Custom GUI plugins for Calibre here...

[janvanmaar]

Since the metadata including cover is updated on any export from the library ("save to disk" and "send to device")

Is this always the case (incl. the cover)? I always run conversion after changing the metadata before sending to the device. I developed this habbit after a book showing up with the old cover after sending it there without conversion. But it is possible that I did something wrong in that case, I should try on a few cases and reevaluate...

[kiwidude]

Originally Posted by Manichean

Since the metadata including cover is updated on any export from the library ("save to disk" and "send to device"), I stand by my reasoning: If you update the metadata from within Calibre, users might be tempted to just go into the library folder and pick the books out themselves for manual transfer. This can lead to Bad Things happening, as we've seen many times on this board. Kiwidude gave the only pro-updating argument I can see as being valid.

Further to that (and again viewer related) - it would mean the ebook viewer uses the same title in it's window that you have in Calibre rather than that used within the metadata. Could make identifying windows for comparing duplicates easier at times.

As for the other point about users being tempted to go direct etc. Personally I don't see that as an issue. There are already "complications" from the internal filenaming standard that Calibre uses which discourage such actions. Plus when I use "export" I know that I can get series information in the name which I could not get from going direct etc.

At the end of the day given the miniscule proportion of Calibre users that visit these forums, then the even smaller proportion who find this thread and are technical/willing to try plugins you are pretty much limited to hardcore Calibre users who know better anyways...

[Manichean]

Originally Posted by janvanmaar

Is this always the case (incl. the cover)? I always run conversion after changing the metadata before sending to the device. I developed this habbit after a book showing up with the old cover after sending it there without conversion. But it is possible that I did something wrong in that case, I should try on a few cases and reevaluate...

From what I know and experienced, that should always be the case. Although I have to say that I only use epub (my device supports nothing else), and that may be a relatively recent addition. I remember conversions being necessary in the 0.6.x versions.

[chaley]

Originally Posted by Manichean

Since the metadata including cover is updated on any export from the library ("save to disk" and "send to device"), I stand by my reasoning:

Well, I learned something new. Since 30/7, the epub metadata writer has updated the cover. There are a few situations where it can't (DRM, cover image is not raster, perhaps others).

I don't see the same change in any other format.

[kovidgoyal]

Covers are updated in *some* MOBI and EPUB files. In the case of EPUB if the original file identified the cover unambiguously in its metadata (and didn't use an svg cover). In the case of MOBI/PRC if the file is modern enough to support EXTH headers and the original file had a cover.

[itimpi]

As far as I know, the cover and metadata is not updated if I download a file via the Calibre server - this would mean that in that case I get the latest metadata and cover.

[kovidgoyal]

it is updated in the server, but only for EPUB, for performance reasons and plugboards are not used in the server.

[phenomshel]

Originally Posted by kovidgoyal

Covers are updated in *some* MOBI and EPUB files. In the case of EPUB if the original file identified the cover unambiguously in its metadata (and didn't use an svg cover). In the case of MOBI/PRC if the file is modern enough to support EXTH headers and the original file had a cover.

Aha! Kovid, I think you answered my question before I asked it again. The last bunch of books I purchased didn't have cover art (only a cover page in black and white with the title/author/publisher) embedded in the ePub files. I added them in calibre, then converted ePub to ePub so the covers would "stick" when I sent them to my Sony (because the last few batches didn't send the covers I chose). However, when I rechecked the files in the ebook viewer application, I got the cover art, but then a big blank broken image for the next page, which would have been the "cover page". I had to use "tweak ePub" and go in and fix them before sending to device. Therefore, I am assuming the ePub files in question probably had the cover identified specifically, and/or an svg cover. Until and unless publishers standardize the way they embed or don't embed cover art or cover pages, I think there's no easy fix for this...

[janvanmaar]

*** This plugin has now been moved to it's own thread - download from the link below ***
https://www.mobileread.com/forums/sho...d.php?t=128011

It is a rather specific plugin but perhaps it will be useful for someone.
The plugin enables sending of books from Calibre using SSH (or RSH) protocol. It requires rsync or similar utility to be installed in the system. It can be used to send books from Calibre wirelessly to your ereader or another computer with SSH server running (on Kindle, this can be achieved by usbNetwork hack; most other readers run Linux too, so it should be possible to install SSH server). The plugin enables customization of the rsync (or equivalent) utility path, remote host details, formats of books to be sent + a Kindle specific option of refreshing the book list after sending. The advantage of rsync is that it only sends the differences between the Calibre book and the book existing on the remote host, so if the book already exists or if only metadata have changed, there will be (almost) no transfer.

