In this project I'm currently working on (I'm building a bridge between a desktop app and a new e-shop) there is a products table that has some spare columns defined that can be used for whatever reason the end user might need some custom data to be stored into.
So, the user needed to set a true/false flag to determine whether the products would appear in three different sliders... Unfortunately, the person who implemented this, didn't even use the same type of spare columns... So,
Slider1's flag is stored in a varchar(50) column
Slider2's flag is stored in a float column
Slider3's flag is stored in a float column
Additionally I ran a SELECT DISTINCT <column> for each one of them to get an idea of the actual data stored in each column and got the following results:
The varchar column has the following data stored in it:
FLDSTRING1
NULL
''
0
1
194276400456
The float column has the following data stored:
FLDFLOAT5
NULL
0
1
And the other float column has this:
FLDFLOAT6
NULL
1
Also, I ran the following query to find the different combinations of the data stored for each column:
SELECT FLDSTRING1, FLDFLOAT5, FLDFLOAT6
FROM MATERIAL
GROUP BY FLDSTRING1, FLDFLOAT5, FLDFLOAT6
and got the following combinations...
FLDSTRING1
FLDFLOAT5
FLDFLOAT6
NULL
NULL
NULL
NULL
NULL
1
NULL
0
NULL
NULL
1
NULL
NULL
1
1
''
NULL
NULL
''
NULL
1
0
NULL
NULL
0
0
NULL
1
NULL
NULL
1
NULL
1
1
0
NULL
1
1
NULL
1
1
1
194276400456
0
NULL
What I need after all this introduction...
I want a concatenated string of three comma-separated values like this
NEWPROD for when FLDSTRING1 would evaluate to true - anything not NULL, 0, or ''
CUSTOM1 for when FLDFLOAT5 would evaluate to true - basically the value 1
CUSTOM2 for when FLDFLOAT6 would evaluate to true - again value 1
After some trial and error I managed to bring this to a point that it kind of works, in the sense that it brings the correct values, not comma-separated though...
SELECT
FLDSTRING1, FLDFLOAT5, FLDFLOAT6,
CONCAT(CASE WHEN ISNULL(FLDSTRING1, '') = '' THEN '' ELSE 'NEWPROD' END,
CASE WHEN ISNULL(FLDFLOAT5, '') = '' THEN '' ELSE 'CUSTOM1' END,
CASE WHEN ISNULL(FLDFLOAT6, '') = '' THEN '' ELSE 'CUSTOM2' END) AS TAGS
FROM
MATERIAL
GROUP BY
FLDSTRING1, FLDFLOAT5, FLDFLOAT6;
FLDSTRING1
FLDFLOAT5
FLDFLOAT6
TAGS
NULL
NULL
NULL
NULL
NULL
1
CUSTOM2
NULL
0
NULL
NULL
1
NULL
CUSTOM1
NULL
1
1
CUSTOM1CUSTOM2
''
NULL
NULL
''
NULL
1
CUSTOM2
0
NULL
NULL
NEWPROD
0
0
NULL
NEWPROD
1
NULL
NULL
NEWPROD
1
NULL
1
NEWPRODCUSTOM2
1
0
NULL
NEWPROD
1
1
NULL
NEWPRODCUSTOM1
1
1
1
NEWPRODCUSTOM1CUSTOM2
194276400456
0
NULL
NEWPROD
Problem #1 is I don't quite understand how this works... I mean, value 0 isn't '', but still for the combination of NULL 0 NULL I get an empty value, which is what I wanted... But how does it do that?
And also, can someone update my final query to comma-separate the calculated TAGS column? Problem #2 is that I don't want it to contain just two commas, like ,,, when the combination wouldn't justify any of the three values to appear... It should work like PHP's implode() works...
To help you help me with this, I'm including a fiddle with the setup of the scenario I describe here... Thanks in advance!
