use std::net::SocketAddr; use std::path::PathBuf; use std::pin::Pin; use gix::bstr::ByteSlice as _; use gix::revision::plumbing::Spec as RevSpec; use gix::revision::walk::Sorting; use tempfile::TempDir; use tokio::sync::mpsc; use tokio_stream::wrappers::ReceiverStream; use tokio_stream::Stream; use tonic::transport::Server; use tonic::{Request, Response, Status}; use tracing::{debug, info}; use crate::diff::{self, prepare_interdiff}; use crate::protocol::v1::diff_request::ComparisonType; use crate::protocol::v1::git_mirror_service_server::{GitMirrorService, GitMirrorServiceServer}; use crate::protocol::v1::{ BlobChunk, CommitLogRequest, CommitLogResponse, DiffRequest, FileDiff, GetBlobRequest, GitCommit, GitSignature, InterdiffRequest, MergeCheckRequest, MergeCheckResponse, ResolveRevRequest, ResolveRevResponse, }; type FileDiffStream = Pin> + Send>>; type BlobChunkStream = Pin> + Send>>; type CommitLogResponseStream = Pin> + Send>>; const BLOB_CHUNK_SIZE: usize = 64 * 1024; const COMMIT_LOG_BATCH_SIZE: usize = 128; struct GitMirror { repo_base: PathBuf, } /// A [`gix::Repository`] backed by a temporary directory that is removed on drop. Derefs to /// `gix::Repository`, so it is used just like one; keeping it alive keeps the scratch dir alive. struct TempRepository { repo: gix::Repository, // Declared AFTER `repo` so `repo` drops first: any file handles into the scratch dir close // before the dir itself is removed (Rust drops struct fields in declaration order). _dir: TempDir, } impl TempRepository { fn new(repo: gix::Repository, dir: TempDir) -> Self { Self { repo, _dir: dir } } } impl std::ops::Deref for TempRepository { type Target = gix::Repository; fn deref(&self) -> &gix::Repository { &self.repo } } impl GitMirror { fn open_repo(&self, did: &str) -> Result { if !did.starts_with("did:") { return Err(Status::invalid_argument("repo must be a DID")); } gix::open(self.repo_base.join(did)) .map_err(|e| Status::not_found(format!("repo not found: {e}"))) } /// Build a throwaway bare repo whose `objects/info/alternates` points read-only at each of the /// given git repositories, so a single `gix::Repository` can see objects from all of them /// without ever mutating them. fn open_scratch(&self, dids: &[&str]) -> Result { let mut seen: Vec<&str> = Vec::new(); let mut object_dirs = Vec::new(); for &did in dids { if seen.contains(&did) { continue } if !did.starts_with("did:") { return Err(Status::invalid_argument("repo must be a DID")); } seen.push(did); let object_dir = std::fs::canonicalize(self.repo_base.join(did).join("objects")) .map_err(|_| Status::not_found(format!("repo not found: {did}")))?; object_dirs.push(object_dir); } if object_dirs.is_empty() { return Err(Status::internal("open_scratch requires at least one repo")); } let scratch = tempfile::tempdir().map_err(|e| Status::internal(e.to_string()))?; gix::init_bare(scratch.path()).map_err(|e| Status::internal(e.to_string()))?; let info_dir = scratch.path().join("objects").join("info"); std::fs::create_dir_all(&info_dir).map_err(|e| Status::internal(e.to_string()))?; let alternates = object_dirs .iter() .map(|p| p.display().to_string()) .collect::>() .join("\n"); std::fs::write(info_dir.join("alternates"), format!