How does one use WITH within a Scalar Function? - tsql

I'm trying to add a WITH clause to a statement within a scalar function but when I do, I get the following syntax errors:
SQL80001: Incorrect syntax near 'WITH'
SQL80001: Incorrect syntax near ')'
The last error refers to the final closing bracket.
Here a greatly-simplified example of the SQL which also fails for the same reason:
CREATE FUNCTION IsSumEqualToTen
(
#number1 INT,
#number2 INT
)
RETURNS BIT
AS
BEGIN
RETURN
(
WITH AddNumbers AS
(
SELECT #number1 + #number2
)
SELECT CASE WHEN AddNumbers = 10
THEN 1
ELSE 0
END
)
END
I know that when changed to this, it has no syntax errors:
CREATE FUNCTION IsSumEqualToTen
(
#number1 INT,
#number2 INT
)
RETURNS BIT
AS
BEGIN
RETURN
(
CASE WHEN (#number1 + #number2) = 10
THEN 1
ELSE 0
END
)
END
But for what I'm actually doing I need a recursive CTE to flatten a heirarchy so the WITH is essential.

First, I must say I agree with Larnu on this one - using a scalar function is probably not the best course of action you can take.
However, I wanted to show you that there's no problem using a common table expression in a user defined function - table valued or scalar.
The problem with the code you've shown is that you do have some syntax errors as well as some conceptual errors. A fixed version of that function (still using a cte) is this:
CREATE FUNCTION IsSumEqualToTen
(
#number1 INT,
#number2 INT
)
RETURNS BIT
AS
BEGIN
DECLARE #Result bit;
WITH AddNumbers(AddNumbers) AS
(
SELECT #number1 + #number2
)
SELECT #Result = CASE WHEN MAX(AddNumbers) = 10
THEN 1
ELSE 0
END
FROM AddNumberes;
RETURN #Result;
END
Notes:
Each column in the cte result set must be named.
A cte is treated as a table, you must take into consideration the fact that it might contain more than one row (even if it can only contain one, the compiler can't know that).
scalar functions are not inlined, this means that you don't do return (body here) but body...; return scalarValue. Infact, this is how the official documentation looks like:
CREATE [ OR ALTER ] FUNCTION [ schema_name. ] function_name
( [ { #parameter_name [ AS ][ type_schema_name. ] parameter_data_type
[ = default ] [ READONLY ] }
[ ,...n ]
]
)
RETURNS return_data_type
[ WITH <function_option> [ ,...n ] ]
[ AS ]
BEGIN
function_body
RETURN scalar_expression
END
[ ; ]

Related

How to fix an expression of non-boolean type specified in a context where a condition is expected, near 'BEGIN'.?

Hey I have this query
IF (select kmkood from or_arved_read where kmkood = '1' or kmkood = '14' or kmkood = '15' or kmkood = '6' )
BEGIN
SET #stat_vat = 21
IF (select kmkood from or_arved_read where kmkood = '2' or kmkood = '7' )
SET #stat_vat = 12
But it returns An expression of non-boolean type specified in a context where a condition is expected, near 'BEGIN'.
how can i fix it?
Firstly, your IF doesn't have a boolean expression. The subquery returns a value (or more accurately for this scenario will return many and cause a different error) for the column kmkood, but then you don't do anything with that value. What value does that column need? The format should be something like this:
IF (SELECT SomeColumn FROM dbo.SomeTable WHERE... ) = 'SomeValue'
Also, if you have a BEGIN you need an END afterwards, which you don't have:
IF (SELECT SomeColumn FROM dbo.SomeTable WHERE... ) = 'SomeValue'
BEGIN
{Do several statements}
END
As, however, you are just doing a SET statement then you don't actually need the BEGIN and END statements.
I suspect what you are really after here, however, in an EXISTS:
IF EXISTS(SELECT 1 FROM dbo.or_arved_read WHERE kmkood IN (1,14,15,6))
SET #stat_vat = 21;
IF EXISTS(SELECT 1 FROM dbo.or_arved_read WHERE kmkood IN (2,7))
SET #stat_vat = 12;
You don't need all those ORs either, an IN works fine, and numbers shouldn't be in single quotes, so that makes things a bit shorter.

