PostgreSQL checking a previous record's element - postgresql

I need to check the previous record's element to make sure the date I query doesn't fall within a specific range between ending date and 7 days before starting date. I have the following code:
create or replace function eight (date) returns text as $$
declare
r record;
checkDate alias for $1;
begin
for r in
select * from periods
order by startDate
loop
if (checkDate between r.startDate and r.endDate) then
return q3(r.id);
elsif (checkDate between (r.startDate - interval '7 days') and r.startDate) then
return q3(r.id);
elsif (checkDate between (lag(r.endDate) over (order by r.startDate)) and (r.startDate - interval '8 days')) then
return q3(r.id);
end if;
end loop;
return null;
end;
$$ language plpgsql;
So basically, I need to check for the following:
If the query date is between the starting and ending dates
If the query date is 7 days before the start of the starting date
If the query date is between ending date and the starting date
and return the id that is associated with that date.
My function seems to work fine in most cases, but there are cases that seem to give me 0 results (when there should always be 1 result) is there something missing in my function? I'm iffy about the last if statement. That is, trying to check from previous records ending date to current records starting date (with the 7 day gap)
EDIT: no dates overlap.

Edit: Removed the part about RETURN NEXT - I had misread the question there.
Doesn't work the way you have it. A window function cannot be called like that. Your record variable r is like a built-in cursor in a FOR loop. Only the current row of the result is visible inside the loop. You would have to integrate the window function lag() it into the initial SELECT.
But since you are looping through the rows in a matching order anyway, you can do it another way.
Consider this largely rewritten example. Returns at the first violating row:
CREATE OR REPLACE FUNCTION q8(_day date)
RETURNS text AS
$BODY$
DECLARE
r record;
last_enddate date;
BEGIN
FOR r IN
SELECT *
-- ,lag(r.endDate) OVER (ORDER BY startDate) AS last_enddate
-- commented, because I supply an alternative solution
FROM periods
ORDER BY startDate
LOOP
IF _day BETWEEN r.startDate AND r.endDate THEN
RETURN 'Violates condition 1'; -- I return differing results
ELSIF _day BETWEEN (r.startDate - 7) AND r.startDate THEN
RETURN 'Violates condition 2';
ELSIF _day BETWEEN last_enddate AND (r.startDate) THEN
-- removed "- 7 ", that is covered above
RETURN 'Violates condition 3';
END IF;
last_enddate := r.enddate; -- remember for next iteration
END LOOP;
RETURN NULL;
END;
$BODY$ LANGUAGE plpgsql;
More hints
Why the alias for $1? You named it _day in the declaration already. Stick to it.
Be sure to know how PostgreSQL handles case in identifiers. ( I only use lower case.)
You can just add / subtract integers (for days) from a date.

Are you sure that lag() will return you something? I'm pretty sure that this is out of context here. Given that rows from periods are selected in order, you can store the current startDate in a variable, and use it in the if statement of the next cycle.

SET search_path='tmp';
DROP table period;
CREATE table period
( start_date DATE NOT NULL
, end_date DATE
);
INSERT INTO period(start_date ,end_date) VALUES
( '2012-01-01' , '2012-02-01' )
, ( '2012-02-01' , '2012-02-07' )
, ( '2012-03-01' , '2012-03-15' )
, ( '2012-04-01' , NULL )
, ( '2012-04-17' , '2012-04-21' )
;
DROP FUNCTION valid_date(DATE) ;
CREATE FUNCTION valid_date(DATE) RETURNS boolean
AS $body$
declare
found boolean ;
zdate ALIAS FOR $1;
begin
found = false;
SELECT true INTO found
WHERE EXISTS (
SELECT * FROM period p
WHERE (p.start_date > zdate
AND p.start_date < zdate + interval '7 day' )
OR ( p.start_date < zdate AND p.end_date > zdate )
OR ( p.start_date < zdate AND p.end_date IS NULL
AND p.start_date >= zdate - interval '7 day' )
)
;
if (found = true) then
return false;
else
return true;
end if;
end;
$body$ LANGUAGE plpgsql;
\echo 2011-01-01:true
SELECT valid_date('2011-01-01' );
\echo 2012-04-08:false
SELECT valid_date('2012-04-08' );
\echo 2012-04-30:true
SELECT valid_date('2012-04-30' );
BTW: I really think that the required functionality should be implemented as a table constraint, imposed by a trigger function (that might be based on the above function).

