reth_era_downloader/
client.rs

1use 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
14/// Accesses the network over HTTP.
15pub trait HttpClient {
16    /// Makes an HTTP GET request to `url`. Returns a stream of response body bytes.
17    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/// An HTTP client with features for downloading ERA files from an external HTTP accessible
38/// endpoint.
39#[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    /// Constructs [`EraClient`] using `client` to download from `url` into `folder`.
50    pub fn new(client: Http, url: Url, folder: impl Into<Box<Path>>) -> Self {
51        Self { client, url, folder: folder.into() }
52    }
53
54    /// Performs a GET request on `url` and stores the response body into a file located within
55    /// the `folder`.
56    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    /// Recovers index of file following the latest downloaded file from a different run.
104    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    /// Deletes files that are outside-of the working range.
122    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                    reth_fs_util::remove_file(entry.path())?;
132                }
133            }
134        }
135
136        Ok(())
137    }
138
139    /// Returns a download URL for the file corresponding to `number`.
140    pub async fn url(&self, number: usize) -> eyre::Result<Option<Url>> {
141        Ok(self.number_to_file_name(number).await?.map(|name| self.url.join(&name)).transpose()?)
142    }
143
144    /// Returns the number of files in the `folder`.
145    pub async fn files_count(&self) -> usize {
146        let mut count = 0usize;
147
148        if let Ok(mut dir) = fs::read_dir(&self.folder).await {
149            while let Ok(Some(entry)) = dir.next_entry().await {
150                if entry.path().extension() == Some("era1".as_ref()) {
151                    count += 1;
152                }
153            }
154        }
155
156        count
157    }
158
159    /// Fetches the list of ERA1 files from `url` and stores it in a file located within `folder`.
160    pub async fn fetch_file_list(&self) -> eyre::Result<()> {
161        let (mut index, mut checksums) = try_join!(
162            self.client.get(self.url.clone()),
163            self.client.get(self.url.clone().join(Self::CHECKSUMS)?),
164        )?;
165
166        let index_path = self.folder.to_path_buf().join("index.html");
167        let checksums_path = self.folder.to_path_buf().join(Self::CHECKSUMS);
168
169        let (mut index_file, mut checksums_file) =
170            try_join!(File::create(&index_path), File::create(&checksums_path))?;
171
172        loop {
173            let (index, checksums) = join!(index.next(), checksums.next());
174            let (index, checksums) = (index.transpose()?, checksums.transpose()?);
175
176            if index.is_none() && checksums.is_none() {
177                break;
178            }
179            let index_file = &mut index_file;
180            let checksums_file = &mut checksums_file;
181
182            try_join!(
183                async move {
184                    if let Some(index) = index {
185                        io::copy(&mut index.as_ref(), index_file).await?;
186                    }
187                    Ok::<(), eyre::Error>(())
188                },
189                async move {
190                    if let Some(checksums) = checksums {
191                        io::copy(&mut checksums.as_ref(), checksums_file).await?;
192                    }
193                    Ok::<(), eyre::Error>(())
194                },
195            )?;
196        }
197
198        let file = File::open(&index_path).await?;
199        let reader = io::BufReader::new(file);
200        let mut lines = reader.lines();
201
202        let path = self.folder.to_path_buf().join("index");
203        let file = File::create(&path).await?;
204        let mut writer = io::BufWriter::new(file);
205
206        while let Some(line) = lines.next_line().await? {
207            if let Some(j) = line.find(".era1") &&
208                let Some(i) = line[..j].rfind(|c: char| !c.is_alphanumeric() && c != '-')
209            {
210                let era = &line[i + 1..j + 5];
211                writer.write_all(era.as_bytes()).await?;
212                writer.write_all(b"\n").await?;
213            }
214        }
215        writer.flush().await?;
216
217        Ok(())
218    }
219
220    /// Returns ERA1 file name that is ordered at `number`.
