| 1 | //! Landing a pull request by making commits on the branch it merges into: |
| 2 | //! a merge commit, one squashed commit, or the branch's commits made again |
| 3 | //! on top (`land` with a `method`); and rebasing a pull request's branch |
| 4 | //! onto that branch (`update_pull_branch` with `rebase`). |
| 5 | //! |
| 6 | //! Each is worked out here, without a sandbox, from trees. The changes one |
| 7 | //! side made since the two last agreed are applied to the other side's |
| 8 | //! tree ([`apply`]): a file only that side changed is taken from it, and a |
| 9 | //! file both changed is merged line by line, as git merges text |
| 10 | //! (test_merge.rs `merge_text`). Lines both changed differently, a file |
| 11 | //! one side deleted and the other changed, or a binary file both changed |
| 12 | //! are a conflict, and nothing is made: the conflict names the files. Only |
| 13 | //! the trees on the way to changed files are read and written. What is |
| 14 | //! made is pushed as one pack, with whatever a fork's commits bring that |
| 15 | //! the repository lacks, and the branch moves only if it is still where it |
| 16 | //! was. |
| 17 | //! |
| 18 | //! The work is bounded for a Worker: at most [`MAX_LAND_FILES`] files |
| 19 | //! applied at a time, [`MAX_LINE_MERGES`] of them merged line by line |
| 20 | //! (each at most test_merge.rs `MAX_MERGE_BYTES`), and at most |
| 21 | //! [`MAX_REBASE_COMMITS`] commits made again by a rebase. Beyond those g1t |
| 22 | //! says it cannot do it itself, and nothing is changed. |
| 23 | |
| 24 | use std::cell::RefCell; |
| 25 | use std::collections::{BTreeSet, HashMap, HashSet}; |
| 26 | |
| 27 | use futures_util::future::{try_join, try_join_all}; |
| 28 | use g1t_contracts::repos::{ |
| 29 | Branch, BuildQueueStepArgs, Commit, EntryKind, GitAccess, LandArgs, Landed, NeedsAgentReason, PullBranchUpdate, QueueStep, |
| 30 | Repo, TreeEntry, |
| 31 | }; |
| 32 | use g1t_contracts::rules::MergeMethod; |
| 33 | use g1t_contracts::time::parse_rfc3339; |
| 34 | use g1t_contracts::{FailureCode, Outcome, User}; |
| 35 | use g1t_kit::now_ms; |
| 36 | use g1t_scan::pack::{ObjectKind, extend_pack, object_id, parse_tree, write_pack}; |
| 37 | use worker::Result; |
| 38 | |
| 39 | use crate::catch_up::{BranchUpdate, Change, Signature, ancestors, commit_object_by, merge_tree, moved_since, read_dirs}; |
| 40 | use crate::diff::{FileEntry, changed_entries_up_to}; |
| 41 | use crate::registry::store_key; |
| 42 | use crate::store::{GitRepo, GitStore, Scope}; |
| 43 | use crate::test_merge::{as_text, merge_text}; |
| 44 | use crate::{MAX_ANCESTRY, Repos, descends_from, land, nearest_ancestor_in}; |
| 45 | |
| 46 | /// The most files one side's change may touch for g1t to apply it. |
| 47 | pub(crate) const MAX_LAND_FILES: usize = 1000; |
| 48 | /// The most files both sides changed that are merged line by line at once. |
| 49 | pub(crate) const MAX_LINE_MERGES: usize = 100; |
| 50 | /// The most commits a rebase makes again. |
| 51 | pub(crate) const MAX_REBASE_COMMITS: usize = 250; |
| 52 | |
| 53 | /// Reads objects from a repository, then from a second one (a fork, or the |
| 54 | /// repository a fork came from) for what the first lacks, and first of all |
| 55 | /// from the objects made here and not yet pushed. |
| 56 | pub(crate) struct Reader<'a, R> { |
| 57 | first: &'a R, |
| 58 | second: Option<&'a R>, |
| 59 | trees: RefCell<HashMap<String, Vec<TreeEntry>>>, |
| 60 | blobs: RefCell<HashMap<String, Vec<u8>>>, |
| 61 | } |
| 62 | |
| 63 | /// What git's tree mode says an entry is. |
| 64 | fn kind_of(mode: &str) -> EntryKind { |
| 65 | match mode { |
| 66 | "40000" | "040000" => EntryKind::Tree, |
| 67 | "100755" => EntryKind::Exec, |
| 68 | "120000" => EntryKind::Symlink, |
| 69 | "160000" => EntryKind::Gitlink, |
| 70 | _ => EntryKind::Blob, |
| 71 | } |
| 72 | } |
| 73 | |
| 74 | impl<'a, R: GitRepo> Reader<'a, R> { |
| 75 | pub(crate) fn new(first: &'a R, second: Option<&'a R>) -> Self { |
| 76 | Reader { first, second, trees: RefCell::default(), blobs: RefCell::default() } |
| 77 | } |
| 78 | |
| 79 | /// Keeps `objects`, made here, to be read before they are pushed. |
| 80 | fn keep(&self, objects: &[(ObjectKind, Vec<u8>)]) { |
| 81 | for (kind, bytes) in objects { |
| 82 | match kind { |
| 83 | ObjectKind::Tree => { |
| 84 | let entries = parse_tree(bytes) |
| 85 | .into_iter() |
| 86 | .map(|item| TreeEntry { kind: kind_of(&item.mode), name: item.name, hash: item.id }) |
| 87 | .collect(); |
| 88 | self.trees.borrow_mut().insert(object_id(ObjectKind::Tree, bytes), entries); |
| 89 | } |
| 90 | ObjectKind::Blob => { |
| 91 | self.blobs.borrow_mut().insert(object_id(ObjectKind::Blob, bytes), bytes.clone()); |
| 92 | } |
| 93 | _ => {} |
| 94 | } |
| 95 | } |
| 96 | } |
| 97 | } |
| 98 | |
| 99 | impl<R: GitRepo> GitRepo for Reader<'_, R> { |
| 100 | async fn access(&self, scope: Scope) -> Result<GitAccess> { |
| 101 | self.first.access(scope).await |
| 102 | } |
| 103 | async fn branches(&self) -> Result<Vec<Branch>> { |
| 104 | self.first.branches().await |
| 105 | } |
| 106 | async fn log(&self, git_ref: &str, limit: u32) -> Result<Vec<Commit>> { |
| 107 | let found = self.first.log(git_ref, limit).await?; |
| 108 | match self.second { |
| 109 | Some(second) if found.is_empty() => second.log(git_ref, limit).await, |
| 110 | _ => Ok(found), |
| 111 | } |
| 112 | } |
| 113 | async fn parents(&self, commit_hash: &str) -> Result<Option<Vec<String>>> { |
| 114 | match (self.first.parents(commit_hash).await?, self.second) { |
| 115 | (None, Some(second)) => second.parents(commit_hash).await, |
| 116 | (found, _) => Ok(found), |
| 117 | } |
| 118 | } |
| 119 | async fn read_tree(&self, tree_hash: &str) -> Result<Option<Vec<TreeEntry>>> { |
| 120 | let made = self.trees.borrow().get(tree_hash).cloned(); |
| 121 | if made.is_some() { |
| 122 | return Ok(made); |
| 123 | } |
| 124 | match (self.first.read_tree(tree_hash).await?, self.second) { |
| 125 | (None, Some(second)) => second.read_tree(tree_hash).await, |
| 126 | (found, _) => Ok(found), |
| 127 | } |
| 128 | } |
| 129 | async fn read_blob(&self, blob_hash: &str) -> Result<Option<Vec<u8>>> { |
| 130 | let made = self.blobs.borrow().get(blob_hash).cloned(); |
| 131 | if made.is_some() { |
| 132 | return Ok(made); |
| 133 | } |
| 134 | match (self.first.read_blob(blob_hash).await?, self.second) { |
| 135 | (None, Some(second)) => second.read_blob(blob_hash).await, |
| 136 | (found, _) => Ok(found), |
| 137 | } |
| 138 | } |
| 139 | async fn read_file(&self, git_ref: &str, path: &str) -> Result<Option<Vec<u8>>> { |
| 140 | self.first.read_file(git_ref, path).await |
| 141 | } |
| 142 | async fn fork(&self, target_key: &str, default_branch_only: bool) -> Result<()> { |
| 143 | self.first.fork(target_key, default_branch_only).await |
| 144 | } |
| 145 | } |
| 146 | |
| 147 | /// What becomes of one file a change touched, applied to another tree. |
| 148 | #[derive(Debug, PartialEq, Eq)] |
| 149 | pub(crate) enum Resolution { |
| 150 | /// The other tree has it as the change started from: put the change's |
| 151 | /// version there (`None`: remove it). |
| 152 | Take(Option<FileEntry>), |
| 153 | /// The other tree already has it as the change made it. |
| 154 | Keep, |
| 155 | /// Both changed the text: merge the three, as `kind`. |
| 156 | Content { kind: EntryKind, base: Option<String>, ours: String, theirs: String }, |
| 157 | /// Both changed it in ways that cannot be merged. |
| 158 | Conflict, |
| 159 | } |
| 160 | |
| 161 | fn text_kind(kind: EntryKind) -> bool { |
| 162 | matches!(kind, EntryKind::Blob | EntryKind::Exec) |
| 163 | } |
| 164 | |
| 165 | /// How a file the change took from `base` to `ours` comes out where it is |
| 166 | /// `theirs` (a directory's entry included, which a file cannot replace). |
| 167 | pub(crate) fn resolve(base: Option<&FileEntry>, ours: Option<&FileEntry>, theirs: Option<&FileEntry>) -> Resolution { |
| 168 | if theirs == base { |
| 169 | return Resolution::Take(ours.cloned()); |
| 170 | } |
| 171 | if theirs == ours { |
| 172 | return Resolution::Keep; |
| 173 | } |
| 174 | // Deleted on one side and changed on the other. |
| 175 | let (Some(ours), Some(theirs)) = (ours, theirs) else { |
| 176 | return Resolution::Conflict; |
| 177 | }; |
| 178 | if ours.0 == EntryKind::Tree || theirs.0 == EntryKind::Tree { |
| 179 | return Resolution::Conflict; |
| 180 | } |
| 181 | let base_kind = base.map(|(kind, _)| *kind); |
| 182 | // A change of mode on one side is kept with the other's content. |
| 183 | let kind = if ours.0 == theirs.0 { |
| 184 | ours.0 |
| 185 | } else if base_kind == Some(ours.0) { |
| 186 | theirs.0 |
| 187 | } else if base_kind == Some(theirs.0) { |
| 188 | ours.0 |
| 189 | } else { |
| 190 | return Resolution::Conflict; |
| 191 | }; |
| 192 | if ours.1 == theirs.1 { |
| 193 | return Resolution::Take(Some((kind, ours.1.clone()))); |
| 194 | } |
| 195 | if !text_kind(ours.0) || !text_kind(theirs.0) || base_kind.is_some_and(|kind| !text_kind(kind)) { |
| 196 | return Resolution::Conflict; |
| 197 | } |
| 198 | Resolution::Content { kind, base: base.map(|(_, hash)| hash.clone()), ours: ours.1.clone(), theirs: theirs.1.clone() } |
| 199 | } |
| 200 | |
| 201 | /// A change applied to a tree. |
| 202 | #[derive(Debug, PartialEq, Eq)] |
| 203 | pub(crate) enum Applied { |
| 204 | /// The tree it makes. |
| 205 | Tree(String), |
| 206 | /// The files whose changes clash, by path. |
| 207 | Conflict(Vec<String>), |
| 208 | /// What g1t cannot apply itself. |
| 209 | Unsupported(String), |
| 210 | } |
| 211 | |
