Handle Content-Type and Content-Disposition headers

giftless/storage/__init__.py

@@ -1,3 +1,4 @@
+import mimetypes
 from abc import ABC
 from typing import Any, BinaryIO, Dict, Iterable, Optional
 
@@ -32,6 +33,9 @@ def exists(self, prefix: str, oid: str) -> bool:
     def get_size(self, prefix: str, oid: str) -> int:
         pass
 
+    def get_mime_type(self, prefix: str, oid: str) -> Optional[str]:
+        return "application/octet-stream"
+
     def verify_object(self, prefix: str, oid: str, size: int):
         """Verify that an object exists
         """
@@ -85,3 +89,7 @@ def verify_object(self, prefix: str, oid: str, size: int) -> bool:
             return self.get_size(prefix, oid) == size
         except exc.ObjectNotFound:
             return False
+
+
+def guess_mime_type_from_filename(filename: str) -> Optional[str]:
+    return mimetypes.guess_type(filename)[0]

giftless/storage/amazon_s3.py

@@ -79,9 +79,16 @@ def get_download_action(self, prefix: str, oid: str, size: int, expires_in: int,
             'Bucket': self.bucket_name,
             'Key': self._get_blob_path(prefix, oid)
         }
-        if extra and 'filename' in extra:
-            filename = safe_filename(extra['filename'])
+
+        filename = extra.get('filename') if extra else None
+        disposition = extra.get('disposition', 'attachment') if extra else 'attachment'
+
+        if filename and disposition:
+            filename = safe_filename(filename)
             params['ResponseContentDisposition'] = f'attachment; filename="{filename}"'
+        elif disposition:
+            params['ResponseContentDisposition'] = disposition
+
         response = self.s3_client.generate_presigned_url('get_object',
                                                          Params=params,
                                                          ExpiresIn=expires_in

giftless/storage/azure.py

@@ -10,7 +10,7 @@
 from azure.core.exceptions import ResourceNotFoundError
 from azure.storage.blob import BlobClient, BlobSasPermissions, BlobServiceClient, generate_blob_sas  # type: ignore
 
-from giftless.storage import ExternalStorage, MultipartStorage, StreamingStorage
+from giftless.storage import ExternalStorage, MultipartStorage, StreamingStorage, guess_mime_type_from_filename
 
 from .exc import ObjectNotFound
 
@@ -63,28 +63,50 @@ def get_size(self, prefix: str, oid: str) -> int:
         except ResourceNotFoundError:
             raise ObjectNotFound("Object does not exist")
 
+    def get_mime_type(self, prefix: str, oid: str) -> Optional[str]:
+        try:
+            blob_client = self.blob_svc_client.get_blob_client(container=self.container_name,
+                                                               blob=self._get_blob_path(prefix, oid))
+            props = blob_client.get_blob_properties()
+            mime_type = props.content_settings.get(
+                "content_type", "application/octet-stream")
+            return mime_type  # type: ignore
+        except ResourceNotFoundError:
+            raise ObjectNotFound("Object does not exist")
+
     def get_upload_action(self, prefix: str, oid: str, size: int, expires_in: int,
                           extra: Optional[Dict[str, Any]] = None) -> Dict[str, Any]:
         filename = extra.get('filename') if extra else None
-        return {
+        headers = {
+            "x-ms-blob-type": "BlockBlob",
+        }
+        reply = {
             "actions": {
                 "upload": {
                     "href": self._get_signed_url(prefix, oid, expires_in, filename, create=True),
-                    "header": {
-                        "x-ms-blob-type": "BlockBlob",
-                    },
                     "expires_in": expires_in
                 }
             }
         }
 
+        if filename:
+            mime_type = guess_mime_type_from_filename(filename)
+            if mime_type:
+                headers["x-ms-blob-content-type"] = mime_type
+
+        reply["actions"]["upload"]["header"] = headers
+
+        return reply
+
     def get_download_action(self, prefix: str, oid: str, size: int, expires_in: int,
                             extra: Optional[Dict[str, Any]] = None) -> Dict[str, Any]:
         filename = extra.get('filename') if extra else None
+        disposition = extra.get('disposition', 'attachment') if extra else 'attachment'
+
         return {
             "actions": {
                 "download": {
-                    "href": self._get_signed_url(prefix, oid, expires_in, filename, read=True),
+                    "href": self._get_signed_url(prefix, oid, expires_in, filename, disposition=disposition, read=True),
                     "header": {},
                     "expires_in": expires_in
                 }
@@ -104,7 +126,6 @@ def get_multipart_actions(self, prefix: str, oid: str, size: int, part_size: int
         _log.info("There are %d uncommitted blocks pre-uploaded; %d parts still need to be uploaded",
                   len(uncommitted), len(parts))
         commit_body = self._create_commit_body(blocks)
-
         reply: Dict[str, Any] = {
             "actions": {
                 "commit": {
@@ -123,6 +144,10 @@ def get_multipart_actions(self, prefix: str, oid: str, size: int, part_size: int
                 }
             }
         }
+        if filename:
+            mime_type = guess_mime_type_from_filename(filename)
+            if mime_type:
+                reply["actions"]["commit"]["header"]["x-ms-blob-content-type"] = mime_type
 
