I have a problem sorting some data in an Elastic Search index.
My data in ES are documents with a default name and possibly translation of the name can be found in a translations nested structure:
[
{
"defaultName": "item1 default name",
"translations": [
{
"languageCode": "en",
"name": "Item1 in english"
},
{
"languageCode": "da",
"name": "Item1 in danish"
}
]
},
{
"defaultName": "item2 default name",
"translations": [
{
"languageCode": "en",
"name": "Item2 in english"
}
]
},
{
"defaultName": "item3 default name",
"translations": []
}
]
When I perform a blank search I need to sort the search result by default name or translated name depending on whether or not the name has been translated or not.
When I search in english (en) the sorting should be done between:
- Item1 in english
- Item2 in english
- Item3 default name
When I search in danish (da) the sorting should be done between:
- Item1 in danish
- Item2 default name
- Item3 default name
I have tried much to get there, but I would like hear if this is possible at all? and maybe a way to do it. I believe I cannot calculate custom scores in this scenario as the sorting is based on text comparison.
If everything fails my fallback solution is to populate non-existing translations with the default translation, but I don't like this solution as it seems much more clean to avoid duplicate data.
Can anyone help me out? 