Quote:
Originally Posted by itimpi
Definitely a great feature.
A query on expected behavior: I had two books today where the title was identical, but the author was stored "lastname, firstname" one one book and "firstname lastname" on the other. Should these have been detected as duplicates and merged or is this potentially a new case to test for?
|
I've written two similar functions. The newest is the Merge feature, found under Edit Metadata. That function always keeps the author from the first selected book, unless that author is "Unknown", in which case it uses the second author and overwrites the first "Unknown" (unless the second author is also "Unknown", in which case .... 3...4...5 )
You're asking about the earlier function I wrote, which automatically puts a new format of a book into the record of a previous book when the two books have the same author, and a similar enough title. That function only works for exact author matches (or IIRC, possibly ignores case). No attempt was made to match "similar" authors, although the "swap author firstname, lastname" option is used before the match attempt is made.
When I wrote it, I thought about matching similar author names. I also thought about automatically doing a firstname, lastname swap when there is a comma in the author name, but I decided to leave that for later ... partly because I wasn't sure how much Kovid would change things. (For example, I originally had the new book automatically overwrite an existing same format. Now it does not. I was very concerned that fuzzy matching an author might inadvertently overwrite a book format, so I required exact author matching).
I see the author matching with a comma as a reasonable new case to test for, but there is an option that "sort of" does that - the swap option for author name. Plus, it's similar to the question of whether exact author name matches should be required. Do we want "Arthur C. Clarke" to match "Arthur Clarke?" How about "Arthur C Clarke" or "Art Clarke."
In the end, I decided I would write the new Merge function, believing that manually using it to solve any problems with the older one would be the easiest and safest.