Have a SQL query on Oracle 11g which returns the count of whether a record having certain ID and status exists within +/- 15 minutes range in a table.
Now I wish to ignore the current date by adding a condition like AND TIMESTAMP < trunc(sysdate).
However, for cases where the record exists in todays date I wish to ignore the date comparison check in the query '2010-07-20 19:15:11' >= TO_CHAR(TIMESTAMP - (1/1440*15), 'YYYY-MM-DD HH24:MI:SS')
AND '2010-07-20 19:15:11' <= (TO_CHAR(TIMESTAMP + (1/1440*15), 'YYYY-MM-DD HH24:MI:SS'))
SELECT count(1) AS COUNT
FROM MASTER_ONE
WHERE ID='123' AND STATUS= 'ACTIVE'
AND '2010-07-20 19:15:11' >= TO_CHAR(TIMESTAMP - (1/1440*15), 'YYYY-MM-DD HH24:MI:SS')
AND '2010-07-20 19:15:11' <= (TO_CHAR(TIMESTAMP + (1/1440*15), 'YYYY-MM-DD HH24:MI:SS'))
UNION ALL
SELECT count(1) AS COUNT
FROM MASTER_TWO
WHERE ID='321' AND STATUS= 'ACTIVE'
AND '2010-07-20 19:15:11' >= TO_CHAR(TIMESTAMP - (1/1440*15), 'YYYY-MM-DD HH24:MI:SS')
AND '2010-07-20 19:15:11' <= (TO_CHAR(TIMESTAMP + (1/1440*15), 'YYYY-MM-DD HH24:MI:SS'))
How do I do this?
The first problem with your query is that you're doing a string comparison on the date. Use to_date instead of to_char and let Oracle help you out.
SELECT
to_date('2010-07-20 19:15:11', 'YYYY-MM-DD HH24:MI:SS') AS orig_date
, to_date('2010-07-20 19:15:11', 'YYYY-MM-DD HH24:MI:SS') - 1 / 24 / 4 AS fifteen_min_prior
, to_date('2010-07-20 19:15:11', 'YYYY-MM-DD HH24:MI:SS') + 1 / 24 / 4 AS fifteen_min_after
FROM dual;
Output:
ORIG_DATE FIFTEEN_MIN_PRIOR FIFTEEN_MIN_AFTER
------------------------- ------------------------- -------------------------
20-JUL-10 07:15:11 PM 20-JUL-10 07:00:11 PM 20-JUL-10 07:30:11 PM
Then use can use those dates in a BETWEEN condition in the predicate. See Oracle date "Between" Query.
I'm not quite clear what you mean by "However, for cases where the record exists in todays date I wish to ignore the date comparison check in the query." You'd just written that you want to exclude values from the current day. Either you're excluding today's records or you're not.
Ok, you can try something like this, if I understood you correctly:
SELECT count(1) AS COUNT
FROM MASTER_ONE
WHERE ID='123' AND STATUS= 'ACTIVE'
AND (timestamp > trunc(sysdate)
OR (timestamp < trunc(sysdate)
AND timestamp BETWEEN to_date(:yourInputDate,'DD/MM/YYYY HH24:MI:SS') - (1/1440*15)
AND to_date(:yourInputDate,'DD/MM/YYYY HH24:MI:SS') + (1/1440*15)))
UNION ALL
SELECT count(1) AS COUNT
FROM MASTER_TWO
WHERE ID='321' AND STATUS= 'ACTIVE'
AND (timestamp > trunc(sysdate)
OR (timestamp < trunc(sysdate)
AND timestamp BETWEEN to_date(:yourInputDate,'DD/MM/YYYY HH24:MI:SS') - (1/1440*15)
AND to_date(:yourInputDate,'DD/MM/YYYY HH24:MI:SS') + (1/1440*15)))
In this Select, you only apply the 15 minutes condition if your timestamp column has a date prior to sysdate.
Related
I have a sales table with created datetime, my business hours are from 9 AM to 2 AM in the night on the following day. I am trying to convert the dates into my business date.
01/08/22 09:39:12.000000000 AM +04:00
Lets say I have a sale at 1 AM, this sale has to be considered in the previous day.
