使用正则表达式将多个值从列转换为行

Question

I am new to the world of Regex and trying to extract a pattern from a free form text column of a table.

There are two things that I am trying to achieve 1.Extract multiple occurrences of a pattern. The pattern that I am trying to extract is of a URL which is one that starts with http or https. 2. After finding multiple occurrences of the URL I would have to explode them to multiple rows.

Input Table

ip_table (user_id, notes)

(123, 'Here are notes - he owns url https://123.com/asda/32/1221 and http://www.facebook.com/page1 so on')
(234, 'this one has http://www.instagram.com/page3/12321213 (https://example.com/1233/qwerty)

Output Table

op_table(user_id, urls)

(123, 'https://123.com/asda/32/1221')
(123, 'http://www.facebook.com/page1')
(234, 'http://www.instagram.com/page3/12321213')
(234, 'https://example.com/1233/qwerty')

Here is what I have so far for the regexp, with no success.

select user_id, regexp_substr(notes, '(https?)://.*[\s]')

Can you please help give me some direction on how I could find repeated patterns of url patterns? The only thing I would need to check is if there is something that starts with http|https and capture that pattern(s) and repear it multiple times in the notes column. Once I find that string I would have to explode that to multiple rows with the matching user ids.

Answer 1

Try this:

WITH
input(id,str) AS (
          SELECT 123, 'Here are notes - he owns url https://123.com/asda/32/1221 and http://www.facebook.com/page1 so on'
UNION ALL SELECT 234, 'this one has http://www.instagram.com/page3/12321213 (https://example.com/1233/qwerty)'
)

-- create a series of 4 integers in Vertica 
-- keep this code-snippet as you'll need it often
-- without Vertica, UNION SELECT 1, 2, 3 and 4
,l(l) AS (
          SELECT TIMESTAMPADD(us,  1  , DATE '2000-01-01' )
UNION ALL SELECT TIMESTAMPADD(us,  4  , DATE '2000-01-01' )
)
,i(i) AS (
  SELECT
    MICROSECOND(ts)
  FROM l
  TIMESERIES ts AS '1 us' OVER(ORDER BY l)
)
-- end create series of 4 integers: CTE i, column i.

-- verticalise the series of tokens found - one row per URL
-- CROSS JOIN input with series of integers to extract the i-th occurrence
SELECT
  id
, i
, regexp_substr(str,'https?://[^\s):"'']+',1, i ) AS url_found
FROM input CROSS JOIN i
WHERE regexp_substr(str,'https?://[^\s):"'']+',1, i ) IS NOT NULL
ORDER BY 1,2;
-- out  id  | i |                url_found                
-- out -----+---+-----------------------------------------
-- out  123 | 1 | https://123.com/asda/32/1221
-- out  123 | 2 | http://www.facebook.com/page1
-- out  234 | 1 | http://www.instagram.com/page3/12321213
-- out  234 | 2 | https://example.com/1233/qwerty
-- out (4 rows)
-- out 
-- out Time: First fetch (4 rows): 12.555 ms. All rows formatted: 12.599 ms