| 212 | /// The change from tree `from` (`None`: nothing) to tree `to`, applied to |
| 213 | /// tree `onto`. The objects it makes are added to `objects`, and kept in |
| 214 | /// `git` for what is applied after it. |
| 215 | pub(crate) async fn apply<R: GitRepo>( |
| 216 | git: &Reader<'_, R>, |
| 217 | from: Option<&str>, |
| 218 | to: &str, |
| 219 | onto: &str, |
| 220 | objects: &mut Vec<(ObjectKind, Vec<u8>)>, |
| 221 | ) -> Result<Applied> { |
| 222 | if from == Some(onto) || to == onto { |
| 223 | return Ok(Applied::Tree(to.to_owned())); |
| 224 | } |
| 225 | if from == Some(to) { |
| 226 | return Ok(Applied::Tree(onto.to_owned())); |
| 227 | } |
| 228 | let (ours, cut) = changed_entries_up_to(git, from, to, MAX_LAND_FILES).await?; |
| 229 | if cut { |
| 230 | return Ok(Applied::Unsupported(format!("it changes more than {MAX_LAND_FILES} files"))); |
| 231 | } |
| 232 | // Each changed path as `onto` has it. |
| 233 | let dirs: BTreeSet<String> = ours.iter().flat_map(|(path, _, _)| ancestors(path).map(str::to_owned)).collect(); |
| 234 | let found = read_dirs(git, onto, &dirs).await?; |
| 235 | let theirs_of = |path: &str| -> Option<FileEntry> { |
| 236 | let (parent, name) = path.rsplit_once('/').unwrap_or(("", path)); |
| 237 | let (_, entries) = found.get(parent)?; |
| 238 | entries.iter().find(|entry| entry.name == name).map(|entry| (entry.kind, entry.hash.clone())) |
| 239 | }; |
| 240 | let mut changes = Vec::new(); |
| 241 | let mut conflicts = Vec::new(); |
| 242 | let mut merges = Vec::new(); |
| 243 | for (path, base, mine) in ours { |
| 244 | let theirs = theirs_of(&path); |
| 245 | match resolve(base.as_ref(), mine.as_ref(), theirs.as_ref()) { |
| 246 | Resolution::Take(entry) => changes.push(Change { path, entry }), |
| 247 | Resolution::Keep => {} |
| 248 | Resolution::Conflict => conflicts.push(path), |
| 249 | Resolution::Content { kind, base, ours, theirs } => merges.push((path, kind, base, ours, theirs)), |
| 250 | } |
| 251 | } |
| 252 | if merges.len() > MAX_LINE_MERGES { |
| 253 | return Ok(Applied::Unsupported(format!("both sides changed more than {MAX_LINE_MERGES} of the same files"))); |
| 254 | } |
| 255 | let read = try_join_all(merges.iter().map(|(_, _, base, ours, theirs)| async move { |
| 256 | let base = async { |
| 257 | match base { |
| 258 | Some(hash) => git.read_blob(hash).await, |
| 259 | None => Ok(Some(Vec::new())), |
| 260 | } |
| 261 | }; |
| 262 | let (base, (ours, theirs)) = try_join(base, try_join(git.read_blob(ours), git.read_blob(theirs))).await?; |
| 263 | Ok::<_, worker::Error>((base, ours, theirs)) |
| 264 | })) |
| 265 | .await?; |
| 266 | let mut made: Vec<(ObjectKind, Vec<u8>)> = Vec::new(); |
| 267 | for ((path, kind, ..), (base, ours, theirs)) in merges.into_iter().zip(read) { |
| 268 | let texts = (as_text(base), as_text(ours), as_text(theirs)); |
| 269 | let (Ok(base), Ok(ours), Ok(theirs)) = texts else { |
| 270 | if matches!(texts, (Err(true), _, _) | (_, Err(true), _) | (_, _, Err(true))) { |
| 271 | conflicts.push(path); |
| 272 | continue; |
| 273 | } |
| 274 | return Ok(Applied::Unsupported(format!("{path} is too large to merge line by line"))); |
| 275 | }; |
| 276 | match merge_text(&base, &ours, &theirs) { |
| 277 | Some(merged) => { |
| 278 | let blob = merged.into_bytes(); |
| 279 | let id = object_id(ObjectKind::Blob, &blob); |
| 280 | made.push((ObjectKind::Blob, blob)); |
| 281 | changes.push(Change { path, entry: Some((kind, id)) }); |
| 282 | } |
| 283 | None => conflicts.push(path), |
| 284 | } |
| 285 | } |
| 286 | if !conflicts.is_empty() { |
| 287 | conflicts.sort(); |
| 288 | return Ok(Applied::Conflict(conflicts)); |
| 289 | } |
| 290 | if changes.is_empty() { |
| 291 | return Ok(Applied::Tree(onto.to_owned())); |
| 292 | } |
| 293 | let trees: HashMap<String, Vec<TreeEntry>> = found.into_values().collect(); |
| 294 | let merged = match merge_tree(onto, &trees, &changes) { |
| 295 | Ok(merged) => merged, |
| 296 | Err(why) => return Ok(Applied::Unsupported(why)), |
| 297 | }; |
| 298 | made.extend(merged.objects.into_iter().map(|bytes| (ObjectKind::Tree, bytes))); |
| 299 | git.keep(&made); |
| 300 | objects.extend(made); |
| 301 | Ok(Applied::Tree(merged.tree)) |
| 302 | } |
| 303 | |
| 304 | /// The commits of a branch the branch it merges into does not have, |
| 305 | /// oldest first and each after its parents: those reachable from the |
| 306 | /// newest of `history` (the branch's first-parent chain) and not from |
| 307 | /// `shared` (the other branch's). `None` when there are more than `limit`. |
| 308 | pub(crate) async fn branch_commits<R: GitRepo>( |
| 309 | git: &R, |
| 310 | history: &[Commit], |
| 311 | shared: &HashSet<String>, |
| 312 | limit: usize, |
| 313 | ) -> Result<Option<Vec<Commit>>> { |
| 314 | let mut known: HashMap<String, Commit> = history.iter().map(|commit| (commit.hash.clone(), commit.clone())).collect(); |
| 315 | let Some(head) = history.first() else { |
| 316 | return Ok(Some(Vec::new())); |
| 317 | }; |
| 318 | let mut order = Vec::new(); |
| 319 | let mut seen = HashSet::new(); |
| 320 | // Depth first, a commit written once its parents are: the first |
| 321 | // parent's line before the others. |
| 322 | let mut stack = vec![(head.hash.clone(), false)]; |
| 323 | while let Some((hash, parents_done)) = stack.pop() { |
| 324 | if parents_done { |
| 325 | if let Some(commit) = known.get(&hash) { |
| 326 | order.push(commit.clone()); |
| 327 | } |
| 328 | continue; |
| 329 | } |
| 330 | if shared.contains(&hash) || !seen.insert(hash.clone()) { |
| 331 | continue; |
| 332 | } |
| 333 | if seen.len() > limit { |
| 334 | return Ok(None); |
| 335 | } |
| 336 | let commit = match known.get(&hash) { |
| 337 | Some(commit) => commit.clone(), |
| 338 | None => match git.log(&hash, 1).await?.into_iter().next() { |
| 339 | Some(commit) => { |
| 340 | known.insert(hash.clone(), commit.clone()); |
| 341 | commit |
| 342 | } |
| 343 | // Not there to read: where the branch's history ends. |
| 344 | None => continue, |
| 345 | }, |
| 346 | }; |
| 347 | stack.push((hash, true)); |
| 348 | for parent in commit.parents.iter().rev() { |
| 349 | stack.push((parent.clone(), false)); |
| 350 | } |
| 351 | } |
| 352 | Ok(Some(order)) |
| 353 | } |
| 354 | |
| 355 | /// A branch's commits, made again. |
| 356 | #[derive(Debug)] |
| 357 | pub(crate) struct Replayed { |
| 358 | /// The last commit made, or where they were made onto if none was. |
| 359 | pub head: String, |
| 360 | /// The commits made, oldest first, each with the one it was made from. |
| 361 | pub made: Vec<(String, String)>, |
| 362 | /// The messages and authors of what was made, for the rules. |
| 363 | pub facts: Vec<g1t_contracts::rules::CommitFacts>, |
| 364 | } |
| 365 | |
| 366 | /// How replaying went. |
| 367 | #[derive(Debug)] |
| 368 | pub(crate) enum Replay { |
| 369 | Done(Replayed), |
| 370 | /// A commit's changes clash with what is there: the commit, and the files. |
| 371 | Conflict { commit: Commit, paths: Vec<String> }, |
| 372 | Unsupported { commit: Commit, why: String }, |
| 373 | } |
| 374 | |
| 375 | /// When a commit says it was authored, in seconds. |
| 376 | fn authored_seconds(commit: &Commit) -> u64 { |
| 377 | parse_rfc3339(&commit.authored_at).unwrap_or(0) / 1000 |
| 378 | } |
| 379 | |
| 380 | /// Makes each of `commits` (oldest first) again on top of `onto` (a commit |
| 381 | /// and its tree): its changes applied to the commit before, its author, |
| 382 | /// date and message kept, committed by `committer` with `sign_off` added. |
| 383 | /// Merge commits are left out, their changes already in what they merged, |
| 384 | /// and so is a commit whose changes are all there already. |
| 385 | pub(crate) async fn replay<R: GitRepo>( |
| 386 | git: &Reader<'_, R>, |
| 387 | commits: &[Commit], |
| 388 | onto: (&str, &str), |
| 389 | committer: &Signature<'_>, |
| 390 | sign_off: Option<&str>, |
| 391 | objects: &mut Vec<(ObjectKind, Vec<u8>)>, |
| 392 | ) -> Result<Replay> { |
| 393 | let trees: HashMap<&str, &str> = commits.iter().map(|commit| (commit.hash.as_str(), commit.tree_hash.as_str())).collect(); |
| 394 | let (mut head, mut tree) = (onto.0.to_owned(), onto.1.to_owned()); |
| 395 | let mut made = Vec::new(); |
| 396 | let mut facts = Vec::new(); |
| 397 | for commit in commits.iter().filter(|commit| commit.parents.len() <= 1) { |
| 398 | let from = match commit.parents.first() { |
| 399 | None => None, |
| 400 | Some(parent) => match trees.get(parent.as_str()) { |
| 401 | Some(tree) => Some((*tree).to_owned()), |
| 402 | None => match git.log(parent, 1).await?.into_iter().next() { |
| 403 | Some(parent) => Some(parent.tree_hash), |
| 404 | None => { |
| 405 | return Ok(Replay::Unsupported { commit: commit.clone(), why: format!("its parent {parent} could not be read") }); |
| 406 | } |
| 407 | }, |
| 408 | }, |
| 409 | }; |
| 410 | let applied = apply(git, from.as_deref(), &commit.tree_hash, &tree, objects).await?; |
| 411 | let new_tree = match applied { |
| 412 | Applied::Tree(new_tree) => new_tree, |
| 413 | Applied::Conflict(paths) => return Ok(Replay::Conflict { commit: commit.clone(), paths }), |
| 414 | Applied::Unsupported(why) => return Ok(Replay::Unsupported { commit: commit.clone(), why }), |
| 415 | }; |
| 416 | if new_tree == tree { |
| 417 | continue; |
| 418 | } |
| 419 | let author = Signature { name: &commit.author.name, email: &commit.author.email, seconds: authored_seconds(commit) }; |
| 420 | let message = with_trailers(&commit.message, sign_off.into_iter().map(str::to_owned).collect::<Vec<_>>().as_slice()); |
| 421 | let bytes = commit_object_by(&new_tree, &[&head], &author, committer, &message); |
| 422 | let id = object_id(ObjectKind::Commit, &bytes); |
| 423 | objects.push((ObjectKind::Commit, bytes)); |
| 424 | let mut fact = crate::rules::made_commit(&id, &message, committer.email, 1, Vec::new()); |
| 425 | fact.author_email = Some(commit.author.email.clone()); |
| 426 | facts.push(fact); |
| 427 | made.push((id.clone(), commit.hash.clone())); |
| 428 | head = id; |
| 429 | tree = new_tree; |
| 430 | } |
| 431 | Ok(Replay::Done(Replayed { head, made, facts })) |
| 432 | } |
| 433 | |
| 434 | /// A trailer's key and who or what it names, compared without case: the |
| 435 | /// address in `<...>` when there is one. |
| 436 | fn trailer_parts(line: &str) -> Option<(String, String)> { |
| 437 | let (key, value) = line.trim().split_once(':')?; |
| 438 | if key.is_empty() || !key.chars().all(|c| c.is_ascii_alphanumeric() || c == '-') || !value.starts_with(' ') { |
| 439 | return None; |
| 440 | } |
| 441 | let value = value.trim(); |
| 442 | let named = match (value.rfind('<'), value.rfind('>')) { |
| 443 | (Some(open), Some(close)) if open < close => &value[open + 1..close], |
| 444 | _ => value, |
| 445 | }; |
| 446 | Some((key.to_ascii_lowercase(), named.to_lowercase())) |
| 447 | } |
| 448 | |
| 449 | /// `message` with `trailers` added at its end, each unless one with the |
| 450 | /// same key naming the same address is there already. They join the |
| 451 | /// message's own trailers when its last paragraph is made of them, and |
| 452 | /// otherwise start a paragraph of their own. |
| 453 | pub(crate) fn with_trailers(message: &str, trailers: &[String]) -> String { |
| 454 | let message = message.trim_end(); |
| 455 | let mut present: Vec<(String, String)> = message.lines().filter_map(trailer_parts).collect(); |
| 456 | let mut adding: Vec<&str> = Vec::new(); |
| 457 | for trailer in trailers { |
| 458 | match trailer_parts(trailer) { |
| 459 | Some(parts) if present.contains(&parts) => {} |
| 460 | Some(parts) => { |
| 461 | present.push(parts); |
| 462 | adding.push(trailer); |
| 463 | } |
| 464 | None => adding.push(trailer), |
| 465 | } |
| 466 | } |
| 467 | if adding.is_empty() { |
| 468 | return message.to_owned(); |
| 469 | } |
| 470 | if message.is_empty() { |
| 471 | return adding.join("\n"); |
| 472 | } |
| 473 | let ends_in_trailers = message |
| 474 | .rsplit_once("\n\n") |
| 475 | .is_some_and(|(_, last)| !last.trim().is_empty() && last.lines().all(|line| trailer_parts(line).is_some())); |
| 476 | let gap = if ends_in_trailers { "\n" } else { "\n\n" }; |
| 477 | format!("{message}{gap}{}", adding.join("\n")) |
| 478 | } |
| 479 | |
| 480 | /// The `Signed-off-by:` trailer for someone. |
| 481 | pub(crate) fn sign_off_trailer(name: &str, email: &str) -> String { |
| 482 | format!("Signed-off-by: {name} <{email}>") |
| 483 | } |
| 484 | |
| 485 | /// A `Co-authored-by:` trailer for everyone who wrote one of `commits` |
| 486 | /// other than `author` (by address) and g1t itself, once each, in the |
| 487 | /// order they first appear. Merge commits change nothing of their own and |
| 488 | /// credit no one. |
| 489 | pub(crate) fn co_authors(commits: &[Commit], author: &str) -> Vec<String> { |
| 490 | let mut seen: HashSet<String> = [author.to_lowercase(), g1t_contracts::system::EMAIL.to_owned()].into_iter().collect(); |
| 491 | commits |
| 492 | .iter() |
| 493 | .filter(|commit| commit.parents.len() <= 1 && !commit.author.email.trim().is_empty()) |
| 494 | .filter(|commit| seen.insert(commit.author.email.trim().to_lowercase())) |
| 495 | .map(|commit| format!("Co-authored-by: {} <{}>", commit.author.name, commit.author.email.trim())) |
| 496 | .collect() |
| 497 | } |
| 498 | |
| 499 | /// `message` with its `Co-authored-by:` trailers made right, as a |
| 500 | /// template may have written them from addresses alone: none for the |
| 501 | /// commit's own author (`author_email`) or g1t, and one naming someone in |
| 502 | /// `known` (trailers g1t made from the commits) as that one names them. |
| 503 | pub(crate) fn tidy_co_authors(message: &str, author_email: &str, known: &[String]) -> String { |
| 504 | let email_of = |line: &str| { |
| 505 | line.rsplit_once('<') |
| 506 | .and_then(|(_, rest)| rest.split_once('>')) |
| 507 | .map(|(email, _)| email.trim().to_lowercase()) |
| 508 | }; |
| 509 | let author = author_email.trim().to_lowercase(); |
| 510 | message |
| 511 | .lines() |
| 512 | .filter_map(|line| { |
| 513 | if !line.trim_start().to_lowercase().starts_with("co-authored-by:") { |
| 514 | return Some(line.to_owned()); |
| 515 | } |
| 516 | let Some(email) = email_of(line) else { return Some(line.to_owned()) }; |
| 517 | if email == author || email == g1t_contracts::system::EMAIL { |
| 518 | return None; |
| 519 | } |
| 520 | Some(known.iter().find(|trailer| email_of(trailer).as_deref() == Some(email.as_str())).cloned().unwrap_or_else(|| line.to_owned())) |
| 521 | }) |
| 522 | .collect::<Vec<_>>() |
| 523 | .join("\n") |
| 524 | } |
| 525 | |
| 526 | /// Files named in a sentence: a few, then how many more. |
| 527 | pub(crate) fn name_files(paths: &[String]) -> String { |
| 528 | const SHOWN: usize = 5; |
| 529 | match paths { |
| 530 | [] => "some files".to_owned(), |
| 531 | [one] => one.clone(), |
| 532 | _ if paths.len() <= SHOWN => { |
| 533 | let (last, rest) = paths.split_last().expect("two or more"); |
| 534 | format!("{} and {last}", rest.join(", ")) |
| 535 | } |
| 536 | _ => format!("{} and {} more files", paths[..SHOWN].join(", "), paths.len() - SHOWN), |
| 537 | } |
| 538 | } |
| 539 | |
| 540 | /// A commit's short hash. |
| 541 | fn short(hash: &str) -> &str { |
| 542 | &hash[..hash.len().min(7)] |
| 543 | } |
| 544 | |
| 545 | /// A commit as people know it: its short hash and title. |
| 546 | fn named_commit(commit: &Commit) -> String { |
| 547 | let title = commit.message.lines().next().unwrap_or_default().trim(); |
| 548 | format!("{} ({title})", &commit.hash[..commit.hash.len().min(7)]) |
| 549 | } |
| 550 | |
| 551 | /// What was made to land, ready to push. |
| 552 | struct Made { |
| 553 | commit: String, |
| 554 | rebased: Vec<(String, String)>, |
| 555 | /// How many commits it brings, when that is known. |
| 556 | commits: Option<u32>, |
| 557 | } |
| 558 | |
| 559 | /// Why a landing was refused, or that it was there already. |
| 560 | enum NotMade { |
| 561 | Refused(Outcome<Landed>), |
| 562 | /// Lines clash in these files. |
| 563 | Clash { message: String, paths: Vec<String> }, |
| 564 | /// Nothing to make: the branch already has every change. |
| 565 | AlreadyThere, |
| 566 | } |
| 567 | |
| 568 | fn refused(code: FailureCode, message: String) -> NotMade { |
| 569 | NotMade::Refused(Outcome::fail(code, message)) |
| 570 | } |
| 571 | |
| 572 | /// The pack that sends what was made: with what the pull request's commits |
| 573 | /// bring that the repository lacks, from the fork, when it does not hold |
| 574 | /// them (`held`). |
| 575 | async fn pack_for<R: GitRepo>( |
| 576 | held: bool, |
| 577 | source_git: &R, |
| 578 | new: &str, |
| 579 | merge_base: &str, |
| 580 | objects: &[(ObjectKind, Vec<u8>)], |
| 581 | ) -> Result<Vec<u8>> { |
| 582 | if held { |
| 583 | return Ok(write_pack(objects)); |
| 584 | } |
| 585 | let source_access = source_git.access(Scope::Read).await?; |
| 586 | let fetched = land::fetch_pack(&source_access, new, Some(merge_base)).await?; |
| 587 | extend_pack(&fetched, objects).map_err(worker::Error::RustError) |
| 588 | } |
| 589 | |
| 590 | /// What `method` (any but `fast_forward`) makes of a pull request's |
| 591 | /// commits (`history`, its first-parent chain, headed by `new`) on the |
| 592 | /// head of `target_history`, with the objects made added to `objects`; |
| 593 | /// and the merge base, which a pack from a fork starts at. `branch` |
| 594 | /// names where it lands in what is said about a refusal. |
| 595 | #[allow(clippy::too_many_arguments)] |
| 596 | async fn make_on<R: GitRepo>( |
| 597 | method: MergeMethod, |
| 598 | reader: &Reader<'_, R>, |
| 599 | target_history: &[Commit], |
| 600 | history: &[Commit], |
| 601 | new: &str, |
| 602 | message: &str, |
| 603 | (author, by): (&Signature<'_>, &Signature<'_>), |
| 604 | sign_off: Option<&str>, |
| 605 | branch: &str, |
| 606 | objects: &mut Vec<(ObjectKind, Vec<u8>)>, |
| 607 | ) -> Result<std::result::Result<(Made, String), NotMade>> { |
| 608 | let Some(base) = target_history.first() else { |
| 609 | return Ok(Err(refused(FailureCode::Conflict, format!("{branch} has no commits yet.")))); |
| 610 | }; |
| 611 | let shared: HashSet<String> = target_history.iter().map(|commit| commit.hash.clone()).collect(); |
| 612 | let merge_base = nearest_ancestor_in(reader, history, &shared).await?; |
| 613 | let Some((merge_base, merge_base_tree)) = merge_base.and_then(|hash| { |
| 614 | target_history.iter().find(|commit| commit.hash == hash).map(|commit| (hash, commit.tree_hash.clone())) |
| 615 | }) else { |
| 616 | return Ok(Err(refused( |
| 617 | FailureCode::Conflict, |
| 618 | format!("g1t could not find where this pull request left {branch}, so it cannot make its commits there."), |
| 619 | ))); |
| 620 | }; |
| 621 | let made = match method { |
| 622 | MergeMethod::Merge | MergeMethod::Squash => { |
| 623 | let squash = method == MergeMethod::Squash; |
| 624 | let co_authored = if squash { |
| 625 | let commits = match branch_commits(reader, history, &shared, MAX_ANCESTRY as usize).await? { |
| 626 | Some(commits) => commits, |
| 627 | // Too many to walk: those on the branch's own line. |
| 628 | None => { |
| 629 | let mut own: Vec<Commit> = history.iter().take_while(|commit| !shared.contains(&commit.hash)).cloned().collect(); |
| 630 | own.reverse(); |
| 631 | own |
| 632 | } |
| 633 | }; |
| 634 | co_authors(&commits, author.email) |
| 635 | } else { |
| 636 | Vec::new() |
| 637 | }; |
| 638 | let head = &history[0]; |
| 639 | made_merge( |
| 640 | reader, |
| 641 | (&merge_base_tree, &head.tree_hash, base), |
| 642 | new, |
| 643 | squash, |
| 644 | message, |
| 645 | &co_authored, |
| 646 | sign_off, |
| 647 | (author, by), |
| 648 | branch, |
| 649 | objects, |
| 650 | ) |
| 651 | .await? |
| 652 | } |
| 653 | _ => { |
| 654 | let commits = match branch_commits(reader, history, &shared, MAX_REBASE_COMMITS).await? { |
| 655 | Some(commits) => commits, |
| 656 | None => { |
| 657 | return Ok(Err(refused( |
| 658 | FailureCode::Conflict, |
| 659 | format!( |
| 660 | "This pull request has more than {MAX_REBASE_COMMITS} commits, too many for g1t to rebase itself. Squash or merge it instead." |
| 661 | ), |
| 662 | ))); |
| 663 | } |
| 664 | }; |
| 665 | match replay(reader, &commits, (&base.hash, &base.tree_hash), by, sign_off, objects).await? { |
| 666 | Replay::Done(replayed) if replayed.made.is_empty() => Err(NotMade::AlreadyThere), |
| 667 | Replay::Done(replayed) => Ok(Made { |
| 668 | commit: replayed.head, |
| 669 | commits: Some(replayed.made.len() as u32), |
| 670 | rebased: replayed.made, |
| 671 | }), |
| 672 | Replay::Conflict { commit, paths } => Err(NotMade::Clash { |
| 673 | message: format!( |
| 674 | "Rebasing stopped at {}: it and {branch} changed the same lines of {}. Rebase the branch onto {branch} and resolve them, or merge it another way.", |
| 675 | named_commit(&commit), |
| 676 | name_files(&paths) |
| 677 | ), |
| 678 | paths, |
| 679 | }), |
| 680 | Replay::Unsupported { commit, why } => Err(refused( |
| 681 | FailureCode::Conflict, |
| 682 | format!("g1t cannot rebase {} itself: {why}. Rebase the branch onto {branch}, or merge it another way.", named_commit(&commit)), |
| 683 | )), |
| 684 | } |
| 685 | } |
| 686 | }; |
| 687 | Ok(made.map(|made| (made, merge_base))) |
| 688 | } |
| 689 | |
| 690 | /// A merge commit (parents `base` and `new`) or a squashed commit |
| 691 | /// (parent `base`) whose tree is the pull request's changes since |
| 692 | /// `trees.0` (the merge base's tree; `trees.1` is its head's) applied |
| 693 | /// to `base`'s. |
| 694 | #[allow(clippy::too_many_arguments)] |
| 695 | async fn made_merge<R: GitRepo>( |
| 696 | reader: &Reader<'_, R>, |
| 697 | trees: (&str, &str, &Commit), |
| 698 | new: &str, |
| 699 | squash: bool, |
| 700 | message: &str, |
| 701 | co_authored: &[String], |
| 702 | sign_off: Option<&str>, |
| 703 | (author, committer): (&Signature<'_>, &Signature<'_>), |
| 704 | branch: &str, |
| 705 | objects: &mut Vec<(ObjectKind, Vec<u8>)>, |
| 706 | ) -> Result<std::result::Result<Made, NotMade>> { |
| 707 | let (merge_base_tree, head_tree, base) = trees; |
| 708 | let verb = if squash { "squash" } else { "merge" }; |
| 709 | let tree = match apply(reader, Some(merge_base_tree), head_tree, &base.tree_hash, objects).await? { |
| 710 | Applied::Tree(tree) => tree, |
| 711 | Applied::Conflict(paths) => { |
| 712 | return Ok(Err(NotMade::Clash { |
| 713 | message: format!( |
| 714 | "This pull request and {branch} changed the same lines of {}. Bring the pull request up to date with {branch} and resolve them, then merge again.", |
| 715 | name_files(&paths) |
| 716 | ), |
| 717 | paths, |
| 718 | })); |
| 719 | } |
| 720 | Applied::Unsupported(why) => { |
| 721 | return Ok(Err(refused( |
| 722 | FailureCode::Conflict, |
| 723 | format!("g1t cannot {verb} this pull request itself: {why}. Bring it up to date with {branch}, or merge it another way."), |
| 724 | ))); |
| 725 | } |
| 726 | }; |
| 727 | // Squashing what is there already would make an empty commit. |
| 728 | if squash && tree == base.tree_hash { |
| 729 | return Ok(Err(NotMade::AlreadyThere)); |
| 730 | } |
| 731 | let mut trailers: Vec<String> = co_authored.to_vec(); |
| 732 | trailers.extend(sign_off.map(str::to_owned)); |
| 733 | let by = if squash { &author.email } else { &committer.email }; |
| 734 | let message = with_trailers(&tidy_co_authors(message, by, co_authored), &trailers); |
| 735 | let bytes = if squash { |
| 736 | commit_object_by(&tree, &[&base.hash], author, committer, &message) |
| 737 | } else { |
| 738 | commit_object_by(&tree, &[&base.hash, new], committer, committer, &message) |
| 739 | }; |
| 740 | let commit = object_id(ObjectKind::Commit, &bytes); |
| 741 | objects.push((ObjectKind::Commit, bytes)); |
| 742 | Ok(Ok(Made { commit, rebased: Vec::new(), commits: squash.then_some(1) })) |
| 743 | } |
| 744 | |
| 745 | /// What one merge queue entry made: the state with it in, and the merge |
| 746 | /// base a pack from its fork starts at; or why it cannot be built. |
| 747 | #[derive(Debug)] |
| 748 | pub(crate) enum StepMade { |
| 749 | Made { commit: String, rebased: Vec<(String, String)>, merge_base: String }, |
| 750 | Conflict { message: String, paths: Vec<String> }, |
| 751 | Refused(String), |
| 752 | } |
| 753 | |
| 754 | /// One merge queue entry (`history`, its first-parent chain from the |
| 755 | /// commit queued) landed by `method` on the head of `onto_history`, a |
| 756 | /// state of the queue: its commits made as landing would make them there, |
| 757 | /// with the objects made added to `objects`. A fast-forward makes none and |
| 758 | /// needs its head to contain that state; an entry whose changes are all |
| 759 | /// there already leaves the state as it is. `held`: the repository holds |
| 760 | /// the entry's commits (`git` reads a fork's otherwise). |
| 761 | #[allow(clippy::too_many_arguments)] |
| 762 | pub(crate) async fn queue_step_on<R: GitRepo>( |
| 763 | git: &Reader<'_, R>, |
| 764 | held: bool, |
| 765 | method: MergeMethod, |
| 766 | (onto_history, history): (&[Commit], &[Commit]), |
| 767 | message: Option<&str>, |
| 768 | signatures: (&Signature<'_>, &Signature<'_>), |
| 769 | sign_off: Option<&str>, |
| 770 | place: &str, |
| 771 | objects: &mut Vec<(ObjectKind, Vec<u8>)>, |
| 772 | ) -> Result<StepMade> { |
| 773 | let (Some(head), Some(base)) = (history.first(), onto_history.first()) else { |
| 774 | return Ok(StepMade::Refused("Its commit or the state it is built on is no longer there.".to_owned())); |
| 775 | }; |
| 776 | let there = |commit: &str| StepMade::Made { commit: commit.to_owned(), rebased: Vec::new(), merge_base: base.hash.clone() }; |
| 777 | if base.hash == head.hash || (held && descends_from(git, onto_history, &head.hash).await?) { |
| 778 | return Ok(there(&base.hash)); |
| 779 | } |
| 780 | if method == MergeMethod::FastForward { |
| 781 | return Ok(if descends_from(git, history, &base.hash).await? { |
| 782 | there(&head.hash) |
| 783 | } else { |
| 784 | StepMade::Refused(format!( |
| 785 | "It lands by fast-forward, so its branch must contain {place} with everything queued ahead of it. Bring it up to date, or merge it another way." |
| 786 | )) |
| 787 | }); |
| 788 | } |
| 789 | let message = match message.map(str::trim_end).filter(|text| !text.trim().is_empty()) { |
| 790 | Some(text) => text.to_owned(), |
| 791 | None if method == MergeMethod::Rebase => String::new(), |
| 792 | None => return Ok(StepMade::Refused("Say what the commit's message is: its first line is its title.".to_owned())), |
| 793 | }; |
| 794 | Ok(match make_on(method, git, onto_history, history, &head.hash, &message, signatures, sign_off, place, objects).await? { |
| 795 | Ok((made, merge_base)) => StepMade::Made { commit: made.commit, rebased: made.rebased, merge_base }, |
| 796 | Err(NotMade::AlreadyThere) => there(&base.hash), |
| 797 | Err(NotMade::Clash { message, paths }) => StepMade::Conflict { message, paths }, |
| 798 | Err(NotMade::Refused(Outcome::Fail(failure))) => StepMade::Refused(failure.message), |
| 799 | Err(NotMade::Refused(Outcome::Ok(_))) => StepMade::Refused("It could not be built.".to_owned()), |
| 800 | }) |
| 801 | } |
| 802 | |
| 803 | impl<S: GitStore> Repos<S> { |
| 804 | /// `land` by `method`, any but `fast_forward`: makes the commits on |
| 805 | /// `branch` of `target` from `new`, the checked head of the pull |
| 806 | /// request's branch (`history`, its first-parent chain, in |
| 807 | /// `source_git`), and moves `branch` to the last of them if it is still |
| 808 | /// where it was. If another change lands first, it is worked out again |
| 809 | /// once on the new head. The caller has checked who may do it. |
| 810 | #[allow(clippy::too_many_arguments)] |
| 811 | pub(crate) async fn land_by( |
| 812 | &self, |
| 813 | method: MergeMethod, |
| 814 | a: &LandArgs, |
| 815 | target: &Repo, |
| 816 | source_git: &S::Repo, |
| 817 | target_git: &S::Repo, |
| 818 | branch: &str, |
| 819 | history: &[Commit], |
| 820 | new: String, |
| 821 | from_fork: bool, |
| 822 | ) -> Result<Outcome<Landed>> { |
| 823 | let message = match method { |
| 824 | MergeMethod::Merge | MergeMethod::Squash => match a.message.as_deref().map(str::trim_end).filter(|text| !text.trim().is_empty()) { |
| 825 | Some(text) => text.to_owned(), |
| 826 | None => { |
| 827 | return Ok(Outcome::fail(FailureCode::Invalid, "Say what the commit's message is: its first line is its title.")); |
| 828 | } |
| 829 | }, |
| 830 | _ => String::new(), |
| 831 | }; |
| 832 | let committer = self.commit_identity(&a.actor).await; |
| 833 | let author = match (&a.author, method) { |
| 834 | (Some(author), MergeMethod::Squash) => self.commit_identity(author).await, |
| 835 | _ => committer.clone(), |
