在Postgres中使用嵌套数组高效查询JSON

Question

I'm trying to query a JSON data structure in Postgres (9.6 on Amazon RDS). The data contains an array of objects, which in turn has an element that contains an array of objects. I'd like to find all records that match a key + value in one of those nested arrays.

Given rows like this:

{"drinkers" : [
  {"name" : "geoff", 
   "beers" : [
     {"name": "PBR"},
     {"name" : "Bud Select"}
    ]}, 
  {"name" : "tom", 
   "beers" : [
     {"name": "Bud Light"},
     {"name" : "Busch"}
    ]}
]}

I want to find all rows where there's a drinkers.beers object that's name is "PBR". The closest I've come is this:

select jsonb_data 
from bars 
where jsonb_array_elements(jsonb_array_elements(data -> 'drinkers') -> 'beers') ->> 'name' = 'PBR'``

But this does not work since the where returns a list instead of a single true/false to match on. I've come up other solutions using sub queries, lateral joins, etc but all these solutions have performance issues, even with proper gin indexes. Any suggestions on how to query a data structure like this in Postgres?

Answer 1

Did you try using the IN operator?

WHERE 'PBR' IN (
  SELECT jsonb_array_elements(jsonb_array_elements(data -> 'drinkers') -> 'beers') ->> 'name'
);

Another approach would be to use JSONB contain (ie @> ):

WHERE data -> 'drinkers' @> '[{"beers": [{"name": "PBR"}]}]';

Answer 2

对不起，这应该只是一个注释，但如果返回名称的语句给你一个列表或数组，你可以使用'PBR' = ANY jsonb_array_elements(... ？