1use std::io;
20use std::num::NonZeroUsize;
21use std::sync::Arc;
22
23use sha2::{Digest, Sha256};
25
26use super::backend::ScanPage;
28use super::manager::{Inner, Operation, encoded_size};
29use super::query::FileSelection;
30use super::snapshot::Snapshot;
31use super::{ByteRange, FileBatch, FileCompletion, FileReference, StorageManager};
32use crate::FileReadBatch;
33use crate::local::LocalReader;
34use crate::{BackendFailure, Error, Result};
35use harmos_domain::model::storage::files::{FileHash, FileSource};
36
37#[derive(Clone, Debug)]
46pub struct FileRead {
47 pub source: FileSource,
48 pub range: ByteRange,
49}
50
51pub struct FileReader {
52 context: crate::execution::Context,
53 source: Source,
54}
55
56enum Source {
57 Stored(Box<StoredReader>),
58 Local(LocalReader),
59}
60
61struct StoredReader {
62 inner: Arc<Inner>,
63
64 snapshot: Option<Snapshot>,
65 file: FileReference,
66 offset: u64,
67 end: u64,
68 bytes: u64,
69 verify: bool,
70 hash: Sha256,
71}
72
73impl<C> StorageManager<C> {
80 #[tracing::instrument(name = "storage.file_reads.read_file", level = "debug", skip_all)]
81 pub async fn read_file(&self, request: FileRead) -> Result<FileReader> {
82 let context = crate::execution::Context::current();
83 let span = context.sync(|| tracing::debug_span!("storage.file.read_file", outcome = "cancelled"));
84 context.result(span, async { self.open_file(request).await }).await
85 }
86
87 pub async fn read_opened_file(&self, file: std::fs::File, range: ByteRange) -> Result<FileReader> {
90 self.open_file_handle(file, range).await
91 }
92
93 pub(crate) async fn open_file_handle(&self, file: std::fs::File, range: ByteRange) -> Result<FileReader> {
94 let reader = LocalReader::opened(&self.inner, file, range).await?;
95 Ok(FileReader {
96 context: crate::execution::Context::current(),
97 source: Source::Local(reader),
98 })
99 }
100
101 pub(crate) async fn open_file(&self, request: FileRead) -> Result<FileReader> {
102 let source = match request.source {
103 FileSource::Stored(file) => Source::Stored(Box::new(self.read_stored_file(file, request.range).await?)),
104 FileSource::Local(path) => Source::Local(LocalReader::open(&self.inner, path, request.range).await?),
105 };
106 Ok(FileReader {
107 context: crate::execution::Context::current(),
108 source,
109 })
110 }
111
112 async fn read_stored_file(&self, file: FileReference, range: ByteRange) -> Result<StoredReader> {
113 let end = range.end(file.length)?;
114 self.inner.primary.history().check(&file.position.history)?;
115
116 let (saved, pending) = {
118 let state = self.inner.state.lock().unwrap();
119 if !state.status.accepted.covers(&file.position)? {
120 return Err(invalid("file has not been admitted"));
121 }
122 let pending = state.queue.iter().find_map(|entry| match &entry.operation {
123 Operation::File(actual) if actual.as_slice()[0].position == file.position => Some(actual.as_slice()[0] == file),
124 _ => None,
125 });
126 (state.status.saved.clone(), pending)
127 };
128 if pending == Some(false) {
129 return Err(invalid("immutable metadata does not match a finished file"));
130 }
131
132 let snapshot = self.snapshot(&saved).await?;
133
134 if pending.is_none() {
136 let selection = FileSelection {
137 scope: file.metadata.scope.clone(),
138 item: Some(file.metadata.id.clone()),
139 };
140 let page = ScanPage {
141 after: file.position.sequence.saturating_sub(1),
142 rows: NonZeroUsize::MIN,
143 bytes: NonZeroUsize::new(encoded_size(&file)?.max(1)).unwrap(),
144 };
145 let (stored, _) = snapshot.scan::<FileReference>(&selection, page).await?;
146 if stored.first() != Some(&file) {
147 return Err(invalid("immutable metadata does not match the stored file"));
148 }
149 }
150 snapshot.read_file(&file, range.offset, NonZeroUsize::MIN).await?;
153 Ok(StoredReader {
154 inner: Arc::clone(&self.inner),
155
156 snapshot: Some(snapshot),
157 offset: range.offset,
158 verify: range.offset == 0 && end == file.length,
159 end,
160 file,
161 bytes: 0,
162 hash: Sha256::new(),
163 })
164 }
165}
166
167impl FileReader {
