SQL大查询脚本问题

Question

你能帮我吗，我不明白我的 WHERE 有什么问题。我尝试了几件事，但无法得出正确的结论。

SELECT 
CONCAT(fullvisitorId, "-", CAST(visitstarttime AS STRING)) as session,
RANK() OVER(PARTITION BY CONCAT(fullvisitorId, "-", CAST(visitstarttime AS STRING)) ORDER BY hitNumber ASC) AS Rank, 
(SELECT MAX(cd.value) FROM hits.customDimensions as cd where cd.index = 12) as cd1,
(SELECT cd.value FROM hits.customDimensions as cd where cd.index=29) as division,
hits.hitnumber

FROM 
  `dl-training-bigquery.NDA_CHAN*******51.ga_sessions_20191202` as session,
  unnest(hits) as hits
  
WHERE division = 'fsh'
ORDER BY hitnumber ASC

enter image description here

Answer 1

division 似乎是在select 中定义的列别名。

这在 SQL 中是不允许的（一般来说，不仅仅是 BigQuery）。使用子查询或 CTE 定义 division。然后在外部查询中过滤。

Answer 2

您不能在同一范围内重用SELECT 子句中定义的列（ORDER BY 子句除外）。在这里，最简单的方法可能是将子查询移动到FROM 子句：

SELECT 
    CONCAT(fullvisitorId, "-", CAST(visitstarttime AS STRING)) as session,
    RANK() OVER(PARTITION BY fullvisitorId, visitstarttime ORDER BY hitNumber) as rank, 
    dim1.cd1,
    dim2.division,
    hits.hitnumber
FROM `dl-training-bigquery.NDA_CHAN*******51.ga_sessions_20191202` as session
CROSS JOIN unnest(hits) as hits
CROSS JOIN (SELECT MAX(cd.value) as cd1 FROM hits.customDimensions as cd where cd.index = 12) as dim1
CROSS JOIN (SELECT cd.value as division FROM hits.customDimensions as cd where cd.index = 29) as dim2  
WHERE dim2.division = 'fsh'
ORDER BY hitnumber ASC

注意我简化了窗口函数的PARTITION BY子句：我认为你这里不需要字符串连接，你可以使用两级分区。

我还将隐式连接更改为显式cross joins。这不会改变逻辑，但在我看来使意图更清晰。