Filter domain-like garbage out of genres

MusicBrainz community tags sometimes carry review-site domain names
(e.g. "musikexpress.de", "rollingstone.de") that leaked into the genre
field. Add a shared looks_like_url predicate and apply it at ingestion
(is_genre_tag, pick_genre_string) and at report read-time
(split_genre_list, top_genre_labels) so both new and already-cached
garbage are dropped.

src/db.rs

@@ -919,11 +919,26 @@ pub fn is_deprioritised_genre(genre: &str) -> bool {
     canonical_genre_key(genre).split_whitespace().count() < 2
 }
 
+/// Return true if a genre label looks like a website domain or URL rather than
+/// a real genre.
+///
+/// MusicBrainz community tags sometimes carry magazine / review-site names
+/// (e.g. "musikexpress.de", "rollingstone.de", "pitchfork.com") that leak into
+/// the genre field. Genuine genre labels never contain '.', '/' or ':', so any
+/// label containing one of those is rejected.
+pub fn looks_like_url(label: &str) -> bool {
+    label.contains('.') || label.contains('/') || label.contains(':')
+}
+
 /// Split a comma-separated genre string into cleaned labels.
+///
+/// Domain-like garbage (see [`looks_like_url`]) is dropped here so that any
+/// such values already cached in the database never reach the report.
 fn split_genre_list(csv: &str) -> Vec<String> {
     csv.split(',')
         .map(str::trim)
         .filter(|s| !s.is_empty())
+        .filter(|s| !looks_like_url(s))
         .map(|s| s.to_string())
         .collect()
 }

src/enrich.rs

@@ -719,11 +719,15 @@ const TAG_BLOCKLIST: &[&str] = &[
 ///
 /// Rejects:
 /// - Tags containing digits (years like "2022", encodings like "iso-8859-1")
+/// - Domain-like tags (review-site names such as "musikexpress.de")
 /// - Known non-genre technical / language tags
 fn is_genre_tag(tag: &str) -> bool {
     if tag.chars().any(|c| c.is_ascii_digit()) {
         return false;
     }
+    if db::looks_like_url(tag) {
+        return false;
+    }
     let lower = tag.to_ascii_lowercase();
     !TAG_BLOCKLIST.contains(&lower.as_str())
 }
@@ -735,7 +739,11 @@ fn is_genre_tag(tag: &str) -> bool {
 /// 2. Community `tags` (filtered to entries with at least one vote and
 ///    passing the genre-tag heuristic to strip technical metadata)
 fn pick_genre_string(genres: Vec<MbGenre>, tags: Vec<MbTag>) -> Option<String> {
-    let mut names: Vec<String> = genres.into_iter().map(|g| g.name).collect();
+    let mut names: Vec<String> = genres
+        .into_iter()
+        .map(|g| g.name)
+        .filter(|n| !db::looks_like_url(n))
+        .collect();
     if names.is_empty() {
         names = tags
             .into_iter()
@@ -1553,6 +1561,43 @@ mod tests {
         assert_eq!(from_tags.as_deref(), Some("electronic"));
     }
 
+    #[test]
+    fn test_pick_genre_string_rejects_domain_tags() {
+        // Review-site tags such as "musikexpress.de" must never be picked.
+        let from_tags = pick_genre_string(
+            vec![],
+            vec![
+                MbTag {
+                    name: "musikexpress.de".to_string(),
+                    count: 3,
+                },
+                MbTag {
+                    name: "rollingstone.de".to_string(),
+                    count: 2,
+                },
+                MbTag {
+                    name: "shoegaze".to_string(),
+                    count: 1,
+                },
+            ],
+        );
+        assert_eq!(from_tags.as_deref(), Some("shoegaze"));
+
+        // Domains must also be stripped from the curated-genres path.
+        let from_genres = pick_genre_string(
+            vec![
+                MbGenre {
+                    name: "pitchfork.com".to_string(),
+                },
+                MbGenre {
+                    name: "dream pop".to_string(),
+                },
+            ],
+            vec![],
+        );
+        assert_eq!(from_genres.as_deref(), Some("dream pop"));
+    }
+
     #[test]
     fn test_uncached_albums_query() {
         // Verify the uncached_albums query works correctly.

src/report.rs

@@ -1727,6 +1727,7 @@ fn top_genre_labels(genre_csv: &str, max_labels: usize) -> Vec<String> {
         .split(',')
         .map(str::trim)
         .filter(|s| !s.is_empty())
+        .filter(|s| !db::looks_like_url(s))
         .map(|s| s.to_string())
         .collect();