know if they are still useful.

Thank you for your help!

[pdurrant]

Originally Posted by Fstab

Maybe Sigil is the best way, but I was reading on the website it was not updates for years.

The most recent release of Sigil was just 19 days ago on 10th August 2019.

I'd recommend Sigil for creating the ePub itself. Creating good quality contents for the ePub is a lot harder.

[Fstab]

Thank you pdurrant.
I was writing about the history of Sigil.
Ok, thanks for your recommendation, I will look further into Sigil.
I have read I can import an ODT (LibreOffice) with ODTImport. I don't know if it is the same as creating the file straight in Sigil.

[pdurrant]

Sigil is an excellent way to create an ePub with valid structure. Creating clean, valid XHTML and CSS files is rather up to you. A simple import of any kind of word processing document is unlikely to produce clean results, unless the author has applied styles incredibly consistently, and even then the CSS is very likely to need adjustments.

[pdurrant]

Sigil was taken over by new developers some time ago. It now has a forum here on MobileRead, and a github repository.

[Fstab]

Ok, thank you for the precision.

[Notjohn]

No word processor makes clean html, though I'm told that Atlantis Word Processor ($40 or so) is better than most. Personally, I finish up writing my books in Word 2007, then I run the Word doc through Word2CleanHtml online, which produces gloriously clean html to open in Sigil and there massage. There are many other ways to achieve the same end, including a Word plug-in by Toxaris who regularly posts here.

[JSWolf]

If the formatting is not complex, then exporting the text from the word processor as text could be a lot easier to work with then the mess you can sometimes get from a word processor.

I've seen HTML Word documents where a text export and then message it from there in Sigil would be so much easier.

[AliceWonder]

I recently started the same journey. I am not using software, other than a text editor. I may go that route but I want to manually do things first to help me understand it better.

Understanding how to properly do metadata within the .opf file has been my personal most time consuming task, especially the Dublin Core stuff. Have to use one tag to define a creator and that creator has to have a unique ID and then you need a second meta tag to refine the dc:creator tag and associate the unique ID with the type of creator. e.g.

    <dc:creator id="editor">Alice Wonder</dc:creator>
    <dc:creator id="author0">Alice Wonder</dc:creator>
    <dc:creator id="author1">Enchantrix Brighton</dc:creator>
    <!-- princess andi illustration 1 -->
    <dc:creator id="illustrator01">Annabelle Brito</dc:creator>
    <!-- french maid with dusting brush -->
    <dc:creator id="illustrator02">Robert DeJesus</dc:creator>
    <!-- Misaki Ayuzawa portrait -->
    <dc:creator id="illustrator03">Kayleigh Vaughn</dc:creator>
    <!-- Rem Portrait -->
    <dc:creator id="illustrator04">Professor Purrv</dc:creator>
    <!-- princess andi illustration 2 -->
    <dc:creator id="illustrator05">Placeholder</dc:creator>
    <!-- back page illustration -->
    <dc:creator id="illustrator99">Gustavo Rezende</dc:creator>
      <!-- http://www.loc.gov/marc/relators/relacode.html -->
      <meta refines="#editor" property="role" scheme="marc:relators">edt</meta>
      <meta refines="#author0" property="role" scheme="marc:relators">aut</meta>
      <meta refines="#author1" property="role" scheme="marc:relators">ctb</meta>
      <meta refines="#illustrator01" property="role" scheme="marc:relators">ill</meta>
      <meta refines="#illustrator02" property="role" scheme="marc:relators">ill</meta>
      <meta refines="#illustrator03" property="role" scheme="marc:relators">ill</meta>
      <meta refines="#illustrator04" property="role" scheme="marc:relators">ill</meta>
      <meta refines="#illustrator05" property="role" scheme="marc:relators">ill</meta>
      <meta refines="#illustrator99" property="role" scheme="marc:relators">ill</meta>

I made comments to myself so I'd know who was what, etc. and so I could always find the URL with the marc:relators that are needed to refine the dc:creator tags.

