I've been going through some of my books looking for bad metadata links from Fantastic Fiction and I'm seeing the weaknesses of their search.
Turns out though that they are fairly consistent in their naming, not perfectly consistent, but you can build a link with a calibre template that works a good percentage of the time and is a lot less prone to bad matches. From what I know of their search interface checking it first would probably reduce the number of times caliber has to hit the website.
Any chance of having an option in the Metadata downloader to try a "built" identifier link before using the FF search interface?
Here's my rough build of the template:
Code:
ff= transliterate(
lowercase(
re(
strcat(
shorten(field('author_sort'),1,'',0), '/',
re(sublist(field('authors'),0,1,'&'),'\.',' '), '/',
re(re(field('title'),'&','and'),'(\ba ?\b|\ban ?\b|\bthe ?\b|\.(?= )|\,|’|!|\?|:(?= )|‒|-|…)','')
),'[: \.|\']+','-')
)
);