1

I'm using Postgres 9.5. I have a table with a column that records URLs. Sometimes the URL has a query string and sometimes it does not. I would like to extract the URL, minus any query string, so I came up with:

select substring(url, 0, case position('?' in url) when 0 then length(url)+1 else position('?' in url) end) 
from article;

This seems a little wordy and I was wondering if there's a more concise way to do this. My table column is of type TEXT.

1
  • Based on stackoverflow.com/questions/41633436/… you may try using a specific uri type instead of text, or look at FTS. Separately, if you need often to do that, why not storing the URL already without the query string, or at least creating another field to store it, with proper triggers for update. Other solution: create a VIEW on your table that hardcodes one for all the above manipulation and then just query this VIEW giving you immediately the URL without the query string. Mar 15, 2019 at 21:21

3 Answers 3

5

You can use regexp_replace() to replace everything after the first ? with nothing:

select regexp_replace(url, '\?.*$', '')

The following example:

with data (url) as (
   values 
    ('http://foo.bar/some/page'),
    ('http://foo.bar/some/page?id=42&name=Marvin')
)
select url, regexp_replace(url, '\?.*$', '') as clean_url
from data;

returns:

url                                        | clean_url               
-------------------------------------------+-------------------------
http://foo.bar/some/page                   | http://foo.bar/some/page
http://foo.bar/some/page?id=42&name=Marvin | http://foo.bar/some/page
3

I suggest split_part() for this, certainly more concise and simpler and substantially cheaper than regular expressions, too:

SELECT split_part(url, '?', 1) AS stripped_url FROM tbl;

Does exactly what you want. If there is no '?', the whole string is the result.

db<>fiddle here

Related:

1

Or, you could do it using a simple CASE statement.

I'm not sure of the performance implications, but I have a feeling that CASE should be more performant - will try and test when I get home - bank holiday today!

Check the fiddle here:

WITH my_data (url) AS 
(
   VALUES 
    ('http://foo.bar/some/page'),
    ('http://foo.bar/some/page?id=42&name=Marvin')
)
SELECT
  CASE
    WHEN POSITION('?' IN url) > 0 THEN
      LEFT(url, POSITION('?' IN url) - 1)
    ELSE
      url
  END AS my_result
FROM my_data;

Result:

my_result
----------------------------
http://foo.bar/some/page
http://foo.bar/some/page

This solution relies on no other assumption than that the requirement that the deletion takes place from the first question mark till the end, which is same assumption as the other solution makes.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.