snix_castore/directoryservice/
object_store.rs1use std::collections::HashMap;
2use std::collections::hash_map;
3use std::sync::Arc;
4
5use data_encoding::HEXLOWER;
6use futures::SinkExt;
7use futures::StreamExt;
8use futures::TryStreamExt;
9use futures::stream::BoxStream;
10use object_store::ObjectStoreExt;
11use object_store::{ObjectStore, path::Path};
12use prost::Message;
13use tokio::io::AsyncWriteExt;
14use tokio_util::codec::LengthDelimitedCodec;
15use tonic::async_trait;
16use tracing::{Level, instrument, trace, warn};
17use url::Url;
18
19use super::{Directory, DirectoryPutter, DirectoryService};
20use crate::composition::{CompositionContext, ServiceBuilder};
21use crate::directoryservice::directory_graph::DirectoryGraphBuilder;
22use crate::directoryservice::order_validator::{self, LeavesToRoot, OrderValidator, RootToLeaves};
23use crate::{B3Digest, Node, proto};
24
25#[derive(Clone)]
33pub struct ObjectStoreDirectoryService {
34 instance_name: String,
35 object_store: Arc<dyn ObjectStore>,
36 base_path: Path,
37}
38
39#[instrument(level=Level::TRACE, skip_all,fields(base_path=%base_path,blob.digest=%digest),ret(Display))]
40fn derive_dirs_path(base_path: &Path, digest: &B3Digest) -> Path {
41 base_path
42 .clone()
43 .join("dirs")
44 .join("b3")
45 .join(HEXLOWER.encode(&digest.as_slice()[..2]))
46 .join(HEXLOWER.encode(digest.as_slice()))
47}
48
49fn parse_proto_directory(encoded_directory: &[u8]) -> Result<crate::Directory, Error> {
51 let directory_proto = proto::Directory::decode(encoded_directory)?;
52
53 Ok(Directory::try_from(directory_proto)?)
54}
55
56#[allow(clippy::identity_op)]
57const MAX_FRAME_LENGTH: usize = 1 * 1024 * 1024 * 1000; impl ObjectStoreDirectoryService {
60 pub fn parse_url_opts<I, K, V>(url: &Url, options: I) -> Result<Self, object_store::Error>
66 where
67 I: IntoIterator<Item = (K, V)>,
68 K: AsRef<str>,
69 V: Into<String>,
70 {
71 let (object_store, path) = object_store::parse_url_opts(url, options)?;
72
73 Ok(Self {
74 instance_name: "root".into(),
75 object_store: Arc::new(object_store),
76 base_path: path,
77 })
78 }
79
80 pub fn parse_url(url: &Url) -> Result<Self, object_store::Error> {
82 Self::parse_url_opts(url, Vec::<(String, String)>::new())
83 }
84
85 pub fn new(instance_name: String, object_store: Arc<dyn ObjectStore>, base_path: Path) -> Self {
86 Self {
87 instance_name,
88 object_store,
89 base_path,
90 }
91 }
92}
93
94#[async_trait]
95impl DirectoryService for ObjectStoreDirectoryService {
96 #[instrument(level = "trace", skip_all, fields(directory.digest = %digest, instance_name = %self.instance_name))]
99 async fn get(&self, digest: &B3Digest) -> Result<Option<Directory>, super::Error> {
100 self.get_recursive(digest).take(1).next().await.transpose()
101 }
102
103 #[instrument(level = "trace", skip_all, fields(directory.digest = %directory.digest(), instance_name = %self.instance_name))]
104 async fn put(&self, directory: Directory) -> Result<B3Digest, super::Error> {
105 if directory
107 .nodes()
108 .any(|(_, e)| matches!(e, Node::Directory { .. }))
