[GUI Plugin] Extract ISBN

[theducks]

Originally Posted by Comfy.n

brazilian portuguese was 85 until 2018 from what I gathered, then it's been replaced by 65, from what I learned just now that you asked!

for the reply. 0 and 1 are all I can read

[Wiggo]

Originally Posted by kiwidude

Wiggo - if you do something constructive (like follow the instructions I gave above and provide the debug and plugin logs) then maybe I can try to help. However the *fact* is that I can't replicate the issue - the plugin works perfectly fine for me and clearly the vast majority of users or else there would be far more posts about it. As for why no response before from me - in 2021 I wasn't maintaining this plugin at that time, davidfor was so I never saw that post.

There has been a bug where it reports a string formatting error rather than the real underlying error. So the actual "final" error that people have posted is useless from a debugging perspective. Until I see debug logs to say otherwise I believe the problem will lie in trying to run pdftohtml or pdfinfo executables on your system. Which are calibre provided executables and nothing I have control over. As I said in the post above if they are failing to execute then the first step is to find out why that is - perhaps you have antivirus interfering for instance. You could try running the command line it is executing manually to see if that works.

But only *you* can do anything about this when I can't replicate the issue.

In my defence, I have to say that until now it didn't really seem as if you were interested in the problem.
The second last mention of this problem with cross-reference was only 4 posts before your reappearance in this thread.

The fact that pdfinfo.exe is the culprit has also been mentioned there, hence the workaround with replacement of the same.

[Shet]

Originally Posted by kiwidude

@shet - thx for trying and posting that. Now we can see the actual underlying problem - for whatever reason the calibre pdfinfo.exe is returning no data for your pdf on your system, whereas it works perfectly fine on mine. This falls into the “a question for Kovid” domain - as I said above the pdfinfo and pdftohtml executables are not part of this plugin, they are supplied with calibre. Perhaps he may have an explanation as to what might be the problem here.

For info: the best way is for me to bulk convert pdfs to epubs, run ExISBN, than bulk delete epubs. accetable workaround for me

[kiwidude]

Shet - glad you have a workaround in the meantime. Perhaps Kovid may have a suggestion, or perhaps he like me cannot replicate issues with pdfinfo.

[kovidgoyal]

Originally Posted by kiwidude

Shet - glad you have a workaround in the meantime. Perhaps Kovid may have a suggestion, or perhaps he like me cannot replicate issues with pdfinfo.

Works for me as well. pdfinfo.exe is built from poppler in a pretty standard way, see my bypy repository in github for the script used to build it. This will be something interfering on the affected computers. Note that nowadays, you shouldnt need to use pdfinfo for most things, you can use the podofo bindings for getting metadata for example, this avoids needing a separate process and so should be faster and possibly more robust, though I suspect podofo will fail on more malformed PDFs than pdfinfo, so you could fallback to pdfinfo. See podofo/__init__.py for the various python wrapper functions that can be used.

[kiwidude]

Thanks very much Kovid for the confirmation and the suggestion of an alternative. I will take a look.

@Comfy.N - pdfinfo is only used to get page count metadata. After that it is a separate pdftohtml executable which is responsible for producing the book content that is parsed for ISBNs. So you will see zero difference between different pdfinfo versions in how many isbns get reported. Other than of course if the calibre bundled pdfinfo doesn't work for you at all.

[kiwidude]

Release Notes
https://github.com/kiwidude68/calibr...ct_isbn-v1.6.2

This now uses podofo as per Kovid's suggestion. @Shet please let me know how you get on with this one.

To anyone who had tried using the zip I had attached yesterday of a test build - please make sure you replace that version with this one (it had the same version number so you will have to force install it).

[Shet]

Originally Posted by kiwidude

Release Notes
https://github.com/kiwidude68/calibr...ct_isbn-v1.6.2

This now uses podofo as per Kovid's suggestion. @Shet please let me know how you get on with this one.

To anyone who had tried using the zip I had attached yesterday of a test build - please make sure you replace that version with this one (it had the same version number so you will have to force install it).

This one works perfectly with pdfs for me. Maybe this is the right way as Kovid said. THX

[kiwidude]

Release Notes
https://github.com/kiwidude68/calibr...ct_isbn-v1.6.3

Thaks to Paul Harden for the contribution which may help for those scanning PDFs.

[davidjoseph1]

I don't know how out of bounds my asking about this is, but LCCNs - Library of Congress Control Numbers - are very useful for grabbing metadata. They're described at Wikidata.

Books with cataloguing in publication data have it on the copyright page.

[kiwidude]

Release Notes:
https://github.com/kiwidude68/calibr...ct_isbn-v1.6.4

Handles edge cases where valid matches were being rejected because the book listed them as "ISBN-13 xxx" and it would test "-13 xxx" rather than "xxx"

[Bertrand]

Hi there,

Since the last update (I think), I have a problem with the plugin.
Every time I run it on one or more books, the plugin systematically deletes the “Title” field of the book.

I don't see anything in the options to correct this behavior.
Did I miss something ?


Before :



After : title = "Inconnu" (Unknown)


Thank you for your help.

[Robert78]

Hi there,

Like Bertand, I have a problem with the plugin since I upgraded to 1.6.4.

Title column is replaced with an "unknown" value.

I have downgraded to the previous version and the problem has disappeared.

[kiwidude]

Nothing has changed in this version of the plugin as far as how metadata is updated. If you are seeing a different behaviour I believe it is because (a) you are now seeing an ISBN being extracted where it was being missed previously (due to the change I made), and/or (b) you are also using a different calibre version.

I don't see this behaviour myself, and it is impossible to guess if you don't actually include log file output, calibre version and do some basic testing yourself - does it happen for all books, is whether they currently have an ISBN relevant, is it epubs or PDFs or both etc.

[ownedbycats]

I'm unable to replicate in Calibre 7.17.100.