snix_castore/directoryservice/
object_store.rs1use std::collections::HashMap;
2use std::collections::hash_map;
3use std::sync::Arc;
4
5use async_stream::try_stream;
6use data_encoding::HEXLOWER;
7use futures::SinkExt;
8use futures::StreamExt;
9use futures::TryFutureExt;
10use futures::TryStreamExt;
11use futures::future::Either;
12use futures::stream::BoxStream;
13use object_store::{ObjectStore, path::Path};
14use prost::Message;
15use tokio::io::AsyncWriteExt;
16use tokio_util::codec::LengthDelimitedCodec;
17use tonic::async_trait;
18use tracing::{Level, instrument, trace, warn};
19use url::Url;
20
21use super::{Directory, DirectoryPutter, DirectoryService, RootToLeavesValidator};
22use crate::composition::{CompositionContext, ServiceBuilder};
23use crate::directoryservice::directory_graph::DirectoryGraphBuilder;
24use crate::directoryservice::directory_graph::DirectoryOrder;
25use crate::{B3Digest, Error, Node, proto};
26
27#[derive(Clone)]
35pub struct ObjectStoreDirectoryService {
36 instance_name: String,
37 object_store: Arc<dyn ObjectStore>,
38 base_path: Path,
39}
40
41#[instrument(level=Level::TRACE, skip_all,fields(base_path=%base_path,blob.digest=%digest),ret(Display))]
42fn derive_dirs_path(base_path: &Path, digest: &B3Digest) -> Path {
43 base_path
44 .child("dirs")
45 .child("b3")
46 .child(HEXLOWER.encode(&digest.as_slice()[..2]))
47 .child(HEXLOWER.encode(digest.as_slice()))
48}
49
50fn parse_proto_directory<F>(
53 encoded_directory: &[u8],
54 digest_allowed: F,
55) -> Result<crate::Directory, Error>
56where
57 F: Fn(&B3Digest) -> bool,
58{
59 let actual_digest = B3Digest::from(blake3::hash(encoded_directory).as_bytes());
60 if !digest_allowed(&actual_digest) {
61 return Err(crate::Error::StorageError(
62 "unexpected directory digest".to_string(),
63 ));
64 }
65
66 let directory_proto = proto::Directory::decode(encoded_directory).map_err(|e| {
67 warn!("unable to parse directory {}: {}", actual_digest, e);
68 Error::StorageError(e.to_string())
69 })?;
70
71 Directory::try_from(directory_proto).map_err(|e| {
72 warn!("unable to convert directory {}: {}", actual_digest, e);
73 Error::StorageError(e.to_string())
74 })
75}
76
77#[allow(clippy::identity_op)]
78const MAX_FRAME_LENGTH: usize = 1 * 1024 * 1024 * 1000; impl ObjectStoreDirectoryService {
81 pub fn parse_url_opts<I, K, V>(url: &Url, options: I) -> Result<Self, object_store::Error>
87 where
88 I: IntoIterator<Item = (K, V)>,
89 K: AsRef<str>,
90 V: Into<String>,
91 {
92 let (object_store, path) = object_store::parse_url_opts(url, options)?;
93
94 Ok(Self {
95 instance_name: "root".into(),
96 object_store: Arc::new(object_store),
97 base_path: path,
98 })
99 }
100
101 pub fn parse_url(url: &Url) -> Result<Self, object_store::Error> {
103 Self::parse_url_opts(url, Vec::<(String, String)>::new())
104 }
105
106 pub fn new(instance_name: String, object_store: Arc<dyn ObjectStore>, base_path: Path) -> Self {
107 Self {
108 instance_name,
109 object_store,
110 base_path,
111 }
112 }
113}
114
115#[async_trait]
116impl DirectoryService for ObjectStoreDirectoryService {
117 #[instrument(level = "trace", skip_all, fields(directory.digest = %digest, instance_name = %self.instance_name))]
120 async fn get(&self, digest: &B3Digest) -> Result<Option<Directory>, Error> {
121 self.get_recursive(digest).take(1).next().await.transpose()
122 }
123
124 #[instrument(level = "trace", skip_all, fields(directory.digest = %directory.digest(), instance_name = %self.instance_name))]
