Question

So, while asciifolding would turn è into e it won’t filter out a single accent like `, right? So how can I get rid of them? For example I want to index O`Learys and be able to search for any variation of olearys, O learys, o`learys.

Answer 1

I don’t think it’s possible to make O`Learys to match both olearys and O learys at the same time using analyzers, tokenizers or filters that come with elasticsearch. It would require to detect presence of ` and indexing O`Learys twice: first as one token olearys and then as two tokens o learys. It can be done by writing only with a custom token filter though.

As a workaround, you can make both olearys and o`learys to match O`Learys by removing character ` from the input using Mapping Char Filter.