Gutenberg books: Kobo doesn't remember position

[franklekens]

I've complained about Gutenberg books here before. Then, the problem was the formatting.

I don't seem to have any trouble with that anymore. But I now have a few Gutenberg epubs for which two Kobo readers (a Forma and an Aura One) refuse to remember my reading position. So each time I open the book, I am sent back to the front page. Or to the cover, actually.

This one is a case in point:
http://www.gutenberg.org/ebooks/15466

I didn't really edit these manually. The only thing I did was have Calibre look for a cover to add to this epub. I do this as a matter of course now, since it is the simplest remedy for another problem with the Gutenberg epub formatting, explained in this thread:
https://www.mobileread.com/forums/sh...d.php?t=279948

Basically, just having Calibre add a cover to the file seems to solve the problem of the file's text having practically no margins.

But I get the impression that this cover added by Calibre creates the problem of Kobo not remembering my reading position. Because I just tried it out: the same file with added cover will display fine, but always open on the first page.
The Gutenberg file without added cover will remember my reading position -- except I'll never read it because the text runs all the way up to the margin of the screen and sometimes even beyond it with half a letter or so.

Sigh. I guess there's no easy way around this then? Except extensive manual fiddling with the epub (which I don't have the time or the inclination to do)?

(I'm not asking because of this particular book per se. But Gutenberg has so many texts that are not easily available elsewhere, or not for free.)

[stumped]

you need to examine the books formatting, for clues as to why that is
e.g. maybe it it just one humongous paragraph ?
maybe just doing an epub to epub conversion in calibre will standardise it enough that your reading position is then remembered,
if not, then you will need some basic editing skills, to make the code resemble other books that do work ok

[stumped]

update, so I took a quick look at the epub , no images version of that book
each para begins with this construct, where the id values are all different
p id="id00416"

I would use find and replace to take those about and to substitute class="bodytext",
then define bodytext in the CSS to have your preferred margins, indents....


so each paragraph would then look like <p class="bodytext"> some text <\p>

NB if you run a calibre epub to epub conversion, calibre will add class statements for you, which you can then redefine

i strongly suspect it the repetitive ID= that cause the issue, because most books do not have those in every paragraph; and also that it only affects/confuses the kobo . that other readers would remember reading position OK.

[Quoth]

I've had problems with formatting of ePubs from Gutenberg. Often no margin and the GUI margin control ineffective.
So now I only download "Kindle with images", the mobi version, and convert to epub with Calibre. Never any problem.
Sometimes I read them on a Kindle.
I've also had weird format issues with the Driver "on the fly" silent conversion to kepub (First Line Indent half correct value visually), so I disabled kepub in the driver. Now epub on the Kobo H2O original match Sony, Nook, Binatone and App based epub readers.

[franklekens]

The mobi version! What a good idea. I'll try doing that in the future.

Thanks for the other tips too. Maybe I'll have a look, but I try to a have as little to do with manual editing of the formatting.

[rashkae]

*How*, exactly, are you adding the cover to the epub? Better yet, why not provide the actual file you're having trouble with.

Another idea, if you have an problematic epub, is to try an Epub to Epub conversion in Calibre. Calibre has built up a vast repertoir of common pitfalls it 'fixes' on the fly when doing this. If course, GIBO still applies, so results may vary.

[stumped]

Originally Posted by rashkae

*How*, exactly, are you adding the cover to the epub? Better yet, why not provide the actual file you're having trouble with.

Another idea, if you have an problematic epub, is to try an Epub to Epub conversion in Calibre. Calibre has built up a vast repertoir of common pitfalls it 'fixes' on the fly when doing this. If course, GIBO still applies, so results may vary.

1 He gave us a download link for the file at Gutenberg. I examined it and commented on the structure. Others have suggested workarounds, like using the. MObi version
2 As for "another idea". I already suggested that, a few posts back

[Quoth]

Originally Posted by rashkae

Another idea, if you have an problematic epub, is to try an Epub to Epub conversion in Calibre. Calibre has built up a vast repertoir of common pitfalls it 'fixes' on the fly when doing this. If course, GIBO still applies, so results may vary.

