Question: count all and count in specific date range

Hello, sorry for the noise.I’ve not been able to add a new metric for a specific date range. I need to count two things. The first one is count by month and I could do it. But the second one is to count in the last five months.

The code that is workings is:

now = timezone.now()
five_month_ago = now - relativedelta(months=5)
 
query_banks = Q("terms", bank__id=ids_of_banks)
aggs_by_month = A('date_histogram', field='created_at', interval='month', format='YYYY-MM')
search = search.query("nested", path="bank", query=[1, 2, 3])     
search.aggs.metric('count', 'cardinality', field='id')
search.aggs.bucket('by_month', aggs_by_month)
result = search.execute()

I tried a lot of things, for counting the last five month without success.

aggs_five_month = A('filter', Q('range', created_at={'gt': five_month_ago, 'lt': now}))
search.aggs.bucket('last_five_months', aggs_five_month).metric('count', 'cardinality', field='id')

Is it possible to do both things with one query?

[honzakral]

This should do the trick:

from elasticsearch_dsl import Search
s = Search()
s.aggs.bucket('per_month', 'date_histogram', field='created_at', interval='month')\
    .metric('count', 'cardinality', field='id')
s.aggs.bucket('last_5_months', 'filter', Q('range', created_at={'gt': 'now-5m'}))\
    .metric('count', 'cardinality', field='id')

It will give you a count per month and also for the last 5 months.

Note that you also have some weird constructs with the nested query. If bank is a nested field then you need to query for the bank ids as:

s = s.query('nested', path='bank', query=Q('terms', bank__id=[1, 2, 3]))

if not then just:

s = s.query('terms', bank__id=[1, 2, 3])

Hope this helps!

[eduzen]

👍 😃 Thanks!