From dee296143f4a23dd41d2d5363feab474f6288d9c Mon Sep 17 00:00:00 2001
From: Boming Zhang <bomingzh@sjtu.edu.cn>
Date: Fri, 4 Oct 2024 02:06:32 -0400
Subject: [PATCH] feat: non-ascii not in the first line of commit msg

---
 pkg/healthcheck/commit.go      | 73 +++++++++++++++---------------
 pkg/healthcheck/commit_test.go | 83 ++++++++++++++++++++++++++++++++++
 2 files changed, 119 insertions(+), 37 deletions(-)
 create mode 100644 pkg/healthcheck/commit_test.go

diff --git a/pkg/healthcheck/commit.go b/pkg/healthcheck/commit.go
index 375046e..4bfecac 100644
--- a/pkg/healthcheck/commit.go
+++ b/pkg/healthcheck/commit.go
@@ -9,6 +9,41 @@ import (
 	"github.com/go-git/go-git/v5"
 )
 
+func checkMsg(msg string) bool {
+	// List of prefixes to ignore in the commit message
+	ignoredPrefixes := []string{
+		"Co-authored-by:",
+		"Reviewed-by:",
+		"Co-committed-by:",
+		"Reviewed-on:",
+	}
+
+	// Split message by lines and ignore specific lines with prefixes
+	lines := strings.Split(msg, "\n")
+	for i, line := range lines {
+		trimmedLine := strings.TrimSpace(line)
+		ignore := false
+		if i != 0 {
+			for _, prefix := range ignoredPrefixes {
+				if strings.HasPrefix(trimmedLine, prefix) {
+					ignore = true
+					break
+				}
+			}
+		}
+		if ignore {
+			continue
+		}
+		// Check for non-ASCII characters in the rest of the lines
+		for _, c := range line {
+			if c > unicode.MaxASCII {
+				return false
+			}
+		}
+	}
+	return true
+}
+
 // nonAsciiMsg checks for non-ASCII characters in the commit message.
 // If the message starts with "Merge pull request", it skips the non-ASCII characters check.
 // Otherwise, it iterates over each character in the message and checks if it is a non-ASCII character.
@@ -35,43 +70,7 @@ func NonAsciiMsg(root string) error {
 	}
 
 	msg := commit.Message
-	if msg == "" {
-		return nil
-	}
-
-	var isCommitLegal bool = true
-	// List of prefixes to ignore in the commit message
-	ignoredPrefixes := []string{
-		"Co-authored-by:",
-		"Reviewed-by:",
-		"Co-committed-by:",
-		"Reviewed-on:",
-	}
-
-	// Split message by lines and ignore specific lines with prefixes
-	lines := strings.Split(msg, "\n")
-	for _, line := range lines {
-		trimmedLine := strings.TrimSpace(line)
-		ignore := false
-		for _, prefix := range ignoredPrefixes {
-			if strings.HasPrefix(trimmedLine, prefix) {
-				ignore = true
-				break
-			}
-		}
-		if ignore {
-			continue
-		}
-		// Check for non-ASCII characters in the rest of the lines
-		for _, c := range line {
-			if c > unicode.MaxASCII {
-				isCommitLegal = false
-				break
-			}
-		}
-	}
-
-	if !isCommitLegal {
+	if !checkMsg(msg) {
 		return fmt.Errorf("Non-ASCII characters in commit messages:\n%s", msg)
 	}
 	return nil
diff --git a/pkg/healthcheck/commit_test.go b/pkg/healthcheck/commit_test.go
new file mode 100644
index 0000000..a987a78
--- /dev/null
+++ b/pkg/healthcheck/commit_test.go
@@ -0,0 +1,83 @@
+package healthcheck
+
+import (
+	"testing"
+)
+
+func TestCheckMsg(t *testing.T) {
+	tests := []struct {
+		name     string
+		message  string
+		expected bool
+	}{
+		{
+			name:     "Valid ASCII message",
+			message:  "This is a valid commit message",
+			expected: true,
+		},
+		{
+			name:     "Message with non-ASCII character",
+			message:  "This message contains a non-ASCII character: é",
+			expected: false,
+		},
+		{
+			name:     "Message with ignored prefix",
+			message:  "First line\nCo-authored-by: John Doe <john@example.com>\nThis is a valid message",
+			expected: true,
+		},
+		{
+			name:     "Message with ignored prefix and non-ASCII character in content",
+			message:  "First line\nCo-authored-by: John Doe <john@example.com>\nThis message has a non-ASCII character: ñ",
+			expected: false,
+		},
+		{
+			name:     "Message with ignored prefix in the first line",
+			message:  "Co-authored-by: Jöhn Döe <john@example.com>",
+			expected: false,
+		},
+		{
+			name:     "Multi-line message with all valid ASCII",
+			message:  "First line\nSecond line\nThird line",
+			expected: true,
+		},
+		{
+			name:     "Multi-line message with non-ASCII in middle",
+			message:  "First line\nSecond line with ö\nThird line",
+			expected: false,
+		},
+		{
+			name:     "Multi-line message with non-ASCII in the first line",
+			message:  "First line with ö\nSecond line",
+			expected: false,
+		},
+		{
+			name:     "Message with multiple ignored prefixes",
+			message:  "First line\nCo-authored-by: John Doe <john@example.com>\nReviewed-by: Jane Smith <jane@example.com>\nValid content",
+			expected: true,
+		},
+		{
+			name:     "Empty message",
+			message:  "",
+			expected: true,
+		},
+		{
+			name:     "Message with only whitespace",
+			message:  "   \n  \t  ",
+			expected: true,
+		},
+		{
+			name:     "Message with non-ASCII after ignored prefix",
+			message:  "First line\nCo-authored-by: John Doe <john@example.com>\nReviewed-by: Jöhn Döe <john@example.com>",
+			expected: true,
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			result := checkMsg(tt.message)
+			if result != tt.expected {
+				t.Errorf("checkMsg(%q) = %v, want %v", tt.message, result, tt.expected)
+			}
+		})
+	}
+}