fix: deduplicate draft revisions in search results

Different revisions of the same draft (e.g. draft-foo-bar-00, -01, -02)
were showing up as separate results. Now keeps only the highest-scoring
revision per base draft name.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

src/ietf_analyzer/search.py

@@ -93,21 +93,40 @@ class HybridSearch:
         self._embeddings_cache_time = now
         return self._embeddings_cache
 
+    @staticmethod
+    def _base_draft_name(name: str) -> str:
+        """Strip revision suffix to get the base draft name for dedup.
+
+        e.g. 'draft-wang-cats-odsi-02' → 'draft-wang-cats-odsi'
+        """
+        return re.sub(r'-\d{2,3}$', '', name)
+
     def search(self, query: str, top_k: int = 10) -> list[dict]:
         """Combine FTS5 keyword search + embedding similarity search.
 
         Returns ranked list of {name, title, score, excerpt, match_type}.
         Falls back to FTS5-only if Ollama is unavailable.
+        Deduplicates draft revisions, keeping the highest-scoring version.
         """
-        fts_results = self._fts_search(query, limit=top_k * 2)
+        fts_results = self._fts_search(query, limit=top_k * 3)
-        embed_results = self._embedding_search(query, limit=top_k * 2)
+        embed_results = self._embedding_search(query, limit=top_k * 3)
 
         if embed_results:
             merged = self._reciprocal_rank_fusion(fts_results, embed_results)
         else:
             merged = fts_results
 
-        return merged[:top_k]
+        # Deduplicate revisions of the same draft, keep best score
+        seen_bases: dict[str, int] = {}
+        deduped = []
+        for r in merged:
+            base = self._base_draft_name(r["name"])
+            if base not in seen_bases:
+                seen_bases[base] = len(deduped)
+                deduped.append(r)
+            # else: skip lower-ranked revision
+
+        return deduped[:top_k]
 
     @staticmethod
     def sanitize_fts_query(query: str) -> str: