postgres-为T2中的每一行选择时间戳之前的T1中的最新一行

Question

我研究了多个问题，但没有找到答案。因此，如果能提供类似解决方案的答案/链接，我将不胜感激。我有两张桌子：车辆和活动。每辆车都有多个GPS位置——比如玩具示例中有4辆车有11排GPS ping。但说玩具测试只有两个项目。

例如车辆(每行有唯一的gid，每辆车有vehicle_id，还有时间戳date_time和位置geom)：

SELECT gid, vehicle_id, date_time, geom FROM test_v vehicles ORDER BY 2,3,1;

gid   | vehicle_id |      date_time      |                        geom                        
---------+------------+---------------------+----------------------------------------------------
1257854 |        742 | 2009-06-06 19:02:01 | 01010000208A7D0000667CC002A7262741A444A87995324041
1258089 |        742 | 2009-06-06 19:03:29 | 01010000208A7D00007ABCA45AD22227415A5ADEB5662E4041
1258842 |        742 | 2009-06-06 19:07:23 | 01010000208A7D000083D5151C7414274196D0D21AE7284041
1259213 |        742 | 2009-06-06 19:09:11 | 01010000208A7D000054160E1FE4052741D6E2E74CD2284041
1257939 |       3055 | 2009-06-06 19:02:31 | 01010000208A7D0000327D6277FAF426418EA975ECB6504041
1258898 |       3055 | 2009-06-06 19:07:41 | 01010000208A7D000003519FB5D5F6264166B1B943C3514041
1259313 |       3425 | 2009-06-06 19:09:35 | 01010000208A7D0000418D23C296442741FCFCEF0069394041
1258008 |       3426 | 2009-06-06 19:02:56 | 01010000208A7D000011958634824027414D5D3823A43C4041
1258307 |       3426 | 2009-06-06 19:04:37 | 01010000208A7D0000C8017753AB38274175E727CBA43A4041
1258744 |       3426 | 2009-06-06 19:06:56 | 01010000208A7D00005F9811D7903A274145135F0589384041
1259018 |       3426 | 2009-06-06 19:08:13 | 01010000208A7D0000546B478B323A2741F4CB8DB6AD374041
(11 rows)

事件(唯一gid、时间戳t_date和位置geom):

SELECT gid, t_date, geom FROM test_c events;
gid  |       t_date        |                        geom                        
-------+---------------------+----------------------------------------------------
13009 | 2009-06-06 19:04:31 | 01010000208A7D000034F06A423F18274123DF6756B8194041
13021 | 2009-06-06 19:08:53 | 01010000208A7D0000D093E2A470FE26413E935C3A5E304041

我们的想法是了解每次活动时的情况，哪些车辆更接近，然后优化活动车辆的分配

因此，我想加入这些表格，以便我为每个事件获得最多4行(作为车辆数量)，每个车辆在事件发生前120分钟和事件发生前的时间段内的最后已知位置，按两者之间的距离排序。我想把它们放在同一张表中，这样我就可以按时间、地点、事件类型等进行比较。

现在我陷入了困境。我知道如何找到特定时间段(比如直到19:10)内车辆的最新记录。但这不是我所需要的，因为它在19:10时切断了车辆表。但对于gid= 13009时间为19:04:31的事件，可能会包含一个晚于该事件的位置。我想要的是到活动时间为止的最新位置，显然每个活动都不同。

我试过这个：

SELECT DISTINCT ON (1)
v.vehicle_id
, row_number() OVER() as gid
, st_distance(v.geom, c.geom)::float4 as distance_m
, c.gid as c_gid 
, v.gid as v_gid
, c.t_date as dt_c 
, v.date_time as dt_v 
, (c.t_date - v.date_time) as d_t
FROM (SELECT * FROM test_c ) as c
,  (SELECT * FROM test_v ORDER BY date_time DESC )  v
WHERE date_time >= (c.t_date -  '120 minute' :: INTERVAL) AND date_time <= (c.t_date + '0 minute' :: INTERVAL)
ORDER BY 1, 7  DESC , 5, 6 ,  c.geom <-> v.geom ASC ;

这就是我得到的：

vehicle_id | gid | distance_m | c_gid |  v_gid  |        dt_c         |        dt_v         |   d_t    
------------+-----+------------+-------+---------+---------------------+---------------------+----------
742 |   3 |     4748.6 | 13021 | 1258842 | 2009-06-06 19:08:53 | 2009-06-06 19:07:23 | 00:01:30
3055 |   2 |    17125.8 | 13021 | 1258898 | 2009-06-06 19:08:53 | 2009-06-06 19:07:41 | 00:01:12
3426 |   1 |    8515.65 | 13021 | 1259018 | 2009-06-06 19:08:53 | 2009-06-06 19:08:13 | 00:00:40
(3 rows)

我得到了不正确的日期，每辆车只在那里一次(我想是因为DISTINCT)，但我需要在最后一个合适的地点为每次活动提供一次(因此在本例中，2次活动需要2次)。

理想情况下，我还希望下一步为每辆车提供3个最新职位，但也许这是以后的事。

Answer 1

这就是我所理解的：

select *
from (
select distinct on (c.gid, v.vehicle_id)
v.vehicle_id
, row_number() over(order by v.gid) as gid
, st_distance(v.geom, c.geom)::float4 as distance_m
, c.gid as c_gid 
, v.gid as v_gid
, c.t_date as dt_c 
, v.date_time as dt_v 
, (c.t_date - v.date_time) as d_t
from
test_c c
inner join 
test_v v on
date_time between c.t_date - '120 minute' :: interval and c.t_date
order by c.gid, v.vehicle_id, v.date_time desc
) s
order by c_gid, distance_m
;
vehicle_id | gid | distance_m | c_gid |  v_gid  |        dt_c         |        dt_v         |   d_t    
------------+-----+------------+-------+---------+---------------------+---------------------+----------
742 |   8 |    10674.9 | 13009 | 1258089 | 2009-06-06 19:04:31 | 2009-06-06 19:03:29 | 00:01:02
3426 |   5 |    18607.5 | 13009 | 1258008 | 2009-06-06 19:04:31 | 2009-06-06 19:02:56 | 00:01:35
3055 |   4 |    28516.8 | 13009 | 1257939 | 2009-06-06 19:04:31 | 2009-06-06 19:02:31 | 00:02:00
742 |  11 |     4748.6 | 13021 | 1258842 | 2009-06-06 19:08:53 | 2009-06-06 19:07:23 | 00:01:30
3426 |  13 |    8515.65 | 13021 | 1259018 | 2009-06-06 19:08:53 | 2009-06-06 19:08:13 | 00:00:40
3055 |  12 |    17125.8 | 13021 | 1258898 | 2009-06-06 19:08:53 | 2009-06-06 19:07:41 | 00:01:12