("{alternates}\n")) .map_err(|e| Status::internal(e.to_string()))?; let repo = gix::open(scratch.path()) .map_err(|e| Status::internal(e.to_string()))? .with_object_memory(); Ok(TempRepository::new(repo, scratch)) } } #[tonic::async_trait] impl GitMirrorService for GitMirror { type DiffStream = FileDiffStream; async fn diff( &self, request: Request, ) -> Result, Status> { let req = request.into_inner(); let head = req .head .ok_or_else(|| Status::invalid_argument("head is required"))?; let base = req.base; debug!( base_repo = ?base.as_ref().map(|b| b.repo.as_str()), base_commit = ?base.as_ref().map(|b| b.commit.as_bstr()), head_repo = %head.repo, head_commit = %head.commit.as_bstr(), "diff request", ); let scratch = match &base { Some(base) => self.open_scratch(&[head.repo.as_str(), base.repo.as_str()])?, None => self.open_scratch(&[head.repo.as_str()])?, }; let head_commit_id = find_commit_by_sha(&scratch, &head.commit)?; let base_commit_id = base .as_ref() .map(|b| find_commit_by_sha(&scratch, &b.commit)) .transpose()?; let comparison_type = ComparisonType::try_from(req.comparison_type).unwrap_or(ComparisonType::Unspecified); let (tx, rx) = mpsc::channel::>(16); tokio::task::spawn_blocking(move || { let repo = &*scratch; let run = || -> anyhow::Result<()> { let head_commit = repo.find_commit(head_commit_id)?; let old = match base_commit_id { Some(base_commit_id) => match comparison_type { ComparisonType::OnlyInHead => { repo.find_tree(repo.find_commit(base_commit_id)?.tree_id()?)? } ComparisonType::Unspecified | ComparisonType::Intersection => { let merge_base = repo.merge_base(base_commit_id, head_commit_id)?.detach(); repo.find_tree(repo.find_commit(merge_base)?.tree_id()?)? } }, None => match head_commit.parent_ids().next() { Some(parent_id) => { repo.find_tree(repo.find_commit(parent_id.detach())?.tree_id()?)? } None => repo.empty_tree(), }, }; let new = repo.find_tree(head_commit.tree_id()?)?; for item in diff::diff(repo, &old, &new, false)? { if tx.blocking_send(Ok(item?.into())).is_err() { break; } } Ok(()) }; if let Err(e) = run() { let _ = tx.blocking_send(Err(Status::internal(e.to_string()))); } }); Ok(Response::new(Box::pin(ReceiverStream::new(rx)))) } type InterdiffStream = FileDiffStream; async fn interdiff( &self, request: Request, ) -> Result, Status> { let req = request.into_inner(); let repo = self.open_repo(&req.repo)?; let from_base_id = resolve_commit_oid(&repo, &req.from_base)?; let from_head_id = resolve_commit_oid(&repo, &req.from_head)?; let to_base_id = resolve_commit_oid(&repo, &req.to_base)?; let to_head_id = resolve_commit_oid(&repo, &req.to_head)?; let safe = repo.into_sync(); let (tx, rx) = mpsc::channel::>(16); tokio::task::spawn_blocking(move || { let repo = safe.to_thread_local(); let run = || -> anyhow::Result<()> { let to_head = repo.find_commit(to_head_id)?; let to_head_tree = to_head.tree()?; let rebased_tree = prepare_interdiff(&repo, (from_base_id, from_head_id), to_base_id)?; for item in diff::diff(&repo, &rebased_tree, &to_head_tree, true)? { if tx.blocking_send(Ok(item?.into())).is_err() { break; } } Ok(()) }; if let Err(e) = run() { let _ = tx.blocking_send(Err(Status::internal(e.to_string()))); } }); Ok(Response::new(Box::pin(ReceiverStream::new(rx)))) } type GetBlobStream = BlobChunkStream; async fn get_blob( &self, request: Request, ) -> Result, Status> { let req = request.into_inner(); let repo = self.open_repo(&req.repo)?; let oid = gix::ObjectId::from_hex(req.oid.as_bytes()) .map_err(|e| Status::invalid_argument(format!