1use std::io::SeekFrom;
3use std::path::Path;
4use vfsi_core::api::*;
5fn vector_index(error: vfsi_core::api::Error, index: usize) -> vfsi_core::api::Error {
6 if error.index().is_some() {
7 error.with_index(index)
8 } else {
9 error
10 }
11}
12struct VectorBudget {
13 entries: usize,
14 bytes: usize,
15}
16impl VectorBudget {
17 fn new(entries: usize, bytes: usize) -> Self {
18 Self { entries, bytes }
19 }
20 fn charge(&mut self, path: &Path) -> Result<()> {
21 if self.entries == 0 {
22 return Err(vfsi_core::api::Error::client(0, libc::EFBIG as u32)
23 .with_context("vector traversal", path));
24 }
25 self.charge_path(path)?;
26 self.entries -= 1;
27 Ok(())
28 }
29 fn charge_path(&mut self, path: &Path) -> Result<()> {
30 let bytes = path.as_os_str().len();
31 if bytes > self.bytes {
32 return Err(vfsi_core::api::Error::client(0, libc::EFBIG as u32)
33 .with_context("vector traversal", path));
34 }
35 self.bytes -= bytes;
36 Ok(())
37 }
38}
39pub fn visit_directory_pages<P: AsRef<Path>>(
40 roots: &[P],
41 policy: vfsi_core::api::VisitOptions,
42 limits: ResourceLimits,
43 mut capacity: impl FnMut(&[&Path]) -> Result<usize>,
44 mut validate_root: impl FnMut(&Path) -> Result<()>,
45 mut fetch: impl FnMut(
46 &[&Path],
47 Vec<Option<crate::DirPageCursor>>,
48 usize,
49 usize,
50 ) -> Result<Vec<crate::DirectoryPage>>,
51 mut callback: impl FnMut(usize, DirectoryListing) -> Result<std::ops::ControlFlow<()>>,
52) -> Result<Vec<crate::TraversalCompletion>> {
53 let directory = policy.directory_options(limits);
54 let walk = policy.walk_options(limits);
55 let mut budget = VectorBudget::new(directory.entry_limit(), directory.path_byte_limit());
56 let mut completed = Vec::new();
57 let mut pending = std::collections::VecDeque::new();
58 for (index, root) in roots.iter().enumerate() {
59 pending.push_back((index, root.as_ref().to_path_buf(), 0usize, None));
60 }
61 while !pending.is_empty() {
62 let owner = pending.front().unwrap().0;
63 if policy.is_recursive() && pending.front().unwrap().2 == 0 {
64 validate_root(&pending.front().unwrap().1)
65 .map_err(|error| vector_index(error, owner))?;
66 }
67 let candidate_paths: Vec<_> = pending
69 .iter()
70 .take_while(|state| !policy.is_recursive() || state.0 == owner)
71 .take(32)
72 .map(|state| state.1.as_path())
73 .collect();
74 let capacity = capacity(&candidate_paths)?.clamp(1, 32);
75 let cohort_size = if policy.is_recursive() {
76 pending
77 .iter()
78 .take(capacity)
79 .take_while(|state| state.0 == owner)
80 .count()
81 } else {
82 pending.len().min(capacity)
83 };
84 let mut cohort: Vec<_> = (0..cohort_size)
85 .map(|_| pending.pop_front().unwrap())
86 .collect();
87 let cursors = cohort.iter_mut().map(|state| state.3.take()).collect();
88 let paths: Vec<_> = cohort
89 .iter()
90 .map(|(_, path, _, _)| path.as_path())
91 .collect();
92 let requested = budget.entries.saturating_add(1);
93 let mut deferred_error = None;
94 let pages = match fetch(&paths, cursors, 1, requested) {
95 Ok(pages) => pages,
96 Err(error)
97 if !policy.is_recursive()
98 && !error.is_transport()
99 && error.index().is_some_and(|i| i > 0 && i < cohort.len()) =>
100 {
101 let failed = error.index().unwrap();
