reth_era_downloader/
client.rs1use alloy_primitives::{hex, hex::ToHexExt};
2use bytes::Bytes;
3use eyre::{eyre, OptionExt};
4use futures_util::{stream::StreamExt, Stream, TryStreamExt};
5use reqwest::{Client, IntoUrl, Url};
6use sha2::{Digest, Sha256};
7use std::{future::Future, path::Path, str::FromStr};
8use tokio::{
9 fs::{self, File},
10 io::{self, AsyncBufReadExt, AsyncRead, AsyncReadExt, AsyncWriteExt},
11 join, try_join,
12};
13
14pub trait HttpClient {
16 fn get<U: IntoUrl + Send + Sync>(
18 &self,
19 url: U,
20 ) -> impl Future<
21 Output = eyre::Result<impl Stream<Item = eyre::Result<Bytes>> + Send + Sync + Unpin>,
22 > + Send
23 + Sync;
24}
25
26impl HttpClient for Client {
27 async fn get<U: IntoUrl + Send + Sync>(
28 &self,
29 url: U,
30 ) -> eyre::Result<impl Stream<Item = eyre::Result<Bytes>> + Unpin> {
31 let response = Self::get(self, url).send().await?;
32
33 Ok(response.bytes_stream().map_err(|e| eyre::Error::new(e)))
34 }
35}
36
37#[derive(Debug, Clone)]
40pub struct EraClient<Http> {
41 client: Http,
42 url: Url,
43 folder: Box<Path>,
44}
45
46impl<Http: HttpClient + Clone> EraClient<Http> {
47 const CHECKSUMS: &'static str = "checksums.txt";
48
49 pub fn new(client: Http, url: Url, folder: impl Into<Box<Path>>) -> Self {
51 Self { client, url, folder: folder.into() }
52 }
53
54 pub async fn download_to_file(&mut self, url: impl IntoUrl) -> eyre::Result<Box<Path>> {
57 let path = self.folder.to_path_buf();
58
59 let url = url.into_url()?;
60 let client = self.client.clone();
61 let file_name = url
62 .path_segments()
63 .ok_or_eyre("cannot-be-a-base")?
64 .next_back()
65 .ok_or_eyre("empty path segments")?;
66 let path = path.join(file_name);
67
68 if !self.is_downloaded(file_name, &path).await? {
69 let number = self
70 .file_name_to_number(file_name)
71 .ok_or_eyre("Cannot parse number from file name")?;
72
73 let mut tries = 1..3;
74 let mut actual_checksum: eyre::Result<_>;
75 loop {
76 actual_checksum = async {
77 let mut file = File::create(&path).await?;
78 let mut stream = client.get(url.clone()).await?;
79 let mut hasher = Sha256::new();
80
81 while let Some(item) = stream.next().await.transpose()? {
82 io::copy(&mut item.as_ref(), &mut file).await?;
83 hasher.update(item);
84 }
85
86 Ok(hasher.finalize().to_vec())
87 }
88 .await;
89
90 if actual_checksum.is_ok() || tries.next().is_none() {
91 break;
92 }
93 }
94
95 self.assert_checksum(number, actual_checksum?)
96 .await
97 .map_err(|e| eyre!("{e} for {file_name} at {}", path.display()))?;
98 }
99
100 Ok(path.into_boxed_path())
101 }
102
103 pub async fn recover_index(&self) -> Option<usize> {
105 let mut max = None;
106
107 if let Ok(mut dir) = fs::read_dir(&self.folder).await {
108 while let Ok(Some(entry)) = dir.next_entry().await {
109 if let Some(name) = entry.file_name().to_str() &&
110 let Some(number) = self.file_name_to_number(name) &&
111 (max.is_none() || matches!(max, Some(max) if number > max))
112 {
113 max.replace(number + 1);
114 }
115 }
116 }
117
118 max
119 }
120
121 pub async fn delete_outside_range(&self, index: usize, max_files: usize) -> eyre::Result<()> {
123 let last = index + max_files;
124
125 if let Ok(mut dir) = fs::read_dir(&self.folder).await {
126 while let Ok(Some(entry)) = dir.next_entry().await {
127 if let Some(name) = entry.file_name().to_str() &&
128 let Some(number) = self.file_name_to_number(name) &&
129 (number < index || number >= last)
130 {
131 eprintln!("Deleting file {}", entry.path().display());
132 eprintln!("{number} < {index} || {number} >= {last}");
133 reth_fs_util::remove_file(entry.path())?;
