Match Jellyfin punctuation variants using provider metadata and strict fallback

backend/app/clients/jellyfin.py

@@ -1,3 +1,4 @@
+import re
 from typing import Any, Dict, Optional
 import httpx
 import time
@@ -192,15 +193,27 @@ class JellyfinClient(ApiClient):
             "SearchTerm": term,
             "IncludeItemTypes": ",".join(item_types or []),
             "Recursive": "true",
-            "Fields": "Path,MediaSources",
+            "Fields": "Path,MediaSources,ProviderIds,OriginalTitle,SortName",
             "Limit": limit,
         }
         headers = self._emby_headers()
         try:
             async with httpx.AsyncClient(timeout=10.0) as client:
-                response = await client.get(url, headers=headers, params=params)
-                response.raise_for_status()
-                result = response.json()
+                normalized = ' '.join(re.sub(r"[^\w\s]", ' ', term, flags=re.UNICODE).split())
+                terms = list(dict.fromkeys([term, normalized]))
+                if normalized != term and normalized.split():
+                    terms.append(max(normalized.split(), key=len))
+                items = {}
+                for search_term in dict.fromkeys(terms):
+                    if not search_term:
+                        continue
+                    response = await client.get(url, headers=headers, params={**params, "SearchTerm": search_term})
+                    response.raise_for_status()
+                    payload = response.json()
+                    for item in payload.get('Items', []):
+                        if isinstance(item, dict) and item.get('Id'):
+                            items[item['Id']] = item
+                result = {'Items': list(items.values()), 'TotalRecordCount': len(items)}
             duration_ms = round((time.perf_counter() - started_at) * 1000, 2)
             finish_remote_call(
                 operation_event_id,

backend/app/services/snapshot.py

@@ -137,12 +137,10 @@ def jellyfin_item_matches_request(
     request_provider_ids = extract_request_provider_ids(request_payload or {})
     item_provider_ids = extract_request_provider_ids(item)
 
-    provider_priority = ("tmdb", "tvdb", "imdb")
-    for key in provider_priority:
-        request_id = request_provider_ids.get(key)
-        item_id = item_provider_ids.get(key)
-        if request_id and item_id and request_id == item_id:
-            return True
+    shared = set(request_provider_ids) & set(item_provider_ids)
+    if shared:
+        # Conflicting metadata must never fall through to title matching.
+        return all(request_provider_ids[key] == item_provider_ids[key] for key in shared)
 
     request_title = _normalize_media_title(title)
     if not request_title:
@@ -169,11 +167,6 @@ def jellyfin_item_matches_request(
     if request_title in item_titles:
         return True
 
-    if request_type == RequestType.tv:
-        for candidate in item_titles:
-            if candidate and (candidate.startswith(request_title) or request_title.startswith(candidate)):
-                return True
-
     return False
 
 

backend/tests/test_jellyfin_matching.py

@@ -0,0 +1,30 @@
+import unittest
+from unittest.mock import patch
+import httpx
+from backend.app.clients.jellyfin import JellyfinClient
+from backend.app.services.snapshot import jellyfin_item_matches_request
+from backend.app.models import RequestType
+
+class JellyfinMatchingTests(unittest.IsolatedAsyncioTestCase):
+    async def test_search_includes_punctuation_variant_and_provider_metadata(self):
+        calls=[]
+        def handle(request):
+            calls.append(request)
+            items=[{'Id':'animated','Name':'Avatar: The Last Airbender','ProductionYear':2005,'ProviderIds':{'Tmdb':'246'}}] if ':' in request.url.params['SearchTerm'] else [{'Id':'live','Name':'Avatar the Last Airbender','ProductionYear':2024,'ProviderIds':{'Tmdb':'82452'}}]
+            return httpx.Response(200,json={'Items':items})
+        original=httpx.AsyncClient
+        with patch('backend.app.clients.jellyfin.httpx.AsyncClient',side_effect=lambda **kw:original(transport=httpx.MockTransport(handle),**kw)):
+            result=await JellyfinClient('http://jellyfin','test').search_items('Avatar: The Last Airbender',['Series'])
+        self.assertEqual({i['Id'] for i in result['Items']},{'live','animated'})
+        self.assertTrue(all('ProviderIds' in r.url.params['Fields'] for r in calls))
+        matches=[i for i in result['Items'] if jellyfin_item_matches_request(i,title='Avatar: The Last Airbender',year=2024,request_type=RequestType.tv,request_payload={'tmdbId':82452})]
+        self.assertEqual([i['Id'] for i in matches],['live'])
+
+    def test_fallback_rejects_remakes_prefixes_and_conflicting_ids(self):
+        def match(item,payload=None):
+            return jellyfin_item_matches_request(item,title='Avatar: The Last Airbender',year=2024,request_type=RequestType.tv,request_payload=payload)
+        self.assertTrue(match({'Name':'Avatar the Last Airbender','ProductionYear':2024}))
+        self.assertFalse(match({'Name':'Avatar the Last Airbender','ProductionYear':2005}))
+        self.assertFalse(match({'Name':'Avatar','ProductionYear':2024}))
+        self.assertFalse(match({'Name':'Avatar the Last Airbender','ProductionYear':2024,'ProviderIds':{'Tmdb':'246'}},{'tmdbId':82452}))
+        self.assertTrue(match({'Name':'Localized title','ProviderIds':{'Tmdb':'82452'}},{'tmdbId':82452}))