105 deferred_error = Some(vector_index(error, cohort[failed].0));
106 let prefix = fetch(
107 &paths[..failed],
108 (0..failed).map(|_| None).collect(),
109 1,
110 requested,
111 )
112 .map_err(|error| error.map_index(|i| cohort.get(i).map_or(i, |state| state.0)))?;
113 cohort.truncate(failed);
114 prefix
115 }
116 Err(error) => return Err(error.map_index(|i| cohort.get(i).map_or(i, |state| state.0))),
117 };
118 if pages.len() != cohort.len() {
119 return Err(vfsi_core::api::Error::transport(
120 None,
121 "invalid directory page result count",
122 ));
123 }
124 let mut children = Vec::new();
125 let mut wave: Vec<_> = cohort
126 .into_iter()
127 .map(|(index, path, depth, _)| (index, path, depth, true))
128 .zip(pages)
129 .collect();
130 while !wave.is_empty() {
131 let mut continuations = Vec::new();
132 for ((index, path, depth, first), (mut page, next, seeds)) in wave {
133 if first && policy.is_recursive() {
134 budget
135 .charge_path(&path)
136 .map_err(|error| vector_index(error, index))?;
137 }
138 if page.path != path || (page.entries.is_empty() && next.is_some()) {
139 return Err(vfsi_core::api::Error::transport(
140 Some(index),
141 "invalid directory page parent or progress",
142 ));
143 }
144 let mut seeds: std::collections::HashMap<_, _> = seeds.into_iter().collect();
145 let mut page_error = None;
146 let mut accepted = 0;
147 for entry in &page.entries {
148 if let Err(error) = budget.charge(entry.path()) {
149 page_error = Some(vector_index(error, index));
150 break;
151 }
152 accepted += 1;
153 if policy.is_recursive() && entry.metadata().is_dir() {
154 if depth >= walk.depth_limit() {
155 if !walk.truncates_at_depth_limit() {
156 page_error = Some(
157 vfsi_core::api::Error::client(index, libc::EFBIG as u32)
158 .with_context("visit_dirs", entry.path()),
159 );
160 break;
161 }
162 } else {
163 children.push((
164 index,
165 entry.path().to_path_buf(),
166 depth + 1,
167 seeds.remove(entry.path()),
168 ));
169 }
170 }
171 }
172 if let Some(error) = &page_error {
173 if accepted == 0 {
174 return Err(error.clone());
175 }
176 page.entries.truncate(accepted);
177 }
178 if !policy.is_recursive() && completed.len() <= index {
179 completed.resize(index + 1, crate::TraversalCompletion::Stopped);
180 }
181 if callback(index, page)
182 .map_err(|error| vector_index(error, index))?
183 .is_break()
184 {
185 if policy.is_recursive() {
186 completed.truncate(index);
187 completed.push(crate::TraversalCompletion::Stopped);
188 } else {
189 completed[index] = crate::TraversalCompletion::Stopped;
190 }
191 return Ok(completed);
192 }
193 if let Some(error) = page_error {
194 return Err(error);
195 }
196 match next {
197 Some(cursor) => continuations.push(((index, path, depth, false), Some(cursor))),
198 None if !policy.is_recursive() => {
199 completed[index] = crate::TraversalCompletion::Complete
200 }
201 None => {}
202 }
203 }
204 if continuations.is_empty() {
205 break;
206 }
207 let cursors = continuations
208 .iter_mut()
209 .map(|state| state.1.take())
210 .collect();
211 let paths: Vec<_> = continuations
212 .iter()
213 .map(|state| state.0.1.as_path())
214 .collect();
215 let requested = budget.entries.saturating_add(1);
216 let pages = fetch(&paths, cursors, requested.min(128), requested).map_err(|error| {
