use std::io;
use std::num::NonZeroUsize;
use std::sync::Arc;
use sha2::{Digest, Sha256};
use super::backend::ScanPage;
use super::manager::{Inner, Operation, encoded_size};
use super::query::FileSelection;
use super::snapshot::Snapshot;
use super::{ByteRange, FileBatch, FileCompletion, FileReference, StorageManager};
use crate::FileReadBatch;
use crate::local::LocalReader;
use crate::{BackendFailure, Error, Result};
use harmos_domain::model::storage::files::{FileHash, FileSource};
#[derive(Clone, Debug)]
pub struct FileRead {
pub source: FileSource,
pub range: ByteRange,
}
pub struct FileReader {
context: crate::execution::Context,
source: Source,
}
enum Source {
Stored(Box<StoredReader>),
Local(LocalReader),
}
struct StoredReader {
inner: Arc<Inner>,
snapshot: Option<Snapshot>,
file: FileReference,
offset: u64,
end: u64,
bytes: u64,
verify: bool,
hash: Sha256,
}
impl<C> StorageManager<C> {
#[tracing::instrument(name = "storage.file_reads.read_file", level = "debug", skip_all)]
pub async fn read_file(&self, request: FileRead) -> Result<FileReader> {
let context = crate::execution::Context::current();
let span = context.sync(|| tracing::debug_span!("storage.file.read_file", outcome = "cancelled"));
context.result(span, async { self.open_file(request).await }).await
}
pub async fn read_opened_file(&self, file: std::fs::File, range: ByteRange) -> Result<FileReader> {
self.open_file_handle(file, range).await
}
pub(crate) async fn open_file_handle(&self, file: std::fs::File, range: ByteRange) -> Result<FileReader> {
let reader = LocalReader::opened(&self.inner, file, range).await?;
Ok(FileReader {
context: crate::execution::Context::current(),
source: Source::Local(reader),
})
}
pub(crate) async fn open_file(&self, request: FileRead) -> Result<FileReader> {
let source = match request.source {
FileSource::Stored(file) => Source::Stored(Box::new(self.read_stored_file(file, request.range).await?)),
FileSource::Local(path) => Source::Local(LocalReader::open(&self.inner, path, request.range).await?),
};
Ok(FileReader {
context: crate::execution::Context::current(),
source,
})
}
async fn read_stored_file(&self, file: FileReference, range: ByteRange) -> Result<StoredReader> {
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.as_slice()[0].position == file.position => Some(actual.as_slice()[0] == file),
_ => None,
});
(state.status.saved.clone(), pending)
};
if pending == Some(false) {
return Err(invalid("immutable metadata does not match a finished file"));
}
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"));
}
}
snapshot.read_file(&file, range.offset, NonZeroUsize::MIN).await?;
Ok(StoredReader {
inner: Arc::clone(&self.inner),
snapshot: Some(snapshot),
offset: range.offset,
verify: range.offset == 0 && end == file.length,
end,
file,
bytes: 0,
hash: Sha256::new(),
})
}
}
impl FileReader {
pub(crate) fn observation(&self) -> Option<(u64, std::time::SystemTime)> {
match &self.source {
Source::Local(reader) => Some(reader.observation()),
Source::Stored(_) => None,
}
}
pub(crate) fn start_opened(&mut self, offset: u64) -> Result<()> {
match &mut self.source {
Source::Local(reader) => reader.start(offset),
Source::Stored(_) => Err(Error::InvalidInput {
field: "opened capture",
reason: "requires an opened source",
}),
}
}
pub(crate) async fn observation_unchanged(&self) -> Result<bool> {
match &self.source {
Source::Local(reader) => reader.observation_unchanged().await,
Source::Stored(_) => Ok(true),
}
}
pub(crate) fn stored_hash(&self) -> Result<Option<FileHash>> {
self.completion()?;
Ok(match &self.source {
Source::Stored(reader) => Some(FileHash {
bytes_read: reader.bytes,
sha256: reader.hash.clone().finalize().into(),
}),
Source::Local(_) => None,
})
}
pub(crate) fn is_local(&self) -> bool {
matches!(self.source, Source::Local(_))
}
pub fn offset(&self) -> u64 {
match &self.source {
Source::Stored(reader) => reader.offset,
Source::Local(reader) => reader.offset(),
}
}
pub async fn next(&mut self) -> Result<Option<FileReadBatch>> {
Ok(self.next_raw(false).await?.map(FileReadBatch::new))
}
pub(crate) async fn next_raw(&mut self, event: bool) -> Result<Option<FileBatch>> {
let context = self.context.clone();
let span = context.sync(|| tracing::trace_span!("storage.reader.page", outcome = "cancelled"));
let future = async {
match &mut self.source {
Source::Stored(reader) => reader.next(event).await,
Source::Local(reader) => reader.next(event).await,
}
};
context.result(span, future).await
}
pub(crate) async fn check_unchanged(&self) -> Result<()> {
match &self.source {
Source::Stored(_) => Ok(()),
Source::Local(reader) => reader.check_unchanged().await,
}
}
pub fn completion(&self) -> Result<FileCompletion> {
match &self.source {
Source::Stored(reader) => reader.completion(),
Source::Local(reader) => reader.completion(),
}
}
}
impl StoredReader {
async fn next(&mut self, event: bool) -> 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 maximum = crate::file_batch::batch_limit(ceiling, self.offset, self.end - self.offset, event)?;
let bytes = snapshot.read_file(&self.file, self.offset, maximum).await?;
if bytes.is_empty() || bytes.capacity() > 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 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))
}