Fix translation detection for Cyrillic and ISO-8859-1

- Skip ISO-8859-1 Latin chars (0xC0-0xFF) to avoid Swedish false positives
- Require 3+ non-Latin chars to trigger translation
- Handle both UTF-8 and single-byte Cyrillic encodings
- Update README with ai_port option

README.md

@@ -103,7 +103,7 @@ ai_target_lang=english
 ai_skip_langs=swedish,english
 ```
 
-When configured, foreign-script messages are shown immediately with an italic translation appearing below after 1-5 seconds. Leave values empty to disable.
+Host can include port (`host:port`) or use a separate `ai_port=` line. When configured, foreign-script messages are shown immediately with an italic translation appearing below after 1-5 seconds. Leave values empty to disable.
 
 ## License
 

main.c

@@ -173,24 +173,27 @@ static void ai_config_load(void)
 static int needs_translation(const char *text)
 {
 	if (!ai_cfg.enabled) return 0;
-	/* Check for non-ASCII characters that suggest non-Latin script */
+	int non_latin_count = 0;
 	for (const unsigned char *p = (const unsigned char *)text; *p; p++) {
-		if (*p >= 0xC0) {
+		if (*p >= 0x80) {
-			/* Multi-byte UTF-8: check if it's beyond Latin Extended (U+0250+) */
+			if ((*p & 0xE0) == 0xC0 && (p[1] & 0xC0) == 0x80) {
-			unsigned int cp = 0;
+				unsigned int cp = ((*p & 0x1F) << 6) | (p[1] & 0x3F);
-			if ((*p & 0xE0) == 0xC0 && p[1]) {
+				if (cp > 0x024F) non_latin_count++;
-				cp = ((*p & 0x1F) << 6) | (p[1] & 0x3F);
+				p++;
-			} else if ((*p & 0xF0) == 0xE0 && p[1] && p[2]) {
+			} else if ((*p & 0xF0) == 0xE0 && (p[1] & 0xC0) == 0x80 && (p[2] & 0xC0) == 0x80) {
-				cp = ((*p & 0x0F) << 12) | ((p[1] & 0x3F) << 6) | (p[2] & 0x3F);
+				non_latin_count++;
-			} else if ((*p & 0xF8) == 0xF0 && p[1] && p[2] && p[3]) {
+				p += 2;
-				cp = ((*p & 0x07) << 18) | ((p[1] & 0x3F) << 12) |
+			} else if ((*p & 0xF8) == 0xF0 && (p[1] & 0xC0) == 0x80 && (p[2] & 0xC0) == 0x80 && (p[3] & 0xC0) == 0x80) {
-				     ((p[2] & 0x3F) << 6) | (p[3] & 0x3F);
+				non_latin_count++;
-			}
+				p += 3;
-			/* Skip Latin Extended, keep Cyrillic, CJK, Arabic, etc. */
+			} else {
-			if (cp > 0x024F) return 1;
+				/* Not valid UTF-8 — skip ISO-8859-1 Latin chars (0xC0-0xFF) */
+				if (*p < 0xC0) non_latin_count++;
 			}
 		}
-	return 0;
+	}
+	/* Require at least 3 non-Latin chars to avoid triggering on stray åäö */
+	return non_latin_count >= 3;
 }
 
 static void translate_async(const char *text, int level)