How can I pass both a scalar and a set of values in the bind.data
parameter of dbGetQuery()
for an SQL statement like
select * from tst where x = ? and y in (?)
Here's what I tried:
> library("RSQLite")
> c <- dbConnect (SQLite())
> dbGetQuery(c, "create table tst (x int, y int)")
> dbGetQuery(c, "insert into tst values (?, ?)", data.frame(x=c (1,2,1,2), y=c(3, 4, 5, 6)))
> dbReadTable(c, "tst")
x y
1 1 3
2 2 4
3 1 5
4 2 6
> dbGetQuery (c, "select * from tst where x = ? and y not in (?)", data.frame(x=2, y=I (list(7,6))))
Error in sqliteFetch(rs, n = -1, ...) :
RAW() can only be applied to a 'raw', not a 'double'
From reading the source, any non-data.frame bind.data
parameter gets forced through as.data.frame()
, so I guess there's little point in trying anything other than data frames.
NOTE: heck, it seems even binding a single set is problematic:
> dbGetQuery(c, "select * from tst where y not in (?)", c(7,6))
x y
1 1 3
2 2 4
3 1 5
4 2 6
5 1 3
6 2 4
7 1 5
This makes it clear that 2 separate queries (one of which returns 4 and one of which returns 3 results) are send from R; SQLite never sees a set parameter.
Earlier note: I want the database engine to filter the appropriate row, i don't want R to compute the cartesian product. In the above example, simply getting rid of I()
creates a 2-row dataframe (thanks to R's recycling), one of which is the solution. R sends each of these 2 rows to sqlite, and of course the second one matches. But the following shows the SQLite engine doesn't actually receive set parameters with regular data.frames:
> dbGetQuery(c, "select * from tst where x in (?) and y in (?)", data.frame(x=c(3,2), y=c(6,7)))
[1] x y
<0 rows> (or 0-length row.names)
> dbGetQuery(c, "select * from tst where x in (?) and y in (?)", data.frame(x=c(3,2), y=c(7,6)))
x y
1 2 6