arachne-node 0.1.0

Iroh-based peer connectivity and bounded data transport for Arachne.
Documentation
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
//! Authorized immutable byte transfers. Catalogs and recipient grants belong to
//! the caller; this module binds a short-lived read capability to a verified
//! workspace endpoint and runs the standard iroh-blobs range/verification engine.
//! No file bytes enter the message journal and no file-size ceiling is imposed.
use std::{
    collections::{BTreeMap, HashSet},
    io,
    path::{Path, PathBuf},
    sync::{
        Arc, Mutex as StdMutex,
        atomic::{AtomicBool, Ordering},
    },
    time::{Duration, Instant},
};

use bytes::Bytes;
use arachne_routing::RoutingTable;
use futures_util::StreamExt;
use iroh::endpoint::{Connection, RecvStream, SendStream, VarInt};
use iroh_blobs::{
    Hash,
    api::{
        TempTag,
        blobs::{AddPathOptions, ExportMode, ExportOptions, ImportMode},
        remote::GetProgressItem,
    },
    get,
    protocol::Request,
    provider::{self, events::EventSender},
    store::{fs::FsStore, gc_run_once},
    util::{RecvStream as BlobReader, SendStream as BlobWriter},
};
use serde::{Deserialize, Serialize};
use tokio::sync::{Mutex, OnceCell, Semaphore, watch};

use crate::{ALPN, Error, PeerId, Result, WorkspaceId, connections::Connections, transport};

pub(crate) const STREAM_KIND: u8 = 1;
const ADMISSION_LIFETIME: Duration = Duration::from_secs(120);
const IDLE_TIMEOUT: Duration = Duration::from_secs(60);

/// Transmitted only inside an authenticated, recipient-scoped application
/// message. The hash identifies bytes; neither a hash nor this token alone is
/// authority. Serving also requires the bound endpoint and current membership.
#[derive(Clone, Debug, Serialize, Deserialize)]
#[serde(deny_unknown_fields)]
pub struct ResourceTicket {
    pub hash: [u8; 32],
    pub size: u64,
    pub grant: [u8; 32],
}

struct ReadGrant {
    workspace: WorkspaceId,
    revision: u64,
    peer: PeerId,
    path: PathBuf,
    expires: Instant,
    tag: TempTag,
}

struct Shared {
    connections: Connections,
    routing: Arc<Mutex<RoutingTable>>,
    store: OnceCell<(PathBuf, FsStore)>,
    grants: StdMutex<BTreeMap<[u8; 32], Arc<ReadGrant>>>,
    changed: watch::Sender<u64>,
    serving: Semaphore,
    downloading: Mutex<()>,
    collecting: Mutex<()>,
    closed: AtomicBool,
}

#[derive(Clone)]
pub struct ResourceTransfers(Arc<Shared>);

impl ResourceTransfers {
    pub(crate) fn new(connections: Connections, routing: Arc<Mutex<RoutingTable>>) -> Self {
        Self(Arc::new(Shared {
            connections,
            routing,
            store: OnceCell::new(),
            grants: StdMutex::new(BTreeMap::new()),
            changed: watch::channel(0).0,
            serving: Semaphore::new(4),
            downloading: Mutex::new(()),
            collecting: Mutex::new(()),
            closed: AtomicBool::new(false),
        }))
    }

    async fn store(&self, root: &Path) -> Result<FsStore> {
        if !root.is_absolute() {
            return Err(Error::Rejected);
        }
        let (path, store) = self
            .0
            .store
            .get_or_try_init(|| async {
                let store = FsStore::load(root).await.map_err(transport)?;
                // One resumable partial per workspace. Unreferenced old import
                // indexes/outboards are reclaimed; immutable source files are external.
                gc_run_once(&store, &mut HashSet::new())
                    .await
                    .map_err(transport)?;
                Ok::<_, Error>((root.to_owned(), store))
            })
            .await?;
        if path != root {
            return Err(Error::Rejected);
        }
        Ok(store.clone())
    }

    async fn authorize(&self, workspace: WorkspaceId, revision: u64, peer: PeerId) -> Result<()> {
        if self.0.closed.load(Ordering::Acquire) {
            return Err(Error::Rejected);
        }
        let routing = self.0.routing.lock().await;
        routing.authorizes_endpoint(workspace, revision, self.0.connections.id())?;
        routing.authorizes_endpoint(workspace, revision, peer)?;
        Ok(())
    }

