fast_pull/core/
multi.rs

1extern crate alloc;
2use super::macros::poll_ok;
3use crate::{DownloadResult, Event, ProgressEntry, RandPuller, RandPusher, Total, WorkerId};
4use alloc::{sync::Arc, vec::Vec};
5use bytes::Bytes;
6use core::{
7    num::NonZeroUsize,
8    sync::atomic::{AtomicUsize, Ordering},
9    time::Duration,
10};
11use fast_steal::{Executor, Handle, Task, TaskList};
12use futures::TryStreamExt;
13use tokio::task::AbortHandle;
14
15#[derive(Debug, Clone)]
16pub struct DownloadOptions {
17    pub download_chunks: Vec<ProgressEntry>,
18    pub concurrent: NonZeroUsize,
19    pub retry_gap: Duration,
20    pub push_queue_cap: usize,
21    pub min_chunk_size: u64,
22}
23
24pub async fn download_multi<R, W>(
25    puller: R,
26    mut pusher: W,
27    options: DownloadOptions,
28) -> DownloadResult<R::Error, W::Error>
29where
30    R: RandPuller + 'static + Sync,
31    W: RandPusher + 'static,
32{
33    let (tx, event_chain) = kanal::unbounded_async();
34    let (tx_push, rx_push) =
35        kanal::bounded_async::<(WorkerId, ProgressEntry, Bytes)>(options.push_queue_cap);
36    let tx_clone = tx.clone();
37    let push_handle = tokio::spawn(async move {
38        while let Ok((id, spin, data)) = rx_push.recv().await {
39            poll_ok!(
40                {},
41                pusher.push(spin.clone(), data.clone()).await,
42                id @ tx_clone => PushError,
43                options.retry_gap
44            );
45            tx_clone.send(Event::PushProgress(id, spin)).await.unwrap();
46        }
47        poll_ok!(
48            {},
49            pusher.flush().await,
50            tx_clone => FlushError,
51            options.retry_gap
52        );
53    });
54    let executor: TokioExecutor<R, W> = TokioExecutor {
55        tx,
56        tx_push,
57        puller,
58        retry_gap: options.retry_gap,
59        id: Arc::new(AtomicUsize::new(0)),
60        min_chunk_size: options.min_chunk_size,
61    };
62    let task_list = TaskList::run(
63        options.concurrent.get(),
64        options.min_chunk_size,
65        &options.download_chunks[..],
66        executor,
67    );
68    DownloadResult::new(
69        event_chain,
70        push_handle,
71        &task_list
72            .handles()
73            .iter()
74            .map(|h| h.0.clone())
75            .collect::<Arc<[_]>>(),
76    )
77}
78
79#[derive(Clone)]
80pub struct TokioHandle(AbortHandle);
81impl Handle for TokioHandle {
82    type Output = ();
83    fn abort(&mut self) -> Self::Output {
84        self.0.abort();
85    }
86}
87pub struct TokioExecutor<R, W>
88where
89    R: RandPuller + 'static,
90    W: RandPusher + 'static,
91{
92    tx: kanal::AsyncSender<Event<R::Error, W::Error>>,
93    tx_push: kanal::AsyncSender<(WorkerId, ProgressEntry, Bytes)>,
94    puller: R,
95    retry_gap: Duration,
96    id: Arc<AtomicUsize>,
97    min_chunk_size: u64,
98}
99impl<R, W> Executor for TokioExecutor<R, W>
100where
101    R: RandPuller + 'static + Sync,
102    W: RandPusher + 'static,
103{
104    type Handle = TokioHandle;
105    fn execute(self: Arc<Self>, task: Arc<Task>, task_list: Arc<TaskList<Self>>) -> Self::Handle {
106        let id = self.id.fetch_add(1, Ordering::SeqCst);
107        let handle = tokio::spawn(async move {
108            'steal_task: loop {
109                let mut start = task.start();
110                if start >= task.end() {
111                    if task_list.steal(&task, 2 * self.min_chunk_size) {
112                        continue;
113                    }
114                    break;
115                }
116                self.tx.send(Event::Pulling(id)).await.unwrap();
117                let download_range = start..task.end();
118                let mut puller = self.puller.clone();
119                let mut stream = puller.pull(&download_range);
120                loop {
121                    match stream.try_next().await {
122                        Ok(Some(mut chunk)) => {
123                            let len = chunk.len() as u64;
124                            task.fetch_add_start(len);
125                            let range_start = start;
126                            start += len;
127                            let range_end = start.min(task.end());
128                            if range_start >= range_end {
129                                continue 'steal_task;
130                            }
131                            let span = range_start..range_end;
132                            let len = span.total() as usize;
133                            self.tx
134                                .send(Event::PullProgress(id, span.clone()))
135                                .await
136                                .unwrap();
137                            self.tx_push
138                                .send((id, span, chunk.split_to(len)))
139                                .await
140                                .unwrap();
141                        }
142                        Ok(None) => break,
143                        Err(e) => {
144                            self.tx.send(Event::PullError(id, e)).await.unwrap();
145                            tokio::time::sleep(self.retry_gap).await;
146                        }
147                    }
148                }
149            }
150            self.tx.send(Event::Finished(id)).await.unwrap();
151            task_list.remove(&task);
152        });
153        TokioHandle(handle.abort_handle())
154    }
155}
156
157#[cfg(test)]
158mod tests {
159    extern crate std;
160    use super::*;
161    use crate::{
162        MergeProgress, ProgressEntry,
163        core::mock::{MockRandPuller, MockRandPusher, build_mock_data},
164    };
165    use alloc::vec;
166    use std::dbg;
167
168    #[tokio::test]
169    async fn test_concurrent_download() {
170        let mock_data = build_mock_data(3 * 1024);
171        let puller = MockRandPuller::new(&mock_data);
172        let pusher = MockRandPusher::new(&mock_data);
173        #[allow(clippy::single_range_in_vec_init)]
174        let download_chunks = vec![0..mock_data.len() as u64];
175        let result = download_multi(
176            puller,
177            pusher.clone(),
178            DownloadOptions {
179                concurrent: NonZeroUsize::new(32).unwrap(),
180                retry_gap: Duration::from_secs(1),
181                push_queue_cap: 1024,
182                download_chunks: download_chunks.clone(),
183                min_chunk_size: 1,
184            },
185        )
186        .await;
187
188        let mut pull_progress: Vec<ProgressEntry> = Vec::new();
189        let mut push_progress: Vec<ProgressEntry> = Vec::new();
190        let mut pull_ids = [false; 32];
191        let mut push_ids = [false; 32];
192        while let Ok(e) = result.event_chain.recv().await {
193            match e {
194                Event::PullProgress(id, p) => {
195                    pull_ids[id] = true;
196                    pull_progress.merge_progress(p);
197                }
198                Event::PushProgress(id, p) => {
199                    push_ids[id] = true;
200                    push_progress.merge_progress(p);
201                }
202                _ => {}
203            }
204        }
205        dbg!(&pull_progress);
206        dbg!(&push_progress);
207        assert_eq!(pull_progress, download_chunks);
208        assert_eq!(push_progress, download_chunks);
209        assert_eq!(pull_ids, [true; 32]);
210        assert_eq!(push_ids, [true; 32]);
211
212        result.join().await.unwrap();
213        pusher.assert().await;
214    }
215}