| 1 | //! `dogfood-hook` — the pre-receive hook (spec §4). | |
| 2 | //! | |
| 3 | //! > Hooks are the compiled `dogfood-hook` binary, installed into each repo's | |
| 4 | //! > `hooks/` directory at creation and connecting to the database over a local | |
| 5 | //! > socket. Do not use shell scripts — argument handling and error propagation | |
| 6 | //! > are both worse. | |
| 7 | //! | |
| 8 | //! Runs synchronously on the push path with a <200ms budget, so it does the | |
| 9 | //! cheap checks first and only touches the database when it must. | |
| 10 | //! | |
| 11 | //! stdin is one line per ref update: `<old-oid> <new-oid> <refname>`. | |
| 12 | //! A non-zero exit rejects the entire push. | |
| 13 | ||
| 14 | use std::io::{BufRead, Write}; | |
| 15 | use std::process::ExitCode; | |
| 16 | ||
| 17 | use df_store::refname; | |
| 18 | ||
| 19 | /// Hard cap on the whole hook, so a database stall cannot hang a push | |
| 20 | /// indefinitely. On timeout we *allow* the push: the alternative is that a | |
| 21 | /// database blip blocks all pushes, and the indexer re-derives state anyway. | |
| 22 | /// Ref-name validation is local and has already run by then. | |
| 23 | const DB_TIMEOUT: std::time::Duration = std::time::Duration::from_millis(1500); | |
| 24 | ||
| 25 | fn main() -> ExitCode { | |
| 26 | let mut stderr = std::io::stderr(); | |
| 27 | ||
| 28 | let updates = match read_updates() { | |
| 29 | Ok(u) => u, | |
| 30 | Err(e) => { | |
| 31 | let _ = writeln!(stderr, "dogfood: could not read ref updates: {e}"); | |
| 32 | return ExitCode::FAILURE; | |
| 33 | } | |
| 34 | }; | |
| 35 | ||
| 36 | if updates.is_empty() { | |
| 37 | return ExitCode::SUCCESS; | |
| 38 | } | |
| 39 | ||
| 40 | // ── 1. ref-name validation (local, no I/O) ─────────────────────────────── | |
| 41 | for u in &updates { | |
| 42 | if let Err(e) = refname::validate_pushed_ref(&u.refname) { | |
| 43 | // The message is printed to the pusher's terminal. `RefError`'s | |
| 44 | // Display never echoes the offending byte, which is the point. | |
| 45 | let _ = writeln!(stderr, "dogfood: rejected `{}`: {e}", sanitise(&u.refname)); | |
| 46 | return ExitCode::FAILURE; | |
| 47 | } | |
| 48 | } | |
| 49 | ||
| 50 | // ── 2. protected bookmarks (needs the database) ────────────────────────── | |
| 51 | let repo_id = std::env::var("DOGFOOD_REPO_ID").ok(); | |
| 52 | let database_url = std::env::var("DATABASE_URL").ok(); | |
| 53 | ||
| 54 | // Set by whichever transport spawned git, to say "this check is mandatory". | |
| 55 | // Its absence means an old hook script or a transport that forgot to pass | |
| 56 | // the environment — the exact failure that made bookmark protection a | |
| 57 | // no-op over HTTPS without anything reporting it. | |
| 58 | let enforced = std::env::var("DOGFOOD_ENFORCE").is_ok(); | |
| 59 | ||
| 60 | match (repo_id, database_url) { | |
| 61 | (Some(repo_id), Some(url)) => match check_protected(&url, &repo_id, &updates) { | |
| 62 | Ok(Some(violation)) => { | |
| 63 | let _ = writeln!(stderr, "dogfood: {violation}"); | |
| 64 | return ExitCode::FAILURE; | |
| 65 | } | |
| 66 | Ok(None) => {} | |
| 67 | Err(e) => { | |
