feat: Add stream_file_content parameter to upload methods

Author: lukaszsocha2

boxsdk/object/folder.py

@@ -263,6 +263,7 @@ def upload_stream(
             additional_attributes: Optional[dict] = None,
             sha1: Optional[str] = None,
             etag: Optional[str] = None,
+            stream_file_content: bool = True,
     ) -> 'File':
         """
         Upload a file to the folder.
@@ -298,6 +299,9 @@ def upload_stream(
             A sha1 checksum for the file.
         :param etag:
             If specified, instruct the Box API to update the item only if the current version's etag matches.
+        :param stream_file_content:
+            If True, the upload will be performed as a stream request. If False, the file will be read into memory
+            before being uploaded, but this may be required if using some proxy servers to handle redirects correctly.
         :returns:
             The newly uploaded file.
         """
@@ -335,7 +339,7 @@ def upload_stream(
         if not headers:
             headers = None
         file_response = self._session.post(
-            url, data=data, files=files, expect_json_response=False, headers=headers
+            url, data=data, files=files, expect_json_response=False, headers=headers, stream_file_content=stream_file_content,
         ).json()
         if 'entries' in file_response:
             file_response = file_response['entries'][0]
@@ -358,6 +362,7 @@ def upload(
             additional_attributes: Optional[dict] = None,
             sha1: Optional[str] = None,
             etag: Optional[str] = None,
+            stream_file_content: bool = True,
     ) -> 'File':
         """
         Upload a file to the folder.
@@ -394,6 +399,9 @@ def upload(
             A sha1 checksum for the new content.
         :param etag:
             If specified, instruct the Box API to update the item only if the current version's etag matches.
+        :param stream_file_content:
+            If True, the upload will be performed as a stream request. If False, the file will be read into memory
+            before being uploaded, but this may be required if using some proxy servers to handle redirects correctly.
         :returns:
             The newly uploaded file.
         """
@@ -412,6 +420,7 @@ def upload(
                 additional_attributes=additional_attributes,
                 sha1=sha1,
                 etag=etag,
+                stream_file_content=stream_file_content,
             )
 
     @api_call

boxsdk/session/session.py

@@ -468,7 +468,8 @@ def _send_request(self, request: '_BoxRequest', **kwargs: Any) -> 'NetworkRespon
         """
         # Reset stream positions to what they were when the request was made so the same data is sent even if this
         # is a retried attempt.
-        files, file_stream_positions = kwargs.get('files'), kwargs.pop('file_stream_positions')
+        files, file_stream_positions, stream_file_content = (
+            kwargs.get('files'), kwargs.pop('file_stream_positions'), kwargs.pop('stream_file_content', True))
         request_kwargs = self._default_network_request_kwargs.copy()
         request_kwargs.update(kwargs)
         proxy_dict = self._prepare_proxy()
@@ -477,11 +478,12 @@ def _send_request(self, request: '_BoxRequest', **kwargs: Any) -> 'NetworkRespon
         if files and file_stream_positions:
             for name, position in file_stream_positions.items():
                 files[name][1].seek(position)
-            data = request_kwargs.pop('data', {})
-            multipart_stream = MultipartStream(data, files)
-            request_kwargs['data'] = multipart_stream
-            del request_kwargs['files']
-            request.headers['Content-Type'] = multipart_stream.content_type
+            if stream_file_content:
+                data = request_kwargs.pop('data', {})
+                multipart_stream = MultipartStream(data, files)
+                request_kwargs['data'] = multipart_stream
+                del request_kwargs['files']
+                request.headers['Content-Type'] = multipart_stream.content_type
         request.access_token = request_kwargs.pop('access_token', None)
 
         # send the request

test/integration_new/object/folder_itest.py

@@ -103,8 +103,8 @@ def test_auto_chunked_upload_NOT_using_upload_session_urls(parent_folder, large_
 
 
 def test_get_items(parent_folder, small_file_path):
-    with BoxTestFolder(parent_folder=parent_folder) as subfolder,\
-            BoxTestFile(parent_folder=parent_folder, file_path=small_file_path) as file,\
+    with BoxTestFolder(parent_folder=parent_folder) as subfolder, \
+            BoxTestFile(parent_folder=parent_folder, file_path=small_file_path) as file, \
             BoxTestWebLink(parent_folder=parent_folder, url='https://box.com') as web_link:
 
         assert set(parent_folder.get_items()) == {subfolder, file, web_link}
@@ -130,6 +130,17 @@ def test_upload_small_file_to_folder(parent_folder, small_file_name, small_file_
         util.permanently_delete(uploaded_file)
 
