CODEWITHSUNDEEP
oracle-databaseregexp-substr
user747291
user747291

Reputation: 821

Extract date from string oracle

I have a column in which a string starts with - 'Chicago, IL, April 20, 2015 — and so on text here'. I want to extract the Date part from this string in Oracle. Any ideas on how to do this. I was able to find something for mm/dd/yyyy like below, but not for long date format.

SELECT REGEXP_SUBSTR(' the meeting will be on 8/8/2008', '[0-9]{1,}/[0-9]{1,}/[0-9]{2,}') FROM dual

Upvotes: 0

Views: 12804

Answers (3)

Pham X. Bach
Pham X. Bach

Reputation: 5442

If your columns value is always start with 'Chicago, IL, April 20, 2015 — and so on text here' then you could simly use SUBSTR instead of REGEXP_SUBSTR 

SELECT 
    SUBSTR(column_name
        ,INSTR(column_name, ',', 1, 2) + 1
        ,INSTR(column_name, '—') - INSTR(column_name, ',', 1, 2) - 1
    ) 
FROM 
    dual;

If not then you could use REGEXP_SUBSTR as other answer mention, my original answer is wrong as @MTO comment

Upvotes: 1

MT0
MT0

Reputation: 167981

You could use:

SELECT  TO_DATE(
          REGEXP_SUBSTR(
            'Chicago, IL, April 20, 2015 — and so on text here',
            '(JANUARY|FEBRUARY|MARCH|APRIL|MAY|JUNE|JULY|AUGUST|SEPTEMBER|'
              || 'OCTOBER|NOVEMBER|DECEMBER)'
              || '[[:space:]]+([012]?[0-9]|3[01])'
              || '[[:punct:][:space:]]+\d{4}',
            1,
            1,
            'i'
          ),
          'MONTH DD YYYY'
        )
FROM    DUAL;

If you want to validate the dates as well (so you don't get an error for February 29, 2001) then you could use a user-defined function:

CREATE FUNCTION parse_Date(
  in_string     VARCHAR2,
  in_format     VARCHAR2 DEFAULT 'YYYY-MM-DD',
  in_nls_params VARCHAR2 DEFAULT NULL
) RETURN DATE DETERMINISTIC
AS
BEGIN
  RETURN TO_DATE( in_string, in_format, in_nls_params );
EXCEPTION
  WHEN OTHERS THEN
    RETURN NULL;
END;
/

And replace the TO_DATE( ... ) function with PARSE_DATE( ... )

Upvotes: 2

Timekiller
Timekiller

Reputation: 3126

Well, you can take a direct approach and use a regular expression like in the example that you've found:

SELECT 
  REGEXP_SUBSTR('Chicago, IL, April 20, 2015 - etc etc', '(January|February|March|April|May|June|July|August|September|October|November|December) [0-9]{1,2}, [0-9]{4}')
FROM dual;

But this will only work properly if all the dates are in the exact same format. Full month name with first letter uppercased, space, day, comma, space, 4-digit year. If there can be more than one space or no space at all, use \s* instead of spaces in the regular expression. If the month name isn't necessarily initcap, use initcap() on source or case-insensitive flag for regexp_substr function.

Additionally, this will catch bogus dates that fit the format, like "April 99, 1234", you'll have to filter them later.

Upvotes: 1

Related Questions