将行转换为列

Question

我有如下数据：

    created_at | 地位    
---------------------+------------
 2016-04-05 1:27:15 | 信息
 2016-04-05 3:27:15 | 信息
 2016-04-05 5:27:15 | 警告
 2016-04-05 10:27:15 | 信息
 2016-04-05 11:27:15 | 警告

有了这些数据，我想转换如下：

状态 | 2016-04-05 1:00:00 | 2016-04-05 4:00:00 | 2016-04-05 8:00:00 | 2016-04-05 12:00:00
---------+--------------------+------------ -+--------------------+--------------------
 信息 | 1 | 1 | 0 | 1
 警告 | 0 | 0 | 1 | 1                  

谁能建议最好的方法来做到这一点？

Answer 1

假设2016-04-05 0:27:15而不是2016-04-05 1:27:15在基础表中，这个问题对我来说更有意义：

CREATE TABLE tbl (created_at timestamp, status text);
INSERT INTO tbl VALUES
  ('2016-04-05 00:27:15', 'info')
, ('2016-04-05 03:27:15', 'info')
, ('2016-04-05 05:27:15', 'warn')
, ('2016-04-05 10:27:15', 'info')
, ('2016-04-05 11:27:15', 'warn');

逻辑是计算发生在并排除下一个边界的事件。这与经常被忽视的功能width_bucket()完美契合。准确地说，它需要Postgres 9.5引入的具有任意边界的变体（因为在 OP 的边界中没有常规模式）。直接从手册解释：

width_bucket(operand anyelement, thresholds anyarray)

operand给定一个列出桶的下限的数组，返回将分配给的桶号；返回0小于第一个下限的输入；所述thresholds阵列必须进行排序，最小第一，或意外的结果，将获得

对于常规存储桶，您也可以使用Postgres 9.1 中提供的另一种变体。
将其与crosstab()重新使用与列名相同的边界结合起来（查询的其余部分适用于 Postgres 9.1）：

SELECT * FROM crosstab(
 $$SELECT status
        , width_bucket(created_at, '{2016-04-05 01:00
                                   , 2016-04-05 04:00
                                   , 2016-04-05 08:00
                                   , 2016-04-05 12:00}'::timestamp[])
        , count(*)::int
   FROM   tbl
   WHERE  created_at < '2016-04-05 12:00'  -- exclude later rows
   GROUP  BY 1, 2
   ORDER  BY 1, 2$$
, 'SELECT generate_series(0,3)'
   ) AS t(status text, "2016-04-05 01:00" int
                     , "2016-04-05 04:00" int
                     , "2016-04-05 08:00" int
                     , "2016-04-05 12:00" int);

结果：

 status | 2016-04-05 01:00 | 2016-04-05 04:00 | 2016-04-05 08:00 | 2016-04-05 12:00
--------+------------------+------------------+------------------+------------------
 info   |                1 |                1 |                  |  1
 warn   |                  |                  |                1 |  1

第二个交叉表参数 ( 'SELECT generate_series(0,3)') 是一个查询字符串，在执行时为每个目标列返回一行。在任一侧找不到的每个值 - 不在原始数据中或不是由第二个参数生成的 - 都会被忽略。

基础知识crosstab()：

• PostgreSQL 交叉表查询

用 0 替换 NULL

如果你需要0而不是NULL在结果中，用 修复COALESCE()，但这只是一个表面问题：

SELECT status
     , COALESCE(t0, 0) AS "2016-04-05 01:00"
     , COALESCE(t1, 0) AS "2016-04-05 04:00"
     , COALESCE(t2, 0) AS "2016-04-05 08:00"
     , COALESCE(t3, 0) AS "2016-04-05 12:00"
FROM   crosstab(
 $$SELECT status
        , width_bucket(created_at, '{2016-04-05 01:00
                                   , 2016-04-05 04:00
                                   , 2016-04-05 08:00
                                   , 2016-04-05 12:00}'::timestamp[])
        , count(*)::int
   FROM   tbl
   WHERE  created_at < '2016-04-05 12:00'
   GROUP  BY 1, 2
   ORDER  BY 1, 2$$
, 'SELECT generate_series(0,3)'
   ) AS t(status text, t0 int, t1 int, t2 int, t3 int);

