From ed1b940486a09e2088207922ea4ce129658028a7 Mon Sep 17 00:00:00 2001
From: Cong <congusbongus@gmail.com>
Date: Sun, 15 Sep 2024 14:44:43 +1000
Subject: [PATCH] Add some blacklists/aliases and misc fixes for OSGL scraper
 #2706

---
 games/d.yaml                        |  2 +-
 originals/0scumm.yaml               |  9 --------
 originals/d.yaml                    | 11 ++++++++++
 originals/s.yaml                    |  2 ++
 scripts/scrape_osgl_inspirations.py | 32 ++++++++++++++++++++++++++++-
 5 files changed, 45 insertions(+), 11 deletions(-)

diff --git a/games/d.yaml b/games/d.yaml
index 38f9c284e..e09b20030 100644
--- a/games/d.yaml
+++ b/games/d.yaml
@@ -293,7 +293,7 @@
   frameworks:
   - SDL
   originals:
-  - 3D Deathchase
+  - Deathchase
   status: playable
   repo: https://github.com/BachToTheFuture/DeathChase3D
   development: halted
diff --git a/originals/0scumm.yaml b/originals/0scumm.yaml
index 83f7ac959..d4b9e8b42 100644
--- a/originals/0scumm.yaml
+++ b/originals/0scumm.yaml
@@ -65,15 +65,6 @@
   platforms:
   - DOS
 
-- name: 3D Deathchase
-  external:
-    wikipedia: 3D Deathchase
-  meta:
-    genres:
-    - Racing
-    subgenres:
-    - Vehicular Combat
-
 - name: The 7th Guest
   external:
     website: https://www.mobygames.com/game/283
diff --git a/originals/d.yaml b/originals/d.yaml
index 422e45368..2c9dbec80 100644
--- a/originals/d.yaml
+++ b/originals/d.yaml
@@ -137,6 +137,17 @@
   platforms:
   - Windows
 
+- name: Deathchase
+  names:
+  - 3D Deathchase
+  external:
+    wikipedia: Deathchase
+  meta:
+    genres:
+    - Racing
+    subgenres:
+    - Vehicular Combat
+
 - name: Death Rally
   external:
     wikipedia: Death Rally
diff --git a/originals/s.yaml b/originals/s.yaml
index cfde56aea..1cda809a2 100644
--- a/originals/s.yaml
+++ b/originals/s.yaml
@@ -444,6 +444,8 @@
     - Turn-Based Strategy
 
 - name: Sid Meier's Colonization
+  names:
+  - Colonization
   external:
     wikipedia: Sid Meier's Colonization
   meta:
diff --git a/scripts/scrape_osgl_inspirations.py b/scripts/scrape_osgl_inspirations.py
index 83fb8fb61..a45c2edec 100644
--- a/scripts/scrape_osgl_inspirations.py
+++ b/scripts/scrape_osgl_inspirations.py
@@ -7,6 +7,33 @@
 
 INSPIRATION_PATTERN = re.compile(r"(.+) \[\d+\]")
 INSPIRED_PATTERN = re.compile(r"Inspired entries: (.+)")
+# OSGL name to OSGC alias
+ALIASES = {
+    "Alone in the Dark series": "Alone in the Dark",
+    "Anno (series)": "Anno series",
+    "BioWare's Aurora engine": "Neverwinter Nights",
+    "Blake Stone: Aliens of Gold": "Blake Stone: Planet Strike",
+    "Blasteroids": "Asteroids",
+    "Caesar 3": "Caesar III",
+    "Civilization series": "Civilization",
+    "Company of Heroes: Opposing Fronts": "Company of Heroes",
+    "Company of Heroes: Tales of Valor": "Company of Heroes",
+}
+# Games that aren't interesting enough or weren't closed source
+BLACKLIST = {
+    "arithmetic",
+    "Black Shades",
+    "Blob Wars Attrition",
+    "Blobby Volley",
+    "Chromium B.S.U.",
+    "CorsixTH",
+    "Crossfire",
+    "Cube",
+    "Cube 2: Sauerbraten",
+    "CUBE engine",
+    "Daimonin",
+    "TuxMath",
+}
 
 
 def main():
@@ -33,7 +60,10 @@ def main():
         for name in original.get("names", []):
             osgc_originals.add(name)
     for game in osgl_games:
-        if game not in osgc_originals:
+        if game in BLACKLIST:
+            continue
+        alias = ALIASES.get(game)
+        if game not in osgc_originals and (not alias or alias not in osgc_originals):
             print(f"Missing original: {game}")
     osgc_games = set(game["name"] for game in games())
     for game, inspireds in osgl_games.items():