Related

Incorporating a LOOP into a SQL

I am fairly new at SQL and have not incorporated a Loop into a SQL statement previously. This SQL query from elan.elig returns data as shown in the grid below.
select (extract(year from age(case when terminationdate is null then
CURRENT_DATE else terminationdate END ,effectivedate ))) *12 +
(extract(month from age(case when terminationdate is null then
CURRENT_DATE else terminationdate END ,effectivedate )) +1)
as "mbrmonths" ,effectivedate
from elan.elig
Mbr Months Effective Date
1 10/1/2018
10 11/1/2018
2 11/1/2018
8 11/1/2018
8 11/1/2018
8 11/1/2018
2 11/1/2018
2 11/1/2018
7 11/1/2018
For each row from the query I need to execute the subsequent LOOP that spreads the memberMonth counts into the Year/Month buckets. The following Do LOOP does exactly this. I have been trying for some time now to determine how to incorporate the Loop into the SQL statement so that for each row read, it will pass the two variables and execute the Loop and then read the next row and continue on..
DO $$
declare
nbr_mem_months integer=5;
effectivedate date ='20190401';
ym char(6) =to_char(effectivedate,'YYYYMM');
begin
for r in 1..nbr_mem_months loop
update elan.pmpm set mbrmonths=mbrmonths+1 where yyyyymm=ym;
effectivedate=effectivedate + interval '1 month';
ym=to_char(effectivedate,'YYYYMM');
end loop;
end;
$$;
PMPM Buckets
yyyymm mbrmonths
201901 0
201902 0
201903 0
201904 1
201905 1
201906 1
201907 1
201908 1
201909 0
201910 0
201911 0
CREATE FUNCTION "UpdatePMPM"() RETURNS boolean
LANGUAGE plpgsql
AS
$$
DECLARE
nbr_mem_months NUMERIC;
effectivedate date;
ym char(6);
BEGIN
LOOP
ym=to_char(effectivedate,'YYYYMM');
nbr_mem_months=5;
UPDATE elan.pmpm set mbrmonths=mbrmonths+1 where yyyyymm=ym;
effectivedate=effectivedate + interval '1 month';
END LOOP;
RETURN TRUE;
END
$$;
*Response from the Select statement:
ERROR: function public.UpdatePMPM(integer, date, text) does not exist
Select public."UpdatePMPM"(5,cast('20190101' as date),cast('...
^
HINT: No function matches the given name and argument types.
The issue is calling the function with arguments but not specifying any when creating the function. So you need something like(not tested):
CREATE FUNCTION "UpdatePMPM"(nbr_mem_months integer, effectivdate date, some_arg varchar) RETURNS void
LANGUAGE plpgsql
AS
$$
DECLARE
ym varchar := to_char(effectivedate,'YYYYMM');
BEGIN
FOR r IN 1..nbr_mem_months LOOP
UPDATE elan.pmpm set mbrmonths = mbrmonths+1 where yyyyymm = ym;
effectivedate = effectivedate + interval '1 month';
ym=to_char(effectivedate,'YYYYMM');
END LOOP;
RETURN;
END
$$;
From the error it is not clear what the third argument is supposed to be, so that will clarification from you.

EXTRACT INTO with multiple rows (PostgreSQL)

