sql - 如何通过两列聚合多个点并从中创建 LineString / MultiLineString

Question

我有一个名为位置的表，其中包含以下行：

id uuid NOT NULL,
"deviceId" text COLLATE pg_catalog."default",
"userId" uuid,
"userName" text COLLATE pg_catalog."default",
"creationDateTime" timestamp with time zone,
shape geometry,
CONSTRAINT id PRIMARY KEY (id)

想象一下，我的用户每小时将点数注册到该表的shape列中。当注册该点的时间注册到表中时，creationDateTime像这样保存到列2018-08-22 00:03:41.649+04:30中。

我如何提取此信息：

each User ---- each day ---- list of geometry(shape column) 例如：

第一天的 User1 有几何点列表。第二天的用户1有几何点列表等等......

我通过 mongo 对同一个项目进行了此查询：

{$project: {
    _id: 0,
    uId : "$UserId",
    dId : "$DeviceId",
    ts :"$CreationDateTime",
    point : "$Point"
    }
 }, 
{$group: {
    _id :{
        did: "$dId",
        day: { $dayOfMonth: "$ts" }
    },
    docs: { $push: "$$ROOT" }
     }
 },

 {
    $sort:{"_id.day": -1}
 }

但是我怎么能用 postgresql 做到这一点？postgre 上不存在这种聚合，我是 postgresql 的新手。这是我的查询：

(Select test1."deviceId",test1."shape", test1."creationDateTime" From   
    (Select * from locations) as test1 Group By test1."deviceId",test1."shape",test1."creationDateTime"
ORDER BY  test1."creationDateTime")

这个查询没有合适的结果，我知道这个查询有问题。 用户的deviceId经常与其他行重复。我该如何处理？

最后我想创建多折线per user - per day - multi poly line

标签： sqlpostgresqlpostgis

Answer 1

可能有一百万种方法可以回答这个问题。这是其中之一：

考虑到您的表结构..

CREATE TEMPORARY TABLE locations
(id uuid,
deviceId text COLLATE pg_catalog."default",
userId uuid,
userName text COLLATE pg_catalog."default",
creationDateTime timestamp with time zone,
shape geometry);

..还有这些样本数据..

INSERT INTO locations (userId, creationDateTime, shape) 
VALUES ('d1166a84-ab66-11e8-98d0-529269fb1459',CURRENT_DATE,'POINT(-1.25 51.75)'),
       ('d1166a84-ab66-11e8-98d0-529269fb1459',CURRENT_DATE,'POINT(-1.15 52.96)'),
       ('d1166a84-ab66-11e8-98d0-529269fb1459',CURRENT_DATE,'POINT(-0.13 50.82)'),
       ('d1166a84-ab66-11e8-98d0-529269fb1459',CURRENT_DATE-1,'POINT(-2.22 53.48)'),
       ('d1166a84-ab66-11e8-98d0-529269fb1459',CURRENT_DATE-1,'POINT(-0.11 51.51)');

..您可以汇总每个用户+日期的积分并LINESTRING使用以下方法创建ST_MakeLine一个GROUP BY：

SELECT userId, creationDateTime, ST_AsText(ST_MakeLine(shape))
FROM locations
GROUP BY userId, creationDateTime
ORDER BY  creationDateTime;

                userid                |    creationdatetime    |                    st_astext                    
--------------------------------------+------------------------+-------------------------------------------------
 d1166a84-ab66-11e8-98d0-529269fb1459 | 2018-08-28 00:00:00+02 | LINESTRING(-2.22 53.48,-0.11 51.51)
 d1166a84-ab66-11e8-98d0-529269fb1459 | 2018-08-29 00:00:00+02 | LINESTRING(-1.25 51.75,-1.15 52.96,-0.13 50.82)
(2 Zeilen)

d1166a84-ab66-11e8-98d0-529269fb1459用户的图形描述2018-08-28 00:00:00+02

以同样的方式，您可以创建MULTIPOINTusing ST_Collect：

SELECT userId, creationDateTime, ST_AsText(ST_Collect(shape))
FROM locations
GROUP BY userId, creationDateTime
ORDER BY  creationDateTime;

                userid                |    creationdatetime    |                    st_astext                    
--------------------------------------+------------------------+-------------------------------------------------
 d1166a84-ab66-11e8-98d0-529269fb1459 | 2018-08-28 00:00:00+02 | MULTIPOINT(-2.22 53.48,-0.11 51.51)
 d1166a84-ab66-11e8-98d0-529269fb1459 | 2018-08-29 00:00:00+02 | MULTIPOINT(-1.25 51.75,-1.15 52.96,-0.13 50.82)
(2 Zeilen)

编辑-使用（又名 WITH 子句）为每个用户（）创建一组LINESTRINGS每天：MULTILINESTRINGCTE

WITH j AS (
  SELECT userId, creationDateTime, ST_MakeLine(shape) AS shape
  FROM locations
  GROUP BY userId, creationDateTime)
SELECT userId, ST_AsText(ST_Collect(shape))
FROM j
GROUP BY userId

                userid                |                                    st_astext                                     
--------------------------------------+----------------------------------------------------------------------------------
 d1166a84-ab66-11e8-98d0-529269fb1459 | MULTILINESTRING((-2.22 53.48,-0.11 51.51),(-1.25 51.75,-1.15 52.96,-0.13 50.82))
(1 

基本上你要做的就是对你需要的记录进行分组（在这种情况下是用户和日期）并使用你选择的聚合函数，例如ST_MergeLine, ST_Collect, ST_Union,ST_Multi等。