Very disappointed with ePub to mobi conversion

[ldolse]

I think your best approach is to do this:

1. Convert to Mobi using Calibre
2. Use the utilities on the mobi subforum - I believe it's mobiunpack, to convert the mobi to Mobipocket Creator's native format.
3. Do your final edits using Mobipocket Creator.

You should be able to get blockquotes working if you follow the steps I outlined before, they're working for me. A bugfix just went in for scenarios where the 'ignore margins' setting is enabled, but it works fine without ignore margins today.

Superscript may be a bug, I don't know how you've defined your css for that, but in general mobi doesn't appear to allow much in the way of formatting superscript. Check this bug for more details:
https://bugs.launchpad.net/calibre/+bug/758667

If you've defined it in epub css matching those parameters then a fix will likely be in the next release.

The font issue is most likely a bug, but we'll need the details on how you defined it in the CSS.

[AlexBell]

Originally Posted by ldolse

I think your best approach is to do this:

1. Convert to Mobi using Calibre
2. Use the utilities on the mobi subforum - I believe it's mobiunpack, to convert the mobi to Mobipocket Creator's native format.
3. Do your final edits using Mobipocket Creator.

You should be able to get blockquotes working if you follow the steps I outlined before, they're working for me. A bugfix just went in for scenarios where the 'ignore margins' setting is enabled, but it works fine without ignore margins today.

Superscript may be a bug, I don't know how you've defined your css for that, but in general mobi doesn't appear to allow much in the way of formatting superscript. Check this bug for more details:
https://bugs.launchpad.net/calibre/+bug/758667

If you've defined it in epub css matching those parameters then a fix will likely be in the next release.

The font issue is most likely a bug, but we'll need the details on how you defined it in the CSS.

Thanks so much for your continuing help, Idolse. I really appreciate it, and apologise for my sometimes crabby responses.

I've just tried a suggestion from another thread, which was to convert the original Word doc to rtf and then convert that to mobi with calibre. Didn't work - the left margin of the blockquotes was far too big and the there was no right indent. And although the end note numbers were not pushed too high they were all [1] and didn't work anyway.

So I'll certainly look at your three step suggestion above. But shouldn't there be a fourth step - to pack the file back into mobi format?

So far as blockquotes are concerned: I have the text within <p></p> tags. Will the process you've mention remove those tags, or will I have to do it manually? And if I remove the tags how do I separate the text into paragraphs? Some of the quotes in the book go on for a page or two with many paragraphs.

No, I don't think the 'superscript' problem is a bug for calibre, apart from not using my CSS. I have this class in the stylesheet:
.fn { font-size: 67%; vertical-align: text-top; } /* With thanks to Weatherwax */

and this markup in the XHTML:
<span class="fn" id="C1F1"><a href="Endmatter.html#C1N1">[1]</a></span>

<p id="C1N1"><a href="Ch01.html#C1F1">[1]</a> Text of endnote</p>

C1F1 is the id of the number in the text of the book, and C1N1 is the location of the end note. The user will see a number with a grey background in the text, and when he or she selects that it brings up the endnote. The same number has a grey background in the endnote file, and when it is selected it takes the reader back to the text. I know it's cumbersome, and would be glad to hear of a simpler method.

I'm really not fussed about the font, except as another example of calibre using its own rules and not my stylesheet. But as I wrote in another post perhaps I'm asking calibre to do something it wasn't designed for.

Regards, Alex

[DoctorOhh]

Thought these two links (1, 2) might give you further insight into mobi.

[ldolse]

Originally Posted by AlexBell

So I'll certainly look at your three step suggestion above. But shouldn't there be a fourth step - to pack the file back into mobi format?

True, that's a fourth step, but I think it's one Mobipocket Creator does more or less automatically upon save.

Originally Posted by AlexBell

So far as blockquotes are concerned: I have the text within <p></p> tags. Will the process you've mention remove those tags, or will I have to do it manually? And if I remove the tags how do I separate the text into paragraphs? Some of the quotes in the book go on for a page or two with many paragraphs.

