Purpose of UNION and BIND in WHERE clause, implementing auto-numbering - sparql

Question

I am trying to implement increment/decrement using sparql/sesame query. I came across How to model consecutive numbering without transactions?, which proposes this query:
INSERT {
_:ex a ex:Invoice ;
ex:number ?next .
}
WHERE {
{
SELECT ((MAX(?number) + 1) AS ?next)
WHERE {
{ ?x ex:number ?number ;
a ex:Invoice }
UNION
{ BIND (0 AS ?number) }
}
}
}
I do not understand the purpose of union in the where clause. What is happening in where clause and with the bind? Is the value bound by bind used as a default if there is no triple present for it, meaning that 0 will be the first number?
EDIT:FOLLOW-UP:
Quick clarification....the query in my post (with coalesce) will insert multiple rows (which was the intent of that query). How can I modify that query to just increment/decrement 1 row?? If I add DELETE before insert, then what ld be the behavior? would the ?next be deleted before its selected???/
DELETE {
_:ex a ex:Invoice ;
ex:number ?object .
}
INSERT {
_:ex a ex:Invoice ;
ex:number ?next .
}
WHERE {
{
SELECT ((MAX(?number) + 1) AS ?next)
WHERE {
{ ?x ex:number ?number ;
a ex:Invoice }
UNION
{ BIND (0 AS ?number) }
}
}
}
Can I use query above to just increment 1 row and for the first time insert a row with value 0??

Answer 1

Yes.
What happens is that both arguments of the union provide a solution (the second arg always providing 0), and then the MAX in the projection makes sure that the highest number is chosen. Thus in most cases the second UNION argument does effectively nothing. However, in the case where the first argument does not provide a solution (because no invoice number is present yet), only the second argument will provide a solution and this will be picked as the max and then inserted (+1) as the first number.