JSONDecodeError on buffer allocation failure

ijl · ijl · commit 52300ddfe830 · 2025-07-15T14:39:19.000Z
diff --git a/README.md b/README.md
@@ -616,6 +616,9 @@ which the standard library allows, but is not valid JSON.
 It raises `JSONDecodeError` if a combination of array or object recurses
 1024 levels deep.
 
+It raises `JSONDecodeError` if unable to allocate a buffer large enough
+to parse the document.
+
 `JSONDecodeError` is a subclass of `json.JSONDecodeError` and `ValueError`.
 This is for compatibility with the standard library.
 
diff --git a/src/deserialize/backend/yyjson.rs b/src/deserialize/backend/yyjson.rs
@@ -50,7 +50,7 @@ const MINIMUM_BUFFER_CAPACITY: usize = 4096;
 
 fn buffer_capacity_to_allocate(len: usize) -> usize {
     // The max memory size is (json_size / 2 * 16 * 1.5 + padding).
-    (((len / 2) * 32) + 256 + (MINIMUM_BUFFER_CAPACITY - 1)) & !(MINIMUM_BUFFER_CAPACITY - 1)
+    (((len / 2) * 24) + 256 + (MINIMUM_BUFFER_CAPACITY - 1)) & !(MINIMUM_BUFFER_CAPACITY - 1)
 }
 
 fn unsafe_yyjson_is_ctn(val: *mut yyjson_val) -> bool {
@@ -73,6 +73,13 @@ pub(crate) fn deserialize(
     assume!(!data.is_empty());
     let buffer_capacity = buffer_capacity_to_allocate(data.len());
     let buffer_ptr = ffi!(PyMem_Malloc(buffer_capacity));
+    if unlikely!(buffer_ptr.is_null()) {
+        return Err(DeserializeError::from_yyjson(
+            Cow::Borrowed("Not enough memory to allocate buffer for parsing"),
+            0,
+            data,
+        ));
+    }
     let mut alloc = crate::ffi::yyjson::yyjson_alc {
         malloc: None,
         realloc: None,
diff --git a/test/test_buffer.py b/test/test_buffer.py
@@ -0,0 +1,29 @@
+# SPDX-License-Identifier: (Apache-2.0 OR MIT)
+
+import os
+
+import pytest
+
+import orjson
+
+ORJSON_RUNNER_MEMORY_GIB = os.getenv("ORJSON_RUNNER_MEMORY_GIB", "")
+
+
+@pytest.mark.skipif(
+    not ORJSON_RUNNER_MEMORY_GIB, reason="ORJSON_RUNNER_MEMORY_GIB not defined"
+)
+def test_memory_loads():
+    buffer_factor = 12
+    segment = "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa"
+    size = (
+        (int(ORJSON_RUNNER_MEMORY_GIB) * 1024 * 1024 * 1024)
+        // buffer_factor
+        // len(segment)
+    )
+    doc = "".join((segment for _ in range(0, size)))
+    with pytest.raises(orjson.JSONDecodeError) as exc_info:
+        _ = orjson.loads(doc)
+    assert (
+        str(exc_info.value)
+        == "Not enough memory to allocate buffer for parsing: line 1 column 1 (char 0)"
+    )
