Wrong metadata after merging

[dancal]

Hi,

I have enabled a single plugin (Google). The log shows calibre downloads the correct metadata. Yet (after the metadata download) in the Edit Metadata > Download Metadata dialog, I end up with an older publication date, one that isn't even in the log:

Running identify query with parameters:
{u'authors': None, u'identifiers': {u'isbn': u'0321545893'}, u'timeout': 30, u'title': None}
Using plugins: Google
The log from individual plugins is below

****************************** Google ******************************
Request extra headers: [('User-agent', 'Mozilla/5.0 (Windows; U; Windows NT 6.1; en-US) AppleWebKit/532.5 (KHTML, like Gecko) Chrome/4.0.249.78 Safari/532.5')]
Found 1 results
Downloading from Google took 1.23421812057


---
Title : Artificial Intelligence: Structures and Strategies for Complex Problem Solving
Author(s) : George F. Luger
Publisher : Pearson Addison-Wesley
Tags : Computers, Intelligence (AI) & Semantics, General
Languages : eng
Published : 2009-12-15T17:53:41.284761+00:00
Identifiers : google:bZwdAQAAIAAJ, isbn:9780321545893
Comments :

************************************************** ******************************
The identify phase took 1.41 seconds
The longest time (1.234218) was taken by: Google
Merging results from different sources and finding earliest publication dates from the xisbn service
We have 1 merged results, merging took: 0.67 seconds

In the dialog I get 1993 instead of 2009 as the publication date (which is the date of an older edition with a different ISBN) -- yet the ISBN is correct.

Is this due to the "xisbn service"? How come the log doesn't even mention a 1993 book? Can I disable merging or xisbn?

[kovidgoyal]

From the log: Merging results from different sources and finding earliest publication dates from the xisbn service

And look in Preferences->Metadata dowload

[dancal]

Originally Posted by kovidgoyal

From the log: Merging results from different sources and finding earliest publication dates from the xisbn service

And look in Preferences->Metadata dowload

OK, but I don't see a way to disable "finding earliest publication dates from the xisbn service" in Preferences->Metadata dowload. If I uncheck "Published date", I suppose I won't get any date at all from the sources.

[theducks]

Originally Posted by dancal

OK, but I don't see a way to disable "finding earliest publication dates from the xisbn service" in Preferences->Metadata dowload. If I uncheck "Published date", I suppose I won't get any date at all from the sources.

You can ALSO configure each source.

finding earliest publication dates from the xisbn service

[dancal]

Originally Posted by theducks

You can ALSO configure each source.

The Configure selected source dialog "enables" me to select which fields will be downloaded from that specific source.

There is no xisbn source. I have Google, Amazon and Open Library, but as the attached log shows, only Google is queried (since I have disabled the other 2 plugins)

Again, I have disabled all plugins except Google; Google gives me the correct data in the log; then apparently xisbn overwrites the Published Date field. How do I disable "xisbn service" or whatever it is that overwrites the pubdate?

I would appreciate a specific answer (as opposed to a riddle).

[theducks]

Originally Posted by dancal

The Configure selected source dialog "enables" me to select which fields will be downloaded from that specific source.

There is no xisbn source. I have Google, Amazon and Open Library, but as the attached log shows, only Google is queried (since I have disabled the other 2 plugins)

Again, I have disabled all plugins except Google; Google gives me the correct data in the log; then apparently xisbn overwrites the Published Date field. How do I disable "xisbn service" or whatever it is that overwrites the pubdate?

I would appreciate a specific answer (as opposed to a riddle).

Sorry, I was thinking it was the ISBNDB.com source

[DoctorOhh]

Originally Posted by dancal

OK, but I don't see a way to disable "finding earliest publication dates from the xisbn service" in Preferences->Metadata dowload.

Ensure in Preferences->Metadata download you have unchecked the Use published date of "first edition" (from worldcat.org) box.

[dancal]

Originally Posted by DoctorOhh

Ensure in Preferences->Metadata download you have unchecked the Use published date of "first edition" (from worldcat.org) box.

Thanks. This option is not available in my 0.9.0 on Ubuntu, but is available in 0.9.9.

