[GUI Plugin] Extract ISBN

[tlindgren]

Apparently this plugin does "Mark Book" on all books it changes or adds ISBN to.
This has not been visible before but Calibre 1.6 is now showing these temporary marks.

IMHO the plugin should default to NOT do this because it may well interfere with actual marks set by the users. It might provide a configuration option to do so or offer it as one of the buttons (after it's run).

It's also why it sometimes triggers the Calibre 1.6 bug referenced above (which relates to marking and unmarking books).

[kiwidude]

@tlindgren - loads of my plugins use the "Marked Books" feature including this one - Find Duplicates, Quality Check, Reading List and no doubt others I am forgetting (or theoretically deprecating, like Temp Marker).

What am I supposed to do - completely disable all those plugins, crippling their functionality?

Until or unless Kovid/others come up with some other variation of Marked Books to replace the functionality which so many plugins rely upon, I have no intention of changing a thing. Marking Books is a *temporary* feature, always has been and AFAIK always will be. If you want to retain your marked books across multiple actions in calibre (be they my plugins or something else that changes markings) then right now your best bet is to simply Tag them, do your actions, then untag them afterwards.

[DoctorOhh]

Originally Posted by tlindgren

Apparently this plugin does "Mark Book" on all books it changes or adds ISBN to.
This has not been visible before but Calibre 1.6 is now showing these temporary marks.

The marks have been visible since 1.0 as a Star in the Cover Grid. With this release temp marks were expanded to include the list view and the icon switched to a pinhead. I have used Temp marks for 2 years, but because you can briefly see them now when you run a plugin you think there may be a problem?

Originally Posted by tlindgren

IMHO the plugin should default to NOT do this because it may well interfere with actual marks set by the users. It might provide a configuration option to do so or offer it as one of the buttons (after it's run).

So you're seriously suggesting to the person who created the Temporary Marks plugin about 2 years ago to stop using temp marks in his plugins because now you don't need his plugin to use the feature and you can see them?

Many of us, including myself, have been using temporary marks for 2 years (This is not a new feature) and not once have I ever had a problem so please let us know when you run into a conflict.

Originally Posted by tlindgren

It's also why it sometimes triggers the Calibre 1.6 bug referenced above (which relates to marking and unmarking books).

It is also why the plugin works. It would seem that 2 + 2 still equals 4, congrats on doing the math.

There are always bugs that crop up. The bug you mentioned has been fixed by Kovid and will be in the next calibre upgrade.

[BetterRed]

@tlindgren, @kiwidude and @DoctorOhh

Prior to 1.6, if you had books marked via the Temp Marker PI, and then you ran the Extract ISDN PI and you elected to update the books with the values, then a side-effect would be that the books that you had previously marked with Temp Marker would lose their marks.

So in that respect nothing has changed, this is the way the two plugins have always interacted, so if it's a 'bug' today, how come it wasn't a 'bug' yesterday.

The difference is that the 'marks' are now very visible in the Book List (spreadsheet) view, previously they were only visible in the recently added Cover Grid view via the presence of a gold star pasted onto the cover which was easily missed.

Marks put there by the ISDN PI are temporary - ie they don't persist across calibre sessions, just like the 'marks' put there by the Temp Marker PI and the new Mark Books feature.

But not all 'marks' are equal, at least one type of 'mark' does persist across calibre sessions. The Find Duplicates PI has an option to exclude apparent duplicates from reporting by future executions of Find Duplicates. It 'flags' the 'mark' with a not_book_duplicate attribute.

The Temp Marker PI would NOT include books 'marked' in this way in its Show marked books option. And the Find Duplicates PI would ONLY include books marked in this way in its Show all book duplicate exemptions option.

But this mutual exclusivity did not seem to 'work' in the Cover Grid view with respect to its Gold stars, which lead me to start the Show Marked as a Column thread. It was that thread that led to the Mark Books feature being implemented in the next release. And the 'problem' I saw with the Gold stars in the Cover Grid view are now manifested as Gold pins in the Book List View.

Feel free to blame me for the problems the Mark Books feature may have created, however I did foreshadow some of the potential consequences in the third paragraph of this post.

@kiwidude - I assume adding a not_book_duplicate tag to replace 'marks flagged' that way wont work unless Find Duplicates is updated to interrogate tags... whilst Ψing

BR

[DoctorOhh]

Originally Posted by BetterRed

But not all 'marks' are equal, at least one type of 'mark' does persist across calibre sessions. The Find Duplicates PI has an option to exclude apparent duplicates from reporting by future executions of Find Duplicates. It 'flags' the 'mark' with a not_book_duplicate attribute.

