I was practicing SQL with PostgreSQL and I got stuck using COALESCE, maybe someone can tell me where I am going wrong.
I have a table with four columns: title, height_cm, length_cm, and width_cm. I want to multiply three last ones and get the column "size". The problem is that the whole dataset has many NULL values and 0s which I would like to skip and return the biggest value possible with the three columns (i.e. if only height has value, return that value, if the three columns have values multiply the three). If no column has a value return "NO VALUE FOUND".
I was not able to manage "skipping" the Null values (I haven't really started with the 0s yet since I am stuck in the beginning). I thought that this would have done the trick but it did not work.
SELECT title, height_cm, length_cm, width_cm,
COALESCE(
height_cm * length_cm * width_cm,
length_cm * width_cm,
height_cm * length_cm,
height_cm * width_cm,
height_cm,
length_cm,
width_cm, 'NO VALUE FOUND') AS size
FROM artworks
ORDER BY size ASC
LIMIT 10
First of all, the system does not recognize "NO VALUE FOUND" (but when I add a 0 instead, it recognizes it), secondly, when I take the text out, the system still considers the NULL values. I am able to make it work adding
WHERE height_cm IS NOT NULL AND length_cm IS NOT NULL AND width_cm IS NOT NULL
But I thought that the main idea of COALESCE was to be able to skip the NULL values. Any suggestions?
Regarding the 0s, if I add:
WHERE height_cm != 0 AND length_cm != 0 AND width_cm != 0
I lose the rows that have values but which also have one 0.
Thanks!
EDIT with the solution from the answers
In the end I used a CTE in combination with the answers from the people who helped below, this is the final query:
WITH query AS (SELECT title, height_cm, length_cm, width_cm,
(CASE WHEN height_cm IS NULL AND length_cm IS NULL AND width_cm IS NULL
THEN 0
ELSE (COALESCE(height_cm, 1) * COALESCE(length_cm, 1) * COALESCE(width_cm, 1))
END) AS size
FROM artworks)
SELECT *
FROM query
WHERE size > 0
ORDER BY size ASC
LIMIT 10
The datatypes of all elements passed to coalesce should be the same, but I would apply coalesce to each column separately and use the default value of 1 if it's null:
coalesce(height_cm, 1) * coalesce(length_cm, 1) * coalesce(width_cm, 1) as size
which is logically the same as what you've coded.
You'll need a special check in case they are all null, for which you may (ab)use coalesce too:
case
when coalesce(height_cm, length_cm, width_cm) is null then null -- or 0
else coalesce(height_cm, 1) * coalesce(length_cm, 1) * coalesce(width_cm, 1)
end as size
You'll have to pick either null or some special value (eg 0) when they are all null. You may render it as "NO VALUE FOUND" to the user as you wish.
"NO VALUE FOUND" is a string type on the other hand 0 is numeric that's why when you are using 0 it does not create problem because in case of 0 all the data for this column is numeric data type but when you are trying to use string it has been created issue of mixing two different data types which will not be allowed by sql engine
I agree with Bohemian on the use of a single expression. However, I think the final expression you want returns a string:
(case when height_cm is null and length_cm is null and width_cm is null
then 'NO VALUE FOUND'
else (coalesce(height_cm, 1) * coalesce(length_cm, 1) * coalesce(width_cm, 1))::text
end) as size
Or, just live with a NULL value meaning that the value is not available.
It sounds like the function you actually want to use here is GREATEST:
SELECT
title,
height_cm,
length_cm,
width_cm,
COALESCE(
GREATEST(
height_cm * length_cm * width_cm,
length_cm * width_cm,
height_cm * length_cm,
height_cm * width_cm,
height_cm,
length_cm,
width_cm)::text, 'NO VALUE FOUND') AS size
FROM artworks
ORDER BY size
LIMIT 10;
The GREATEST function will automatically pick up on the largest non NULL value already. So, you only need COALESCE here as a catch-all for the case where the height, length, and width, all happen to be NULL at the same time.
Related
SELECT ID, VOLUME, TYPEOF(VOLUME) FROM DBT.BASE
When I select these columns, I see that the results have some NULL values. They don't seem to be strings. However, when I try to filter the NULL values out with a where statement:
SELECT ID, VOLUME, TYPEOF(VOLUME) FROM DBT.BASE WHERE VOLUME = NULL
I don't see any results. What might be the possible causes? I also tried filtering with 'NULL' but that would throw an error since the column type is double.
use this for only want null recode
SELECT ID, VOLUME, TYPEOF(VOLUME) FROM DBT.BASE WHERE VOLUME IS NULL
or
SELECT ID, VOLUME, TYPEOF(VOLUME) FROM DBT.BASE WHERE ISNULL(VOLUME,'') = ''
if you get not null value then use
SELECT ID, VOLUME, TYPEOF(VOLUME) FROM DBT.BASE WHERE ISNULL(VOLUME,'') <> ''
or
SELECT ID, VOLUME, TYPEOF(VOLUME) FROM DBT.BASE WHERE VOLUME IS NOT NULL
a more full answer is NULL is not directly comparable, much like NaN is not comparable in floating point numbers. Both represent the "lack of a value" if you "have not value here" how can you compare it to something.