134 }
135 }
136 }
137
138 Ok(())
139 }
140
141 pub async fn url(&self, number: usize) -> eyre::Result<Option<Url>> {
143 Ok(self.number_to_file_name(number).await?.map(|name| self.url.join(&name)).transpose()?)
144 }
145
146 pub async fn files_count(&self) -> usize {
148 let mut count = 0usize;
149
150 if let Ok(mut dir) = fs::read_dir(&self.folder).await {
151 while let Ok(Some(entry)) = dir.next_entry().await {
152 if entry.path().extension() == Some("era1".as_ref()) {
153 count += 1;
154 }
155 }
156 }
157
158 count
159 }
160
161 pub async fn fetch_file_list(&self) -> eyre::Result<()> {
163 let (mut index, mut checksums) = try_join!(
164 self.client.get(self.url.clone()),
165 self.client.get(self.url.clone().join(Self::CHECKSUMS)?),
166 )?;
167
168 let index_path = self.folder.to_path_buf().join("index.html");
169 let checksums_path = self.folder.to_path_buf().join(Self::CHECKSUMS);
170
171 let (mut index_file, mut checksums_file) =
172 try_join!(File::create(&index_path), File::create(&checksums_path))?;
173
174 loop {
175 let (index, checksums) = join!(index.next(), checksums.next());
176 let (index, checksums) = (index.transpose()?, checksums.transpose()?);
177
178 if index.is_none() && checksums.is_none() {
179 break;
180 }
181 let index_file = &mut index_file;
182 let checksums_file = &mut checksums_file;
183
184 try_join!(
185 async move {
186 if let Some(index) = index {
187 io::copy(&mut index.as_ref(), index_file).await?;
188 }
189 Ok::<(), eyre::Error>(())
190 },
191 async move {
192 if let Some(checksums) = checksums {
193 io::copy(&mut checksums.as_ref(), checksums_file).await?;
194 }
195 Ok::<(), eyre::Error>(())
196 },
197 )?;
198 }
199
200 let file = File::open(&index_path).await?;
201 let reader = io::BufReader::new(file);
202 let mut lines = reader.lines();
203
204 let path = self.folder.to_path_buf().join("index");
205 let file = File::create(&path).await?;
206 let mut writer = io::BufWriter::new(file);
207
208 while let Some(line) = lines.next_line().await? {
209 if let Some(j) = line.find(".era1") &&
210 let Some(i) = line[..j].rfind(|c: char| !c.is_alphanumeric() && c != '-')
211 {
212 let era = &line[i + 1..j + 5];
213 writer.write_all(era.as_bytes()).await?;
214 writer.write_all(b"\n").await?;
215 }
216 }
217 writer.flush().await?;
218
219 Ok(())
220 }
221
222 pub async fn number_to_file_name(&self, number: usize) -> eyre::Result<Option<String>> {
224 let path = self.folder.to_path_buf().join("index");
225 let file = File::open(&path).await?;
226 let reader = io::BufReader::new(file);
227 let mut lines = reader.lines();
228 for _ in 0..number {
229 lines.next_line().await?;
230 }
231
232 Ok(lines.next_line().await?)
233 }
234
235 async fn is_downloaded(&self, name: &str, path: impl AsRef<Path>) -> eyre::Result<bool> {
236 let path = path.as_ref();
237
238 match File::open(path).await {
239 Ok(file) => {
240 let number = self
241 .file_name_to_number(name)
242 .ok_or_else(|| eyre!("Cannot parse ERA number from {name}"))?;
243
244 let actual_checksum = checksum(file).await?;
245 let is_verified = self.verify_checksum(number, actual_checksum).await?;
246
247 if !is_verified {
248 fs::remove_file(path).await?;
249 }
250
251 Ok(is_verified)
252 }
253 Err(e) if e.kind() == io::ErrorKind::NotFound => Ok(false),
254 Err(e) => Err(e)?,
255 }
256 }
257
258 async fn verify_checksum(&self, number: usize, actual_checksum: Vec<u8>) -> eyre::Result<bool> {
263 Ok(actual_checksum == self.expected_checksum(number).await?)
264 }
265
266 async fn assert_checksum(&self, number: usize, actual_checksum: Vec<u8>) -> eyre::Result<()> {
271 let expected_checksum = self.expected_checksum(number).await?;
272
273 if actual_checksum == expected_checksum {
274 Ok(())
275 } else {
276 Err(eyre!(
277 "Checksum mismatch, got: {}, expected: {}",
278 actual_checksum.encode_hex(),
279 expected_checksum.encode_hex()
280 ))
281 }
282 }
283
284 async fn expected_checksum(&self, number: usize) -> eyre::Result<Vec<u8>> {
288 let file = File::open(self.folder.join(Self::CHECKSUMS)).await?;
289 let reader = io::BufReader::new(file);
290 let mut lines = reader.lines();
291
292 for _ in 0..number {
293 lines.next_line().await?;
294 }
295 let expected_checksum =
296 lines.next_line().await?.ok_or_else(|| eyre!("Missing hash for number {number}"))?;
297 let expected_checksum = hex::decode(expected_checksum)?;
298
299 Ok(expected_checksum)
300 }
301
302 fn file_name_to_number(&self, file_name: &str) -> Option<usize> {
303 file_name.split('-').nth(1).and_then(|v| usize::from_str(v).ok())
304 }
305}
306
307async fn checksum(mut reader: impl AsyncRead + Unpin) -> eyre::Result<Vec<u8>> {
308 let mut hasher = Sha256::new();
309
310 let mut data = vec![0; 64 * 1024];
312
313 loop {
314 let len = reader.read(&mut data).await?;
316 if len == 0 {
317 break;
318 } hasher.update(&data[..len]);
322 }
323
324 let hash = hasher.finalize().to_vec();
326
327 Ok(hash)
328}
329
330#[cfg(test)]
331mod tests {
332 use super::*;
333 use std::path::PathBuf;
334 use test_case::test_case;
335
336 impl EraClient<Client> {
337 fn empty() -> Self {
338 Self::new(Client::new(), Url::from_str("file:///").unwrap(), PathBuf::new())
339 }
340 }
341
342 #[test_case("mainnet-00600-a81ae85f.era1", Some(600))]
343 #[test_case("mainnet-00000-a81ae85f.era1", Some(0))]
344 #[test_case("00000-a81ae85f.era1", None)]
345 #[test_case("", None)]
346 fn test_file_name_to_number(file_name: &str, expected_number: Option<usize>) {
347 let client = EraClient::empty();
348
349 let actual_number = client.file_name_to_number(file_name);
350
351 assert_eq!(actual_number, expected_number);
352 }
353}