fix: relocate PR review comments using git blame --reverse, improving comment placement (#12015)

When a review comment is placed on a PR in Forgejo, Forgejo performs a `git blame` to identify which commit originated the line, and records that commit and line number in the comment's database record. Later when the review is viewed, Forgejo currently makes no effort to place that comment in the correct *current* location, which may vary -- for example, if a PR had two commits and the comment was made on a line in the first commit, but the second commit changes line numbers in that file, the comment will appear in the incorrect location. This PR adds the usage of `git blame --reverse` to calculate the correct location to display the comment in the current view (whether reviewing the PR commit-by-commit, or "Files changed"). It certainly does not fix all problems with comment placement (see comments). Another major addition in this PR is a test harness for making relatively complex PRs and reviewing the diffs on the per-commit view and PR-diff views. ## Checklist The [contributor guide](https://forgejo.org/docs/next/contributor/) contains information that will be helpful to first time contributors. All work and communication must conform to Forgejo's [AI Agreement](https://codeberg.org/forgejo/governance/src/branch/main/AIAgreement.md). There also are a few [conditions for merging Pull Requests in Forgejo repositories](https://codeberg.org/forgejo/governance/src/branch/main/PullRequestsAgreement.md). You are also welcome to join the [Forgejo development chatroom](https://matrix.to/#/#forgejo-development:matrix.org). ### Tests for Go changes - I added test coverage for Go changes... - [x] in their respective `*_test.go` for unit tests. - [x] in the `tests/integration` directory if it involves interactions with a live Forgejo server. - I ran... - [ ] `make pr-go` before pushing ### Documentation - [ ] I created a pull request [to the documentation](https://codeberg.org/forgejo/docs) to explain to Forgejo users how to use this change. - [x] I did not document these changes and I do not expect someone else to do it. ### Release notes - [x] This change will be noticed by a Forgejo user or admin (feature, bug fix, performance, etc.). I suggest to include a release note for this change. - [ ] This change is not visible to a Forgejo user or admin (refactor, dependency upgrade, etc.). I think there is no need to add a release note for this change.  ## Release notes  - Bug fixes - [PR](https://codeberg.org/forgejo/forgejo/pulls/12015): relocate PR review comments using `git blame --reverse`, improving comment placement  Reviewed-on: https://codeberg.org/forgejo/forgejo/pulls/12015 Reviewed-by: Andreas Ahlenstorf <aahlenst@noreply.codeberg.org> Co-authored-by: Mathieu Fenniak <mathieu@fenniak.net> Co-committed-by: Mathieu Fenniak <mathieu@fenniak.net>
2026-04-11 21:45:39 +02:00
parent 92b95414e8
commit 9fe0cbee02
9 changed files with 936 additions and 19 deletions
@@ -19,7 +19,9 @@ import (
 	project_model "forgejo.org/models/project"
 	repo_model "forgejo.org/models/repo"
 	user_model "forgejo.org/models/user"
+	"forgejo.org/modules/cache"
 	"forgejo.org/modules/container"
+	"forgejo.org/modules/git"
 	"forgejo.org/modules/gitrepo"
 	"forgejo.org/modules/json"
 	"forgejo.org/modules/log"
@@ -321,6 +323,8 @@ type Comment struct {
 	CommitsNum  int64                               `xorm:"-"`
 	IsForcePush bool                                `xorm:"-"`

+	reverseLineBlame *git.ReverseLineBlame `xorm:"-"`
+
 	// If you add new fields that might be used to store abusive content (mainly string fields),
 	// please also add them in the CommentData struct and the corresponding constructor.
 }
@@ -744,6 +748,55 @@ func (c *Comment) UnsignedLine() uint64 {
 	return uint64(c.Line)
 }

+func (c *Comment) ResolveCurrentLine(ctx context.Context, repo *repo_model.Repository, currentHead string) (*git.ReverseLineBlame, error) {
+	if c.reverseLineBlame != nil {
+		return c.reverseLineBlame, nil
+	}
+
+	// When a PR is viewed, the requirement to perform `git blame --reverse...` on every comment is a bit of a
+	// performance risk. To minimize this risk, cache the results relative to the requested head, so it only needs to be
+	// recalculated when head changes (or on cache eviction).
+	//
+	// Some performance testing was done which showed that a hot cache is much faster than the blame reverse
+	// operation -- 500-1000x runtime difference:
+	//
+	// - cache miss (Forgejo repo)      took 7,690,574 ns
+	// - cache miss (~1000 commit repo) took 1,671,223 ns
+	// - cache hit (in-memory adapter)  took     3,710 ns
+	// - cache hit (redis adapter)      took    77,311 ns
+	resolveJSON, err := cache.GetString(fmt.Sprintf("comment.Resolve;ID=%d;HEAD=%s", c.ID, currentHead), func() (string, error) {
+		gitRepo, closer, err := gitrepo.RepositoryFromContextOrOpen(ctx, repo)
+		if err != nil {
+			return "", fmt.Errorf("failed to open repo: %w", err)
+		}
+		defer closer.Close()
+
+		reverseBlame, err := gitRepo.ReverseLineBlame(c.CommitSHA, c.TreePath, c.UnsignedLine(), currentHead)
+		if err != nil {
+			return "", fmt.Errorf("failed to perform `git blame --reverse` to resolve current line for comment (id=%d): %w", c.ID, err)
+		}
+
+		data, err := json.Marshal(reverseBlame)
+		if err != nil {
+			return "", err
+		}
+
+		return string(data), nil
+	})
+	if err != nil {
+		return nil, err
+	}
+
+	var reverseBlame *git.ReverseLineBlame
+	err = json.Unmarshal([]byte(resolveJSON), &reverseBlame)
+	if err != nil {
+		return nil, err
+	}
+
+	c.reverseLineBlame = reverseBlame
+	return c.reverseLineBlame, nil
+}
+
 // CodeCommentLink returns the url to a comment in code
 func (c *Comment) CodeCommentLink(ctx context.Context) string {
 	err := c.LoadIssue(ctx)