109 {
110 Err(Error::PutForDirectoryWithChildren)?
111 }
112
113 let mut handle = self.put_multiple_start();
114 handle.put(directory).await?;
115 handle.close().await
116 }
117
118 #[instrument(level = "trace", skip_all, fields(directory.digest = %root_directory_digest, instance_name = %self.instance_name))]
119 fn get_recursive(
120 &self,
121 root_directory_digest: &B3Digest,
122 ) -> BoxStream<'_, Result<Directory, super::Error>> {
123 let dir_path = derive_dirs_path(&self.base_path, root_directory_digest);
126 let object_store = &self.object_store;
127 let root_directory_digest = *root_directory_digest;
128
129 async_stream::try_stream! {
130 let bytes_stream = match object_store.get(&dir_path).await {
131 Ok(v) => v.into_stream(),
132 Err(object_store::Error::NotFound { .. }) => {
133 return;
134 }
135 Err(e) => Err(Error::ObjectStore(e))?,
136 };
137
138 let r = tokio_util::io::StreamReader::new(bytes_stream);
140 let decompressed_stream = async_compression::tokio::bufread::ZstdDecoder::new(r);
141
142 let mut encoded_directories = LengthDelimitedCodec::builder()
144 .max_frame_length(MAX_FRAME_LENGTH)
145 .length_field_type::<u32>()
146 .new_read(decompressed_stream)
147 .err_into::<Error>();
148
149 let mut order_validator = RootToLeaves::new_with_root_digest(root_directory_digest);
150 while let Some(encoded_directory) = encoded_directories.try_next().await? {
151 let digest = B3Digest::from(blake3::hash(&encoded_directory));
153 if !order_validator.would_accept(&digest) {
154 Err(Error::UnexpectedDigest(digest))?;
155 }
156
157 let directory = parse_proto_directory(&encoded_directory)?;
159
160 order_validator.try_accept(&directory).map_err(Error::DirectoryOrdering)?;
162
163 yield directory;
164 }
165
166 order_validator.finalize().map_err(Error::DirectoryOrdering)?;
167 }
168 .boxed()
169 }
170
171 #[instrument(skip_all)]
172 fn put_multiple_start(&self) -> Box<dyn DirectoryPutter + '_>
173 where
174 Self: Clone,
175 {
176 Box::new(ObjectStoreDirectoryPutter::new(
177 self.object_store.clone(),
178 &self.base_path,
179 ))
180 }
181}
182
183#[derive(thiserror::Error, Debug)]
184enum Error {
185 #[error("wrong arguments: {0}")]
186 WrongConfig(&'static str),
187 #[error("put() may only be used for directories without children")]
188 PutForDirectoryWithChildren,
189
190 #[error("Directory Graph ordering error")]
191 DirectoryOrdering(#[from] order_validator::OrderingError),
192 #[error("next directory in batch has unexpected digest {0}")]
193 UnexpectedDigest(B3Digest),
194 #[error("failed to decode protobuf: {0}")]
195 ProtobufDecode(#[from] prost::DecodeError),
196 #[error("failed to validate directory: {0}")]
197 DirectoryValidation(#[from] crate::DirectoryError),
198
199 #[error("DirectoryPutter already closed")]
200 DirectoryPutterAlreadyClosed,
201
202 #[error("ObjectStore error: {0}")]
203 ObjectStore(#[from] object_store::Error),
204
205 #[error("io error: {0}")]
206 IO(#[from] std::io::Error),
207}
208impl From<Error> for super::Error {
209 fn from(value: Error) -> Self {
210 Self(Box::new(value))
211 }
212}
213
214#[derive(serde::Deserialize)]
215#[serde(deny_unknown_fields)]
216pub struct ObjectStoreDirectoryServiceConfig {
217 object_store_url: String,
218 #[serde(default)]
219 object_store_options: HashMap<String, String>,
220}
221
222impl TryFrom<url::Url> for ObjectStoreDirectoryServiceConfig {
223 type Error = Box<dyn std::error::Error + Send + Sync>;
224 fn try_from(url: url::Url) -> Result<Self, Self::Error> {
225 let trimmed_url = {
228 let s = url.to_string();
229 let mut url = Url::parse(s.strip_prefix("objectstore+").ok_or(Error::WrongConfig(