I just re-tested the blockquotes - I'm starting to think this must be a bug in Calibre related to the css styles you're using. I just tried using both <p> and <div> and both converted identically - and work fine - for me. This is my source:

<body>
  <h1 id="heading_id_2">Blockquote Test</h1>

  <blockquote>
    <div>
      First level indent
    </div>

    <blockquote>
      <div>
        Second level indent
      </div>
    </blockquote>
  </blockquote>

  <p class="sgc-1">Using &lt;p&gt; tags instead:</p>

  <blockquote>
    <p class="noindent">First level indent</p>

    <blockquote>
      <p class="noindent">Second level indent</p>
    </blockquote>
  </blockquote>
</body>

My guess would be that it's somehow related to the css margins you have defined. You'd need to open a bug with a sample file attached to see. Alternatively a fix got submitted for the 'ignore margins' option, which will be in the next release, so you could give that a shot, as it will ignore css margins but preserve existing blockquote nesting.

Originally Posted by AlexBell

No, I don't think the 'superscript' problem is a bug for calibre, apart from not using my CSS. I have this class in the stylesheet:
.fn { font-size: 67%; vertical-align: text-top; } /* With thanks to Weatherwax */

Calibre will automatically turn that into a superscript, so that's definitely not a bug. Unfortunately mobi doesn't allow any fine grained control of positioning and text size for superscript/subscript. The only thing that you could potentially do is wrap the superscript in <small></small> tags in mobipocket Creator for a small font size reduction - Calibre doesn't do this today.

Originally Posted by AlexBell

I'm really not fussed about the font, except as another example of calibre using its own rules and not my stylesheet. But as I wrote in another post perhaps I'm asking calibre to do something it wasn't designed for.

The only reason I'm asking about this is there is definitely conversion code that should handle this, so it appears to be a limitation with the way you've defined sans-serif in your stylesheet vs. what Calibre is expecting.

[dwig]

Originally Posted by user_none

EPUB does have a standard and uses XHTML. To be a valid EPUB the HTML should be valid XHTML. ...

Yes, that is well and good for ePub output but validating "HTML" is not validating "XHTML" and, my major point, not all that usefull in creating a file for the purposes of converting to MOBI. The EPUB validation standards don't apply to HTML code written for conversion to MOBI.

[user_none]

Originally Posted by dwig

Yes, that is well and good for ePub output but validating "HTML" is not validating "XHTML"

Originally Posted by AlexBell

I prepare all my ePub ebooks by hand from XHTML 1.1 strict files

Originally Posted by dwig

The EPUB validation standards don't apply to HTML code written for conversion to MOBI.

Yes it does. When converting using calibre and using an EPUB as the input format validation does matter. calibre is forgiving to an extent but if it has to guess you are hoping calibre guesses right.

[AlexBell]

Thanks again for all your help and suggestions.

The author for whom I prepared the ePub version of his text book wants the mobi version to appear much the same as the print version - indents and all. So far as I can tell the only way I could do that is to prepare another version which will work around mobi's limitations. After completing a version based on valid XHTML 1.1 I'm not prepared to do the book all over again based on invalid HTML 3.2. So I'm going to suggest to the author that he sell the pdf version to those students who use Kindles.

I have downloaded the Amazon Kindle publishing guide just to have as a reference.

Regards, Alex

[ldolse]

PDF is a valid route, but remember if you're planning to publish pdf for the Kindle you need to publish to the Kindle's physical pages size, otherwise any attempts to actually read the document would be worse than a poor mobi conversion. You'll probably want one pdf for the regular Kindle and another for the DX.

Several fixes are now going into the next release based on this thread which will improve mobi conversion somewhat, but I think most of your concerns relate to inherent issues with the mobi format itself, and less with Calibre's conversion.

[AlexBell]

Originally Posted by ldolse

Several fixes are now going into the next release based on this thread which will improve mobi conversion somewhat, but I think most of your concerns relate to inherent issues with the mobi format itself, and less with Calibre's conversion.

Thanks again. I'm sure you're right.

Regards, Alex