Multiple AND statements in a FileMaker calculation

I want to calculate if a value falls outside of 10% of the last two values added to a database. This calculation is not giving me correct feedback when I have 'Weight" values close to 10, or from 100-110. Otherwise it works fine.
Case (
Get(RecordNumber) ≤ 2 ; "Continue Collecting Data" ;
(((GetNthRecord ( Weight ; Get(RecordNumber)-2))*.9) ≤ Weight) and
(((GetNthRecord ( Weight ; Get(RecordNumber)-2))*1.1) ≥ Weight) and
(((GetNthRecord ( Weight ; Get(RecordNumber)-1))*.9) ≤ Weight) and
(((GetNthRecord ( Weight ; Get(RecordNumber)-1))*1.1) ≥ Weight);
"Stable";
"Unstable")
I’m going to start with the assumption that your table includes fields for both the primary key and the creation timestamp. If not, I would highly recommend adding both to this table and every other table you create.
Assuming these fields are in place, you need to create another occurrence of the table on which this layout is based, then relate the primary key to itself via a Cartesian (×) join. Sort the relationship by creation timestamp descending. Your calculation is then:
Case (
(((GetNthRecord ( Weight ; 1 ) *.9 ) ≤ Weight) and
(((GetNthRecord ( Weight ; 1 ) *1.1 ) ≥ Weight) and
(((GetNthRecord ( Weight ; 2 ) *.9 ) ≤ Weight) and
(((GetNthRecord ( Weight ; 2 ) *1.1 ) ≥ Weight);
"Stable";
"Unstable")
Another thing I noticed is that your code is kind of complex. The Let function can make things easier to read, and your four criteria can be cut down to whether either difference is out of range. So, a simpler version becomes:
Let ( [
#weight1 = GetNthRecord ( all::weight ; 1 ) ;
#weight2 = GetNthRecord ( all::weight ; 2 )
] ; //end define Let
Case (
Abs ( #weight1 - weight ) > .1 ; "Unstable" ;
Abs ( #weight2 - weight ) > .1 ; "Unstable" ;
"Stable"
) //end Case
) //end Let
Does that help?
Assuming you are using FileMaker v12 or later, this looks like a good place to use the ExecuteSQL function (not the script step) to retrieve the last two values. You could do something like this:
Let (
sqlQuery = "
SELECT t.weight
FROM MyTable t
WHERE t.id <> ?
ORDER BY t.id DESC
FETCH FIRST 2 ROWS ONLY
" ;
ExecuteSQL ( sqlQuery ; "" ; "" ; MyTable::id )
)
This query assumes you have a unique 'id' field (i.e. a primary key) that's defined as an 'auto-enter serial' value. The WHERE clause makes sure that the current record (presumably the one the user is entering) is not included in the query. The ORDER BY DESC clause forces the last two records to the top where we can fetch the 'weight' values easily into a value list.
Assuming you use a 'Set Variable' script step to put the query results into $lastValues, you can then test for whether they are in range like so:
Let ( [
lastValue1 = GetValue ( $lastValues ; 1 ) ;
lastValue2 = GetValue ( $lastValues ; 2 ) ;
Value1_InRange = lastValue1 - Abs ( lastValue1 - weight ) >= ( 0.9 * lastValue1 ) ;
Value2_InRange = lastValue2 - Abs ( lastValue2 - weight ) >= ( 0.9 * lastValue2 )
] ;
Value1_InRange and Value2_InRange // returns 1 if both values within range, 0 if not
)
If I were doing this, I would put the above range-checking code into a custom function so it's generic and can be easily reused:
IsWithinRange ( valueToTest ; lastValue ; range ) =
lastValue - Abs ( lastValue - valueToTest ) >= ( ( 1 - range ) * lastValue )
Then the above range-checking code can be reduced to:
IsWithinRange ( MyTable::weight ; GetValue ( $lastValues ; 1 ) ; 0.1 ) &
IsWithinRange ( MyTable::weight ; GetValue ( $lastValues ; 2 ) ; 0.1 )
And one last note.. if you use the ExecuteSQL function in a calculated field, be sure to make it 'unstored' so that it only executes when needed. However, I would recommend you avoid that altogether and simply call it from a script step like 'Set Variable'. That way you can control exactly when it executes.
Hope that helps!

