Sony PRS-T1 in Calibre

[kovidgoyal]

That error message is displayed when the query

SELECT file_path, _id FROM books

fails on your database.

[log]

Thanks kovid, there actually might be a problem, just haven't noticed anything on my reader yet.

When I do select count(*) from books, I get 45, when I do select * from books I get only 5 rows back. Maybe it really is corrupted?

[itimpi]

Normally if you delete the Sony database file a new (clean) file gets rebuilt from scanning the books when you disconnect from the USB.

[Kolenka]

Originally Posted by log

Thanks kovid, there actually might be a problem, just haven't noticed anything on my reader yet.

When I do select count(*) from books, I get 45, when I do select * from books I get only 5 rows back. Maybe it really is corrupted?

The issue is that there is SOMETHING about the SQLite DB that has made it unreadable to the SQLite version on the PC. But apparently what the T1 itself is using is more tolerant and can work with the DB anyhow. It may be related to the version of SQLite being used or something else.

But the reality is that when it happens, even my nice DB analysis tools that use SQLite cannot read the DB without dumping it and re-creating it. I can certainly look into trying to automate it, but I have very little free time as it is. The work I did on the driver was mostly to help jump start support on the T1, as I simply don't have the time to keep a close eye on the community. Not nearly as much as I'd like.

EDIT: It could also be due to how the T1 caches DB information, since I've seen the official app run into this "corrupt" DB as well. There is a bug with the T1 firmware where plugging it in while it is asleep will kick you out to the home screen instead of the "you are connected" screen. This creates a situation where both the T1 and the PC are reading/writing the DB at the same time. This is a recipe for problems.

[Orthonormal]

I have found a quirk with book covers. I imported an RTF document into Calibre, and used Calibre's edit metadata window to add a JPG book cover. Then I converted to ePUB format for the Sony Reader. Then I used Calibre to save the ePUB to the PRS-T1.

The cover graphic shows up in thumbnail views (the library, most recently read) but it is not used for the standby screen -- instead I get a blank screen.

I have noticed on books that do display the cover on the standby screen, that the first page when I read the book is the cover image. On books that do not display the cover on the standby screen, the first page is blank.

It may or may not be relevant that the books with the blank-first-page problem are the ones converted from RTF, and the books that do not have the problem are all native ePUB. And that may or may not be universally true -- it is just what I have observed so far.

[Kolenka]

Originally Posted by Orthonormal

I have found a quirk with book covers. I imported an RTF document into Calibre, and used Calibre's edit metadata window to add a JPG book cover. Then I converted to ePUB format for the Sony Reader. Then I used Calibre to save the ePUB to the PRS-T1.

The cover graphic shows up in thumbnail views (the library, most recently read) but it is not used for the standby screen -- instead I get a blank screen.

I have noticed on books that do display the cover on the standby screen, that the first page when I read the book is the cover image. On books that do not display the cover on the standby screen, the first page is blank.

It may or may not be relevant that the books with the blank-first-page problem are the ones converted from RTF, and the books that do not have the problem are all native ePUB. And that may or may not be universally true -- it is just what I have observed so far.

By default, Calibre can copy over the thumbnail it has to the T1 for you. This means browsing will show the correct thumbnail (otherwise the T1 will render the first page). When showing the standby screen, the T1 always renders the first page. My guess is that your conversions don't actually include a cover as the first page.

[Orthonormal]

Originally Posted by Kolenka

When showing the standby screen, the T1 always renders the first page. My guess is that your conversions don't actually include a cover as the first page.

I looked around for options to include the cover on the first page, and didn't see anything likely. Any tips?

[JSWolf]

Originally Posted by Orthonormal

I looked around for options to include the cover on the first page, and didn't see anything likely. Any tips?

Your best to add the cover when you do the conversion. There is an option someplace to add a separate graphic as the cover when you convert.

[Orthonormal]

The real issue in my case was something odd about the JPG file. The ePUB plugin for Firefox displayed the cover on page 1, but Digital Editions (and the PRS-T1) did not. After cleaning up the JPG file format using jpegtran, the covers were correctly displayed on the PRS-T1 and in Digital Editions.

I don't know what was wrong with the JPG files, but running jpegtran with -optimize, -progressive, and -copy none fixed it.

