harmos_storage/
file_reads.rs1use 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::{FileBatch, FileCompletion, FileReference, StorageManager};
33use crate::{BackendFailure, Error, Result};
34use harmos_domain::operations::storage::ReadFile;
35
36pub struct FileReader {
45 inner: Arc<Inner>,
46 held: Held,
47 snapshot: Option<Snapshot>,
48 file: FileReference,
49 offset: u64,
50 end: u64,
51 bytes: u64,
52 verify: bool,
53 hash: Sha256,
54 _metadata: Reservation,
55}
56
57impl<C> StorageManager<C> {
64 pub async fn read_file(&self, request: ReadFile) -> Result<FileReader> {
65 let ReadFile { file, range } = request;
66 let end = range.end(file.length)?;
67 self.inner.primary.history().check(&file.position.history)?;
68
69 let (saved, pending) = {
71 let state = self.inner.state.lock().unwrap();
72 if !state.status.accepted.covers(&file.position)? {
73 return Err(invalid("file has not been admitted"));
74 }
75 let pending = state.queue.iter().find_map(|entry| match &entry.operation {
76 Operation::File(actual) if actual.position == file.position => Some(actual == &file),
77 _ => None,
78 });
79 (state.status.saved.clone(), pending)
80 };
81 if pending == Some(false) {
82 return Err(invalid("immutable metadata does not match a finished file"));
83 }
84 let held = Held::default();
85 let metadata =
86 Reservation::wait(&self.inner, Pool::Read, &held, encoded_size(&file)?.saturating_mul(2).saturating_add(ENTRY_OVERHEAD), None)
87 .await
88 .map_err(Refusal::into_error)?;
89 let snapshot = self.snapshot(&saved).await?;
90
91 if pending.is_none() {
93 let selection = FileSelection { scope: file.metadata.scope.clone(), item: Some(file.metadata.id.clone()) };
94 let page = ScanPage {
95 after: file.position.sequence.saturating_sub(1),
96 rows: NonZeroUsize::MIN,
97 bytes: NonZeroUsize::new(encoded_size(&file)?.max(1)).unwrap(),
98 };
99 let (stored, _) = snapshot.scan::<FileReference>(&selection, page).await?;
100 if stored.first() != Some(&file) {
101 return Err(invalid("immutable metadata does not match the stored file"));
102 }
103 }
104 Ok(FileReader {
105 inner: Arc::clone(&self.inner),
106 held,
107 snapshot: Some(snapshot),
108 offset: range.offset,
109 verify: range.offset == 0 && end == file.length,
110 end,
111 file,
112 bytes: 0,
113 hash: Sha256::new(),
114 _metadata: metadata,
115 })
116 }
117}
118
119impl FileReader {
120 pub fn reference(&self) -> &FileReference {
121 &self.file
122 }
123
124 pub fn offset(&self) -> u64 {
125 self.offset
126 }
127
128 pub async fn next(&mut self) -> Result<Option<FileBatch>> {
129 let Some(snapshot) = &self.snapshot else {
130 return Ok(None);
131 };
132 if self.offset == self.end {
133 if self.verify && self.end == 0 && <[u8; 32]>::from(self.hash.clone().finalize()) != self.file.sha256 {
135 return Err(corrupt("file digest mismatch"));
136 }
137 self.snapshot = None;
138 return Ok(None);
139 }
140
141 let ceiling = self.inner.settings.response.batch_bytes.get();
143 let overhead = 2 + varint_bytes(self.offset) + varint_bytes(ceiling as u64);
144 let remaining = usize::try_from(self.end - self.offset).unwrap_or(usize::MAX);
145 let maximum = ceiling
146 .checked_sub(overhead)
147 .map(|bytes| bytes.min(remaining))
148 .and_then(NonZeroUsize::new)
149 .ok_or(Error::ResourceExhausted { resource: "file response batch bytes" })?;
150 let charge = maximum.get().saturating_mul(2).saturating_add(std::mem::size_of::<FileBatch>() + ENTRY_OVERHEAD);
151 let _transfer = Reservation::wait(&self.inner, Pool::Read, &self.held, charge, None).await.map_err(Refusal::into_error)?;
152 let bytes = snapshot.read_file(&self.file, self.offset, maximum).await?;
153 if bytes.is_empty() || bytes.len() > maximum.get() || bytes.len() as u64 > self.end - self.offset {
154 return Err(corrupt("file batch is empty or outside the immutable length"));
155 }
156 let mut hash = self.hash.clone();
158 hash.update(&bytes);
159 let next = self.offset + bytes.len() as u64;
160 if self.verify && next == self.end && <[u8; 32]>::from(hash.clone().finalize()) != self.file.sha256 {
161 return Err(corrupt("file digest mismatch"));
162 }
163 let offset = self.offset;
164 self.hash = hash;
165 self.offset = next;
166 self.bytes += bytes.len() as u64;
167 Ok(Some(FileBatch { offset, bytes }))
168 }
169
170 pub fn completion(&self) -> Result<FileCompletion> {
171 if self.offset != self.end || self.snapshot.is_some() {
172 return Err(Error::NotReady { domain: "file read completion" });
173 }
174 Ok(FileCompletion { bytes: self.bytes })
175 }
176}
177
178fn varint_bytes(mut value: u64) -> usize {
183 let mut bytes = 1;
184 while value >= 128 {
185 value >>= 7;
186 bytes += 1;
187 }
188 bytes
189}
190
191fn invalid(reason: &'static str) -> Error {
192 Error::InvalidInput { field: "file reference", reason }
193}
194
195fn corrupt(reason: &'static str) -> Error {
196 Error::backend("file read", BackendFailure::Corrupt, io::Error::new(io::ErrorKind::InvalidData, reason))
197}
198
199