Behavior changes

[kiwidude]

Thx for the debate on this guys. So to summarise the state of play as I understand it...

• Find Duplicates can be easily sorted with Kovids additions today.
• Count Pages and Modify ePub I should be able to change without drama.
• With Extract ISBN there could be a large number of files to copy up front (if the user selects several thousand books as they have in the past it will result in many more thousands of book formats being copied up front) but it's the "right" thing to do anyway as a background job.
• Quality Check could potentially be changed to use the SpooledTempFile approach since it runs in process. But will hold off making changes until the db API settles in case some other idea crops up
• Open With remains a sticky problem, but again some bright idea may come in down the track.

[kovidgoyal]

Seems about right

[jackie_w]

H-e-e-l-l-p-p! Most of this thread is completely over my head but the keyword cover(), from post #1 has made me reply rather than go get my duvet and sit under it singing la-la-lah-la-lah until it's all over

I have no idea whether my cosmetic Copy Cover plugin will be affected by your plans and I don't run from source to suck-it-and-see.

Please can someone help me figure out whether I need to do anything? I think the following info may be relevant.

1. Plugin currently runs modally.
   
2. It does not update any data in the calibre database but it does process multiple records (1-99) sequentially.
   
3. I'm not aware of using a format() function.
   
4. I'm not exactly sure what you mean by the cover() function but I do use the following code:
   

   db = self.gui.library_view.model().db
   cover_path = db.cover(row, index_is_id=False, as_path=True)
   cover_data = db.cover(row, index_is_id=False, as_path=False)

5. The cover_path is only used for testing existence of a cover. It is not used to copy the cover from calibre to device.
   
6. Instead, the cover_data is used, in one of 2 ways depending on attached device:
   • - directly, using calibre function save_cover_data_to from calibre.utils.magick.draw
   • - indirectly by using an instance of the Image() class from calibre.utils.magick to load the cover_data followed by certain actions with the Image methods size and quantize. Image instance is then copied to device using save_cover_data_to.
7. I use os.path quite a lot but (I think) always referring to files/folders on the device rather than in calibre. I can check more carefully if necessary.

Now, do I need to get the duvet?

[kovidgoyal]

You're fine, the only difference for you is that every time you use cover(..., as_path=True) a temp file is created and the cover is copied to it. The tempfile is automatically deleted when you close calibre.

I would suggest that you use a single call to both get the cover data and test for its existence, like this:

cover_data = db.cover(row, index_is_id=False, as_path=False)
if cover_data is None:
   # no cover present
else:
   # cover present

And let me emphasize that the changes should not break any existing plugins. The worst that will happen is that a plugin will run slower and create lots of temp files.

[jackie_w]

Well, that's made me feel a bit more relaxed

I should obviously have coded it that way to begin with. Do you advise me to do a new release just to make that change or can I leave it until I actually have a new feature to offer (if ever).

[kovidgoyal]

It's up to you, if you are the type that cannot live with the creation of all those extra temp files on all your users computers, then make an updated release.

If I were you, I would make an updated release, but then, I have a disease, just look at the number of calibre releases there are...

[kiwidude]

Luckily I don't have that disease. I can go months without releasing.

[jackie_w]

Long may you be thus afflicted.

OK, boss, I'll get it sorted. I've been interested to see the number of my plugin downloads rocket since you released v0.8.6. About 10/day before, about 100/day since. Presumably, not a coincidence as I haven't added any functionality.