fix: clone MAME and FBNeo sources per process

scripts/scraper/fbneo_hash_scraper.py

@@ -29,7 +29,8 @@ log = logging.getLogger(__name__)
 
 REPO_URL = "https://github.com/finalburnneo/FBNeo.git"
 REPO_ROOT = Path(__file__).resolve().parent.parent.parent
-CLONE_DIR = REPO_ROOT / "tmp" / "fbneo"
+# One clone per process, for the reason mame_hash_scraper gives.
+CLONE_DIR = REPO_ROOT / "tmp" / f"fbneo-{os.getpid()}"
 CACHE_PATH = REPO_ROOT / "data" / "fbneo-hashes.json"
 EMULATORS_DIR = REPO_ROOT / "emulators"
 STALE_HOURS = 24

scripts/scraper/mame_hash_scraper.py

@@ -31,7 +31,9 @@ log = logging.getLogger(__name__)
 
 _ROOT = Path(__file__).resolve().parents[2]
 _CACHE_PATH = _ROOT / "data" / "mame-hashes.json"
-_CLONE_DIR = _ROOT / "tmp" / "mame"
+# One clone per process: a fixed tmp/mame was removed and recloned at another
+# release by a second run while the first was still parsing it.
+_CLONE_DIR = _ROOT / "tmp" / f"mame-{os.getpid()}"
 _EMULATORS_DIR = _ROOT / "emulators"
 _REPO_URL = "https://github.com/mamedev/mame.git"
 _STALE_HOURS = 24

tests/test_hash_merge.py

@@ -759,5 +759,20 @@ class WritersPutTheValuesOnDisk(unittest.TestCase):
         self.assertEqual(member["crc32"], "9036d879")
 
 
+class OneCloneDirectoryPerRun(unittest.TestCase):
+    """Two runs shared tmp/mame: the second removed and recloned it at
+    another release while the first parsed it, and the first wrote 0.289
+    sets into a 0.278 profile."""
+
+    def test_the_clone_directory_names_the_process(self):
+        import os
+
+        from scripts.scraper import fbneo_hash_scraper, mame_hash_scraper
+
+        for clone in (mame_hash_scraper._CLONE_DIR, fbneo_hash_scraper.CLONE_DIR):
+            with self.subTest(clone=clone.name):
+                self.assertTrue(clone.name.endswith(f"-{os.getpid()}"))
+
+
 if __name__ == "__main__":
     unittest.main()