Refactor gazetteer annotation format

The gazetteer now annotates text with `Canonical [?original]`. This
prioritizes the corrected term for the LLM while keeping the original
transcription as a fallback.

This change aligns the gazetteer's annotation strategy with the LLM's
prompt, which expects corrections to be marked for review. Previously,
the format was `original [?Canonical]`, which could lead to the LLM
using the potentially incorrect original term.

server/src/gazetteer/mod.rs

@@ -4,8 +4,10 @@
 //! The LLM cannot know invented proper names (drug brands, uncommon
 //! technical terms). Whisper misrenders them, usually within a couple
 //! of edits of the real spelling. This module finds those tokens and
-//! rewrites them as `token [?Canonical]`. The LLM decides in context
-//! whether to adopt the hint.
+//! rewrites them as `Canonical [?original]` — the vocabulary form
+//! becomes the anchor, the original Whisper token sits in the bracket
+//! as a veto option. The LLM decides in context whether to keep the
+//! canonical or revert to the original.
 //!
 //! Matcher: single-stage linear Damerau-Levenshtein scan over all
 //! entries, filtered by distance ≤ MAX_EDIT_DISTANCE. At typical sizes
@@ -91,9 +93,12 @@ impl Gazetteer {
         self.entries.push(entry.into());
     }
 
-    /// Annotate `text` by appending ` [?Canonical]` after any word-token
-    /// that's within edit-distance 2 of a gazetteer entry (and not an
-    /// exact match). Non-word segments pass through byte-identical.
+    /// Annotate `text` by replacing any word-token within edit-distance
+    /// 2 of a gazetteer entry with `Canonical [?original]`. The
+    /// canonical takes the anchor position so the LLM's default action
+    /// is to adopt it; the original Whisper token stays available in
+    /// the bracket as a veto option. Exact matches and non-word
+    /// segments pass through unchanged.
     pub fn annotate(&self, text: &str) -> String {
         if self.is_empty() {
             return text.to_string();
@@ -102,10 +107,16 @@ impl Gazetteer {
         for seg in tokenize(text) {
             match seg {
                 Segment::Word(w) => match self.find_candidate(w) {
-                    Some(c) => {
-                        out.push_str(w);
-                        out.push_str(" [?");
+                    Some((c, d)) => {
+                        tracing::info!(
+                            token = w,
+                            canonical = c,
+                            distance = d,
+                            "gazetteer hit"
+                        );
                         out.push_str(c);
+                        out.push_str(" [?");
+                        out.push_str(w);
                         out.push(']');
                     }
                     None => out.push_str(w),
@@ -116,7 +127,7 @@ impl Gazetteer {
         out
     }
 
-    fn find_candidate(&self, token: &str) -> Option<&str> {
+    fn find_candidate(&self, token: &str) -> Option<(&str, usize)> {
         if token.chars().count() < MIN_TOKEN_LEN {
             return None;
         }
@@ -130,7 +141,7 @@ impl Gazetteer {
             .map(|c| (damerau_levenshtein(&lower, &c.to_lowercase()), c.as_ref()))
             .filter(|(d, _)| *d >= 1 && *d <= MAX_EDIT_DISTANCE)
             .min_by(|a, b| a.0.cmp(&b.0).then_with(|| a.1.cmp(b.1)))
-            .map(|(_, c)| c)
+            .map(|(d, c)| (c, d))
     }
 }
 
@@ -191,7 +202,7 @@ mod tests {
         let g = from_entries(&["Cerebrum"]);
         let out = g.annotate("Patient mit Blutung im Zerebrum.");
         assert!(
-            out.contains("Zerebrum [?Cerebrum]"),
+            out.contains("Cerebrum [?Zerebrum]"),
             "expected annotation, got: {out:?}"
         );
     }
@@ -239,7 +250,7 @@ mod tests {
         let g = from_entries(&["Carvedilol"]);
         let out = g.annotate("Patient nimmt Carvenilol.");
         assert!(
-            out.contains("Carvenilol [?Carvedilol]"),
+            out.contains("Carvedilol [?Carvenilol]"),
             "got: {out:?}"
         );
     }
@@ -250,7 +261,7 @@ mod tests {
         let g = from_entries(&["Berodual"]);
         let out = g.annotate("Patient bekommt Erodual.");
         assert!(
-            out.contains("Erodual [?Berodual]"),
+            out.contains("Berodual [?Erodual]"),
             "got: {out:?}"
         );
     }
@@ -262,7 +273,7 @@ mod tests {
         let g = from_entries(&["Cerebrum", "Cerebrom"]);
         let out = g.annotate("Patient mit Blutung im Zerebrum.");
         assert!(
-            out.contains("Zerebrum [?Cerebrum]"),
+            out.contains("Cerebrum [?Zerebrum]"),
             "expected Cerebrum as closer candidate, got: {out:?}"
         );
     }