extract dates and times from string in Redshift - amazon-redshift

Question

I have a column like the one below. The last two sets of numbers are date and time. I want to create date-time column by extracting values from the column.
1002206391240385-sponsoredProducts-SameDayPull-20190627-012313.json
The started by extracting the date but it does not give what I need
Select regexp_substr('1002206391240385-sponsoredProducts-SameDayPull-20190627-012313.json','-[\\d{8}]-')

Answer 1

This substring extracts the date time part from your string.
SELECT substring(col_name,regexp_instr(col_name,'-',1,regexp_count(col_name,'-')-1)+1,
regexp_instr(col_name,'.json',1)-regexp_instr(col_name,'-',1,regexp_count(col_name,'-')-1)-1)
regexp_count counts have many hyphens in the string
regexp_instr gives the position of the hyphen
substring returns starting from second to last hyphen till .json in the string
To test I have used
WITH test(col_name) AS (
SELECT '1002206391240385-sponsoredProducts-SameDayPull-20190627-012313.json'::TEXT
)
SELECT col_name,
substring(col_name,regexp_instr(col_name,'-',1,regexp_count(col_name,'-')-1)+1,
regexp_instr(col_name,'.json',1)-regexp_instr(col_name,'-',1,regexp_count(col_name,'-')-1)-1) datetime
FROM test
Output is
col_name datetime
1002206391240385-sponsoredProducts-SameDayPull-20190627-012313.json 20190627-012313

Answer 2

As an alternative, and if the filename format is consistent, you could use a non-regex solution e.g. extract the part of the filename string that contains the date and then use TO_TIMESTAMP with a format string to extract the date and time:
SELECT TO_TIMESTAMP(RIGHT('1002206391240385-sponsoredProducts-SameDayPull-20190627-012313.json', 20), 'YYYYMMDD-HH24MISS.json') AS extracted_datetime
which returns
extracted_datetime |
----------------------|
2019-06-27 01:23:13+00|