My question is somewhat related to :
Display multiple values of a column in one row (SQL Oracle)
However, I could not achieve desired results. Following is my Problem Statement;
I have a SQL Query ;
SELECT initiator_msisdn, trx_type || '/' || SUM(trx_amt/100) || '/' || SUM(merchant_comm_amt/100) agent_data
FROM LBI_DM_MK.T_M_INTERNAL_AUDIT_D
WHERE DATA_DATE = '20180401'
AND trx_status ='Completed'
GROUP BY initiator_msisdn, trx_type
;
That returns these rows;
The SQL That brings this data is ;
But, I want following result.
Please help to sort out this issue;
You could use LISTAGG:
WITH cte AS (
SELECT initiator_msisdn,
trx_type || '/' || SUM(trx_amt/100) || '/' ||
SUM(merchant_comm_amt/100) agent_data
FROM LBI_DM_MK.T_M_INTERNAL_AUDIT_D
WHERE DATA_DATE = '20180401'
AND trx_status ='Completed'
GROUP BY initiator_msisdn, trx_type
)
SELECT initiator_msisdn,
LISTAGG(agent_data, '|') WITHIN GROUP (ORDER BY agent_data) AS agent_data
FROM cte
GROUP BY initiator_msisdn;
Related
I have an update query which works fine. once I put this in my PLSQL block it returns NULL and there is no error. Prior to this part of the block I insert a default value into the target table, therefore I know the update is actually updating the fields to NULL. Since this query is counting the rows of each column group by another column, then there is always a value for it.and I get the correct count from the query outside of the string..
when I place the query in my plsql Block I use the Execute immediate and the query would be a string in which my_table_name and my_column_names and Table_1 are variable.
I searched alot and found things like I should commit and etc. But problem still there.
for example:
Need help in execute immediate update query
update Table_FINAL r
set column_1 =
(
select max(totalcount)
from (
select 'my_table_name' as table_name, 'my_collumn_name' as column_name, column_3, count(*) as totalcount
from my_table_name a
where exists (select 1 from Table_2 where Table_2.column_x = a.column_x)
group by column_3
) s
where r.column_3 = s.column_3
)
;
and here the string:
execute immediate 'update Table_FINAL r
set column_1 =
(
select max(totalcount)
from (
select ''' || my_table_name || ''' as table_name, ''' || my_collumn_name || ''' as column_name, column_3, count(*) as totalcount
from ' || my_table_name || ' a
where exists (select 1 from Table_2 where Table_2.column_x = a.column_x)
group by column_3
) s
where r.column_3 = s.column_3
)'
;
it updates column_1 to NULL
If your dynamic SQL returns no rows, totalcount will be null, so max(totalcount) will be null and in that circumstance column_1 will be updated to null.
There are a couple of obvious solutions:
only execute the update if there's a value: ... where r.column_3 = s.column_3 and totalcount is not null.
handle the null : select max(nvl(totalcount,0)) …
Now, you assert that the update query "works fine". Does it work fine for all values you pass as my_table_name? Another reason why dynamic SQL is hard is that we can't look at the source code and know for certain what it's going to do at runtime. So you need to so some debugging. Run the dynamic SELECT statement without the update and see what you're actually executing:
execute immediate ' select max(totalcount) from (
select ''' || my_table_name || ''' as table_name, ''' || my_collumn_name || ''' as column_name, column_3, count(*) as totalcount
from ' || my_table_name || ' a
where exists (select 1 from Table_2 where Table_2.column_x = a.column_x)
group by column_3)' into l_total_count;
dbms_ouput.put_line(my_table_name ||'.'|| my_collumn_name ||' max(totalcount) = ' || l_total_count);
Remember to enable SERVEROUTPUT in whatever client you're using.
the update string is inside a loop and it seems that in each loop round it indeed does the correct update but then ir updates the rest to NULL … But I do not fully get what I should do since I hvae hundreds of fields.. (sic)
I guessed you were calling this code from a loop. So, it's not that you don't overwrite the "correct update" with null but that you don't want to over-write any value with any subsequent value. I can offer a few suggestions but really this is your data model and your business logic, so only you can decide the correct way to handle this.
