sql 删除包含“test”的行但保留包含“contest”的行

Question

I want to remove all campaign containing the word 'test' in their campaign names but keep the campaigns that have 'contest' in their names. Some campaigns don't have either test or contest in their names and i'd like to keep them as well. How can I achieve this in sql in Snowflake environment? Since upper/lower case can be completely random I'm thinking of using NAME (ilike any('%_test_%', '%_test%', 'test_%', 'test')) however when I do this 'contest' is also removed which is not the result I want.

sample data:

|                    Name                      |
|AuroraEnterprisesInc_AuroraSepTestCampaign    |
|_TEST                                         |
|test                                          |
|001_test_XP_GutterDoctor_June-OctoberCampaign|
|FR_TEST_                                      |
|002_XP_HunterKnifton_21ElectionContestCampaign|
|001_XP_Kijiji_BuyandSell                      |

the result should be

|                    Name                      |
|002_XP_HunterKnifton_21ElectionContestCampaign|
|001_XP_Kijiji_BuyandSell                      |

Answer 1

You can use a regular expression in your where clause to filter these rows:

select * from T1 where
regexp_instr(NAME, '\\btest\\b|_test_|\\btest_|_test\\b', 1, 1, 0, 'i') = 0;

NAME
002_XP_HunterKnifton_21ElectionContestCampaign
001_XP_Kijiji_BuyandSell

Answer 2

If I understand correctly you want to exclude all records where the NAME column contains the substring "test" unless that substring is specifically part of the larger string "contest".

In other words, you're not trying to only exclude the stand-alone word "test", but any record that contains "test" not preceded by "con". Eg, you want to exclude SeptTestCampaign and testament but retain SeptContestCampaign or contestament .

Assuming that's correct, a WHERE clause such as:

 SELECT ...
 WHERE ((name ILIKE '%contest%') OR (NOT (name ILIKE '%test%')))

should suffice. (I've added parens just to make the logical grouping unambiguous but most or maybe all of those are technically optional.)

The first criterion ( name ILIKE '%contest%' ) yields the set of records that expressly include the substring "contest".

The second ( NOT name ILIKE '%test%' ) yields the set of all records that do not contain the substring "test" at all.

The OR operator yields the union of those two sets.

Or if you want to delete the test records rather than just filtering them out of a result set, reverse the conditions:

 DELETE FROM my_table 
 WHERE (name ILIKE '%test%') AND (NOT (name ILIKE '%contest%'))

One small caveat: if a record contains both contest and (independently) test , the logic above will include that record in the final results. Eg, a record with the NAME Test_ElectionContestCampaign will be included in the select (and will not be removed by the delete). Based on your context I assume that's unlikely enough to be OK, but if you literally want to exclude all records that contain any instance of the substring test without a leading con please clarify.