("bad oid '{}': {e}", req.oid)))?; let safe = repo.into_sync(); let (tx, rx) = mpsc::channel::>(16); tokio::task::spawn_blocking(move || { let repo = safe.to_thread_local(); let run = || -> anyhow::Result<()> { let blob = repo.find_object(oid)?.try_into_blob()?; for chunk in blob.data.chunks(BLOB_CHUNK_SIZE) { let msg = BlobChunk { data: chunk.to_vec(), }; if tx.blocking_send(Ok(msg)).is_err() { break; } } Ok(()) }; if let Err(e) = run() { let _ = tx.blocking_send(Err(Status::internal(e.to_string()))); } }); Ok(Response::new(Box::pin(ReceiverStream::new(rx)))) } type CommitLogStream = CommitLogResponseStream; async fn commit_log( &self, request: Request, ) -> Result, Status> { let req = request.into_inner(); debug!( repo = %req.repo, all_refs = req.all_refs, ranges = ?req.ranges, after = ?req.after, before = ?req.before, max_commits = req.max_commits, skip = req.skip, "commit_log request", ); if req.all_refs != req.ranges.is_empty() { return Err(Status::invalid_argument( "specify exactly one of `ranges` or `all_refs`", )); } let repo = self.open_repo(&req.repo)?; let safe = repo.into_sync(); let (tx, rx) = mpsc::channel::>(16); tokio::task::spawn_blocking(move || { let repo = safe.to_thread_local(); let run = || -> anyhow::Result<()> { let after = req.after.map(|t| t.seconds); let before = req.before.map(|t| t.seconds); let max_commits = req.max_commits as usize; let mut skip = req.skip as usize; let mut sent = 0usize; let mut batch = Vec::with_capacity(COMMIT_LOG_BATCH_SIZE); for info in commit_log_walk(&repo, &req)? { let info = info?; let commit_time = info.commit_time(); if before.is_some_and(|b| commit_time > b) { continue; } if after.is_some_and(|a| commit_time < a) { break; } if skip > 0 { skip -= 1; continue; } if max_commits != 0 && sent >= max_commits { break; } batch.push(GitCommit::try_from(&info.object()?)?); sent += 1; if batch.len() == COMMIT_LOG_BATCH_SIZE { let msg = CommitLogResponse { commits: std::mem::take(&mut batch), }; if tx.blocking_send(Ok(msg)).is_err() { return Ok(()); // client hung up } } } if !batch.is_empty() { let _ = tx.blocking_send(Ok(CommitLogResponse { commits: batch })); } Ok(()) }; if let Err(e) = run() { let _ = tx.blocking_send(Err(Status::internal(e.to_string()))); } }); Ok(Response::new(Box::pin(ReceiverStream::new(rx)))) } async fn merge_check( &self, request: Request, ) -> Result, Status> { let req = request.into_inner(); let target = req .target .ok_or_else(|| Status::invalid_argument("target is required"))?; let source = req .source .ok_or_else(|| Status::invalid_argument("source is required"))?; debug!( target_repo = %target.repo, target_commit = %target.commit.as_bstr(), source_repo = %source.repo, source_commit = %source.commit.as_bstr(), "merge_check request", ); let scratch = self.open_scratch(&[target.repo.as_str(), source.repo.as_str()])?; let response = tokio::task::spawn_blocking(move || -> Result { let target_oid = find_commit_by_sha(&scratch, &target.commit)?; let source_oid = find_commit_by_sha(&scratch, &source.commit)?; crate::merge::merge_check(&scratch, target_oid, source_oid) .map_err(|e| Status::internal(e.to_string())) }) .await .map_err(|e| Status::internal(format!("merge task panicked: {e}")))??; Ok(Response::new(response)) } async fn resolve_rev( &self, request: Request, ) -> Result, Status> { let req = request.into_inner(); debug!(repo = %req.repo, rev = %req.rev.as_bstr(), "resolve_rev request"); let repo = self.open_repo(&req.repo)?; let commit = tokio::task::spawn_blocking(move || -> Result { Ok(resolve_simple_rev(&repo, &req.rev)?.to_string()) }) .await .map_err(|e| Status::internal(format!