Notes:
The implementation is simple, so it has a few shortcomings currently: Calibre cannot be used while books are being sent by SSH; the button is not integrated into the "Send to device" group; applicability on Windows is questionable (no command line rsync utility for Windows as far as I know?), thus I disabled the plugin for Windows by default; only one remote host at a time can be specified at the moment; only SSH login with preshared keys is currently supported. Some of the shortcomings could possibly be sorted out on request - I did not want to spend too much time implementing without knowing that somebody would actually use it.

Installation instructions:
First follow the general instructions from kiwidude mentioned at the beginning of this thread: download the zip file, install the plugin, add the 'Send by RSYNC' button by customizing the toolbar.
Finally, customize the plugin in Preferences->Plugins->User Interface Actions Plugin->Send by RSYNC. A Calibre restart seems to be necessary for any Calibre plugin customization to take effect.

EDIT: kiwidude, could you perhaps add link to this plugin in your first post? thanks

[kiwidude]

Originally Posted by janvanmaar

EDIT: kiwidude, could you perhaps add link to this plugin in your first post? thanks

Done.

[Bertrand]

Hi

I'm using the plugin Search the Internet, and it works great.

One problem though. I have french books with accented letters and when doing a search on Amazon, here's an example of what I get :

"De l'inconvénient d'être né".

Which of course doesn't return any result.

When I perform a search directly on the website there's no problem.

On the other hand, I've added an other site (www.fnac.fr) in the site list and and it works well.

So what can cause the problem ?

Any thing I can do ?

Thanks.

[kiwidude]

Originally Posted by Bertrand

Hi

I'm using the plugin Search the Internet, and it works great.

One problem though. I have french books with accented letters and when doing a search on Amazon, here's an example of what I get :

"De l'inconvénient d'être né".

Which of course doesn't return any result.

When I perform a search directly on the website there's no problem.

On the other hand, I've added an other site (www.fnac.fr) in the site list and and it works well.

So what can cause the problem ?

Any thing I can do ?

Thanks.

I only have English books sorry so when it comes to handling other languages I have no experience. That's why all my plugins come with a "it works for me" warning

Without having a specific title name to test (put one here if you can) I can't verify this but my guess would be some sort of unicode characterset issue. My plugin does no unicode handling currently.

As to why it is one website and not the other. Ultimately what the plugin is trying to achieve is to generate a URL that matches what you see in the browser after you type that same title into the website's own search box. It could be that Amazon is doing some sort of translation that your other website is not or vice versa. Grab the search URLs that the website generates when you directly search on them and have a look at them, and then you can figure out what code needs to be changed in the plugin to match. The convert_to_search_text method is my very basic and crude conversion that works for a number of websites that I needed but clearly is not every situation.

It maybe that you need to add support for some alternate "tokens" for certain websites if unicode handling does not generically solve your problem across the sites you search. For instance this website http://www.isfdb.org/ has a very bespoke and picky search syntax that the plugin is not compatible with. If I decide I want it in the menu (for now I get to it via Wikipedia) then I would have to add a new token like {author_fn_ln} and corresponding code within the plugin for it so as to support both variations.

[Bertrand]

Thanks for your quick answer.

Here's the title, with accented letters : De l'inconvénient d'être né

Grab the search URLs that the website generates when you directly search on them and have a look at them, and then you can figure out what code needs to be changed in the plugin to match.

Hmm, I could try... The less I can say is that I'm not familiar with python...

:-)

[kiwidude]

Originally Posted by Bertrand

Thanks for your quick answer.

Here's the title, with accented letters : De l'inconvénient d'être né

Right, so chucking that at Amazon and you can see that it translates all those accented characters, so for instance é becomes "%E9".

I also am not a Python person (just figured enough out to get by), nor am I an expert on encoding URLs. I'll have a google later if I find the time unless some guru on either of the above can pick up and suggest a "generic" way to correctly convert such text to the encoded version. There is possibly some function or simple algorithm for doing it but I have no clue without further investigation.

[kovidgoyal]

from urllib import quote

quote(title.encode('utf-8))