How To Combine Multiple Queries In ElasticSearch - elasticsearch

Question

I am encountering an issue trying to correctly combine elastic search queries, in SQL my query would look something like this:
Select * from ABPs where (PId = 10 and PUId = 1130) or (PId = 30 and PUId = 2000) or (PlayerID = '12345')
I can achieve each of these by themselves and get correct results.
Query A) (PId = 10 and PUId = 1130)
translates to
{
"query": {
"bool": {
"must": [
{
"term": {
"PId": "1366"
}
},
{
"term": {
"PUId": "10"
}
}
]
}
}
}
Query B) (PId = 10 and PUId = 1130)
translates the same as above just with different values
Query C) (PlayerID = '12345')
translates to
{
"query": {
"match": {
"PlayerUuid": "62fe0832-7881-477c-88bb-9cbccdbfb3c3"
}
}
}
I have been trying to figure out how to get all of these into the same ES search query and I am just not having any luck at all and was hoping someone with more extensive ES experience would be able to give me a hand.

Answer 1

You can make use of Bool query using should(Logical OR) and must(Logical AND) clause.
Below is the ES query representation of the clause Select * from ABPs where (PId = 10 and PUId = 1130) or (PId = 30 and PUId = 2000) or (PlayerID = '12345')
POST <your_index_name>/_search
{
"query": {
"bool": {
"should": [
{
"bool": {
"must": [
{
"term": {
"PId": "10"
}
},
{
"term": {
"PUId": {
"value": "1130"
}
}
}
]
}
},
{
"bool": {
"must": [
{
"term": {
"PId": "30"
}
},
{
"term": {
"PUId": "2000"
}
}
]
}
},
{
"bool": {
"must": [
{
"term": {
"PlayerId": "12345"
}
}
]
}
}
]
}
}
}
Note that I'm assuming the fields PId, PUId and PlayerId are all of type keyword.

Answer 2

Wrap all your queries into a should-clause of a bool-query which you put in the filter-clause of another top-level bool-query.
Pseudo-code (as I’m typing on a cell phone):
“bool”: {
“filter”: {
“bool”: {
“should”: [
{query1},
{query2},
{query3}
]
}
}
}
In a bool- query made up of only should-clauses, will make it a requirement that at least one of the queries in the should-clause has to match (minimum_should_match-will be in such a scenario).
Update with the actual query (additional explanation):
POST <your_index_name>/_search
{
"query": {
"bool": {
"filter": {
"bool": {
"should": [
{"bool": {{"must": [ {"term": {"PId": "10"}},{"term": {"PUId": "1130"}} ]}}},
{"bool": {{"must": [ {"term": {"PId": "30"}},{"term": {"PUId": "2000"}} ]}}},
{"term": {"PlayerId": "12345"}}
]
}
}
}
}
}
The example above is wrapping your actual bool-query in a filer-clause of another top-level bool-query to follow best-practices and guarantee for a better performance: whenever you don't care about the score, especially when it's always about exact-matching queries, you should put them into filter-clauses. For those queries Elasticsearch will not calculate a score and therefore can even potentially cache the results of that query for even better performance.