| 68 | // A *transient* failure fails open, loudly: a database blip | |
| 69 | // blocking every push is worse than one missed check, and ref | |
| 70 | // names were already validated locally above. | |
| 71 | let _ = writeln!( | |
| 72 | stderr, | |
| 73 | "dogfood: warning: could not verify bookmark protection ({e}); allowing push" | |
| 74 | ); | |
| 75 | } | |
| 76 | }, | |
| 77 | // Misconfiguration, not a blip. Failing open here is how a protected | |
| 78 | // bookmark stays unprotected forever with nobody noticing, so when the | |
| 79 | // caller said the check was mandatory, refuse. | |
| 80 | _ if enforced => { | |
| 81 | let _ = writeln!( | |
| 82 | stderr, | |
| 83 | "dogfood: push validation is not configured on this server \ | |
| 84 | (missing DOGFOOD_REPO_ID or DATABASE_URL); refusing the push" | |
| 85 | ); | |
| 86 | return ExitCode::FAILURE; | |
| 87 | } | |
| 88 | _ => {} | |
| 89 | } | |
| 90 | ||
| 91 | ExitCode::SUCCESS | |
| 92 | } | |
| 93 | ||
| 94 | struct Update { | |
| 95 | old: String, | |
| 96 | new: String, | |
| 97 | refname: String, | |
| 98 | } | |
| 99 | ||
| 100 | impl Update { | |
| 101 | /// Whether this update deletes the ref. | |
| 102 | fn is_delete(&self) -> bool { | |
| 103 | self.new.chars().all(|c| c == '0') | |
| 104 | } | |
| 105 | ||
| 106 | /// Whether this update is a non-fast-forward (a force push). | |
| 107 | /// | |
| 108 | /// The hook cannot cheaply prove ancestry without object access, so this is | |
| 109 | /// the conservative signal: any update to an existing ref that is not a | |
| 110 | /// creation. Protected bookmarks reject both. | |
| 111 | fn is_update_of_existing(&self) -> bool { | |
| 112 | !self.old.chars().all(|c| c == '0') | |
| 113 | } | |
| 114 | } | |
| 115 | ||
| 116 | fn read_updates() -> std::io::Result<Vec<Update>> { | |
| 117 | let stdin = std::io::stdin(); | |
| 118 | let mut out = Vec::new(); | |
| 119 | ||
| 120 | for line in stdin.lock().lines() { | |
| 121 | let line = line?; | |
| 122 | let line = line.trim_end_matches(['\r', '\n']); | |
| 123 | if line.is_empty() { | |
| 124 | continue; | |
| 125 | } | |
| 126 | let mut parts = line.split(' '); | |
| 127 | let (Some(old), Some(new), Some(refname)) = (parts.next(), parts.next(), parts.next()) | |
| 128 | else { | |
| 129 | // A malformed line means we do not understand the push; refuse | |
| 130 | // rather than let it through unvalidated. | |
| 131 | return Err(std::io::Error::new( | |
| 132 | std::io::ErrorKind::InvalidData, | |
| 133 | "malformed ref update line", | |
| 134 | )); | |
| 135 | }; | |
| 136 | out.push(Update { | |
| 137 | old: old.to_string(), | |
| 138 | new: new.to_string(), | |
| 139 | refname: refname.to_string(), | |
| 140 | }); | |
| 141 | } | |
| 142 | Ok(out) | |
| 143 | } | |
| 144 | ||
| 145 | /// Whether `old` is an ancestor of `new` — i.e. whether this update is a | |
| 146 | /// fast-forward. | |
| 147 | /// | |
| 148 | /// Asked of git rather than reasoned about: the hook already runs inside the | |
| 149 | /// repository (git sets `GIT_DIR`), and re-implementing ancestry against a | |
| 150 | /// pack we do not otherwise read would be a lot of code to get subtly wrong. | |
