book loop back at the end of chapter

[mauriziofantino]

Hello jgoguen,

first of all let me say great job! I opened this issue cause I am having some problem in the majority of the books (not all but with only one at the moment I do not have problem) I upload with your extended module. In practice I am always stuck at the cover page once I open a book. I can read the book by using the toc jumping on a specific chapter, but when I reach its end i loop back to the cover page and again the only way to get rid of it is to use pass through the table of content.

Another thing I wanted to say is related to the problem of the bookmarks. With the last firmware upgrade many people have exerienced the problem of the bookmarks. A possible solution is to set the margin size to zero. Indeed all the original kepub I have taken from my kobo has the margin size set to zero.

Another comment is on the kobo tags which are added. I have noticed that your plugin as well as the suggestions of dsandrei in his blog adds only the hash tag kobo.x.y, while I have noticed that in the original kepub the string is more complex and appear in the form of: . I have no idea if the class="KoboSpan" has any effect on the file.

Thanks for your work and let me know if I can help you somehow!

Maurizio.

[mauriziofantino]

Ok in the meantime I have made some test. When I upload the file something is messed in the xhtml file and for this reason the books presents that problem I was mentioning in my previous comment. I try to post it what is inserted:

As you can see that CDATA part is something which should not be there and I believe it is the piece which bring all my problem.

Regards,

Maurizio.

[giorgio130]

Regarding the "Cover loop" issue, it seems the same problem I've pointed out in another issue, where the chapters go in random order. I don't think it's coming from how the kepub is formatted. This all stems from the code adding entries referring to the single TOC entries in the Kobo.sqlite database. As jgoguen told me to, try to use the "no_database" branch of the code and the thing should be solved. As for the other information, it seems quite interesting...

[jgoguen]

@mauriziofantino Can you try the no-database branch and see if the cover loop issue still exists for you after removing the book from your Kobo device and re-adding it? I will see if I can get one of my purchased books to re-download, I was having problems with that a couple weeks ago. I will most likely need a book for testing; do you have a book purchased from Kobo with the digital restrictions management stripped (or that was sold without DRM) and not changed/converted in any other way that you would be willing to send me? I'll also take the one you modified manually, but I will need a decrypted book as close to its original state as possible. I've added my email address to my public profile page here.

It should be possible to rebuild the kepub structure, but if you don't mind I'd like you to make a separate bug report for that. The cover loop issue I hope will be fixed with bug #5. Ideally I'd like to try to keep one issue per bug report and I think it'll be the bookmark issue for this one.

[jgoguen]

The CDATA blocks usually indicate character data (CDATA means "character data") within a XML tag that must be displayed as-is and not parsed as XML. There are valid reasons why that might be added. I'll take a look at the results of conversions as I work through other bugs, see if I notice it and try to figure out why it's happening.

As a note, if a tag has the string "<![CDATA[Hello, World!]]>" as its text, what you should see displayed on screen is "Hello, World!". Is that the case for you or are you seeing the entire CDATA block as well? Hopefully the text shows properly :)

Joel Goguen Email: jgoguen@jgoguen.ca Web: http://www.jgoguen.ca/

On 2013-01-29, at 15:49, mauriziofantino notifications@github.com wrote:

I am using the no-database branch... and i am having the issue with it. I dunno if the problem is related to this or not but if I check the structure of the file sideloaded on the kobo glo before the tag I the conversion put this strange string: ;![CDATA[

As far as the other information a quick update: I have taken on of the book and rebuild the structure of an origianl kepub. I have kept the page style of the epub as it was, but also added the kobo.css (which keep a proper formatting which allows to have no problem with the bookmarks) and the kobo.js (which has a script for the bookmark evaluation) and manually corrected each header of each xhtml file in the text folder. What I can tell you is that the epub so correct is perfect. Any problem with the bookmark, page style well formatted. No error in the page formatting (by just changing the extension file sometimes it happened to me that to have at the bottom of the page half row in height and then the same row repeted in the following page, problem that now I have not anymore).