I tried that on problematic Gutenberg ePubs. Didn't work. I'm sure it would if you added the appropriate rules.

It was simpler just to have sane ePub conversion defaults and convert the mobi from Gutenberg, as I need it anyway.

If that is problematic (rare), I can export and reformat in LibreOffice rather than peering at CSS. I've had to do it with one Kindle book bought from Amazon. It was poor on a Kindle too. It also had proof errors suggesting it was OCR from a scan. I forget title, but it was one of the books Jane Austin mentions in Northanger Abbey. Generally Archive.ie books are not proofed and some cheaply done old books sold in Kindle format on Amazon are worse proofed than the same title on Gutenberg.

Contrary to the belief of companies using DRM and the DMCA (USA only), copyright DOES expire, though Disney & Friends keep getting it extended.

Note that the words can be out of copyright, but a book may have new formatting (rather than being a facsimile) and/or new illustrations, or new proof corrections, making it a fresh copyright derived work.

[PeterT]

FYI I downloaded the version "with images" and am seeing no issues with location not being remembered.

I've tried both just exitting the bóok and going back into it, and exiting and powering off and back on.

This is on a Clara HD running a slightly newer f/w version.

Looking at the book with Sigil the only thing I wondered about was the name of the various internal files (for instance @public@vhost@g@gutenberg@html@files@15466@15466-8-0.txt.html). I realize this is a legal name but still....

*Edit* I am loading this as an Epub not kepub.

*Edit2* Bother I didn't realize your issue came AFTER adding a cover Any chance you could just attach your editted version of the book?

*Edit3* If you are converting to kepub also attach a copy of the book from your Kobo.

Sent from my Nexus 7 using Tapatalk

[GeoffR]

The fact that you have the "no-margin" problem if you don't add a cover means that you must be loading the book as a KePub (this problem is caused by the KePub reader treating the first file as a cover image and displaying it full-screen.)

However the book (like most Gutenberg books) also has a TOC structure that will cause navigation problems in the KePub reader -- specifically it has 12 ncx entries but only 3 spine entries.

I think books with this sort of problem work better if you load them as ordinary ePubs, rather than as KePubs.

(Edit: Also, adding a cover to a book that doesn't have one, or converting a book with this sort of TOC structure using Calibre will change the internal file structure, so be sure to delete any old version of the book from the device using the device menus before loading the new version. Using Calibre to load a new version over top of old version can cause the sort of problems you describe.)

[rashkae]

Originally Posted by stumped

1 He gave us a download link for the file at Gutenberg. I examined it and commented on the structure. Others have suggested workarounds, like using the. MObi versionp

Per the first post, the problem with not remembering the reading position only happens *after* he added the covers. That leaves us trying to guess what exactly it is about the final file that is causing the issue, or trying to replicate it by assuming what he's doing.

Originally Posted by stumped

2 As for "another idea". I already suggested that, a few posts back

So you did, sorry

[Quoth]

I've never ever had a problem related to covers added via Calibre.
I've used the Calibre Cover Creator, browsed images (my own), the Calibre downloader and used Gimp via Calibre GUI to edit a cover.
The normal way is either to browse or download via the interface in Edit Metadata for a book, or use the Cover Creator plugin from main GUI. CC main use is to quickly add text to a browsed image or cropped downloaded image. It's better to use Gimp and have layers, save as XCF and export a suitable size JPEG. Then text can be edited or image changed without changing text.

[davidfor]

Originally Posted by rashkae

Per the first post, the problem with not remembering the reading position only happens *after* he added the covers. That leaves us trying to guess what exactly it is about the final file that is causing the issue, or trying to replicate it by assuming what he's doing.

I think it's a combination of the internal file names, adding the cover and sending as a kepub. I think the NCX file gets rewritten along the way and the "bad" characters are escaped. With the way Kobo stores and uses the ToC in the database, and the reading position, I think there is a conflict between where characters have been escaped and where they haven't been.

But, I'm having problems reproducing it. I had if failed as described once, fiddled and fixed. But, I cannot revert and have it break. But, I am sure that cleaning/simplifying the internal file names will fix it.