Merge pull request #545 from PyO3/replace-PyByteArray-data

Replace unsound PyByteArray::data with PyByteArray::to_vec

CHANGELOG.md

@@ -21,6 +21,7 @@ and this project adheres to [Semantic Versioning](http://semver.org/spec/v2.0.0.
 ### Changed
 
  * Implementing the Using the `gc` parameter for `pyclass` (e.g. `#[pyclass(gc)]`) without implementing the `class::PyGCProtocol` trait is now a compile-time error. Failing to implement this trait could lead to segfaults. [#532](https://github.com/PyO3/pyo3/pull/532)
+ * `PyByteArray::data` has been replaced with `PyDataArray::to_vec` because returning a `&[u8]` is unsound. (See [this comment](https://github.com/PyO3/pyo3/issues/373#issuecomment-512332696) for a great write-up for why that was unsound)
 
 ## [0.7.0] - 2018-05-26
 

src/types/bytearray.rs

@@ -18,8 +18,6 @@ pyobject_native_type!(PyByteArray, ffi::PyByteArray_Type, ffi::PyByteArray_Check
 impl PyByteArray {
     /// Creates a new Python bytearray object.
     /// The byte string is initialized by copying the data from the `&[u8]`.
-    ///
-    /// Panics if out of memory.
     pub fn new<'p>(py: Python<'p>, src: &[u8]) -> &'p PyByteArray {
         let ptr = src.as_ptr() as *const c_char;
         let len = src.len() as ffi::Py_ssize_t;
@@ -47,16 +45,37 @@ impl PyByteArray {
         self.len() == 0
     }
 
-    /// Gets the Python bytearray data as byte slice.
-    pub fn data(&self) -> &[u8] {
-        unsafe {
+    /// Copies the contents of the bytearray to a rust vector
+    ///
+    /// # Example
+    ///
+    /// ```
+    /// # use pyo3::prelude::*;
+    /// # use pyo3::types::PyByteArray;
+    /// # use pyo3::types::IntoPyDict;
+    /// # let gil = GILGuard::acquire();
+    /// # let py = gil.python();
+    /// #
+    /// let bytearray = PyByteArray::new(py, b"Hello World.");
+    /// let mut copied_message = bytearray.to_vec();
+    /// assert_eq!(b"Hello World.", copied_message.as_slice());
+    ///
+    /// copied_message[11] = b'!';
+    /// assert_eq!(b"Hello World!", copied_message.as_slice());
+    ///
+    /// let locals = [("bytearray", bytearray)].into_py_dict(py);
+    /// py.run("assert bytearray == b'Hello World.'", None, Some(locals)).unwrap();
+    /// ```
+    pub fn to_vec(&self) -> Vec<u8> {
+        let slice = unsafe {
             let buffer = ffi::PyByteArray_AsString(self.0.as_ptr()) as *mut u8;
             let length = ffi::PyByteArray_Size(self.0.as_ptr()) as usize;
             slice::from_raw_parts_mut(buffer, length)
-        }
+        };
+        slice.to_vec()
     }
 
-    /// Resize bytearray object.
+    /// Resize bytearray object to `len`.
     pub fn resize(&self, len: usize) -> PyResult<()> {
         unsafe {
             let result = ffi::PyByteArray_Resize(self.0.as_ptr(), len as ffi::Py_ssize_t);
@@ -84,13 +103,13 @@ mod test {
         let src = b"Hello Python";
         let bytearray = PyByteArray::new(py, src);
         assert_eq!(src.len(), bytearray.len());
-        assert_eq!(src, bytearray.data());
+        assert_eq!(src, bytearray.to_vec().as_slice());
 
         let ba: PyObject = bytearray.into();
         let bytearray = PyByteArray::from(py, &ba).unwrap();
 
         assert_eq!(src.len(), bytearray.len());
-        assert_eq!(src, bytearray.data());
+        assert_eq!(src, bytearray.to_vec().as_slice());
 
         bytearray.resize(20).unwrap();
         assert_eq!(20, bytearray.len());