("resolve task panicked: {e}")))??; Ok(Response::new(ResolveRevResponse { commit })) } } fn resolve_commit_oid(repo: &gix::Repository, spec: &[u8]) -> Result { let commit = repo .rev_parse_single(spec.as_bstr()) .map_err(|e| { Status::invalid_argument(format!("bad rev '{}': {e}", String::from_utf8_lossy(spec))) })? .object() .map_err(|e| Status::internal(e.to_string()))? .peel_to_commit() .map_err(|e| Status::internal(e.to_string()))?; Ok(commit.id) } fn resolve_simple_rev(repo: &gix::Repository, rev: &[u8]) -> Result { let unsupported = rev == b"HEAD" || rev == b"@" || rev.windows(2).any(|w| w == b"@{") || rev.iter().any(|&b| matches!(b, b'~' | b'^' | b':')); if unsupported { return Err(Status::invalid_argument(format!( "unsupported rev '{}': only a branch, tag, or commit oid is allowed", rev.as_bstr() ))); } resolve_commit_oid(repo, rev) } fn find_commit_by_sha(repo: &gix::Repository, sha: &[u8]) -> Result { let oid = gix::ObjectId::from_hex(sha).map_err(|e| { Status::invalid_argument(format!("bad commit sha '{}': {e}", sha.as_bstr())) })?; repo.find_commit(oid) .map_err(|_| Status::not_found(format!("commit not found: {oid}")))?; Ok(oid) } fn commit_log_walk<'repo>( repo: &'repo gix::Repository, req: &CommitLogRequest, ) -> anyhow::Result> { let mut tips = Vec::new(); let mut hidden = Vec::new(); if req.all_refs { for r in repo.references()?.all()? { let mut r = r.map_err(|e| anyhow::anyhow!(e))?; if let Ok(commit) = r.peel_to_commit() { tips.push(commit.id); } } } else { for range in &req.ranges { let revspec = repo.rev_parse(range.as_bstr())?; let spec = revspec.detach(); match spec { RevSpec::Include(id) => tips.push(id), RevSpec::Range { from, to } => { tips.push(to); hidden.push(from); } _ => { anyhow::bail!("The spec isn't currently supported: {spec:?}") } } } } Ok(repo .rev_walk(tips) .sorting(Sorting::ByCommitTime(Default::default())) .with_hidden(hidden) .all()?) } impl TryFrom<&gix::Commit<'_>> for GitCommit { type Error = anyhow::Error; fn try_from(commit: &gix::Commit<'_>) -> anyhow::Result { let decoded = commit.decode()?; Ok(GitCommit { oid: commit.id.to_string(), author: Some(decoded.author()?.try_into()?), committer: Some(decoded.committer()?.try_into()?), message: decoded.message.to_vec(), parents: decoded .parents .into_iter() .map(|id| id.to_string()) .collect(), extra_headers: decoded .extra_headers .iter() .map(|(k, v)| (k.to_string(), v.to_string())) .collect(), }) } } impl TryFrom> for GitSignature { type Error = anyhow::Error; fn try_from(sig: gix::actor::SignatureRef<'_>) -> anyhow::Result { let time = sig.time()?; Ok(GitSignature { name: sig.name.to_vec(), email: sig.email.to_vec(), date: Some(prost_types::Timestamp { seconds: time.seconds, nanos: 0, }), }) } } pub async fn serve(addr: SocketAddr, repo_base: PathBuf) -> anyhow::Result<()> { let service = GitMirror { repo_base }; info!(addr = %addr, "gitmirror gRPC server listening"); Server::builder() .add_service(GitMirrorServiceServer::new(service)) .serve(addr) .await?; Ok(()) } #[cfg(test)] mod tests { use super::*; #[test] fn temp_repository_removes_scratch_dir_on_drop() { let dir = tempfile::tempdir().unwrap(); gix::init_bare(dir.path()).unwrap(); let repo = gix::open(dir.path()).unwrap().with_object_memory(); let temp = TempRepository::new(repo, dir); let path = temp.git_dir().to_path_buf(); assert!(path.exists()); // Derefs to gix::Repository. assert!(temp.object_hash() == gix::hash::Kind::Sha1); drop(temp); assert!(!path.exists(), "scratch dir should be gone after drop"); } }