PostgreSQL ERROR: invalid input syntax for integer: "1e+06"

The full error message is:
ERROR: invalid input syntax for integer: "1e+06"
SQL state: 22P02
Context: In PL/R function sample
The query I'm using is:
WITH a as
(
SELECT a.tract_id_alias,
array_agg(a.pgid ORDER BY a.pgid) as pgids,
array_agg(a.sample_weight_geo ORDER BY a.pgid) as block_weights
FROM results_20161109.block_microdata_res_joined a
WHERE a.tract_id_alias in (66772, 66773, 66785, 66802, 66805, 66806, 66813)
AND a.bldg_count_res > 0
GROUP BY a.tract_id_alias
)
SELECT NULL::INTEGER agent_id,
a.tract_id_alias,
b.year,
unnest(shared.sample(a.pgids,
b.n_agents,
1 * b.year,
True,
a.block_weights)
) as pgid
FROM a
LEFT JOIN results_20161109.initial_agent_count_by_tract_res_11 b
ON a.tract_id_alias = b.tract_id_alias
ORDER BY b.year, a.tract_id_alias, pgid;
And the shared.sample function I'm using is:
CREATE OR REPLACE FUNCTION shared.sample(ids bigint[], size integer, seed integer DEFAULT 1, with_replacement boolean DEFAULT false, probabilities numeric[] DEFAULT NULL::numeric[])
RETURNS integer[] AS
$BODY$
set.seed(seed)
if (length(ids) == 1) {
s = rep(ids,size)
} else {
s = sample(ids,size, with_replacement,probabilities)
}
return(s)
$BODY$
LANGUAGE plr VOLATILE
COST 100;
ALTER FUNCTION shared.sample(bigint[], integer, integer, boolean, numeric[])
OWNER TO "server-superusers";
I'm pretty new to this stuff, so any help would be appreciated.
Not a problem of the function. Like the error messages says: The string '1e+06' cannot be cast to integer.
Obviously, the columns n_agents in your table results_20161109.initial_agent_count_by_tract_res_11 is not an integer column. Probably type text or varchar? (That info would help in your question.)
Either way, the assignment cast does not work for the target type integer. But it does for numeric:
Does not work:
SELECT '1e+06'::text::int; -- error as in question
Works:
SELECT '1e+06'::text::numeric::int;
If my assumptions hold, you can use this as stepping stone.
Replace b.n_agents in your query with b.n_agents::numeric::int.
It's your responsibility that numbers stay in integer range, or you get the next exception.
If that did not nail it, you need to look into function overloading:
Is there a way to disable function overloading in Postgres
And function type resolution:
PostgreSQL function call
The schema search path is relevant in many related cases, but you did schema-qualify all objects, so we can rule that out.
How does the search_path influence identifier resolution and the "current schema"
Your query generally looks good. I had a look and only found minor improvements:
SELECT NULL::int AS agent_id -- never omit the AS keyword for column alias
, a.tract_id_alias
, b.year
, s.pgid
FROM (
SELECT tract_id_alias
, array_agg(pgid) AS pgids
, array_agg(sample_weight_geo) AS block_weights
FROM ( -- use a subquery, cheaper than CTE
SELECT tract_id_alias
, pgid
, sample_weight_geo
FROM results_20161109.block_microdata_res_joined
WHERE tract_id_alias IN (66772, 66773, 66785, 66802, 66805, 66806, 66813)
AND bldg_count_res > 0
ORDER BY pgid -- sort once in a subquery. cheaper.
) sub
GROUP BY 1
) a
LEFT JOIN results_20161109.initial_agent_count_by_tract_res_11 b USING (tract_id_alias)
LEFT JOIN LATERAL
unnest(shared.sample(a.pgids
, b.n_agents
, b.year -- why "1 * b.year"?
, true
, a.block_weights)) s(pgid) ON true
ORDER BY b.year, a.tract_id_alias, s.pgid;

