feat: add whitelist char support to nonascii check

This commit brings support of whitelisted characters during repo
healthcheck/non-ascii file check. Supported by an extra switch to
`repo-health-checker`, `-whitelistedChars`. The argument takes a
comma-separated list of non-ASCII characters and ignores them during
repo healthcheck. Illegal cmdline input is logged by the logger.

cmd/repo-health-checker/main.go

@@ -45,6 +45,7 @@ var (
 	checkFileNameList string
 	checkFileSumList  string
 	metaFile          []string
+	whitelistedChars  string
 	allowedDomainList string
 	actorCsvPath      string
 	showVersion       *bool
@@ -57,6 +58,7 @@ 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(&whitelistedChars, "whitelistedChars", "", "comma-separated list of non-ASCII characters allowed in files")
 	flag.StringVar(&allowedDomainList, "allowedDomainList", "sjtu.edu.cn", "comma-separated list of allowed domains for commit author email")
 	flag.StringVar(&actorCsvPath, "actorCsvPath", "/home/tt/.config/joj/students.csv", "path to actor csv file")
 	parseMultiValueFlag(&metaFile, "meta", "meta files to check")
@@ -74,12 +76,14 @@ func main() {
 		"repoSize", repoSize,
 		"checkFileNameList", checkFileNameList,
 		"checkFileSumList", checkFileSumList,
+		"whitelistedChars", whitelistedChars,
 		"meta", metaFile,
 	)
 	res := healthcheck.All(
 		rootDir,
 		checkFileNameList,
 		checkFileSumList,
+		whitelistedChars,
 		allowedDomainList,
 		actorCsvPath,
 		metaFile,

pkg/healthcheck/all.go

@@ -12,7 +12,8 @@ type Result struct {
 }
 
 func All(
-	rootDir, checkFileNameList, checkFileSumList, allowedDomainList, actorCsvPath string,
+	rootDir, checkFileNameList, checkFileSumList, whitelistedCharsCSV,
+	allowedDomainList, actorCsvPath string,
 	metaFile []string, repoSize float64,
 ) (res Result) {
 	var err error
@@ -44,7 +45,7 @@ func All(
 	} else {
 		res.Msg += "### Meta File Check Passed\n"
 	}
-	err = NonASCIIFiles(rootDir)
+	err = NonASCIIFiles(rootDir, whitelistedCharsCSV)
 	if err != nil {
 		res.Msg += fmt.Sprintf("### Non-ASCII Characters File Check Failed:\n%s\n", err.Error())
 		res.Failed = true

pkg/healthcheck/nonascii.go

@@ -8,13 +8,49 @@ import (
 	"path/filepath"
 	"strings"
 	"unicode"
+	"unicode/utf8"
 
 	"github.com/go-git/go-git/v5/plumbing/format/gitattributes"
 )
 
+// Read the list of comma-separated allowed characters from command line and convert it to a hashmap.
+func parseWhitelistedChars(csv string) map[rune]struct{} {
+	whitelist := make(map[rune]struct{})
+	if strings.TrimSpace(csv) == "" {
+		return whitelist
+	}
+
+	for _, raw := range strings.Split(csv, ",") {
+		elem := strings.TrimSpace(raw)
+		if elem == "" {
+			slog.Warn("ignoring invalid whitelisted-chars element", "element", raw, "reason", "empty element")
+			continue
+		}
+
+		if utf8.RuneCountInString(elem) != 1 {
+			slog.Warn("ignoring invalid whitelisted-chars element", "element", elem, "reason", "element must be exactly one character")
+			continue
+		}
+
+		ch, _ := utf8.DecodeRuneInString(elem)
+		if ch == utf8.RuneError {
+			slog.Warn("ignoring invalid whitelisted-chars element", "element", elem, "reason", "invalid utf-8 rune")
+			continue
+		}
+		if ch <= unicode.MaxASCII {
+			slog.Warn("ignoring invalid whitelisted-chars element", "element", elem, "reason", "ASCII characters are not allowed")
+			continue
+		}
+
+		whitelist[ch] = struct{}{}
+	}
+
+	return whitelist
+}
+
 // getNonASCII retrieves a list of files in the specified root directory that contain non-ASCII characters.
 // It searches for non-ASCII characters in each file's content and returns a list of paths to files containing non-ASCII characters.
-func getNonASCII(root string) ([]string, error) {
+func getNonASCII(root string, whitelist map[rune]struct{}) ([]string, error) {
 	var nonASCII []string
 	gitattrExist := true
 	var matcher gitattributes.Matcher
@@ -70,6 +106,9 @@ func getNonASCII(root string) ([]string, error) {
 		for scanner.Scan() {
 			cont := true
 			for _, c := range scanner.Text() {
+				if _, ok := whitelist[c]; ok {
+					continue
+				}
 				if c > unicode.MaxASCII {
 					nonASCII = append(nonASCII, "\t"+path)
 					cont = false
@@ -89,8 +128,10 @@ func getNonASCII(root string) ([]string, error) {
 
 // NonASCIIFiles checks for non-ASCII characters in files within the specified root directory.
 // It prints a message with the paths to files containing non-ASCII characters, if any.
-func NonASCIIFiles(root string) error {
+// Additionally it accept a list of whitelisted characters that are allowed, repo-wide.
-	nonASCII, err := getNonASCII(root)
+func NonASCIIFiles(root, whitelistedCharsCSV string) error {
+	whitelist := parseWhitelistedChars(whitelistedCharsCSV)
+	nonASCII, err := getNonASCII(root, whitelist)
 	if err != nil {
 		slog.Error("getting non-ascii", "err", err)
 		return fmt.Errorf("error getting non-ascii: %w", err)