The Temp Marker PI would NOT include books 'marked' in this way in its Show marked books option. And the Find Duplicates PI would ONLY include books marked in this way in its Show all book duplicate exemptions option.

But this mutual exclusivity did not seem to 'work' in the Cover Grid view with respect to its Gold stars, which lead me to start the Show Marked as a Column thread. It was that thread that led to the Mark Books feature being implemented in the next release. And the 'problem' I saw with the Gold stars in the Cover Grid view are now manifested as Gold pins in the Book List View.

I pointed this out during beta testing in this thread. As you pointed out functionally they didn't seem to cause a problem.

Originally Posted by BetterRed

Feel free to blame me for the problems the Mark Books feature may have created, however I did foreshadow some of the potential consequences in the third paragraph of this post.

Blame you because now the marks are visible in the list view? In the third paragraph you said, "How will the new feature 'interoperate' with the more widely used Find Duplicates plug-in." The feature is not new (it has been mainstreamed into calibre), being able to see the marks are new. I agree that it may be unsettling to see the persistent Find duplicate marks and perhaps that plugin should be modified to somehow hide the marks, but that is a topic for the Find duplicates thread. That is why I didn't bring it up in my post.

The point is that if you notice temp marks from this plugin, that is by design and if someone now thinks this is a possible problem when for the last two years it was not a problem ...

[BetterRed]

@DoctorOhh - I wasn't even aware this PI marked the books for which it found an IDSN value - I had to do a test to see what it did.

I use this PI in my Receiving Dock library where I assign the core metadata values (author(s), title, strapline and ISDN) before moving the 'books' to their home libraries. In the Receiving Dock I display the ISDN as a column so its obvious which books get updated.

I missed your "Books Marked on Every Restart" thread. Had I seen it I would have remembered it and bought it to chaley's attention at an earlier point our discussion in the "Show Marked as a Column", rather than right at the end of the thread.

I have a workaround to the books marked as not_book_duplicate issue - uninstall/reinstall Find Duplicates as required.

BR

[timtam]

First, I would like to say that I found Extract ISBN plugin very useful.
I would really appreciate it, though, if an extra feature could be added in extracting ISBN from PDF files.

Currently, according to the first post of this thread, the plugin scans only the first 10 pages of the PDF file, then if ISBN not found, the last 5 pages in reverse order.

Could you please scan the "keywords" metadata field of the PDF file as well? Many books put their ISBN in the keywords field. Scanning the content in this field should be much smaller than scanning even one page of the PDF file and should be more authoritative.

Thanks for developing this great plugin!

[Townk]

I found a small problem with this plugin:

In some books they have 2 ISBN numbers. One for the printed edition and one for the eBook one. For instance the 11/22/63 book from Stephen King has this in the book:

ISBN 978-1-4516-2728-2 (print)
ISBN 978-1-4516-2730-5 (eBook)

The plugin successfully finds the ISBN, but since the printed version comes before the eBook one, it uses the wrong ISBN number even though the book contains the right one.

Unfortunately I don't believe there is an easy way to fix this problem without creating a different type of confirmation dialog. If a new dialog confirmation could be created than I would suggest the following:

After finishing the job, present a table with column 1 being book title, column 2 being the founded ISBN numbers as a combo box and column 3 a checkbox that indicates what user wants to do about (update it or not).

[kiwidude]

@Townk - this plugin offers just a simple and crude way to grab an ISBN primarily for metadata gathering purposes. As soon as you download metadata it will overwrite the ISBN again with one of the matches based on what you selected. So it is fairly pointless for most people to try to get too picky about which ISBN they have associated with the book, because of the way ISBN groups and calibre metadata download works.

So I am not going to try to make this plugin more interactive in that way (it gets really complicated when you have multiple books selected etc and becomes a whole lot of UI work for something few people really care about).

In the first iterations of this plugin it tried to have some logic around detecting certain text to reduce the false positives like searching for "ISBN" etc. However that was a failure as there is no universal standard by which publishers put this in their books. It just all gets too hard basically - so instead the plugin searches to find the first valid ISBN it can find and that is what will get used. And if your book unfortunately has an advertisement for some other book inside it listing it's ISBN on that page, well then you will end up with the wrong one from that too.