I have backported 0.9.0 (the latest Debian/Ubuntu version) to Ubuntu Precise. I guess I'll have to either install from source or wait.

[PeterT]

That was added in 0.9.2

• Metadata download: Turn off the use of the published date for the earliest edition a book as the published date. The earliest edition was identified via worldcat.org, which has rather poor data, leading to the occasional incorrect result. If you want this feature back, you can turn it on again via Preferences->Metadata download.

Can't you just follow the install procedure on the Calibre download page?

Binary install

calibre has a binary install that includes private versions of all its dependencies. It runs on 32-bit and 64-bit Intel compatible machines. To install or upgrade, simply copy paste the following command into a terminal and press Enter:

sudo python -c "import sys; py3 = sys.version_info[0] > 2; u = __import__('urllib.request' if py3 else 'urllib', fromlist=1); exec(u.urlopen('http://status.calibre-ebook.com/linux_installer').read()); main()"

[DoctorOhh]

Originally Posted by dancal

I have backported 0.9.0 (the latest Debian/Ubuntu version) to Ubuntu Precise. I guess I'll have to either install from source or wait.

Or do a binary install of 0.9.10 via the command line indicated on calibre's download page.



FYI the feature was added in version 0.9.2

Release: 0.9.2 [11 Oct, 2012]

New Features

• Metadata download: Turn off the use of the published date for the earliest edition a book as the published date. The earliest edition was identified via worldcat.org, which has rather poor data, leading to the occasional incorrect result. If you want this feature back, you can turn it on again via Preferences->Metadata download.

[dancal]

[rant]
Sure, but it's the same thing -- you lose package management for when you want to uninstall (or query file / package relationships)

Since you have to apt-get uninstall calibre first, you also end up with "orphaned" dependcies, which means apt "helpfully" prints

The following packages were automatically installed and are no longer required: ...

... every time you use it.

I guess it's not a big deal for a single package, but once you start making exception after exception the filesystem gets pretty messy.

Of course, it's quite surprising that some popular (and high quality) packages are so neglected by Debian. It sometimes looks like the maintainer has pretty much abandoned the package without formally giving up maintainership.
[/rant]

Anyway, thanks to all who replied.

[Adoby]

Why not remove the unused dependencies? That way everything will be nice and tidy.

Research the command apt-get autoremove

As long as you have been careful and not been messing with your install, it should work fine and leave you with a lean and clean system without any orphaned packages. I use it without any hesitation on my Ubuntu install.

[dancal]

I know my Debian, thanks

What do you accomplish by removing calibre's dependecies as well? Even more unmanaged files spread around the filesystem?

What if some of those dependencies are needed by some other (future or present) package? You may not even be able to uninstall the deps, unless you "go manual" with that other package too.

The only clean solution is to make an updated Debian package (which often requires updating the dependencies too).

[Adoby]

There really is no point in having packages installed that are not needed. Just unnecessary bloat.

Why would you want to keep unused packages? If they ever are needed again, by some other package, apt will install them again.

The binary install for Calibre includes private versions of all the dependencies. So there is no need to keep, maintain or install anything else but the binary install. When you update that, all the private dependencies are updated as well. It may be a little wasteful of hard disk space, but it sure saves a lot of time not having to install and update and maintain a large set of dependencies. Kovid has all that sorted and tested for you in each new release.

And Calibre doesn't spread a lot of files everywhere. Just remove one folder and you almost all of Calibre is gone. Not anything that can cause any problems.

[dancal]

Well, it turns out building a calibre-0.9.10 .deb package was as easy as uscan / uupdate. I will test it and upload it to my PPA.

I don't keep unused packages. But I *prefer* to keep my USED packages in the apt database (as opposed to informally-managed manual installations). Package managers like dpkg and rpm where invented for a good reason (and apt and yum for *another* good reason).

As to the option of having both a manually-installed and an apt-installed package (with different versions): that can lead to problems, because it's unclear how programs locate shared libraries / configuration files / dependent binaries. You never know which version ends up taking precedence.

Perhaps the above may not apply for calibre, but I've had nasty experiences with other packages (e.g. jack)