| 836 | }; |
| 837 | let sign_off = a.sign_off.then(|| sign_off_trailer(&committer.name, &committer.email)); |
| 838 | // Whether the repository holds the pull request's commits: always |
| 839 | // for its own branch, and for a fork's once its head was kept there |
| 840 | // (test_merge.rs). Otherwise they are read from the fork, and sent |
| 841 | // with what is made. |
| 842 | let held = !from_fork || !target_git.log(&new, 1).await?.is_empty(); |
| 843 | let reader = Reader::new(target_git, (!held).then_some(source_git)); |
| 844 | let mut attempt = 0; |
| 845 | loop { |
| 846 | attempt += 1; |
| 847 | let target_history = target_git.log(branch, MAX_ANCESTRY).await?; |
| 848 | let Some(base) = target_history.first().cloned() else { |
| 849 | return Ok(Outcome::fail( |
| 850 | FailureCode::Conflict, |
| 851 | format!("{branch} has no commits yet, so there is nothing to make this pull request's commits on. Land it by fast-forward instead."), |
| 852 | )); |
| 853 | }; |
| 854 | // There already: an attempt that published and stopped before |
| 855 | // the merge was recorded, or a branch with nothing new. |
| 856 | if base.hash == new || (held && descends_from(target_git, &target_history, &new).await?) { |
| 857 | let commit = match method { |
| 858 | MergeMethod::Merge => target_history |
| 859 | .iter() |
| 860 | .find(|commit| commit.parents.get(1) == Some(&new)) |
| 861 | .map_or_else(|| base.hash.clone(), |commit| commit.hash.clone()), |
| 862 | _ => base.hash.clone(), |
| 863 | }; |
| 864 | return Ok(Outcome::Ok(Landed { commit, previous: None, method, rebased: Vec::new() })); |
| 865 | } |
| 866 | let mut objects: Vec<(ObjectKind, Vec<u8>)> = Vec::new(); |
| 867 | let seconds = now_ms() / 1000; |
| 868 | let by = Signature { name: &committer.name, email: &committer.email, seconds }; |
| 869 | let author = Signature { name: &author.name, email: &author.email, seconds }; |
| 870 | let made = |
| 871 | make_on(method, &reader, &target_history, history, &new, &message, (&author, &by), sign_off.as_deref(), branch, &mut objects) |
| 872 | .await?; |
| 873 | let (made, merge_base) = match made { |
| 874 | Ok(made) => made, |
| 875 | Err(NotMade::Refused(outcome)) => return Ok(outcome), |
| 876 | Err(NotMade::Clash { message, .. }) => return Ok(Outcome::fail(FailureCode::Conflict, message)), |
| 877 | Err(NotMade::AlreadyThere) => { |
| 878 | return Ok(Outcome::Ok(Landed { commit: base.hash.clone(), previous: None, method, rebased: Vec::new() })); |
| 879 | } |
| 880 | }; |
| 881 | let pack = pack_for(held, source_git, &new, &merge_base, &objects).await?; |
| 882 | let target_access = target_git.access(Scope::Write).await?; |
| 883 | if !self.registry.reserve_public(target, pack.len() as u64, self.free_public_bytes).await? { |
| 884 | return Ok(Outcome::fail(FailureCode::Forbidden, "Public repository storage is full. Make room before adding more data.")); |
| 885 | } |
| 886 | let bytes = pack.len() as u64; |
| 887 | let pushed = land::push_pack(&target_access, branch, Some(&base.hash), &made.commit, pack).await?; |
| 888 | self.registry.finish_public(target, bytes, bytes).await?; |
| 889 | self.refs_moved(&target.id).await; |
| 890 | match pushed { |
| 891 | Ok(()) => { |
| 892 | self.publish_push(target, &format!("refs/heads/{branch}"), Some(&base.hash), &made.commit, Some(&a.actor), made.commits) |
| 893 | .await?; |
| 894 | return Ok(Outcome::Ok(Landed { |
| 895 | commit: made.commit, |
| 896 | previous: Some(base.hash.clone()), |
| 897 | method, |
| 898 | rebased: made.rebased, |
| 899 | })); |
| 900 | } |
| 901 | // Most often another pull request landed meanwhile: once |
| 902 | // more, on top of it. |
| 903 | Err(_) if attempt < 2 => continue, |
| 904 | Err(reason) => { |
| 905 | return Ok(Outcome::fail(FailureCode::Conflict, format!("{branch} could not be updated: {reason}"))); |
| 906 | } |
| 907 | } |
| 908 | } |
| 909 | } |
| 910 | |
| 911 | /// `build_queue_step`: one merge queue entry landed by its method on |
| 912 | /// `onto`, a commit of the target, as `land_by` would land it on a |
| 913 | /// branch whose head that is, and `push_branch` pointed at what was |
| 914 | /// made. Nothing else moves. The work service decides who may merge; |
| 915 | /// the committer must still be able to. |
| 916 | pub(crate) async fn build_queue_step(&self, a: BuildQueueStepArgs) -> Result<Outcome<QueueStep>> { |
| 917 | let actor = Some(a.actor.clone()); |
| 918 | let (Some(target), Some(source)) = (self.registry.by_id(&a.target_id).await?, self.registry.by_id(&a.source_id).await?) else { |
| 919 | return Ok(Outcome::fail(FailureCode::NotFound, "No such repository.")); |
| 920 | }; |
| 921 | if !crate::registry::can(&target, &actor, g1t_contracts::access::Capability::Merge) || !a.actor.verified { |
| 922 | return Ok(Outcome::fail(FailureCode::Forbidden, "Whoever queued it can no longer merge here.")); |
| 923 | } |
| 924 | // A fork's change lands only on a repository in its network. |
| 925 | if source.id != target.id && self.pull_target(&source, Some(&target.id)).await?.is_none_or(|found| found.id != target.id) { |
| 926 | return Ok(Outcome::fail(FailureCode::NotFound, "No such repository.")); |
| 927 | } |
| 928 | let from_fork = source.id != target.id; |
| 929 | self.live(&source).await?; |
| 930 | let source_git = self.store.open(&store_key(&source)).await?; |
| 931 | let target_git = self.store.open(&store_key(&target)).await?; |
| 932 | let refused = |message: String| Ok(Outcome::Ok(QueueStep::Refused { message })); |
| 933 | // What was queued is the commit: its branch may have moved since. |
| 934 | let (history, onto_history) = try_join(source_git.log(&a.commit, MAX_ANCESTRY), target_git.log(&a.onto, MAX_ANCESTRY)).await?; |
| 935 | let (Some(head), Some(base)) = (history.first(), onto_history.first()) else { |
| 936 | return refused(format!("{} or the state it is built on is no longer there.", short(&a.commit))); |
| 937 | }; |
| 938 | let new = head.hash.clone(); |
| 939 | let base = base.hash.clone(); |
| 940 | let held = !from_fork || !target_git.log(&new, 1).await?.is_empty(); |
| 941 | let reader = Reader::new(&target_git, (!held).then_some(&source_git)); |
| 942 | let place = target.default_branch.clone(); |
| 943 | let mut objects: Vec<(ObjectKind, Vec<u8>)> = Vec::new(); |
| 944 | let committer = self.commit_identity(&a.actor).await; |
| 945 | let author = match (&a.author, a.method) { |
| 946 | (Some(author), MergeMethod::Squash) => self.commit_identity(author).await, |
| 947 | _ => committer.clone(), |
| 948 | }; |
| 949 | let sign_off = a.sign_off.then(|| sign_off_trailer(&committer.name, &committer.email)); |
| 950 | let seconds = now_ms() / 1000; |
| 951 | let by = Signature { name: &committer.name, email: &committer.email, seconds }; |
| 952 | let author = Signature { name: &author.name, email: &author.email, seconds }; |
| 953 | let step = queue_step_on( |
| 954 | &reader, |
| 955 | held, |
| 956 | a.method, |
| 957 | (&onto_history, &history), |
| 958 | a.message.as_deref(), |
| 959 | (&author, &by), |
| 960 | sign_off.as_deref(), |
| 961 | &place, |
| 962 | &mut objects, |
| 963 | ) |
| 964 | .await?; |
| 965 | let (commit, rebased, merge_base) = match step { |
| 966 | StepMade::Made { commit, rebased, merge_base } => (commit, rebased, merge_base), |
| 967 | StepMade::Conflict { message, paths } => return Ok(Outcome::Ok(QueueStep::Conflict { message, paths })), |
| 968 | StepMade::Refused(message) => return refused(message), |
| 969 | }; |
| 970 | // Nothing new when it was there already. |
| 971 | let pack = if commit == base { |
| 972 | write_pack(&objects) |
| 973 | } else { |
| 974 | pack_for(held, &source_git, &new, &merge_base, &objects).await? |
| 975 | }; |
| 976 | let old = target_git.log(&a.push_branch, 1).await?.into_iter().next().map(|commit| commit.hash); |
| 977 | let target_access = target_git.access(Scope::Write).await?; |
| 978 | if !self.registry.reserve_public(&target, pack.len() as u64, self.free_public_bytes).await? { |
| 979 | return Ok(Outcome::fail(FailureCode::Forbidden, "Public repository storage is full. Make room before adding more data.")); |
| 980 | } |
| 981 | let bytes = pack.len() as u64; |
| 982 | let pushed = land::push_pack(&target_access, &a.push_branch, old.as_deref(), &commit, pack).await?; |
| 983 | self.registry.finish_public(&target, bytes, bytes).await?; |
| 984 | self.refs_moved(&target.id).await; |
| 985 | Ok(match pushed { |
| 986 | Ok(()) => Outcome::Ok(QueueStep::Built { commit, rebased }), |
| 987 | Err(reason) => Outcome::fail(FailureCode::Conflict, format!("{} could not be updated: {reason}", a.push_branch)), |
| 988 | }) |
| 989 | } |
| 990 | |
| 991 | /// `update_pull_branch` by `rebase`: makes the commits of `branch` of |
| 992 | /// `source` again on the head of `base_branch` of `target` and moves |
| 993 | /// the branch there (a force push), if it is still at `expected_head` |
| 994 | /// (else its head when read). The caller has checked who may do it. |
| 995 | pub(crate) async fn rebase_up( |
| 996 | &self, |
| 997 | source: Repo, |
| 998 | target: Repo, |
| 999 | branch: String, |
| 1000 | base_branch: String, |
| 1001 | actor: &User, |
| 1002 | expected_head: Option<&str>, |
| 1003 | ) -> Result<Outcome<PullBranchUpdate>> { |
| 1004 | let from_fork = source.id != target.id; |
| 1005 | // A working copy removed after its pull request closed is made again. |
| 1006 | self.live(&source).await?; |
| 1007 | let source_git = self.store.open(&store_key(&source)).await?; |
| 1008 | let target_git = self.store.open(&store_key(&target)).await?; |
| 1009 | let (history, target_history) = |
