sql query to get only friday dates between two dates - sql

Question

Question: How do i select only friday dates between two dates in oracle
SELECT dates,TO_CHAR(dates,'day-mon-yyyy')
FROM
(SELECT to_date('01-jan-12','dd-mon-yy')+rownum -1 AS dates
FROM addresses
WHERE rownum <= to_date('31-jan-12','dd-mon-yy')- to_date('01-jan-12','dd-mon-yy')+1)
WHERE upper(regexp_substr(TO_CHAR(dates,'day-mon-yy'),'([[:alpha:]])+'))=upper('FRIDAY');
i need Output like:
06-JAN-12 FRIDAY
13-JAN-12 FRIDAY
20-JAN-12 FRIDAY
27-JAN-12 FRIDAY
03-FEB-12 FRIDAY
10-FEB-12 FRIDAY
17-FEB-12 FRIDAY
24-FEB-12 FRIDAY
02-MAR-12 FRIDAY
09-MAR-12 FRIDAY
16-MAR-12 FRIDAY
23-MAR-12 FRIDAY
30-MAR-12 FRIDAY
visit more sql queries: SQL Query Interview Questions

Answer 1

You will get an error from what you posted because the values you've used as the first argument to to_char() aren't enclosed in single quotes:
select to_date(01-jan-12,'dd-mon-yy') from dual;
ORA-00904: "JAN": invalid identifier
Because there are no quotes, jan is interpreted as an identifier, and there is (presumably) no column call JAN in your addresses table. It's also bad practice to use 2-digit years, and where you must (from really old data) it's usually better to use RR than YY. The month names are also affected by your NLS settings, so it's safer to use month numbers than names; if you really want names the to_char() function has a third argument to control the language.
You are doing this in quite a complicated way, and you're relying on the addresses table having enough rows. Specifying that you want the day name in lower case via day instead of DAY, then making it uppercase, then stripping off the bits of the string - that you specified in the first place! - to only get the day name, and then comparing that on the assumption (again) that the NLS settings will give you English day names anyway is... unnecessarily convoluted. As is calling upper() against a fixed string literal you you can (and are) supply in uppercase already.
Instead of
WHERE upper(regexp_substr(TO_CHAR(dates,'day-mon-yy'),'([[:alpha:]])+'))=upper('FRIDAY');
you could do any of these, or other variations:
WHERE regexp_substr(TO_CHAR(dates,'DAY-mon-
y'),'([[:alpha:]])+')=upper('FRIDAY');
WHERE TO_CHAR(dates,'DAY')='FRIDAY ';
WHERE TRIM(TO_CHAR(dates,'DAY'))=upper('FRIDAY');
WHERE TO_CHAR(dates,'FMDAY','NLS_DATE_LANGUAGE=ENGLISH')='FRIDAY';
You could avoid relying on the addresses table by using a hierarchical query, against the dual table:
SELECT next_day(date '2012-01-01' - 1, 'FRIDAY') + (7 * (level - 1))
FROM dual
CONNECT BY next_day(date '2012-01-01' - 1, 'FRIDAY') + (7 * (level - 1))
<= date '2012-03-31';
Using next_day also relies on NLS settings though, so unless you can always control the session date language, it might be safer (if a little less efficient) to get all the dates and then filter them in an NLS-independent way:
SELECT dates, to_char(dates, 'FMDAY')
FROM (
SELECT date '2012-01-01' + level - 1 AS dates
FROM dual
CONNECT BY level <= date '2012-03-31' - date '2012-01-01'
)
WHERE to_char(dates, 'FMDAY', 'NLS_DATE_LANGUAGE=ENGLISH') = 'FRIDAY';
DATES TO_CHAR(DATES,'FMDAY')
--------- ------------------------------------
06-JAN-12 FRIDAY
13-JAN-12 FRIDAY
20-JAN-12 FRIDAY
27-JAN-12 FRIDAY
03-FEB-12 FRIDAY
10-FEB-12 FRIDAY
17-FEB-12 FRIDAY
24-FEB-12 FRIDAY
02-MAR-12 FRIDAY
09-MAR-12 FRIDAY
16-MAR-12 FRIDAY
23-MAR-12 FRIDAY
30-MAR-12 FRIDAY
13 rows selected.
As #mathguy pointed out in a comment, although next_day() is NLS-sensitive, you can use an expression for the second argument, so instead of hard-coding the day name you can do this:
next_day(date '2012-01-01' - 1, to_char(date '1999-12-31', 'FMDAY'))
where 1999-12-31 can be any date known to be a Friday; and if you don't mind the expression in the select list and connect-by clause being different (and really, you - and I - shouldn't!) you can reduce the computational cost of that check with:
SELECT dates, to_char(dates, 'FMDAY', 'NLS_DATE_LANGUAGE=ENGLISH')
FROM (
SELECT next_day(date '2012-01-01' - 1,
to_char(date '1999-12-31', 'FMDAY')) + (7 * (level - 1)) AS dates
FROM dual
CONNECT BY level <= 1 + (date '2012-03-31' - next_day(date '2012-01-01' - 1,
to_char(date '1999-12-31', 'FMDAY')))/7
);
which gets the same 13 rows as above, regardless of the session's date language. If you want the output to be in the session language too, just remove the overriding third argument to to_char().