Using the ePubCheck validator has GREATLY helped. It would give me errors and I could search google for the error string and 90% of the time find an explanation for the error and then have a better understanding of how ePub works.

I do use the editor that came with Calibre to run the ePubCheck validator (as a plugin) and I may at some point use that editor instead of just a text editor once I better understand ePub.

But anyway, understand the metadata was the biggest hurdle for me, the rest seems fairly easy to grasp.

[JSWolf]

The first thing to learn is HTML/CSS. Then from there you can learn the technical stuff to do with ePub.

[AliceWonder]

Originally Posted by JSWolf

The first thing to learn is HTML/CSS. Then from there you can learn the technical stuff to do with ePub.

Good point - one thing many don't get is that there is a difference between XML and HTML and then XHTML muddies the water as it is sometimes served w/ HTML mime type instead of XML mime type.

With HTML certain tags MUST have a child node, even if the child node is empty - examples are a and script and style. That means they can not be self-closing in HTML.

With XML any node can be self closing and be legal. Not all make sense to be, but they can be.

With XHTML though - those tags where HTML requires a child node can only be self closing IF served with an XML mime type. WordPress for example does this, it uses XHTML but is served with HTML mime type so the HTML rules apply, not the XML rules.

But with ePub since an XML mime type is specified in the .opf, the XML rules apply, any tag can be self closing so you don't have to use a separate closing tag unless it has a child node (including text) and some tags that don't have to be closed in HTML (like br and img and hr) do have to be self closing in ePub.

Also - with HTML tags are not case sensitive but with XML they are, so with ePub since XML mime type is specified, only use lower case tags for the XHTML and I think the .opf also only ever uses lower case tags.

[lumpynose]

Originally Posted by JSWolf

The first thing to learn is HTML/CSS. Then from there you can learn the technical stuff to do with ePub.

You have to be careful though, most if not all web sites or books about HTML/CSS are targeted at making web pages and include lots of stuff that won't work with an epub. And that's not including all the javascript stuff they probably talk about.

I would love to find a web site about making epubs, epub2 vs epub3, what to watch out for for the different ereaders, and making good epubs that convert well into kindle formats, etc.

[JSWolf]

Originally Posted by lumpynose

You have to be careful though, most if not all web sites or books about HTML/CSS are targeted at making web pages and include lots of stuff that won't work with an epub. And that's not including all the javascript stuff they probably talk about.

I would love to find a web site about making epubs, epub2 vs epub3, what to watch out for for the different ereaders, and making good epubs that convert well into kindle formats, etc.

Keep it simple. That's what works in most cases. Don't use any extra code in the CSS. Don't leave any unused classes. That's what works in most cases. Don't try to be fancy where you don't need to be.

[Fstab]

Thank you all!

@Notjohn
Ok, good to know there is different tools working. I still prefer to use recent and free tools.

@JSWolf
Seems better to export only the text, but I suppose it is a lot more work for the layout?
I have to make some tests before, because for now I don't know the process.
And I did not know it is better to know HTML and CSS (which I know a little) to make good ePub.

@AliceWonder
Ok, I like text editors like those for coding but I don't know if it takes a lot more time to manage the layout after, because it seems more efficient in a word processor like LibreOffice Writer (or maybe I don't understand well, because I am not fluent in English).

@lumpynose
I would like to see this kind of website myself too! ^^

To begin, I don't want to have some personal fonts, only to use what is provided by the devices. Maybe it will be simple that way?

[Turtle91]

There are a few tutorials here in the MR wiki. There are also many examples in our library that you can use to learn.

I probably would stay away from the pure text method - that is a lot of extra work going back and re-inserting all the formatting tags. JSWolf's comment was "If the formatting is not complex" meaning very few uses of italics, bold, or small-caps, etc. Otherwise you can use Toxaris' Word plugin that does an excellent job of cleaning up all the crud from word....it'll even save as a basic epub that you can open with Sigil and fine-tune.