epub -> azw3 conversion resulting in blank spaces where text belongs

genoasalami · 04-01-2017, 12:54 PM

Hi folks- very longtime lurker. I love the software. Kudos to Kovid Goyal for keeping it going for this long, and for keeping it free. It's thanks to him I get to use the best ereader hardware out there without any of the compromises or sacrifices amazon would otherwise force on us. So, I'm very grateful.

My issue:

Recently, using version 2.82 on OS X, I have been converting ePubs into AZW3 format to be used on my Oasis. Every so often, the bottom half of a page will be blank. If I advance the page, a whole paragraph or more of text will have been skipped. However, if I then go backwards one page, the missing text appears, and in the correct place.

Sort of a strange problem, but I can't figure it out. Appreciate any thoughts.

Thanks all

Edit: I found the below discussion of this issue, but no solution. (Note, I am running Running 5.8.7.0.1 and still have the issue described below.)

Quote:

Originally Posted by odamizu

After updating to 5.8.5.0.2, my Oasis has starting missing paragraphs. A screen will show the top half of the page, with the bottom half blank. Going forward one screen then back again will restore the full page. This happens sporadically.

At first I thought it was a problem with the ebook (epub converted to azw3 using KindleGen), but then it happened with two more ebooks. Looking at the html in each book reveals nothing unusual.

Restarting my Oasis will temporarily fix it, but then it will start happening again with greater and greater frequency until I restart again.

I have just reset my Oasis and will see if that helps, but wondering if anyone else is experiencing this. This has never happened to me before on my Oasis or Voyage. Just started with this latest update.

Thanks for any insights.

Quote:

Originally Posted by NullNix

I see the same thing. It's clearly a bug. (For me, it happens intermittently and is always refreshed by switching pages, with no increasing frequency over time.)

Quote:

Originally Posted by septentrion

It happens with my Voyage, not often enough to bother me. I hadn't realised ti might be caused by the latest software update.

Quote:

Originally Posted by NullNix

I never saw it before 5.8.5.*, so I'd bet it's a regression somewhere in there, indeed.

https://www.mobileread.com/forums/sh...0&postcount=22

theducks · 04-01-2017, 01:32 PM

Crazy text that gets interpreted as (coding ) a HTML comment??

Except I would not expect that in a 'conversion' . Find the code (in the EPUB) where things go bad , then open the AZW3 in the editor and locate the same spot. Anything odd (I would expect the same code)?

genoasalami · 04-03-2017, 08:22 PM

Quote:

Originally Posted by theducks

Crazy text that gets interpreted as (coding ) a HTML comment??

Except I would not expect that in a 'conversion' . Find the code (in the EPUB) where things go bad , then open the AZW3 in the editor and locate the same spot. Anything odd (I would expect the same code)?

Here are some examples. Copyrighted text removed.

See post 5

theducks · 04-03-2017, 09:02 PM

There appear that there are Un balanced (curly quotes) in the Paragraph " "

That must be what is driving the parser off the end

BTW I don't think you need to obscure text in that small of a snippets (2 or 3 lines)

BetterRed · 04-03-2017, 09:08 PM

Quote:

Originally Posted by genoasalami

Here are some examples . . .

Moderator Notice
Please take a few moments to read our Posting Guidelines, in particular the part about the maximum image size allowed (600x600 pixels). Images that violate those guidelines may be deleted.

It is every member's responsibility to read our Posting Guidelines and follow them.

© Dr. Drib

If you use the boards attachment facility (use the paper click icon in Advanced mode, to upload and insert in-line) it will post thumbnails. See my attachment.

Rather than posting code fragments you could install and use the ScrambleEbook plugin, and attach its output epub.

