diff options
| author | Rutger Broekhoff | 2024-01-22 22:52:01 +0100 |
|---|---|---|
| committer | Rutger Broekhoff | 2024-01-22 22:52:01 +0100 |
| commit | f5ff2803af0e03f57ab3093a9384d91abb9de083 (patch) | |
| tree | edb5a5c28c183c421bdf2b3e2c05c36ecb77bacd /rs/server | |
| parent | efe9d55ea6c01c718d3c53dbcee6b48899b48267 (diff) | |
| download | gitolfs3-f5ff2803af0e03f57ab3093a9384d91abb9de083.tar.gz gitolfs3-f5ff2803af0e03f57ab3093a9384d91abb9de083.zip | |
Finish basic implementation of Rust LFS server
Diffstat (limited to 'rs/server')
| -rw-r--r-- | rs/server/Cargo.toml | 1 | ||||
| -rw-r--r-- | rs/server/src/main.rs | 548 |
2 files changed, 442 insertions, 107 deletions
diff --git a/rs/server/Cargo.toml b/rs/server/Cargo.toml index 9a2a9a9..987e154 100644 --- a/rs/server/Cargo.toml +++ b/rs/server/Cargo.toml | |||
| @@ -15,5 +15,6 @@ mime = "0.3" | |||
| 15 | serde = { version = "1", features = ["derive"] } | 15 | serde = { version = "1", features = ["derive"] } |
| 16 | serde_json = "1" | 16 | serde_json = "1" |
| 17 | tokio = { version = "1.35", features = ["full"] } | 17 | tokio = { version = "1.35", features = ["full"] } |
| 18 | tokio-util = "0.7" | ||
| 18 | tower = "0.4" | 19 | tower = "0.4" |
| 19 | tower-service = "0.3" | 20 | tower-service = "0.3" |
diff --git a/rs/server/src/main.rs b/rs/server/src/main.rs index 0266f61..99805a4 100644 --- a/rs/server/src/main.rs +++ b/rs/server/src/main.rs | |||
| @@ -1,7 +1,9 @@ | |||
| 1 | use std::collections::HashMap; | 1 | use std::collections::HashMap; |
| 2 | use std::collections::HashSet; | 2 | use std::collections::HashSet; |
| 3 | use std::process::ExitCode; | ||
| 3 | use std::sync::Arc; | 4 | use std::sync::Arc; |
| 4 | 5 | ||
| 6 | use aws_sdk_s3::error::SdkError; | ||
| 5 | use aws_sdk_s3::operation::head_object::HeadObjectOutput; | 7 | use aws_sdk_s3::operation::head_object::HeadObjectOutput; |
| 6 | use axum::extract::rejection; | 8 | use axum::extract::rejection; |
| 7 | use axum::extract::FromRequest; | 9 | use axum::extract::FromRequest; |
| @@ -15,7 +17,6 @@ use axum::Json; | |||
| 15 | use axum::ServiceExt; | 17 | use axum::ServiceExt; |
| 16 | use base64::prelude::*; | 18 | use base64::prelude::*; |
| 17 | use chrono::DateTime; | 19 | use chrono::DateTime; |
| 18 | use chrono::Duration; | ||
| 19 | use chrono::Utc; | 20 | use chrono::Utc; |
| 20 | use common::HexByte; | 21 | use common::HexByte; |
| 21 | use serde::de; | 22 | use serde::de; |
| @@ -29,12 +30,10 @@ use axum::{ | |||
| 29 | extract::{FromRequestParts, OriginalUri, Request}, | 30 | extract::{FromRequestParts, OriginalUri, Request}, |
| 30 | http::{request::Parts, StatusCode, Uri}, | 31 | http::{request::Parts, StatusCode, Uri}, |
| 31 | response::IntoResponse, | 32 | response::IntoResponse, |
| 32 | routing::{get, post, put}, | 33 | routing::{get, post}, |
| 33 | Extension, Router, | 34 | Extension, Router, |
| 34 | }; | 35 | }; |
| 35 | 36 | ||
| 36 | use serde_json::json; | ||
| 37 | |||
| 38 | #[derive(Clone)] | 37 | #[derive(Clone)] |
| 39 | struct RepositoryName(String); | 38 | struct RepositoryName(String); |
| 40 | 39 | ||
| @@ -65,7 +64,10 @@ async fn rewrite_url<B>( | |||
| 65 | let uri = req.uri(); | 64 | let uri = req.uri(); |
| 66 | let original_uri = OriginalUri(uri.clone()); | 65 | let original_uri = OriginalUri(uri.clone()); |
| 67 | 66 | ||
| 68 | let path_and_query = uri.path_and_query().unwrap(); | 67 | let Some(path_and_query) = uri.path_and_query() else { |
| 68 | // L @ no path & query | ||
| 69 | return Err(StatusCode::BAD_REQUEST); | ||
| 70 | }; | ||
| 69 | let Some((repo, path)) = path_and_query.path().split_once("/info/lfs/objects") else { | 71 | let Some((repo, path)) = path_and_query.path().split_once("/info/lfs/objects") else { |
| 70 | return Err(StatusCode::NOT_FOUND); | 72 | return Err(StatusCode::NOT_FOUND); |
| 71 | }; | 73 | }; |
| @@ -73,7 +75,7 @@ async fn rewrite_url<B>( | |||
| 73 | .trim_start_matches('/') | 75 | .trim_start_matches('/') |
| 74 | .trim_end_matches('/') | 76 | .trim_end_matches('/') |
| 75 | .to_string(); | 77 | .to_string(); |
| 76 | if !path.starts_with("/") || !repo.ends_with(".git") { | 78 | if !path.starts_with('/') || !repo.ends_with(".git") { |
| 77 | return Err(StatusCode::NOT_FOUND); | 79 | return Err(StatusCode::NOT_FOUND); |
| 78 | } | 80 | } |
| 79 | 81 | ||
| @@ -82,7 +84,9 @@ async fn rewrite_url<B>( | |||
| 82 | None => path.try_into().ok(), | 84 | None => path.try_into().ok(), |
| 83 | Some(q) => format!("{path}?{q}").try_into().ok(), | 85 | Some(q) => format!("{path}?{q}").try_into().ok(), |
| 84 | }; | 86 | }; |
| 85 | let new_uri = Uri::from_parts(parts).unwrap(); | 87 | let Ok(new_uri) = Uri::from_parts(parts) else { |
