Option to check similarity with unused/extraneous characters

[whamer100]

When doing searches on a QuickScore class, if a query contains extra characters than the options, it will simply fail, even if its just one character off.

Example: Wanted word in list is "Apple" and the query is "Xpple", even though its one character away from being a 1.0, it will fail to have it in results.

[fwextensions]

Thanks for the feedback. The original Quicksilver algorithm that QuickScore is based on always required an exact, in-order match, even if the matches were widely spaced. While allowing for typos may have some benefit in theory, in practice I've found it usually returns too many nonsensical results when searching across many long strings. URLs, in particular, often have long tokens that contain almost the entire alphabet, so allowing for typos can mean that almost any URL matches.

Fuse.js is a string scoring library that allows for typos or out-of-order matches, and you can compare it against QuickScore on the demo page. To my mind, the ordering and number of its results aren't as useful.

It may be possible to adjust the Quicksilver algorithm to ignore characters in a query that aren't in a target string while giving it a lower score, but I don't see how to allow for out-of-order characters. So xpple might get a non-zero score against apple, but paple wouldn't. But I'm not sure how useful that would be in practice.