| 151 | /// | |
| 152 | /// On any doubt this returns `false`, so an update we cannot classify is | |
| 153 | /// treated as a force-push and refused. For a *protected* bookmark that is the | |
| 154 | /// safe direction. | |
| 155 | fn is_fast_forward(old: &str, new: &str) -> bool { | |
| 156 | std::process::Command::new("git") | |
| 157 | .args(["merge-base", "--is-ancestor", old, new]) | |
| 158 | .stdin(std::process::Stdio::null()) | |
| 159 | .stdout(std::process::Stdio::null()) | |
| 160 | .stderr(std::process::Stdio::null()) | |
| 161 | .status() | |
| 162 | .map(|s| s.success()) | |
| 163 | .unwrap_or(false) | |
| 164 | } | |
| 165 | ||
| 166 | /// Reject pushes to an archived repository, and deletes or force-pushes of | |
| 167 | /// protected bookmarks. | |
| 168 | /// | |
| 169 | /// "Protected" means what the settings page says it means: the bookmark cannot | |
| 170 | /// be **deleted** or **force-updated**. An ordinary fast-forward is allowed — | |
| 171 | /// refusing those too would block the normal `jj git push -b main` workflow | |
| 172 | /// while the UI promised it was fine. | |
| 173 | fn check_protected( | |
| 174 | database_url: &str, | |
| 175 | repo_id: &str, | |
| 176 | updates: &[Update], | |
| 177 | ) -> Result<Option<String>, String> { | |
| 178 | let runtime = tokio::runtime::Builder::new_current_thread() | |
| 179 | .enable_all() | |
| 180 | .build() | |
| 181 | .map_err(|e| e.to_string())?; | |
| 182 | ||
| 183 | runtime.block_on(async { | |
| 184 | let fut = async { | |
| 185 | let pool = df_db::connect(database_url, 1) | |
| 186 | .await | |
| 187 | .map_err(|e| e.to_string())?; | |
| 188 | ||
| 189 | let repo: uuid::Uuid = repo_id.parse().map_err(|_| "bad repo id".to_string())?; | |
| 190 | ||
| 191 | // An archived repository accepts nothing. Checked here as well as on | |
| 192 | // each transport, because the hook is the one place both of them | |
| 193 | // pass through. | |
| 194 | let archived: Option<(bool,)> = | |
| 195 | sqlx::query_as("SELECT archived FROM repos WHERE id = $1") | |
| 196 | .bind(repo) | |
| 197 | .fetch_optional(&pool) | |
| 198 | .await | |
| 199 | .map_err(|e| e.to_string())?; | |
| 200 | ||
| 201 | match archived { | |
| 202 | Some((true,)) => { | |
| 203 | return Ok::<_, String>(Some( | |
| 204 | "this repository is archived and does not accept pushes".to_string(), | |
| 205 | )) | |
| 206 | } | |
| 207 | // No row: the repository was deleted between the push starting | |
| 208 | // and this check. Refusing is the only safe reading. | |
| 209 | None => return Ok(Some("this repository no longer exists".to_string())), | |
| 210 | Some((false,)) => {} | |
| 211 | } | |
| 212 | ||
| 213 | // The default bookmark is protected whether or not it carries the | |
| 214 | // flag. The settings page says "always" for it, and a `bookmarks` | |
| 215 | // row created by the indexer defaults to unprotected — so reading | |
| 216 | // the flag alone would quietly make that promise false. | |
| 217 | let protected: Vec<(String,)> = sqlx::query_as( | |
| 218 | "SELECT name FROM bookmarks WHERE repo_id = $1 AND protected | |
| 219 | UNION | |
