Make use of cache while transform builtin containers

astroid/brain/brain_builtin_inference.py

@@ -154,7 +154,7 @@ def _container_generic_inference(node, context, node_type, transform):
     if len(node.args) > 1:
         raise UseInferenceDefault()
 
-    arg, = args
+    (arg,) = args
     transformed = transform(arg)
     if not transformed:
         try:
@@ -169,15 +169,17 @@ def _container_generic_inference(node, context, node_type, transform):
     return transformed
 
 
-def _container_generic_transform(arg, klass, iterables, build_elts):
+def _container_generic_transform(arg, context, klass, iterables, build_elts):
     if isinstance(arg, klass):
         return arg
     elif isinstance(arg, iterables):
         if all(isinstance(elt, nodes.Const) for elt in arg.elts):
             elts = [elt.value for elt in arg.elts]
         else:
             # TODO: Does not handle deduplication for sets.
-            elts = filter(None, map(helpers.safe_infer, arg.elts))
+            elts = filter(
+                None, map(partial(helpers.safe_infer, context=context), arg.elts)
+            )
     elif isinstance(arg, nodes.Dict):
         # Dicts need to have consts as strings already.
         if not all(isinstance(elt[0], nodes.Const) for elt in arg.items):
@@ -197,6 +199,7 @@ def _infer_builtin_container(
 ):
     transform_func = partial(
         _container_generic_transform,
+        context=context,
         klass=klass,
         iterables=iterables,
         build_elts=build_elts,

tests/unittest_inference.py

@@ -5502,5 +5502,19 @@ class A:
     assert inferred.value == 42
 
 
+def test_recursion_error_inferring_builtin_containers():
+    node = extract_node(
+        """
+    class Foo:
+        a = "foo"
+    inst = Foo()
+
+    b = tuple([inst.a]) #@
+    inst.a = b
+    """
+    )
+    helpers.safe_infer(node.targets[0])
+
+
 if __name__ == "__main__":
     unittest.main()