I did nothing straordinary, and I can share it with you. The problem is that I did it manually cause I can program in C, but not in python (not yet :-P ) and I have no expertise in calibre plugins.

If you want me I can share the result of the upload, the same book after my header correction and one of the original (free) kepub which I used as example.

Send me an email to maurizio.fantino@gmail.com and I will send to you the books.

Good good work I really see the potential to have an epub almost identical to a kepub on my kobo glo without doing anything than simply press a button!!!

Maurizio

— Reply to this email directly or view it on GitHubhttps://github.com/jgoguen/calibre-kobo-driver/issues/6#issuecomment-12853988.

[mauriziofantino]

@jgoguen Errata Corrige I was not using the no-database branch.

@giorgio130 I have read your post here http://forum.simplicissimus.it/kobo/calibre-driver-per-kobo-con-funzioni-aggiuntive-%28kepub%29/ Good Spread the word.

[jgoguen]

Can you see if this is fixed now for you? This issue is only dealing with the book looping on the cover page, I've created bug #7 for investigating the rebuild of the kepub and adding additional files.

[mauriziofantino]

The problem of the cover loop seems to be fixed. I have tried with 5, 6 books and I had no problem with them. One strange thing is that for all of them but one the cover was correctly uploaded on my kobo. For one I got the blank cover with the title. To get the cover of this last book I had to plug back my kobo and re-sync the meta data, unplug it and here there was the cover. I have tried several times this procedure (also using the modify epub calibre plugin to correct cover and metadata issues) but I got always the same behavior.

Mauri

[mauriziofantino]

I have tried the no-database branch... Should I have used the master one?

[jgoguen]

Can you try with the master branch? That's where the database code is that was causing problems. It's good that you tried with no-database as well, that validates that the no-database branch is working properly.

The examples weren't bad. The covers are based on what's defined in the ePub file itself; if there's a tag with id="cover", I add the properties="cover-image" which is what tells the Kobo to show that as a cover image. There are some books, even professional books direct from major publishers, without any cover specified at all; for these, when using calibre and this plugin, what happens is:

• The ePub file is converted, no cover image is marked since none exists.
• The ePub file and cover image is uploaded to the appropriate places on the Kobo device.
• The database is updated with the proper image ID.
• After ejecting the Kobo device, it processes the ePub file, finds no cover image, displays the generic cover. This causes the image ID field (among others) to be emptied.
• Plugging the Kobo device back into calibre causes calibre to update the metadata, including re-setting the image ID.

[mauriziofantino]

ok I tested the master and it works as the no-database. Which branch shall we use for testing in order to help you with the development?

Thanks for the explanation on the id="cover", it might be boring but I belive this can be easily fxed by editing the book with Sigil or simply by doing the procedure you mentioned to resynch the metadata.

I have also noticed that the original kepub they also have the class KoboSpan before the id="kobo.x.y" in the form . The converted epub they are missing the class part. Do you know the scope of this missing part?

Maurizio

[jgoguen]

I'm pretty sure you actually can't use Sigil for adding the cover-image property. That is actually not valid according to the ePub specification, so Sigil warns you about a validation issue. I'm not sure why Kobo decided to require a non-standard attribute, but readers simply ignore the extra attribute so it's not as bad as it could be. It's only validating applications (like Sigil) that are unhappy.

Adding the koboSpan class is something I'm dealing with in bug #7, since it wasn't relevant to the cover page loop issue. Now that this bug is fixed, I'll start on #7 which will include the koboSpan class.

For what branch to use, it depends. For normal day-to-day use, definitely use the no-database branch. The database code is currently interesting, but has no lasting effect while the device is still processing new books. Although as we've seen, getting the database entries wrong can really cause problems. As a tester, which branch to use depends on what code each bug touches. For any bug that touches the database code (like #5 and this one) the master branch must be used. For bugs that don't touch the database code, both branches must be tested.

[jgoguen]

The "koboSpan" class has been added to both branches.