| 220 | SELECT default_bookmark FROM repos WHERE id = $1", | |
| 221 | ) | |
| 222 | .bind(repo) | |
| 223 | .fetch_all(&pool) | |
| 224 | .await | |
| 225 | .map_err(|e| e.to_string())?; | |
| 226 | ||
| 227 | for u in updates { | |
| 228 | let Some(short) = u.refname.strip_prefix("refs/heads/") else { | |
| 229 | continue; | |
| 230 | }; | |
| 231 | if !protected.iter().any(|(n,)| n == short) { | |
| 232 | continue; | |
| 233 | } | |
| 234 | if u.is_delete() { | |
| 235 | return Ok(Some(format!( | |
| 236 | "`{short}` is a protected bookmark and cannot be deleted" | |
| 237 | ))); | |
| 238 | } | |
| 239 | // Creating it is fine — that is the first push to a new | |
| 240 | // repository. Updating it is fine too, as long as it only moves | |
| 241 | // forward. | |
| 242 | if u.is_update_of_existing() && !is_fast_forward(&u.old, &u.new) { | |
| 243 | return Ok(Some(format!( | |
| 244 | "`{short}` is a protected bookmark and cannot be force-updated; \ | |
| 245 | rebase onto it, or open a change" | |
| 246 | ))); | |
| 247 | } | |
| 248 | } | |
| 249 | Ok(None) | |
| 250 | }; | |
| 251 | ||
| 252 | match tokio::time::timeout(DB_TIMEOUT, fut).await { | |
| 253 | Ok(r) => r, | |
| 254 | Err(_) => Err("timed out".to_string()), | |
| 255 | } | |
| 256 | }) | |
| 257 | } | |
| 258 | ||
| 259 | /// Render a ref name safely for terminal output. | |
| 260 | /// | |
| 261 | /// Validation has usually already rejected anything dangerous, but this runs on | |
| 262 | /// the error path where the name is by definition untrusted. | |
| 263 | fn sanitise(s: &str) -> String { | |
| 264 | s.chars() | |
| 265 | .map(|c| { | |
| 266 | if c.is_control() || c == '\u{7f}' { | |
| 267 | '?' | |
| 268 | } else { | |
| 269 | c | |
| 270 | } | |
| 271 | }) | |
| 272 | .take(200) | |
| 273 | .collect() | |
| 274 | } | |
| 275 | ||
| 276 | #[cfg(test)] | |
| 277 | mod tests { | |
| 278 | use super::*; | |
| 279 | ||
| 280 | fn u(old: &str, new: &str, name: &str) -> Update { | |
| 281 | Update { | |
| 282 | old: old.into(), | |
| 283 | new: new.into(), | |
| 284 | refname: name.into(), | |
| 285 | } | |
| 286 | } | |
| 287 | ||
| 288 | const ZERO: &str = "0000000000000000000000000000000000000000"; | |
| 289 | const OID: &str = "1111111111111111111111111111111111111111"; | |
| 290 | ||
| 291 | #[test] | |
| 292 | fn detects_deletes() { | |
| 293 | assert!(u(OID, ZERO, "refs/heads/main").is_delete()); | |
| 294 | assert!(!u(OID, OID, "refs/heads/main").is_delete()); | |
| 295 | } | |
| 296 | ||
| 297 | #[test] | |
| 298 | fn detects_creation_versus_update() { | |
| 299 | assert!(!u(ZERO, OID, "refs/heads/new").is_update_of_existing()); | |
| 300 | assert!(u(OID, OID, "refs/heads/main").is_update_of_existing()); | |
| 301 | } | |
| 302 | ||
| 303 | #[test] | |
| 304 | fn sanitise_strips_control_bytes() { | |
| 305 | assert_eq!(sanitise("main\x1b[31m"), "main?[31m"); | |
| 306 | assert_eq!(sanitise("main\n"), "main?"); | |
| 307 | assert_eq!(sanitise("ok"), "ok"); | |
| 308 | } | |
| 309 | ||
| 310 | #[test] | |
| 311 | fn sanitise_bounds_its_output() { | |
| 312 | assert_eq!(sanitise(&"a".repeat(500)).len(), 200); | |
| 313 | } | |
| 314 | } |
314 lines · Rust