How do I get a column with consecutive, increasing numbers, without having any numbers missing?
Asked Answered
R

1

8

Possible Duplicates:
How to show row numbers in PostgreSQL query?
Resequencing a column with identifier in Postgresql
PostgreSQL Record Reordering using Update with a Sub-Select

I am just asking if in PostgreSQL exist such possibility that if I have 5 rows and in one column there are numbers 1, 2, 3, 4, 5 and in those column is not primary key if I delete say tird row that postgreSQL re enumerate this column so I can have 1, 2, 3, 4 instead of 1, 2, 4, 5?

Ramonramona answered 29/12, 2012 at 19:40 Comment(0)
T
11

Adapted this from a previous answer. This kind of stuff happens often when applications want a tabbing-order for variables (read: records in an EAV model) , which could also be (part of) an alternate key.

  • the priority field needs to be kept consecutive. [this is the tabbing-order]
  • on INSERT: all records with priority >= the new record should have their priorities incremented
  • similarly: on DELETE -> decremented
  • if a record's priority is altered by an UPDATE, the records between the old and the new priority value should have their priorities shifted up or down.
  • to avoid recursive trigger invocation:
    • the trigger-based updates flip the flipflag of any record they touch.
    • And they test for old.flipflag=new.flipflag to detect real updates. (those not caused by a trigger)

        -- Make some data
DROP SCHEMA tmp CASCADE;
CREATE SCHEMA tmp ;
SET search_path=tmp;

CREATE TABLE fruits
        ( id INTEGER NOT NULL PRIMARY KEY
        , priority INTEGER NOT NULL
        , flipflag boolean NOT NULL default false
        , zname varchar NOT NULL
        , CONSTRAINT unique_priority UNIQUE (priority) DEFERRABLE INITIALLY DEFERRED
        );
INSERT INTO fruits(id,zname,priority) VALUES
 (1  , 'Pear' ,4)
,(2  , 'Apple' ,2)
,(3  , 'Orange' ,1)
,(4  , 'Banana' ,3)
,(5  , 'Peach' ,5)
        ;

        -- Trigger functions for Insert/update/delete
CREATE function shift_priority()
RETURNS TRIGGER AS $body$

BEGIN
        UPDATE fruits fr
        SET priority = priority +1
        , flipflag = NOT flipflag       -- alternating bit protocol ;-)
        WHERE NEW.priority < OLD.priority
        AND OLD.flipflag = NEW.flipflag -- redundant condition
        AND fr.priority >= NEW.priority
        AND fr.priority < OLD.priority
        AND fr.id <> NEW.id             -- exlude the initiating row
                ;
        UPDATE fruits fr
        SET priority = priority -1
        , flipflag = NOT flipflag
        WHERE NEW.priority > OLD.priority
        AND OLD.flipflag = NEW.flipflag
        AND fr.priority <= NEW.priority
        AND fr.priority > OLD.priority
        AND fr.id <> NEW.id
        ;
        RETURN NEW;
END;

$body$
language plpgsql;

CREATE function shift_down_priority()
RETURNS TRIGGER AS $body$

BEGIN

        UPDATE fruits fr
        SET priority = priority -1
        , flipflag = NOT flipflag       -- alternating bit protocol ;-)
        WHERE fr.priority > OLD.priority
                ;
        RETURN NEW;
END;

$body$
language plpgsql;

CREATE function shift_up_priority()
RETURNS TRIGGER AS $body$

BEGIN
        UPDATE fruits fr
        SET priority = priority +1
        , flipflag = NOT flipflag       -- alternating bit protocol ;-)
        WHERE fr.priority >= NEW.priority
                ;
        RETURN NEW;
END;

$body$
language plpgsql;

        -- Triggers for Insert/Update/Delete
CREATE TRIGGER shift_priority_u
        AFTER UPDATE OF priority ON fruits
        FOR EACH ROW
        WHEN (OLD.flipflag = NEW.flipflag AND OLD.priority <> NEW.priority)
        EXECUTE PROCEDURE shift_priority()
        ;
CREATE TRIGGER shift_priority_d
        AFTER DELETE ON fruits
        FOR EACH ROW
        EXECUTE PROCEDURE shift_down_priority()
        ;
CREATE TRIGGER shift_priority_i
        BEFORE INSERT ON fruits
        FOR EACH ROW
        EXECUTE PROCEDURE shift_up_priority()
        ;

        -- Do some I/U/D operations
\echo Pears are Okay
UPDATE fruits
SET priority = 1
WHERE id=1; -- 1,4

SELECT * FROM fruits ORDER BY priority;

\echo dont want bananas
DELETE FROM fruits WHERE id = 4;
SELECT * FROM fruits ORDER BY priority;

\echo  We want Kiwis
INSERT INTO fruits(id,zname,priority) VALUES (4  , 'Kiwi' ,3) ;
SELECT * FROM fruits ORDER BY priority;

Result:

Pears are Okay
UPDATE 1
 id | priority | flipflag | zname  
----+----------+----------+--------
  1 |        1 | f        | Pear
  3 |        2 | t        | Orange
  2 |        3 | t        | Apple
  4 |        4 | t        | Banana
  5 |        5 | f        | Peach
(5 rows)

dont want bananas
DELETE 1
 id | priority | flipflag | zname  
----+----------+----------+--------
  1 |        1 | f        | Pear
  3 |        2 | t        | Orange
  2 |        3 | t        | Apple
  5 |        4 | t        | Peach
(4 rows)

We want Kiwis
INSERT 0 1
 id | priority | flipflag | zname  
----+----------+----------+--------
  1 |        1 | f        | Pear
  3 |        2 | t        | Orange
  4 |        3 | f        | Kiwi
  2 |        4 | f        | Apple
  5 |        5 | f        | Peach
(5 rows)
Thuthucydides answered 30/12, 2012 at 17:17 Comment(5)
Adapted from a previous answer... So this is a duplicate too?Camey
No, I added code for delete and insert. The older version only had updates -->> bubbling up/down.Thuthucydides
You endless-loop-prevention by alternating-bit reminds me of this related answer, where I used deferred triggers. But I don't have the time to dive in right now.Danseuse
This could probably be a really great answer if it had some kind of explanation beyond "I adapted this from an unnamed source". It's all very well putting a winking smiley next to the flipflag column, but I have absolutely no idea why it's there.Territerrible
It is there to distinguish real updates from updates caused by a trigger, thus avoiding infinite recursion (other mechanisms to avoid infinite recursion have arisen since) And the other answer was my own, IIRC.Thuthucydides

© 2022 - 2024 — McMap. All rights reserved.