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::{ObjectStore, path::Path};
11use prost::Message;
12use tokio::io::AsyncWriteExt;
13use tokio_util::codec::LengthDelimitedCodec;
14use tonic::async_trait;
15use tracing::{Level, instrument, trace, warn};
16use url::Url;
17
18use super::{Directory, DirectoryPutter, DirectoryService, RootToLeavesValidator};
19use crate::composition::{CompositionContext, ServiceBuilder};
20use crate::directoryservice::directory_graph::DirectoryGraphBuilder;
21use crate::{B3Digest, Node, proto};
22
23#[derive(Clone)]
31pub struct ObjectStoreDirectoryService {
32 instance_name: String,
33 object_store: Arc<dyn ObjectStore>,
34 base_path: Path,
35}
36
37#[instrument(level=Level::TRACE, skip_all,fields(base_path=%base_path,blob.digest=%digest),ret(Display))]
38fn derive_dirs_path(base_path: &Path, digest: &B3Digest) -> Path {
39 base_path
40 .child("dirs")
41 .child("b3")
42 .child(HEXLOWER.encode(&digest.as_slice()[..2]))
43 .child(HEXLOWER.encode(digest.as_slice()))
44}
45
46fn parse_proto_directory<F>(
49 encoded_directory: &[u8],
50 digest_allowed: F,
51) -> Result<crate::Directory, Error>
52where
53 F: Fn(&B3Digest) -> bool,
54{
55 let actual_digest = B3Digest::from(blake3::hash(encoded_directory).as_bytes());
56 if !digest_allowed(&actual_digest) {
57 return Err(Error::UnexpectedDigest(actual_digest));
58 }
59
60 let directory_proto =
61 proto::Directory::decode(encoded_directory).map_err(Error::ProtobufDecode)?;
62
63 Directory::try_from(directory_proto).map_err(Error::DirectoryValidation)
64}
65
66#[allow(clippy::identity_op)]
67const MAX_FRAME_LENGTH: usize = 1 * 1024 * 1024 * 1000; impl ObjectStoreDirectoryService {
70 pub fn parse_url_opts<I, K, V>(url: &Url, options: I) -> Result<Self, object_store::Error>
76 where
77 I: IntoIterator<Item = (K, V)>,
78 K: AsRef<str>,
79 V: Into<String>,
80 {
81 let (object_store, path) = object_store::parse_url_opts(url, options)?;
82
83 Ok(Self {
84 instance_name: "root".into(),
85 object_store: Arc::new(object_store),
86 base_path: path,
87 })
88 }
89
90 pub fn parse_url(url: &Url) -> Result<Self, object_store::Error> {
92 Self::parse_url_opts(url, Vec::<(String, String)>::new())
93 }
94
95 pub fn new(instance_name: String, object_store: Arc<dyn ObjectStore>, base_path: Path) -> Self {
96 Self {
97 instance_name,
98 object_store,
99 base_path,
100 }
101 }
102}
103
104#[async_trait]
105impl DirectoryService for ObjectStoreDirectoryService {
106 #[instrument(level = "trace", skip_all, fields(directory.digest = %digest, instance_name = %self.instance_name))]
109 async fn get(&self, digest: &B3Digest) -> Result<Option<Directory>, super::Error> {
110 self.get_recursive(digest).take(1).next().await.transpose()
111 }
112
113 #[instrument(level = "trace", skip_all, fields(directory.digest = %directory.digest(), instance_name = %self.instance_name))]
114 async fn put(&self, directory: Directory) -> Result<B3Digest, super::Error> {
115 if directory
117 .nodes()
118 .any(|(_, e)| matches!(e, Node::Directory { .. }))