Can Firebirds SELECT FIRST accept a variable?

http://www.firebirdsql.org/refdocs/langrefupd20-select.html#langrefupd20-first-skip
The manual says that FIRST accepts "Any expression evaluating to an integer." Shouldn't this mean a variable too?
In the following stored procedure, I get an error trying to supply :DAYS to FIRST.
Token unknown - line 10, column 18
:
Line 10 column 18 is the : in front of DAYS...
SET TERM ^ ;
CREATE PROCEDURE P_STOCK_MDA
( STOCK BIGINT, TRADE_DATE DATE, DAYS SMALLINT )
RETURNS
( AVG_CLOSE NUMERIC(6,2) )
AS
BEGIN
SELECT AVG(STOCK_ADJ_CLOSE) FROM (
SELECT FIRST :DAYS STOCK_ADJ_CLOSE
FROM STOCK_DAILY yd
WHERE yd.STOCK_STOCK=:STOCK AND yd.TRADE_DATE<=:TRADE_DATE
ORDER BY yd.TRADE_DATE DESC
) INTO AVG_CLOSE;
END^
You need to enclose the parameter in parentheses to get it working:
SELECT FIRST (:DAYS) STOCK_ADJ_CLOSE
Full code:
SET TERM ^ ;
CREATE PROCEDURE P_STOCK_MDA
( STOCK BIGINT, TRADE_DATE DATE, DAYS SMALLINT )
RETURNS
( AVG_CLOSE NUMERIC(6,2) )
AS
BEGIN
SELECT AVG(STOCK_ADJ_CLOSE) FROM (
SELECT FIRST (:DAYS) STOCK_ADJ_CLOSE
FROM STOCK_DAILY yd
WHERE yd.STOCK_STOCK=:STOCK AND yd.TRADE_DATE<=:TRADE_DATE
ORDER BY yd.TRADE_DATE DESC
) INTO AVG_CLOSE;
END^
SET TERM ; ^
The documentation does say :
If <int-expr> is an integer literal or a query parameter, the “()” may be omitted
However I assume this only applies to the ? query parameter in DSQL, not to the named parameters in PSQL.

Assign case statement result to variable

When I try to run the following statement, I get this error:
"A SELECT statement that assigns a value to a variable must not be combined with data-retrieval operations."
DECLARE #OFR DECIMAL(18,2)
;
SELECT #OFR =
CASE SUM(ofr.Quantity)
WHEN 0 THEN 0
ELSE SUM(ofr.Gross) / SUM(ofr.Quantity)
END
FROM DistributionCosts ofr
;
SELECT #OFR
I just want to stick the result of that case statement into a variable and I'm getting stuck here.
Any ideas?
Thanks.
Are you sure this is where your error is? I just ran the statement below and it worked:
DECLARE #OFR DECIMAL(18,2)
;
with DistributionCosts as
(
select 1 as quantity, 5 as gross
union
select 5 , 20
)
SELECT #OFR =
CASE SUM(ofr.Quantity)
WHEN 0 THEN 0
ELSE SUM(ofr.Gross) / SUM(ofr.Quantity)
END
FROM DistributionCosts ofr
;
SELECT #OFR
If I select a column in addition to setting a variable I do get your error... If you are selecting another column you will need to either stop selecting that column or assign that column to a local variable as well.