Merge branch 'main' into PYTHON-6016

Author: NoahStapp

README.md

@@ -98,7 +98,7 @@ package that is incompatible with PyMongo.
 
 ## Dependencies
 
-PyMongo supports CPython 3.9+ and PyPy3.9+.
+PyMongo supports CPython 3.9+ and PyPy3.9+. PyPy support is deprecated and will be removed in a future release.
 
 Required dependencies:
 

bson/__init__.py

@@ -597,18 +597,32 @@ def _element_to_dict(
 _T = TypeVar("_T", bound=MutableMapping[str, Any])
 
 
-def _raw_to_dict(
-    data: Any,
-    position: int,
-    obj_end: int,
-    opts: CodecOptions[RawBSONDocument],
-    result: _T,
-    raw_array: bool = False,
-) -> _T:
-    data, view = get_data_and_view(data)
-    return cast(
-        _T, _elements_to_dict(data, view, position, obj_end, opts, result, raw_array=raw_array)
-    )
+if _USE_C:
+
+    def _raw_to_dict(
+        data: Any,
+        position: int,
+        obj_end: int,
+        opts: CodecOptions[RawBSONDocument],
+        result: _T,
+        raw_array: bool = False,
+    ) -> _T:
+        return cast(_T, _cbson._raw_to_dict(data, position, obj_end, opts, result, raw_array))
+
+else:
+
+    def _raw_to_dict(
+        data: Any,
+        position: int,
+        obj_end: int,
+        opts: CodecOptions[RawBSONDocument],
+        result: _T,
+        raw_array: bool = False,
+    ) -> _T:
+        data, view = get_data_and_view(data)
+        return cast(
+            _T, _elements_to_dict(data, view, position, obj_end, opts, result, raw_array=raw_array)
+        )
 
 
 def _elements_to_dict(

bson/_cbsonmodule.c

@@ -2867,9 +2867,6 @@ static int _element_to_dict(PyObject* self, const char* string,
 }
 
 static PyObject* _cbson_element_to_dict(PyObject* self, PyObject* args) {
-    /* TODO(PYTHON-6038): buffer-protocol inputs are copied
-     * upstream in get_data_and_view. Native RawBSONDocument inflation in C
-     * should accept the buffer directly. */
     char* string;
     PyObject* bson;
     PyObject* options_obj = NULL;
@@ -3094,6 +3091,98 @@ static PyObject* _prepare_input_buffer(PyObject* self, PyObject* bson,
     return PyBytes_FromObject(bson);
 }
 
+/* Decode all elements of a raw BSON document into `result`. */
+static PyObject* _cbson_raw_to_dict(PyObject* self, PyObject* args) {
+    PyObject* data;
+    unsigned position;
+    unsigned obj_end;
+    PyObject* options_obj;
+    PyObject* result;
+    int raw_array = 0;
+    codec_options_t options;
+    PyObject* bson = NULL;
+    Py_buffer view = {0};
+    PyObject* ret = NULL;
+    const char* string;
+    unsigned end;
+
+    if (!(PyArg_ParseTuple(args, "OIIOOp", &data, &position, &obj_end,
+                           &options_obj, &result, &raw_array) &&
+            convert_codec_options(self, options_obj, &options))) {
+        return NULL;
+    }
+
+    bson = _prepare_input_buffer(self, data, &options);
+    if (!bson) {
+        destroy_codec_options(&options);
+        return NULL;
+    }
+
+    if (!_get_buffer(bson, &view)) {
+        Py_DECREF(bson);
+        destroy_codec_options(&options);
+        return NULL;
+    }
+
+    string = (char*)view.buf;
+    /* obj_end must be the index of the document's eoo byte. The eoo check
+     * also protects against field scans running past the end of the buffer. */
+    if (obj_end < 1 || (Py_ssize_t)obj_end >= view.len || position > obj_end ||
+            string[obj_end]) {
+        PyObject* InvalidBSON = _error("InvalidBSON");
+        if (InvalidBSON) {
+            PyErr_SetString(InvalidBSON, "bad object or element length");
+            Py_DECREF(InvalidBSON);
+        }
+        goto done;
+    }
+
+    options.buffer_owner = bson;
+
+    end = obj_end - 1;
+    while (position < end) {
+        PyObject* name = NULL;
+        PyObject* value = NULL;
+        int new_position = _element_to_dict(
+            self, string, position, obj_end, &options, raw_array, &name, &value);
+        if (new_position < 0) {
+            goto done;
+        }
+        position = (unsigned)new_position;
+
+        if (PyDict_CheckExact(result)) {
+            if (PyDict_SetItem(result, name, value) < 0) {
+                Py_DECREF(name);
+                Py_DECREF(value);
+                goto done;
+            }
+        } else {
+            if (PyObject_SetItem(result, name, value) < 0) {
+                Py_DECREF(name);
+                Py_DECREF(value);
+                goto done;
+            }
+        }
+        Py_DECREF(name);
+        Py_DECREF(value);
+    }
+    if (position != obj_end) {
+        PyObject* InvalidBSON = _error("InvalidBSON");
+        if (InvalidBSON) {
+            PyErr_SetString(InvalidBSON, "bad object or element length");
+            Py_DECREF(InvalidBSON);
+        }
+        goto done;
+    }
+    Py_INCREF(result);
+    ret = result;
+done:
+    PyBuffer_Release(&view);
+    Py_DECREF(bson);
+    destroy_codec_options(&options);
+    return ret;
+}
+
 static PyObject* _cbson_bson_to_dict(PyObject* self, PyObject* args) {
     int32_t size;
     Py_ssize_t total_size;
@@ -3423,6 +3512,8 @@ static PyMethodDef _CBSONMethods[] = {
      "convert binary data to a sequence of documents."},
     {"_element_to_dict", _cbson_element_to_dict, METH_VARARGS,
      "Decode a single key, value pair."},
+    {"_raw_to_dict", _cbson_raw_to_dict, METH_VARARGS,
+     "Decode all elements of a raw BSON document into a result mapping."},
     {"_array_of_documents_to_buffer", _cbson_array_of_documents_to_buffer, METH_VARARGS, "Convert raw array of documents to a stream of BSON documents"},
     {"_test_long_long_to_str", _test_long_long_to_str, METH_VARARGS, "Test conversion of extreme and common Py_ssize_t values to str."},
     {NULL, NULL, 0, NULL}