this is my function:
CREATE OR REPLACE FUNCTION SANDBOX.DAILYVERIFY_DATE(TABLE_NAME regclass, DATE_DIFF INTEGER)
RETURNS void AS $$
DECLARE
RESULT BOOLEAN;
DATE DATE;
BEGIN
EXECUTE 'SELECT VORHANDENES_DATUM AS DATE, CASE WHEN DATUM IS NULL THEN FALSE ELSE TRUE END AS UPDATED FROM
(SELECT DISTINCT DATE VORHANDENES_DATUM FROM ' || TABLE_NAME ||
' WHERE DATE > CURRENT_DATE -14-'||DATE_DIFF|| '
) A
RIGHT JOIN
(
WITH compras AS (
SELECT ( NOW() + (s::TEXT || '' day'')::INTERVAL )::TIMESTAMP(0) AS DATUM
FROM generate_series(-14, -1, 1) AS s
)
SELECT DATUM::DATE
FROM compras)
B
ON DATUM = VORHANDENES_DATUM'
INTO date,result;
RAISE NOTICE '%', result;
INSERT INTO SANDBOX.UPDATED_TODAY VALUES (TABLE_NAME, DATE, RESULT);
END;
$$ LANGUAGE plpgsql;
It is supposed to upload rows into the table SANDBOX.UPDATED_TODAY, which contains table name, a date and a boolean.
The boolean shows, whether there was an entry for that date in the table. The whole part, which is inside of EXECUTE ... INTO works fine and gives me those days.
However, this code only inserts the first row of the query's result. What I want is that all 14 rows get inserted. Obviously, I need to change it into something like a loop or something completely different, but how exactly would that work?
Side note: I removed some unnecessary parts regarding those 2 parameters you can see. It does not have to do with that at all.
Put the INSERT statement inside the EXECUTE. You don't need the result of the SELECT for anything other than inserting it into that table, right? So just insert it directly as part of the same query:
CREATE OR REPLACE FUNCTION SANDBOX.DAILYVERIFY_DATE(TABLE_NAME regclass, DATE_DIFF INTEGER)
RETURNS void AS
$$
BEGIN
EXECUTE
'INSERT INTO SANDBOX.UPDATED_TODAY
SELECT ' || QUOTE_LITERAL(TABLE_NAME) || ', VORHANDENES_DATUM, CASE WHEN DATUM IS NULL THEN FALSE ELSE TRUE END
FROM (
SELECT DISTINCT DATE VORHANDENES_DATUM FROM ' || TABLE_NAME ||
' WHERE DATE > CURRENT_DATE -14-'||DATE_DIFF|| '
) A
RIGHT JOIN (
WITH compras AS (
SELECT ( NOW() + (s::TEXT || '' day'')::INTERVAL )::TIMESTAMP(0) AS DATUM
FROM generate_series(-14, -1, 1) AS s
)
SELECT DATUM::DATE
FROM compras
) B
ON DATUM = VORHANDENES_DATUM';
END;
$$ LANGUAGE plpgsql;
The idiomatic way to loop through dynamic query results would be
FOR date, result IN
EXECUTE 'SELECT ...'
LOOP
INSERT INTO ...
END LOOP;

how to get the next date for a day of month

thanks for reading, this is the situation
I have a current_date and a day of month, so i need to know what will be the next date for this day of month, having in mind that some month don't have 30 and 31.
Example:
current_date = '2018-09-24'
day_of_week = 31
Expected result: '2018-12-31'
Currently i have this:
create or replace function next_diff(vals int[], current_val int) returns int as
$$
declare v int;
declare o int := vals[1];
begin
foreach v in array vals loop
if current_val >= o and current_val < v then
return v - current_val;
end if;
o := v;
end loop;
return vals[1] - current_val;
end;
$$ language plpgsql;
and this:
create or replace function next_day_of_month(days_of_month int[], curr_date date) returns date as
$$
declare cur_dom int := extract(day from curr_date);
declare next_diff int := next_diff(days_of_month, cur_dom);
begin
if next_diff < 0 then
curr_date := curr_date + '1 months'::interval;
end if;
curr_date := curr_date + (next_diff || 'days')::interval;
return curr_date;
end;
$$ language plpgsql;
but for this calling:
select next_day_of_month(array[31], '2018-09-24');
i am getting:
"2018-10-01"
Extra example
If i have this value
current_date = '2018-02-01'
day_of_week = 31
i will need the next month with 31th but i can't get '2018-02-31' because February don't have 31th then i should get '2018-02-31' because March have 31th.
Conclusion
if the month don't have the specified day must ignore the month and jump to the next.
thanks for all
Final method
Using Carlos Gomez answer, i create this PostgreSQL function and work perfectly:
create or replace function next_day_date(curr_date date, day_of_month int) returns date as
$$
declare next_day date;
begin
SELECT next_day_date into next_day FROM (
SELECT make_date_nullable(EXTRACT(year from n.month)::int, EXTRACT(month from n.month)::int, day_of_month) AS next_day_date
FROM (
SELECT generate_series(curr_date, curr_date + '3 months'::interval, '1 month'::interval) as month
) n
) results
WHERE results.next_day_date IS NOT NULL and results.next_day_date > curr_date LIMIT 1;
return next_day;
end;
$$ language plpgsql;
just add other filter in where clause and results.next_day_date > curr_date to prevent get the same or previous values for specified date
Thanks everyone for helping
Thenks Carlos you are the best
Gracias carlos eres el mejor :)
Your examples don't really match up but I think I know what you are trying to solve for (your first example result should be '2018-10-31' since October has 31 days and your second example result should be '2018-03-31'). It seems that given a date and a day of month you want to find the next month that has that day of month. To do this, I would do the following:
This function just wraps make_date to let it return null since it throws an exception if a date given to it is out of bounds (like February 30).
CREATE OR REPLACE FUNCTION make_date_nullable(year int, month int, day int)
RETURNS date as $$
BEGIN
RETURN make_date(year, month, day);
EXCEPTION WHEN others THEN RETURN null;
END;
$$ language plpgsql;
This SELECT first generates the next three months starting with the current one, then makes date out of them with your provided day_of_month and finally gets the first one that isn't null (exists according to postgresql.
SELECT next_day_date FROM (
SELECT make_date_nullable(EXTRACT(year from n.month)::int, EXTRACT(month from n.month)::int, day_of_month) AS next_day_date
FROM (
SELECT generate_series(current_date, current_date + '3 months'::interval, '1 month'::interval) as month
) n
) results
WHERE results.next_day_date IS NOT NULL LIMIT 1;
Hope this helps!