"There is nobody next to you, what is their name?" it just doesn't make sense.
So to test you ask column IS NULL or column IS NOT NULL or you can use a compact logic expression see Conditional Expressions but some common ones in Snowflake are:
short form
ANSI long
snowflake long
NVL(column,'')
CASE WHEN column IS NOT NULL THEN column ELSE '' END
IFF(column IS NOT NULL, column, '')
NVL2(column,'a','b')
CASE WHEN column IS NOT NULL THEN 'a' ELSE 'b' END
IFF(column IS NOT NULL, 'a', 'b')
ZEROIFNULL(column)
CASE WHEN column IS NOT NULL THEN column ELSE 0 END
IFF(column IS NOT NULL, column, 0)
COALESCE/NVL/IFNULL are all interchangable so I will only show one (expect COALESCE can handle N items which are checked in order)
You can use the where is function or is not function to filter all the null values.
I want to know if the 4th integer in the ID, is even, or if its odd.
If the 4th number is even (if the number is either 0,2,4,6,8 I want to put the ID into a new column named 'even'
IF the 4th number is odd, the column should have the name 'Odd'
select ID as 'Female'
from Users2
where ID LIKE '%[02468]'
This shows if any of the numbers are even. I want to specify the 4th number
Try this:
select *, OddOrEven = iif(substring(ID,4,1) in ('0','2','4','6','8') , 'Even', 'Odd') from Users2
This will tell you whether the 4th character is Odd or Even.
This is of course assuming that the 4th character of ID column will be numeric.
To make it permanently part of the table, you can add a computed column as shown below.
alter table Users2
add OddOrEven as iif(substring(ID,4,1) in ('0','2','4','6','8'), 'Even', 'Odd')
Substring the character you are interested in
Convert to an int
Check whether modulus 2 returns 0 (i.e. even).
select id
, case when convert(int,substring(id, 4, 1)) % 2 = 0 then 'Even' else 'Odd' end
from Users;
Example:
select id
, case when convert(int,substring(id, 4, 1)) % 2 = 0 then 'Even' else 'Odd' end
from (values ('4545-4400'), ('4546-4400')) X (id);
Returns
id
4545-4400
Odd
4546-4400
Even
Thats assuming there is always a 4th character. If not you would need to check for it.
You were close, but only need to check a single character against a set of characters:
where Substring( Id, 4, 1 ) like '[02468]'
Note that there is no wildcard (%) in the pattern.
It can be used in an expression like:
case when Substring( Id, 4, 1 ) like '[02468]' then 'Even' else 'Odd' end as Oddity
I need to filter out rows in table where numer_lini column has number in it and it is between 100 and 999, below code works just fine when i comment out line where i cast marsnr to integer. However when i try to use it i get error: Invalid character found in a character string argument of the function "INTEGER". when looking at the list seems like replace and translate filters only numbers just fine and select only contains legit numbers (list of unique values is not long so its easy to scan by eye). So why does it fail to cast something? I also tried using integer(marsnr), but it produces the same error. I need casting because i need numeric range, otherwise i get results like 7,80 and so on. As I mentioned Im using IBM DB2 database.
select numer_lini, war_trasy, id_prz1, id_prz2
from alaska.trasa
where numer_lini in (
select marsnr
from (
select
distinct numer_lini marsnr
from alaska.trasa
where case
when replace(translate(numer_lini, '0','123456789','0'),'0','') = ''
then numer_lini
else 'no'
end <> 'no'
)
where cast(marsnr as integer) between 100 and 999
)
fetch first 300 rows only
If you look at the optimized SQL from the Db2 explain, you will see that Db2 has collapsed your code into a single select.
SELECT DISTINCT Q2.NUMER_LINI AS "NUMER_LINI",
Q2.WAR_TRASY AS "WAR_TRASY",
Q2.ID_PRZ1 AS "ID_PRZ1",
Q2.ID_PRZ2 AS "ID_PRZ2",
Q1.NUMER_LINI
FROM ALASKA.TRASA AS Q1,
ALASKA.TRASA AS Q2
WHERE (Q2.NUMER_LINI = Q1.NUMER_LINI)
AND (100 <= INTEGER(Q1.NUMER_LINI))
AND (INTEGER(Q1.NUMER_LINI) <= 999)
AND (CASE WHEN (REPLACE(TRANSLATE(Q1.NUMER_LINI,
'0',
'123456789',
'0'),
'0',
'') = '') THEN Q1.NUMER_LINI
ELSE 'no' END <> 'no')
Use a CASE to force Db2 to do the "is integer" check first. Also, you don't check for the empty string.
