BigQuery 使用动态键和值提取嵌套 JSON

Question

我想用动态键提取嵌套的 JSON。我目前可以借助和提取密钥，但解析 JSON 对象的值会产生意外的结果。“[对象对象]”12

使用动态键和值解析嵌套 JSON 的正确方法是什么？（我不想使用自定义 JS UDF，但我不确定现有的 JSON 函数是否可以处理该问题。）

记录的输入栏

{
    "key1":{"ItemID":1,"UseCount":4,"ItemCount":7},
    "key2":{"ItemID":2,"UseCount":5,"ItemCount":8},
    "key3":{"ItemID":3,"UseCount":6,"ItemCount":9}
    ...
}

当前查询

bigquery-utils： json_extract_keys() json_extract_values()

WITH
sample_logs AS (
    SELECT '{"key1":{"ItemID":1,"UseCount":4,"ItemCount":7},"key2":{"ItemID":2,"UseCount":5,"ItemCount":8},"key3":{"ItemID":3,"UseCount":6,"ItemCount":9}}' as json_string,
    UNION ALL SELECT '{"key4":{"ItemID":1,"UseCount":4,"ItemCount":7},"key5":{"ItemID":2,"UseCount":5,"ItemCount":8}}'
)
SELECT
    json_string,
    key,
    TO_JSON_STRING(value) as value,
FROM sample_logs
CROSS JOIN UNNEST(bqutil.fn.json_extract_keys(json_string)) as key WITH OFFSET
INNER JOIN UNNEST(bqutil.fn.json_extract_values(json_string)) as value WITH OFFSET USING (OFFSET)
;

结果

预期成绩

JSON_STRING1  |  "key1"   |  {"ItemID":1,"UseCount":4,"ItemCount":7}  -- <- not [object Object]
JSON_STRING1  |  "key2"   |  {"ItemID":2,"UseCount":5,"ItemCount":8}
JSON_STRING1  |  "key3"   |  {"ItemID":3,"UseCount":6,"ItemCount":9}
JSON_STRING2  |  "key4"   |  {"ItemID":1,"UseCount":4,"ItemCount":7}
JSON_STRING2  |  "key5"   |  {"ItemID":2,"UseCount":5,"ItemCount":8}

Answer 1

你可以考虑下面。

WITH
sample_logs AS (
    SELECT '{"key1":{"ItemID":1,"UseCount":4,"ItemCount":7},"key2":{"ItemID":2,"UseCount":5,"ItemCount":8},"key3":{"ItemID":3,"UseCount":6,"ItemCount":9}}' as json_string,
    UNION ALL SELECT '{"key4":{"ItemID":1,"UseCount":4,"ItemCount":7},"key5":{"ItemID":2,"UseCount":5,"ItemCount":8}}'
)
SELECT json_string,
       key,
       PARSE_JSON(json_string)[key] value
  FROM sample_logs, 
UNNEST(bqutil.fn.json_extract_keys(json_string)) as key;

• JSON 函数要求 jsonpath 为字符串文字，而JSON 类型允许对象键为变量。

查询结果