结果：

 status | 2016-04-05 01:00 | 2016-04-05 04:00 | 2016-04-05 08:00 | 2016-04-05 12:00
--------+------------------+------------------+------------------+------------------
 info   |                1 |                1 |                0 |  1
 warn   |                0 |                0 |                1 |  1

加总

在 Postgres 9.5+ 中添加每次status使用的总数GROUPING SETS

SELECT status
     , COALESCE(t0, 0) AS "2016-04-05 01:00"
     , COALESCE(t1, 0) AS "2016-04-05 04:00"
     , COALESCE(t2, 0) AS "2016-04-05 08:00"
     , COALESCE(t3, 0) AS "2016-04-05 12:00"
     , COALESCE(t4, 0) AS total
FROM   crosstab(
 $$SELECT status, COALESCE(slot, -1), ct  -- special slot for totals
   FROM  (
      SELECT status
           , width_bucket(created_at, '{2016-04-05 01:00
                                      , 2016-04-05 04:00
                                      , 2016-04-05 08:00
                                      , 2016-04-05 12:00}'::timestamp[]) AS slot
           , count(*)::int AS ct
      FROM   tbl
      WHERE  created_at < '2016-04-05 12:00'
      GROUP  BY GROUPING SETS ((1, 2), 1)  -- add totals per status
      ORDER  BY 1, 2
      ) sub$$
 , 'VALUES (0), (1), (2), (3), (-1)'  -- switched to VALUES for more sophisticated series
   ) AS t(status text, t0 int, t1 int, t2 int, t3 int, t4 int);

结果如上，加上：

...  | total
... -+-------
...  |     3
...  |     2

请注意，total包括聚合前未排除的所有行，即使按crosstab().

这是对@Vérace 在评论中的请求的答复，而不是对不清楚的问题的答复。

Answer 2

为了回答这个问题，我做了以下事情。（我发现以下三个线程很有帮助- 1、2和3。我还发现分别位于generate_series和CROSSTAB 这里和这里的文档很有用）。这应该适用于 9.1 - 未测试，但文档表明没有使用 9.1 后的东西。

创建了一个表：

ntest=# create table pv_tab(created_at timestamp, status varchar(10));

并填充它。

INSERT INTO pv_tab VALUES('2016-04-05 01:27:15', 'info');
INSERT INTO pv_tab VALUES('2016-04-05 03:27:15', 'info');
INSERT INTO pv_tab VALUES('2016-04-05 05:27:15', 'warn');
INSERT INTO pv_tab VALUES('2016-04-05 10:27:15', 'info');
INSERT INTO pv_tab VALUES('2016-04-05 11:27:15', 'warn');
INSERT INTO pv_tab VALUES('2016-04-05 00:27:15', 'info');
INSERT INTO pv_tab VALUES('2016-04-05 00:24:15', 'info');
INSERT INTO pv_tab VALUES('2016-04-05 00:24:13', 'warn');
INSERT INTO pv_tab VALUES('2016-04-05 00:24:13', 'warn');
INSERT INTO pv_tab VALUES('2016-04-05 01:24:13', 'warn');
INSERT INTO pv_tab VALUES('2016-04-05 01:24:13', 'info');
INSERT INTO pv_tab VALUES('2016-04-05 01:12:13', 'info');
INSERT INTO pv_tab VALUES('2016-04-05 01:12:22', 'info');
INSERT INTO pv_tab VALUES('2016-04-05 02:05:45', 'info');
INSERT INTO pv_tab VALUES('2016-04-05 02:34:45', 'warn');
INSERT INTO pv_tab VALUES('2016-04-05 10:34:45', 'warn');
INSERT INTO pv_tab VALUES('2016-04-05 10:35:45', 'warn');
INSERT INTO pv_tab VALUES('2016-04-05 10:36:45', 'warn');
INSERT INTO pv_tab VALUES('2016-04-05 10:36:45', 'info');
INSERT INTO pv_tab VALUES('2016-04-05 10:36:34', 'info');
(20 rows)