Any function that can help me solve this issue would be appreciated
It might be a bit of an overkill, but you could just use EXTRACT:
WITH dat AS
(
SELECT to_date('01/08/22 09:39:12','DD/MM/YY HH24:MI:SS') AS t_stmp FROM dual UNION ALL
SELECT to_date('02/08/22 01:03:15','DD/MM/YY HH24:MI:SS') FROM dual UNION ALL
SELECT to_date('02/08/22 08:27:33','DD/MM/YY HH24:MI:SS') FROM dual UNION ALL
SELECT to_date('02/08/22 14:11:51','DD/MM/YY HH24:MI:SS') FROM dual UNION ALL
SELECT to_date('02/08/22 02:01:15','DD/MM/YY HH24:MI:SS') FROM dual
)
SELECT CASE WHEN EXTRACT(HOUR FROM CAST(t_stmp AS TIMESTAMP)) BETWEEN 2 AND 8 THEN -1
ELSE 0
END + TRUNC(t_stmp,'DD') AS business_date
FROM dat;
business_date
01.08.2022
02.08.2022
01.08.2022
02.08.2022
01.08.2022
It looks like you just need to make a 2 hour shift to get your sales in the right date. You can add or substract hours from DATE/DATETIME/TIMESTAMP data type. If your column is TIMESTAMP then it would be like this:
-- when selecting data for date of sales
SELECT TRUNC(your_column_name - INTERVAL '2' HOUR, 'dd') "SALE_DATE"
-- And/Or
WHERE TRUNC(your_column_name - INTERVAL '2' HOUR, 'dd') = :DATE_OF_SALES
-- TRUNC function always returns DATE datatype
--
-- The opposite conversion would be
CAST(your_datetime_column + INTERVAL '2' HOUR as TIMESTAMP) ...
Here is the small sample with result:
SELECT
to_char(SYSDATE, 'dd.mm.yyyy hh24:mi:ss') "DATETIME",
to_char(SYSDATE - INTERVAL '2' HOUR, 'dd.mm.yyyy hh24:mi:ss') "DATETIME_MINUS_2H",
to_char(SYSDATE + INTERVAL '2' HOUR, 'dd.mm.yyyy hh24:mi:ss') "DATETIME_PLUS_2H",
to_char(SYSDATE - INTERVAL '10' HOUR, 'dd.mm.yyyy hh24:mi:ss') "DATETIME_MINUS_10H"
FROM
DUAL
--
-- R e s u l t
--
-- DATETIME DATETIME_MINUS_2H DATETIME_PLUS_2H DATETIME_MINUS_10H
-- ------------------- ------------------- ------------------- -------------------
-- 07.08.2022 09:58:38 07.08.2022 07:58:38 07.08.2022 11:58:38 06.08.2022 23:58:38
The last column now has the date from day before.
Given:
INSERT INTO EP_ACCESS (PROFILE_ID, EPISODE_ID, START_TIMESTAMP, DISCONNECT_TIMESTAMP)
VALUES ('1', '1', TO_DATE('2020-01-01 00:00:01','yyyy/mm/dd hh24:mi:ss'), TO_DATE('2020-01-01 00:00:02','yyyy/mm/dd hh24:mi:ss'));
How can I select those who start_timestamp is in 2020?
You would use:
where start_timestamp >= date '2020-01-01' and
start_timestamp < date '2021-01-01'
Of course, you can use a timestamp literal if you prefer typing longer strings.
There are several options.
1 - Use BETWEEN
SELECT *
FROM EP_ACCESS
WHERE START_TIMESTAMP BETWEEN TO_DATE('2020-01-01 00:00:00', 'YYYY-MM-DD HH24:MI:SS')
AND TO_DATE('2020-12-31 23:59:59', 'YYYY-MM-DD HH24:MI:SS')
or
SELECT *
FROM EP_ACCESS
WHERE START_TIMESTAMP BETWEEN DATE '2020-01-01'
AND DATE '2021-01-01' - INTERVAL '1' SECOND
2 - Use EXTRACT
SELECT *
FROM EP_ACCESS
WHERE EXTRACT(YEAR FROM START_TIMESTAMP) = 2020
3 - Use TRUNC
SELECT *
FROM EP_ACCESS
WHERE TRUNC(START_TIMESTAMP, 'YYYY') = DATE '2020-01-01'
Of these options, BETWEEN will probably provide the best performance as the other two require executing a function against the START_TIMESTAMP field in every row in the table.