doc/changelog.rst

@@ -8,6 +8,7 @@ PyMongo 4.18 brings a number of changes including:
 
 - Dropped support for MongoDB 4.2.
 - Added support for MongoDB 9.0.
+- PyPy support is deprecated and will be removed in a future release.
 - Improved TLS connection performance by reusing TLS sessions across connections
   to the same server, avoiding a full handshake on each new connection.
   Session resumption is supported on all Python versions for synchronous clients
@@ -37,6 +38,8 @@ PyMongo 4.18 brings a number of changes including:
   :class:`bytearray` are always :class:`bytes` copies.
 - :func:`bson.get_data_and_view` now returns a view of a private :class:`bytes` copy
   for buffer-protocol inputs other than :class:`bytes` or :class:`bytearray`.
+- Improved the performance of lazily decoding a
+  :class:`~bson.raw_bson.RawBSONDocument` when the C extension is available.
 - Fixed a potential out-of-bounds read in the C extension when decoding an
   array of BSON documents. An embedded document whose declared length exceeds
   the bytes remaining in the array now raises

pymongo/asynchronous/bulk.py

@@ -33,7 +33,7 @@
 from bson.raw_bson import RawBSONDocument
 from pymongo import _csot, common
 from pymongo._telemetry import _generate_op_id_or_none
-from pymongo.asynchronous.client_session import AsyncClientSession, _validate_session_write_concern
+from pymongo.asynchronous.client_session import AsyncClientSession
 from pymongo.asynchronous.command_runner import (
     run_bulk_write_command,
 )
@@ -45,6 +45,7 @@
     _raise_bulk_write_error,
     _Run,
 )
+from pymongo.client_session_shared import _validate_session_write_concern
 from pymongo.common import (
     validate_is_document_type,
     validate_ok_for_replace,

pymongo/asynchronous/client_bulk.py

@@ -33,10 +33,7 @@
 from bson.raw_bson import RawBSONDocument
 from pymongo import _csot, common
 from pymongo._telemetry import _generate_op_id_or_none
-from pymongo.asynchronous.client_session import (
-    AsyncClientSession,
-    _validate_session_write_concern,
-)
+from pymongo.asynchronous.client_session import AsyncClientSession
 from pymongo.asynchronous.collection import AsyncCollection
 from pymongo.asynchronous.command_cursor import AsyncCommandCursor
 from pymongo.asynchronous.command_runner import (
@@ -52,6 +49,7 @@
     _merge_command,
     _throw_client_bulk_write_exception,
 )
+from pymongo.client_session_shared import _validate_session_write_concern
 from pymongo.common import (
     validate_is_document_type,
     validate_ok_for_replace,