Merge pull request #12 from aoxolotl/fixMimetype

Support automatic mimetype inference

README.md

@@ -13,6 +13,12 @@
   <img alt="Build Status" src="https://img.shields.io/github/actions/workflow/status/datatorch/python/package.yml?branch=master">
 </p>
 
+## Installation
+
+*Note*: Newer versions (>=0.4.8.2) of DataTorch client require `libmagic` for file uploads
+to datasets. Your system may or may not have `libmagic` pre-installed. Please
+make sure you have `libmagic` installed before proceeding.
+
 ```bash
 pip install datatorch
 ```

datatorch/api/api.py

@@ -4,6 +4,7 @@
 from typing import IO, overload, cast
 from urllib import request
 from urllib.parse import urlencode
+import magic
 
 from datatorch.api.entity.dataset import Dataset
 from datatorch.api.entity.storage_link import StorageLink
@@ -141,10 +142,16 @@ def upload_to_default_filesource(
         datasetId = "" if dataset is None else dataset.id
         importFiles = "false" if dataset is None else "true"
         endpoint = f"{self.api_url}/file/v1/upload/{storageId}?path={storageFolderName}&import={importFiles}&datasetId={datasetId}"
-        # r = requests.post(endpoint, files={"file": file}, user=self.viewer)
+
+        # save the current position
+        tell = file.tell()
+        # read 1024 bytes and get the mimetype
+        mimetype = magic.from_buffer(file.read(1024), mime=True)
+        # go back to the saved position
+        file.seek(tell)
         r = requests.post(
             endpoint,
-            files={"file": file},
+            files={"file": (os.path.basename(file.name), file, mimetype)},
             headers={self.token_header: self._api_token},
             stream=True,
         )

setup.py

@@ -8,8 +8,6 @@
 
 requirements = [
     "Click==8.0.0",
-    "numpy",
-    "docker",
     "gql==3.4.0",
     "websockets==10.4",
     "websocket-client",
@@ -26,13 +24,16 @@
     "shapely==2.0.1",
     "tqdm~=4.65.0",
     "urllib3==1.26.15",
+    "numpy",
+    "docker",
+    "python-magic",
 ]
 
 requirements_agents = []
 
 setup(
     name="datatorch",
-    version="0.4.8.1",
+    version="0.4.8.2",
     description="A CLI and library for interacting with DataTorch.",
     author="DataTorch",
     author_email="support@datatorch.io",