Cursors and streaming
By default prepared query execution fetches all rows, you can use a
Cursor
to control the amount of rows you want to read:
connection.prepare("SELECT * FROM users WHERE first_name LIKE \$$1", { ar0 ->
if (ar0.succeeded()) {
var pq = ar0.result()
// Cursors require to run within a transaction
connection.begin({ ar1 ->
if (ar1.succeeded()) {
var tx = ar1.result()
// Create a cursor
var cursor = pq.cursor(Tuple.of("julien"))
// Read 50 rows
cursor.read(50, { ar2 ->
if (ar2.succeeded()) {
var rows = ar2.result()
// Check for more ?
if (cursor.hasMore()) {
// Repeat the process...
} else {
// No more rows - commit the transaction
tx.commit()
}
}
})
}
})
}
})
Cursors shall be closed when they are released prematurely:
cursor.read(50, { ar2 ->
if (ar2.succeeded()) {
// Close the cursor
cursor.close()
}
})
A stream API is also available for cursors, which can be more convenient, specially with the Rxified version.
connection.prepare("SELECT * FROM users WHERE first_name LIKE \$$1", { ar0 ->
if (ar0.succeeded()) {
var pq = ar0.result()
// Streams require to run within a transaction
connection.begin({ ar1 ->
if (ar1.succeeded()) {
var tx = ar1.result()
// Fetch 50 rows at a time
var stream = pq.createStream(50, Tuple.of("julien"))
// Use the stream
stream.exceptionHandler({ err ->
println("Error: ${err.getMessage()}")
})
stream.endHandler({ v ->
tx.commit()
println("End of stream")
})
stream.handler({ row ->
println("User: ${row.getString("last_name")}")
})
}
})
}
})
The stream read the rows by batch of 50
and stream them, when the rows have been passed to the handler,
a new batch of 50
is read and so on.
The stream can be resumed or paused, the loaded rows will remain in memory until they are delivered and the cursor will stop iterating.