use std::io;
use std::num::NonZeroUsize;
use std::sync::Arc;
use sha2::{Digest, Sha256};
use super::backend::ScanPage;
use super::capacity::{Held, Pool, Refusal, Reservation};
use super::manager::{ENTRY_OVERHEAD, Inner, Operation, encoded_size};
use super::query::FileSelection;
use super::snapshot::Snapshot;
use super::{FileBatch, FileCompletion, FileReference, StorageManager};
use crate::{BackendFailure, Error, Result};
use harmos_domain::operations::storage::ReadFile;
pub struct FileReader {
inner: Arc<Inner>,
held: Held,
snapshot: Option<Snapshot>,
file: FileReference,
offset: u64,
end: u64,
bytes: u64,
verify: bool,
hash: Sha256,
_metadata: Reservation,
}
impl<C> StorageManager<C> {
pub async fn read_file(&self, request: ReadFile) -> Result<FileReader> {
let ReadFile { file, range } = request;
let end = range.end(file.length)?;
self.inner.primary.history().check(&file.position.history)?;
let (saved, pending) = {
let state = self.inner.state.lock().unwrap();
if !state.status.accepted.covers(&file.position)? {
return Err(invalid("file has not been admitted"));
}
let pending = state.queue.iter().find_map(|entry| match &entry.operation {
Operation::File(actual) if actual.position == file.position => Some(actual == &file),
_ => None,
});
(state.status.saved.clone(), pending)
};
if pending == Some(false) {
return Err(invalid("immutable metadata does not match a finished file"));
}
let held = Held::default();
let metadata =
Reservation::wait(&self.inner, Pool::Read, &held, encoded_size(&file)?.saturating_mul(2).saturating_add(ENTRY_OVERHEAD), None)
.await
.map_err(Refusal::into_error)?;
let snapshot = self.snapshot(&saved).await?;
if pending.is_none() {
let selection = FileSelection { scope: file.metadata.scope.clone(), item: Some(file.metadata.id.clone()) };
let page = ScanPage {
after: file.position.sequence.saturating_sub(1),
rows: NonZeroUsize::MIN,
bytes: NonZeroUsize::new(encoded_size(&file)?.max(1)).unwrap(),
};
let (stored, _) = snapshot.scan::<FileReference>(&selection, page).await?;
if stored.first() != Some(&file) {
return Err(invalid("immutable metadata does not match the stored file"));
}
}
Ok(FileReader {
inner: Arc::clone(&self.inner),
held,
snapshot: Some(snapshot),
offset: range.offset,
verify: range.offset == 0 && end == file.length,
end,
file,
bytes: 0,
hash: Sha256::new(),
_metadata: metadata,
})
}
}
impl FileReader {
pub fn reference(&self) -> &FileReference {
&self.file
}
pub fn offset(&self) -> u64 {
self.offset
}
pub async fn next(&mut self) -> Result<Option<FileBatch>> {
let Some(snapshot) = &self.snapshot else {
return Ok(None);
};
if self.offset == self.end {
if self.verify && self.end == 0 && <[u8; 32]>::from(self.hash.clone().finalize()) != self.file.sha256 {
return Err(corrupt("file digest mismatch"));
}
self.snapshot = None;
return Ok(None);
}
let ceiling = self.inner.settings.response.batch_bytes.get();
let overhead = 2 + varint_bytes(self.offset) + varint_bytes(ceiling as u64);
let remaining = usize::try_from(self.end - self.offset).unwrap_or(usize::MAX);
let maximum = ceiling
.checked_sub(overhead)
.map(|bytes| bytes.min(remaining))
.and_then(NonZeroUsize::new)
.ok_or(Error::ResourceExhausted { resource: "file response batch bytes" })?;
let charge = maximum.get().saturating_mul(2).saturating_add(std::mem::size_of::<FileBatch>() + ENTRY_OVERHEAD);
let _transfer = Reservation::wait(&self.inner, Pool::Read, &self.held, charge, None).await.map_err(Refusal::into_error)?;
let bytes = snapshot.read_file(&self.file, self.offset, maximum).await?;
if bytes.is_empty() || bytes.len() > maximum.get() || bytes.len() as u64 > self.end - self.offset {
return Err(corrupt("file batch is empty or outside the immutable length"));
}
let mut hash = self.hash.clone();
hash.update(&bytes);
let next = self.offset + bytes.len() as u64;
if self.verify && next == self.end && <[u8; 32]>::from(hash.clone().finalize()) != self.file.sha256 {
return Err(corrupt("file digest mismatch"));
}
let offset = self.offset;
self.hash = hash;
self.offset = next;
self.bytes += bytes.len() as u64;
Ok(Some(FileBatch { offset, bytes }))
}
pub fn completion(&self) -> Result<FileCompletion> {
if self.offset != self.end || self.snapshot.is_some() {
return Err(Error::NotReady { domain: "file read completion" });
}
Ok(FileCompletion { bytes: self.bytes })
}
}
fn varint_bytes(mut value: u64) -> usize {
let mut bytes = 1;
while value >= 128 {
value >>= 7;
bytes += 1;
}
bytes
}
fn invalid(reason: &'static str) -> Error {
Error::InvalidInput { field: "file reference", reason }
}
fn corrupt(reason: &'static str) -> Error {
Error::backend("file read", BackendFailure::Corrupt, io::Error::new(io::ErrorKind::InvalidData, reason))
}