| 1010 | try_join(source_git.log(&branch, MAX_ANCESTRY), target_git.log(&base_branch, MAX_ANCESTRY)).await?; |
| 1011 | let (Some(head), Some(base)) = (history.first(), target_history.first()) else { |
| 1012 | return Ok(Outcome::fail(FailureCode::Conflict, format!("{branch} has no commits to bring up to date."))); |
| 1013 | }; |
| 1014 | if let Some(refused) = moved_since(&branch, &head.hash, expected_head) { |
| 1015 | return Ok(refused); |
| 1016 | } |
| 1017 | let needs_agent = |reason, detail: String, paths| Outcome::Ok(PullBranchUpdate::NeedsAgent { reason, detail, paths }); |
| 1018 | let shared: HashSet<String> = target_history.iter().map(|commit| commit.hash.clone()).collect(); |
| 1019 | let reader = Reader::new(&source_git, from_fork.then_some(&target_git)); |
| 1020 | let Some(merge_base) = nearest_ancestor_in(&reader, &history, &shared).await? else { |
| 1021 | return Ok(needs_agent( |
| 1022 | NeedsAgentReason::Unsupported, |
| 1023 | format!("g1t could not find where this pull request left {base_branch}."), |
| 1024 | Vec::new(), |
| 1025 | )); |
| 1026 | }; |
| 1027 | // On the base's head already, with no merge commits to straighten. |
| 1028 | let commits = match branch_commits(&reader, &history, &shared, MAX_REBASE_COMMITS).await? { |
| 1029 | Some(commits) => commits, |
| 1030 | None => { |
| 1031 | return Ok(needs_agent( |
| 1032 | NeedsAgentReason::Unsupported, |
| 1033 | format!("{branch} has more than {MAX_REBASE_COMMITS} commits, too many for g1t to rebase itself."), |
| 1034 | Vec::new(), |
| 1035 | )); |
| 1036 | } |
| 1037 | }; |
| 1038 | if merge_base == base.hash && commits.iter().all(|commit| commit.parents.len() <= 1) { |
| 1039 | return Ok(Outcome::Ok(PullBranchUpdate::UpToDate { commit: head.hash.clone() })); |
| 1040 | } |
| 1041 | let committer = self.commit_identity(actor).await; |
| 1042 | let by = Signature { name: &committer.name, email: &committer.email, seconds: now_ms() / 1000 }; |
| 1043 | let mut objects: Vec<(ObjectKind, Vec<u8>)> = Vec::new(); |
| 1044 | let replayed = match replay(&reader, &commits, (&base.hash, &base.tree_hash), &by, None, &mut objects).await? { |
| 1045 | Replay::Done(replayed) => replayed, |
| 1046 | Replay::Conflict { commit, paths } => { |
| 1047 | return Ok(needs_agent( |
| 1048 | NeedsAgentReason::Conflicting, |
| 1049 | format!("{} and {base_branch} changed the same lines of some files.", named_commit(&commit)), |
| 1050 | paths, |
| 1051 | )); |
| 1052 | } |
| 1053 | Replay::Unsupported { commit, why } => { |
| 1054 | return Ok(needs_agent( |
| 1055 | NeedsAgentReason::Unsupported, |
| 1056 | format!("g1t could not rebase {} itself: {why}.", named_commit(&commit)), |
| 1057 | Vec::new(), |
| 1058 | )); |
| 1059 | } |
| 1060 | }; |
| 1061 | let update = BranchUpdate { |
| 1062 | branch: &branch, |
| 1063 | head: &head.hash, |
| 1064 | commit: &replayed.head, |
| 1065 | fast_forward: false, |
| 1066 | commits: replayed.made.len() as u32, |
| 1067 | }; |
| 1068 | if let Some(refused) = self.update_refused(&source, actor, &update, replayed.facts.clone(), true).await? { |
| 1069 | return Ok(refused); |
| 1070 | } |
| 1071 | // A fork lacks what the base branch gained since it was made. |
| 1072 | let pack = if from_fork { |
| 1073 | let target_access = target_git.access(Scope::Read).await?; |
| 1074 | let fetched = land::fetch_pack(&target_access, &base.hash, Some(&merge_base)).await?; |
| 1075 | extend_pack(&fetched, &objects).map_err(worker::Error::RustError)? |
| 1076 | } else { |
| 1077 | write_pack(&objects) |
| 1078 | }; |
| 1079 | self.push_update(&source, &source_git, actor, &update, pack).await |
| 1080 | } |
| 1081 | } |
| 1082 | |
| 1083 | #[cfg(test)] |
| 1084 | mod tests { |
| 1085 | use super::*; |
| 1086 | use g1t_contracts::repos::Signature as Person; |
| 1087 | |
| 1088 | fn blob(hash: &str) -> Option<FileEntry> { |
| 1089 | Some((EntryKind::Blob, hash.to_owned())) |
| 1090 | } |
| 1091 | |
| 1092 | #[test] |
| 1093 | fn a_file_the_other_side_left_alone_takes_the_change() { |
| 1094 | assert_eq!(resolve(blob("1").as_ref(), blob("2").as_ref(), blob("1").as_ref()), Resolution::Take(blob("2"))); |
| 1095 | // Added, where nothing was. |
| 1096 | assert_eq!(resolve(None, blob("2").as_ref(), None), Resolution::Take(blob("2"))); |
| 1097 | // Deleted. |
| 1098 | assert_eq!(resolve(blob("1").as_ref(), None, blob("1").as_ref()), Resolution::Take(None)); |
| 1099 | // Made the same on both sides. |
| 1100 | assert_eq!(resolve(blob("1").as_ref(), blob("2").as_ref(), blob("2").as_ref()), Resolution::Keep); |
| 1101 | } |
| 1102 | |
| 1103 | #[test] |
| 1104 | fn a_file_both_changed_is_merged_or_conflicts() { |
| 1105 | assert_eq!( |
| 1106 | resolve(blob("1").as_ref(), blob("2").as_ref(), blob("3").as_ref()), |
| 1107 | Resolution::Content { kind: EntryKind::Blob, base: Some("1".into()), ours: "2".into(), theirs: "3".into() } |
| 1108 | ); |
| 1109 | // Deleted on one side, changed on the other. |
| 1110 | assert_eq!(resolve(blob("1").as_ref(), None, blob("3").as_ref()), Resolution::Conflict); |
| 1111 | assert_eq!(resolve(blob("1").as_ref(), blob("2").as_ref(), None), Resolution::Conflict); |
| 1112 | // A file where the other side has a directory. |
| 1113 | assert_eq!(resolve(None, blob("2").as_ref(), Some(&(EntryKind::Tree, "t".into()))), Resolution::Conflict); |
| 1114 | // Submodules moved two ways. |
| 1115 | let link = |hash: &str| Some((EntryKind::Gitlink, hash.to_owned())); |
| 1116 | assert_eq!(resolve(link("1").as_ref(), link("2").as_ref(), link("3").as_ref()), Resolution::Conflict); |
| 1117 | // Made executable on one side, changed on the other. |
| 1118 | assert_eq!( |
| 1119 | resolve(blob("1").as_ref(), Some(&(EntryKind::Exec, "1".into())), blob("2").as_ref()), |
| 1120 | Resolution::Content { kind: EntryKind::Exec, base: Some("1".into()), ours: "1".into(), theirs: "2".into() } |
| 1121 | ); |
| 1122 | } |
| 1123 | |
| 1124 | #[test] |
| 1125 | fn trailers_are_added_once_after_the_message() { |
| 1126 | let sign = sign_off_trailer("Ana", "ana@example.com"); |
| 1127 | assert_eq!(with_trailers("Fix it\n", std::slice::from_ref(&sign)), "Fix it\n\nSigned-off-by: Ana <ana@example.com>"); |
| 1128 | // Already signed off (case aside): not again. |
| 1129 | let signed = "Fix it\n\nsigned-off-by: Ana <ANA@example.com>\n"; |
| 1130 | assert_eq!(with_trailers(signed, std::slice::from_ref(&sign)), signed.trim_end()); |
| 1131 | // Joins trailers already ending the message. |
| 1132 | // A template's trailers, from addresses alone: the author's own and |
| 1133 | // g1t's go, and a known one takes its real name. |
| 1134 | let known = ["Co-authored-by: Bo Lin <bo@example.com>".to_owned()]; |
| 1135 | let g1t = format!("Co-authored-by: g1t <{}>", g1t_contracts::system::EMAIL); |
| 1136 | let message = format!( |
| 1137 | "Fix\n\nCo-authored-by: ana <ANA@example.com>\nCo-authored-by: bo <bo@example.com>\n{g1t}\nCo-authored-by: cy <cy@example.com>" |
| 1138 | ); |
| 1139 | assert_eq!( |
| 1140 | tidy_co_authors(&message, "ana@example.com", &known), |
| 1141 | "Fix\n\nCo-authored-by: Bo Lin <bo@example.com>\nCo-authored-by: cy <cy@example.com>" |
| 1142 | ); |
| 1143 | let co = "Co-authored-by: Bo <bo@example.com>".to_owned(); |
| 1144 | assert_eq!( |
| 1145 | with_trailers("Fix it\n\nBody.\n\nCo-authored-by: Cy <cy@example.com>", &[co.clone(), sign.clone()]), |
| 1146 | "Fix it\n\nBody.\n\nCo-authored-by: Cy <cy@example.com>\nCo-authored-by: Bo <bo@example.com>\nSigned-off-by: Ana <ana@example.com>" |
| 1147 | ); |
| 1148 | // A body that only looks like it has a colon is not a trailer. |
| 1149 | assert_eq!(with_trailers("Fix it\n\nNote: see below\nmore text", std::slice::from_ref(&co)), "Fix it\n\nNote: see below\nmore text\n\nCo-authored-by: Bo <bo@example.com>"); |
| 1150 | // Twice in one list: once. |
| 1151 | assert_eq!(with_trailers("Fix", &[co.clone(), co.clone()]), "Fix\n\nCo-authored-by: Bo <bo@example.com>"); |
| 1152 | // Adding it to a message already holding it changes nothing. |
| 1153 | let once = with_trailers("Fix", std::slice::from_ref(&sign)); |
| 1154 | assert_eq!(with_trailers(&once, std::slice::from_ref(&sign)), once); |
| 1155 | } |
| 1156 | |
| 1157 | fn commit(hash: &str, parents: &[&str], name: &str, email: &str) -> Commit { |
| 1158 | Commit { |
| 1159 | hash: hash.to_owned(), |
| 1160 | tree_hash: String::new(), |
| 1161 | message: format!("{hash}\n"), |
| 1162 | author: Person { name: name.to_owned(), email: email.to_owned() }, |
| 1163 | parents: parents.iter().map(|&parent| parent.to_owned()).collect(), |
| 1164 | authored_at: String::new(), |
| 1165 | } |
| 1166 | } |
| 1167 | |
| 1168 | #[test] |
| 1169 | fn everyone_else_whose_commits_are_squashed_is_credited_once() { |
| 1170 | let commits = [ |
| 1171 | commit("a", &["m"], "Ana", "ana@example.com"), |
| 1172 | commit("b", &["a"], "Bo", "bo@example.com"), |
| 1173 | commit("c", &["b"], "Bo again", "BO@example.com"), |
| 1174 | commit("d", &["c"], "g1t", g1t_contracts::system::EMAIL), |
| 1175 | // A merge commit brings nothing of its own. |
| 1176 | commit("e", &["d", "x"], "Cy", "cy@example.com"), |
| 1177 | commit("f", &["e"], "Di", "di@example.com"), |
| 1178 | ]; |
| 1179 | assert_eq!( |
| 1180 | co_authors(&commits, "ANA@example.com"), |
| 1181 | ["Co-authored-by: Bo <bo@example.com>", "Co-authored-by: Di <di@example.com>"] |
| 1182 | ); |
| 1183 | } |
| 1184 | |
| 1185 | #[test] |
| 1186 | fn files_are_named_a_few_at_a_time() { |
| 1187 | let paths = |n: usize| (1..=n).map(|i| format!("f{i}")).collect::<Vec<_>>(); |
| 1188 | assert_eq!(name_files(&paths(1)), "f1"); |
