如何在bigquery中使用group_concat生成json字符串？

Question

我有一张桌子：

user   device   minutes
====   ======   =======
ben    phone    1.9
ben    tablet   2.3
anna   phone    1.6

我希望用户将设备的使用打包成json字符串，如：

user   json
====   =====
ben    {"phone":1.9, "tablet":2.3}
anna   {"phone":1.6, "tablet":2.3}

我怎样才能在bigquery中做到这一点？

Answer 1

下面是BigQuery Standard SQL的版本

#standardSQL
WITH `yourproject.yourdataset.yourtable` AS (
  SELECT 'ben' user, 'phone' device, 1.9 minutes UNION ALL
  SELECT 'ben', 'tablet', 2.3 UNION ALL
  SELECT 'anna', 'phone', 1.6 
)
SELECT 
  user, 
  CONCAT('{', STRING_AGG(CONCAT('"',device, '":', CAST(minutes AS STRING))), '}') json,
  ROUND(SUM(IF(device='tablet',minutes,0)),2) tablet_minutes
FROM `yourproject.yourdataset.yourtable`
GROUP BY user 

输出如下

user    json                        tablet_minutes   
ben     {"phone":1.9,"tablet":2.3}  2.3  
anna    {"phone":1.6}               0.0  

Answer 2

使用BigQuery Legacy SQL

# legacySQL
SELECT
  user,
  json,
  JSON_EXTRACT_SCALAR(json, '$.tablet') AS tablet_minutes
FROM (
  SELECT
    user,
    CONCAT('{', GROUP_CONCAT_UNQUOTED(CONCAT('"', device,'":', minutes)), '}') AS json,
  FROM 
    (SELECT 'ben' user, 'phone' device, '1.9' minutes),
    (SELECT 'ben' user, 'tablet' device, '2.3' minutes),
    (SELECT 'anna' user, 'phone' device, '1.6' minutes),
  GROUP BY
    user)

使用标准SQL（感谢Mikhail）

# standardSQL
WITH `yourproject.yourdataset.yourtable` AS (
  SELECT 'ben' user, 'phone' device, 1.9 minutes UNION ALL
  SELECT 'ben', 'tablet', 2.3 UNION ALL
  SELECT 'anna', 'phone', 1.6 
)
SELECT 
  user, 
  CONCAT('{', STRING_AGG(CONCAT('"',device, '":', CAST(minutes AS STRING))), '}') json,
  ROUND(SUM(IF(device='tablet',minutes,0)),2) tablet_minutes
FROM `yourproject.yourdataset.yourtable`
GROUP BY user 

两个输出

user    json                        tablet_minutes
====    ====                        ==============   
ben     {"phone":1.9,"tablet":2.3}  2.3  
anna    {"phone":1.6}               0.0