217 error.map_index(|i| continuations.get(i).map_or(i, |state| state.0.0))
218 })?;
219 if pages.len() != continuations.len() {
220 return Err(vfsi_core::api::Error::transport(
221 None,
222 "invalid directory continuation count",
223 ));
224 }
225 wave = continuations
226 .into_iter()
227 .map(|state| state.0)
228 .zip(pages)
229 .collect();
230 }
231 if let Some(error) = deferred_error {
232 return Err(error);
233 }
234 for child in children.into_iter().rev() {
237 pending.push_front(child);
238 }
239 if policy.is_recursive() && pending.front().is_none_or(|state| state.0 != owner) {
240 completed.push(crate::TraversalCompletion::Complete);
241 }
242 }
243 Ok(completed)
244}
245
246macro_rules! file_methods {
247 ($file:ty) => {
248 fn path(&self) -> &Path {
249 <$file>::path(self)
250 }
251 fn metadata(&self) -> Result<Metadata> {
252 <$file>::metadata(self)
253 }
254 fn read_request_at(&self, offset: u64, length: usize) -> Self::ReadRequest<'_> {
255 <$file>::read_request_at(self, offset, length)
256 }
257 fn read_request_at_into<'a>(
258 &'a self,
259 offset: u64,
260 buffer: &'a mut [u8],
261 ) -> Self::ReadIntoRequest<'a> {
262 <$file>::read_request_at_into(self, offset, buffer)
263 }
264 fn read_at(&self, buffer: &mut [u8], offset: u64) -> Result<usize> {
265 <$file>::read_at(self, buffer, offset)
266 }
267 fn write_at(&self, buffer: &[u8], offset: u64) -> Result<usize> {
268 <$file>::write_at(self, buffer, offset)
269 }
270 fn read_native(&mut self, buffer: &mut [u8]) -> Result<usize> {
271 <$file>::read_native(self, buffer)
272 }
273 fn read_to_end_with_limit(&mut self, max_bytes: usize) -> Result<Vec<u8>> {
274 <$file>::read_to_end_with_limit(self, max_bytes)
275 }
276 fn write_native(&mut self, buffer: &[u8]) -> Result<usize> {
277 <$file>::write_native(self, buffer)
278 }
279 fn seek_native(&mut self, position: SeekFrom) -> Result<u64> {
280 <$file>::seek_native(self, position)
281 }
282 fn sync_data(&self) -> Result<()> {
283 <$file>::sync_data(self)
284 }
285 fn sync_all(&self) -> Result<()> {
286 <$file>::sync_all(self)
287 }
288 fn try_close(&mut self) -> Result<()> {
289 <$file>::try_close(self)
290 }
291 fn is_closed(&self) -> bool {
292 <$file>::is_closed(self)
293 }
294 fn close(self) -> Result<()> {
295 <$file>::close(self)
296 }
297 };
298}
299
300impl<F: crate::FileSystem> FileHandle for crate::FsFile<F> {
301 type ReadRequest<'a>
302 = crate::FsRead<'a, F>
303 where
304 Self: 'a;
305 type ReadIntoRequest<'a>
306 = crate::FsReadInto<'a, F>
307 where
308 Self: 'a;
309 file_methods!(crate::FsFile<F>);
310}
311
312pub(crate) trait NativeHooks: Vfsi {
313 fn page_capacity(&self, paths: &[&Path]) -> Result<usize>;
314 fn open_native(&self, request: OpenRequest) -> Result<Self::File>;
315 fn stream_native(
316 &self,
317 path: impl AsRef<Path>,
318 options: crate::ReadStreamOptions,
319 callback: impl FnMut(u64, &[u8]) -> Result<bool>,
320 ) -> Result<crate::StreamCompletion>;
321}
322
323macro_rules! client_methods {
324 ($client:ty, $receiver:path) => {
325 client_methods!($client, $receiver, <$client>::vread);
326 };
327 ($client:ty, $receiver:path, $readv:expr) => {
328 client_methods!($client, $receiver, $readv, $receiver);
