autobib: fix sorting of author-less books

[fare]

Convert title using content->string before it may be used as a sortable name by autobib.

[github-actions]

Resyntax analyzed 1 file in this pull request and found no issues.

[github-actions]

Resyntax analyzed 1 file in this pull request and found no issues.

[fare]

@mflatt is this right, or should the content->string be moved to within (one of?) the fields of org-author-name ?

[mflatt]

I think you're right that content->string belongs around the third argument to make-author-element in org-author-names. It looks like the names field was meant to be a string, always.

There's a use of content->string composed with author-element-names on line 708. That was added in commit 6a8986f in reaction to (I think) the same underlying issue. It looks like fixing org-author-names would have been the right change then (but that commit has some other useful corrections, too).

[fare]

I'll move the fix to org-author-name and remove the kluge line 708—though how do I test that? (I'll assume the test that @bennn added is enough.)

Also—should I use define-struct/contract to force names to be a string? What's the contract for cite?

[github-actions]

Resyntax analyzed 1 file in this pull request and found no issues.

[mflatt]

Your revised changes look good to me.

Here's an example that works both before and after your changes, but doesn't work if I only adjust authors to drop the compose1. We could add this to the test documents in "scribble-test/tests/scriblib". (I'm happy to do that after merging your change, if you prefer.)

#lang scribble/base
@(begin
   (require scriblib/autobib)
   (define-cite ~cite citet generate-bibliography)
   (define great-pub
     (make-bib
      #:title "The Publication"
      #:author (authors (org-author-name (list (italic "Great") ", Inc."))
                        "Scribble Volunteer")
      #:date 2026)))

@title{Example}

@citet[great-pub]

@generate-bibliography[]

[fare]

I'm willing to add this change, if you tell me how to test it. Otherwise, LGTM. I think that the test that @bennn added in the commit you mentioned might already test this case, though.