Composite Agg after Date Filter

[adaley222 (Andrew Daley)]

I am trying to count documents over a week, for each hour/day of the week. So essentially, I'd like the output to be 7x24 =168 doc_counts, for a specified date range.

The code below nearly gives me what I need, however, it only returns the top terms when I need all of them.

{"size": 0,
"aggs" : {
		"booked_filter" : {
            "filter" : { "term": {"type":"booked"} },
    "aggs": {
        "range": {
        	"date_range": {
            	"field": "start",
                "format": "date_hour_minute_second",
            	"ranges": [{ "to": "now",  "from": "now-7d"}]},
	"aggs": {
        "day-of-week-count": {
            "terms": {
                "script": "return doc[\"start\"].value.getDayOfWeekEnum();"},
    "aggs": {
        "hour-of_day-count": {
            "terms": {
            	"script": "return doc[\"start\"].value.getHourOfDay();"
                				}
            				}
						}
					}
				}
			}
		}
    }
}
}

The documentation says that instead of just setting the size greater than the cardinality of the field, I should use a composite aggregation. This is what I've tried:

{"size": 0,
"aggs" : {
		"booked_filter" : {
            "filter" : { "term": {"type":"booked"} },
"aggs": {
    	"date-filter": {
    		"filter": { 
    			"range": {
    				"startu": {
    					"gte": 1583440081000,
    					"lte": 1584131281000}}}},
	"aggs": {
		"composite": {
			"sources": [
				{"day-of-week-count": { "terms": {"script": "return doc[\"start\"].value.getDayOfWeekEnum();"}}},
        		{"hour-count": {"terms": {"script": "return doc[\"start\"].value.getHourOfDay();"}}}]
					}
				}
			}
		}
    }
}

However, this gives me the following error:

`"[composite] aggregation cannot be used with a parent aggregation of type: [FilterAggregatorFactory]"

Am I approaching this entirely incorrectly? Or is this something Elasticsearch isn't able to do at the moment?

Adjusting the size of the first query should give me the info I need, but I'd like to know the proper way to handle a use case like this if there is one.