Question

我在BigQuery上运行以下查询：

WITH allTables AS ( 
  SELECT 
    CONCAT(user_dim.app_info.app_id, ':', user_dim.app_info.app_platform) AS app,
    user_dim.app_info.app_instance_id AS users
  FROM `dataset1.app_events_*`, UNNEST(event_dim) AS event
  WHERE _TABLE_SUFFIX BETWEEN '20170406' AND '20170406'
  OR _TABLE_SUFFIX BETWEEN 'intraday_20170406' AND 'intraday_20170406'

  UNION ALL 
  SELECT 
    CONCAT(user_dim.app_info.app_id, ':', user_dim.app_info.app_platform) AS app,
    user_dim.app_info.app_instance_id AS users
  FROM `dataset2.app_events_*`, UNNEST(event_dim) AS event
  WHERE _TABLE_SUFFIX BETWEEN '20170406' AND '20170406'
  OR _TABLE_SUFFIX BETWEEN 'intraday_20170406' AND 'intraday_20170406'
) 
SELECT 
  app AS target,
  COUNT(DISTINCT(users)) AS datapoint_value,
  UNIX_SECONDS(PARSE_TIMESTAMP('%Y%m%d', '20170406')) AS datapoint_time
FROM allTables
GROUP BY app

最终查询会比这要大得多，但这只是一个简单的例子。我遇到的问题是，如果不满足where条件，则不会返回任何内容。我想更改此查询，以便在不满足where时，它将返回不同的数据。有没有办法在BigQuery中执行此操作？任何帮助都会很棒，谢谢！

Answer 1

不确定，但使用where条件作为app和用户的case语句可能会有所帮助。手动将日期作为时间戳插入并解析，以确保您获取每个子查询的数据。

WITH allTables AS ( 
  SELECT 
    CASE WHEN _TABLE_SUFFIX BETWEEN '20170406' AND '20170406' 
         OR _TABLE_SUFFIX BETWEEN 'intraday_20170406' AND 'intraday_20170406' 
         THEN CONCAT(user_dim.app_info.app_id, ':', user_dim.app_info.app_platform) 
         ELSE NULL END AS app,
    CASE WHEN _TABLE_SUFFIX BETWEEN '20170406' AND '20170406' 
         OR _TABLE_SUFFIX BETWEEN 'intraday_20170406' AND 'intraday_20170406' 
         THEN user_dim.app_info.app_instance_id 
         ELSE NULL END AS users,
    '20170406' as timestamp
  FROM `dataset1.app_events_*`, UNNEST(event_dim) AS event   
  UNION ALL 
  SELECT 
    CASE WHEN _TABLE_SUFFIX BETWEEN '20170406' AND '20170406' 
         OR _TABLE_SUFFIX BETWEEN 'intraday_20170406' AND 'intraday_20170406' 
         THEN CONCAT(user_dim.app_info.app_id, ':', user_dim.app_info.app_platform) 
         ELSE NULL END AS app,
    CASE WHEN _TABLE_SUFFIX BETWEEN '20170406' AND '20170406' 
         OR _TABLE_SUFFIX BETWEEN 'intraday_20170406' AND 'intraday_20170406' 
         THEN user_dim.app_info.app_instance_id 
         ELSE NULL END AS users,
    '20170406' as timestamp
  FROM `dataset2.app_events_*`, UNNEST(event_dim) AS event
) 
SELECT 
  app AS target,
  COUNT(DISTINCT(users)) AS datapoint_value,
  UNIX_SECONDS(PARSE_TIMESTAMP('%Y%m%d', timestamp)) AS datapoint_time
FROM allTables
GROUP BY app

Answer 2

结束不需要为我的项目执行此操作，而是我只是每天为表运行一个插入，然后更新这些表。

在非匹配where子句时返回不同的数据？

2 个答案: