CROSSTAB PostgreSQL - Oracle 中 PIVOT 的替代方案

Question

我正在将 Oracle 数据透视表的查询迁移到 PostgreSQL 交叉表。

create table(cntry numeric,week numeric,year numeric,days text,day text);
insert into x_c values(1,15,2015,'DAY1','MON');
...
insert into x_c values(1,15,2015,'DAY7','SUN');
insert into x_c values(2,15,2015,'DAY1','MON');
...
                values(4,15,2015,'DAY7','SUN');

我有 4 周的时间在 table 中处理 28 行这样的行。我的 Oracle 查询如下所示：

SELECT * FROM(select * from x_c)
PIVOT (MIN(DAY) FOR (DAYS) IN
   ('DAY1' AS DAY1 ,'DAY2' DAY2,'DAY3' DAY3,'DAY4' DAY4,'DAY5' DAY5,'DAY6' DAY6,'DAY7' DAY7 ));

结果：

cntry|week|year|day1|day2|day3|day4|day4|day6|day7|
---------------------------------------------------
   1 | 15 |2015| MON| TUE| WED| THU| FRI| SAT| SUN|
...
   4 | 18 |2015| MON| ...

现在我写了一个这样的 Postgres 交叉表查询：

select *
from crosstab('select cntry,week,year,days,min(day) as day
               from x_c
               group by cntry,week,year,days'
             ,'select distinct days from x_c order by 1'
             ) as (cntry numeric,week numeric,year numeric
                  ,day1 text,day2 text,day3 text,day4 text, day5 text,day6 text,day7 text);

我只得到一行作为输出：

  1|17|2015|MON|TUE| ...   -- only this row is coming

我哪里做错了？

Answer 1

我从 http://www.postgresonline.com/journal/categories/24-tablefunc

中得到了这个

  select year_wk_cntry.t[1],year_wk_cntry.t[2],year_wk_cntry.t[3],day1,day2,day3,day4,day5,day6,day7
   from crosstab('select  ARRAY[country :: numeric,week,year] as t,days,min(day) as day
                            from x_c group by country,week,year,days order by 1,2
                            ','select distinct days from x_c order by 1')
                            as year_wk_cntry (t numeric[],day1 text,day2 text,day3 text,
         day4 text, day5 text,day6 text,day7 text);

谢谢！！

Answer 2

ORDER BY 在您的原始查询中缺失。手册：

In practice the SQL query should always specify ORDER BY 1,2 to ensure that the input rows are properly ordered, that is, values with the same row_name are brought together and correctly ordered within the row.

更重要的是（也更棘手），crosstab() 恰好需要 one row_name 列。这个密切相关的答案中的详细解释：

Crosstab splitting results due to presence of unrelated field

是将多个列嵌套在一个数组中，然后再取消嵌套。这是不必要的昂贵、容易出错和有限的（仅适用于具有相同数据类型的列，否则您需要转换并可能丢失正确的排序顺序）。

相反，使用 rank() 或 dense_rank()（在我的示例中为 rnk）生成代理项 row_name 列：

SELECT cntry, week, year, day1, day2, day3, day4, day5, day6, day7
FROM   crosstab (
  'SELECT dense_rank() OVER (ORDER BY cntry, week, year)::int AS rnk
        , cntry, week, year, days, day
   FROM   x_c
   ORDER  BY rnk, days'
 , $$SELECT unnest('{DAY1,DAY2,DAY3,DAY4,DAY5,DAY6,DAY7}'::text[])$$
   ) AS ct (rnk int, cntry int, week int, year int
          , day1 text, day2 text, day3 text, day4 text, day5 text, day6 text, day7 text)
ORDER  BY rnk;

我对输出列 cntry、week、year 使用数据类型 integer，因为这似乎是（更便宜的）合适的类型。您也可以像以前一样使用数字。

这里是交叉表查询的基础知识：

PostgreSQL Crosstab Query

CROSSTAB PostgreSQL - Oracle 中 PIVOT 的替代方案

CROSSTAB PostgreSQL - Alternative for PIVOT in Oracle

postgresql

crosstab

postgresql-9.4