Treat it as a 98% plugin - it gets it right awy more times than it gets it wrong, but it will always require human interaction to have a 100% success rate.

[Townk]

@Kiwidude thanks for the reply. I understand and in some point agree with you about the work Vs. the benefits to work in this area (I can't 100% agree because this feature would be awesome for me :P).
Anyway, thanks again for taking the time to explain that and regardless of its 98% utility I think it is a heck of a plugin.

[uiuiii]

Thanks for the pluggin, in general, it's great. However, I got an error:

I got a few PDF ebooks (from the same publisher) with the same ISBN extracted ....from a fax number, i.e:

Tel: +...
Fax: +... (0)1279 431059 ---> this is extracted as a ISBN number

The line of ISBN is a bit later, preceeding with ISBN ....

I wonder how the pluggin analyses the line, would it be a bug, as it should search for the keyword ISBN first (?)

Suggestion:
- Would it be possible to give an option, checking against online metadata bases with title/author/publisher... as inputs to make sure the extracted number is correct?
- When more than one ISBN are found, would it be possible to display and let the user choosing one?
- The field ISBN in Calibre can contain only one value, I wonder how to support more than one value, i.e ISBN-10, ISBN-13..

[Noobish]

Originally Posted by uiuiii

Thanks for the pluggin, in general, it's great. However, I got an error:

I got a few PDF ebooks (from the same publisher) with the same ISBN extracted ....from a fax number, i.e:

Tel: +...
Fax: +... (0)1279 431059 ---> this is extracted as a ISBN number

The line of ISBN is a bit later, preceeding with ISBN ....

I wonder how the pluggin analyses the line, would it be a bug, as it should search for the keyword ISBN first (?)

Suggestion:
- Would it be possible to give an option, checking against online metadata bases with title/author/publisher... as inputs to make sure the extracted number is correct?
- When more than one ISBN are found, would it be possible to display and let the user choosing one?
- The field ISBN in Calibre can contain only one value, I wonder how to support more than one value, i.e ISBN-10, ISBN-13..

it's not a bug, some books stores information in a different way, which means that what you see isnt exactly stored in the order you see it, for example :
inside the book it could be stored like : ISBN (101)-101010100 FAX 1234567890908X
but when you see it u see like:
ISBN 1234567890908X
FAX (101)-101010100

in this case text search for isbn might fail despite of the search algorithm, you will need ocr search utility.

[uiuiii]

@Noobish and @kiwidude:
Thanks for the information. However, what I mean is this:

Here is an example I mentioned. The PDF content is:
Tel: ... (0)1279 623623
Fax: ... (0)1279 431059
...
...
ISBN 0 273 65667 8
...

Then from ISBN extract log:
Invalid ISBN match: 1279 623623
Valid ISBN10: 1279431059
Invalid ISBN match: ...
Invalid ISBN match: ...
Valid ISBN10: 0273656678
Invalid ISBN match: ...
...
New ISBN extracted of: 1279431059

Which means it uses the 1st 'Valid ISBN10' it finds (ignoring the 2nd 'true' ISBN), which makes sense, unfortunately it finds the one that's not so "valid".

Another eBook, with the same publisher:
Tel: ... (0)1279 623623
Fax: ... (0)1279 431059
...
...
ISBN: 978-0-273-71492-7
...

And the log:
Invalid ISBN match: 1279 623623
Valid ISBN10: 1279431059
Valid ISBN13: 9780273714927
Invalid ISBN match: ...
Invalid ISBN match: ...
Invalid ISBN match: ...
...
New ISBN extracted of: 9780273714927

Here it prioritizes on ISBN13, then it works well.

But as kiwidude said, "false positives" when searching for "ISBN", therefore my proposal to check against online resources, but this increases complexity for a pluggin then. Perhaps a selection/rejection among "Valid ISBN' found identified by the keyword preceding the 'ISBN' could be an improvement for the pluggin with not so much development effort.

By the way, these books are true PDF therefore the pluggin can read them.

Anyway, thanks for the pluggin, we can work with it so even if it will stay so.

[LikeBooks]

@kiwidude
Can you please make (all) your plugins translatable?
We´ll translate them on https://www.transifex.com/organizati...alibre-plugins

I started to localize and translate "Extract ISBN" -> see attatchement
(It´s possible, that I missed some strings, but it´s sth. to start with)

[JSWolf]

Has this on plugin ever been fixed to extract the eBook's ISBN when it lists both the electronic ISBN And that for the paper version?