| 88 | return Err(StatusCode::INTERNAL_SERVER_ERROR); | ||
| 89 | }; | ||
| 86 | 90 | ||
| 87 | *req.uri_mut() = new_uri; | 91 | *req.uri_mut() = new_uri; |
| 88 | req.extensions_mut().insert(original_uri); | 92 | req.extensions_mut().insert(original_uri); |
| @@ -95,11 +99,45 @@ struct AppState { | |||
| 95 | s3_client: aws_sdk_s3::Client, | 99 | s3_client: aws_sdk_s3::Client, |
| 96 | s3_bucket: String, | 100 | s3_bucket: String, |
| 97 | authz_conf: AuthorizationConfig, | 101 | authz_conf: AuthorizationConfig, |
| 102 | // Should not end with a slash. | ||
| 103 | base_url: String, | ||
| 104 | } | ||
| 105 | |||
| 106 | struct Env { | ||
| 107 | s3_access_key_id: String, | ||
| 108 | s3_secret_access_key: String, | ||
| 109 | s3_bucket: String, | ||
| 110 | s3_endpoint: String, | ||
| 111 | base_url: String, | ||
| 112 | key_path: String, | ||
| 113 | listen_host: String, | ||
| 114 | listen_port: String, | ||
| 115 | trusted_forwarded_hosts: String, | ||
| 116 | } | ||
| 117 | |||
| 118 | fn require_env(name: &str) -> Result<String, String> { | ||
| 119 | std::env::var(name).map_err(|_| format!("environment variable {name} should be defined and valid")) | ||
| 120 | } | ||
| 121 | |||
| 122 | impl Env { | ||
| 123 | fn load() -> Result<Env, String> { | ||
| 124 | Ok(Env { | ||
| 125 | s3_secret_access_key: require_env("GITOLFS3_S3_ACCESS_KEY_FILE")?, | ||
| 126 | s3_access_key_id: require_env("GITOLFS3_S3_ACCESS_KEY_ID_FILE")?, | ||
| 127 | s3_endpoint: require_env("GITOLFS3_S3_ENDPOINT")?, | ||
| 128 | s3_bucket: require_env("GITOLFS3_S3_BUCKET")?, | ||
| 129 | base_url: require_env("GITOLFS3_BASE_URL")?, | ||
| 130 | key_path: require_env("GITOLFS3_KEY_PATH")?, | ||
| 131 | listen_host: require_env("GITOLFS3_LISTEN_HOST")?, | ||
| 132 | listen_port: require_env("GITOLFS3_LISTEN_PORT")?, | ||
| 133 | trusted_forwarded_hosts: std::env::var("GITOLFS3_TRUSTED_FORWARDED_HOSTS").unwrap_or_default(), | ||
| 134 | }) | ||
| 135 | } | ||
| 98 | } | 136 | } |
| 99 | 137 | ||
| 100 | fn get_s3_client() -> aws_sdk_s3::Client { | 138 | fn get_s3_client(env: &Env) -> Result<aws_sdk_s3::Client, std::io::Error> { |
| 101 | let access_key_id = std::env::var("S3_ACCESS_KEY_ID").unwrap(); | 139 | let access_key_id = std::fs::read_to_string(&env.s3_access_key_id)?; |
| 102 | let secret_access_key = std::env::var("S3_SECRET_ACCESS_KEY").unwrap(); | 140 | let secret_access_key = std::fs::read_to_string(&env.s3_secret_access_key)?; |
| 103 | 141 | ||
| 104 | let credentials = aws_sdk_s3::config::Credentials::new( | 142 | let credentials = aws_sdk_s3::config::Credentials::new( |
| 105 | access_key_id, | 143 | access_key_id, |
| @@ -109,51 +147,85 @@ fn get_s3_client() -> aws_sdk_s3::Client { | |||
| 109 | "gitolfs3-env", | 147 | "gitolfs3-env", |
| 110 | ); | 148 | ); |
| 111 | let config = aws_config::SdkConfig::builder() | 149 | let config = aws_config::SdkConfig::builder() |
| 112 | .endpoint_url(std::env::var("S3_ENDPOINT").unwrap()) | 150 | .endpoint_url(&env.s3_endpoint) |
| 113 | .credentials_provider(aws_sdk_s3::config::SharedCredentialsProvider::new( | 151 | .credentials_provider(aws_sdk_s3::config::SharedCredentialsProvider::new( |
| 114 | credentials, | 152 | credentials, |
| 115 | )) | 153 | )) |
| 116 | .build(); | 154 | .build(); |
| 117 | aws_sdk_s3::Client::new(&config) | 155 | Ok(aws_sdk_s3::Client::new(&config)) |
| 118 | } | 156 | } |
| 119 | 157 | ||
| 120 | #[tokio::main] | 158 | #[tokio::main] |
| 121 | async fn main() { | 159 | async fn main() -> ExitCode { |
| 122 | // run our app with hyper, listening globally on port 3000 | 160 | let env = match Env::load() { |
| 123 | let listener = tokio::net::TcpListener::bind("0.0.0.0:3000").await.unwrap(); | 161 | Ok(env) => env, |
| 124 | 162 | Err(e) => { | |
| 125 | let key_path = std::env::var("GITOLFS3_KEY_PATH").unwrap(); | 163 | println!("Failed to load configuration: {e}"); |
| 126 | let key = common::load_key(&key_path).unwrap(); | 164 | return ExitCode::from(2); |
| 127 | let trusted_forwarded_hosts = std::env::var("GITOLFS3_TRUSTED_FORWARDED_HOSTS").unwrap(); | 165 | } |
| 128 | let trusted_forwarded_hosts: HashSet<String> = trusted_forwarded_hosts | 166 | }; |
| 167 | |||
| 168 | let s3_client = match get_s3_client(&env) { | ||
| 169 | Ok(s3_client) => s3_client, | ||
| 170 | Err(e) => { | ||
| 171 | println!("Failed to create S3 client: {e}"); | ||
| 172 | return ExitCode::FAILURE; | ||
| 173 | }, | ||
| 174 | }; | ||
| 175 | let key = match common::load_key(&env.key_path) { | ||
| 176 | Ok(key) => key, | ||
| 177 | Err(e) => { | ||
| 178 | println!("Failed to load Gitolfs3 key: {e}"); | ||
| 179 | return ExitCode::FAILURE; | ||