119 {
120 Err(Error::PutForDirectoryWithChildren)?
121 }
122
123 let mut handle = self.put_multiple_start();
124 handle.put(directory).await?;
125 handle.close().await
126 }
127
128 #[instrument(level = "trace", skip_all, fields(directory.digest = %root_directory_digest, instance_name = %self.instance_name))]
129 fn get_recursive(
130 &self,
131 root_directory_digest: &B3Digest,
132 ) -> BoxStream<'_, Result<Directory, super::Error>> {
133 let dir_path = derive_dirs_path(&self.base_path, root_directory_digest);
136 let object_store = &self.object_store;
137 let root_directory_digest = *root_directory_digest;
138
139 async_stream::try_stream! {
140 let bytes_stream = match object_store.get(&dir_path).await {
141 Ok(v) => v.into_stream(),
142 Err(object_store::Error::NotFound { .. }) => {
143 return;
144 }
145 Err(e) => Err(Error::ObjectStore(e))?,
146 };
147
148 let r = tokio_util::io::StreamReader::new(bytes_stream);
150 let decompressed_stream = async_compression::tokio::bufread::ZstdDecoder::new(r);
151
152 let mut encoded_directories = LengthDelimitedCodec::builder()
154 .max_frame_length(MAX_FRAME_LENGTH)
155 .length_field_type::<u32>()
156 .new_read(decompressed_stream)
157 .err_into::<Error>();
158
159 let mut order_validator = RootToLeavesValidator::new_with_root_digest(root_directory_digest);
160 while let Some(encoded_directory) = encoded_directories.try_next().await? {
161 let directory = parse_proto_directory(&encoded_directory, |digest| {
162 order_validator.would_accept(digest)
163 })?;
164
165 order_validator.try_accept(&directory).map_err(Error::DirectoryOrdering)?;
166
167 yield directory;
168 }
169
170 order_validator.finalize().map_err(Error::DirectoryOrdering)?;
171 }.boxed()
172 }
173
174 #[instrument(skip_all)]
175 fn put_multiple_start(&self) -> Box<dyn DirectoryPutter + '_>
176 where
177 Self: Clone,
178 {
179 Box::new(ObjectStoreDirectoryPutter::new(
180 self.object_store.clone(),
181 &self.base_path,
182 ))
183 }
184}
185
186#[derive(thiserror::Error, Debug)]
187enum Error {
188 #[error("wrong arguments: {0}")]
189 WrongConfig(&'static str),
190 #[error("put() may only be used for directories without children")]
191 PutForDirectoryWithChildren,
192
193 #[error("Directory Graph ordering error")]
194 DirectoryOrdering(#[from] crate::directoryservice::OrderingError),
195 #[error("requested directory has unexpected digest {0}")]
196 UnexpectedDigest(B3Digest),
197 #[error("failed to decode protobuf: {0}")]
198 ProtobufDecode(#[from] prost::DecodeError),
199 #[error("failed to validate directory: {0}")]
200 DirectoryValidation(#[from] crate::DirectoryError),
201
202 #[error("DirectoryPutter already closed")]
203 DirectoryPutterAlreadyClosed,
204
205 #[error("ObjectStore error: {0}")]
206 ObjectStore(#[from] object_store::Error),
207
208 #[error("io error: {0}")]
209 IO(#[from] std::io::Error),
210}
211
212#[derive(serde::Deserialize)]
213#[serde(deny_unknown_fields)]
214pub struct ObjectStoreDirectoryServiceConfig {
215 object_store_url: String,
216 #[serde(default)]
217 object_store_options: HashMap<String, String>,
218}
219
220impl TryFrom<url::Url> for ObjectStoreDirectoryServiceConfig {
221 type Error = Box<dyn std::error::Error + Send + Sync>;
222 fn try_from(url: url::Url) -> Result<Self, Self::Error> {
223 let trimmed_url = {
226 let s = url.to_string();
227 let mut url = Url::parse(s.strip_prefix("objectstore+").ok_or(Error::WrongConfig(