Since you are using SQL Server 2014, instead of CONCAT_WS you may try STUFF as shown below. By prepending the delimiter , before all strings ,the STUFF will remove the first comma found.
SELECT
FLDSTRING1,
FLDFLOAT5,
FLDFLOAT6,
STUFF(
CONCAT(
CASE WHEN FLDSTRING1 IS NULL OR FLDSTRING1 IN ('0','') THEN '' THEN '' ELSE ',NEWPROD' END,
CASE WHEN FLDFLOAT5 IS NULL THEN '' ELSE ',CUSTOM1' END,
CASE WHEN FLDFLOAT6 IS NULL THEN '' ELSE ',CUSTOM2' END
),
1,1,''
) AS TAGS
FROM #MATERIAL
GROUP BY FLDSTRING1, FLDFLOAT5, FLDFLOAT6;
View working demo db fiddle
Let me know if this works for you.
Use the CONCAT_WS() function to concat values into a comma (or other separator) separated list, which ignores nulls.
To use CONCAT_WS(), you want to pass it a true NULL if the value is "blank" (by your definition), otherwise your custom label:
SELECT DISTINCT
FLDSTRING1,
FLDFLOAT5,
FLDFLOAT6,
CONCAT_WS(',',
CASE WHEN FLDSTRING1 IS NULL OR FLDSTRING1 = '' OR FLDSTRING1 = '0' THEN NULL ELSE 'NEWPROD' END,
CASE WHEN FLDFLOAT5 IS NULL OR FLDFLOAT5 = 0 THEN NULL ELSE 'CUSTOM1' END,
CASE WHEN FLDFLOAT6 IS NULL OR FLDFLOAT6 = 0 THEN NULL ELSE 'CUSTOM2' END) AS TAGS
FROM MATERIAL
Replaced GROUP BY with DISTINCT because it's simpler and (here) achieves the same thing.
If CONCAT_WS is not available:
SELECT DISTINCT
FLDSTRING1,
FLDFLOAT5,
FLDFLOAT6,
REPLACE(REPLACE(REPLACE(CONCAT(
CASE WHEN FLDSTRING1 IS NULL OR FLDSTRING1 = '' OR FLDSTRING1 = '0' THEN 'X' ELSE 'NEWPROD' END,
',',
CASE WHEN FLDFLOAT5 IS NULL OR FLDFLOAT5 = 0 THEN 'X' ELSE 'CUSTOM1' END,
',',
CASE WHEN FLDFLOAT6 IS NULL OR FLDFLOAT6 = 0 THEN 'X' ELSE 'CUSTOM2' END
), ',X', ''), 'X,', ''), 'X', '') AS TAGS
FROM MATERIAL
See dbfiddle.
Related
Forced update of first letters if present else mark it as null when empty
IF LENGTH([Column])>0 THEN UPDATE [Table] SET Position = Upper([Column]) ELSE RETURN NULL END IF;
Col A
abcdef
defghi
efgijg
''
elllef
''
ijkmnk
Expected Result (first two as capital and marked as null if row value is empty)
Col A
ABcdef
DEfghi
EFgijg
NULL
ELllef
NULL
IJkmnk
You can do this with a CASE expression in the SET part
UPDATE the_table
SET column_a = CASE
WHEN column_a = '' THEN NULL
ELSE upper(left(column_a,2))||substr(column_a,3)
END;
If you want to treat a value with only spaces as "empty" as well, you can use WHEN trim(column_a) = '' THEN NULL instead.
