From 9cf640dee5fc3f7cb1fb549be3584bc8b8660dc9 Mon Sep 17 00:00:00 2001
From: Boming Zhang <bomingzh@sjtu.edu.cn>
Date: Wed, 4 Jun 2025 06:05:11 -0400
Subject: [PATCH] feat(healthcheck): check commit author email

---
 cmd/repo-health-checker/main.go |  9 ++++++++-
 pkg/healthcheck/all.go          | 10 +++++++++-
 pkg/healthcheck/commit.go       | 32 ++++++++++++++++++++++++++++++++
 3 files changed, 49 insertions(+), 2 deletions(-)

diff --git a/cmd/repo-health-checker/main.go b/cmd/repo-health-checker/main.go
index c4a16cd..e170c13 100644
--- a/cmd/repo-health-checker/main.go
+++ b/cmd/repo-health-checker/main.go
@@ -45,6 +45,7 @@ var (
 	checkFileNameList string
 	checkFileSumList  string
 	metaFile          []string
+	allowedDomainList string
 	showVersion       *bool
 	Version           string
 )
@@ -55,6 +56,7 @@ func init() {
 	flag.Float64Var(&repoSize, "repoSize", 2, "maximum size of the repo in MiB")
 	flag.StringVar(&checkFileNameList, "checkFileNameList", "", "comma-separated list of files to check")
 	flag.StringVar(&checkFileSumList, "checkFileSumList", "", "comma-separated list of expected checksums")
+	flag.StringVar(&allowedDomainList, "allowedDomainList", "sjtu.edu.cn", "comma-separated list of allowed domains for commit author email")
 	parseMultiValueFlag(&metaFile, "meta", "meta files to check")
 }
 
@@ -73,7 +75,12 @@ func main() {
 		"meta", metaFile,
 	)
 	res := healthcheck.All(
-		rootDir, checkFileNameList, checkFileSumList, metaFile, repoSize,
+		rootDir,
+		checkFileNameList,
+		checkFileSumList,
+		allowedDomainList,
+		metaFile,
+		repoSize,
 	)
 	jsonRes, err := json.Marshal(res)
 	if err != nil {
diff --git a/pkg/healthcheck/all.go b/pkg/healthcheck/all.go
index 68ce56f..14d3fb7 100644
--- a/pkg/healthcheck/all.go
+++ b/pkg/healthcheck/all.go
@@ -3,6 +3,7 @@ package healthcheck
 
 import (
 	"fmt"
+	"strings"
 )
 
 type Result struct {
@@ -11,7 +12,7 @@ type Result struct {
 }
 
 func All(
-	rootDir, checkFileNameList, checkFileSumList string,
+	rootDir, checkFileNameList, checkFileSumList, allowedDomainList string,
 	metaFile []string, repoSize float64,
 ) (res Result) {
 	var err error
@@ -57,5 +58,12 @@ func All(
 	} else {
 		res.Msg += "### Repo File Check Passed\n"
 	}
+	err = AuthorEmailCheck(rootDir, strings.Split(allowedDomainList, ","))
+	if err != nil {
+		res.Msg += fmt.Sprintf("### Author Email Check Failed:\n%s\n", err.Error())
+		res.Failed = true
+	} else {
+		res.Msg += "### Author Email Check Passed\n"
+	}
 	return res
 }
diff --git a/pkg/healthcheck/commit.go b/pkg/healthcheck/commit.go
index d27b6e4..0a96890 100644
--- a/pkg/healthcheck/commit.go
+++ b/pkg/healthcheck/commit.go
@@ -74,3 +74,35 @@ func NonASCIIMsg(root string) error {
 	}
 	return nil
 }
+
+func AuthorEmailCheck(root string, allowedDomains []string) error {
+	repo, err := git.PlainOpen(root)
+	if err != nil {
+		slog.Error("opening git repo", "err", err)
+		return fmt.Errorf("error opening git repo: %v", err)
+	}
+
+	ref, err := repo.Head()
+	if err != nil {
+		slog.Error("getting reference", "err", err)
+		return fmt.Errorf("error getting reference: %v", err)
+	}
+
+	commit, err := repo.CommitObject(ref.Hash())
+	if err != nil {
+		slog.Error("getting latest commit", "err", err)
+		return fmt.Errorf("error getting latest commit: %v", err)
+	}
+
+	email := commit.Author.Email
+	for _, domain := range allowedDomains {
+		if strings.HasSuffix(email, "@"+domain) {
+			return nil
+		}
+	}
+	return fmt.Errorf(
+		"Author email %s is not in the allowed domains: %v",
+		email,
+		allowedDomains,
+	)
+}