I have a column in a postgres database which logs search querystrings for a page on our website.
The column contains data like
"a=2&b=4"
"a=2,3"
"b=4&a=3"
"a=4&a=3"
I'd like to work out the frequency of each value for a certain parameter (a).
value | freq
------|------
3 | 3
2 | 2
4 | 1
Anyway to do this in a single SQL statement?
Something like this:
with all_values as (
select string_to_array(split_part(parameter, '=', 2), ',') as query_params
from the_table d,
unnest(string_to_array(d.querystring, '&')) as x(parameter)
where x.parameter like 'a%'
)
select t.value, count(*)
from all_values av, unnest(av.query_params) as t(value)
group by t.value
order by t.value;
Online example: http://rextester.com/OXM67442
try something like this :
select data_value,count(*) from (
select data_name,unnest(string_to_array(data_values,',')) data_value from (
select split_part(data_array,'=',1) data_name ,split_part(data_array,'=',2) data_values from (
select unnest(string_to_array(mydata,'&')) data_array from mytable
) a
) b
) c where data_name='a' group by 1 order by 1
Assuming tha table that keeps the counts is called paramcount:
WITH vals(v) AS
(SELECT regexp_replace(p, '^.*=', '')
FROM regexp_split_to_table(
'b=4&a=3,2',
'&|,'
) p(p)
)
INSERT INTO paramcount (value, freq)
SELECT v, 1 FROM vals
ON CONFLICT (value)
DO UPDATE SET freq = paramcount.freq + 1
WHERE paramcount.value = EXCLUDED.value;
get csv integer after 'a='
split that to numbers
stat values
select v, count(*) from (
SELECT c,unnest(string_to_array(unnest(regexp_matches(c,'a=([0-9,]+)','g')),',')) as v FROM qrs
) x group by v;
Parametrize:
WITH argname(aname) as (values ('a'::TEXT))
select v, count(*) from (SELECT c,unnest(string_to_array(unnest(regexp_matches(c,aname||'=([0-9,]+)','g')),',')) as v FROM qrs,argname) x group by v;
Related
I have Car table. Car has is_sold and is_shipped. A Car belongs to a dealership, dealership_id (FK).
I want to run a query that tells me the count of sold cars and the count of shipped cars for a given dealership all in one result.
sold_count | shipped_count
10 | 4
The single queries I have look like this:
select count(*) as sold_count
from car
where dealership_id=25 and is_sold=true;
and
select count(*) as shipped_count
from car
where dealership_id=25 and is_shipped=true;
How do I combine the two to get both counts in one result?
This will do:
select dealership_id,
sum(case when is_sold is true then 1 else 0 end),
sum(case when is_shipped is true then 1 else 0 end)
from cars group by dealership_id;
You can use the filter clause of the Aggregate function. (see demo)
select dealership_id
, count(*) filter (where is_sold) cars_sold
, count(*) filter (where is_shipped) cars_shipped
from cars
where dealership_id = 25
group by dealership_id;
You can also using cross join.
select 'hello' as col1, 'world' as col2;
return:
col1 | col2
-------+-------
hello | world
(1 row)
similarly,
with a as
(
select count(*) as a1 from emp where empid> 5),
b as (
select count(*) as a2 from emp where salary > 6000)
select * from a, b;
or you can even apply to different table. like:
with a as
(select count(*) as a1 from emp where empid> 5),
b as
(select count(*) as a2 from ab )
select * from a, b;
with a as
(
select count(*) as sold_count
from car
where dealership_id=25 and is_sold=true
),
b as
(
select count(*) as shipped_count
from car
where dealership_id=25 and is_shipped=true
)
select a,b;
further reading: https://www.postgresql.org/docs/current/queries-table-expressions.html.
https://stackoverflow.com/a/26369295/15603477
I can't replace every 2 characters of a string with a '.'
select STUFF('abcdefghi', 3, 1, '.') c3,STUFF('abcdefghi', 5, 1,
'.') c5,STUFF('abcdefghi', 7, 1, '.') c7,STUFF('abcdefghi', 9, 1, '.')
