SQL 查询使用 databricks 分解嵌套 json 中的多个值

Question

我正在尝试使用 databricks sql 获取嵌套 json 的数据，但无法对多个数组列执行爆炸，并且抛出以下错误

[UNSUPPORTED_GENERATOR.MULTI_GENERATOR] 不支持生成器：仅允许使用一台生成器

尝试执行以下查询时

SELECT
explode(array_join(from_json(food:Fruits.children\[*\].`Fruit Quantity`, 'array\<string\>'), ' - ')) AS `Fruit Quantity`,
explode(array_join(from_json(food:Fruits.children\[*\].`Fruit Weight`, 'array\<string\>'), ' - ')) AS `Fruit Weight`,
explode(array_join(from_json(food:Fruits.children\[\*\].`Fruit Packaging`, 'array\<string\>'), ' - ')) AS `Fruit Packaging`
FROM
input_table

输入

| Day | Food |
| -------- | -------- |
| Day 1   | {
  "breakfast": {
    "carbs": [
    {
    "carbs name": "No food"
    }

    ]
  },
  "Fruits": {
    "batch": "batch01",
    "children": [
      {
        "Fruit name": "apple01",
        "Fruit Quantity": 3,
        "Fruit Weight": "50g",
        "Fruit Packaging": "Basket"
      },
      {
        "Fruit name": "apple01",
        "Fruit Quantity": 3,
        "Fruit Weight": "50g",
        "Fruit Packaging": ""
      },
      {
        "Fruit name": "apple02",
        "Fruit Quantity": 5,
        "Fruit Weight": "100g",
        "Fruit Packaging": "Foil"
      }
    ]
  }
}   |

输出

注意：我需要没有任何 CTE 或子查询的 sql。

Answer 1

检查以下查询。

SELECT
  day,  
  FROM_JSON(food, 'Fruits struct<batch string>').Fruits.batch AS Batch,
  children['Fruit name'] AS `Fruit name`,
  children['Fruit Quantity'] AS `Fruit Quantity`,
  children['Fruit Weight'] AS `Fruit Weight`,
  children['Fruit Packaging'] AS `Fruit Packaging`,
  carbs['carbs name'] AS `carbs name`
FROM VALUES 
  ('Day 1 ', '{ "breakfast": { "carbs": [ { "carbs name": "No food" } ] }, "Fruits": { "batch": "batch01", "children": [ { "Fruit name": "apple01", "Fruit Quantity": 3, "Fruit Weight": "50g", "Fruit Packaging": "Basket" }, { "Fruit name": "apple01", "Fruit Quantity": 3, "Fruit Weight": "50g", "Fruit Packaging": "" }, { "Fruit name": "apple02", "Fruit Quantity": 5, "Fruit Weight": "100g", "Fruit Packaging": "Foil" } ] } }') AS (day, food) 
LATERAL VIEW EXPLODE(FROM_JSON(food, 'Fruits struct<children array<map<string,string>>>').Fruits.children) AS children
LATERAL VIEW EXPLODE(from_json(food, 'breakfast struct<carbs array<map<string,string>>>').breakfast.carbs) AS carbs

+------+-------+----------+--------------+------------+---------------+----------+
|day   |Batch  |Fruit name|Fruit Quantity|Fruit Weight|Fruit Packaging|carbs name|
+------+-------+----------+--------------+------------+---------------+----------+
|Day 1 |batch01|apple01   |3             |50g         |Basket         |No food   |
|Day 1 |batch01|apple01   |3             |50g         |               |No food   |
|Day 1 |batch01|apple02   |5             |100g        |Foil           |No food   |
+------+-------+----------+--------------+------------+---------------+----------+