Normalize oauth email username (#28561)

models/user/user.go

@@ -10,8 +10,10 @@ import (
 	"fmt"
 	"net/url"
 	"path/filepath"
+	"regexp"
 	"strings"
 	"time"
+	"unicode"
 
 	_ "image/jpeg" // Needed for jpeg support
 
@@ -29,6 +31,9 @@ import (
 	"code.gitea.io/gitea/modules/util"
 	"code.gitea.io/gitea/modules/validation"
 
+	"golang.org/x/text/runes"
+	"golang.org/x/text/transform"
+	"golang.org/x/text/unicode/norm"
 	"xorm.io/builder"
 )
 
@@ -515,6 +520,26 @@ func GetUserSalt() (string, error) {
 	return hex.EncodeToString(rBytes), nil
 }
 
+// Note: The set of characters here can safely expand without a breaking change,
+// but characters removed from this set can cause user account linking to break
+var (
+	customCharsReplacement    = strings.NewReplacer("Æ", "AE")
+	removeCharsRE             = regexp.MustCompile(`['´\x60]`)
+	removeDiacriticsTransform = transform.Chain(norm.NFD, runes.Remove(runes.In(unicode.Mn)), norm.NFC)
+	replaceCharsHyphenRE      = regexp.MustCompile(`[\s~+]`)
+)
+
+// normalizeUserName returns a string with single-quotes and diacritics
+// removed, and any other non-supported username characters replaced with
+// a `-` character
+func NormalizeUserName(s string) (string, error) {
+	strDiacriticsRemoved, n, err := transform.String(removeDiacriticsTransform, customCharsReplacement.Replace(s))
+	if err != nil {
+		return "", fmt.Errorf("Failed to normalize character `%v` in provided username `%v`", s[n], s)
+	}
+	return replaceCharsHyphenRE.ReplaceAllLiteralString(removeCharsRE.ReplaceAllLiteralString(strDiacriticsRemoved, ""), "-"), nil
+}
+
 var (
 	reservedUsernames = []string{
 		".",

models/user/user_test.go

@@ -544,3 +544,31 @@ func Test_ValidateUser(t *testing.T) {
 		assert.EqualValues(t, expected, err == nil, fmt.Sprintf("case: %+v", kase))
 	}
 }
+
+func Test_NormalizeUserFromEmail(t *testing.T) {
+	testCases := []struct {
+		Input             string
+		Expected          string
+		IsNormalizedValid bool
+	}{
+		{"test", "test", true},
+		{"Sinéad.O'Connor", "Sinead.OConnor", true},
+		{"Æsir", "AEsir", true},
+		// \u00e9\u0065\u0301
+		{"éé", "ee", true},
+		{"Awareness Hub", "Awareness-Hub", true},
+		{"double__underscore", "double__underscore", false}, // We should consider squashing double non-alpha characters
+		{".bad.", ".bad.", false},
+		{"new😀user", "new😀user", false}, // No plans to support
+	}
+	for _, testCase := range testCases {
+		normalizedName, err := user_model.NormalizeUserName(testCase.Input)
+		assert.NoError(t, err)
+		assert.EqualValues(t, testCase.Expected, normalizedName)
+		if testCase.IsNormalizedValid {
+			assert.NoError(t, user_model.IsUsableUsername(normalizedName))
+		} else {
+			assert.Error(t, user_model.IsUsableUsername(normalizedName))
+		}
+	}
+}