大查询的不同和分组依据

Question

从选择每个 GROUP BY 组中的第一行开始？我正在尝试在 Google 大查询中做一个非常相似的事情。

数据集：fh-bigquery:reddit_comments.2018_01

目标：对于每个 link_id（Reddit 提交），根据 created_utc 选择第一条评论

SELECT body,link_id 
FROM [fh-bigquery:reddit_comments.2018_01] 
where subreddit_id == "t5_2zkvo"  
group by  link_id ,body, created_utc  
order by link_id ,body,  created_utc desc 

目前它不起作用，因为它仍然没有给我唯一/不同的 parent_id(s)

谢谢，麻烦您了！

---

编辑： 我说 parent_id 是 == 提交是不正确的，它实际上是 link_id

Answer 1

下面是 BigQuery 标准 SQL

#standardSQL
SELECT 
  ARRAY_AGG(body ORDER BY created_utc LIMIT 1)[OFFSET(0)] body, 
  link_id
FROM `fh-bigquery.reddit_comments.2018_01`
WHERE subreddit_id = 't5_2zkvo'
GROUP BY link_id
-- ORDER BY link_id

Answer 2

我们可以在这里使用ROW_NUMBER()：

SELECT body, parent_id, created_utc
FROM
(
    SELECT *, ROW_NUMBER() OVER (PARTITION BY parent_id ORDER BY created_utc) rn
    FROM [fh-bigquery:reddit_comments.2018_01]
    WHERE subreddit_id = 't5_2zkvo'
) t
WHERE rn = 1
ORDER BY parent_id ,body, created_utc DESC;

请注意，您可以继续使用当前的方法，但是您必须将查询表述为表和子查询之间的联接，该子查询查找每个评论的最早条目：

SELECT t1.*
FROM [fh-bigquery:reddit_comments.2018_01] t1
INNER JOIN
(
    SELECT parent_id, MIN(created_utc) AS first_created_utc
    FROM [fh-bigquery:reddit_comments.2018_01]
    GROUP BY parent_id
) t2
    ON t1.parent_id = t2.parent_id AND t1.created_utc = t2.first_created_utc;