大查询 distinct on 和 group by

Question

继 Select first row in each GROUP BY group? 之后，我正在尝试在 Google 大查询中做一件非常相似的事情。

数据集：fh-bigquery:reddit_comments.2018_01

Aim：对于每个 link_id（Reddit 提交）select created_utc

的第一个评论

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 unique/distinct parent_id(s)

拜托，谢谢！

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

Answer 1

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，但随后您必须将查询表述为 table 和为每个评论找到最早条目的子查询之间的连接：

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;

Answer 2

以下适用于 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

大查询 distinct on 和 group by

Big query distinct on and group by

sql

google-bigquery

reddit