 
+def test_upload_small_file_to_folder_with_disabled_streaming_file_content(
+        parent_folder, small_file_name, small_file_path
+):
+    uploaded_file = parent_folder.upload(file_path=small_file_path, file_name=small_file_name, stream_file_content=False)
+    try:
+        assert uploaded_file.id
+        assert uploaded_file.parent == parent_folder
+    finally:
+        util.permanently_delete(uploaded_file)
+
+
 def test_create_subfolder(parent_folder):
     created_subfolder = parent_folder.create_subfolder(name=util.random_name())
     try:
@@ -199,7 +210,7 @@ def test_delete_folder(parent_folder):
 
 
 def test_cascade_and_get_metadata_cascade_policies(parent_folder):
-    with BoxTestMetadataTemplate(display_name="test_template") as metadata_template,\
+    with BoxTestMetadataTemplate(display_name="test_template") as metadata_template, \
             BoxTestFolder(parent_folder=parent_folder) as folder:
         folder.cascade_metadata(metadata_template)
 

test/unit/object/test_folder.py

@@ -2,7 +2,7 @@
 from datetime import datetime
 from io import BytesIO
 from os.path import basename
-from unittest.mock import mock_open, patch, Mock, MagicMock
+from unittest.mock import mock_open, patch, Mock, MagicMock, ANY
 import pytest
 import pytz
 
@@ -334,7 +334,14 @@ def test_upload(
     # in Python 2 tests
     attributes.update(additional_attributes)
     data = {'attributes': json.dumps(attributes)}
-    mock_box_session.post.assert_called_once_with(expected_url, expect_json_response=False, files=mock_files, data=data, headers=if_match_sha1_header)
+    mock_box_session.post.assert_called_once_with(
+        expected_url,
+        expect_json_response=False,
+        files=mock_files,
+        data=data,
+        headers=if_match_sha1_header,
+        stream_file_content=True
+    )
     assert isinstance(new_file, File)
     assert new_file.object_id == mock_object_id
     assert 'id' in new_file
@@ -438,6 +445,27 @@ def test_upload_does_preflight_check_if_specified(
             assert not test_folder.preflight_check.called
 
 
+@patch('boxsdk.object.folder.open', mock_open(read_data=b'some bytes'), create=True)
+@pytest.mark.parametrize('stream_file_content', (True, False))
+def test_upload_if_flag_stream_file_content_is_passed_to_session(
+        mock_box_session,
+        test_folder,
+        stream_file_content,
+):
+    expected_url = f'{API.UPLOAD_URL}/files/content'
+
+    test_folder.upload('foo.txt', file_name='foo.txt', stream_file_content=stream_file_content)
+
+    mock_files = {'file': ('unused', ANY)}
+    mock_box_session.post.assert_called_once_with(
+        expected_url,
+        data=ANY,
+        files=mock_files,
+        expect_json_response=False,
+        headers=None,
+        stream_file_content=stream_file_content)
+
+
 def test_create_subfolder(test_folder, mock_box_session, mock_object_id, mock_folder_response):
     expected_url = test_folder.get_type_url()
     mock_box_session.post.return_value = mock_folder_response

test/unit/session/test_session.py

@@ -1,8 +1,10 @@
 from functools import partial
-from io import IOBase
+from io import IOBase, BytesIO
 from numbers import Number
+import os
 from unittest.mock import MagicMock, Mock, PropertyMock, call, patch, ANY
 from requests.exceptions import RequestException, SSLError, ConnectionError as RequestsConnectionError
+from requests_toolbelt import MultipartEncoder
 
 import pytest
 
@@ -449,3 +451,48 @@ def test_proxy_malformed_dict_does_not_attach(box_session, monkeypatch, mock_net
 
 def test_proxy_network_config_property(box_session):
     assert isinstance(box_session.proxy_config, Proxy)
+
+
+def test_multipart_request_with_disabled_streaming_file_content(
+        box_session, mock_network_layer, generic_successful_response):
+    test_url = 'https://example.com'
+    file_bytes = os.urandom(1024)
+    mock_network_layer.request.side_effect = [generic_successful_response]
+    box_session.post(
+        url=test_url,
+        files={'file': ('unused', BytesIO(file_bytes))},
+        data={'attributes': '{"name": "test_file"}'},
+        stream_file_content=False
+    )
+    mock_network_layer.request.assert_called_once_with(
+        'POST',
+        test_url,
+        access_token='fake_access_token',
+        headers=ANY,
+        log_response_content=True,
+        files={'file': ('unused', ANY)},
+        data={'attributes': '{"name": "test_file"}'},
+    )
+
+
+def test_multipart_request_with_enabled_streaming_file_content(
+        box_session, mock_network_layer, generic_successful_response):
+    test_url = 'https://example.com'
+    file_bytes = os.urandom(1024)
+    mock_network_layer.request.side_effect = [generic_successful_response]
+    box_session.post(
+        url=test_url,
+        files={'file': ('unused', BytesIO(file_bytes))},
+        data={'attributes': '{"name": "test_file"}'},
+        stream_file_content=True
+    )
+    call_args = mock_network_layer.request.call_args[0]
+    call_kwargs = mock_network_layer.request.call_args[1]
+    assert call_args[0] == 'POST'
+    assert call_args[1] == test_url
+    assert call_kwargs['access_token'] == 'fake_access_token'
+    assert call_kwargs['log_response_content'] is True
+    assert isinstance(call_kwargs['data'], MultipartEncoder)
+    assert call_kwargs['data'].fields['attributes'] == '{"name": "test_file"}'
+    assert call_kwargs['data'].fields['file'][0] == 'unused'
+    assert isinstance(call_kwargs['data'].fields['file'][1], BytesIO)