| 180 | } | ||
| 181 | }; | ||
| 182 | |||
| 183 | let trusted_forwarded_hosts: HashSet<String> = env.trusted_forwarded_hosts | ||
| 129 | .split(',') | 184 | .split(',') |
| 130 | .map(|s| s.to_owned()) | 185 | .map(|s| s.to_owned()) |
| 186 | .filter(|s| !s.is_empty()) | ||
| 131 | .collect(); | 187 | .collect(); |
| 188 | let base_url = env.base_url.trim_end_matches('/').to_string(); | ||
| 132 | 189 | ||
| 133 | let authz_conf = AuthorizationConfig { | 190 | let authz_conf = AuthorizationConfig { |
| 134 | key, | 191 | key, |
| 135 | trusted_forwarded_hosts, | 192 | trusted_forwarded_hosts, |
| 136 | }; | 193 | }; |
| 137 | 194 | ||
| 138 | let s3_client = get_s3_client(); | ||
| 139 | let s3_bucket = std::env::var("S3_BUCKET").unwrap(); | ||
| 140 | let shared_state = Arc::new(AppState { | 195 | let shared_state = Arc::new(AppState { |
| 141 | s3_client, | 196 | s3_client, |
| 142 | s3_bucket, | 197 | s3_bucket: env.s3_bucket, |
| 143 | authz_conf, | 198 | authz_conf, |
| 199 | base_url, | ||
| 144 | }); | 200 | }); |
| 145 | let app = Router::new() | 201 | let app = Router::new() |
| 146 | .route("/batch", post(batch)) | 202 | .route("/batch", post(batch)) |
| 147 | .route("/:oid0/:oid1/:oid", get(obj_download)) | 203 | .route("/:oid0/:oid1/:oid", get(obj_download)) |
| 148 | .route("/:oid0/:oid1/:oid", put(obj_upload)) | ||
| 149 | .with_state(shared_state); | 204 | .with_state(shared_state); |
| 150 | 205 | ||
| 151 | let middleware = axum::middleware::map_request(rewrite_url); | 206 | let middleware = axum::middleware::map_request(rewrite_url); |
| 152 | let app_with_middleware = middleware.layer(app); | 207 | let app_with_middleware = middleware.layer(app); |
| 153 | 208 | ||
| 154 | axum::serve(listener, app_with_middleware.into_make_service()) | 209 | let Ok(listen_port): Result<u16, _> = env.listen_port.parse() else { |
| 155 | .await | 210 | println!("Configured LISTEN_PORT should be an unsigned integer no higher than 65535"); |
| 156 | .unwrap(); | 211 | return ExitCode::from(2); |
| 212 | }; | ||
| 213 | let addr: (String, u16) = (env.listen_host, listen_port); | ||
| 214 | let listener = match tokio::net::TcpListener::bind(addr).await { | ||
| 215 | Ok(listener) => listener, | ||
| 216 | Err(e) => { | ||
| 217 | println!("Failed to listen: {e}"); | ||
| 218 | return ExitCode::FAILURE; | ||
| 219 | } | ||
| 220 | }; | ||
| 221 | |||
| 222 | match axum::serve(listener, app_with_middleware.into_make_service()).await { | ||
| 223 | Ok(_) => ExitCode::SUCCESS, | ||
| 224 | Err(e) => { | ||
| 225 | println!("Error serving: {e}"); | ||
| 226 | ExitCode::FAILURE | ||
| 227 | } | ||
| 228 | } | ||
| 157 | } | 229 | } |
| 158 | 230 | ||
| 159 | #[derive(Debug, Serialize, Deserialize, PartialEq, Eq, Clone, Copy)] | 231 | #[derive(Debug, Serialize, Deserialize, PartialEq, Eq, Clone, Copy)] |
| @@ -174,11 +246,9 @@ impl Default for HashAlgo { | |||
| 174 | } | 246 | } |
| 175 | } | 247 | } |
| 176 | 248 | ||
| 177 | type Oid = common::Digest<32>; | ||
| 178 | |||
| 179 | #[derive(Debug, Deserialize, Clone)] | 249 | #[derive(Debug, Deserialize, Clone)] |
| 180 | struct BatchRequestObject { | 250 | struct BatchRequestObject { |
| 181 | oid: Oid, | 251 | oid: common::Oid, |
| 182 | size: i64, | 252 | size: i64, |
| 183 | } | 253 | } |
| 184 | 254 | ||
| @@ -196,8 +266,6 @@ struct BatchRequest { | |||
| 196 | operation: common::Operation, | 266 | operation: common::Operation, |
| 197 | #[serde(default = "default_transfers")] | 267 | #[serde(default = "default_transfers")] |
| 198 | transfers: Vec<TransferAdapter>, | 268 | transfers: Vec<TransferAdapter>, |
| 199 | #[serde(rename = "ref")] | ||
| 200 | reference: Option<BatchRef>, | ||
| 201 | objects: Vec<BatchRequestObject>, | 269 | objects: Vec<BatchRequestObject>, |
| 202 | #[serde(default)] | 270 | #[serde(default)] |
| 203 | hash_algo: HashAlgo, | 271 | hash_algo: HashAlgo, |
| @@ -206,7 +274,7 @@ struct BatchRequest { | |||
| 206 | #[derive(Clone)] | 274 | #[derive(Clone)] |
| 207 | struct GitLfsJson<T>(Json<T>); | 275 | struct GitLfsJson<T>(Json<T>); |
| 208 | 276 | ||
| 209 | const LFS_MIME: &'static str = "application/vnd.git-lfs+json"; | 277 | const LFS_MIME: &str = "application/vnd.git-lfs+json"; |
| 210 | 278 | ||
| 211 | enum GitLfsJsonRejection { | 279 | enum GitLfsJsonRejection { |
| 212 | Json(rejection::JsonRejection), | 280 | Json(rejection::JsonRejection), |
| @@ -246,7 +314,7 @@ fn has_git_lfs_json_content_type(req: &Request) -> bool { | |||
| 246 | let Ok(content_type) = content_type.to_str() else { | 314 | let Ok(content_type) = content_type.to_str() else { |
| 247 | return false; | 315 | return false; |
| 248 | }; | 316 | }; |
| 249 | return is_git_lfs_json_mimetype(content_type); | 317 | is_git_lfs_json_mimetype(content_type) |
| 250 | } | 318 | } |
| 251 | 319 | ||
| 252 | #[async_trait] | 320 | #[async_trait] |
| @@ -287,7 +355,7 @@ struct GitLfsErrorData<'a> { | |||
| 287 | 355 | ||
| 288 | type GitLfsErrorResponse<'a> = (StatusCode, GitLfsJson<GitLfsErrorData<'a>>); | 356 | type GitLfsErrorResponse<'a> = (StatusCode, GitLfsJson<GitLfsErrorData<'a>>); |
| 289 | 357 | ||
| 290 | const fn make_error_resp<'a>(code: StatusCode, message: &'a str) -> GitLfsErrorResponse { | 358 | const fn make_error_resp(code: StatusCode, message: &str) -> GitLfsErrorResponse { |
| 291 | (code, GitLfsJson(Json(GitLfsErrorData { message }))) | 359 | (code, GitLfsJson(Json(GitLfsErrorData { message }))) |
| 292 | } | 360 | } |
| 293 | 361 | ||
| @@ -309,13 +377,36 @@ struct BatchResponseObjectActions { | |||
| 309 | verify: Option<BatchResponseObjectAction>, | 377 | verify: Option<BatchResponseObjectAction>, |
| 310 | } | 378 | } |
| 311 | 379 | ||
| 380 | #[derive(Debug, Clone, Serialize)] | ||
| 381 | struct BatchResponseObjectError { | ||
| 382 | code: u16, | ||
| 383 | message: String, | ||
| 384 | } | ||
| 385 | |||
| 312 | #[derive(Debug, Serialize, Clone)] | 386 | #[derive(Debug, Serialize, Clone)] |
| 313 | struct BatchResponseObject { | 387 | struct BatchResponseObject { |
| 314 | oid: Oid, | 388 | oid: common::Oid, |
| 315 | size: i64, | 389 | size: i64, |
| 316 | #[serde(skip_serializing_if = "Option::is_none")] | 390 | #[serde(skip_serializing_if = "Option::is_none")] |
| 317 | authenticated: Option<bool>, | 391 | authenticated: Option<bool>, |
| 318 | actions: BatchResponseObjectActions, | 392 | actions: BatchResponseObjectActions, |
| 393 | #[serde(skip_serializing_if = "Option::is_none")] | ||
| 394 | error: Option<BatchResponseObjectError>, | ||
| 395 | } | ||
| 396 | |||
| 397 | impl BatchResponseObject { | ||
| 398 | fn error(obj: &BatchRequestObject, code: StatusCode, message: String) -> BatchResponseObject { | ||
| 399 | BatchResponseObject { | ||
| 400 | oid: obj.oid, | ||
| 401 | size: obj.size, | ||
| 402 | authenticated: None, | ||
| 403 | actions: Default::default(), | ||
| 404 | error: Some(BatchResponseObjectError { | ||
| 405 | code: code.as_u16(), | ||
| 406 | message, | ||
| 407 | }), | ||
| 408 | } | ||
| 409 | } | ||
| 319 | } | 410 | } |
| 320 | 411 | ||
| 321 | #[derive(Debug, Serialize, Clone)] | 412 | #[derive(Debug, Serialize, Clone)] |
| @@ -325,11 +416,11 @@ struct BatchResponse { | |||
| 325 | hash_algo: HashAlgo, | 416 | hash_algo: HashAlgo, |
| 326 | } | 417 | } |
| 327 | 418 | ||
| 328 | fn validate_checksum(oid: Oid, obj: &HeadObjectOutput) -> bool { | 419 | fn validate_checksum(oid: common::Oid, obj: &HeadObjectOutput) -> bool { |
| 329 | if let Some(checksum) = obj.checksum_sha256() { | 420 | if let Some(checksum) = obj.checksum_sha256() { |
| 330 | if let Ok(checksum) = BASE64_STANDARD.decode(checksum) { | 421 | if let Ok(checksum) = BASE64_STANDARD.decode(checksum) { |
| 331 | if let Ok(checksum32b) = TryInto::<[u8; 32]>::try_into(checksum) { | 422 | if let Ok(checksum32b) = TryInto::<[u8; 32]>::try_into(checksum) { |
| 332 | return Oid::from(checksum32b) == oid; | 423 | return common::Oid::from(checksum32b) == oid; |
| 333 | } | 424 | } |
| 334 | } | 425 | } |
| 335 | } | 426 | } |
| @@ -343,11 +434,15 @@ fn validate_size(expected: i64, obj: &HeadObjectOutput) -> bool { | |||
| 343 | true | 434 | true |
| 344 | } | 435 | } |
| 345 | 436 | ||
| 346 | async fn handle_download_object(state: &AppState, repo: &str, obj: &BatchRequestObject, trusted: bool) -> BatchResponseObject { | 437 | async fn handle_upload_object( |
| 438 | state: &AppState, | ||
| 439 | repo: &str, | ||
| 440 | obj: &BatchRequestObject, | ||
| 441 | ) -> Option<BatchResponseObject> { | ||
| 347 | let (oid0, oid1) = (HexByte(obj.oid[0]), HexByte(obj.oid[1])); | 442 | let (oid0, oid1) = (HexByte(obj.oid[0]), HexByte(obj.oid[1])); |
| 348 | let full_path = format!("{repo}/lfs/objects/{}/{}/{}", oid0, oid1, obj.oid); | 443 | let full_path = format!("{repo}/lfs/objects/{}/{}/{}", oid0, oid1, obj.oid); |
| 349 | 444 | ||
| 350 | let result = state | 445 | match state |
| 351 | .s3_client | 446 | .s3_client |
| 352 | .head_object() | 447 | .head_object() |
| 353 | .bucket(&state.s3_bucket) | 448 | .bucket(&state.s3_bucket) |
| @@ -355,36 +450,189 @@ async fn handle_download_object(state: &AppState, repo: &str, obj: &BatchRequest | |||
| 355 | .checksum_mode(aws_sdk_s3::types::ChecksumMode::Enabled) | 450 | .checksum_mode(aws_sdk_s3::types::ChecksumMode::Enabled) |
| 356 | .send() | 451 | .send() |
| 357 | .await | 452 | .await |
| 358 | .unwrap(); // TODO: don't unwrap() | 453 | { |
| 454 | Ok(result) => { | ||
| 455 | if validate_size(obj.size, &result) && validate_checksum(obj.oid, &result) { | ||
| 456 | return None; | ||
| 457 | } | ||
| 458 | } | ||
| 459 | Err(SdkError::ServiceError(e)) if e.err().is_not_found() => {} | ||
| 460 | _ => { | ||
| 461 | return Some(BatchResponseObject::error( | ||
| 462 | obj, | ||
| 463 | StatusCode::INTERNAL_SERVER_ERROR, | ||
| 464 | "Failed to query object information".to_string(), | ||
| 465 | )); | ||
| 466 | } | ||
| 467 | }; | ||
| 468 | |||
| 469 | let expires_in = std::time::Duration::from_secs(5 * 60); | ||
| 470 | let expires_at = Utc::now() + expires_in; | ||
| 471 | |||
| 472 | let Ok(config) = aws_sdk_s3::presigning::PresigningConfig::expires_in(expires_in) else { | ||
| 473 | return Some(BatchResponseObject::error( | ||
| 474 | obj, | ||
| 475 | StatusCode::INTERNAL_SERVER_ERROR, | ||
| 476 | "Failed to generate upload URL".to_string(), | ||
| 477 | )); | ||
| 478 | }; | ||
| 479 | let Ok(presigned) = state | ||
| 480 | .s3_client | ||
| 481 | .put_object() | ||
| 482 | .checksum_sha256(obj.oid.to_string()) | ||
| 483 | .content_length(obj.size) | ||
| 484 | .presigned(config) | ||
| 485 | .await | ||
| 486 | else { | ||
| 487 | return Some(BatchResponseObject::error( | ||
| 488 | obj, | ||
| 489 | StatusCode::INTERNAL_SERVER_ERROR, | ||
| 490 | "Failed to generate upload URL".to_string(), | ||
| 491 | )); | ||
| 492 | }; | ||
| 493 | Some(BatchResponseObject { | ||
| 494 | oid: obj.oid, | ||
| 495 | size: obj.size, | ||
| 496 | authenticated: Some(true), | ||
| 497 | actions: BatchResponseObjectActions { | ||
| 498 | download: Some(BatchResponseObjectAction { | ||
| 499 | header: presigned | ||
| 500 | .headers() | ||
| 501 | .map(|(k, v)| (k.to_owned(), v.to_owned())) | ||
| 502 | .collect(), | ||
| 503 | expires_at, | ||
| 504 | href: presigned.uri().to_string(), | ||
| 505 | }), | ||
| 506 | ..Default::default() | ||
| 507 | }, | ||
| 508 | error: None, | ||
| 509 | }) | ||
| 510 | } | ||
| 511 | |||
| 512 | async fn handle_download_object( | ||
| 513 | state: &AppState, | ||
| 514 | repo: &str, | ||
| 515 | obj: &BatchRequestObject, | ||
| 516 | trusted: bool, | ||
| 517 | ) -> BatchResponseObject { | ||
| 518 | let (oid0, oid1) = (HexByte(obj.oid[0]), HexByte(obj.oid[1])); | ||
| 519 | let full_path = format!("{repo}/lfs/objects/{}/{}/{}", oid0, oid1, obj.oid); | ||
| 520 | |||
| 521 | let result = match state | ||
| 522 | .s3_client | ||
| 523 | .head_object() | ||
| 524 | .bucket(&state.s3_bucket) | ||
| 525 | .key(full_path) | ||
| 526 | .checksum_mode(aws_sdk_s3::types::ChecksumMode::Enabled) | ||
| 527 | .send() | ||
| 528 | .await | ||
| 529 | { | ||
| 530 | Ok(result) => result, | ||
| 531 | Err(_) => { | ||
| 532 | return BatchResponseObject::error( | ||
| 533 | obj, | ||
| 534 | StatusCode::INTERNAL_SERVER_ERROR, | ||
| 535 | "Failed to query object information".to_string(), | ||
| 536 | ) | ||
| 537 | } | ||
| 538 | }; | ||
| 539 | |||
| 359 | // Scaleway actually doesn't provide SHA256 suport, but maybe in the future :) | 540 | // Scaleway actually doesn't provide SHA256 suport, but maybe in the future :) |
| 360 | if !validate_checksum(obj.oid, &result) { | 541 | if !validate_checksum(obj.oid, &result) { |
| 361 | todo!(); | 542 | return BatchResponseObject::error( |
| 543 | obj, | ||
| 544 | StatusCode::UNPROCESSABLE_ENTITY, | ||
| 545 | "Object corrupted".to_string(), | ||
| 546 | ); | ||
| 362 | } | 547 | } |
| 363 | if !validate_size(obj.size, &result) { | 548 | if !validate_size(obj.size, &result) { |
| 364 | todo!(); | 549 | return BatchResponseObject::error( |
| 550 | obj, | ||
| 551 | StatusCode::UNPROCESSABLE_ENTITY, | ||
| 552 | "Incorrect size specified (or object corrupted)".to_string(), | ||
| 553 | ); | ||
| 365 | } | 554 | } |
| 366 | 555 | ||
| 367 | let expires_in = std::time::Duration::from_secs(5 * 60); | 556 | let expires_in = std::time::Duration::from_secs(5 * 60); |
| 368 | let expires_at = Utc::now() + expires_in; | 557 | let expires_at = Utc::now() + expires_in; |
| 369 | 558 | ||
| 370 | if trusted { | 559 | if trusted { |
| 371 | let config = aws_sdk_s3::presigning::PresigningConfig::expires_in(expires_in).unwrap(); | 560 | let Ok(config) = aws_sdk_s3::presigning::PresigningConfig::expires_in(expires_in) else { |
| 372 | let presigned = state.s3_client.get_object().presigned(config).await.unwrap(); | 561 | return BatchResponseObject::error( |
| 373 | return BatchResponseObject{ | 562 | obj, |
| 563 | StatusCode::INTERNAL_SERVER_ERROR, | ||
| 564 | "Failed to generate upload URL".to_string(), | ||
| 565 | ); | ||
| 566 | }; | ||
| 567 | let Ok(presigned) = state.s3_client.get_object().presigned(config).await else { | ||
| 568 | return BatchResponseObject::error( | ||
| 569 | obj, | ||
| 570 | StatusCode::INTERNAL_SERVER_ERROR, | ||
| 571 | "Failed to generate upload URL".to_string(), | ||
| 572 | ); | ||
| 573 | }; | ||
| 574 | return BatchResponseObject { | ||
| 374 | oid: obj.oid, | 575 | oid: obj.oid, |
| 375 | size: obj.size, | 576 | size: obj.size, |
| 376 | authenticated: Some(true), | 577 | authenticated: Some(true), |
| 377 | actions: BatchResponseObjectActions { | 578 | actions: BatchResponseObjectActions { |
| 378 | download: Some(BatchResponseObjectAction{ | 579 | download: Some(BatchResponseObjectAction { |
| 379 | header: presigned.headers().map(|(k, v)| (k.to_owned(), v.to_owned())).collect(), | 580 | header: presigned |
| 581 | .headers() | ||
| 582 | .map(|(k, v)| (k.to_owned(), v.to_owned())) | ||
| 583 | .collect(), | ||
| 380 | expires_at, | 584 | expires_at, |
| 381 | href: presigned.uri().to_string(), | 585 | href: presigned.uri().to_string(), |
| 382 | }), | 586 | }), |
| 383 | ..Default::default() | 587 | ..Default::default() |
| 384 | } | 588 | }, |
| 589 | error: None, | ||
| 385 | }; | 590 | }; |
| 386 | } | 591 | } |
| 387 | todo!(); | 592 | |
| 593 | let Some(tag) = common::generate_tag( | ||
| 594 | common::Claims { | ||
| 595 | specific_claims: common::SpecificClaims::Download(obj.oid), | ||
| 596 | repo_path: repo, | ||
| 597 | expires_at, | ||
| 598 | }, | ||
| 599 | &state.authz_conf.key, | ||
| 600 | ) else { | ||
| 601 | return BatchResponseObject::error( | ||
| 602 | obj, | ||
| 603 | StatusCode::INTERNAL_SERVER_ERROR, | ||
| 604 | "Internal server error".to_string(), | ||
| 605 | ); | ||
| 606 | }; | ||
| 607 | |||
| 608 | let upload_path = format!( | ||
| 609 | "{repo}/info/lfs/objects/{}/{}/{}", | ||
| 610 | HexByte(obj.oid[0]), | ||
| 611 | HexByte(obj.oid[1]), | ||
| 612 | obj.oid, | ||
| 613 | ); | ||
| 614 | |||
| 615 | BatchResponseObject { | ||
| 616 | oid: obj.oid, | ||
| 617 | size: obj.size, | ||
| 618 | authenticated: Some(true), | ||
| 619 | actions: BatchResponseObjectActions { | ||
| 620 | download: Some(BatchResponseObjectAction { | ||
| 621 | header: { | ||
| 622 | let mut map = HashMap::new(); | ||
| 623 | map.insert( | ||
| 624 | "Authorization".to_string(), | ||
| 625 | format!("Gitolfs3-Hmac-Sha256 {tag} {}", expires_at.timestamp()), | ||
| 626 | ); | ||
| 627 | map | ||
| 628 | }, | ||
| 629 | expires_at, | ||
| 630 | href: format!("{}/{upload_path}", state.base_url), | ||
| 631 | }), | ||
| 632 | ..Default::default() | ||
| 633 | }, | ||
| 634 | error: None, | ||
| 635 | } | ||
| 388 | } | 636 | } |
| 389 | 637 | ||
| 390 | struct AuthorizationConfig { | 638 | struct AuthorizationConfig { |
| @@ -410,17 +658,17 @@ fn forwarded_for_trusted_host( | |||
| 410 | )); | 658 | )); |
| 411 | } | 659 | } |
| 412 | } | 660 | } |
| 413 | return Ok(false); | 661 | Ok(false) |
| 414 | } | 662 | } |
| 415 | const REPO_NOT_FOUND: GitLfsErrorResponse = | 663 | const REPO_NOT_FOUND: GitLfsErrorResponse = |
| 416 | make_error_resp(StatusCode::NOT_FOUND, "Repository not found"); | 664 | make_error_resp(StatusCode::NOT_FOUND, "Repository not found"); |
| 417 | 665 | ||
| 418 | fn authorize( | 666 | fn authorize_batch( |
| 419 | conf: &AuthorizationConfig, | 667 | conf: &AuthorizationConfig, |
| 420 | headers: &HeaderMap, | ||
| 421 | repo_path: &str, | 668 | repo_path: &str, |
| 422 | public: bool, | 669 | public: bool, |
| 423 | operation: common::Operation, | 670 | operation: common::Operation, |
| 671 | headers: &HeaderMap, | ||
| 424 | ) -> Result<Trusted, GitLfsErrorResponse<'static>> { | 672 | ) -> Result<Trusted, GitLfsErrorResponse<'static>> { |
| 425 | // - No authentication required for downloading exported repos | 673 | // - No authentication required for downloading exported repos |
| 426 | // - When authenticated: | 674 | // - When authenticated: |
| @@ -428,46 +676,12 @@ fn authorize( | |||
| 428 | // - When accessing over Tailscale: | 676 | // - When accessing over Tailscale: |
| 429 | // - No authentication required for downloading from any repo | 677 | // - No authentication required for downloading from any repo |
| 430 | 678 | ||
| 431 | const INVALID_AUTHZ_HEADER: GitLfsErrorResponse = | 679 | let claims = VerifyClaimsInput { |
| 432 | make_error_resp(StatusCode::BAD_REQUEST, "Invalid authorization header"); | 680 | specific_claims: common::SpecificClaims::BatchApi(operation), |
| 433 | 681 | repo_path, | |
| 434 | if let Some(authz) = headers.get(header::AUTHORIZATION) { | 682 | }; |
| 435 | if let Ok(authz) = authz.to_str() { | 683 | if verify_claims(conf, &claims, headers)? { |
| 436 | if let Some(val) = authz.strip_prefix("Gitolfs3-Hmac-Sha256 ") { | 684 | return Ok(Trusted(true)); |
| 437 | let Some((tag, expires_at)) = val.split_once(' ') else { | ||
| 438 | return Err(INVALID_AUTHZ_HEADER); | ||
| 439 | }; | ||
| 440 | let Ok(tag): Result<common::Digest<32>, _> = tag.parse() else { | ||
| 441 | return Err(INVALID_AUTHZ_HEADER); | ||
| 442 | }; | ||
| 443 | let Ok(expires_at): Result<i64, _> = expires_at.parse() else { | ||
| 444 | return Err(INVALID_AUTHZ_HEADER); | ||
| 445 | }; | ||
| 446 | let Some(expires_at) = DateTime::<Utc>::from_timestamp(expires_at, 0) else { | ||
| 447 | return Err(INVALID_AUTHZ_HEADER); | ||
| 448 | }; | ||
| 449 | let Some(expected_tag) = common::generate_tag( | ||
| 450 | common::Claims { | ||
| 451 | auth_type: common::AuthType::GitLfsAuthenticate, | ||
| 452 | repo_path, | ||
| 453 | expires_at, | ||
| 454 | operation, | ||
| 455 | }, | ||
| 456 | &conf.key, | ||
| 457 | ) else { | ||
| 458 | return Err(INVALID_AUTHZ_HEADER); | ||
| 459 | }; | ||
| 460 | if tag == expected_tag { | ||
| 461 | return Ok(Trusted(true)); | ||
| 462 | } else { | ||
| 463 | return Err(INVALID_AUTHZ_HEADER); | ||
| 464 | } | ||
| 465 | } else { | ||
| 466 | return Err(INVALID_AUTHZ_HEADER); | ||
| 467 | } | ||
| 468 | } else { | ||
| 469 | return Err(INVALID_AUTHZ_HEADER); | ||
| 470 | } | ||
| 471 | } | 685 | } |
| 472 | 686 | ||
| 473 | let trusted = forwarded_for_trusted_host(headers, &conf.trusted_forwarded_hosts)?; | 687 | let trusted = forwarded_for_trusted_host(headers, &conf.trusted_forwarded_hosts)?; |
| @@ -495,7 +709,7 @@ fn repo_exists(name: &str) -> bool { | |||
| 495 | let Ok(metadata) = std::fs::metadata(name) else { | 709 | let Ok(metadata) = std::fs::metadata(name) else { |
| 496 | return false; | 710 | return false; |
| 497 | }; | 711 | }; |
| 498 | return metadata.is_dir(); | 712 | metadata.is_dir() |
| 499 | } | 713 | } |
| 500 | 714 | ||
| 501 | fn is_repo_public(name: &str) -> Option<bool> { | 715 | fn is_repo_public(name: &str) -> Option<bool> { |
| @@ -517,12 +731,12 @@ async fn batch( | |||
| 517 | let Some(public) = is_repo_public(&repo) else { | 731 | let Some(public) = is_repo_public(&repo) else { |
| 518 | return REPO_NOT_FOUND.into_response(); | 732 | return REPO_NOT_FOUND.into_response(); |
| 519 | }; | 733 | }; |
| 520 | let Trusted(trusted) = match authorize( | 734 | let Trusted(trusted) = match authorize_batch( |
| 521 | &state.authz_conf, | 735 | &state.authz_conf, |
| 522 | &headers, | ||
| 523 | &repo, | 736 | &repo, |
| 524 | public, | 737 | public, |
| 525 | payload.operation, | 738 | payload.operation, |
| 739 | &headers, | ||
| 526 | ) { | 740 | ) { |
| 527 | Ok(authn) => authn, | 741 | Ok(authn) => authn, |
| 528 | Err(e) => return e.into_response(), | 742 | Err(e) => return e.into_response(), |
| @@ -547,16 +761,24 @@ async fn batch( | |||
| 547 | return make_error_resp(StatusCode::CONFLICT, message).into_response(); | 761 | return make_error_resp(StatusCode::CONFLICT, message).into_response(); |
| 548 | } | 762 | } |
| 549 | 763 | ||
| 550 | let resp: BatchResponse; | 764 | let mut resp = BatchResponse { |
| 765 | transfer: TransferAdapter::Basic, | ||
| 766 | objects: vec![], | ||
| 767 | hash_algo: HashAlgo::Sha256, | ||
| 768 | }; | ||
| 551 | for obj in payload.objects { | 769 | for obj in payload.objects { |
| 552 | handle_download_object(&state, &repo, &obj, trusted).await; | 770 | match payload.operation { |
| 553 | // match payload.operation { | 771 | common::Operation::Download => resp |
| 554 | // Operation::Download => resp.objects.push(handle_download_object(repo, obj));, | 772 | .objects |
| 555 | // Operation::Upload => resp.objects.push(handle_upload_object(repo, obj)), | 773 | .push(handle_download_object(&state, &repo, &obj, trusted).await), |
| 556 | // }; | 774 | common::Operation::Upload => { |
| 775 | if let Some(obj_resp) = handle_upload_object(&state, &repo, &obj).await { | ||
| 776 | resp.objects.push(obj_resp); | ||
| 777 | } | ||
| 778 | } | ||
| 779 | }; | ||
| 557 | } | 780 | } |
| 558 | 781 | GitLfsJson(Json(resp)).into_response() | |
| 559 | format!("hi from {repo}\n").into_response() | ||
| 560 | } | 782 | } |
| 561 | 783 | ||
| 562 | #[derive(Deserialize, Copy, Clone)] | 784 | #[derive(Deserialize, Copy, Clone)] |
| @@ -564,7 +786,7 @@ async fn batch( | |||
| 564 | struct FileParams { | 786 | struct FileParams { |
| 565 | oid0: HexByte, | 787 | oid0: HexByte, |
| 566 | oid1: HexByte, | 788 | oid1: HexByte, |
| 567 | oid: Oid, | 789 | oid: common::Oid, |
| 568 | } | 790 | } |
| 569 | 791 | ||
| 570 | impl<'de> Deserialize<'de> for FileParams { | 792 | impl<'de> Deserialize<'de> for FileParams { |
| @@ -591,6 +813,118 @@ impl<'de> Deserialize<'de> for FileParams { | |||
| 591 | } | 813 | } |
| 592 | } | 814 | } |
| 593 | 815 | ||
| 594 | async fn obj_download(Path(FileParams { oid0, oid1, oid }): Path<FileParams>) {} | 816 | pub struct VerifyClaimsInput<'a> { |
| 817 | pub specific_claims: common::SpecificClaims, | ||
| 818 | pub repo_path: &'a str, | ||
| 819 | } | ||
| 820 | |||
| 821 | // Note: expires_at is ignored. | ||
| 822 | fn verify_claims( | ||
| 823 | conf: &AuthorizationConfig, | ||
| 824 | claims: &VerifyClaimsInput, | ||
| 825 | headers: &HeaderMap, | ||
| 826 | ) -> Result<bool, GitLfsErrorResponse<'static>> { | ||
| 827 | const INVALID_AUTHZ_HEADER: GitLfsErrorResponse = | ||
| 828 | make_error_resp(StatusCode::BAD_REQUEST, "Invalid authorization header"); | ||
| 829 | |||
| 830 | if let Some(authz) = headers.get(header::AUTHORIZATION) { | ||
| 831 | if let Ok(authz) = authz.to_str() { | ||
| 832 | if let Some(val) = authz.strip_prefix("Gitolfs3-Hmac-Sha256 ") { | ||
| 833 | let (tag, expires_at) = val.split_once(' ').ok_or(INVALID_AUTHZ_HEADER)?; | ||
| 834 | let tag: common::Digest<32> = tag.parse().map_err(|_| INVALID_AUTHZ_HEADER)?; | ||
| 835 | let expires_at: i64 = expires_at.parse().map_err(|_| INVALID_AUTHZ_HEADER)?; | ||
| 836 | let expires_at = | ||
| 837 | DateTime::<Utc>::from_timestamp(expires_at, 0).ok_or(INVALID_AUTHZ_HEADER)?; | ||
| 838 | let Some(expected_tag) = common::generate_tag( | ||
| 839 | common::Claims { | ||
| 840 | specific_claims: claims.specific_claims, | ||
| 841 | repo_path: claims.repo_path, | ||
| 842 | expires_at, | ||
| 843 | }, | ||
| 844 | &conf.key, | ||
| 845 | ) else { | ||
| 846 | return Err(make_error_resp( | ||
| 847 | StatusCode::INTERNAL_SERVER_ERROR, | ||
| 848 | "Internal server error", | ||
| 849 | )); | ||
| 850 | }; | ||
| 851 | if tag == expected_tag { | ||
| 852 | return Ok(true); | ||
| 853 | } | ||
| 854 | } | ||
| 855 | } | ||
| 856 | return Err(INVALID_AUTHZ_HEADER); | ||
| 857 | } | ||
| 858 | Ok(false) | ||
| 859 | } | ||
| 860 | |||
| 861 | fn authorize_get( | ||
| 862 | conf: &AuthorizationConfig, | ||
| 863 | repo_path: &str, | ||
| 864 | oid: common::Oid, | ||
| 865 | headers: &HeaderMap, | ||
| 866 | ) -> Result<(), GitLfsErrorResponse<'static>> { | ||
| 867 | let claims = VerifyClaimsInput { | ||
| 868 | specific_claims: common::SpecificClaims::Download(oid), | ||
| 869 | repo_path, | ||
| 870 | }; | ||
| 871 | if !verify_claims(conf, &claims, headers)? { | ||
| 872 | return Err(make_error_resp( | ||
| 873 | StatusCode::UNAUTHORIZED, | ||
| 874 | "Repository not found", | ||
| 875 | )); | ||
| 876 | } | ||
| 877 | Ok(()) | ||
| 878 | } | ||
| 595 | 879 | ||
| 596 | async fn obj_upload(Path(FileParams { oid0, oid1, oid }): Path<FileParams>) {} | 880 | async fn obj_download( |
| 881 | State(state): State<Arc<AppState>>, | ||
| 882 | headers: HeaderMap, | ||
| 883 | RepositoryName(repo): RepositoryName, | ||
| 884 | Path(FileParams { oid0, oid1, oid }): Path<FileParams>, | ||
| 885 | ) -> Response { | ||
| 886 | if let Err(e) = authorize_get(&state.authz_conf, &repo, oid, &headers) { | ||
| 887 | return e.into_response(); | ||
| 888 | } | ||
| 889 | |||
| 890 | let full_path = format!("{repo}/lfs/objects/{}/{}/{}", oid0, oid1, oid); | ||
| 891 | let result = match state | ||
| 892 | .s3_client | ||
| 893 | .get_object() | ||
| 894 | .bucket(&state.s3_bucket) | ||
| 895 | .key(full_path) | ||
| 896 | .checksum_mode(aws_sdk_s3::types::ChecksumMode::Enabled) | ||
| 897 | .send() | ||
| 898 | .await | ||
| 899 | { | ||
| 900 | Ok(result) => result, | ||
| 901 | Err(_) => { | ||
| 902 | return ( | ||
| 903 | StatusCode::INTERNAL_SERVER_ERROR, | ||
| 904 | "Failed to query object information", | ||
| 905 | ) | ||
| 906 | .into_response(); | ||
| 907 | } | ||
| 908 | }; | ||
| 909 | |||
| 910 | let mut headers = header::HeaderMap::new(); | ||
| 911 | if let Some(content_type) = result.content_type { | ||
| 912 | let Ok(header_value) = content_type.try_into() else { | ||
| 913 | return ( | ||
| 914 | StatusCode::INTERNAL_SERVER_ERROR, | ||
| 915 | "Object has invalid content type", | ||
| 916 | ) | ||
| 917 | .into_response(); | ||
| 918 | }; | ||
| 919 | headers.insert(header::CONTENT_TYPE, header_value); | ||
| 920 | } | ||
| 921 | if let Some(content_length) = result.content_length { | ||
| 922 | headers.insert(header::CONTENT_LENGTH, content_length.into()); | ||
| 923 | } | ||
| 924 | |||
| 925 | let async_read = result.body.into_async_read(); | ||
| 926 | let stream = tokio_util::io::ReaderStream::new(async_read); | ||
| 927 | let body = axum::body::Body::from_stream(stream); | ||
| 928 | |||
| 929 | (headers, body).into_response() | ||
| 930 | } | ||