230 "Missing objectstore+ part in URI scheme",
231 ))?)?;
232 url.set_query(None);
234 url
235 };
236 Ok(ObjectStoreDirectoryServiceConfig {
237 object_store_url: trimmed_url.into(),
238 object_store_options: url
239 .query_pairs()
240 .into_iter()
241 .map(|(k, v)| (k.to_string(), v.to_string()))
242 .collect(),
243 })
244 }
245}
246
247#[async_trait]
248impl ServiceBuilder for ObjectStoreDirectoryServiceConfig {
249 type Output = dyn DirectoryService;
250 async fn build<'a>(
251 &'a self,
252 instance_name: &str,
253 _context: &CompositionContext,
254 ) -> Result<Arc<Self::Output>, Box<dyn std::error::Error + Send + Sync>> {
255 let opts = {
256 let mut opts: HashMap<&str, _> = self
257 .object_store_options
258 .iter()
259 .map(|(k, v)| (k.as_str(), v.as_str()))
260 .collect();
261
262 if let hash_map::Entry::Vacant(e) =
263 opts.entry(object_store::ClientConfigKey::UserAgent.as_ref())
264 {
265 e.insert(crate::USER_AGENT);
266 }
267
268 opts
269 };
270
271 let (object_store, path) =
272 object_store::parse_url_opts(&self.object_store_url.parse()?, opts)?;
273 Ok(Arc::new(ObjectStoreDirectoryService::new(
274 instance_name.to_string(),
275 Arc::new(object_store),
276 path,
277 )))
278 }
279}
280
281struct ObjectStoreDirectoryPutter<'a> {
282 object_store: Arc<dyn ObjectStore>,
283 base_path: &'a Path,
284
285 builder: Option<DirectoryGraphBuilder<LeavesToRoot>>,
286}
287
288impl<'a> ObjectStoreDirectoryPutter<'a> {
289 fn new(object_store: Arc<dyn ObjectStore>, base_path: &'a Path) -> Self {
290 Self {
291 object_store,
292 base_path,
293 builder: Some(DirectoryGraphBuilder::<LeavesToRoot>::new()),
294 }
295 }
296}
297
298#[async_trait]
299impl DirectoryPutter for ObjectStoreDirectoryPutter<'_> {
300 #[instrument(level = "trace", skip_all, fields(directory.digest=%directory.digest()), err)]
301 async fn put(&mut self, directory: Directory) -> Result<(), super::Error> {
302 let builder = self
303 .builder
304 .as_mut()
305 .ok_or_else(|| Error::DirectoryPutterAlreadyClosed)?;
306
307 builder
308 .try_insert(directory)
309 .map_err(Error::DirectoryOrdering)?;
310
311 Ok(())
312 }
313
314 #[instrument(level = "trace", skip_all, ret, err)]
315 async fn close(&mut self) -> Result<B3Digest, super::Error> {
316 let builder = self
317 .builder
318 .take()
319 .ok_or_else(|| Error::DirectoryPutterAlreadyClosed)?;
320
321 let directory_graph = builder.build().map_err(Error::DirectoryOrdering)?;
323 let root_digest = directory_graph.root().digest();
324
325 let dir_path = derive_dirs_path(self.base_path, &root_digest);
326
327 match self.object_store.head(&dir_path).await {
328 Ok(_) => {
330 trace!("directory tree already exists");
331 }
332
333 Err(object_store::Error::NotFound { .. }) => {
335 trace!("uploading directory tree");
336
337 let object_store_writer =
338 object_store::buffered::BufWriter::new(self.object_store.clone(), dir_path);
339 let compressed_writer =
340 async_compression::tokio::write::ZstdEncoder::new(object_store_writer);
341 let mut directories_sink = LengthDelimitedCodec::builder()
342 .max_frame_length(MAX_FRAME_LENGTH)
343 .length_field_type::<u32>()
344 .new_write(compressed_writer);
345
346 for directory in directory_graph.drain_root_to_leaves() {
348 directories_sink
349 .send(proto::Directory::from(directory).encode_to_vec().into())
350 .await
351 .map_err(Error::IO)?;
352 }
353
354 let mut compressed_writer = directories_sink.into_inner();
355 compressed_writer.shutdown().await.map_err(Error::IO)?;
356 }
357 Err(err) => Err(Error::ObjectStore(err))?,
359 }
360
361 Ok(root_digest)
362 }
363}