BR

genoasalami · 04-03-2017, 09:24 PM

Quote:

Originally Posted by theducks

There appear that there are Un balanced (curly quotes) in the Paragraph " "

That must be what is driving the parser off the end

BTW I don't think you need to obscure text in that small of a snippets (2 or 3 lines)

Thinking, then, to just remove all smart punctuation when converting?

theducks · 04-03-2017, 09:35 PM

Quote:

Originally Posted by genoasalami

Thinking, then, to just remove all smart punctuation when converting?

IIRC Removing Smart punctuation, will still leave the unbalanced problem.
Remove only converts to straight

Why not fix 1 or 3 by hand and see if the appear where they were bad before

genoasalami · 04-03-2017, 10:07 PM

Quote:

Originally Posted by theducks

IIRC Removing Smart punctuation, will still leave the unbalanced problem.
Remove only converts to straight

Why not fix 1 or 3 by hand and see if the appear where they were bad before

I'll give it a go. However, I am thinking (based on the edited information in my first post) that this may be a Kindle firmware bug. (sigh...) Thanks for taking the time to help out

AnemicOak · 04-04-2017, 11:40 PM

Does this only happen on Calibre generated AZW3's or also with ones generated with KindleGen/Kindle Previewer?

EDIT: Nevermind. I see odamizu is having the same problem with KindleGen generated files.

kovidgoyal · 04-05-2017, 12:16 AM

This is a long standing issue with kindle firmware. Basically, under some circumstances parts of the text fail to render and you have to go back a page and forward again to work around it.

IIRC this usually happens when an anchor (the destination for a link) is close to/spans an internal text boundary in the MOBI/AZW3 file. MOBI files have the text/markup split up internally into fixed size chunks and the Kindle renderer basically renders one chunk at a time. THere can be issues if yo follow a link/TOC entry that goes to alocation close to the boundary of one of those chunks.

I dont think it is anything in the HTML markup itself that causes this -- however there is no way to be sure of that since we cant see the source code of the kindle renderer

odamizu · 04-09-2017, 10:32 PM

That's interesting. I used ;dm to download error logs and it seems the problem lies with this error message:

framework: [41m[PRE_RENDER] fillUpCache(533): ERROR: NO page cache list or current page is a special page that shouldn't use cache[0m

Any insights into what "No page cache list" might mean? (There's nothing special about the page, just plain straight text, so I don't think the second half of the error message applies.)

More details here

Thanks!

kovidgoyal · 04-09-2017, 10:47 PM

Probably it means that the renderer has not pre-rendered one of the chunks that are displayed on the current page, which is presumably why it shows up as blank.

odamizu · 04-09-2017, 11:54 PM

Yes, that makes sense. Given this error message, do you think there's anything in my HTML that might be causing this or that I should look for? Or do you still think it's related to the anchors in relation to internal text boundaries, as you say above?

(I've scrutinized the HTML and there doesn't seem to be anything unusual about it. Also, the source EPUB validates.)

Thank you!

kovidgoyal · 04-10-2017, 12:02 AM

Quoting myself

Quote:

Originally Posted by kovidgoyal

I dont think it is anything in the HTML markup itself that causes this -- however there is no way to be sure of that since we cant see the source code of the kindle renderer

odamizu · 04-10-2017, 01:50 PM

Okay, thank you.

I was hoping the error message ("No page cache list") would suggest an HTML fix, but it appears no such luck <sigh>

04-04-2017, 11:40 PM	#9
AnemicOak Bookaholic Posts: 14,391 Karma: 54969924 Join Date: Oct 2007 Location: Minnesota Device: iPad Mini 4, AuraHD, iPhone XR +	Does this only happen on Calibre generated AZW3's or also with ones generated with KindleGen/Kindle Previewer? EDIT: Nevermind. I see odamizu is having the same problem with KindleGen generated files. Last edited by AnemicOak; 04-04-2017 at 11:43 PM.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
pdf to epub - blank lines between text ?	Vortex	Conversion	2	06-13-2014 01:03 PM
Conversion ePub -> azw3, text centered	apeiron75	Calibre	0	05-18-2013 12:28 PM
Conversion from epub to azw3	Joy736	Conversion	12	01-01-2013 11:00 AM
[Old Thread] PDF to Epub conversion (spaces between letters)	mastroalex	Conversion	8	10-09-2011 10:39 PM
Paragraph spaces in ePub to Mobi conversion disrupts indent formatting	markpearl	Conversion	34	09-21-2011 02:42 PM