    /// Caller has already checked this resource's public/direct audience. The
    /// source must be an application-owned immutable snapshot, not a live file.
    pub async fn prepare(
        &self,
        root: &Path,
        path: &Path,
        workspace: WorkspaceId,
        revision: u64,
        peer: PeerId,
    ) -> Result<ResourceTicket> {
        let generation = *self.0.changed.borrow();
        self.authorize(workspace, revision, peer).await?;
        if !path.is_absolute() {
            return Err(Error::Rejected);
        }
        let metadata = tokio::fs::metadata(path).await.map_err(transport)?;
        if !metadata.is_file() {
            return Err(Error::Rejected);
        }
        let store = self.store(root).await?;
        let tag = store
            .add_path_with_opts(AddPathOptions {
                path: path.to_owned(),
                mode: ImportMode::TryReference,
                format: iroh_blobs::BlobFormat::Raw,
            })
            .temp_tag()
            .await
            .map_err(transport)?;
        self.authorize(workspace, revision, peer).await?;
        let mut token = [0; 32];
        getrandom::fill(&mut token).map_err(transport)?;
        let ticket = ResourceTicket {
            hash: *tag.hash().as_bytes(),
            size: metadata.len(),
            grant: token,
        };
        {
            let mut grants = self.0.grants.lock().unwrap();
            if generation != *self.0.changed.borrow() {
                return Err(Error::Rejected);
            }
            grants
                .retain(|_, grant| grant.expires > Instant::now() || Arc::strong_count(grant) > 1);
            if grants.len() >= 64 {
                return Err(Error::Backpressure);
            }
            grants.insert(
                token,
                Arc::new(ReadGrant {
                    workspace,
                    revision,
                    peer,
                    path: path.to_owned(),
                    expires: Instant::now() + ADMISSION_LIFETIME,
                    tag,
                }),
            );
        }
        self.collect(&store).await?;
        Ok(ticket)
    }

    /// Withdraw before deleting a source, or pass None when serving preferences
    /// change. Active streams are cancelled too, not just future admissions.
    pub fn revoke(&self, path: Option<&Path>) {
        let mut grants = self.0.grants.lock().unwrap();
        grants.retain(|_, grant| path.is_some_and(|path| grant.path != path));
        self.policy_changed();
    }

    pub(crate) fn policy_changed(&self) {
        self.0
            .changed
            .send_modify(|generation| *generation = generation.wrapping_add(1));
    }

    async fn collect(&self, store: &FsStore) -> Result<()> {
        let _serial = self.0.collecting.lock().await;
        gc_run_once(store, &mut HashSet::new())
            .await
            .map_err(transport)
    }

    async fn admitted(&self, token: &[u8; 32], grant: &ReadGrant) -> Result<()> {
        self.authorize(grant.workspace, grant.revision, grant.peer)
            .await?;
        if !self.0.grants.lock().unwrap().contains_key(token) {
            return Err(Error::Rejected);
        }
        Ok(())
    }

    pub(crate) async fn serve(
        &self,
        connection: &Connection,
        send: &mut SendStream,
        recv: &mut RecvStream,
    ) -> Result<()> {
        let _slot = self
            .0
            .serving
            .try_acquire()
            .map_err(|_| Error::Backpressure)?;
        let mut changes = self.0.changed.subscribe();
        let mut token = [0; 32];
        tokio::time::timeout(crate::TIMEOUT, recv.read_exact(&mut token))
            .await
            .map_err(|_| Error::Timeout("resource admission"))?
            .map_err(transport)?;
        let grant = self
            .0
            .grants
            .lock()
            .unwrap()
            .get(&token)
            .cloned()
            .ok_or(Error::Rejected)?;
        if grant.peer != *connection.remote_id().as_bytes() || grant.expires <= Instant::now() {
            return Err(Error::Rejected);
        }
        self.admitted(&token, &grant).await?;
        let (_, store) = self.0.store.get().ok_or(Error::Rejected)?;
        let mut pair = provider::StreamPair::new(
            connection.stable_id() as u64,
            SizedReader {
                inner: recv,
                size: None,
                header: [0; 8],
                read: 0,
            },
            IdleWriter(send),
            EventSender::default(),
        );
        let request = tokio::time::timeout(crate::TIMEOUT, pair.read_request())
            .await
            .map_err(|_| Error::Timeout("blob request"))?
            .map_err(transport)?;
        let Request::Get(request) = request else {
            return Err(Error::Rejected);
        };
        // No push, collections, observe or get-many surface. A grant covers only
        // one raw immutable blob, including its authenticated resume ranges.
        if request.hash != grant.tag.hash() || !request.ranges.is_blob() {
            return Err(Error::Rejected);
        }
        let transfer = provider::handle_get(pair, store.as_ref().clone(), request);
        tokio::pin!(transfer);
        loop {
            tokio::select! {
                result = &mut transfer => return result.map_err(transport),
                result = changes.changed() => {
                    result.map_err(transport)?;
                    self.admitted(&token, &grant).await?;
                }
            }
        }
    }