Aggregate totalcount like this: set column_1 = column_1 + total_count. For this to work you'll need to apply nvl(max(totalcount),0).
Add a column to TABLE_FINAL to store values of my_collumn_value. Reference that column in the UPDATE statement's WHERE clause. (Obviously this supposes a completely different result set from the previous suggestion). You may need a column for values of my_table_name too.
Our vendor's database has Number types for all numbers including whole numbers and decimal numbers. Literally, every numeric type column is created as NUMBER without precision and scale.
This is a big problem as we need to map these columns to proper data types on our target system, we are loading data from these tables into.
We need to know if a number is an integer or decimal.
Other than doing a random sampling/data profiling, is it possible to infer proper data types?
UPDATE:
I accepted the answer below and suggestion from #Bohemian. In addition to that, I will use SAMPLE clause that will do a random sampling of the table since my source tables are huge (many billions of rows).
SELECT
MAX(CASE WHEN col1 IS NOT NULL AND col1 <> round(col1, 0) then 1 else 0 end) as col1,
MAX(CASE WHEN col2 IS NOT NULL AND col2 <> round(col2, 0) then 1 else 0 end) as col2
FROM TABLE
SAMPLE(0.05)
If I want to sample only X rows, use formula below to SAMPLE(N):
Xrows*100/table_rows_total
You can try selecting each FIELD, and seeing if all values of FIELD are equal to ROUND(FIELD, 0). If they are, then that field should be integer. If not, decimal.
I have answered it in this other post and the query that you would use to find the maximum number of decimal places in all of the number columns is the same as that one.
To identify the columns with the their maximum decimal position digits, you can run the SQL below after substituting the MY_SCHEMA, MY_TABLE and the number 10 with say 25 to identify columns that have values over 25 decimal places. This SQL will generate a SQL that should be run to get your result.
SELECT 'SELECT ' || LISTAGG('MAX(LENGTH(TO_CHAR(ABS(' || column_name || ') - FLOOR(ABS(' || column_name || '))))) - 1 AS decimals_' || column_name || CHR(13)
, CHR(9)|| ', ') WITHIN GROUP (ORDER BY rn) ||
' FROM ' || owner || '.' || table_name || CHR(13) ||
' WHERE ' || CHR(13) ||
LISTAGG('(LENGTH(TO_CHAR(ABS(' || column_name || ') - FLOOR(ABS(' || column_name || ')))) - 1) > 10 ' || CHR(13)
, CHR(9)|| ' OR ')
WITHIN GROUP (ORDER BY rn) AS Nasty_Numbers_Finder_Query
FROM
(
SELECT owner, table_name, column_name,
row_number() OVER ( PARTITION BY table_name ORDER BY rownum) rn
FROM dba_tab_columns
WHERE
OWNER = 'MY_SCHEMA'
AND table_name = 'MY_TABLE'
AND (data_type LIKE '%FLOAT%'
OR data_type LIKE '%NUMERIC%')
) a
GROUP BY owner, table_name
For more information, I have blogged about it here.
I have following sql query in code, which is makes query go slow:
SELECT a.*,
(SELECT CASE
WHEN (score IS NOT NULL OR comments IS NOT NULL)
THEN
' ( score : ' || TO_CHAR (SCORE) || ' )'
ELSE
' ( hələ )'
END
FROM t_lecture_task_present b
WHERE b.t_idx = a.t_idx AND B.STUDENT_ID = '{$member_code}')
AS task_score
FROM t_lecture_task a
WHERE a.lec_open_idx = '24422'
ORDER BY s_date ASC, t_idx ASC
(16 seconds)
If I try query without
(SELECT CASE
WHEN (score IS NOT NULL OR comments IS NOT NULL)
THEN
' ( score : ' || TO_CHAR (SCORE) || ' )'
ELSE
' ( hələ )'
END
FROM t_lecture_task_present b
WHERE b.t_idx = a.t_idx AND B.STUDENT_ID = '{$member_code}')
as task_score
it works faster.