04-01-2017, 01:32 PM	#2
theducks Well trained by Cats Posts: 30,094 Karma: 57259780 Join Date: Aug 2009 Location: The Central Coast of California Device: Kobo Libra2,Kobo Aura2v1, K4NT(Fixed: New Bat.), Galaxy Tab A	Crazy text that gets interpreted as (coding ) a HTML comment?? Except I would not expect that in a 'conversion' . Find the code (in the EPUB) where things go bad , then open the AZW3 in the editor and locate the same spot. Anything odd (I would expect the same code)?

04-03-2017, 09:02 PM	#4
theducks Well trained by Cats Posts: 30,094 Karma: 57259780 Join Date: Aug 2009 Location: The Central Coast of California Device: Kobo Libra2,Kobo Aura2v1, K4NT(Fixed: New Bat.), Galaxy Tab A	There appear that there are Un balanced (curly quotes) in the Paragraph " " That must be what is driving the parser off the end BTW I don't think you need to obscure text in that small of a snippets (2 or 3 lines)

04-05-2017, 12:16 AM	#10
kovidgoyal creator of calibre Posts: 44,138 Karma: 22670164 Join Date: Oct 2006 Location: Mumbai, India Device: Various	This is a long standing issue with kindle firmware. Basically, under some circumstances parts of the text fail to render and you have to go back a page and forward again to work around it. IIRC this usually happens when an anchor (the destination for a link) is close to/spans an internal text boundary in the MOBI/AZW3 file. MOBI files have the text/markup split up internally into fixed size chunks and the Kindle renderer basically renders one chunk at a time. THere can be issues if yo follow a link/TOC entry that goes to alocation close to the boundary of one of those chunks. I dont think it is anything in the HTML markup itself that causes this -- however there is no way to be sure of that since we cant see the source code of the kindle renderer

04-09-2017, 10:32 PM	#11
odamizu just an egg Posts: 1,609 Karma: 4924808 Join Date: Mar 2015 Device: Kindle, iOS	That's interesting. I used ;dm to download error logs and it seems the problem lies with this error message: framework: [41m[PRE_RENDER] fillUpCache(533): ERROR: NO page cache list or current page is a special page that shouldn't use cache[0m Any insights into what "No page cache list" might mean? (There's nothing special about the page, just plain straight text, so I don't think the second half of the error message applies.) More details here Thanks!

04-09-2017, 10:47 PM	#12
kovidgoyal creator of calibre Posts: 44,138 Karma: 22670164 Join Date: Oct 2006 Location: Mumbai, India Device: Various	Probably it means that the renderer has not pre-rendered one of the chunks that are displayed on the current page, which is presumably why it shows up as blank.

04-09-2017, 11:54 PM	#13
odamizu just an egg Posts: 1,609 Karma: 4924808 Join Date: Mar 2015 Device: Kindle, iOS	Yes, that makes sense. Given this error message, do you think there's anything in my HTML that might be causing this or that I should look for? Or do you still think it's related to the anchors in relation to internal text boundaries, as you say above? (I've scrutinized the HTML and there doesn't seem to be anything unusual about it. Also, the source EPUB validates.) Thank you!

04-10-2017, 01:50 PM	#15
odamizu just an egg Posts: 1,609 Karma: 4924808 Join Date: Mar 2015 Device: Kindle, iOS	Okay, thank you. I was hoping the error message ("No page cache list") would suggest an HTML fix, but it appears no such luck <sigh>

Advert

Advert