PostgreSQL Query with dynamic columns and counts from join - postgresql

Question

I'm probably overthinking this, because I'm not really sure where to start... But here goes:
I have the following
Tables
students
assessments
students_assessments
Expected output
First query for "Attempts"
student_id | Assessment 1 | Assessment 2 | Assessment 3 | Assessment 4
1 3 1 2 0
2 1 0 0 0
3 2 1 1 0
4 5 3 3 0
5 1 5 0 0
6 2 1 2 0
Second query for "passed"
student_id | Assessment 1 | Assessment 2 | Assessment 3 | Assessment 4
1 t t f f
2 t t f f
3 t t f f
4 t t t f
5 t t f f
6 t t t f
The part that is tripping me up is not doing a join for EVERY assessment, and not even defining the columns manually but rather "generating" each column for assessments that exist.
I feel like it's simple, and I am just too overworked to figure it out right now :) thank you in advance for your help, and here's a SQL Fiddle of data as an example

Answer 1

Simple query for "Attempts",
select student_id,sum(case when assessment_id=1 then 1 else 0 end) as "Assessment 1",
sum(case when assessment_id=2 then 1 else 0 end) as "Assessment 2",
sum(case when assessment_id=3 then 1 else 0 end) as "Assessment 3",
sum(case when assessment_id=4 then 1 else 0 end) as "Assessment 4",
sum(case when assessment_id=5 then 1 else 0 end) as "Assessment 5",
sum(case when assessment_id=6 then 1 else 0 end) as "Assessment 6"
from assessments_students
group by student_id
order by student_id
In crosstab() function also, need to define columns name explicitly like "Assessment 1","Assessment 2" and so on.
Or write custom function for creating dynamic query, and execute using EXECUTE statement.
DROP FUNCTION get_Attempts() ;
CREATE OR REPLACE FUNCTION get_Attempts() RETURNS text AS
$BODY$
DECLARE
r1 record;
str_query text := '';
BEGIN
str_query :='select student_id,';
FOR r1 IN SELECT "_id" , "name" FROM Assessments
LOOP
str_query:= str_query ||
'sum(case when assessment_id=' || r1."_id" || ' then 1 else 0 end) as "' || r1.name ||'",' ;
END LOOP;
str_query:=trim( trailing ',' from str_query); -- remove last semicolon
str_query:= str_query || ' from assessments_students group by student_id order by student_id';
return str_query;
END
$BODY$
LANGUAGE 'plpgsql' ;
SELECT * FROM get_Attempts();
Second query for "passed"
select student_id,
max(case when assessment_id=1 and passed='t' then 't' else 'f' end) as "Assessment 1",
max(case when assessment_id=2 and passed='t' then 't' else 'f' end) as "Assessment 2",
max(case when assessment_id=3 and passed='t' then 't' else 'f' end) as "Assessment 3",
max(case when assessment_id=4 and passed='t' then 't' else 'f' end) as "Assessment 4",
max(case when assessment_id=5 and passed='t' then 't' else 'f' end) as "Assessment 5",
max(case when assessment_id=6 and passed='t' then 't' else 'f' end) as "Assessment 6"
from assessments_students
group by student_id
order by student_id
and it's function look likes,
DROP FUNCTION get_passed() ;
CREATE OR REPLACE FUNCTION get_passed() RETURNS text AS
$BODY$
DECLARE
r1 record;
str_query text := '';
BEGIN
str_query :='select student_id,';
FOR r1 IN SELECT "_id" , "name" FROM Assessments
LOOP
str_query:= str_query ||
'max(case when assessment_id=' || r1."_id" || ' and passed=''t'' then ''t'' else ''f'' end) as "' || r1.name ||'",' ;
END LOOP;
str_query:=trim( trailing ',' from str_query); -- remove last semicolon
str_query:= str_query || ' from assessments_students group by student_id order by student_id';
return str_query;
END
$BODY$
LANGUAGE 'plpgsql' ;
SELECT * FROM get_passed();

Answer 2

SELECT * FROM crosstab(
'Select studentid, assessmentname,count (studentassessmentid) from ....[do ur joins here] group by studentid,assessmentname order by 1,2' AS ct (studentid, assesment1, assessment2,assesent3,assessment4);

Answer 3

I realized it doesn't work anymore. I'm using PostgreSQL 12.
So here you cannot return non-predefined table type (I mean variable columns) either select from varying char.
An example with using anonymous code block, crosstab and temp tables.
Cursors are redundand, I was solving a task that meant of using them.
CREATE EXTENSION IF NOT EXISTS tablefunc;
DO
$$
DECLARE
movie_probe CURSOR FOR
SELECT m.name movie_name, count(c.id) cinema_count
FROM movies m
JOIN sessions s ON m.id = s.movie_id
JOIN cinema_halls ch ON s.cinema_hall_id = ch.id
JOIN cinemas c ON ch.cinema_id = c.id
GROUP BY m.name
HAVING count(c.name) > 1
ORDER BY count(c.name) DESC;
movie_rec RECORD;
movie_columns TEXT DEFAULT '';
cinemas_probe CURSOR (cond_movie_name TEXT) FOR
SELECT m.name movie_name, c.name cinema_name
FROM movies m
JOIN sessions s ON m.id = s.movie_id
JOIN cinema_halls ch ON s.cinema_hall_id = ch.id
JOIN cinemas c ON ch.cinema_id = c.id
WHERE cond_movie_name = m.name
ORDER BY c.name;
cinema_rec RECORD;
cinema_row_counter INT DEFAULT 0;
BEGIN
DROP TABLE IF EXISTS cinema_multiples_aev;
CREATE TEMP TABLE cinema_multiples_aev (
row_id INT,
movie_name TEXT,
cinema_name TEXT
);
OPEN movie_probe;
LOOP
FETCH movie_probe INTO movie_rec;
EXIT WHEN NOT FOUND;
OPEN cinemas_probe(movie_rec.movie_name);
LOOP
FETCH cinemas_probe INTO cinema_rec;
EXIT WHEN NOT FOUND;
cinema_row_counter := cinema_row_counter + 1;
INSERT INTO cinema_multiples_aev (row_id, movie_name, cinema_name)
VALUES (cinema_row_counter, cinema_rec.movie_name, cinema_rec.cinema_name);
END LOOP;
CLOSE cinemas_probe;
cinema_row_counter := 0;
movie_columns := movie_columns || ', "' || movie_rec.movie_name || '" TEXT';
END LOOP;
CLOSE movie_probe;
movie_columns := substring(movie_columns FROM 2);
DROP TABLE IF EXISTS movie_multiples;
EXECUTE format('CREATE TEMP TABLE movie_multiples(row_id INT, %s)', movie_columns);
EXECUTE format(E'
INSERT INTO movie_multiples
SELECT *
FROM crosstab(\'select row_id, movie_name, cinema_name
from cinema_multiples_aev
order by 1,2\')
AS cinema_multiples_aev(row_id INT, %s);
', movie_columns, movie_columns);
ALTER TABLE movie_multiples DROP COLUMN row_id;
END
$$ LANGUAGE plpgsql;
SELECT *
FROM movie_multiples;
DROP TABLE IF EXISTS movie_multiples;
DROP TABLE IF EXISTS cinema_multiples_aev;
If it's confusing for the lack of strusture, everything could be found here github