I am trying to get time difference in Oracle database. The time difference for all the rows is okay except for one row. Here is the query
SELECT MAX(REGEXP_SUBSTR (CAST(TO_DATE(call_end, 'YYYY/MM/DD HH24:MI:SS') AS TIMESTAMP) - CAST(TO_DATE(call_start, 'YYYY/MM/DD HH24:MI:SS') AS TIMESTAMP), '\d{2}:\d{2}:\d{2}')) AS call_time
FROM calls
The time difference occurs with the following row:
call_end: '2020-02-20 13:00:20'
call_start: '2020-02-20 12:56:03'
The returned result is '11:55:43' which is wrong. As the correct answer should be '00:04:17'
It looks like your table already stores the call start/end times as dates, and you're doing an implicit conversion from date to string, then converting back to a date. You can see that result if you have HH instead of HH24 in your NLS_DATE_FORMAT setting:
alter session set nls_date_format = 'YYYY-MM-DD HH:MI:SS';
with calls (call_end, call_start) as (
select cast(timestamp '2020-02-20 13:00:20' as date), cast(timestamp '2020-02-20 12:56:03' as date) from dual
)
SELECT MAX(REGEXP_SUBSTR (CAST(TO_DATE(call_end, 'YYYY/MM/DD HH24:MI:SS') AS TIMESTAMP) - CAST(TO_DATE(call_start, 'YYYY/MM/DD HH24:MI:SS') AS TIMESTAMP), '\d{2}:\d{2}:\d{2}')) AS call_time
FROM calls;
CALL_TIME
---------------------------
11:55:43
When you do
TO_DATE(call_end, 'YYYY/MM/DD HH24:MI:SS')
because it's already a date you're really doing:
TO_DATE(TO_CHAR(call_end, <NLS_DATE_FORMAT>), 'YYYY/MM/DD HH24:MI:SS')
so with my setting (and yours must be similar, maybe with slashes instead of dashes) that is:
TO_DATE(TO_CHAR(call_end, 'YYYY-MM-DD HH:MI:SS'), 'YYYY/MM/DD HH24:MI:SS')
and the mismatch between HH and HH24 becomes more obvious. So, you're actually converting the strings 2020-02-20 01:00:20 and 2020-02-20 12:56:03 back to dates, and the time difference between 01:00:20 and 12:56:03 is 11:55:43. Well - actually, it is minus 11 hours:
SELECT CAST(TO_DATE(call_end, 'YYYY/MM/DD HH24:MI:SS') AS TIMESTAMP) - CAST(TO_DATE(call_start, 'YYYY/MM/DD HH24:MI:SS') AS TIMESTAMP)
FROM calls;
CAST(TO_DATE(CALL_E
-------------------
-00 11:55:43.000000
but your regex isn't picking that up.
As they are dates skip that part of the conversion completely, and just cast directly to timestamps if you want intervals to work from:
SELECT MAX(REGEXP_SUBSTR (CAST(call_end AS TIMESTAMP) - CAST(call_start AS TIMESTAMP), '\d{2}:\d{2}:\d{2}')) AS call_time
FROM calls;
CALL_TIME
---------------------------
00:04:17
or use a substring instead of a regex, as #MTO showed.
You could also leave them as dates, get the difference as a fraction of a day, add that back to any nominal date at midnight, and then convert the resulting date to a string:
SELECT TO_CHAR(date '2000-01-01' + MAX(call_end - call_start), 'HH24:MI:SS') AS call_time
FROM calls;
CALL_TIM
--------
00:04:17
This won't work properly for a call that lasts more than 24 hours (used to see that a lot with modem calls, but can still happen); but neither will your interval approach. Both ignore any full days and just show the remainder. There are ways to handle that of course, but you'd need to decide how you want to display it - with separate day count (like interval does anyway), or with the 'hours' number allowed to go above 24... but then you may go above 99 hours...
Your question shows the call end time as 2020-02-20 13:00:20, which suggests that's how your client is displaying it when you query the table. Some clients (I think PL/SQL Developer, but not sure, it's been a while) use their own preferences/settings instead of honouring the session's NLS settings. But that has no effect on how Oracle behaves internally when it has to do implicit conversions.
I'm not sure if you want the result as an interval or timestamp, but this should do what you want:
select t.*, call_end - call_start,
substr(to_char(call_end - call_start, 'HH24:MI:SS'), 12, 8) as str from (select timestamp '2020-02-20 13:00:20.000' as call_end,
timestamp '2020-02-20 12:56:03.000' as call_start
from dual) t
Here is a db<>fiddle.