228 "Missing objectstore+ part in URI scheme",
229 ))?)?;
230 url.set_query(None);
232 url
233 };
234 Ok(ObjectStoreDirectoryServiceConfig {
235 object_store_url: trimmed_url.into(),
236 object_store_options: url
237 .query_pairs()
238 .into_iter()
239 .map(|(k, v)| (k.to_string(), v.to_string()))
240 .collect(),
241 })
242 }
243}
244
245#[async_trait]
246impl ServiceBuilder for ObjectStoreDirectoryServiceConfig {
247 type Output = dyn DirectoryService;
248 async fn build<'a>(
249 &'a self,
250 instance_name: &str,
251 _context: &CompositionContext,
252 ) -> Result<Arc<dyn DirectoryService>, Box<dyn std::error::Error + Send + Sync + 'static>> {
253 let opts = {
254 let mut opts: HashMap<&str, _> = self
255 .object_store_options
256 .iter()
257 .map(|(k, v)| (k.as_str(), v.as_str()))
258 .collect();
259
260 if let hash_map::Entry::Vacant(e) =
261 opts.entry(object_store::ClientConfigKey::UserAgent.as_ref())
262 {
263 e.insert(crate::USER_AGENT);
264 }
265
266 opts
267 };
268
269 let (object_store, path) =
270 object_store::parse_url_opts(&self.object_store_url.parse()?, opts)?;
271 Ok(Arc::new(ObjectStoreDirectoryService::new(
272 instance_name.to_string(),
273 Arc::new(object_store),
274 path,
275 )))
276 }
277}
278
279struct ObjectStoreDirectoryPutter<'a> {
280 object_store: Arc<dyn ObjectStore>,
281 base_path: &'a Path,
282
283 builder: Option<DirectoryGraphBuilder>,
284}
285
286impl<'a> ObjectStoreDirectoryPutter<'a> {
287 fn new(object_store: Arc<dyn ObjectStore>, base_path: &'a Path) -> Self {
288 Self {
289 object_store,
290 base_path,
291 builder: Some(DirectoryGraphBuilder::new_leaves_to_root()),
292 }
293 }
294}
295
296#[async_trait]
297impl DirectoryPutter for ObjectStoreDirectoryPutter<'_> {
298 #[instrument(level = "trace", skip_all, fields(directory.digest=%directory.digest()), err)]
299 async fn put(&mut self, directory: Directory) -> Result<(), super::Error> {
300 let builder = self
301 .builder
302 .as_mut()
303 .ok_or_else(|| Error::DirectoryPutterAlreadyClosed)?;
304
305 builder.try_insert(directory)?;
306
307 Ok(())
308 }
309
310 #[instrument(level = "trace", skip_all, ret, err)]
311 async fn close(&mut self) -> Result<B3Digest, super::Error> {
312 let builder = self
313 .builder
314 .take()
315 .ok_or_else(|| Error::DirectoryPutterAlreadyClosed)?;
316
317 let directory_graph = builder.build()?;
319 let root_digest = directory_graph.root().digest();
320
321 let dir_path = derive_dirs_path(self.base_path, &root_digest);
322
323 match self.object_store.head(&dir_path).await {
324 Ok(_) => {
326 trace!("directory tree already exists");
327 }
328
329 Err(object_store::Error::NotFound { .. }) => {
331 trace!("uploading directory tree");
332
333 let object_store_writer =
334 object_store::buffered::BufWriter::new(self.object_store.clone(), dir_path);
335 let compressed_writer =
336 async_compression::tokio::write::ZstdEncoder::new(object_store_writer);
337 let mut directories_sink = LengthDelimitedCodec::builder()
338 .max_frame_length(MAX_FRAME_LENGTH)
339 .length_field_type::<u32>()
340 .new_write(compressed_writer);
341
342 for directory in directory_graph.drain_root_to_leaves() {
344 directories_sink
345 .send(proto::Directory::from(directory).encode_to_vec().into())
346 .await?;
347 }
348
349 let mut compressed_writer = directories_sink.into_inner();
350 compressed_writer.shutdown().await?;
351 }
352 Err(err) => Err(Error::ObjectStore(err))?,
354 }
355
356 Ok(root_digest)
357 }
358}