So, I tried removing or comments is not null, and it works got 2 seconds faster.
You can't compare the performance of queries that produce different results :)
Depending on data distribution in your tables, you would likely benefit from these indexes:
t_lecture_task(lec_open_idx)
t_lecture_task_present(t_idx, student_id)
Try to re-write your query to use joins instead of scalar sub queries (select as a column). Not only are they more awkward to read, they are more difficult to optimize.
select a.*
,case when score is not null
or comments is not null then ' ( score : ' || to_char (score) || ' )'
else ' ( hələ )'
end as task_score
from t_lecture_task a
left join t_lecture_task_present b on(
b.t_idx = a.t_idx
and b.student_id = '{$member_code}')
)
where a.lec_open_idx = '24422';
I need a query to get Top N employees working on every project in a specific month based on working hours. And I need the same query for all time (without specifying a month).
I get two hints first Hint: Use Substitution variables to prompt the user to enter values for N and a month.
Second Hint: Use the rank analytical function. If two employees tie they should get the same rank.
Just right now i have this not completed solution, and not sure if i should complete it:
SELECT BSR_PROJ.PROJECT_NAME,
BSR_TM.FNAME || ' ' || BSR_TM.MNAME || ' ' || BSR_TM.LNAME EMPLOYEE_NAME,
BSR_TRD.WORK_ITEM_DATE,
RANK() OVER (PARTITION BY BSR_PROJ.PROJECT_NAME ORDER BY BSR_TRD.WORK_ITEM_DATE ASC) EMPRANK
FROM BSR_TEAM_REPORT_DETAILS BSR_TRD,
BSR_PROJECTS BSR_PROJ,
BSR_TEAM_MEMBERS BSR_TM
WHERE BSR_TRD.BSR_TEAM_RES_ID = BSR_TM.ID
AND BSR_TRD.BSR_PRJ_ID = BSR_PROJ.ID
;
You need to include the month in the analytic function, as rank is calculated per project per month; truncating the date with a month format mask is an easy way to achieve this.
You also need to include that truncated month in the projection, so you can filter on it. I have chosen to present the month in the format 2015-12. You may wish to show it differently.
The query you have will generate the whole set of ranks for all employees across all assignments. You need an outer query to apply the filtering requirements. My solution uses SQL Plus substitution variables rather than bind variables, but the principle is the same:
select distinct project_name
, employee_name
from (
SELECT BSR_PROJ.PROJECT_NAME,
BSR_TM.FNAME || ' ' || BSR_TM.MNAME || ' ' || BSR_TM.LNAME EMPLOYEE_NAME,
to_char(trunc(bsr_trd.work_item_date, 'MM'), 'yyyy-mm') as project_month,
RANK() OVER (PARTITION BY BSR_PROJ.PROJECT_NAME,
trunc(bsr_trd.work_item_date, 'MM')
ORDER BY BSR_TRD.WORK_ITEM_DATE ASC) EMPRANK
FROM BSR_TEAM_REPORT_DETAILS BSR_TRD,
BSR_PROJECTS BSR_PROJ,
BSR_TEAM_MEMBERS BSR_TM
WHERE BSR_TRD.BSR_TEAM_RES_ID = BSR_TM.ID
AND BSR_TRD.BSR_PRJ_ID = BSR_PROJ.ID
)
where project_month = '&proj_month'
and emprank <= &rnk;
The answer i found as below :
SELECT *