    /// Standard verified range download; retains at most one partial in this
    /// workspace store. `target` is a private caller-owned temporary path.
    /// The caller reserves disk/quota using the authenticated advertised size.
    #[allow(clippy::too_many_arguments)]
    pub async fn fetch(
        &self,
        root: &Path,
        target: &Path,
        workspace: WorkspaceId,
        revision: u64,
        peer: PeerId,
        ticket: ResourceTicket,
        progress: watch::Sender<u64>,
    ) -> Result<()> {
        let _slot = self
            .0
            .downloading
            .try_lock()
            .map_err(|_| Error::Backpressure)?;
        self.authorize(workspace, revision, peer).await?;
        if !target.is_absolute() {
            return Err(Error::Rejected);
        }
        let store = self.store(root).await?;
        let hash = Hash::from_bytes(ticket.hash);
        store
            .tags()
            .set(b"partial", hash)
            .await
            .map_err(transport)?;
        self.collect(&store).await?;
        let local = store.remote().local(hash).await.map_err(transport)?;
        progress.send_replace(local.local_bytes());
        if !local.is_complete() {
            let mut changes = self.0.changed.subscribe();
            let connection =
                tokio::time::timeout(IDLE_TIMEOUT, self.0.connections.connect(peer, ALPN))
                    .await
                    .map_err(|_| Error::Timeout("resource connect"))??;
            self.authorize(workspace, revision, peer).await?;
            let (mut send, recv) = connection.open_bi().await.map_err(transport)?;
            send.write_all(&[STREAM_KIND]).await.map_err(transport)?;
            send.write_all(&ticket.grant).await.map_err(transport)?;
            let reader = SizedReader {
                inner: recv,
                size: Some(ticket.size),
                header: [0; 8],
                read: 0,
            };
            let pair = get::StreamPair::new(connection.stable_id() as u64, reader, send);
            let transfer = store.remote().fetch(pair, hash);
            let mut events = transfer.stream();
            let idle = tokio::time::sleep(IDLE_TIMEOUT);
            tokio::pin!(idle);
            let result = loop {
                tokio::select! {
                    event = events.next() => match event {
                        Some(GetProgressItem::Progress(bytes)) => {
                            progress.send_replace(bytes);
                            idle.as_mut().reset(tokio::time::Instant::now() + IDLE_TIMEOUT);
                        }
                        Some(GetProgressItem::Done(_)) => break Ok(()),
                        Some(GetProgressItem::Error(error)) => break Err(transport(error)),
                        None => break Err(Error::Rejected),
                    },
                    _ = &mut idle => break Err(Error::Timeout("resource progress")),
                    result = changes.changed() => {
                        if let Err(error) = result.map_err(transport) { break Err(error); }
                        if let Err(error) = self.authorize(workspace, revision, peer).await { break Err(error); }
                    }
                }
            };
            // Flush verified partial ranges even when the peer disappears.
            drop(events);
            store.sync_db().await.map_err(transport)?;
            result?;
        }
        self.authorize(workspace, revision, peer).await?;
        store
            .export_with_opts(ExportOptions {
                hash,
                target: target.to_owned(),
                mode: ExportMode::TryReference,
            })
            .await
            .map_err(transport)?;
        if tokio::fs::metadata(target).await.map_err(transport)?.len() != ticket.size {
            return Err(Error::Rejected);
        }
        store.tags().delete(b"partial").await.map_err(transport)?;
        self.collect(&store).await?;
        progress.send_replace(ticket.size);
        Ok(())
    }