c9
if I use STUFF I should subsequently overlap the strings c3, c5, c7 and c9. but I can't find a method
can you help me?
initial string:
abcdefghi
the result I would like is
ab.de.gh.
the string can be up to 50 characters
Create a numbers / tally / digits table, if you don't have one already, then you can use this to target each character position:
with digits as ( /* This would be a real table, here it's just to test */
select n from (values(1),(2),(3),(4),(5),(6),(7),(8),(9),(10))x(n)
), t as (
select 'abcdefghi' as s
)
select String_Agg( case when d.n%3 = 0 then '.' else Substring(t.s, d.n, 1) end, '')
from t
cross apply digits d
where d.n <Len(t.s)
Using for xml with existing table
with digits as (
select n from (values(1),(2),(3),(4),(5),(6),(7),(8),(9),(10))x(n)
),
r as (
select t.id, case when d.n%3=0 then '.' else Substring(t.s, d.n, 1) end ch
from t
cross apply digits d
where d.n <Len(t.s)
)
select result=(select '' + ch
from r r2
where r2.id=r.id
for xml path('')
)
from r
group by r.id
You can try it like this:
Easiest might be a quirky update ike here:
DECLARE #string VARCHAR(100)='abcdefghijklmnopqrstuvwxyz';
SELECT #string = STUFF(#string,3*A.pos,1,'.')
FROM (SELECT TOP(LEN(#string)/3) ROW_NUMBER() OVER(ORDER BY (SELECT NULL))
FROM master..spt_values) A(pos);
SELECT #string;
Better/Cleaner/Prettier was a recursive CTE:
We use a declared table to have some tabular sample data
DECLARE #tbl TABLE(ID INT IDENTITY, SomeString VARCHAR(200));
INSERT INTO #tbl VALUES('')
,('a')
,('ab')
,('abc')
,('abcd')
,('abcde')
,('abcdefghijklmnopqrstuvwxyz');
--the query
WITH recCTE AS
(
SELECT ID
,SomeString
,(LEN(SomeString)+1)/3 AS CountDots
,1 AS OccuranceOfDot
,SUBSTRING(SomeString,4,LEN(SomeString)) AS RestString
,CAST(LEFT(SomeString,2) AS VARCHAR(MAX)) AS Growing
FROM #tbl
UNION ALL
SELECT t.ID
,r.SomeString
,r.CountDots
,r.OccuranceOfDot+2
,SUBSTRING(RestString,4,LEN(RestString))
,CONCAT(Growing,'.',LEFT(r.RestString,2))
FROM #tbl t
INNER JOIN recCTE r ON t.ID=r.ID
WHERE r.OccuranceOfDot/2<r.CountDots-1
)
SELECT TOP 1 WITH TIES ID,Growing
FROM recCTE
ORDER BY ROW_NUMBER() OVER(PARTITION BY ID ORDER BY OccuranceOfDot DESC);
--the result
1
2 a
3 ab
4 ab
5 ab
6 ab.de
7 ab.de.gh.jk.mn.pq.st.vw.yz
The idea in short
We use a recursive CTE to walk along the string
we add the needed portion together with a dot
We stop, when the remaining length is to short to continue
a little magic is the ORDER BY ROW_NUMBER() OVER() together with TOP 1 WITH TIES. This will allow all first rows (frist per ID) to appear.
I will give you example of table that I have:
Supplier | Value
sup1 | 4
sup2 | 1
sup1 | 0
sup1 | 3
sup2 | 5
I need a result that will do average by supplier, but if there is value 0 for a supplier, do not average, but return 0 instead
It should look like this:
Supplier | Value
sup1 | 0
sup2 | 3
This is a little trick but it should work :
SELECT Supplier,
CASE WHEN MIN(ABS(Value)) = 0 THEN 0 ELSE AVG(Value) END
FROM TableTest
GROUP BY Supplier
EDIT : Using the ABS() function let you avoid having problems with negative values
DECLARE #TAB TABLE (SUPPLIER VARCHAR(50),VALUE INTEGER)
INSERT INTO #TAB
SELECT 'sup1',4
UNION ALL
SELECT 'sup2',1
UNION ALL
SELECT 'sup1',0
UNION ALL
SELECT 'sup1',3
UNION ALL
SELECT 'sup2',5
SELECT * FROM #TAB
SELECT T1.SUPPLIER,CASE WHEN EXISTS(SELECT 1 FROM #TAB T WHERE T.SUPPLIER = T1.SUPPLIER AND T.VALUE = 0) THEN 0 ELSE AVG(T1.VALUE) END AS VALUE
FROM #TAB T1
GROUP BY T1.SUPPLIER
Result
SUPPLIER VALUE
sup1 0
sup2 3
Using the following query is one of the way to do.
