Question

我正在使用Presto，并尝试从嵌套的json结构中从'source'='dd'提取所有'id'，如下所示。

SELECT name from (
SELECT name, row_number() over (partition by name) as RN from TEST group by test having count(Name) >= 1 ) a where RN <= 2

期望将ID [apple1，apple3]提取到Presto中的一列中我想知道在Presto Query中实现此目标的正确方法是什么？

Answer 1

如果数据具有发布示例中所示的规则结构，则可以使用parsing the value as JSON的组合，将其强制转换为结构化的SQL类型（数组/地图/行），并使用{{3} }到array processing functions，filter并提取所需的元素：

WITH data(value) AS (VALUES '{
  "results": [
    {
      "docs": [
        {
          "id": "apple1",
          "source": "dd"
        },
        {
          "id": "apple2",
          "source": "aa"
        },
        {
          "id": "apple3",
          "source": "dd"
        }
      ],
      "group": 99806
    }
  ]
}'),
parsed(value) AS (
  SELECT cast(json_parse(value) AS row(results array(row(docs array(row(id varchar, source varchar)), "group" bigint))))
  FROM data
)
SELECT 
  transform(                                        -- extract the id from the resulting docs
    filter(                                         -- filter docs with source = 'dd'
        flatten(                                    -- flatten all docs arrays into a single doc array
             transform(value.results, r -> r.docs)  -- extract the docs arrays from the result array
        ),
        doc -> doc.source = 'dd'),
    doc -> doc.id)
FROM parsed

上面的查询产生：

      _col0
------------------
 [apple1, apple3]
(1 row)

查询以从Presto中深度嵌套的json数组对象中提取ID

1 个答案: