websocket utf8: move utf8 functions from websocket to encoding.utf8, add utf8_test.v (4/4) (#5924)

vlib/encoding/utf8/utf8.v

@@ -1,8 +1,4 @@
-module websocket
-
-pub fn utf8_validate_str(str string) bool {
-	return utf8_validate(str.str, str.len)
-}
+module utf8
 
 struct Utf8State {
 mut:
@@ -11,7 +7,11 @@ mut:
 	failed   bool
 }
 
-pub fn utf8_validate(data byteptr, len int) bool {
+pub fn validate_str(str string) bool {
+	return validate(str.str, str.len)
+}
+
+pub fn validate(data byteptr, len int) bool {
 	mut state := Utf8State{}
 	for i := 0; i < len; i++ {
 		s := data[i]
@@ -22,7 +22,6 @@ pub fn utf8_validate(data byteptr, len int) bool {
 		if state.failed {
 			return false
 		}
-		// i++ //fast forward
 	}
 	return !state.failed && state.subindex <= 0
 }

vlib/encoding/utf8/utf8_test.v

@@ -0,0 +1,9 @@
+import encoding.utf8 { validate_str }
+
+fn test_validate_str() {
+	assert validate_str('añçá') == true
+	assert validate_str('\x61\xC3\xB1\xC3\xA7\xC3\xA1') == true
+	assert validate_str('\xC0\xC1') == false
+	assert validate_str('\xF5\xFF') == false
+	assert validate_str('\xE0\xEF') == false
+}

vlib/net/websocket/ws.v

@@ -3,6 +3,7 @@ module websocket
 import net
 import net.urllib
 import encoding.base64
+import encoding.utf8
 import eventbus
 import sync
 import log
@@ -478,7 +479,7 @@ pub fn (mut ws Client) read() int {
 			}
 			payload[payload_len] = `\0`
 			if frame.opcode == .text_frame && payload_len > 0 {
-				if !utf8_validate(payload, int(payload_len)) {
+				if !utf8.validate(payload, int(payload_len)) {
 					ws.log.error('malformed utf8 payload')
 					ws.send_error_event('Recieved malformed utf8.')
 					ws.close(1007, 'malformed utf8 payload')
@@ -563,7 +564,7 @@ pub fn (mut ws Client) read() int {
 			payload_len -= 2
 			reason = string(&data[header_len])
 			ws.log.info('Closing with reason: $reason & code: $code')
-			if reason.len > 1 && !utf8_validate(reason.str, reason.len) {
+			if reason.len > 1 && !utf8.validate(reason.str, reason.len) {
 				ws.log.error('malformed utf8 payload')
 				ws.send_error_event('Recieved malformed utf8.')
 				ws.close(1007, 'malformed utf8 payload')