Logo Questions Linux Laravel Mysql Ubuntu Git Menu
 

Howto select (almost) unique values in a specific order

Tags:

sql

distinct

In a trip there are several stops, (a stop = an adress whereone or multiple orders are loaded, or delivered), in a specific order. For example:

Trip A
Trip_order   Action   Place          Ordernumber
10           Load     Paris          394798
20           Load     Milan          657748
30           UnLoad   Athens         657748
40           Unload   Thessaloniki   394798
50           Load     Thessaloniki   10142
60           Load     Thessaloniki   6577
70           Unload   Athens         6577
80           Unload   Athens         10412
90           Load     Thessaloniki   975147
100          Unload   Paris          975147

I want to see the specific stops, in order of the trip:

Load Paris
Load Milan
Unload Athens
Unload Thessaloniki
Load Thessaloniki
Unload Athens
Load Thessaloniki
Unload Paris

I did look at This, but if I do that, I only get the unload Athens, unload Thessaloniki and Load Thessaloniki once.

How do I solve this?

EDIT: 11:11 (UTC +01:00) To be more specific: these are the tables which present this information:

Trips
Trip_ID
100001  
100002
100003
....

Actions
Trip_ID  Action MatNr RoOr RoVlg    OrderID
100001   1      10      10     1     394798
100001   1      10      20     1     657748
100001   1      10      30     1     657748
100001   1      10      40     1     394798
100001   1      10      50     1      10142
100001   1      10      60     1       6577
100001   1      10      70     1       6577
100001   1      10      80     1      10412
100001   1      10      90     1     975147
100001   1      10     100     1     975147

(Action: 1=load, 4=unload) The combination of MatNr, RoOr and RoVlg is the order of the Trip.

Orders
OrderID LoadingPlace UnloadingPlace
6577    Thessaloniki Athens
10142   Thessaloniki Athens
394798  Paris        Thessaloniki
657748  Milan        Athens
975147  Thessaloniki Paris
like image 222
PSVSupporter Avatar asked May 01 '15 07:05

PSVSupporter


People also ask

How do you select unique records from the particular column?

The SELECT DISTINCT statement is used to return only distinct (different) values. Inside a table, a column often contains many duplicate values; and sometimes you only want to list the different (distinct) values.

Can you use select and select distinct in SQL?

The SQL SELECT DISTINCT Statement The DISTINCT clause is used in a SELECT statement to remove duplicates from the result set of the query.

How can I get distinct values for a particular column in SQL?

To get unique or distinct values of a column in MySQL Table, use the following SQL Query. SELECT DISTINCT(column_name) FROM your_table_name; You can select distinct values for one or more columns. The column names has to be separated with comma.


5 Answers

Try this one. No variables, nothing especially fancy:

select a1.action, a1.place
  from trip_a a1
    left join trip_a a2
      on a2.trip_order = 
        (select min(trip_order) 
          from trip_a a3 
          where trip_order > a1.trip_order)
  where a1.action != a2.action or a1.place != a2.place or a2.place is null

Demo here: http://sqlfiddle.com/#!9/4b6dc/13

Hopefully it works on whatever sql you're using, it should, so long as subqueries are supported.

Tt simply finds the next highest trip_id, and joins to it, or joins to null if there is no higher trip_order. It then selects only the rows where either the place, the action, or both are different, or if there is no place in the joined table (a2.place is null).

edited after criteria changed completely

If you want to get the same results, built entirely from your base tables, you can do this:

  select 
         case when a.action = 1 then 'load' when a.action = 0 then 'unload' end as action,
         case when a.action = 1 then o.loadingplace when a.action = 0 then o.unloadingplace end as place
    from trips t
      inner join actions a
        on t.trip_id = a.trip_id
      inner join orders o
        on a.orderid = o.orderid
      left join actions a2
        on a2.roor = 
          (select min(roor) 
             from actions a3
             where a3.roor > a.roor)
      left join orders o2
        on a2.orderid = o2.orderid
      where a.action != a2.action
        or a2.action is null
        or
          case when a.action = 1 then o.loadingplace != o2.loadingplace
               when a.action = 0 then o.unloadingplace != o2.unloadingplace
          end
    order by a.roor asc

And here's an updated fiddle: http://sqlfiddle.com/#!9/fdf9c/14

like image 163
pala_ Avatar answered Oct 04 '22 20:10

pala_


You don't need and you don't want to use distinct for that because we can see in your example that several destinations occur multiple times. What you want: filter out records that match the preceding record in terms of action and place.

This could look something like this:

SELECT *
FROM Trips t1 LEFT JOIN Trips t2 ON t1.Trip_Order = t2.Trip_Order - 10
WHERE t1.Action <> t2.Action OR t1.Place <> t2.Place)
like image 29
lex82 Avatar answered Oct 04 '22 20:10

lex82


In SQL server, you can get difference of ROW_NUMBER() based trip_order and action,place on and try something like this.

You can use it as a reference to create a similar query in USQL.

Sample Data

DECLARE @Trip TABLE (Trip_order INT,   Action VARCHAR(10),  Place  VARCHAR(50),Ordernumber INT)

INSERT INTO @Trip VALUES
(10           ,'Load',     'Paris',          394798),
(20           ,'Load',     'Milan',          657748),
(30           ,'UnLoad',   'Athens',         657748),
(40           ,'UnLoad',   'Thessaloniki',   394798),
(50           ,'Load',     'Thessaloniki',   10142),
(60           ,'Load',     'Thessaloniki',   6577),
(70           ,'UnLoad',   'Athens',         6577),
(80           ,'UnLoad',   'Athens',         10412),
(90           ,'Load',     'Thessaloniki',   975147),
(100          ,'UnLoad',   'Paris',          975147);

Query

SELECT action,place FROM 
(
    SELECT *,ROW_NUMBER()OVER(ORDER BY trip_order) - ROW_NUMBER()OVER(ORDER BY action,place) n
    FROM @trip
)t
GROUP BY n,action,place
ORDER BY MIN(trip_order)
like image 40
ughai Avatar answered Oct 04 '22 21:10

ughai


Try this:

Will work in MySQL:::

SELECT IF(@temp=@temp:=A.TripName, @rank, @rank:=@rank+1) AS rank, A.TripName
FROM (SELECT CONCAT(A.Action, A.Place) AS TripName
      FROM TripA A
     ) A, (SELECT @temp:=0, @rank:=0) AS B
GROUP BY rank
like image 37
Saharsh Shah Avatar answered Oct 04 '22 21:10

Saharsh Shah


SELECT s.*
FROM stops s LEFT JOIN stops prev ON 
     ( prev.Trip_order < s.Trip_order
       AND NOT EXISTS ( SELECT 'a'
                        FROM stops prev2
                        WHERE prev2.Trip_order < s.Trip_order
                        AND prev2.Trip_order > prev.Trip_order
                       )
      )
WHERE s.Action <> COALESCE(prev.Action, '')
OR s.Place <> COALESCE(prev.Place, '')
ORDER BY s.Trip_order
like image 40
Javaluca Avatar answered Oct 04 '22 20:10

Javaluca