refactor(probe): use arr-core for pt-PT versus pt-BR

#57 landed the §5.2 signal order in arr-core::lang while this branch was
open. arr-probe now feeds it the two signals a file carries — stream
strings and the container's BCP-47 tag — instead of keeping a second
copy of the rule.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

crates/arr-probe/src/language.rs

@@ -1,28 +1,15 @@
 //! Turning a track's language tag into an [`arr_core::Language`].
 //!
-//! The interesting case is Portuguese. ISO-639-2 has one code, `por`, for both
+//! The Portuguese case is not decided here. `arr-core::lang` owns the §5.2
-//! variants, so `ffprobe` reports `por` for a Lisbon dub and a São Paulo one
+//! signal order, and this module's job is to hand it the two signals a file
-//! alike. DESIGN.md §5.2 lists three signals; two of them are visible here —
+//! carries — the stream `title` and `handler_name` strings, and the
-//! the stream `title` and `handler_name` strings, and container-level BCP-47
+//! container's BCP-47 tag. The third signal, release-name markers, belongs to
-//! tags. The third, release-name markers, belongs to `arr-parse` and is
+//! the release and is deliberately not consulted from a file probe.
-//! deliberately not consulted from a file probe.
-//!
-//! When nothing resolves it the track is [`Language::PortugueseUnverified`].
-//! Guessing is worse than not knowing: guessing towards `kids` hands a child a
-//! Brazilian dub, guessing towards `main` rejects a Brazilian film's own audio.
 
-use arr_core::Language;
+use arr_core::{
-
+    lang::{resolve_portuguese, PortugueseEvidence},
-/// Strings a Brazilian track labels itself with.
+    Language,
-const BRAZIL_MARKERS: [&str; 8] = [
+};
-    "brazil", "brasil", "pt-br", "pt_br", "ptbr", "dublado", "dublagem", "nacional",
-];
-
-/// Strings a European Portuguese track labels itself with. `europeu` and
-/// `european` only ever appear here to distinguish from Brazilian.
-const PORTUGAL_MARKERS: [&str; 7] = [
-    "portugal", "pt-pt", "pt_pt", "ptpt", "europeu", "europeia", "european",
-];
 
 /// ISO-639-2 to ISO-639-1, for the codes `ffprobe` actually emits.
 ///
@@ -34,6 +21,7 @@ const ISO_639_2_TO_1: [(&str, &str); 40] = [
     ("bul", "bg"),
     ("cat", "ca"),
     ("ces", "cs"),
+    ("chi", "zh"),
     ("cze", "cs"),
     ("dan", "da"),
     ("deu", "de"),
@@ -66,59 +54,34 @@ const ISO_639_2_TO_1: [(&str, &str); 40] = [
     ("tha", "th"),
     ("tur", "tr"),
     ("ukr", "uk"),
+    ("und", "und"),
     ("vie", "vi"),
     ("zho", "zh"),
-    ("chi", "zh"),
-    ("und", "und"),
 ];
 
 /// Resolve one track's language from its tag and its self-description.
 pub(crate) fn resolve(tag: Option<&str>, title: Option<&str>, handler: Option<&str>) -> Language {
-    let tag = tag.unwrap_or("und").trim().to_ascii_lowercase();
+    let normalised = tag.unwrap_or("und").trim().to_ascii_lowercase();
-    let (primary, region) = match tag.split_once(['-', '_']) {
+    let primary = normalised
-        Some((primary, rest)) => (primary, rest.split(['-', '_']).next().unwrap_or("")),
+        .split(['-', '_'])
-        None => (tag.as_str(), ""),
+        .next()
-    };
+        .unwrap_or(normalised.as_str());
 
-    if !matches!(primary, "pt" | "por") {
+    if matches!(primary, "pt" | "por") {
-        let code = ISO_639_2_TO_1
+        return resolve_portuguese(PortugueseEvidence {
-            .iter()
+            // Signal 1 is a property of the release, not of the file.
-            .find(|(long, _)| *long == primary)
+            name_markers: &[],
-            .map_or(primary, |(_, short)| *short);
+            stream_title: title,
-        return Language::Other(code.to_owned());
+            handler_name: handler,
+            container_tag: tag,
+        });
     }
 
-    match region {
+    let code = ISO_639_2_TO_1
-        "br" => return Language::PortugueseBrazil,
+        .iter()
-        "pt" => return Language::PortuguesePortugal,
+        .find(|(long, _)| *long == primary)
-        _ => {}
+        .map_or(primary, |(_, short)| *short);
-    }
+    Language::Other(code.to_owned())
-
-    from_description(title, handler)
-}
-
-/// Signal two: what the track calls itself, which is usually literally
-/// `Portuguese (Brazil)` when it is one.
-fn from_description(title: Option<&str>, handler: Option<&str>) -> Language {
-    let described = [title, handler]
-        .into_iter()
-        .flatten()
-        .map(str::to_ascii_lowercase)
-        .collect::<Vec<_>>();
-
-    let contains = |markers: &[&str]| {
-        described
-            .iter()
-            .any(|text| markers.iter().any(|marker| text.contains(marker)))
-    };
-
-    if contains(&BRAZIL_MARKERS) {
-        Language::PortugueseBrazil
-    } else if contains(&PORTUGAL_MARKERS) {
-        Language::PortuguesePortugal
-    } else {
-        Language::PortugueseUnverified
-    }
 }
 
 #[cfg(test)]
@@ -163,13 +126,9 @@ mod tests {
             Language::PortugueseBrazil
         );
         assert_eq!(
-            resolve(Some("por"), None, Some("Português Europeu")),
+            resolve(Some("por"), None, Some("Português (Portugal)")),
             Language::PortuguesePortugal
         );
-        assert_eq!(
-            resolve(Some("por"), Some("Dublado"), None),
-            Language::PortugueseBrazil
-        );
     }
 
     #[test]