Fix panic on non UTF-8 string (#59)

Author: MarshalX

pytests/test_dag_cbor.py

@@ -137,6 +137,15 @@ def test_recursion_limit_exceed_on_nested_maps() -> None:
     assert 'in DAG-CBOR decoding' in str(exc_info.value)
 
 
+
+def test_dag_cbor_decode_invalid_utf8() -> None:
+    with pytest.raises(ValueError) as exc_info:
+        libipld.decode_dag_cbor(bytes.fromhex('62c328'))
+
+
+    assert 'Invalid UTF-8 string' in str(exc_info.value)
+
+
 def test_dab_cbor_decode_map_int_key() -> None:
     dag_cbor = bytes.fromhex('a10000')
     with pytest.raises(ValueError) as exc_info:
@@ -151,3 +160,4 @@ def test_dab_cbor_encode_map_int_key() -> None:
         libipld.encode_dag_cbor(obj)
 
     assert 'Map keys must be strings' in str(exc_info.value)
+

src/lib.rs

@@ -102,11 +102,13 @@ fn get_bytes_from_py_any<'py>(obj: &'py Bound<'py, PyAny>) -> PyResult<&'py [u8]
     }
 }
 
-fn string_new_bound<'py>(py: Python<'py>, s: &[u8]) -> Bound<'py, PyString> {
+fn string_new_bound<'py>(py: Python<'py>, s: &[u8]) -> Result<Bound<'py, PyString>> {
+    std::str::from_utf8(s).map_err(|e| anyhow!("Invalid UTF-8 string: {}", e))?;
+
     let ptr = s.as_ptr() as *const c_char;
     let len = s.len() as ffi::Py_ssize_t;
     unsafe {
-        Bound::from_owned_ptr(py, ffi::PyUnicode_FromStringAndSize(ptr, len)).downcast_into_unchecked()
+        Ok(Bound::from_owned_ptr(py, ffi::PyUnicode_FromStringAndSize(ptr, len)).downcast_into_unchecked())
     }
 }
 
@@ -135,7 +137,7 @@ fn decode_dag_cbor_to_pyobject<R: Read + Seek>(
         }
         MajorKind::TextString => {
             let len = decode::read_uint(r, major)?;
-            string_new_bound(py, &decode::read_bytes(r, len)?).into_pyobject(py)?.into()
+            string_new_bound(py, &decode::read_bytes(r, len)?)?.into_pyobject(py)?.into()
         }
         MajorKind::Array => {
             let len: ffi::Py_ssize_t = decode_len(decode::read_uint(r, major)?)?.try_into()?;
@@ -173,7 +175,7 @@ fn decode_dag_cbor_to_pyobject<R: Read + Seek>(
                     }
                 }
 
-                let key_py = string_new_bound(py, key.as_slice()).into_pyobject(py)?;
+                let key_py = string_new_bound(py, key.as_slice())?.into_pyobject(py)?;
                 prev_key = Some(key);
 
                 let value_py = decode_dag_cbor_to_pyobject(py, r, depth + 1)?;