#![allow(clippy::type_complexity)]
use crate::node::{InputStreams, Node, NodeExecutionError, OutputStreams};
use crate::nodes::common::BaseNode;
use async_trait::async_trait;
use futures::stream;
use std::any::Any;
use std::collections::HashMap;
use std::pin::Pin;
use std::sync::Arc;
use tokio_stream::{StreamExt, wrappers::ReceiverStream};
#[async_trait]
pub trait GroupByKeyFunction: Send + Sync {
async fn extract_key(&self, value: Arc<dyn Any + Send + Sync>) -> Result<String, String>;
}
pub type GroupByConfig = Arc<dyn GroupByKeyFunction>;
pub struct GroupByConfigWrapper(pub GroupByConfig);
impl GroupByConfigWrapper {
pub fn new(config: GroupByConfig) -> Self {
Self(config)
}
}
struct GroupByKeyFunctionWrapper<F> {
function: F,
}
#[async_trait]
impl<F> GroupByKeyFunction for GroupByKeyFunctionWrapper<F>
where
F: Fn(
Arc<dyn Any + Send + Sync>,
) -> std::pin::Pin<Box<dyn std::future::Future<Output = Result<String, String>> + Send>>
+ Send
+ Sync,
{
async fn extract_key(&self, value: Arc<dyn Any + Send + Sync>) -> Result<String, String> {
(self.function)(value).await
}
}
pub fn group_by_config<F, Fut>(function: F) -> GroupByConfig
where
F: Fn(Arc<dyn Any + Send + Sync>) -> Fut + Send + Sync + 'static,
Fut: std::future::Future<Output = Result<String, String>> + Send + 'static,
{
Arc::new(GroupByKeyFunctionWrapper {
function: move |v| {
Box::pin(function(v))
as std::pin::Pin<Box<dyn std::future::Future<Output = Result<String, String>> + Send>>
},
})
}
#[allow(dead_code)]
enum InputPort {
Config,
In,
KeyFunction,
}
pub struct GroupByNode {
pub(crate) base: BaseNode,
}
impl GroupByNode {
pub fn new(name: String) -> Self {
Self {
base: BaseNode::new(
name,
vec![
"configuration".to_string(),
"in".to_string(),
"key_function".to_string(),
],
vec!["out".to_string(), "error".to_string()],
),
}
}
}
#[async_trait]
impl Node for GroupByNode {
fn name(&self) -> &str {
self.base.name()
}
fn set_name(&mut self, name: &str) {
self.base.set_name(name);
}
fn input_port_names(&self) -> &[String] {
self.base.input_port_names()
}
fn output_port_names(&self) -> &[String] {
self.base.output_port_names()
}
fn has_input_port(&self, name: &str) -> bool {
self.base.has_input_port(name)
}
fn has_output_port(&self, name: &str) -> bool {
self.base.has_output_port(name)
}
fn execute(
&self,
mut inputs: InputStreams,
) -> Pin<
Box<dyn std::future::Future<Output = Result<OutputStreams, NodeExecutionError>> + Send + '_>,
> {
Box::pin(async move {
let _config_stream = inputs.remove("configuration");
let in_stream = inputs.remove("in").ok_or("Missing 'in' input")?;
let key_function_stream = inputs
.remove("key_function")
.ok_or("Missing 'key_function' input")?;
let in_stream = in_stream.map(|item| (InputPort::In, item));
let key_function_stream = key_function_stream.map(|item| (InputPort::KeyFunction, item));
let merged_stream: Pin<
Box<dyn futures::Stream<Item = (InputPort, Arc<dyn Any + Send + Sync>)> + Send>,
> = Box::pin(stream::select_all(vec![
Box::pin(in_stream)
as Pin<Box<dyn futures::Stream<Item = (InputPort, Arc<dyn Any + Send + Sync>)> + Send>>,
Box::pin(key_function_stream)
as Pin<Box<dyn futures::Stream<Item = (InputPort, Arc<dyn Any + Send + Sync>)> + Send>>,
]));
let (out_tx, out_rx) = tokio::sync::mpsc::channel(10);
let (error_tx, error_rx) = tokio::sync::mpsc::channel(10);
let out_tx_clone = out_tx.clone();
let error_tx_clone = error_tx.clone();
tokio::spawn(async move {
let mut merged_stream = merged_stream;
let mut data_buffer: Vec<Arc<dyn Any + Send + Sync>> = Vec::new();
let mut key_function: Option<GroupByConfig> = None;
while let Some((port, item)) = merged_stream.next().await {
match port {
InputPort::Config => {
}
InputPort::KeyFunction => {
if let Ok(wrapper) = Arc::downcast::<GroupByConfigWrapper>(item.clone()) {
key_function = Some(wrapper.0.clone());
} else {
let error_msg = format!(
"Invalid key function type: {} (expected GroupByConfigWrapper)",
std::any::type_name_of_val(&*item)
);
let error_arc = Arc::new(error_msg) as Arc<dyn Any + Send + Sync>;
let _ = error_tx_clone.send(error_arc).await;
return;
}
}
InputPort::In => {
data_buffer.push(item);
}
}
}
if let Some(func) = key_function {
let mut groups: HashMap<String, Vec<Arc<dyn Any + Send + Sync>>> = HashMap::new();
for item in data_buffer {
match func.extract_key(item.clone()).await {
Ok(key) => {
groups.entry(key).or_default().push(item);
}
Err(e) => {
let error_arc = Arc::new(e) as Arc<dyn Any + Send + Sync>;
let _ = error_tx_clone.send(error_arc).await;
}
}
}
let grouped_result: HashMap<String, Arc<dyn Any + Send + Sync>> = groups
.into_iter()
.map(|(key, items)| {
let items_arc = Arc::new(items) as Arc<dyn Any + Send + Sync>;
(key, items_arc)
})
.collect();
let result_arc = Arc::new(grouped_result) as Arc<dyn Any + Send + Sync>;
let _ = out_tx_clone.send(result_arc).await;
} else {
let error_msg = "No key function provided for grouping".to_string();
let error_arc = Arc::new(error_msg) as Arc<dyn Any + Send + Sync>;
let _ = error_tx_clone.send(error_arc).await;
}
});
let mut outputs = HashMap::new();
outputs.insert(
"out".to_string(),
Box::pin(ReceiverStream::new(out_rx))
as Pin<Box<dyn tokio_stream::Stream<Item = Arc<dyn Any + Send + Sync>> + Send>>,
);
outputs.insert(
"error".to_string(),
Box::pin(ReceiverStream::new(error_rx))
as Pin<Box<dyn tokio_stream::Stream<Item = Arc<dyn Any + Send + Sync>> + Send>>,
);
Ok(outputs)
})
}
}