Duplicate Organizer — a calibre plugin

[gmwestrup]

Tidy up a messy calibre library by collapsing scattered copies of the same book into one clean record — combining all their formats, the cover, and every scrap of metadata under a single entry.

It only ever deletes true exact duplicates, keeps anything that genuinely differs, and always shows you a full review plan before changing a single thing.

How it finds duplicates
Any rule can be toggled. Books linked by any rule are grouped together.

Title + Author
Same title and same author(s), ignoring case and spacing.

Close title match
Ignores subtitles, punctuation, and a leading “The/A”, so different-format copies group together even when their metadata differs slightly.

Identifier
Same ISBN, Google, Amazon, DOI, or any other identifier.

File content
Identical files by SHA-256 hash — the most precise check.

What it does with each group

• Keeps one master copy — by default the one with the most formats (or newest / oldest, your choice).
  
• Copies in every missing format — EPUB, MOBI, PDF, and the rest — so all formats of a book end up together under the author.
  
• Merges all metadata — cover image, series, publisher, dates, rating, languages, comments, tags, identifiers, and custom columns. Empty fields are filled and tags/identifiers unioned, without overwriting what's already set.
  
• Deletes only exact duplicates. A copy that shares a title but has genuinely different content in the same format is kept and given a numbered title (e.g. “The Hobbit (2)”), so nothing is ever lost.
  
• Safe by design

—
Always shows a full, itemized review dialog before touching anything. Uncheck any group to skip it.
—
Deletions go through calibre's normal recoverable path, not permanent removal.
—
Run it on your whole library or just the books you've selected.
Works with calibre 6+ (including 9.x). Configure matching rules and behavior under
Preferences → Plugins → Duplicate Organizer.

Now it merges everything into the kept book:

All formats the kept copy is missing.
Cover image — pulled from a duplicate if the kept book has none.
All metadata — series, publisher, dates, rating, languages, comments, plus tags and identifiers (unioned), and any custom columns. Empty fields get filled; existing values on the kept book are never overwritten.
One thing worth knowing: it fills empty fields rather than replacing populated ones, so the "keep the copy with the most formats" master stays authoritative for anything it already has. If you'd rather a specific copy's metadata always win, I can add that as an option.

[JSWolf]

How do you merge in multiple copies of the metadata when they differ? For example, different covers.

[theducks]

Originally Posted by JSWolf

How do you merge in multiple copies of the metadata when they differ? For example, different covers.

Merge does not stomp existing.
Merge is mostly for differing (file) formats. If the cover is different, it probably IS NOT the exact same book (different editions)

If you want both sets comments, the MANUEL copy paste. If you want more, then don't merge.

[JSWolf]

Can you have a list of the proposed duplicates? Can the proposed duplicates be SHA-256 checked to see if they are the same?

[gmwestrup]

Originally Posted by JSWolf

How do you merge in multiple copies of the metadata when they differ? For example, different covers.

It takes one cover, not all of them combined. Current behavior: if the kept ("master") book has no cover, it grabs the first cover it finds among the duplicates being merged in. If the master already has a cover, that one is kept as-is and the others are discarded (there's no such thing as "combining" multiple cover images into one — that's not really an operation that makes sense for a book cover).

If you want different behavior — e.g. always prefer the highest-resolution cover among all copies, regardless of whether the master already has one — I will add that.

[gmwestrup]

Originally Posted by JSWolf

Can you have a list of the proposed duplicates? Can the proposed duplicates be SHA-256 checked to see if they are the same?

It does list the proposed changes. There are also diffrent selection now in prefrences to make changes. I also just added a few more details and options. Version 2 uploaded.

Use this also with the find Duplicates plug in. If you do that you can select which one to process.

[gmwestrup]

This is version 2 added more preferance selections.

[JSWolf]

I'll give this a try when I get some spare time.

[Wiggo]

Am I right in assuming that the merging happens automatically?

[JSWolf]

If I have two duplicates based on and exact author and title match, can there be an option to delete the copy with the earlier date? Thanks.