168 pub(crate) fn observation(&self) -> Option<(u64, std::time::SystemTime)> {
169 match &self.source {
170 Source::Local(reader) => Some(reader.observation()),
171 Source::Stored(_) => None,
172 }
173 }
174
175 pub(crate) fn start_opened(&mut self, offset: u64) -> Result<()> {
176 match &mut self.source {
177 Source::Local(reader) => reader.start(offset),
178 Source::Stored(_) => Err(Error::InvalidInput {
179 field: "opened capture",
180 reason: "requires an opened source",
181 }),
182 }
183 }
184
185 pub(crate) async fn observation_unchanged(&self) -> Result<bool> {
186 match &self.source {
187 Source::Local(reader) => reader.observation_unchanged().await,
188 Source::Stored(_) => Ok(true),
189 }
190 }
191
192 pub(crate) fn stored_hash(&self) -> Result<Option<FileHash>> {
193 self.completion()?;
194 Ok(match &self.source {
195 Source::Stored(reader) => Some(FileHash {
196 bytes_read: reader.bytes,
197 sha256: reader.hash.clone().finalize().into(),
198 }),
199 Source::Local(_) => None,
200 })
201 }
202
203 pub(crate) fn is_local(&self) -> bool {
204 matches!(self.source, Source::Local(_))
205 }
206
207 pub fn offset(&self) -> u64 {
208 match &self.source {
209 Source::Stored(reader) => reader.offset,
210 Source::Local(reader) => reader.offset(),
211 }
212 }
213
214 pub async fn next(&mut self) -> Result<Option<FileReadBatch>> {
215 Ok(self.next_raw(false).await?.map(FileReadBatch::new))
216 }
217
218 pub(crate) async fn next_raw(&mut self, event: bool) -> Result<Option<FileBatch>> {
219 let context = self.context.clone();
220 let span = context.sync(|| tracing::trace_span!("storage.reader.page", outcome = "cancelled"));
221 let future = async {
222 match &mut self.source {
223 Source::Stored(reader) => reader.next(event).await,
224 Source::Local(reader) => reader.next(event).await,
225 }
226 };
227 context.result(span, future).await
228 }
229
230 pub(crate) async fn check_unchanged(&self) -> Result<()> {
231 match &self.source {
232 Source::Stored(_) => Ok(()),
233 Source::Local(reader) => reader.check_unchanged().await,
234 }
235 }
236
237 pub fn completion(&self) -> Result<FileCompletion> {
238 match &self.source {
239 Source::Stored(reader) => reader.completion(),
240 Source::Local(reader) => reader.completion(),
241 }
242 }
243}
244
245impl StoredReader {
246 async fn next(&mut self, event: bool) -> Result<Option<FileBatch>> {
247 let Some(snapshot) = &self.snapshot else {
248 return Ok(None);
249 };
250 if self.offset == self.end {
251 if self.verify && self.end == 0 && <[u8; 32]>::from(self.hash.clone().finalize()) != self.file.sha256 {
253 return Err(corrupt("file digest mismatch"));
254 }
255 self.snapshot = None;
256 return Ok(None);
257 }
258
259 let ceiling = self.inner.settings.response.batch_bytes.get();
261 let maximum = crate::file_batch::batch_limit(ceiling, self.offset, self.end - self.offset, event)?;
262
263 let bytes = snapshot.read_file(&self.file, self.offset, maximum).await?;
264 if bytes.is_empty() || bytes.capacity() > maximum.get() || bytes.len() as u64 > self.end - self.offset {
265 return Err(corrupt("file batch is empty or outside the immutable length"));
266 }
267 let mut hash = self.hash.clone();
269 hash.update(&bytes);
270 let next = self.offset + bytes.len() as u64;
271 if self.verify && next == self.end && <[u8; 32]>::from(hash.clone().finalize()) != self.file.sha256 {
272 return Err(corrupt("file digest mismatch"));
273 }
274 let offset = self.offset;
275 self.hash = hash;
276 self.offset = next;
277 self.bytes += bytes.len() as u64;
278 Ok(Some(FileBatch { offset, bytes }))
279 }
280
281 pub fn completion(&self) -> Result<FileCompletion> {
282 if self.offset != self.end || self.snapshot.is_some() {
283 return Err(Error::NotReady { domain: "file read completion" });
284 }
285 Ok(FileCompletion { bytes: self.bytes })
286 }
287}
288
289fn invalid(reason: &'static str) -> Error {
294 Error::InvalidInput { field: "file reference", reason }
295}
296
297fn corrupt(reason: &'static str) -> Error {
298 Error::backend("file read", BackendFailure::Corrupt, io::Error::new(io::ErrorKind::InvalidData, reason))
299}
300
301