I'm writing an sql constraints about values on columns based on other values in other columns for Example:
if Column Replacement Part = 'N/A', the Replacement Company must be = 'N/A' and verse vice .
if type ='obs' the LTB_Date and LTS_Date Must be Not Null and
if type !='obs' the LTB_Date and LTS_Date Must be Null
Try to use a CHECK constraint like this:
ALTER TABLE PCN_Table ADD CONSTRAINT CHECK(
CASE
WHEN Replacement_Part = 'N/A' THEN CASE WHEN Replacement_Company = 'N/A' THEN 1 ELSE 0 END
ELSE 1
END=1
AND
CASE
WHEN [type] = 'obs' AND LTB_Date IS NOT NULL AND LTS_Date IS NOT NULL THEN 1
WHEN [type] <> 'obs' AND LTB_Date IS NULL AND LTS_Date IS NULL THEN 1
ELSE 0
END=1
)
I have a query where i used case when to KNOWN_AS column stating when its null then 'null' else 'not null' end as known_as2. Now in where clause i want to bring rows which only contain 'Not null' .
SELECT i.individual_ref, 0, 'KNOWNAS', i.FORENAMES, i.KNOWN_AS,
case when KNOWN_AS is null then 'Null' else ' Not null' end as known_as
FROM TestDatabase.dbo.INDIVIDUAL I
JOIN TestDatabase.dbo.MEMBER M ON M.INDIVIDUAL_REF=I.INDIVIDUAL_REF
WHERE m.member_status IN(33,1316)
AND i.KNOWN_AS IS null or i.KNOWN_AS=''
and m.MEMBER_STATUS in (33,1316)
and LEN(i.FORENAMES) > '1' and i.FORENAMES !=''
AND i.FORENAMES IS NOT NULL
Reason i'm want help is :-
I have a table which contain Forename, surname and known_as field.
I want to get members who's known_as field is blank/null and forename is not null or blank and forename length is >1 . How can i achieve it. the member status is from another table call member where i want member who are in active and pending status hence i said WHERE m.member_status IN (33,1316). Any solution please.
Finally I have solved it using
SELECT i.individual_ref,0,'KNOWNAS',
case when KNOWN_AS is null then 'Null' else ' Not null' end as knownas2
FROM TestDatabase.dbo.INDIVIDUAL I
JOIN TestDatabase.dbo.MEMBER M ON M.INDIVIDUAL_REF=I.INDIVIDUAL_REF
WHERE m.member_status IN(33,1316)
and len(i.forenames)>2 and
(IsNull(i.forenames, '') <> '') and (i.known_as is null or i.known_as='')
I had to take len(i.forenames, '')>2 instead of 1 because some members also has forename by mistakenly updated as Mr.
Now in where clause i want to bring rows which only contain 'Not null' .
Just modify your WHERE clause with IS NOT NULL as :
SELECT i.individual_ref, 0, 'KNOWNAS', i.FORENAMES, i.KNOWN_AS
FROM DiTestDatabase.dbo.INDIVIDUAL I JOIN
DiTestDatabase.dbo.MEMBER M
ON M.INDIVIDUAL_REF = I.INDIVIDUAL_REF
WHERE m.member_status IN (33,1316) AND
LEN(i.FORENAMES) > 1 AND i.FORENAMES != '' AND
i.FORENAMES IS NOT NULL AND i.KNOWN_AS IS NOT NULL;
Note :
LEN() will return INT type. So, you don't need to use ''.
Might be below query will help you.
SELECT i.individual_ref, 0, 'KNOWNAS', i.FORENAMES, i.KNOWN_AS
,CASE WHEN i.KNOWN_AS IS NULL THEN 'null'
ELSE 'Not null' END AS KNOWN_AS2
FROM DiTestDatabase.dbo.INDIVIDUAL I JOIN
DiTestDatabase.dbo.MEMBER M
ON M.INDIVIDUAL_REF = I.INDIVIDUAL_REF
WHERE m.member_status IN (33,1316) AND
LEN(i.FORENAMES) > 1 AND i.FORENAMES != '' AND
i.FORENAMES IS NOT NULL AND i.KNOWN_AS IS NOT NULL;
this way we can get solution
SELECT * FROM (
SELECT i.individual_ref, 0, 'KNOWNAS', i.FORENAMES, i.KNOWN_AS,
case when KNOWN_AS is null then 'Null' else 'Not null' end as known_as
FROM DiTestDatabase.dbo.INDIVIDUAL I
JOIN DiTestDatabase.dbo.MEMBER M ON M.INDIVIDUAL_REF=I.INDIVIDUAL_REF
WHERE m.member_status IN(33,1316)
AND i.KNOWN_AS IS null or i.KNOWN_AS=''
and m.MEMBER_STATUS in (33,1316)
and LEN(i.FORENAMES) > 1 and i.FORENAMES !=''
AND i.FORENAMES IS NOT NULL) t
WHERE t.known_as='Not null'
I want to search for non null values from the 'currentsheet' which works fine but some fields are actually blank rather than null. How can I find blank fields using postgreSQL as the below has not worked and still displays blank values under the 'currentsheet' field.
