Elasticsearch字词过滤器，但带前缀

Question

I have a field that I want to filter for multiple values. I'm currently doing this with a terms filter:

"filter" : {
            "terms" : { "user" : ["kimchy", "elasticsearch"]}
        }

But I want the filter to match on prefixes as well as whole terms, something like this ( prefixes is just me pseudocoding what I want):

 "filter" : {
            "prefixes" : { "user" : ["kim", "elast"]}
        }

Or do I have to resort to multiple or filters?

"filter" : {
            "or" : {
                "filters" : [
                    {
                        "prefix" : { "user" : "kim" }
                    },
                    {
                        "prefix" : { "user" : "elast" }
                    }
                ],

Answer 1

ES does not have anything like prefixes but

1) you could use query string query rather than writing multiple or filters

You could write your query as

{
  "query": {
    "query_string": {
      "default_field": "users",
      "query": "kim* OR elas*"
    }
  }
}

There is one problem though, if your user field is mapped as not_analyzed then it wont find Kimchy , Elasticsearch as by default lowercase_expanded_terms is true and it will lowercase all your prefixes.

For this to work 100%, I would recommend using keyword tokenizer with lowercase token filter so that we can make case insensitive search, after that above query will give you desired result. You could add more prefix with OR

2) Another approach could be with keyword tokenizer mixed with lowercase and edgengram token filter . Lets say you use min_gram : 2 and max_gram : 7 then you could use terms query itself act like multiple prefix query like this

"filter": {
    "terms": {
        "user": ["kim", "elast"]
    }
}

Hope this helps!