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