Bug in author sort

[JSWolf]

In calibre 9.3, I found an author sort bug.

The OPF code is

<dc:creator opf:file-as="Dana Stabenow" opf:role="aut">Dana Stabenow</dc:creator>

It picks up Dana Stabenow as the author sort instead of Stabenow, Dana.

[un_pogaz]

Always as been.
Calibre has always imported "author sort" stored in the OPF as it, even if is incorect.

Use Quality check to found incorect value.

[JSWolf]

Originally Posted by un_pogaz

Always as been.
Calibre has always imported author sort sorted in the OPF as it, even if is incorect.

It's the first time I've noticed this issue. The author name and author sort should not be the same. It should be that if the author sort and author name is the same without a comma then it gets sorted in author sort and if it has a comma, it should get sorted in author name.

[un_pogaz]

Originally Posted by JSWolf

It's the first time I've noticed this issue. The author name and author sort should not be the same. It should be that if the author sort and author name is the same without a comma then it gets sorted in author sort and if it has a comma, it should get sorted in author name.

Damn it, JSWolf read and understand what I write:
Calibre has always import "author sort" stored in the OPF as it, even if is incorect.
this mean:
If the epub already contain a stored value "author sort", Calibre will use it, even if it incorect. You need to manualy check if the "author sort" value of imported book is the good one or not.

You imported book has an incorrect value, Calibre read and use it. If you never check the "author sort" of your book, their is probably plenty of incorect values in your library.

Is not a bug or a issue, is the excepted behavior. Hence the usefulness of the Quality check action.

[gbm]

Sign. You do not check your Author sort in Edit metadata?
https://manual.calibre-ebook.com/met...book-at-a-time

You can click the button next to author sort to have calibre automatically fill it in using the sort values stored with each author. Use the Manage authors dialog to see and change the authors’ sort values. This dialog can be opened by clicking and holding the button next to author sort.

berne

Originally Posted by JSWolf

It's the first time I've noticed this issue. The author name and author sort should not be the same. It should be that if the author sort and author name is the same without a comma then it gets sorted in author sort and if it has a comma, it should get sorted in author name.

[theducks]

Originally Posted by un_pogaz

Always as been.
Calibre has always imported "author sort" stored in the OPF as it, even if is incorect.

Use Quality check to found incorect value.

Yep. Calibre only calculates upon import if there was no value. Been that way since Guttenberg started printing

[Comfy.n]

Out of curiosity I made a search in my largest library and got these results, where fileas-search is a tag applied to all EPUBs containing the file-as OPF metadata property, and fileas-search-comma is a tag applied to all in which such property contains a comma:

(tags:"=fileas-search" not tags:"=fileas-search-comma") l:en not a:~^\(\\p{L}\\p{M}*\)+$
3050 items

(tags:"=fileas-search" l:en) not a:~^\(\\p{L}\\p{M}*\)+$
10373 items

So, that's definitely not unusual for publishers to use the FN LN pattern in file-as metadata. One third at least.

(l:en is alias for language:english; BTW the FN LN pattern was found to be frequent if the language is not english)

[DNSB]

I remember going through my library a couple of years back using Quality Check on the author names/sort. Out of ~14,000 ebooks, 130 had an error in the author/author sort. I corrected those and currently show no issues.

For any new books I do individual metadata updates. One recent anthology for my spouse showed FaRo Authors as the author and author sort. Following my usual habit that was replaced with the full list of authors just to make it more searchable.

[Comfy.n]

hmm, let me correct my stats:

(l:en tags:"=fileas-search-comma") not a:~^\(\\p{L}\\p{M}*\)+$
13345 items - 47%

(l:en not tags:"=fileas-search-comma") not a:~^\(\\p{L}\\p{M}*\)+$
15021 items - 53%

l:en not a:~^\(\\p{L}\\p{M}*\)+$
28366 items

So... it's actually kind of split who uses file-as FN LN - LN, FN in this sample universe.

[ShellShock]

The possible formats of author names is so varied, especially with initials versus full names, multiple authors etc, that it is safer for Calibre to assume that if an author name is defined in the OPF it should use it and not override it with what may be an incorrectly derived value.

[JSWolf]

Originally Posted by ShellShock

The possible formats of author names is so varied, especially with initials versus full names, multiple authors etc, that it is safer for Calibre to assume that if an author name is defined in the OPF it should use it and not override it with what may be an incorrectly derived value.

But if it matches the author name, it's incorrect. If it's all caps it's incorrect. Those are situations that should be fixed by calibre when getting the metadata from the OPF.