PR comment causes 500 server error

[otbutz]

• Gitea version (or commit ref): 1.8.1
• Git version: 2.21
• Operating system: Ubuntu 18.04
• Database (use [x]):
  • PostgreSQL
  • MySQL
  • MSSQL
  • SQLite
• Can you reproduce the bug at https://try.gitea.io:
  • Yes (provide example URL)
  • No
  • Not relevant
• Log gist:

2019/06/03 07:48:37 [.../repo/pull_review.go:80 CreateCodeComment()] [E] CreateCodeComment: pq: invalid byte sequence for encoding "UTF8": 0xb3

Description

No matter which file i'm trying to comment in this particular PR, the same error happens. Simply writing test in the comment generates the same issue.

[zeripath]

I suspect the issue is that the underlying code line and Patch component here is not UTF-8:

gitea/models/issue_comment.go

Lines 928 to 939 in 45f588e

	return CreateComment(&CreateCommentOptions{
	Type: CommentTypeCode,
	Doer: doer,
	Repo: repo,
	Issue: issue,
	Content: content,
	LineNum: line,
	TreePath: treePath,
	CommitSHA: commitID,
	ReviewID: reviewID,
	Patch: patch,
	})

[zeripath]

OK looking at:

gitea/models/issue_comment.go

Lines 922 to 926 in 45f588e

	patchBuf := new(bytes.Buffer)
	if err := GetRawDiffForFile(gitRepo.Path, pr.MergeBase, headCommitID, RawDiffNormal, treePath, patchBuf); err != nil {
	return nil, fmt.Errorf("GetRawDiffForLine[%s, %s, %s, %s]: %v", err, gitRepo.Path, pr.MergeBase, headCommitID, treePath)
	}
	patch = CutDiffAroundLine(strings.NewReader(patchBuf.String()), int64((&Comment{Line: line}).UnsignedLine()), line < 0, setting.UI.CodeCommentLines)

The issue is that patchBuf.String() doesn't actually guarantee that the String is a valid UTF-8 string. (Thanks Go(!)) Git diff gives you the raw bytes - it doesn't care about the encoding.

[zeripath]

@otbutz is there any chance you could upload a minimal test file to try.gitea.io it would be nice to know which encoding breaks this.

[otbutz]

Not really. Corporate project.

[lafriks]

Duplicate of #6236