Calibre's merge algorithm on title and author(s)

[feuille]

While working on a fork of the isfdb2 metadata import plugin, I stumbled again upon Caliber's metadata merge algorithm.

The publication search in the ISFDB database with http://www.isfdb.org/cgi-bin/adv_search_results.cgi?ORDERBY=pub_title&START=0&T YPE=Publication&USE_1=pub_title&OPERATOR_1=contain s&TERM_1=Zur+besonderen+Verwendung&USE_2=author_ca nonical&OPERATOR_2=contains&TERM_2=K.+H.+Scheer&CO NJUNCTION_1=AND
finds 7 publications, all with the same "title record" (in the ISFDB nomenclature), of course with the same author and the same (book) title, but with different publishers, publication dates, covers and slightly different content. There are 4 books of them in my Calibre library and of course I want to pass the correct metadata to the correct book.

But Calibre's merge algorithm for title and author(s) merged the 7 results into one (see merge_metadata_results() in https://github.com/kovidgoyal/calibre/blob/master/src/calibre/ebooks/metadata/sources/identify.py).

As a workaround in my version of the isfdb2 plugin, I added a sequence number to each title field of the seven publications found before putting it into the result queue, so that the merge algorithm does not take place.

The plugin with the original behavior is on https://github.com/confluence/isfdb2-calibre, my fork with the tricked merge algorithm on https://github.com/bertholdm/isfdb2-calibre.

In the metadata search configuration dialog there is a checkbox "several entries per source". The help text says: "Normally, the metadata download system will keep only a single result per metadata source. This option will cause it to keep all results returned from every metadata source. Useful if you only use one or two sources and want to select individual results from them by hand. Note that result with identical title/author/identifiers are still merged."

I wish if there was another checkbox with the choice "Results with identical title / author will not be merged", or, perhaps the possibility to configure the fields for the merge algorithm (title + author(s) + publishing date + ...).