Clening up my old ebook library

[Kedryn]

I have a library that got a bit messed up by readarr-calibre-cwa, that i want to use with Bindery
I have so many unmatched book (nearly 2k) and i'm startint to slowly fix every author folder
Starting with William Gibson, i cleaned up all multiple unwanted folder inside his root, and left one folder\epub for every book i have, some in italian, some in english

now, doing a rescan does not seem to fix the bindery db.
For example, in bindery i have a book "Burning Chrome". it's status is "wanted" (and not "imported"), it's folder is
/books/William Gibson/Burning Chrome/Burning Chrome - William Gibson [eng] (1981).epub
the file exists in that path, but aven after a scan, it's still "wanted".

How i can have bindery check the new folder/files in the author root folder?

The oddity is that filtering for imported files, it shows also that book but still with "wanted" status

[Kedryn]

Ok, there were 2 files in that folder, one epub and one txt, and it was matching the txt, not the epub.
That fixed it, but i've found another "not updated by scan" folder.
The book "Distrust That Particular Flavor by William Gibson" is imported, and it's folder is
"/books/William Gibson/Distrust That Particular Flavor by William Gibson/Distrust That Particular Flavor by William Gibson - William Gibson [eng] (2012).epub"

But that folder does not exists enymore on the filesystem, it's
/books/William Gibson/Distrust That Particular Flavor/Distrust That Particular Flavor - William Gibson [eng] (2012).epub

Scanning the library does not fix that

[vavallee]

Late reply, sorry. Both things you hit are real bugs, now filed and fixed.

The txt winning over the epub. Bindery treats .txt and .rtf as ebook containers, and when several files compete for the same book it takes whichever the directory listing yields first. So a notes file sorting ahead of the epub takes the book's ebook slot and the epub gets reported unmatched. There is already a rule that calls those files supplements rather than ebooks, but it only applies when the folder also contains audio, so an ebook only folder had no protection at all. Filed as #2188, fix in #2190. Real containers now claim the book ahead of sidecars, and the result no longer depends on scan order. A text only or PDF only library still works, since a sidecar still claims a book when nothing better is competing for it.

The renamed folder that never updated. This one is worse than it looks. Rescanning does find the file at its new path, and it does register it, but the old row stays and the book keeps rendering the dead path forever, while still saying Imported. That is because the derived path is taken from the oldest row rather than the one that exists. Filed as #2186. The ABS importer already prunes rows whose files have vanished, the library scan just never got the same treatment.

Until that ships you can clear a stuck one by deregistering just the dead row, which leaves the file on disk:

DELETE /api/v1/book/{id}/file?path=<the old path exactly as the book reports it>

Get the exact spelling from GET /api/v1/book/{id}/files. Do not use the Delete file button in the UI for this, it deletes every file for that format from disk, including the good one.

Thanks for reporting both, and for spotting that the txt case was about which file got picked rather than about matching in general. That is what made it findable.