- PyList, PyTuple and PySequence's get_item now accepts only usize indices instead of isize.

- `PyList` and `PyTuple`'s `get_item` now always uses the safe API. See `get_item_unchecked` for retrieving index without checks.

Author: Aviram Hassan

CHANGELOG.md

@@ -39,6 +39,13 @@ and this project adheres to [Semantic Versioning](http://semver.org/spec/v2.0.0.
 - Fix compiler warning: deny trailing semicolons in expression macro. [#1762](https://github.com/PyO3/pyo3/pull/1762)
 - Fix incorrect FFI definition of `Py_DecodeLocale`. The 2nd argument is now `*mut Py_ssize_t` instead of `Py_ssize_t`. [#1766](https://github.com/PyO3/pyo3/pull/1766)
 
+### Changed
+- `PyList`, `PyTuple` and `PySequence`'s `get_item` now accepts only `usize` indices instead of `isize`.
+### Added
+- Add `PyList::get_item_unchecked()` and `PyTuple::get_item_unchecked()` to get items without bounds checks. [#1733](https://github.com/PyO3/pyo3/pull/1733)
+
+### Changed 
+- `PyList` and `PyTuple`'s `get_item` now return `PyResult<&PyAny> instead of panicking. [#1733](https://github.com/PyO3/pyo3/pull/1733)
 ## [0.14.1] - 2021-07-04
 
 ### Added

benches/bench_list.rs

@@ -32,7 +32,22 @@ fn list_get_item(b: &mut Bencher) {
     let mut sum = 0;
     b.iter(|| {
         for i in 0..LEN {
-            sum += list.get_item(i as isize).extract::<usize>().unwrap();
+            sum += list.get_item(i).unwrap().extract::<usize>().unwrap();
+        }
+    });
+}
+
+fn list_get_item_unchecked(b: &mut Bencher) {
+    let gil = Python::acquire_gil();
+    let py = gil.python();
+    const LEN: usize = 50_000;
+    let list = PyList::new(py, 0..LEN);
+    let mut sum = 0;
+    b.iter(|| {
+        for i in 0..LEN {
+            unsafe {
+                sum += list.get_item_unchecked(i).extract::<usize>().unwrap();
+            }
         }
     });
 }
@@ -41,6 +56,7 @@ fn criterion_benchmark(c: &mut Criterion) {
     c.bench_function("iter_list", iter_list);
     c.bench_function("list_new", list_new);
     c.bench_function("list_get_item", list_get_item);
+    c.bench_function("list_get_item_unchecked", list_get_item_unchecked);
 }
 
 criterion_group!(benches, criterion_benchmark);

benches/bench_tuple.rs

@@ -32,7 +32,22 @@ fn tuple_get_item(b: &mut Bencher) {
     let mut sum = 0;
     b.iter(|| {
         for i in 0..LEN {
-            sum += tuple.get_item(i).extract::<usize>().unwrap();
+            sum += tuple.get_item(i).unwrap().extract::<usize>().unwrap();
+        }
+    });
+}
+
+fn tuple_get_item_unchecked(b: &mut Bencher) {
+    let gil = Python::acquire_gil();
+    let py = gil.python();
+    const LEN: usize = 50_000;
+    let tuple = PyTuple::new(py, 0..LEN);
+    let mut sum = 0;
+    b.iter(|| {
+        for i in 0..LEN {
+            unsafe {
+                sum += tuple.get_item_unchecked(i).extract::<usize>().unwrap();
+            }
         }
     });
 }
@@ -41,6 +56,7 @@ fn criterion_benchmark(c: &mut Criterion) {
     c.bench_function("iter_tuple", iter_tuple);
     c.bench_function("tuple_new", tuple_new);
     c.bench_function("tuple_get_item", tuple_get_item);
+    c.bench_function("tuple_get_item_unchecked", tuple_get_item_unchecked);
 }
 
 criterion_group!(benches, criterion_benchmark);

pyo3-macros-backend/src/from_pyobject.rs

@@ -243,7 +243,7 @@ impl<'a> Container<'a> {
         for i in 0..len {
             let error_msg = format!("failed to extract field {}.{}", quote!(#self_ty), i);
             fields.push(quote!(
-                s.get_item(#i).extract().map_err(|inner| {
+                s.get_item(#i).and_then(PyAny::extract).map_err(|inner| {
                 let py = pyo3::PyNativeType::py(obj);
                 let new_err = pyo3::exceptions::PyTypeError::new_err(#error_msg);
                 new_err.set_cause(py, Some(inner));

src/conversions/array.rs

@@ -60,7 +60,7 @@ mod min_const_generics {
         if seq_len != N {
             return Err(invalid_sequence_length(N, seq_len));
         }
-        array_try_from_fn(|idx| seq.get_item(idx as isize).and_then(PyAny::extract))
+        array_try_from_fn(|idx| seq.get_item(idx).and_then(PyAny::extract))
     }
 
     // TODO use std::array::try_from_fn, if that stabilises:

src/types/dict.rs

@@ -628,23 +628,23 @@ mod tests {
     #[test]
     fn test_items() {
         Python::with_gil(|py| {
-            let mut v = HashMap::new();
-            v.insert(7, 32);
-            v.insert(8, 42);
-            v.insert(9, 123);
-            let ob = v.to_object(py);
-            let dict = <PyDict as PyTryFrom>::try_from(ob.as_ref(py)).unwrap();
-            // Can't just compare against a vector of tuples since we don't have a guaranteed ordering.
-            let mut key_sum = 0;
-            let mut value_sum = 0;
-            for el in dict.items().iter() {
-                let tuple = el.cast_as::<PyTuple>().unwrap();
-                key_sum += tuple.get_item(0).extract::<i32>().unwrap();
-                value_sum += tuple.get_item(1).extract::<i32>().unwrap();
-            }
-            assert_eq!(7 + 8 + 9, key_sum);
-            assert_eq!(32 + 42 + 123, value_sum);
-        });
+        let mut v = HashMap::new();
+        v.insert(7, 32);
+        v.insert(8, 42);
+        v.insert(9, 123);
+        let ob = v.to_object(py);
+        let dict = <PyDict as PyTryFrom>::try_from(ob.as_ref(py)).unwrap();
+        // Can't just compare against a vector of tuples since we don't have a guaranteed ordering.
+        let mut key_sum = 0;
+        let mut value_sum = 0;
+        for el in dict.items().iter() {
+            let tuple = el.cast_as::<PyTuple>().unwrap();
+            key_sum += tuple.get_item(0).unwrap().extract::<i32>().unwrap();
+            value_sum += tuple.get_item(1).unwrap().extract::<i32>().unwrap();
+        }
+        assert_eq!(7 + 8 + 9, key_sum);
+        assert_eq!(32 + 42 + 123, value_sum);
+    });
     }
 
     #[test]