329 };
330 ($client:ty, $receiver:path, $readv:expr, $read_receiver:path) => {
331 client_methods!(
332 $client,
333 $receiver,
334 $readv,
335 $read_receiver,
336 <$client>::write_partial_native,
337 <$client>::write_complete
338 );
339 };
340 ($client:ty, $receiver:path, $readv:expr, $read_receiver:path, $writev:expr, $write_allv:expr) => {
341 client_methods!(
342 $client,
343 $receiver,
344 $readv,
345 $read_receiver,
346 $writev,
347 $write_allv,
348 <$client>::vgetattrs
349 );
350 };
351 ($client:ty, $receiver:path, $readv:expr, $read_receiver:path, $writev:expr, $write_allv:expr, $metadata:expr) => {
352 client_methods!(
353 $client,
354 $receiver,
355 $readv,
356 $read_receiver,
357 $writev,
358 $write_allv,
359 $metadata,
360 $receiver
361 );
362 };
363 ($client:ty, $receiver:path, $readv:expr, $read_receiver:path, $writev:expr, $write_allv:expr, $metadata:expr, $write_receiver:path) => {
364 client_methods!(
365 $client,
366 $receiver,
367 $readv,
368 $read_receiver,
369 $writev,
370 $write_allv,
371 $metadata,
372 $write_receiver,
373 vrename,
374 vmkdir,
375 vcopy,
376 vclose,
377 vopen
378 );
379 };
380 ($client:ty, $receiver:path, $readv:expr, $read_receiver:path, $writev:expr, $write_allv:expr, $metadata:expr, $write_receiver:path, $rename:ident, $mkdir:ident, $copy:ident, $close:ident, $open_batch:ident) => {
382 fn vrename<P: AsRef<Path>, Q: AsRef<Path>>(&self, pairs: &[(P, Q)]) -> Result<()> {
383 <$client>::$rename($receiver(self), pairs)
384 }
385 fn vlistdirs<P: AsRef<Path>>(
386 &self,
387 paths: &[P],
388 options: vfsi_core::api::VisitOptions,
389 callback: impl FnMut(usize, DirectoryListing) -> Result<std::ops::ControlFlow<()>>,
390 ) -> Result<Vec<crate::TraversalCompletion>> {
391 visit_directory_pages(
392 paths,
393 options,
394 self.limits(),
395 |paths| <$client as NativeHooks>::page_capacity($receiver(self), paths),
396 |path| {
397 let metadata = self.vgetattrs(
398 &[path],
399 vfsi_core::api::MetadataOptions::new()
400 .fields(vfsi_core::api::MetadataFields::MODE)
401 .follow_symlinks(false),
402 )?;
403 if metadata.len() != 1 {
404 return Err(vfsi_core::api::Error::transport(
405 None,
406 "invalid directory root metadata count",
407 ));
408 }
409 if !metadata[0].is_dir() {
410 return Err(vfsi_core::api::Error::client(0, libc::ENOTDIR as u32)
411 .with_context("visit_dirs", path));
412 }
413 Ok(())
414 },
415 |paths, cursors, page_size, max_entries| {
416 <$client>::read_dir_pages_with_fields(
417 $receiver(self),
418 paths,
419 options.metadata_fields(),
420 cursors,
421 page_size,
422 max_entries,
423 )
424 },
425 callback,
426 )
427 }
428 fn vstream<P: AsRef<Path>>(
429 &self,
430 paths: &[P],
431 options: crate::ReadStreamOptions,
432 mut callback: impl FnMut(usize, u64, &[u8]) -> Result<bool>,
433 ) -> Result<Vec<crate::StreamCompletion>> {
434 let mut output = Vec::new();
435 for (index, path) in paths.iter().enumerate() {
436 let completion = <$client as NativeHooks>::stream_native(
437 $receiver(self),
438 path,
439 options,
440 |offset, data| callback(index, offset, data),
441 )
442 .map_err(|error| vector_index(error, index))?;
443 output.push(completion);
444 if matches!(completion, crate::StreamCompletion::Stopped { .. }) {