我的查询的（正确）结果是：

stat        slot1   slot2   slot3   slot4  slot5  slot6   Total
----        -----   -----   -----   -----  -----  -----   -----
info            6       2       0       3      0      0      11
warn            3       1       1       4      0      0       9

工作查询是：

SELECT * FROM CROSSTAB
(
'
WITH time_slots AS
(
  SELECT
    status,
    CASE

      -- Here I put the "created_at" values into "buckets" - it would
      -- not be desirable to have too many of these buckets - certainly
      -- any more than 12 would make the `SQL` and result unwieldy!

      -- I recommend that you create 2hr slots - 00:00 - 02:00, &c.
      -- This `CTE` splits the times into the various slots sample slots
      -- 1-4 - you can, of course, have more but it makes the `SQL` and 
      -- the answer more messy. Here, I;ve deliberately only used 4
      -- out of 6 in order to illustrate dealing with sparse data in 
      -- the result. (I used the OP;s initial slots - easy to change).

      WHEN created_at <  ''2016-04-05 02:00'' THEN 1
      WHEN created_at >= ''2016-04-05 02:00'' AND created_at < ''2016-04-05 04:00'' THEN 2
      WHEN created_at >= ''2016-04-05 04:00'' AND created_at < ''2016-04-05 08:00'' THEN 3
      WHEN created_at >= ''2016-04-05 08:00'' AND created_at < ''2016-04-05 12:00'' THEN 4
    END AS time_slot,
    COUNT(status) AS stat_count
  FROM pv_tab
  GROUP BY status, time_slot
  ORDER BY status, time_slot
),
statuae AS
  -- Get all statuses. Hardly necessary when there are only two, but
  -- could be an issue later if more values are required ("unknown"..  &c.).
(
  SELECT DISTINCT(status) AS stati
  FROM pv_tab
),
all_slots (slots) AS
  -- This `CTE` is necessary to perform a cross-join between statuses 
  -- and slots. This is because the `CROSSTAB` table function doesn;t 
  -- appear to play well with `NULL`s - see my question to Erwin
  -- Brandstetter in comments.
(
  SELECT generate_series(1, 6) -- six (should be) 2 hour slots. In any case, it is arbitrary!
),
stat_slots AS
  -- Here the statuses slots are cross-joined - i.e. all slots with all statuses.
(
  SELECT statuae.stati, all_slots.slots
  FROM statuae, all_slots
),
individual_stati AS
  -- `Left-join` the complete status/slot table with the actual slots in
  -- the sample table. NULL counts are `COALESCE`ed into 0 - necessary, otherwise
  -- `NULL`s "back up" the result and leaves blanks in the right-most 
  -- columns - and the totals appear in what should be slots. 
(
  SELECT ss.stati AS status, ss.slots AS time_slot, COALESCE(ts.stat_count, 0) AS counts
  FROM stat_slots ss
  LEFT JOIN time_slots ts
  ON ss.stati = ts.status AND ss.slots = ts.time_slot
  ORDER BY 1, 2
),
total_stati AS
  -- This is just pure showing off :-). I;m using this `CTE` to add
  -- a totals field to the query. Not asked for by the OP - can be
  -- ripped out! I got the idea for this from the 3rd link (top of post).
(
  SELECT status, 7 AS time_slot, count(status) AS counts -- 7 - an exta slot for totals
  FROM pv_tab
  GROUP BY status
)
-- Final query bringing it all together - nice, simple and elegant. :-)
SELECT status, time_slot, counts FROM individual_stati
UNION
SELECT status, time_slot, counts FROM total_stati
ORDER BY 1, 2
'
) AS My_Tab("stat" varchar(10), "slot1" bigint, "slot2" bigint, "slot3" bigint, "slot4" bigint, "slot5" bigint, "slot6" bigint, "Total" bigint);