| 1189 | assert_eq!(name_files(&paths(3)), "f1, f2 and f3"); |
| 1190 | assert_eq!(name_files(&paths(8)), "f1, f2, f3, f4, f5 and 3 more files"); |
| 1191 | } |
| 1192 | |
| 1193 | /// A store of whole objects made in the test: commits by name, trees |
| 1194 | /// and blobs by their real ids. |
| 1195 | #[derive(Default)] |
| 1196 | struct Store { |
| 1197 | commits: HashMap<String, Commit>, |
| 1198 | trees: HashMap<String, Vec<TreeEntry>>, |
| 1199 | blobs: HashMap<String, Vec<u8>>, |
| 1200 | } |
| 1201 | |
| 1202 | impl Store { |
| 1203 | fn tree(&mut self, files: &[(&str, &str)]) -> String { |
| 1204 | let mut here: Vec<TreeEntry> = Vec::new(); |
| 1205 | let mut dirs: std::collections::BTreeMap<&str, Vec<(&str, &str)>> = Default::default(); |
| 1206 | for (path, text) in files { |
| 1207 | match path.split_once('/') { |
| 1208 | Some((dir, rest)) => dirs.entry(dir).or_default().push((rest, text)), |
| 1209 | None => { |
| 1210 | let hash = object_id(ObjectKind::Blob, text.as_bytes()); |
| 1211 | self.blobs.insert(hash.clone(), text.as_bytes().to_vec()); |
| 1212 | here.push(TreeEntry { name: (*path).to_owned(), hash, kind: EntryKind::Blob }); |
| 1213 | } |
| 1214 | } |
| 1215 | } |
| 1216 | for (dir, files) in dirs { |
| 1217 | let hash = self.tree(&files); |
| 1218 | here.push(TreeEntry { name: dir.to_owned(), hash, kind: EntryKind::Tree }); |
| 1219 | } |
| 1220 | let id = object_id(ObjectKind::Tree, &crate::catch_up::encode_entries(&here)); |
| 1221 | self.trees.insert(id.clone(), here); |
| 1222 | id |
| 1223 | } |
| 1224 | |
| 1225 | fn commit(&mut self, hash: &str, parents: &[&str], author: &str, files: &[(&str, &str)]) -> Commit { |
| 1226 | let tree_hash = self.tree(files); |
| 1227 | let made = Commit { |
| 1228 | hash: hash.to_owned(), |
| 1229 | tree_hash, |
| 1230 | message: format!("Change {hash}\n"), |
| 1231 | author: Person { name: author.to_owned(), email: format!("{author}@example.com") }, |
| 1232 | parents: parents.iter().map(|&parent| parent.to_owned()).collect(), |
| 1233 | authored_at: "2026-10-01T00:00:00.000Z".to_owned(), |
| 1234 | }; |
| 1235 | self.commits.insert(hash.to_owned(), made.clone()); |
| 1236 | made |
| 1237 | } |
| 1238 | |
| 1239 | /// The files of a tree, read through `git`, by path. |
| 1240 | fn files(git: &Reader<'_, Store>, tree: &str) -> Vec<(String, String)> { |
| 1241 | let mut out = Vec::new(); |
| 1242 | let mut todo = vec![(String::new(), tree.to_owned())]; |
| 1243 | while let Some((prefix, tree)) = todo.pop() { |
| 1244 | for entry in block_on(git.read_tree(&tree)).unwrap().expect("tree") { |
| 1245 | let path = format!("{prefix}{}", entry.name); |
| 1246 | if entry.kind == EntryKind::Tree { |
| 1247 | todo.push((format!("{path}/"), entry.hash)); |
| 1248 | } else { |
| 1249 | let bytes = block_on(git.read_blob(&entry.hash)).unwrap().expect("blob"); |
| 1250 | out.push((path, String::from_utf8(bytes).unwrap())); |
| 1251 | } |
| 1252 | } |
| 1253 | } |
| 1254 | out.sort(); |
| 1255 | out |
| 1256 | } |
| 1257 | } |
| 1258 | |
| 1259 | impl GitRepo for Store { |
| 1260 | async fn access(&self, _scope: Scope) -> Result<GitAccess> { |
| 1261 | unimplemented!() |
| 1262 | } |
| 1263 | async fn branches(&self) -> Result<Vec<Branch>> { |
| 1264 | Ok(Vec::new()) |
| 1265 | } |
| 1266 | async fn log(&self, git_ref: &str, limit: u32) -> Result<Vec<Commit>> { |
| 1267 | let mut out = Vec::new(); |
| 1268 | let mut at = self.commits.get(git_ref); |
| 1269 | while let Some(commit) = at.filter(|_| (out.len() as u32) < limit) { |
| 1270 | out.push(commit.clone()); |
| 1271 | at = commit.parents.first().and_then(|parent| self.commits.get(parent)); |
| 1272 | } |
| 1273 | Ok(out) |
| 1274 | } |
| 1275 | async fn parents(&self, commit_hash: &str) -> Result<Option<Vec<String>>> { |
| 1276 | Ok(self.commits.get(commit_hash).map(|commit| commit.parents.clone())) |
| 1277 | } |
| 1278 | async fn read_tree(&self, tree_hash: &str) -> Result<Option<Vec<TreeEntry>>> { |
| 1279 | Ok(self.trees.get(tree_hash).cloned()) |
| 1280 | } |
| 1281 | async fn read_blob(&self, blob_hash: &str) -> Result<Option<Vec<u8>>> { |
| 1282 | Ok(self.blobs.get(blob_hash).cloned()) |
| 1283 | } |
| 1284 | async fn read_file(&self, _git_ref: &str, _path: &str) -> Result<Option<Vec<u8>>> { |
| 1285 | Ok(None) |
| 1286 | } |
| 1287 | async fn fork(&self, _target_key: &str, _default_branch_only: bool) -> Result<()> { |
| 1288 | Ok(()) |
| 1289 | } |
| 1290 | } |
| 1291 | |
| 1292 | fn block_on<F: std::future::Future>(future: F) -> F::Output { |
| 1293 | let mut future = std::pin::pin!(future); |
| 1294 | let mut cx = std::task::Context::from_waker(std::task::Waker::noop()); |
| 1295 | loop { |
| 1296 | if let std::task::Poll::Ready(output) = future.as_mut().poll(&mut cx) { |
| 1297 | return output; |
| 1298 | } |
| 1299 | } |
| 1300 | } |
| 1301 | |
| 1302 | fn files(list: &[(&str, &str)]) -> Vec<(String, String)> { |
| 1303 | list.iter().map(|(path, text)| ((*path).to_owned(), (*text).to_owned())).collect() |
| 1304 | } |
| 1305 | |
| 1306 | /// `main` at m then b; the branch at m then h1, h2. |
| 1307 | fn diverged() -> (Store, [Commit; 4]) { |
| 1308 | let mut store = Store::default(); |
| 1309 | let m = store.commit("m", &[], "ana", &[("README.md", "hi\n"), ("src/lib.rs", "a\nb\nc\nd\ne\n")]); |
| 1310 | let b = store.commit("b", &["m"], "ana", &[("README.md", "hello\n"), ("src/lib.rs", "a\nb\nc\nd\nE\n")]); |
| 1311 | let h1 = store.commit("h1", &["m"], "bo", &[("README.md", "hi\n"), ("src/lib.rs", "A\nb\nc\nd\ne\n")]); |
| 1312 | let h2 = store.commit( |
| 1313 | "h2", |
| 1314 | &["h1"], |
| 1315 | "cy", |
| 1316 | &[("README.md", "hi\n"), ("src/lib.rs", "A\nb\nc\nd\ne\n"), ("src/new.rs", "new\n")], |
| 1317 | ); |
| 1318 | (store, [m, b, h1, h2]) |
| 1319 | } |
| 1320 | |
| 1321 | #[test] |
| 1322 | fn a_change_applies_line_by_line_where_both_sides_touched_a_file() { |
| 1323 | let (store, [m, b, _, h2]) = diverged(); |
| 1324 | let git = Reader::new(&store, None); |
| 1325 | let mut objects = Vec::new(); |
| 1326 | let Applied::Tree(tree) = block_on(apply(&git, Some(&m.tree_hash), &h2.tree_hash, &b.tree_hash, &mut objects)).unwrap() else { |
| 1327 | panic!("not applied") |
| 1328 | }; |
| 1329 | assert_eq!( |
| 1330 | Store::files(&git, &tree), |
| 1331 | files(&[("README.md", "hello\n"), ("src/lib.rs", "A\nb\nc\nd\nE\n"), ("src/new.rs", "new\n")]) |
| 1332 | ); |
| 1333 | // The merged blob, the root and src: only what changed. |
| 1334 | assert_eq!(objects.iter().filter(|(kind, _)| *kind == ObjectKind::Blob).count(), 1); |
| 1335 | assert_eq!(objects.iter().filter(|(kind, _)| *kind == ObjectKind::Tree).count(), 2); |
| 1336 | } |
| 1337 | |
| 1338 | #[test] |
| 1339 | fn clashing_lines_name_the_file_and_make_nothing() { |
| 1340 | let mut store = Store::default(); |
| 1341 | let m = store.commit("m", &[], "ana", &[("a", "1\n"), ("b", "x\n")]); |
| 1342 | let h = store.commit("h", &["m"], "bo", &[("a", "2\n"), ("b", "y\n")]); |
| 1343 | let b = store.commit("b", &["m"], "ana", &[("a", "3\n"), ("b", "x\n")]); |
| 1344 | let git = Reader::new(&store, None); |
| 1345 | let mut objects = Vec::new(); |
| 1346 | let applied = block_on(apply(&git, Some(&m.tree_hash), &h.tree_hash, &b.tree_hash, &mut objects)).unwrap(); |
| 1347 | assert_eq!(applied, Applied::Conflict(vec!["a".into()])); |
| 1348 | assert!(objects.is_empty()); |
| 1349 | } |
| 1350 | |
| 1351 | #[test] |
| 1352 | fn a_branch_s_commits_come_oldest_first_and_stop_at_the_base() { |
| 1353 | let (mut store, [m, b, h1, h2]) = diverged(); |
| 1354 | // The branch merged main in, then went on. |
| 1355 | let j = store.commit("j", &["h2", "b"], "bo", &[("README.md", "hello\n")]); |
| 1356 | let k = store.commit("k", &["j"], "bo", &[("README.md", "hello\n")]); |
| 1357 | let history = block_on(store.log("k", 100)).unwrap(); |
| 1358 | let shared: HashSet<String> = [m.hash.clone(), b.hash.clone()].into_iter().collect(); |
| 1359 | let commits = block_on(branch_commits(&store, &history, &shared, 10)).unwrap().unwrap(); |
| 1360 | let order: Vec<&str> = commits.iter().map(|commit| commit.hash.as_str()).collect(); |
| 1361 | assert_eq!(order, [h1.hash.as_str(), h2.hash.as_str(), j.hash.as_str(), k.hash.as_str()]); |
| 1362 | // Too many: none. |
| 1363 | assert!(block_on(branch_commits(&store, &history, &shared, 3)).unwrap().is_none()); |
| 1364 | } |
| 1365 | |
| 1366 | #[test] |
| 1367 | fn a_rebase_makes_each_commit_again_on_the_base_keeping_its_author() { |
| 1368 | let (store, [_, b, h1, h2]) = diverged(); |
| 1369 | let git = Reader::new(&store, None); |
| 1370 | let committer = Signature { name: "Di", email: "di@example.com", seconds: 1_800_000_000 }; |
| 1371 | let sign = sign_off_trailer("Di", "di@example.com"); |
| 1372 | let mut objects = Vec::new(); |
| 1373 | let commits = [h1.clone(), h2.clone()]; |
| 1374 | let Replay::Done(replayed) = block_on(replay(&git, &commits, (&b.hash, &b.tree_hash), &committer, Some(&sign), &mut objects)).unwrap() else { |
| 1375 | panic!("not replayed") |
| 1376 | }; |
| 1377 | assert_eq!(replayed.made.iter().map(|(_, original)| original.as_str()).collect::<Vec<_>>(), ["h1", "h2"]); |
| 1378 | let made: Vec<String> = objects |
| 1379 | .iter() |
| 1380 | .filter(|(kind, _)| *kind == ObjectKind::Commit) |
| 1381 | .map(|(_, bytes)| String::from_utf8(bytes.clone()).unwrap()) |
| 1382 | .collect(); |
| 1383 | assert_eq!(made.len(), 2); |
| 1384 | // The first on the base, the second on the first. |
| 1385 | assert!(made[0].contains(&format!("parent {}\n", b.hash)), "{}", made[0]); |
| 1386 | assert!(made[1].contains(&format!("parent {}\n", replayed.made[0].0)), "{}", made[1]); |
