Create the entire dataset as JSON array, e.g.
[
{
"id": "123456",
"country": "US",
"locale": "en_us"
},
{
"id": "7890",
"country": "UK",
"locale": "en_gb"
}
]
base64 encode it and pass it as a parameter, e.g.
GET endpoint.com/entity?set=BASE64_ENCODED_DATASET
then decode on the server and parse the JSON array into Java objects using perhaps Spring Boot.
Based on the valid URL size comment (although 2000 is usable), you could put the data in a header instead, which can be from 8-16kb depending on the server. GETting multiple resources at once is going to involve compromise somewhere in the design.
As Base64 can contain +/=
you can url encode it too although I haven't found the need to do this in practice when using this technique in SAML.
Another approach would be to compromise on searching via country and locale specific IDs:
GET endpoint.com/entity/{country}/{locale}/{id_csv}
so you would search like this:
GET endpoint.com/entity/US/en_us/123456,0349,23421
your backend handles (if using Spring) as @PathParam
for {country}
and {locale}
and it splits {id_csv}
to get the list of IDs for that country/locale combination.
To get another country/locale search:
GET endpoint.com/entity/UK/en_gb/7890,234,123232
URLs are much smaller but you can't query the entire dataset in one go as you need to query based on country/locale each time.