pref: add fulltext index for bookmarkEntries.text

[hi-ogawa]

• closes feat: improve bookmark text search #361

todo

• probably ngram is useless for latin language text
• is ordering deterministic?
• check if planetscale supports it
• check perf, index usage
• migration staging

pnpm knex-staging migrate:up
pnpm cli-staging resetBookmarkEntriesTextCharacters

• migration production

pnpm knex-production migrate:up
pnpm cli-production resetBookmarkEntriesTextCharacters

[hi-ogawa]

It seems we cannot search single character unless you can change server config ngram_token_size (which probably we don't for planetscale db).

• https://stackoverflow.com/questions/57231378/mysql-how-to-search-a-single-character-with-fulltext-index

Unfortunately, for CJK usage, it's very natural desire to look up with single character especially for beginner, so I might need to come up with some hack.

EDIT:

Okay, here is a workaround. Only works to search (character)*. Unfortunately, this means we cannot search for common grammatic suffix e.g. ...요, ...고, etc... But probably such need is for super beginner, so it might be fine to not support it.

[hi-ogawa]

skeema put extra trailing space after magic comment.
It's a bit too annoying. Not sure if there's any workaround.

[hi-ogawa]

It feels very redundant but maybe to support single character search, we can add extra "multi valued" index https://dev.mysql.com/doc/refman/8.0/en/create-index.html#create-index-multi-valued