| 1387 | assert_eq!(replayed.head, replayed.made[1].0); |
| 1388 | // Their authors and dates kept, committed by whoever rebased. |
| 1389 | assert!(made[0].contains("author bo <bo@example.com> 1790812800 +0000\ncommitter Di <di@example.com> 1800000000 +0000\n")); |
| 1390 | assert!(made[1].ends_with("\n\nChange h2\n\nSigned-off-by: Di <di@example.com>\n"), "{}", made[1]); |
| 1391 | // The last holds both sides. |
| 1392 | let tree = made[1].lines().next().unwrap().strip_prefix("tree ").unwrap().to_owned(); |
| 1393 | assert_eq!( |
| 1394 | Store::files(&git, &tree), |
| 1395 | files(&[("README.md", "hello\n"), ("src/lib.rs", "A\nb\nc\nd\nE\n"), ("src/new.rs", "new\n")]) |
| 1396 | ); |
| 1397 | } |
| 1398 | |
| 1399 | #[test] |
| 1400 | fn a_rebase_stops_at_the_commit_that_clashes() { |
| 1401 | let mut store = Store::default(); |
| 1402 | store.commit("m", &[], "ana", &[("a", "1\n"), ("b", "x\n")]); |
| 1403 | let h1 = store.commit("h1", &["m"], "bo", &[("a", "1\n"), ("b", "y\n")]); |
| 1404 | let h2 = store.commit("h2", &["h1"], "bo", &[("a", "2\n"), ("b", "y\n")]); |
| 1405 | let b = store.commit("b", &["m"], "ana", &[("a", "3\n"), ("b", "x\n")]); |
| 1406 | let git = Reader::new(&store, None); |
| 1407 | let committer = Signature { name: "Di", email: "di@example.com", seconds: 1 }; |
| 1408 | let mut objects = Vec::new(); |
| 1409 | let replayed = block_on(replay(&git, &[h1, h2], (&b.hash, &b.tree_hash), &committer, None, &mut objects)).unwrap(); |
| 1410 | let Replay::Conflict { commit, paths } = replayed else { panic!("no conflict") }; |
| 1411 | assert_eq!(commit.hash, "h2"); |
| 1412 | assert_eq!(paths, ["a"]); |
| 1413 | assert_eq!(named_commit(&commit), "h2 (Change h2)"); |
| 1414 | } |
| 1415 | |
| 1416 | #[test] |
| 1417 | fn a_commit_whose_changes_are_there_already_is_left_out() { |
| 1418 | let mut store = Store::default(); |
| 1419 | store.commit("m", &[], "ana", &[("a", "1\n")]); |
| 1420 | let h1 = store.commit("h1", &["m"], "bo", &[("a", "2\n")]); |
| 1421 | // The base made the same change. |
| 1422 | let b = store.commit("b", &["m"], "ana", &[("a", "2\n")]); |
| 1423 | let git = Reader::new(&store, None); |
| 1424 | let committer = Signature { name: "Di", email: "di@example.com", seconds: 1 }; |
| 1425 | let mut objects = Vec::new(); |
| 1426 | let Replay::Done(replayed) = block_on(replay(&git, &[h1], (&b.hash, &b.tree_hash), &committer, None, &mut objects)).unwrap() else { |
| 1427 | panic!("not replayed") |
| 1428 | }; |
| 1429 | assert!(replayed.made.is_empty()); |
| 1430 | assert_eq!(replayed.head, b.hash); |
| 1431 | } |
| 1432 | |
| 1433 | #[test] |
| 1434 | fn a_merge_commit_has_both_parents_and_a_squash_one() { |
| 1435 | let (store, [m, b, _, h2]) = diverged(); |
| 1436 | let git = Reader::new(&store, None); |
| 1437 | let committer = Signature { name: "Di", email: "di@example.com", seconds: 5 }; |
| 1438 | let author = Signature { name: "Bo", email: "bo@example.com", seconds: 5 }; |
| 1439 | let mut objects = Vec::new(); |
| 1440 | let Applied::Tree(tree) = block_on(apply(&git, Some(&m.tree_hash), &h2.tree_hash, &b.tree_hash, &mut objects)).unwrap() else { |
| 1441 | panic!("not applied") |
| 1442 | }; |
| 1443 | let merge = String::from_utf8(commit_object_by(&tree, &[&b.hash, &h2.hash], &committer, &committer, "Merge it")).unwrap(); |
| 1444 | assert!(merge.contains(&format!("parent {}\nparent {}\n", b.hash, h2.hash))); |
| 1445 | let squash = String::from_utf8(commit_object_by(&tree, &[&b.hash], &author, &committer, "Squash it")).unwrap(); |
| 1446 | assert_eq!(squash.matches("parent ").count(), 1); |
| 1447 | assert!(squash.contains("author Bo <bo@example.com> 5 +0000\ncommitter Di <di@example.com> 5 +0000\n")); |
| 1448 | } |
| 1449 | |
| 1450 | #[test] |
| 1451 | fn reads_fall_back_to_the_second_repository_and_the_objects_made() { |
| 1452 | let mut fork = Store::default(); |
| 1453 | let h = fork.commit("h", &[], "bo", &[("a", "fork\n")]); |
| 1454 | let repo = Store::default(); |
| 1455 | let git = Reader::new(&repo, Some(&fork)); |
| 1456 | assert!(block_on(git.read_tree(&h.tree_hash)).unwrap().is_some()); |
| 1457 | assert_eq!(block_on(git.log("h", 1)).unwrap().len(), 1); |
| 1458 | let made = vec![(ObjectKind::Blob, b"made\n".to_vec())]; |
| 1459 | git.keep(&made); |
| 1460 | assert_eq!(block_on(git.read_blob(&object_id(ObjectKind::Blob, b"made\n"))).unwrap().as_deref(), Some(&b"made\n"[..])); |
| 1461 | } |
| 1462 | |
| 1463 | /// A commit made in a test, as the store would list it. |
| 1464 | fn made_commit(objects: &[(ObjectKind, Vec<u8>)], id: &str) -> Commit { |
| 1465 | let bytes = objects |
| 1466 | .iter() |
| 1467 | .find(|(kind, bytes)| *kind == ObjectKind::Commit && object_id(ObjectKind::Commit, bytes) == id) |
| 1468 | .map(|(_, bytes)| String::from_utf8(bytes.clone()).unwrap()) |
| 1469 | .expect("made"); |
| 1470 | let field = |name: &str| bytes.lines().filter_map(|line| line.strip_prefix(name)).map(str::to_owned).collect::<Vec<_>>(); |
| 1471 | Commit { |
| 1472 | hash: id.to_owned(), |
| 1473 | tree_hash: field("tree ").remove(0), |
| 1474 | message: bytes.split_once("\n\n").map(|(_, message)| message.to_owned()).unwrap_or_default(), |
| 1475 | author: Person { name: "Di".to_owned(), email: "di@example.com".to_owned() }, |
| 1476 | parents: field("parent "), |
| 1477 | authored_at: "2026-10-01T00:00:00.000Z".to_owned(), |
| 1478 | } |
| 1479 | } |
| 1480 | |
| 1481 | #[test] |
| 1482 | fn each_queue_entry_lands_by_its_own_method_on_the_state_ahead_of_it() { |
| 1483 | let (mut store, [m, b, h1, h2]) = diverged(); |
| 1484 | // A second change, from the same start, to another file. |
| 1485 | let g = store.commit("g", &["m"], "eve", &[("README.md", "hi\n"), ("src/lib.rs", "a\nb\nc\nd\ne\n"), ("docs.md", "docs\n")]); |
| 1486 | let k = store.commit("k", &["m"], "eve", &[("README.md", "hi\n"), ("src/lib.rs", "a\nb\nc\nd\ne\n"), ("k.md", "k\n")]); |
| 1487 | let git = Reader::new(&store, None); |
| 1488 | let by = Signature { name: "Di", email: "di@example.com", seconds: 7 }; |
| 1489 | let author = Signature { name: "bo", email: "bo@example.com", seconds: 7 }; |
| 1490 | let mut objects = Vec::new(); |
| 1491 | let step = |method, onto: &[Commit], history: &[Commit], objects: &mut Vec<_>| { |
| 1492 | block_on(queue_step_on(&git, true, method, (onto, history), Some("Land it"), (&author, &by), None, "main", objects)).unwrap() |
| 1493 | }; |
| 1494 | let onto = vec![b.clone(), m.clone()]; |
| 1495 | |
| 1496 | // The first squashes onto the default branch: one commit, one parent. |
| 1497 | let StepMade::Made { commit: first, .. } = step(MergeMethod::Squash, &onto, &[h2.clone(), h1.clone(), m.clone()], &mut objects) else { |
| 1498 | panic!("not squashed") |
| 1499 | }; |
| 1500 | let first = made_commit(&objects, &first); |
| 1501 | assert_eq!(first.parents, ["b"]); |
| 1502 | |
| 1503 | // The second is a merge commit on the first's state, so landing it |
| 1504 | // is a fast-forward from there. |
| 1505 | let onto = vec![first.clone(), b.clone(), m.clone()]; |
| 1506 | let StepMade::Made { commit: second, .. } = step(MergeMethod::Merge, &onto, &[g.clone(), m.clone()], &mut objects) else { |
| 1507 | panic!("not merged") |
| 1508 | }; |
| 1509 | let second = made_commit(&objects, &second); |
| 1510 | assert_eq!(second.parents, [first.hash.clone(), "g".to_owned()]); |
| 1511 | assert_eq!( |
| 1512 | Store::files(&git, &second.tree_hash), |
| 1513 | files(&[("README.md", "hello\n"), ("docs.md", "docs\n"), ("src/lib.rs", "A\nb\nc\nd\nE\n"), ("src/new.rs", "new\n")]) |
| 1514 | ); |
| 1515 | |
| 1516 | // Queued again, its changes are there already: the state is kept. |
| 1517 | let onto = vec![second.clone(), first.clone(), b.clone(), m.clone()]; |
| 1518 | let StepMade::Made { commit, .. } = step(MergeMethod::Rebase, &onto, &[h2.clone(), h1.clone(), m.clone()], &mut objects) else { |
| 1519 | panic!("not kept") |
| 1520 | }; |
| 1521 | assert_eq!(commit, second.hash); |
| 1522 | |
| 1523 | // A fast-forward needs its head to hold the state ahead of it. |
| 1524 | let StepMade::Refused(why) = step(MergeMethod::FastForward, &onto, &[k, m.clone()], &mut objects) else { |
| 1525 | panic!("fast-forwarded") |
| 1526 | }; |
| 1527 | assert!(why.contains("fast-forward"), "{why}"); |
| 1528 | } |
| 1529 | |
| 1530 | #[test] |
| 1531 | fn a_fast_forward_entry_is_its_own_head_and_a_clash_names_the_files() { |
| 1532 | let (mut store, [m, b, h1, h2]) = diverged(); |
| 1533 | let ahead = store.commit("f", &["b"], "eve", &[("README.md", "hello\n"), ("src/lib.rs", "a\nb\nc\nd\nE\n"), ("f.md", "f\n")]); |
| 1534 | let clash = store.commit("x", &["m"], "eve", &[("README.md", "hey\n"), ("src/lib.rs", "a\nb\nc\nd\ne\n")]); |
| 1535 | let git = Reader::new(&store, None); |
| 1536 | let by = Signature { name: "Di", email: "di@example.com", seconds: 7 }; |
| 1537 | let mut objects = Vec::new(); |
| 1538 | let onto = vec![b.clone(), m.clone()]; |
| 1539 | let made = block_on(queue_step_on(&git, true, MergeMethod::FastForward, (&onto, &[ahead.clone(), b.clone(), m.clone()]), None, (&by, &by), None, "main", &mut objects)).unwrap(); |
| 1540 | assert!(matches!(made, StepMade::Made { ref commit, .. } if commit == "f"), "{made:?}"); |
| 1541 | assert!(objects.is_empty()); |
| 1542 | let clashed = block_on(queue_step_on(&git, true, MergeMethod::Rebase, (&onto, &[clash, m.clone()]), None, (&by, &by), None, "main", &mut objects)).unwrap(); |
| 1543 | assert!(matches!(clashed, StepMade::Conflict { ref paths, .. } if paths == &["README.md".to_owned()]), "{clashed:?}"); |
| 1544 | let _ = (h1, h2); |
| 1545 | } |
| 1546 | } |