125 async fn put(&self, directory: Directory) -> Result<B3Digest, Error> {
126 if directory
128 .nodes()
129 .any(|(_, e)| matches!(e, Node::Directory { .. }))
130 {
131 return Err(Error::InvalidRequest(
132 "only put_multiple_start is supported by the ObjectStoreDirectoryService for directories with children".into(),
133 ));
134 }
135
136 let mut handle = self.put_multiple_start();
137 handle.put(directory).await?;
138 handle.close().await
139 }
140
141 #[instrument(level = "trace", skip_all, fields(directory.digest = %root_directory_digest, instance_name = %self.instance_name))]
142 fn get_recursive(
143 &self,
144 root_directory_digest: &B3Digest,
145 ) -> BoxStream<'static, Result<Directory, Error>> {
146 let dir_path = derive_dirs_path(&self.base_path, root_directory_digest);
149 let object_store = self.object_store.clone();
150 let root_directory_digest = root_directory_digest.to_owned();
151
152 Box::pin(
153 (async move {
154 let stream = match object_store.get(&dir_path).await {
155 Ok(v) => v.into_stream(),
156 Err(object_store::Error::NotFound { .. }) => {
157 return Ok(Either::Left(futures::stream::empty()));
158 }
159 Err(e) => return Err(std::io::Error::from(e).into()),
160 };
161
162 let r = tokio_util::io::StreamReader::new(stream);
164 let decompressed_stream = async_compression::tokio::bufread::ZstdDecoder::new(r);
165
166 let mut encoded_directories = LengthDelimitedCodec::builder()
168 .max_frame_length(MAX_FRAME_LENGTH)
169 .length_field_type::<u32>()
170 .new_read(decompressed_stream)
171 .err_into::<Error>();
172
173 Ok(Either::Right(try_stream! {
174 let mut order_validator = if let Some(encoded_directory) = encoded_directories.try_next().await? {
175 let directory = parse_proto_directory(&encoded_directory, |digest| {
176 digest == &root_directory_digest
177 })?;
178
179 let order_validator = RootToLeavesValidator::new_with_root(&directory);
180 yield directory;
181 order_validator
182 } else {
183 Err(Error::StorageError("no directories stored".to_string()))?
185 };
186
187 while let Some(encoded_directory) = encoded_directories.try_next().await? {
188 let directory = parse_proto_directory(&encoded_directory, |digest| {
189 order_validator.would_accept(digest)
190 })?;
191
192 order_validator.try_accept(&directory).map_err(|e| Error::StorageError(e.to_string()))?;
193
194 yield directory;
195 }
196
197 order_validator.finalize().map_err(|e| Error::StorageError(e.to_string()))?;
198 }))
199 })
200 .try_flatten_stream(),
201 )
202 }
203
204 #[instrument(skip_all)]
205 fn put_multiple_start(&self) -> Box<dyn DirectoryPutter + '_>
206 where
207 Self: Clone,
208 {
209 Box::new(ObjectStoreDirectoryPutter::new(
210 self.object_store.clone(),
211 &self.base_path,
212 ))
213 }
214}
215
216#[derive(serde::Deserialize)]
217#[serde(deny_unknown_fields)]
218pub struct ObjectStoreDirectoryServiceConfig {
219 object_store_url: String,
220 #[serde(default)]
221 object_store_options: HashMap<String, String>,
222}
223
224impl TryFrom<url::Url> for ObjectStoreDirectoryServiceConfig {
225 type Error = Box<dyn std::error::Error + Send + Sync>;
226 fn try_from(url: url::Url) -> Result<Self, Self::Error> {
227 let trimmed_url = {
230 let s = url.to_string();
231 let mut url = Url::parse(
232 s.strip_prefix("objectstore+")
233 .ok_or(Error::StorageError("Missing objectstore uri".into()))?,
234 )?;
235 url.set_query(None);
237 url
