1use std::io;
20use std::num::NonZeroUsize;
21use std::sync::Arc;
22
23use sha2::{Digest, Sha256};
25
26use super::backend::ScanPage;
28use super::capacity::{Held, Pool, Refusal, Reservation};
29use super::manager::{ENTRY_OVERHEAD, Inner, Operation, encoded_size};
30use super::query::FileSelection;
31use super::snapshot::Snapshot;
32use super::{ByteRange, FileBatch, FileCompletion, FileReference, StorageManager};
33use crate::FileReadBatch;
34use crate::arrow::RetainedMemory;
35use crate::local::LocalReader;
36use crate::{BackendFailure, Error, MemorySize, Result};
37use harmos_domain::model::storage::files::{FileHash, FileSource};
38use harmos_domain::operations::storage::ReadFile;
39
40pub struct FileReader {
49 context: crate::execution::Context,
50 source: Source,
51}
52
53enum Source {
54 Stored(Box<StoredReader>),
55 Local(LocalReader),
56}
57
58struct StoredReader {
59 inner: Arc<Inner>,
60 held: Held,
61 snapshot: Option<Snapshot>,
62 file: FileReference,
63 offset: u64,
64 end: u64,
65 bytes: u64,
66 verify: bool,
67 hash: Sha256,
68 _metadata: RetainedMemory,
69}
70
71impl<C> StorageManager<C> {
78 #[tracing::instrument(name = "storage.file_reads.read_file", level = "debug", skip_all)]
79 pub async fn read_file(&self, request: ReadFile) -> Result<FileReader> {
80 let context = crate::execution::Context::current();
81 let span = context.sync(|| tracing::debug_span!("storage.file.read_file", outcome = "cancelled"));
82 context.result(span, async { self.read_file_with_held(request, Held::default()).await }).await
83 }
84
85 pub(crate) async fn read_file_with_held(&self, request: ReadFile, held: Held) -> Result<FileReader> {
86 let source = match request.source {
87 FileSource::Stored(file) => Source::Stored(Box::new(self.read_stored_file(file, request.range, held).await?)),
88 FileSource::Local(path) => Source::Local(LocalReader::open(&self.inner, path, request.range, held).await?),
89 };
90 Ok(FileReader { context: crate::execution::Context::current(), source })
91 }
92
93 async fn read_stored_file(&self, file: FileReference, range: ByteRange, held: Held) -> Result<StoredReader> {
94 let end = range.end(file.length)?;
95 self.inner.primary.history().check(&file.position.history)?;
96
97 let (saved, pending) = {
99 let state = self.inner.state.lock().unwrap();
100 if !state.status.accepted.covers(&file.position)? {
101 return Err(invalid("file has not been admitted"));
102 }
103 let pending = state.queue.iter().find_map(|entry| match &entry.operation {
104 Operation::File(actual) if actual.as_slice()[0].position == file.position => Some(actual.as_slice()[0] == file),
105 _ => None,
106 });
107 (state.status.saved.clone(), pending)
108 };
109 if pending == Some(false) {
110 return Err(invalid("immutable metadata does not match a finished file"));
111 }
112 let metadata = Reservation::wait(
113 &self.inner,
114 Pool::Read,
115 &held,
116 file.memory_size().saturating_mul(3).saturating_add(std::mem::size_of::<StoredReader>() + ENTRY_OVERHEAD),
117 None,
118 )
119 .await
120 .map_err(Refusal::into_error)?