         if parts:
             reply['actions']['parts'] = parts
@@ -141,14 +166,16 @@ def _get_blob_path(self, prefix: str, oid: str) -> str:
         return os.path.join(storage_prefix, prefix, oid)
 
     def _get_signed_url(self, prefix: str, oid: str, expires_in: int, filename: Optional[str] = None,
-                        **permissions: bool) -> str:
+                        disposition: Optional[str] = None, **permissions: bool) -> str:
         blob_name = self._get_blob_path(prefix, oid)
         permissions = BlobSasPermissions(**permissions)
         token_expires = (datetime.now(tz=timezone.utc) + timedelta(seconds=expires_in))
 
         extra_args = {}
-        if filename:
-            extra_args['content_disposition'] = f'attachment; filename="{filename}"'
+        if filename and disposition:
+            extra_args['content_disposition'] = f'{disposition}; filename="{filename}"'
+        elif disposition:
+            extra_args['content_disposition'] = f'{disposition};"'
 
         sas_token = generate_blob_sas(account_name=self.blob_svc_client.account_name,
                                       account_key=self.blob_svc_client.credential.account_key,

giftless/storage/google_cloud.py

@@ -68,10 +68,13 @@ def get_upload_action(self, prefix: str, oid: str, size: int, expires_in: int,
     def get_download_action(self, prefix: str, oid: str, size: int, expires_in: int,
                             extra: Optional[Dict[str, Any]] = None) -> Dict[str, Any]:
         filename = extra.get('filename') if extra else None
+        disposition = extra.get('disposition', 'attachment') if extra else 'attachment'
+
         return {
             "actions": {
                 "download": {
-                    "href": self._get_signed_url(prefix, oid, expires_in=expires_in, filename=filename),
+                    "href": self._get_signed_url(
+                        prefix, oid, expires_in=expires_in, filename=filename, disposition=disposition),
                     "header": {},
                     "expires_in": expires_in
                 }
@@ -90,10 +93,13 @@ def _get_blob_path(self, prefix: str, oid: str) -> str:
         return os.path.join(storage_prefix, prefix, oid)
 
     def _get_signed_url(self, prefix: str, oid: str, expires_in: int, http_method: str = 'GET',
-                        filename: Optional[str] = None) -> str:
+                        filename: Optional[str] = None, disposition: Optional[str] = None) -> str:
         bucket = self.storage_client.bucket(self.bucket_name)
         blob = bucket.blob(self._get_blob_path(prefix, oid))
         disposition = f'attachment; filename={filename}' if filename else None
+        if filename and disposition:
+            disposition = f'{disposition}; filename="{filename}"'
+
         url: str = blob.generate_signed_url(expiration=timedelta(seconds=expires_in), method=http_method, version='v4',
                                             response_disposition=disposition, credentials=self.credentials)
         return url

giftless/storage/local_storage.py

@@ -42,6 +42,11 @@ def get_size(self, prefix: str, oid: str) -> int:
             return os.path.getsize(self._get_path(prefix, oid))
         raise exc.ObjectNotFound("Object was not found")
 
+    def get_mime_type(self, prefix: str, oid: str) -> str:
+        if self.exists(prefix, oid):
+            return "application/octet-stream"
+        raise exc.ObjectNotFound("Object was not found")
+
     def get_multipart_actions(self, prefix: str, oid: str, size: int, part_size: int, expires_in: int,
                               extra: Optional[Dict[str, Any]] = None) -> Dict[str, Any]:
         return super().get_multipart_actions(prefix, oid, size, part_size, expires_in, extra)

giftless/transfer/basic_streaming.py

@@ -83,10 +83,18 @@ def get(self, organization, repo, oid):
 
         filename = request.args.get('filename')
         filename = safe_filename(filename)
-        headers = {'Content-Disposition': f'attachment; filename="{filename}"'} if filename else None
+        disposition = request.args.get('disposition')
+
+        headers = {}
+        if filename and disposition:
+            headers = {'Content-Disposition': f'attachment; filename="{filename}"'}
+        elif disposition:
+            headers = {'Content-Disposition': disposition}
 
         if self.storage.exists(path, oid):
             file = self.storage.get(path, oid)
+            mime_type = self.storage.get_mime_type(path, oid)
+            headers['Content-Type'] = mime_type
             return Response(file, direct_passthrough=True, status=200, headers=headers)
         else:
             raise NotFound("The object was not found")

tests/test_schema.py

@@ -38,11 +38,15 @@ def test_batch_request_default_transfer():
 
 
 def test_object_schema_accepts_x_fields():
-    payload = {"oid": "123abc", "size": 1212, "x-filename": "foobarbaz", "x-mtime": 123123123123}
+    payload = {
+        "oid": "123abc", "size": 1212, "x-filename": "foobarbaz",
+        "x-mtime": 123123123123, "x-disposition": "inline"
+    }
     parsed = schema.ObjectSchema().load(payload)
     assert "foobarbaz" == parsed['extra']['filename']
     assert 123123123123 == parsed['extra']['mtime']
     assert "123abc" == parsed['oid']
+    assert "inline" == parsed['extra']['disposition']
 
 
 def test_object_schema_rejects_unknown_fields():