238 };
239 Ok(ObjectStoreDirectoryServiceConfig {
240 object_store_url: trimmed_url.into(),
241 object_store_options: url
242 .query_pairs()
243 .into_iter()
244 .map(|(k, v)| (k.to_string(), v.to_string()))
245 .collect(),
246 })
247 }
248}
249
250#[async_trait]
251impl ServiceBuilder for ObjectStoreDirectoryServiceConfig {
252 type Output = dyn DirectoryService;
253 async fn build<'a>(
254 &'a self,
255 instance_name: &str,
256 _context: &CompositionContext,
257 ) -> Result<Arc<dyn DirectoryService>, Box<dyn std::error::Error + Send + Sync + 'static>> {
258 let opts = {
259 let mut opts: HashMap<&str, _> = self
260 .object_store_options
261 .iter()
262 .map(|(k, v)| (k.as_str(), v.as_str()))
263 .collect();
264
265 if let hash_map::Entry::Vacant(e) =
266 opts.entry(object_store::ClientConfigKey::UserAgent.as_ref())
267 {
268 e.insert(crate::USER_AGENT);
269 }
270
271 opts
272 };
273
274 let (object_store, path) =
275 object_store::parse_url_opts(&self.object_store_url.parse()?, opts)?;
276 Ok(Arc::new(ObjectStoreDirectoryService::new(
277 instance_name.to_string(),
278 Arc::new(object_store),
279 path,
280 )))
281 }
282}
283
284struct ObjectStoreDirectoryPutter<'a> {
285 object_store: Arc<dyn ObjectStore>,
286 base_path: &'a Path,
287
288 builder: Option<DirectoryGraphBuilder>,
289}
290
291impl<'a> ObjectStoreDirectoryPutter<'a> {
292 fn new(object_store: Arc<dyn ObjectStore>, base_path: &'a Path) -> Self {
293 Self {
294 object_store,
295 base_path,
296 builder: Some(DirectoryGraphBuilder::new_with_insertion_order(
297 DirectoryOrder::LeavesToRoot,
298 )),
299 }
300 }
301}
302
303#[async_trait]
304impl DirectoryPutter for ObjectStoreDirectoryPutter<'_> {
305 #[instrument(level = "trace", skip_all, fields(directory.digest=%directory.digest()), err)]
306 async fn put(&mut self, directory: Directory) -> Result<(), Error> {
307 let builder = self
308 .builder
309 .as_mut()
310 .ok_or_else(|| Error::StorageError("already closed".to_string()))?;
311
312 builder.insert(directory)?;
313
314 Ok(())
315 }
316
317 #[instrument(level = "trace", skip_all, ret, err)]
318 async fn close(&mut self) -> Result<B3Digest, Error> {
319 let builder = self
320 .builder
321 .take()
322 .ok_or_else(|| Error::StorageError("already closed".to_string()))?;
323
324 let directory_graph = builder.build()?;
326 let root_digest = directory_graph.root().digest();
327
328 let dir_path = derive_dirs_path(self.base_path, &root_digest);
329
330 match self.object_store.head(&dir_path).await {
331 Ok(_) => {
333 trace!("directory tree already exists");
334 }
335
336 Err(object_store::Error::NotFound { .. }) => {
338 trace!("uploading directory tree");
339
340 let object_store_writer =
341 object_store::buffered::BufWriter::new(self.object_store.clone(), dir_path);
342 let compressed_writer =
343 async_compression::tokio::write::ZstdEncoder::new(object_store_writer);
344 let mut directories_sink = LengthDelimitedCodec::builder()
345 .max_frame_length(MAX_FRAME_LENGTH)
346 .length_field_type::<u32>()
347 .new_write(compressed_writer);
348
349 for directory in directory_graph.drain(DirectoryOrder::RootToLeaves) {
350 directories_sink
351 .send(proto::Directory::from(directory).encode_to_vec().into())
352 .await?;
353 }
354
355 let mut compressed_writer = directories_sink.into_inner();
356 compressed_writer.shutdown().await?;
357 }
358 Err(err) => Err(std::io::Error::from(err))?,
360 }
361
362 Ok(root_digest)
363 }
364}