PS - when converting a book individually, you are given the opportunity to change the cover, or add one if there isn't already one assigned. I already had covers assigned to the books, so that wasn't the problem. There is also an option to not include covers (either in single or batch conversion) but I didn't have that checked.

[log]

Originally Posted by Kolenka

The issue is that there is SOMETHING about the SQLite DB that has made it unreadable to the SQLite version on the PC. But apparently what the T1 itself is using is more tolerant and can work with the DB anyhow. It may be related to the version of SQLite being used or something else.

But the reality is that when it happens, even my nice DB analysis tools that use SQLite cannot read the DB without dumping it and re-creating it. I can certainly look into trying to automate it, but I have very little free time as it is. The work I did on the driver was mostly to help jump start support on the T1, as I simply don't have the time to keep a close eye on the community. Not nearly as much as I'd like.

EDIT: It could also be due to how the T1 caches DB information, since I've seen the official app run into this "corrupt" DB as well. There is a bug with the T1 firmware where plugging it in while it is asleep will kick you out to the home screen instead of the "you are connected" screen. This creates a situation where both the T1 and the PC are reading/writing the DB at the same time. This is a recipe for problems.

Ok, this is strange. It looks like the T1 must have done something to fix the corrupt database?

I'm still using the same version of calibre where I had the issue (.832) and I just finished reading the book I was reading when I first noticed the issue, just connected to calibre, had no error message and was able to transfer a new book with no error messages.

Also FYI my previous statement with difference between select count(*) and select * might be wrong. I still get the same results with my client (Base for OSX) but I just downloaded the official command line sqlite3 client for OSX and I get the proper rows when doing a select *.

And Kolenka, you have already done so much for us T1 users by enabling the calibre support so no worries! I will at least try to get some more detailed information on why it happens on how it fixes it self by keeping copies of the dbs it happens again so we can get to the bottom of it.

[IorekB]

Hi everyone,
I recently opened a new thread concerning an issue I came across with calibre-created news epubs and the default behaviour of calibre when creating collections on the PRS-T1. Quick summary:

By default, news epubs are sorted into the exact same organisational structure (according to their respective source) on the reader twice: Once in the "Periodicals" section, and again in the "Collections" section (due to collections being created for all tags). A few attempts of resolving this on my own having failed, I suggested to include a way into calibre to prevent the (unnecessary) duplication mentioned above from happening, thus avoiding clutter in the collections list (genres and news sources all mixed up).

Kovid Goyal seems to approve of the idea (Yay for this happening to my first own thread after joining! ) and suggested posting in this thread, where mentioning it would seem more likely to attract the attention of the right people most probably being able to help out/maybe add an option (as suggested in more detail here) to the device driver. I really would appreciate that!

[cayle]

Originally Posted by requiem_mortis

3) I have a 'collection' on 'Internal Memory' which is displayed with an icon that looks like a 3 book stack with left/right arrows on it which I can neither delete rename or add content to. How do I clear the issue to allow me to create 'Internal' memory collections that work ?

This is a collection created by the Sony Reader software. Just connect to the program, delete the collection and sync collections.

[Kolenka]

Originally Posted by IorekB

By default, news epubs are sorted into the exact same organisational structure (according to their respective source) on the reader twice: Once in the "Periodicals" section, and again in the "Collections" section (due to collections being created for all tags). A few attempts of resolving this on my own having failed, I suggested to include a way into calibre to prevent the (unnecessary) duplication mentioned above from happening, thus avoiding clutter in the collections list (genres and news sources all mixed up).

The behavior was modeled after the behavior in the 505/6xx/7xx/9xx driver, which does pretty much the same thing on the models that have a periodicals section. I can see disabling this being much more useful on the T1 though (especially now that stuff appears in the periodicals section).

Actually, the fastest way to get my attention is to file a Calibre bug/request on it. Kovid has been triaging the T1 stuff to me, and I get an e-mail on it the same day. I might not hear about such requests here for weeks.

[JSWolf]

I have an idea if anyone can implement it. If the file is a periodical, have an option in Calibre to not have collections created and just put it in the periodicals section. That would solve the problem. So the user can decide what to do and that would work best.

[Kolenka]

Originally Posted by JSWolf

I have an idea if anyone can implement it. If the file is a periodical, have an option in Calibre to not have collections created and just put it in the periodicals section. That would solve the problem. So the user can decide what to do and that would work best.

That is exactly what Iorek is asking for 2-3 posts up.