221    pub async fn number_to_file_name(&self, number: usize) -> eyre::Result<Option<String>> {
222        let path = self.folder.to_path_buf().join("index");
223        let file = File::open(&path).await?;
224        let reader = io::BufReader::new(file);
225        let mut lines = reader.lines();
226        for _ in 0..number {
227            lines.next_line().await?;
228        }
229
230        Ok(lines.next_line().await?)
231    }
232
233    async fn is_downloaded(&self, name: &str, path: impl AsRef<Path>) -> eyre::Result<bool> {
234        let path = path.as_ref();
235
236        match File::open(path).await {
237            Ok(file) => {
238                let number = self
239                    .file_name_to_number(name)
240                    .ok_or_else(|| eyre!("Cannot parse ERA number from {name}"))?;
241
242                let actual_checksum = checksum(file).await?;
243                let is_verified = self.verify_checksum(number, actual_checksum).await?;
244
245                if !is_verified {
246                    fs::remove_file(path).await?;
247                }
248
249                Ok(is_verified)
250            }
251            Err(e) if e.kind() == io::ErrorKind::NotFound => Ok(false),
252            Err(e) => Err(e)?,
253        }
254    }
255
256    /// Returns `true` if `actual_checksum` matches expected checksum of the ERA1 file indexed by
257    /// `number` based on the [file list].
258    ///
259    /// [file list]: Self::fetch_file_list
260    async fn verify_checksum(&self, number: usize, actual_checksum: Vec<u8>) -> eyre::Result<bool> {
261        Ok(actual_checksum == self.expected_checksum(number).await?)
262    }
263
264    /// Returns `Ok` if `actual_checksum` matches expected checksum of the ERA1 file indexed by
265    /// `number` based on the [file list].
266    ///
267    /// [file list]: Self::fetch_file_list
268    async fn assert_checksum(&self, number: usize, actual_checksum: Vec<u8>) -> eyre::Result<()> {
269        let expected_checksum = self.expected_checksum(number).await?;
270
271        if actual_checksum == expected_checksum {
272            Ok(())
273        } else {
274            Err(eyre!(
275                "Checksum mismatch, got: {}, expected: {}",
276                actual_checksum.encode_hex(),
277                expected_checksum.encode_hex()
278            ))
279        }
280    }
281
282    /// Returns SHA-256 checksum for ERA1 file indexed by `number` based on the [file list].
283    ///
284    /// [file list]: Self::fetch_file_list
285    async fn expected_checksum(&self, number: usize) -> eyre::Result<Vec<u8>> {
286        let file = File::open(self.folder.join(Self::CHECKSUMS)).await?;
287        let reader = io::BufReader::new(file);
288        let mut lines = reader.lines();
289
290        for _ in 0..number {
291            lines.next_line().await?;
292        }
293        let expected_checksum =
294            lines.next_line().await?.ok_or_else(|| eyre!("Missing hash for number {number}"))?;
295        let expected_checksum = hex::decode(expected_checksum)?;
296
297        Ok(expected_checksum)
298    }
299
300    fn file_name_to_number(&self, file_name: &str) -> Option<usize> {
301        file_name.split('-').nth(1).and_then(|v| usize::from_str(v).ok())
302    }
303}
304
305async fn checksum(mut reader: impl AsyncRead + Unpin) -> eyre::Result<Vec<u8>> {
306    let mut hasher = Sha256::new();
307
308    // Create a buffer to read data into, sized for performance.
309    let mut data = vec![0; 64 * 1024];
310
311    loop {
312        // Read data from the reader into the buffer.
313        let len = reader.read(&mut data).await?;
314        if len == 0 {
315            break;
316        } // Exit loop if no more data.
317
318        // Update the hash with the data read.
319        hasher.update(&data[..len]);
320    }
321
322    // Finalize the hash after all data has been processed.
323    let hash = hasher.finalize().to_vec();
324
325    Ok(hash)
326}
327
328#[cfg(test)]
329mod tests {
330    use super::*;
331    use std::path::PathBuf;
332    use test_case::test_case;
333
334    impl EraClient<Client> {
335        fn empty() -> Self {
336            Self::new(Client::new(), Url::from_str("file:///").unwrap(), PathBuf::new())
337        }
338    }
339
340    #[test_case("mainnet-00600-a81ae85f.era1", Some(600))]
341    #[test_case("mainnet-00000-a81ae85f.era1", Some(0))]
342    #[test_case("00000-a81ae85f.era1", None)]
343    #[test_case("", None)]
344    fn test_file_name_to_number(file_name: &str, expected_number: Option<usize>) {
345        let client = EraClient::empty();
346
347        let actual_number = client.file_name_to_number(file_name);
348
349        assert_eq!(actual_number, expected_number);
350    }
351}