use std::default::Default;
use std::rc::Rc;
use std::cell::RefCell;
use crate::scheduling::{Schedule, Activations};
use crate::progress::{Source, Target};
use crate::progress::{Timestamp, Operate, operate::SharedProgress, Antichain};
use crate::Container;
use crate::dataflow::{StreamCore, Scope};
use crate::dataflow::channels::pushers::Tee;
use crate::dataflow::channels::pact::ParallelizationContract;
use crate::dataflow::operators::generic::operator_info::OperatorInfo;
#[derive(Debug)]
pub struct OperatorShape {
name: String, notify: bool, peers: usize, inputs: usize, outputs: usize, }
impl OperatorShape {
fn new(name: String, peers: usize) -> Self {
OperatorShape {
name,
notify: true,
peers,
inputs: 0,
outputs: 0,
}
}
pub fn inputs(&self) -> usize {
self.inputs
}
pub fn outputs(&self) -> usize {
self.outputs
}
}
#[derive(Debug)]
pub struct OperatorBuilder<G: Scope> {
scope: G,
index: usize,
global: usize,
address: Rc<[usize]>, shape: OperatorShape,
summary: Vec<Vec<Antichain<<G::Timestamp as Timestamp>::Summary>>>,
}
impl<G: Scope> OperatorBuilder<G> {
pub fn new(name: String, mut scope: G) -> Self {
let global = scope.new_identifier();
let index = scope.allocate_operator_index();
let address = scope.addr_for_child(index);
let peers = scope.peers();
OperatorBuilder {
scope,
index,
global,
address,
shape: OperatorShape::new(name, peers),
summary: vec![],
}
}
pub fn index(&self) -> usize {
self.index
}
pub fn global(&self) -> usize {
self.global
}
pub fn shape(&self) -> &OperatorShape {
&self.shape
}
pub fn set_notify(&mut self, notify: bool) {
self.shape.notify = notify;
}
pub fn new_input<C: Container, P>(&mut self, stream: &StreamCore<G, C>, pact: P) -> P::Puller
where
P: ParallelizationContract<G::Timestamp, C> {
let connection = vec![Antichain::from_elem(Default::default()); self.shape.outputs];
self.new_input_connection(stream, pact, connection)
}
pub fn new_input_connection<C: Container, P>(&mut self, stream: &StreamCore<G, C>, pact: P, connection: Vec<Antichain<<G::Timestamp as Timestamp>::Summary>>) -> P::Puller
where
P: ParallelizationContract<G::Timestamp, C> {
let channel_id = self.scope.new_identifier();
let logging = self.scope.logging();
let (sender, receiver) = pact.connect(&mut self.scope, channel_id, self.address.clone(), logging);
let target = Target::new(self.index, self.shape.inputs);
stream.connect_to(target, sender, channel_id);
self.shape.inputs += 1;
assert_eq!(self.shape.outputs, connection.len());
self.summary.push(connection);
receiver
}
pub fn new_output<C: Container>(&mut self) -> (Tee<G::Timestamp, C>, StreamCore<G, C>) {
let connection = vec![Antichain::from_elem(Default::default()); self.shape.inputs];
self.new_output_connection(connection)
}
pub fn new_output_connection<C: Container>(&mut self, connection: Vec<Antichain<<G::Timestamp as Timestamp>::Summary>>) -> (Tee<G::Timestamp, C>, StreamCore<G, C>) {
let (targets, registrar) = Tee::<G::Timestamp,C>::new();
let source = Source::new(self.index, self.shape.outputs);
let stream = StreamCore::new(source, registrar, self.scope.clone());
self.shape.outputs += 1;
assert_eq!(self.shape.inputs, connection.len());
for (summary, entry) in self.summary.iter_mut().zip(connection.into_iter()) {
summary.push(entry);
}
(targets, stream)
}
pub fn build<L>(mut self, logic: L)
where
L: FnMut(&mut SharedProgress<G::Timestamp>)->bool+'static
{
let inputs = self.shape.inputs;
let outputs = self.shape.outputs;
let operator = OperatorCore {
shape: self.shape,
address: self.address,
activations: self.scope.activations(),
logic,
shared_progress: Rc::new(RefCell::new(SharedProgress::new(inputs, outputs))),
summary: self.summary,
};
self.scope.add_operator_with_indices(Box::new(operator), self.index, self.global);
}
pub fn operator_info(&self) -> OperatorInfo {
OperatorInfo::new(self.index, self.global, self.address.clone())
}
}
struct OperatorCore<T, L>
where
T: Timestamp,
L: FnMut(&mut SharedProgress<T>)->bool+'static,
{
shape: OperatorShape,
address: Rc<[usize]>,
logic: L,
shared_progress: Rc<RefCell<SharedProgress<T>>>,
activations: Rc<RefCell<Activations>>,
summary: Vec<Vec<Antichain<T::Summary>>>,
}
impl<T, L> Schedule for OperatorCore<T, L>
where
T: Timestamp,
L: FnMut(&mut SharedProgress<T>)->bool+'static,
{
fn name(&self) -> &str { &self.shape.name }
fn path(&self) -> &[usize] { &self.address[..] }
fn schedule(&mut self) -> bool {
let shared_progress = &mut *self.shared_progress.borrow_mut();
(self.logic)(shared_progress)
}
}
impl<T, L> Operate<T> for OperatorCore<T, L>
where
T: Timestamp,
L: FnMut(&mut SharedProgress<T>)->bool+'static,
{
fn inputs(&self) -> usize { self.shape.inputs }
fn outputs(&self) -> usize { self.shape.outputs }
fn get_internal_summary(&mut self) -> (Vec<Vec<Antichain<T::Summary>>>, Rc<RefCell<SharedProgress<T>>>) {
self.activations.borrow_mut().activate(&self.address[..]);
self.shared_progress
.borrow_mut()
.internals
.iter_mut()
.for_each(|output| output.update(T::minimum(), self.shape.peers as i64));
(self.summary.clone(), self.shared_progress.clone())
}
fn set_external_summary(&mut self) {
self.schedule();
}
fn notify_me(&self) -> bool { self.shape.notify }
}