E.g. with this table and data
create TABLE alaska.trasa (numer_lini VARCHAR(10), war_trasy INT , id_prz1 INT, id_prz2 INT);
insert into alaska.trasa values ('',1,1,1),('99',1,1,1),('500',1,1,1),('3000',1,1,1),('00300',1,1,1),('AXS',1,1,1);
This SQL works
select numer_lini, war_trasy, id_prz1, id_prz2
from alaska.trasa
where case when translate(numer_lini, '','0123456789') = ''
and numer_lini <> ''
then integer(numer_lini) else 0 end
between 100 and 999
Although that does fail if there is an embedded space in the input. E.g. '30 0'. To cater for that, a regular expressing is probably preferred. E.g.
select numer_lini, war_trasy, id_prz1, id_prz2
from alaska.trasa
where case when regexp_like(numer_lini, '^\s*[+-]?\s*((\d+\.?\d*)|(\d*\.?\d+))\s*$'))
then integer(numer_lini) else 0 end
between 100 and 999
I have a table with about a hundred rows. It has a column is_gallery that contains either 1, 0, or NULL. If I do...
SELECT * WHERE is_gallery != 1
or
SELECT * WHERE NOT (is_gallery = 1)
it excludes the rows where is_gallery is null. I can manage to get a proper response if I do
SELECT * WHERE (is_gallery = 0 OR is_gallery is null)
But shouldn't the "!=" or NOT work? Isn't there a way to just return the rows where is_gallery doesn't equal 1 without testing for every other possibility?
You can use the IS and IS NOT operators instead of = and !=. These treat NULL like a normal value.
SELECT * FROM yourTable WHERE is_gallery IS NOT 1
The best thing to use is coalesce as in:
SELECT *
WHERE coalesce(is_gallery,0) != 1;
what coalesce does, is replaces any null value in that column with the second parameter. In the example above, any nulls in the "is_gallery" column will be replaced with 0 before it is compared with 1. So will of course return true.
On NULL realize that a NULL value isn't equal to ANYTHING - not even NULL itself. It cannot be compared - so when "comparing", it always will return FALSE. On NULL, it has a special operator which is "IS NULL" or "IS NOT NULL"
I have this query that i get errors when data gets loaded to this table and the divisor is zero which is the b. living_units column. I want to know if I can do some sort of try catch or if its zero to show null instead of failing or something that will help not error out?
SELECT a.SERVICE_TYPE_GRP,
a.HSIA_TYPE,
a.STAT_DYN_IND,
a.VIDEO_IND,
a.VOICE_IND,
a.CUST_CNT,
b.LIVING_UNIT_CNT,
a.DT_MODIFIED,
a.CUST_CNT / b.LIVING_UNIT_CNT AS ALLRGN_TK_RT_PCT
FROM ( SELECT lp.SERVICE_TYPE_GRP SERVICE_TYPE_GRP,
lp.HSIA_TYPE HSIA_TYPE,
lp.STAT_DYN_IND STAT_DYN_IND,
lp.VIDEO_IND VIDEO_IND,
lp.VOICE_IND VOICE_IND,
lp.DT_MODIFIED DT_MODIFIED,
SUM (lp.CUST_CNT) CUST_CNT
FROM RPT_SUBSCR_REGION_DTL lp
GROUP BY SERVICE_TYPE_GRP,
HSIA_TYPE,
STAT_DYN_IND,
VIDEO_IND,
VOICE_IND,
DT_MODIFIED) a,
( SELECT DT_MODIFIED, SUM (LIVING_UNIT_CNT) LIVING_UNIT_CNT
FROM RPT_REGION_CUST_DTL
WHERE dt_modified = (SELECT dt_modified
FROM ls_dt_modified
WHERE NAME = 'RPT_REGION_CUST_DTL')
GROUP BY DT_MODIFIED) b
WHERE a.DT_MODIFIED = b.DT_MODIFIED;
The error i get is ORA-01476: divisor is equal to zero.
If you have a complex divisor expression and a NULL result is acceptable, there is an alternative that does not require repeating the divisor expression:
a.CUST_CNT / nullif(b.LIVING_UNIT_CNT, 0)
The NULLIF function returns NULL if the first parameter is equal to the second parameter. In this case it returns NULL if b.LIVING_UNIT_CNT is equal to zero. And anything divided by NULL also becomes NULL. It's a bit of a "trick" maybe, but it saves the expression repetition of the CASE statement.
Try a case statement:
case
when b.LIVING_UNIT_CNT = 0 -- the divisor
then 0 -- a default value
else a.CUST_CNT / b.LIVING_UNIT_CNT
end
ALLRGN_TK_RT_PCT