fix: deep-copy cache entries so mutating search results cannot corrupt the cache

tests/test_tinydb.py

@@ -677,6 +677,27 @@ def test_query_cache():
     assert db.search(query) == [{'name': 'foo', 'value': 42}]
 
 
+def test_query_cache_not_corrupted_by_mutation():
+    """Mutating nested values of a search result must not pollute the cache."""
+    db = TinyDB(storage=MemoryStorage)
+    db.insert({'name': 'Alice', 'tags': ['a', 'b']})
+
+    query = where('name') == 'Alice'
+
+    results1 = db.search(query)
+    assert results1 == [{'name': 'Alice', 'tags': ['a', 'b']}]
+
+    # Mutate both a top-level value and a nested mutable inside the document
+    results1[0]['name'] = 'MUTATED'
+    results1[0]['tags'].append('c')
+
+    # A subsequent search hits the cache; it must return the original data
+    results2 = db.search(query)
+    assert results2 == [{'name': 'Alice', 'tags': ['a', 'b']}], (
+        'query cache was corrupted by mutation of a previously returned document'
+    )
+
+
 def test_tinydb_is_iterable(db: TinyDB):
     assert [r for r in db] == db.all()
 

tinydb/table.py

@@ -4,6 +4,7 @@
 """
 
 from collections.abc import Callable, Iterable, Iterator, Mapping
+from copy import deepcopy
 from typing import (
     NoReturn,
     Optional,
@@ -244,7 +245,7 @@ def search(self, cond: QueryLike) -> list[Document]:
         # query
         cached_results = self._query_cache.get(cond)
         if cached_results is not None:
-            return cached_results[:]
+            return deepcopy(cached_results)
 
         # Perform the search by applying the query to all documents.
         # Then, only if the document matches the query, convert it
@@ -271,8 +272,9 @@ def search(self, cond: QueryLike) -> list[Document]:
         is_cacheable: Callable[[], bool] = getattr(cond, 'is_cacheable',
                                                    lambda: True)
         if is_cacheable():
-            # Update the query cache
-            self._query_cache[cond] = docs[:]
+            # Update the query cache with a deep copy so that callers
+            # mutating returned documents cannot corrupt cached results.
+            self._query_cache[cond] = deepcopy(docs)
 
         return docs