ArangoDB AQL deep array scan

Question

I have a collection of customers with their visited places, organised as follows:

{
  "customer_id": 151,
  "first_name": "Nakia",
  "last_name": "Boyle",
  "visited_places": [
    {
      "country": "Liberia",
      "cities": [
        "Mullerside",
        "East Graham"
      ]
    },
    {
      "country": "Rwanda",
      "cities": [
        "West Kristofer",
        "Effertzbury",
        "Stokeston",
        "South Darionfort",
        "Lewisport"
      ]
    }
  ]
}

I am trying to find all customers that have visited a specific city in a specific country. I've got it working like this:

FOR target IN usertable 
FILTER [] != target.visited_places[* FILTER CURRENT.country == @country AND CONTAINS(CURRENT.cities, @city)]
LIMIT @limit 
RETURN target

The query seems cumbersome and I am not sure if it is performant.

Is there any better way to do this in terms of readability and performance?

Answer 1

You could filter by country and create a persistent array index for that on visited_places[*].country but you still need a secondary condition that ensures that the country and city you look for occur in the same array element:

FOR doc IN usertable
  FILTER @country IN doc.visited_places[*].country
  FILTER LENGTH(doc.visited_places[* FILTER CURRENT.country == @country AND @city IN CURRENT.cities])
  RETURN doc