fix(ingest/file-backed-collections): Properly set _use_sqlite_on_conflict

Author: asikowitz

metadata-ingestion/src/datahub/utilities/file_backed_collections.py

@@ -243,7 +243,7 @@ def __post_init__(self) -> None:
             # This was added in 3.24.0 from 2018-06-04.
             # See https://www.sqlite.org/lang_conflict.html
             if OVERRIDE_SQLITE_VERSION_REQUIREMENT:
-                self.use_sqlite_on_conflict = False
+                self._use_sqlite_on_conflict = False
             else:
                 raise RuntimeError("SQLite version 3.24.0 or later is required")
 

metadata-ingestion/tests/unit/utilities/test_file_backed_collections.py

@@ -1,8 +1,11 @@
 import dataclasses
 import json
+import os
 import pathlib
 import random
 import sqlite3
+from unittest.mock import patch
+
 from dataclasses import dataclass
 from typing import Counter, Dict
 
@@ -15,6 +18,36 @@
 )
 
 
+def test_set_use_sqlite_on_conflict():
+    with patch("sqlite3.sqlite_version_info", (3, 24, 0)):
+        cache = FileBackedDict[int](
+            tablename="cache",
+            cache_max_size=10,
+            cache_eviction_batch_size=10,
+        )
+        assert cache._use_sqlite_on_conflict == True
+
+    with pytest.raises(RuntimeError):
+        with patch("sqlite3.sqlite_version_info", (3, 23, 1)):
+            cache = FileBackedDict[int](
+                tablename="cache",
+                cache_max_size=10,
+                cache_eviction_batch_size=10,
+            )
+            assert cache._use_sqlite_on_conflict == False
+
+    with patch("sqlite3.sqlite_version_info", (3, 23, 1)), patch(
+        "datahub.utilities.file_backed_collections.OVERRIDE_SQLITE_VERSION_REQUIREMENT",
+        True,
+    ):
+        cache = FileBackedDict[int](
+            tablename="cache",
+            cache_max_size=10,
+            cache_eviction_batch_size=10,
+        )
+        assert cache._use_sqlite_on_conflict == False
+
+
 @pytest.mark.parametrize("use_sqlite_on_conflict", [True, False])
 def test_file_dict(use_sqlite_on_conflict: bool) -> None:
     cache = FileBackedDict[int](