Elasticsearch date range aggregation over multiple fields - date

Question

My documents' structure look like this:
{
"element": "A",
"date": "2014-01-01",
"valid_until": "2014-02-01"
},
{
"element": "A",
"date": "2014-02-01",
"valid_until": "9999-12-31"
}
The date "9999-12-31" is here to say: "it has not yet expired". There is always range like this, so for a given element "A", date > valid_until can never overlaps. I can therefore count how much element I have by using the pseudo-code like this: COUNT elements WHERE date < date_to_count AND valid_until >= date_to_count
Where "date_to_count" is the date at which I want to count the values for. As I want to calculate this at several points in time, I could either use a date histogram, or a date range aggregation. However, the date range does seem to work only with one kind of field. Ideally, I'd like to be able to do that:
"aggs": {
"foo": {
"date_range": {
"fields": ["date", "valid_until"],
"ranges": [
{"from": "2014-01-01", "to": {"2014-02-01"}},
{"from": "2014-02-01", "to": {"2014-03-01"}},
{"from": "2014-03-01", "to": {"2014-04-01"}}
]
}
}
}
Where the "date" will be used for "from", and the "valid_until" would be used for "to".
I've tried several other ideas with script, but can't find an efficient way to do it this way :/.
I think I could also workaround this if, in a script, I could have access to the current from/to values, but once again, I've tried things like "ctx.to", "context.to", but those variables are undefined.
Thanks!

Answer 1

Since both the date_range and date_histogram aggregations work on a single field, I do not think you can achieve your goal with an aggregation. But if you don't have too many date ranges that you need to query for, you could call the count API with a query for each date range. That would look something like this:
"query": {
"filtered": {
"filter": {
"bool" {
"must": [
{ "range": { "date": { "gte": "2014-01-01" }}},
{ "range": { "valid_until": { "lt": "2014-02-01" }}}
]
}
}
}
}

Answer 2

I was facing the same problem, and wanted to address this by using one single query. Here is the solution that works for me in Elasticsearch 5.2
"aggs": {
"range1": {
"date_range": {
"fields": "date",
"ranges": [
{"from": "2014-01-01", "to": {"2014-02-01"}},
{"from": "2014-02-01", "to": {"2014-03-01"}},
{"from": "2014-03-01", "to": {"2014-04-01"}}
]
},
"range2": {
"date_range": {
"field": "valid_until",
"ranges": [
{"from": "2014-01-01", "to": {"2014-02-01"}},
{"from": "2014-02-01", "to": {"2014-03-01"}},
{"from": "2014-03-01", "to": {"2014-04-01"}}
]
}
}
}