SELECT *
FROM PUBLIC._http_requests
WHERE (_http_requests.currentsheet IS NOT NULL OR _http_requests.currentsheet <> '')
AND _http_requests.session_id IS NOT NULL
AND _http_requests.http_referer IS NOT NULL
You need to use AND to check _http_requests.currentsheet. If it was NULL, then it would always be true for the <> '' check and vice versa.
As a way simpler example, you can use select statements without a table to help debug this sort of thing (from psql or whatever SQL query tool you like):
select ('' is not null or '' <> '') as empty_result,
(null is not null or null <> '') as null_result;
empty_result | null_result
--------------+-------------
t |
If the string is '', you get true. If the string is null, you get null (this is because comparisons with null are SQL oddities -- select null = null; results in null). Let's see what happens when we replace or with and:
select ('' is not null and '' <> '') as empty_result,
(null is not null and null <> '') as null_result;
empty_result | null_result
--------------+-------------
f | f
Neat! With X is not null and X <> '', we get false when X is either '' or null.
So the way to phrase the select statement to do what you actually want is:
SELECT *
FROM PUBLIC._http_requests
WHERE _http_requests.currentsheet IS NOT NULL
AND _http_requests.currentsheet <> ''
AND _http_requests.session_id IS NOT NULL
AND _http_requests.http_referer IS NOT NULL;
I think you just need AND _http_requests.currentsheet <> ''. The AND is important there so that we exclude both.
I have a case inside my stored procedure which I use before executing the data.
DECLARE #Setup nvarchar(50)
SELECT
#ZipCode = CASE
WHEN REPLACE(#ZipCode, '0', '') = ''
THEN NULL ELSE #ZipCode
END,
#ZipCodeEND = CASE
WHEN REPLACE(#ZipCodeEND, '0', '') = ''
THEN NULL ELSE #ZipCodeEND
END,
SELECT
#Setup = CASE WHEN (LEN(ISNULL(#ZipCode, ''))) > 0 THEN '1' ELSE '0' END +
CASE WHEN (LEN(ISNULL(#ZipCodeEND,''))) > 0 THEN '1' ELSE '0' END
IF ISNULL(#ID, 0) = 0
BEGIN
INSERT INTO dbo.MapToStaticValuesTable(ZipCode, ZipCodeEND, Setup)
VALUES(#ZipCode, #ZipCodeEND, #Setup)
END
The problem here is even if zipcode and zipcodeEnd are empty and set to null after being saved into the table I keep getting the value "11" instead of getting "00".
Now if I do the same example with nvarchar values it would work, but since ZipCode and ZipCodeEnd are set to int it's acting weird.
It is acting weird because you are using string functions on integers. Not sure what you are trying to achieve with your code but I'm sure it can be done just by checking the values as integers.
I guess this could be what you are looking for.
select case when nullif(#ZipCode, 0) is null then '0' else '1' end +
case when nullif(#ZipCodeEND, 0) is null then '0' else '1' end
One example of weird
select isNull(#ZipCode, '')
return 0 if #ZipCode is null.