Missing text on Kobo H20 Edition 2

[aquatic_rent]

Hi all,

I just grabbed an H2O edition 2 yesterday and immediately loaded some epubs to try things out.

I've noticed that one epub is missing chunks of text throughout the book.

I've pulled the epub up in Calibre, where it displays fine, to diagnose. It looks like it happens most often after ellipses, which are encoded in the epub with non-breaking spaces, in a line like:

all those stresses&nbsp;.&nbsp;.&nbsp;.</span></p>

It will just choke at the first non-breaking space and then skip several lines of text, sometimes all the way to the next Chapter heading.

To make this harder to diagnose, it does not happen with all non-breaking spaces, or even all ellipses.

I've tried altering fonts and spacing and margins, turning off the device, and reloading the epub. I even trimmed the epub down to the section that was having problems, to isolate it. It reproduced, but didn't give me any clues about what in the epub was causing the issue.

I tried searching the forums, I found some issues about missing text, but they did not seem related. (In a post on an older Kobo, it seems like they could see the top few pixels of the next line of text then empty space on the device. With this issue, it cuts the text then places the very next letter after cut text next, so you might get two words mashed together without a space.)

Appreciate any recommendations for how I can help diagnose what's going on. (I tried emailing Kobo support, the email address in this forum redirects to a live chat that no one answered... I'll try again during other hours, maybe it was just closed.)

Thanks for any recommendations!

[JSWolf]

RMDSK (ADE) does not like code that has errors. So it could be your code has errors. Run one of the ePub with the problem through Calibre's eBook editor and run the book check on it and fix all the errors. Then run the same ePub through epubcheck and fix any errors. Then load it back on your Kobo and see if it displays correctly.

[aquatic_rent]

Oh thanks, that was the perfect recommendation.

The error checker built into Calibre didn't find the issue, but EPUB-Checker confirmed:

Fatal Error while parsing file 'The entity "nbsp" was referenced, but not declared.'.

Googling that error, this StackOverflow post points out that nbsp is not a recognized XML entity (even though it's about something completely different).

It recommends declaring the entity or using the hex:

& # xA0; or & # 160;

(But without spaces, apparently deleting those on this forum breaks the text, even in a CODE block)

There were other errors, lots of "parsing file 'element "span" not allowed here;"

Too many, and I'm not sure how to fix them. Those weren't marked "FATAL" though. I'll try find and replace on the nbsp's and hope that goes most of the way.

Repairing every error could get really time consuming... really hope this is find and replace.

(Maybe this was all because I converted from .mobi?)

Thanks!

StackOverflow link: https://stackoverflow.com/questions/...erenc#13012488

[davidfor]

Originally Posted by aquatic_rent

Hi all,

I just grabbed an H2O edition 2 yesterday and immediately loaded some epubs to try things out.

I've noticed that one epub is missing chunks of text throughout the book.

I've pulled the epub up in Calibre, where it displays fine, to diagnose. It looks like it happens most often after ellipses, which are encoded in the epub with non-breaking spaces, in a line like:

all those stresses&nbsp;.&nbsp;.&nbsp;.</span></p>

It will just choke at the first non-breaking space and then skip several lines of text, sometimes all the way to the next Chapter heading.

That code is not a problem. I have plenty of books with exactly that in it and have not seen anything missing.

To make this harder to diagnose, it does not happen with all non-breaking spaces, or even all ellipses.

I've tried altering fonts and spacing and margins, turning off the device, and reloading the epub. I even trimmed the epub down to the section that was having problems, to isolate it. It reproduced, but didn't give me any clues about what in the epub was causing the issue.

I tried searching the forums, I found some issues about missing text, but they did not seem related. (In a post on an older Kobo, it seems like they could see the top few pixels of the next line of text then empty space on the device. With this issue, it cuts the text then places the very next letter after cut text next, so you might get two words mashed together without a space.)

Appreciate any recommendations for how I can help diagnose what's going on. (I tried emailing Kobo support, the email address in this forum redirects to a live chat that no one answered... I'll try again during other hours, maybe it was just closed.)

As Jon said, it is probably an error in the code. The most likely culprit is incorrect nesting of tags. I've seen this problem with code that looked like:

<p>So I sang to her...</p>
<p><span class="green">First lyric line</p>
<p>Second lyric line</p>
<p>Third lyric line</span></p>
<p>She seemed to like it...</p>

When I read this on the Kobo, everything after the last line of the song was missing. The calibre viewer showed it all (and gave a very different ending to the night than I was expecting). I had to fix the code in the calibre editor to see it on the Kobo.

If you open the book in the calibre editor, run the check function (the little bug in the toolbar). It will show any errors like this for you to fix.

Edit: To slow!

[GeoffR]

&nbsp and other such entities are not allowed in epubs unless you have a suitable DOCTYPE declared at the top of the file. Here is an example of the beginning of a .xhtml file that uses &nbsp in the text and passes epubcheck:

<?xml version="1.0" encoding="utf-8" standalone="no"?>
<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.1//EN"
  "http://www.w3.org/TR/xhtml11/DTD/xhtml11.dtd">
<html xmlns="http://www.w3.org/1999/xhtml">

[aquatic_rent]

Thanks for all your suggestions and help.

For any others that hit this issue:

I did not have errors with tags, calibre's error check didn't show any errors even when I still had this issue. Tag nesting still worth checking to rule things out, could easily cause problems for some.

I'm guessing davidfor's files with nbsp worked because of correct headers.

Some of my embedded html files in the epub had a full header with a DOCTYPE, some didn't, which explains why this behaved inconsistently throughout the book. No idea why my headers were only half correct.

FIX: I used find and replace in Calibre's editor for all instances of nbsp to #160. (You may want a leading ampersand a/o trailing semicolon to avoid false positives.) That was fast, and seems to have fixed it. Double checking the headers of all 30 or so files would probably be the better school solution, but the find/replace was very fast, so if someone else has this issue, I'd try it first.

Thanks again to everyone who weighed in!

(Now how to mark this thread resolved...)

[JSWolf]

You should still check with epubcheck. It catches errors that Calibre doesn't.

[aquatic_rent]

Originally Posted by JSWolf

You should still check with epubcheck. It catches errors that Calibre doesn't.

Oh definitely that too. I tried to mention before, your suggestion was how I confirmed nbsp was causing what epubcheck alarmingly called "FATAL ERRORS," so this is a great case study for epubcheck finding things calibre doesn't. Thanks for the original recommendation there.