Improve web UI

furumi-server/src/web/meta.rs

@@ -137,28 +137,29 @@ fn extract_tags(
     cover: &mut Option<(Vec<u8>, String)>,
 ) {
     for tag in tags {
+        let value = fix_encoding(tag.value.to_string());
         if let Some(key) = tag.std_key {
             match key {
                 StandardTagKey::TrackTitle => {
-                    *title = Some(tag.value.to_string());
+                    *title = Some(value);
                 }
                 StandardTagKey::Artist | StandardTagKey::Performer => {
                     if artist.is_none() {
-                        *artist = Some(tag.value.to_string());
+                        *artist = Some(value);
                     }
                 }
                 StandardTagKey::Album => {
-                    *album = Some(tag.value.to_string());
+                    *album = Some(value);
                 }
                 StandardTagKey::TrackNumber => {
                     if track.is_none() {
-                        *track = tag.value.to_string().parse().ok();
+                        *track = value.parse().ok();
                     }
                 }
                 StandardTagKey::Date | StandardTagKey::OriginalDate => {
                     if year.is_none() {
                         // Parse first 4 characters as year
-                        *year = tag.value.to_string()[..4.min(tag.value.to_string().len())].parse().ok();
+                        *year = value[..4.min(value.len())].parse().ok();
                     }
                 }
                 _ => {}
@@ -173,3 +174,31 @@ fn extract_tags(
         }
     }
 }
+
+/// Heuristic to fix mojibake (CP1251 bytes interpreted as Latin-1/Windows-1252)
+fn fix_encoding(s: String) -> String {
+    // If it's already a valid UTF-8 string that doesn't look like mojibake, return it.
+    // Mojibake looks like characters from Latin-1 Supplement (0xC0-0xFF)
+    // where they should be Cyrillic.
+    
+    let bytes: Vec<u8> = s.chars().map(|c| c as u32).filter(|&c| c <= 255).map(|c| c as u8).collect();
+    
+    // If the length is different, it means there were characters > 255, so it's not simple Latin-1 mojibake.
+    if bytes.len() != s.chars().count() {
+        return s;
+    }
+
+    // Check if it's likely CP1251. Russian characters in CP1251 are 0xC0-0xFF.
+    // In Latin-1 these are characters like À-ÿ.
+    let has_mojibake = bytes.iter().any(|&b| b >= 0xC0);
+    if !has_mojibake {
+        return s;
+    }
+
+    let (decoded, _, errors) = encoding_rs::WINDOWS_1251.decode(&bytes);
+    if errors {
+        return s;
+    }
+    
+    decoded.into_owned()
+}