Assuming your data types are strings (you really should store them as DATE data types), you can use:
SELECT SUBSTR(
MAX(
( TO_DATE( call_end, 'YYYY-MM-DD HH24:MI:SS' )
- TO_DATE( call_start, 'YYYY-MM-DD HH24:MI:SS' )
) DAY(1) TO SECOND
),
4,
8
) AS call_time
FROM calls
If they are already DATE data types then, in the MAX aggregation, just use:
( call_end - call_start ) DAY(1) TO SECOND
So for your data:
CREATE TABLE calls ( call_end, call_start ) AS
SELECT '2020-02-20 13:00:20', '2020-02-20 12:56:03' FROM DUAL
This outputs:
| CALL_TIME |
| :-------- |
| 00:04:17 |
db<>fiddle here
Mr Gyl, You can as well do: db <> fiddle :This solution will also consider when the call has lapsed over days by adding those hours together:
WITH da AS (
SELECT
NUMTODSINTERVAL(TO_DATE('2020-02-20 13:00:20', 'yyyy-mm-dd hh24:mi:ss') - TO_DATE('2020-02-20 12:56:03', 'yyyy-mm-dd hh24:mi:ss'), 'DAY') AS call_diff
FROM
dual ) SELECT
EXTRACT( DAY FROM call_diff )*24 + EXTRACT( HOUR FROM call_diff )|| ':' || EXTRACT( MINUTE FROM call_diff ) || ':' || EXTRACT( SECOND FROM call_diff ) DIFFERENCE
FROM
da
DIFFERENCE|
----------|
0:4:17 |
So you can implement as
WITH da AS (
SELECT
NUMTODSINTERVAL(TO_DATE(call_end, 'yyyy-mm-dd hh24:mi:ss') - TO_DATE(call_start, 'yyyy-mm-dd hh24:mi:ss'), 'DAY') AS call_diff
FROM
calls
) SELECT
EXTRACT( DAY FROM call_diff )*24 + EXTRACT( HOUR FROM call_diff )|| ':' || EXTRACT( MINUTE FROM call_diff ) || ':' || EXTRACT( SECOND FROM call_diff ) TIMESTAMP
FROM
da
Please see below response but this will only work within 24hrs and returns both '00'
WITH sample_lt AS(
SELECT '2020-02-20 12:56:03' START_TIME, '2020-02-20 13:00:20' END_TIME FROM dual
)
SELECT start_time,
end_time,
TO_CHAR (TRUNC (SYSDATE) + (to_date(end_time, 'yyyy-mm-dd HH24:MI:SS') -
to_date(start_time, 'yyyy-mm-dd HH24:MI:SS')
) , 'hh24:mi:ss' ) duration
FROM sample_lt ;
I run this query in oracle:
select TO_CHAR(parsed_on, 'DD-MON-YYYY HH24:MI:SS'), TO_CHAR(creation_date, 'DD-MON-YYYY HH24:MI:SS'),
(parsed_on - creation_date) * 1000
from hotels
where TO_CHAR(CREATION_DATE, 'dd/mm/yyyy') = TO_CHAR(SYSDATE, 'dd/mm/yyyy') and incoming_psd_id = 608423671;
and this is the result, where I was excepting 12 seconds instead of 8.4
10-NOV-2017 07:49:54 10-NOV-2017 07:37:46 8,42592592592
The correct expression for seconds in Oracle would be:
(parsed_on - creation_date) * 24*60*60
The difference is decimal days.
If you want minutes, then use:
(parsed_on - creation_date) * 24*60
select ( to_date('10-NOV-2017 07:49:54', 'DD-MON-YYYY HH:MI:SS') - to_date('10-NOV-2017 07:37:46', 'DD-MON-YYYY HH:MI:SS') )* 86400
from dual
Correctly gives 728 seconds.
And if you want to compare dates, you have this option:
where trunc(CREATION_DATE) = trunc(SYSDATE)
TRUNC cuts out the time portion of dates by default, whereas trunc(SYSDATE, YYYY) would cut it down to years only.
select CASE
WHEN (shipment_date >=((to_char(shipment_date, 'dd-Mon-yyyy 08:00')::timestamp)) and shipment_date < ((to_char(shipment_date, 'dd-Mon-yyyy 20:00')::timestamp )))
THEN to_char(shipment_date, 'dd/MM/yyyy 08:00')
WHEN (shipment_date >=((to_char(shipment_date, 'dd-Mon-yyyy 20:00')::timestamp)) and shipment_date < ((to_char(shipment_date + interval '1' day, 'dd-Mon-yyyy 08:00')::timestamp )))
THEN to_char(shipment_date, 'dd/MM/yyyy 20:00')
END as shipment_date_sort,
COUNT(lotnumber)
from shipping.picklist_lots
inner join shipping.picklist on picklist_lots.picklist_id = picklist.picklist_id
where shipment_date >= '12/23/2014' and shipment_date < '12/24/2014'
GROUP BY shipment_date_sort
it doesnt give me the result which the shipment date between 24/12/2014 00:00 and 24/12/2014 08:00
i need it
where shipment_date >= '12/23/2014' and shipment_date < '12/24/2014'
Add a time value. According to your datetime value, 24/12/2014 00:00, it should be in format (dd/MM/yyyy 00:00)
where shipment_date >= '23/12/2014' and shipment_date <= '24/12/2014 23:59'
Based on logical query processing, you cannot use alias of your SELECT statement in group by because SELECT happens after GROUP BY so you need to duplicate your CASE statement into your GROUP BY part.
LOGICAL QUERY PROCESSING