445 break;
446 }
447 }
448 Ok(output)
449 }
450 fn vgetattrs<P: AsRef<Path>>(
451 &self,
452 paths: &[P],
453 options: vfsi_core::api::MetadataOptions,
454 ) -> Result<Vec<Metadata>> {
455 ($metadata)($receiver(self), paths, options)
456 }
457 fn limits(&self) -> ResourceLimits {
458 <$client>::limits($receiver(self))
459 }
460
461 fn vopen(&self, requests: &[OpenRequest]) -> Result<Vec<Self::File>> {
462 if requests.len() == 1 {
463 return <$client as NativeHooks>::open_native($receiver(self), requests[0].clone())
465 .map(|file| vec![file]);
466 }
467 <$client>::$open_batch($receiver(self), requests)
468 }
469 fn vread<'a>(
470 &self,
471 ops: impl IntoIterator<Item = vfsi_core::api::ReadOp<'a, Self::File>>,
472 options: vfsi_core::api::ReadOptions,
473 ) -> Result<Vec<vfsi_core::api::ReadResult>> {
474 ($readv)($read_receiver(self), ops, options)
475 }
476 fn vwrite<'a>(
477 &self,
478 requests: &[vfsi_core::api::WriteOp<'a, Self::File>],
479 options: vfsi_core::api::WriteOptions,
480 ) -> Result<Vec<WriteResult>> {
481 let result = if options.writes_all() {
482 ($write_allv)($write_receiver(self), requests)
483 } else {
484 ($writev)($write_receiver(self), requests)
485 };
486 result.map_err(public_write_error)
487 }
488 fn vclose(&self, files: &mut [Self::File]) -> Result<()> {
489 <$client>::$close($receiver(self), files)
490 }
491 fn vmkdir<P: AsRef<Path>>(&self, paths: &[P]) -> Result<()> {
492 <$client>::$mkdir($receiver(self), paths)
493 }
494
495 fn vcopy<P: AsRef<Path>, Q: AsRef<Path>>(&self, pairs: &[(P, Q)]) -> Result<()> {
496 <$client>::$copy($receiver(self), pairs)
497 }
498 fn vremove<P: AsRef<Path>>(
499 &self,
500 paths: &[P],
501 mode: vfsi_core::api::RemoveMode,
502 options: crate::RemoveOptions,
503 ) -> Result<()> {
504 match mode {
505 vfsi_core::api::RemoveMode::Entry | vfsi_core::api::RemoveMode::Tree => {
506 <$client>::remove_paths_with_options(
507 $receiver(self),
508 paths,
509 mode == vfsi_core::api::RemoveMode::Tree,
510 options,
511 )
512 }
513 vfsi_core::api::RemoveMode::Contents => {
514 let mut first_error = None;
515 for (index, path) in paths.iter().enumerate() {
516 if let Err(error) = <$client>::remove_dir_contents_with_options(
517 $receiver(self),
518 path,
519 options,
520 ) {
521 let error = vector_index(error, index);
522 if error.is_transport() || !options.continues_on_error() {
523 return Err(error);
524 }
525 first_error.get_or_insert(error);
526 }
527 }
528 first_error.map_or(Ok(()), Err)
529 }
530 }
531 }
532 };
533}
534
535impl<F: crate::NativeFileSystem + crate::VectorFileSystem + crate::VecFs + 'static> Vfsi
536 for crate::FsClient<F>
537{
538 type File = crate::FsFile<F>;
539 client_methods!(
540 crate::FsClient<F>,
541 std::convert::identity,
542 read_backend::<F>,
543 std::convert::identity,
544 write_backend::<F>,
545 write_backend_all::<F>,
546 metadata_backend::<F, _>,
547 std::convert::identity,
548 renamev,
549 mkdirv,
550 copyv,
551 try_closev,
552 openv
553 );
554}
555
556impl<F: crate::NativeFileSystem + crate::VectorFileSystem + crate::VecFs + 'static> NativeHooks
557 for crate::FsClient<F>
558{
559 fn open_native(&self, request: OpenRequest) -> Result<Self::File> {