Update the records if they exist

Periodically I need to check the records in a table and update them, in particular check the records with "payment_pending" status and update them to "payment_pending_expired" status. But I'm not sure how to do it properly:
CREATE OR REPLACE FUNCTION cancel_pending_orders(user_id integer, max_day_amount_allowed integer)
RETURNS SETOF void AS
$BODY$
BEGIN
-- if the records exist at all...
IF EXISTS (
SELECT ph.id FROM payment_history as ph
INNER JOIN payment AS p
ON p.id = ph.payment_id
WHERE p.user_id = $1
AND ph.status = 'payment_pending'
AND ph.added_at + max_day_amount_allowed <= now()
) THEN
-- make all of them status = 'payment_pending_expired'
RETURN;
END IF;
The questions:
1) How do I add max_day_amount_allowed to ph.added_at? If it were a literal I could do this by:
....
AND (ph.added_at + interval '30d') <= now()
but it is not a literal, it is a variable.
2) How do I refer to the found records (in case, the exist)
....
) THEN
-- make all of them ph.status = 'payment_pending_expired'
-- but how do I refer to them?
RETURN;
P.S. ph.status has a type of varchar and not integer only for simplicity.
1) You need to cast the day count to an interval:
AND (ph.added_at + ($2 || ' days')::interval) <= now()
2) You can use CURSORs to do something with each row in a result-set.
But in your case (if you want to only update them) just use a single UPDATE command:
UPDATE payment_history AS ph
SET ph.status = 'payment_pending_expired'
FROM payment AS p
WHERE p.id = ph.payment_id
AND p.user_id = $1
AND ph.status = 'payment_pending'
AND (ph.added_at + ($2 || ' days')::interval) <= now()
How do I add max_day_amount_allowed to ph.added_at?
Assuming the type timestamp for added_at.
Don't convert to text, concatenate and convert back. Just multiply an interval:
ph.added_at + interval '1d' * max_day_amount_allowed <= now()
Or, if added_at is a date, you can just add integer to a date. The date is then coerced to timestamp automatically (according to local time) for the comparison to now():
ph.added_at + max_day_amount_allowed <= now()

Function Getting the right week number of year

I want to create a function to get the right week number of year.
I already posted here to find a 'native' solution, but apparently there is not.
I tryed to create funcrtion based on this mysql example
Here is the code translated to postgresql:
CREATE OR REPLACE FUNCTION week_num_year(_date date)
RETURNS integer AS
$BODY$declare
_year integer;
begin
select date_part('year',_date) into _year;
return ceil((to_char(_date,'DDD')::integer+(to_char(('01-01-'||_year)::date,'D')::integer%7-7))/7);
end;$BODY$
LANGUAGE plpgsql VOLATILE
COST 100;
But it gives wrong result, can someone help me ?
My config: PostgreSQL 9.2
If you want proper week numbers use:
select extract(week from '2012-01-01'::date);
This will produce the result 52, which is correct if you look on a calendar.
Now, if you actually want to define week numbers as "Every 7 days starting with the first day of the year" that's fine, though it doesn't match the week numbers anyone else uses and has some odd quirks:
select floor((extract(doy from '2011-01-01'::date)-1)/7)+1;
By the way, parsing date strings and hacking them up with string functions is almost always a really bad idea.
create or replace function week_num_year(_date date)
returns integer as
$body$
declare
_year date;
_week_number integer;
begin
select date_trunc('year', _date)::date into _year
;
with first_friday as (
select extract(doy from a::date) ff
from generate_series(_year, _year + 6, '1 day') s(a)
where extract(dow from a) = 5
)
select floor(
(extract(doy from _date) - (select ff from first_friday) - 1) / 7
) + 2 into _week_number
;
return _week_number
;
end;
$body$
language plpgsql immutable
You can retrieve the day of the week and also the week of the year by running:
select id,extract(DOW from test_date),extract(week from test_date), testdate,name from yourtable
What about the inbuild extract function?
SELECT extract (week from current_timestamp) FROM A_TABLE_FROM_YOUR_DB;