[BACKEND-829] fix content length (#30)

* package data for upload properly to fix content length being incorrect for file uploads

* basename -> filename

* yapf

* fix mypy

* yapf

* test coverage for video uploads

* test video

* clear parens

Author: rllin

labelbox/client.py

@@ -3,6 +3,7 @@
 import logging
 import mimetypes
 import os
+from typing import Tuple
 
 import requests
 import requests.exceptions
@@ -176,7 +177,7 @@ def check_errors(keywords, *path):
 
         return response["data"]
 
-    def upload_file(self, path):
+    def upload_file(self, path: str) -> str:
         """Uploads given path to local file.
 
         Also includes best guess at the content type of the file.
@@ -190,26 +191,36 @@ def upload_file(self, path):
 
         """
         content_type, _ = mimetypes.guess_type(path)
-        basename = os.path.basename(path)
+        filename = os.path.basename(path)
         with open(path, "rb") as f:
-            return self.upload_data(data=(basename, f.read(), content_type))
-
-    def upload_data(self, data):
+            return self.upload_data(content=f.read(),
+                                    filename=filename,
+                                    content_type=content_type)
+
+    def upload_data(self,
+                    content: bytes,
+                    filename: str = None,
+                    content_type: str = None) -> str:
         """ Uploads the given data (bytes) to Labelbox.
 
         Args:
-            data (bytes): The data to upload.
+            content: bytestring to upload
+            filename: name of the upload
+            content_type: content type of data uploaded
+
         Returns:
             str, the URL of uploaded data.
+
         Raises:
             labelbox.exceptions.LabelboxError: If upload failed.
         """
+
         request_data = {
             "operations":
                 json.dumps({
                     "variables": {
                         "file": None,
-                        "contentLength": len(data),
+                        "contentLength": len(content),
                         "sign": False
                     },
                     "query":
@@ -224,7 +235,10 @@ def upload_data(self, data):
             self.endpoint,
             headers={"authorization": "Bearer %s" % self.api_key},
             data=request_data,
-            files={"1": data})
+            files={
+                "1": (filename, content, content_type) if
+                     (filename and content_type) else content
+            })
 
         try:
             file_data = response.json().get("data", None)

tests/integration/conftest.py

@@ -108,3 +108,10 @@ def iframe_url(environ) -> str:
         Environ.PROD: 'https://editor.labelbox.com',
         Environ.STAGING: 'https://staging-editor.labelbox.com',
     }[environ]
+
+
+@pytest.fixture
+def sample_video() -> str:
+    path_to_video = 'tests/integration/media/cat.mp4'
+    assert os.path.exists(path_to_video)
+    return path_to_video

tests/integration/media/cat.mp4

@@ -1,4 +1,5 @@
 import pytest
+import requests
 
 from labelbox import Dataset
 from labelbox.exceptions import ResourceNotFoundError
@@ -78,3 +79,23 @@ def test_get_data_row_for_external_id(dataset, rand_gen):
 
     with pytest.raises(ResourceNotFoundError):
         data_row = dataset.data_row_for_external_id(external_id)
+
+
+def test_upload_video_file(dataset, sample_video: str) -> None:
+    """
+    Tests that a mp4 video can be uploaded and preserve content length
+    and content type.
+
+    """
+    dataset.create_data_row(row_data=sample_video)
+    task = dataset.create_data_rows([sample_video, sample_video])
+    task.wait_till_done()
+
+    with open(sample_video, 'rb') as video_f:
+        content_length = len(video_f.read())
+
+    for data_row in dataset.data_rows():
+        url = data_row.row_data
+        response = requests.head(url, allow_redirects=True)
+        assert int(response.headers['Content-Length']) == content_length
+        assert response.headers['Content-Type'] == 'video/mp4'