560 self.open_with(request)
561 }
562 fn page_capacity(&self, _paths: &[&Path]) -> Result<usize> {
563 self.directory_page_batch_size()
564 }
565 fn stream_native(
566 &self,
567 path: impl AsRef<Path>,
568 options: crate::ReadStreamOptions,
569 callback: impl FnMut(u64, &[u8]) -> Result<bool>,
570 ) -> Result<crate::StreamCompletion> {
571 self.read_stream_with_options(path, options, callback)
572 }
573}
574
575pub(crate) fn read_backend_owned<
576 F: crate::NativeFileSystem + crate::VectorFileSystem + crate::VecFs + 'static,
577>(
578 client: &crate::FsClient<F>,
579 requests: &[ReadRequest<'_, crate::FsRead<'_, F>>],
580 budget: usize,
581) -> Result<Vec<OwnedReadResult>> {
582 if requests.iter().all(|request| request.range_ref().is_some()) {
583 return client.readv_with_limit_projected(requests, budget, |request| {
584 request.range_ref().expect("checked ranges")
585 });
586 }
587 read_batch(
588 requests,
589 budget,
590 |ranges, bytes| client.readv_with_limit_projected(ranges, bytes, |request| request),
591 |paths, bytes| {
592 client
593 .read_files_with_options(paths, crate::ReadAllOptions::new().max_total_bytes(bytes))
594 },
595 )
596}
597pub(crate) fn read_backend<
598 'a,
599 F: crate::NativeFileSystem + crate::VectorFileSystem + crate::VecFs + 'static,
600>(
601 client: &crate::FsClient<F>,
602 ops: impl IntoIterator<Item = ReadOp<'a, crate::FsFile<F>>>,
603 options: ReadOptions,
604) -> Result<Vec<ReadResult>> {
605 consume_ops(
606 ops,
607 options.limit_or(client.limits().max_read_bytes),
608 |requests, options| read_backend_owned(client, requests, options),
609 |requests, bytes| client.readv_into_with_limit(requests, bytes),
610 )
611}
612
613pub(crate) fn public_write_error(error: vfsi_core::api::Error) -> vfsi_core::api::Error {
616 if error.operation() == Some("write_allv")
617 && let Some(path) = error.path().map(std::path::Path::to_path_buf)
618 {
619 error.with_context("writev", path)
620 } else {
621 error
622 }
623}
624
625pub(crate) fn write_backend<'a, F>(
626 client: &crate::FsClient<F>,
627 ops: &[WriteOp<'a, crate::FsFile<F>>],
628) -> vfsi_core::api::Result<Vec<vfsi_core::api::WriteResult>>
629where
630 F: crate::NativeFileSystem + crate::VectorFileSystem + crate::VecFs + 'static,
631{
632 client.writev_mapped(ops, |op| op.file().write_request_at(op.offset(), op.data()))
633}
634
635pub(crate) fn write_backend_all<'a, F>(
636 client: &crate::FsClient<F>,
637 ops: &[WriteOp<'a, crate::FsFile<F>>],
638) -> vfsi_core::api::Result<Vec<vfsi_core::api::WriteResult>>
639where
640 F: crate::NativeFileSystem + crate::VectorFileSystem + crate::VecFs + 'static,
641{
642 client.write_allv_mapped(ops, |op| op.file().write_request_at(op.offset(), op.data()))
643}
644
645pub(crate) fn metadata_backend<F, P: AsRef<std::path::Path>>(
646 client: &crate::FsClient<F>,
647 paths: &[P],
648 options: MetadataOptions,
649) -> vfsi_core::api::Result<Vec<crate::Metadata>>
650where
651 F: crate::NativeFileSystem + crate::VectorFileSystem + crate::VecFs + 'static,
652{
653 client.metadata_many(
654 paths,
655 options.requested_fields(),
656 options.follows_symlinks(),
657 )
658}
659
660use vfsi_core::api::internal::{OwnedReadResult, ReadRequest, consume_ops, read_batch};