FROM (SELECT RANK ()
OVER (PARTITION BY PROJ_NAME ORDER BY WORKING_HOURS DESC)
AS EMPRANK,
PROJ_NAME,
EMPLOYEE_NAME,
WORKING_HOURS
FROM ( SELECT BSR_PROJ.PROJECT_NAME AS PROJ_NAME,
BSR_TM.FNAME
|| ' '
|| BSR_TM.MNAME
|| ' '
|| BSR_TM.LNAME
EMPLOYEE_NAME,
SUM (WORK_ITEM_CONSUMED_HOURS) AS WORKING_HOURS
FROM BSR_TEAM_REPORT_DETAILS BSR_TRD,
BSR_PROJECTS BSR_PROJ,
BSR_TEAM_MEMBERS BSR_TM
WHERE BSR_TRD.BSR_TEAM_RES_ID = BSR_TM.ID
AND BSR_TRD.BSR_PRJ_ID = BSR_PROJ.ID
AND TO_CHAR (TRUNC (BSR_TRD.WORK_ITEM_DATE, 'MM'),
'YYYY-MM') = '&PROJ_MONTH'
GROUP BY BSR_PROJ.PROJECT_NAME,
BSR_TM.FNAME
|| ' '
|| BSR_TM.MNAME
|| ' '
|| BSR_TM.LNAME)) INNER_TABLE
WHERE EMPRANK <= &RNK;
I am trying to modify a query that someone else wrote which is taking a really long time to run. The problem has to do with the <> portion of the exists query. Any idea how this can be changed to run quicker?
SELECT m.level4 center, cc.description, m.employeename, m.empno,
TO_DATE (ct.tsdate, 'dd-mon-yyyy') tsdate, ct.starttime, ct.endtime,
ct.DURATION,
NVL (DECODE (ct.paycode, ' ', 'REG', ct.paycode), 'REG') paycode,
ct.titlecode, ct.costcenter, m.tsgroup
FROM clairvia_text ct, MASTER m, costcenteroutbound cc
WHERE ct.recordtype = '1'
AND ct.empno = m.empno
AND m.level4 = cc.center
AND EXISTS (
SELECT ct1.recordtype,ct1.empno,ct1.tsdate,ct1.processdate
FROM clairvia_text ct1
WHERE ct.recordtype = ct1.recordtype
AND ct.empno = ct1.empno
AND ct.tsdate = ct1.tsdate
AND ct.processdate = ct1.processdate
group by ct1.recordtype,ct1.empno,ct1.tsdate,ct1.processdate
having count(*) < 2)
Oracle can be finnicky with exists statements and subqueries. A couple of things to try:
Change the exists to an "in"
Change the exists to a group by statement with a "having count(1) > 1". This could even be changed into a join.
I'm assuming indexes are not an issue.
You can use analytic function count here to eliminate duplicated rows.
select * from (
SELECT m.level4 center, cc.description, m.employeename, m.empno,
TO_DATE (ct.tsdate, 'dd-mon-yyyy') tsdate, ct.starttime, ct.endtime,
ct.DURATION,
NVL (DECODE (ct.paycode, ' ', 'REG', ct.paycode), 'REG') paycode,
ct.titlecode, ct.costcenter, m.tsgroup,
count(1) over (partition by ct.recordtype,ct.empno,ct.tsdate,ct.processdate
order by null) cnt
FROM clairvia_text ct, MASTER m, costcenteroutbound cc
WHERE ct.recordtype = '1'
AND ct.empno = m.empno
AND m.level4 = cc.center
) where cnt=1
I do not have your structures and data, so I run similar queries with all_tab_cols and first query took ~500s on my laptop and second query ~2s.
-- slow test
select count(1)
from all_tab_cols atc
where exists (
select 1
from all_tab_cols atc1
where atc1.column_name = atc.column_name
group by column_name
having count(1) = 1)
-- fast test
select count(1)
from (
select column_name,
count(1) over (partition by atc.column_name order by null) cnt
from all_tab_cols atc
)
where cnt = 1