Hello everybody!
Given that types will be dropped in future releases I wanted to move them in a special field for each document (as stated here).
Now for the search query I'll have something like this:
{
"query": {
"bool" : {
"should" : [
{ "match_phrase": { "field1": { "query": "test", "_name": "field1" }}},
{ "match": { "field2": { "query": "test", "_name": "field2"}}},
{ "match_phrase": { "field3": { "query": "test", "_name": "field3"}}},
{ "match": { "field4": { "query": "test", "_name": "field4"}}},
{ "match": { "field5": { "query": "test", "_name": "field5"}}}
],
"filter": {
"match": {
"type_field": "important_type_name"
}
}
}
}
}
The only addition from what I used to have here is the filter for the exact match on the custom field type_field, which has keyword as type in its mapping.
The problem now is that this search will include all documents that have "type_field" = "important_type_name". As stated here in the should section, the should will only act in influencing the score.
How can I make a query that uses the should bool query but that has a strict filter on it? i.e. I want to use the capabilities of should (with fuzziness, boosting and match or match_phrase) on multiple fields but still be sure that it looks only through some already filtered documents, and all the final results will have matched_queries (that's why I used the named queries, to show that right now, some resulted documents don't have any matched queries but are still part of the final result).
Thanks very much in advance!