将不同的 SQL 脚本合并到一个脚本中答案

【问题标题】：Merging different SQL scripts into one script将不同的 SQL 脚本合并到一个脚本中
【发布时间】：2017-06-06 14:20:13
【问题描述】：

我有 3 个查询脚本在同一个表上工作；一个是计算总行数，另一个是计算重复行的总数，另一个是计算不同（唯一）行的总数。

统计总行数：

select count(*) as total_rows from <table_name>

计算重复行数：

select COUNT (<column_name>) as total_duplicate from <table_name> group by (<column_name>) HAVING count(<column_name>) > 1

计算唯一行数：

select COUNT(DISTINCT <column_name>) as total_unique from <table_name>

我正在尝试将所有 3 个合并到一个脚本中。我能够为总计数和唯一计数做到这一点；但是，我不确定如何包含重复计数查询，因为它有一个 group by 语句。

请有什么想法。

【问题讨论】：

标签： sql hive

【解决方案1】：

尝试这样做：

select count(*) as total_rows,COUNT(DISTINCT <column_name>) as total_unique, 
    (select COUNT (<column_name>) 
     from <table_name> 
     group by (<column_name>) 
     HAVING count(<column_name>) > 1) as total_duplicate
from <table_name>

【讨论】：

如所写，您将获得“子查询返回超过 1 个值”。有多个重复项时出错。
@BaconBits - 不，甚至没有。至少目前，Hive 不支持 SELECT 子句中的子查询

【解决方案2】：

select      sum(rows_per_col)   as total_rows
           ,count (is_dup)      as total_duplicate
           ,count(*)            as total_unique

from       (select      column_name
                       ,case when count(*) > 1 then 1 end   as is_dup
                       ,count(*)                            as rows_per_col
            
            from        table_name
            
            group by    column_name
            ) t

演示

with table_name as (select explode(array(1,1,1,2,3,3,4,4,4,4,4,5)) as column_name)

select      sum(rows_per_col)   as total_rows
           ,count (is_dup)      as total_duplicate
           ,count(*)            as total_unique

from       (select      column_name
                       ,case when count(*) > 1 then 1 end   as is_dup
                       ,count(*)                            as rows_per_col

            from        table_name

            group by    column_name
            ) t

+------------+-----------------+--------------+
| total_rows | total_duplicate | total_unique |
+------------+-----------------+--------------+
|         12 |               3 |            5 |
+------------+-----------------+--------------+

【讨论】：

【解决方案3】：

您可以将这些合并在一起：

 select count(*) as count_total
        'total_rows' as count_type
 from <table_name>

 union

 select count (<column_name>)as count_total
        'total_duplicate' as count_type
 from <table_name>
 group by (<column_name>)
 HAVING count(<column_name>) > 1

 union

 select COUNT(DISTINCT <column_name>) as count_total
        'total_unique' as count type
 from <table_name>

【讨论】：