Use Redump + No-Intro DATs together for games; dedupe fuzzy matches

README.md

@@ -52,8 +52,11 @@ python3 -m app.main
 ## First-time setup
 
 1. Put your DAT files in `data/dats/`:
-   - games: Redump `Sony - PlayStation Portable`
+   - games: Redump `Sony - PlayStation Portable` (authoritative hashes)
-   - videos: Redump `Sony - PlayStation Portable (UMD Video)`
+     **and** No-Intro `Sony - PlayStation Portable` (supplies the serials
+     Redump lacks — keeps serial-based matching working).
+   - videos: No-Intro `Sony - PlayStation Portable (UMD Video)` (Redump has
+     no UMD Video DAT).
 2. Ingest them into the local SQLite library:
 
    ```bash

app/matcher.py

@@ -181,6 +181,16 @@ def match(conn, record):
         return "unknown", None
     top = [cand for ratio, cand in scored if ratio >= best_ratio - 0.001]
     top = _narrow(conn, top, size)
+    # The same disc can exist in multiple catalogs (e.g. No-Intro + Redump);
+    # collapse identical entries so they don't read as an ambiguous match.
+    seen = set()
+    deduped = []
+    for cand in top:
+        key = (cand.get("name"), size)
+        if key not in seen:
+            seen.add(key)
+            deduped.append(cand)
+    top = deduped
     if len(top) == 1:
         return f"fuzzy:{best_ratio:.2f}", top[0]
     return "unknown", None

data/dats/Sony - PlayStation Portable (No-Intro).dat

@@ -1,5 +1,5 @@
 clrmamepro (
-	name "Sony - PlayStation Portable"
+	name "Sony - PlayStation Portable (No-Intro)"
 	description "Sony - PlayStation Portable"
 	version "2021.10.24"
 	homepage "http://github.com/robloach/libretro-dats"