    /// Clear only the private resumable transfer, not retained/published files.
    pub async fn clear_partial(&self, root: &Path) -> Result<()> {
        let _slot = self.0.downloading.lock().await;
        if root.exists() {
            let store = self.store(root).await?;
            store.tags().delete(b"partial").await.map_err(transport)?;
            self.collect(&store).await?;
        }
        Ok(())
    }

    pub(crate) async fn close(&self) {
        self.stop();
        if let Some((_, store)) = self.0.store.get() {
            let _ = store.shutdown().await;
        }
    }

    pub(crate) fn stop(&self) {
        self.0.closed.store(true, Ordering::Release);
        self.revoke(None);
    }
}

/// Check the standard blob size header before its decoder/storage sees it.
/// Advertised length is a resource integrity boundary, not a file-size limit.
struct SizedReader<R> {
    inner: R,
    size: Option<u64>,
    header: [u8; 8],
    read: usize,
}

impl<R> SizedReader<R> {
    fn validate(&mut self, bytes: &[u8]) -> io::Result<()> {
        let Some(size) = self.size else {
            return Ok(());
        };
        let length = (8 - self.read).min(bytes.len());
        self.header[self.read..self.read + length].copy_from_slice(&bytes[..length]);
        self.read += length;
        if self.read == 8 && u64::from_le_bytes(self.header) != size {
            return Err(io::Error::new(
                io::ErrorKind::InvalidData,
                "resource length differs from grant",
            ));
        }
        Ok(())
    }
}

impl<R: BlobReader> BlobReader for SizedReader<R> {
    async fn recv_bytes(&mut self, len: usize) -> io::Result<Bytes> {
        let bytes = io_idle(self.inner.recv_bytes(len)).await?;
        self.validate(&bytes)?;
        Ok(bytes)
    }
    async fn recv_bytes_exact(&mut self, len: usize) -> io::Result<Bytes> {
        let bytes = io_idle(self.inner.recv_bytes_exact(len)).await?;
        self.validate(&bytes)?;
        Ok(bytes)
    }
    async fn recv_exact(&mut self, target: &mut [u8]) -> io::Result<()> {
        io_idle(self.inner.recv_exact(target)).await?;
        self.validate(target)
    }
    fn stop(&mut self, code: VarInt) -> io::Result<()> {
        BlobReader::stop(&mut self.inner, code)
    }
    fn id(&self) -> u64 {
        BlobReader::id(&self.inner)
    }
}

async fn io_idle<T>(operation: impl std::future::Future<Output = io::Result<T>>) -> io::Result<T> {
    tokio::time::timeout(IDLE_TIMEOUT, operation)
        .await
        .map_err(|_| io::Error::new(io::ErrorKind::TimedOut, "resource stream stalled"))?
}

/// A peer keeping QUIC alive without reading must release scarce serve slots.
/// This limits idle I/O, never the duration or size of a progressing transfer.
struct IdleWriter<'a>(&'a mut SendStream);
impl BlobWriter for IdleWriter<'_> {
    async fn send_bytes(&mut self, bytes: Bytes) -> io::Result<()> {
        io_idle(BlobWriter::send_bytes(self.0, bytes)).await
    }
    async fn send(&mut self, bytes: &[u8]) -> io::Result<()> {
        io_idle(BlobWriter::send(self.0, bytes)).await
    }
    async fn sync(&mut self) -> io::Result<()> {
        io_idle(BlobWriter::sync(self.0)).await
    }
    fn reset(&mut self, code: VarInt) -> io::Result<()> {
        BlobWriter::reset(self.0, code)
    }
    async fn stopped(&mut self) -> io::Result<Option<VarInt>> {
        io_idle(BlobWriter::stopped(self.0)).await
    }
    fn id(&self) -> u64 {
        BlobWriter::id(self.0)
    }
}

#[tokio::test(start_paused = true)]
async fn idle_deadline_is_not_a_transfer_duration_limit() {
    let error = io_idle(std::future::pending::<io::Result<()>>())
        .await
        .unwrap_err();
    assert_eq!(error.kind(), io::ErrorKind::TimedOut);
    let started = tokio::time::Instant::now();
    for _ in 0..3 {
        io_idle(async {
            tokio::time::sleep(IDLE_TIMEOUT / 2).await;
            Ok(())
        })
        .await
        .unwrap();
    }
    assert!(started.elapsed() > IDLE_TIMEOUT);
}