First I push the supplier which has the Value = 0, then based on the result, I will do the remaining calculation and finally using UNION to get the expected result:
DECLARE #ZeroValue TABLE (Supplier VARCHAR (20));
INSERT INTO #ZeroValue (Supplier)
SELECT Supplier FROM TestTable WHERE Value = 0
SELECT Supplier, 0 AS Value FROM #ZeroValue
UNION
SELECT T.Supplier, AVG(T.Value) AS Value
FROM TestTable T
JOIN #ZeroValue Z ON Z.Supplier != T.Supplier
GROUP BY T.Supplier
Schema used for the sample:
CREATE TABLE TestTable (Supplier VARCHAR (20), Value INT);
INSERT INTO TestTable (Supplier, Value) VALUES
('sup1', 4), ('sup2', 1), ('sup1', 0), ('sup1', 3), ('sup2', 5);
Please find the working demo on db<>fiddle
I would like to produce a string containing some parsed numeric ranges.
I have a table with some data
b_id,s_id
1,50
1,51
1,53
1,61
1,62
1,63
2,91
2,95
2,96
2,97
Using only SQL in PostgreSQL, how could I produce this output:
b_id,s_seqs
1,"50-51,53,61-63"
2,"91,95-97"
How on earth do I do that?
select b_id, string_agg(seq, ',' order by seq_no) as s_seqs
from (
select
b_id, seq_no,
replace(regexp_replace(string_agg(s_id::text, ','), ',.+,', '-'), ',', '-') seq
from (
select
b_id, s_id,
sum(mark) over w as seq_no
from (
select
b_id, s_id,
(s_id- 1 <> lag(s_id, 1, s_id) over w)::int as mark
from my_table
window w as (partition by b_id order by s_id)
) s
window w as (partition by b_id order by s_id)
) s
group by 1, 2
) s
group by 1;
Here you can find a step-by-step analyse from the innermost query towards the outside.
I have a table looks like,
x y
1 2
2 null
3 null
1 null
11 null
I want to fill the null value by conducting a rolling
function to apply y_{i+1}=y_{i}+x_{i+1} with sql as simple as possible (inplace)
so the expected result
x y
1 2
2 4
3 7
1 8
11 19
implement in postgresql. I may encapsulate it in a window function, but the implementation of custom function seems always complex
WITH RECURSIVE t AS (
select x, y, 1 as rank from my_table where y is not null
UNION ALL
SELECT A.x, A.x+ t.y y , t.rank + 1 rank FROM t
inner join
(select row_number() over () rank, x, y from my_table ) A
on t.rank+1 = A.rank
)
SELECT x,y FROM t;
You can iterate over rows using a recursive CTE. But in order to do so, you need a way to jump from row to row. Here's an example using an ID column:
; with recursive cte as
(
select id
, y
from Table1
where id = 1
union all
select cur.id
, prev.y + cur.x
from Table1 cur
join cte prev
on cur.id = prev.id + 1
)
select *
from cte
;
You can see the query at SQL Fiddle. If you don't have an ID column, but you do have another way to order the rows, you can use row_number() to get an ID:
; with recursive sorted as
(
-- Specify your ordering here. This example sorts by the dt column.
select row_number() over (order by dt) as id
, *
from Table1
)
, cte as
(
select id
, y
from sorted
where id = 1
union all
select cur.id
, prev.y + cur.x
from sorted cur
join cte prev
on cur.id = prev.id + 1
)
select *
from cte
;
Here's the SQL Fiddle link.