Hi. Yes this is the default behavior as we try to keep native JSON types. I think we might support an option to opt-out from this feature but in the meantime you can use setItem(key, '"123")

String():

String('1') // '1'
String(true) // 'true'
String(null) // 'null'
String('[]') // '[]'

JSON.stringify & JSON.parse :

JSON.stringify('1') // '"1"'
JSON.stringify(true) // 'true'
JSON.stringify(null) // 'null'
JSON.stringify('[]') // '"[]"'


JSON.parse(JSON.stringify('1')) //  '1'
JSON.parse(JSON.stringify(true)) // true
JSON.parse(JSON.stringify(null)) // null
JSON.parse(JSON.stringify('[]')) // '[]'

idb-keyval:

import { set, get } from 'idb-keyval'

await set('foo', '[]')
await get('foo') // '[]'

await set('bar', null)
await get('bar') // null

Inconsistent behavior of String() and JSON.stringify() will lead to confusion, which makes it impossible to migrate directly from idb-keyval to unstorage

I use jsonr to store the compressed json into storage. Because the value obtained by getItem is not the value originally stored, an error is reported.

@pi0

I tried using JSON.stringify to handle the string value, but this broke too many test cases

Following up on #591, I think we can at least expose a new option for setItem as setItem(key, value, { preserveTypes: true }) (or global unstorage option) and default it to true in v2.