From 78162e7022023132aac3ca0ffdef8191c5c12a38 Mon Sep 17 00:00:00 2001
From: Abdessamad Derraz <3028866+Abdess@users.noreply.github.com>
Date: Sat, 17 Oct 2026 01:54:35 +0200
Subject: [PATCH] fix: write the database cache atomically

---
 scripts/generate_db.py          | 12 ++++++++----
 tests/test_audit_regressions.py | 20 ++++++++++++++++++++
 2 files changed, 28 insertions(+), 4 deletions(-)

diff --git a/scripts/generate_db.py b/scripts/generate_db.py
index 2bb1a663..9d4d10d2 100644
--- a/scripts/generate_db.py
+++ b/scripts/generate_db.py
@@ -29,6 +29,7 @@ from common import (
     require_yaml,
     source_yaml_paths,
     write_if_changed,
+    write_text_atomic,
     yaml_load,
 )
 
@@ -258,11 +259,14 @@ def load_cache(cache_path: str) -> dict:
         return {}
 
 
-def save_cache(cache_path: str, cache: dict):
-    """Save cache to disk."""
+def save_cache(cache_path: str, cache: dict) -> None:
+    """Save the cache whole or not at all.
+
+    Written in place, a crash while serializing left a truncated file that
+    the next run read as empty and rehashed the collection from scratch.
+    """
     os.makedirs(os.path.dirname(cache_path), exist_ok=True)
-    with open(cache_path, "w") as f:
-        json.dump(cache, f)
+    write_text_atomic(cache_path, json.dumps(cache))
 
 
 def _load_gitignored_large_files() -> set[str]:
diff --git a/tests/test_audit_regressions.py b/tests/test_audit_regressions.py
index 4e092800..1e97802c 100644
--- a/tests/test_audit_regressions.py
+++ b/tests/test_audit_regressions.py
@@ -2161,6 +2161,26 @@ class ContributorsSurviveAFailedRequest(unittest.TestCase):
         )
 
 
+class TheDatabaseCacheIsWrittenWhole(unittest.TestCase):
+    """A crash while serializing left a truncated db_cache.json behind, and
+    the next run rehashed the whole collection from scratch."""
+
+    def test_a_failed_save_keeps_the_previous_cache(self):
+        import json
+        from unittest import mock
+
+        from scripts import generate_db
+
+        with tempfile.TemporaryDirectory(dir=TMP_ROOT) as directory:
+            path = os.path.join(directory, "cache", "db_cache.json")
+            generate_db.save_cache(path, {"a": 1})
+            with mock.patch.object(json, "dumps", side_effect=TypeError("miscomputed")):
+                with self.assertRaises(TypeError):
+                    generate_db.save_cache(path, {"a": 2})
+            self.assertEqual(generate_db.load_cache(path), {"a": 1})
+            self.assertEqual(os.listdir(os.path.dirname(path)), ["db_cache.json"])
+
+
 class KidVidTapesStayWithTheContent(unittest.TestCase):
     """Stella opens its KidVid tapes by bare name (stella2014, from the
     working directory) or from its base directory (stella2023, the content