121 .retain();
122 let snapshot = self.snapshot(&saved, &held).await?;
123
124 if pending.is_none() {
126 let selection = FileSelection { scope: file.metadata.scope.clone(), item: Some(file.metadata.id.clone()) };
127 let page = ScanPage {
128 after: file.position.sequence.saturating_sub(1),
129 rows: NonZeroUsize::MIN,
130 bytes: NonZeroUsize::new(encoded_size(&file)?.max(1)).unwrap(),
131 };
132 let (stored, _) = snapshot.scan::<FileReference>(&selection, page, metadata.clone()).await?;
133 if stored.first() != Some(&file) {
134 return Err(invalid("immutable metadata does not match the stored file"));
135 }
136 }
137 Ok(StoredReader {
138 inner: Arc::clone(&self.inner),
139 held,
140 snapshot: Some(snapshot),
141 offset: range.offset,
142 verify: range.offset == 0 && end == file.length,
143 end,
144 file,
145 bytes: 0,
146 hash: Sha256::new(),
147 _metadata: metadata,
148 })
149 }
150}
151
152impl FileReader {
153 pub(crate) fn stored_hash(&self) -> Result<Option<FileHash>> {
154 self.completion()?;
155 Ok(match &self.source {
156 Source::Stored(reader) => Some(FileHash { bytes_read: reader.bytes, sha256: reader.hash.clone().finalize().into() }),
157 Source::Local(_) => None,
158 })
159 }
160
161 pub(crate) fn is_local(&self) -> bool {
162 matches!(self.source, Source::Local(_))
163 }
164
165 pub fn offset(&self) -> u64 {
166 match &self.source {
167 Source::Stored(reader) => reader.offset,
168 Source::Local(reader) => reader.offset(),
169 }
170 }
171
172 pub async fn next(&mut self) -> Result<Option<FileReadBatch>> {
173 Ok(self.next_raw(false).await?.map(|(batch, memory)| FileReadBatch::retained(batch, memory)))
174 }
175
176 pub(crate) async fn next_raw(&mut self, event: bool) -> Result<Option<(FileBatch, RetainedMemory)>> {
177 let context = self.context.clone();
178 let span = context.sync(|| tracing::trace_span!("storage.reader.page", outcome = "cancelled"));
179 let future = async {
180 match &mut self.source {
181 Source::Stored(reader) => reader.next(event).await,
182 Source::Local(reader) => reader.next(event).await,
183 }
184 };
185 context.result(span, future).await
186 }
187
188 pub(crate) async fn check_unchanged(&self) -> Result<()> {
189 match &self.source {
190 Source::Stored(_) => Ok(()),
191 Source::Local(reader) => reader.check_unchanged().await,
192 }
193 }
194
195 pub fn completion(&self) -> Result<FileCompletion> {
196 match &self.source {
197 Source::Stored(reader) => reader.completion(),
198 Source::Local(reader) => reader.completion(),
199 }
200 }
201}
202
203impl StoredReader {
204 async fn next(&mut self, event: bool) -> Result<Option<(FileBatch, RetainedMemory)>> {
205 let Some(snapshot) = &self.snapshot else {
206 return Ok(None);
207 };
208 if self.offset == self.end {
209 if self.verify && self.end == 0 && <[u8; 32]>::from(self.hash.clone().finalize()) != self.file.sha256 {
211 return Err(corrupt("file digest mismatch"));
212 }
213 self.snapshot = None;
214 return Ok(None);
215 }
216
217 let ceiling = self.inner.settings.response.batch_bytes.get();
219 let maximum = crate::file_batch::batch_limit(ceiling, self.offset, self.end - self.offset, event)?;
220 let charge = maximum.get().saturating_mul(2).saturating_add(std::mem::size_of::<FileBatch>() + ENTRY_OVERHEAD);
221 let transfer = Reservation::wait(&self.inner, Pool::Read, &self.held, charge, None).await.map_err(Refusal::into_error)?.retain();
222 let bytes = snapshot.read_file(&self.file, self.offset, maximum, transfer.clone()).await?;
223 if bytes.is_empty() || bytes.capacity() > maximum.get() || bytes.len() as u64 > self.end - self.offset {
224 return Err(corrupt("file batch is empty or outside the immutable length"));
225 }
226 let mut hash = self.hash.clone();
228 hash.update(&bytes);
229 let next = self.offset + bytes.len() as u64;
230 if self.verify && next == self.end && <[u8; 32]>::from(hash.clone().finalize()) != self.file.sha256 {
231 return Err(corrupt("file digest mismatch"));
232 }
233 let offset = self.offset;
234 self.hash = hash;
235 self.offset = next;
236 self.bytes += bytes.len() as u64;
237 Ok(Some((FileBatch { offset, bytes }, transfer)))
238 }
239
240 pub fn completion(&self) -> Result<FileCompletion> {
241 if self.offset != self.end || self.snapshot.is_some() {
242 return Err(Error::NotReady { domain: "file read completion" });
243 }
244 Ok(FileCompletion { bytes: self.bytes })
245 }
246}
247
248fn invalid(reason: &'static str) -> Error {
253 Error::InvalidInput { field: "file reference", reason }
254}
255
256fn corrupt(reason: &'static str) -> Error {
257 Error::backend("file read", BackendFailure::Corrupt, io::Error::new(io::ErrorKind::InvalidData, reason))
258}
259
260