BigQuery 中的 UNION ALL 或 CONCATENATE 数据集答案

【问题标题】：UNION ALL or CONCATENATE Datasets in BigQueryBigQuery 中的 UNION ALL 或 CONCATENATE 数据集
【发布时间】：2017-05-17 14:22:45
【问题描述】：

我正在使用 BigQuery 控制台，需要将 12 个不同的数据集合并，但信息相同，只需更改 de dataset_id，因为日期范围对所有数据集都相同。

我尝试将union all 函数放在第一个查询的末尾，然后放在另一个查询的后面，但不起作用。

Error: SELECT list expression references hits.contentgroup.contentgroup2 which is neither grouped nor aggregated at [2:3]

这是查询：

SELECT
  hits.contentgroup.contentgroup2 CampaignGrouping,
  custd.value member_PK,
  'Web' Canal,
  'ES' AS country_id,
  SUM(hits.contentGroup.contentGroupUniqueViews2) VistasUnicas
FROM
  `id_project.11773102.ga_sessions*`,
  UNNEST(customdimensions) custd,
  UNNEST(hits) AS hits
WHERE
  1 = 1
  AND PARSE_TIMESTAMP('%Y%m%d', REGEXP_EXTRACT(_table_suffix, r'.*_(.*)')) BETWEEN TIMESTAMP('2017-04-25') AND TIMESTAMP('2017-04-30')
  AND custd.index=30
  and hits.contentGroup.contentGroup2 <> '(not set)'
  AND custd.value <> 'null'
  AND hits.contentGroup.contentGroupUniqueViews2 IS NOT NULL
UNION ALL
SELECT
  hits.contentgroup.contentgroup2 CampaignGrouping,
  custd.value member_PK,
  'Web' Canal,
  'ES' AS country_id,
  SUM(hits.contentGroup.contentGroupUniqueViews2) VistasUnicas
FROM
  `id_project.11773102.ga_sessions*`,
  UNNEST(customdimensions) custd,
  UNNEST(hits) AS hits
WHERE
  1 = 1
  AND PARSE_TIMESTAMP('%Y%m%d', REGEXP_EXTRACT(_table_suffix, r'.*_(.*)')) BETWEEN TIMESTAMP('2017-04-25') AND TIMESTAMP('2017-04-30')
  AND custd.index=30
  and hits.contentGroup.contentGroup2 <> '(not set)'
  AND custd.value <> 'null'
  AND hits.contentGroup.contentGroupUniqueViews2 IS NOT NULL
 GROUP BY
  1, 2
ORDER BY 5 ASC

谢谢。

【问题讨论】：

您遇到了什么错误（如果有）。解释“不起作用” - IOW，什么不起作用。
抱歉@SloanThrasher 我现在包含在问题中，
所以，错误信息似乎很清楚。 hits.contentgroup.contentgroup2 来自哪里？
来自我在查询中声明的表。当我在没有联合的情况下运行查询时，运行！我不知道为什么会这样......

标签： mysql google-bigquery

【解决方案1】：

您需要在联合中的第一个查询中使用GROUP BY，例如：

SELECT
  hits.contentgroup.contentgroup2 CampaignGrouping,
  custd.value member_PK,
  'Web' Canal,
  'ES' AS country_id,
  SUM(hits.contentGroup.contentGroupUniqueViews2) VistasUnicas
FROM
  `bigquery-aaaaa-162814.11773102.ga_sessions*`,
  UNNEST(customdimensions) custd,
  UNNEST(hits) AS hits
WHERE
  1 = 1
  AND PARSE_TIMESTAMP('%Y%m%d', REGEXP_EXTRACT(_table_suffix, r'.*_(.*)')) BETWEEN TIMESTAMP('2017-04-25') AND TIMESTAMP('2017-04-30')
  AND custd.index=30
  and hits.contentGroup.contentGroup2 <> '(not set)'
  AND custd.value <> 'null'
  AND hits.contentGroup.contentGroupUniqueViews2 IS NOT NULL
GROUP BY 1, 2
UNION ALL
SELECT ...

作为UNION ALL 和GROUP BY 的具体示例：

#standardSQL
WITH T AS (
  SELECT 1 AS x, 'foo' AS y UNION ALL
  SELECT 1, 'bar' UNION ALL
  SELECT 2, 'foo'
)
SELECT x, STRING_AGG(y, ',') AS y
FROM T
GROUP BY x
UNION ALL
SELECT SUM(x), y
FROM T
GROUP BY y;

【讨论】：