implement mpu file upload

Author: MariusWirtz

TM1py/Services/FileService.py

@@ -1,14 +1,16 @@
 # -*- coding: utf-8 -*-
+import concurrent.futures
 import json
 import warnings
+from io import BytesIO
 from pathlib import Path
 from typing import List, Iterable, Union
 
+from TM1py.Exceptions import TM1pyVersionException
 from TM1py.Services import RestService
 from TM1py.Services.ObjectService import ObjectService
 from TM1py.Utils import format_url
 from TM1py.Utils.Utils import verify_version, require_version
-from TM1py.Exceptions import TM1pyVersionException
 
 
 class FileService(ObjectService):
@@ -37,7 +39,7 @@ def get_names(self, **kwargs) -> bytes:
             version_content_path=self.version_content_path)
 
         return self._rest.GET(url, **kwargs).content
-    
+
     @require_version(version="11.4")
     def get_all_names(self, path: Union[str, Path] = "", **kwargs) -> List[str]:
         """ return list of blob file names
@@ -49,16 +51,16 @@ def get_all_names(self, path: Union[str, Path] = "", **kwargs) -> List[str]:
 
         response = self._rest.GET(url, **kwargs).content
         return [file['Name'] for file in json.loads(response)['value']]
-    
+
     @require_version(version="11.4")
     def get(self, file_name: str, **kwargs) -> bytes:
         """ Get file
 
         :param file_name: file name in root or path to file
         """
         path = Path(file_name)
-        self._check_subfolder_support(path = path, function="FileService.get")
-        
+        self._check_subfolder_support(path=path, function="FileService.get")
+
         url = self._construct_content_url(
             path=path,
             exclude_path_end=False,
@@ -108,23 +110,27 @@ def _construct_content_url(self, path: Path, exclude_path_end: bool = True, exte
             **parent_folders)
 
         return url.rstrip("/")
-    
+
     def _check_subfolder_support(self, path: Path, function: str) -> None:
         REQUIRED_VERSION = "12"
         if len(path.parts) > 1 and not verify_version(required_version=REQUIRED_VERSION, version=self.version):
-            raise TM1pyVersionException(function=function, required_version=REQUIRED_VERSION , feature='Subfolder')
-    
+            raise TM1pyVersionException(function=function, required_version=REQUIRED_VERSION, feature='Subfolder')
+
     @require_version(version="11.4")
-    def create(self, file_name: Union[str, Path], file_content: bytes, **kwargs):
+    def create(self, file_name: Union[str, Path], file_content: bytes, multi_part_upload: bool = False,
+               max_mb_per_part: float = 200, max_workers: int = 1, **kwargs):
         """ Create file
 
         Folders in file_name (e.g. folderA/folderB/file.csv) will be created implicitly
 
         :param file_name: file name in root or path to file
         :param file_content: file_content as bytes or BytesIO
+        :param multi_part_upload: boolean use multipart upload or not (only available from TM1 12 onwards)
+        :param max_mb_per_part: max megabyte per part in multipart upload (only available from TM1 12 onwards)
+        :param max_workers: max parallel workers for multipart upload (only available from TM1 12 onwards)
         """
         path = Path(file_name)
-        self._check_subfolder_support(path = path, function = "FileService.create")
+        self._check_subfolder_support(path=path, function="FileService.create")
 
         # Create folder structure iteratively
         if path.parents:
@@ -141,46 +147,133 @@ def create(self, file_name: Union[str, Path], file_content: bytes, **kwargs):
         }
         self._rest.POST(url, json.dumps(body), **kwargs)
 
+        return self._upload_file_content(path, file_content, multi_part_upload, max_mb_per_part, max_workers, **kwargs)
+
+    def _upload_file_content(
+            self,
+            path: Path,
+            file_content: bytes,
+            multi_part_upload: bool = False,
+            max_mb_per_part: float = 200,
+            max_workers: int = 1,
+            **kwargs):
+
         url = self._construct_content_url(path, exclude_path_end=False, extension="Content")
-        return self._rest.PUT(
-            url=url,
-            data=file_content,
-            headers=self.binary_http_header,
+
+        if not multi_part_upload:
+            return self._rest.PUT(
+                url=url,
+                data=file_content,
+                headers=self.binary_http_header,
+                **kwargs)
+
+        self.upload_file_content_with_mpu(url, file_content, max_mb_per_part, max_workers, **kwargs)
+
+    def upload_file_content_with_mpu(self, content_url: str, file_content: bytes, max_mb_per_part: float,
+                                     max_workers: int = 1, **kwargs):
+        # Initiate multipart upload
+        response = self._rest.POST(
+            url=content_url + "/mpu.CreateMultipartUpload",
+            data="{}",
+            async_requests_mode=False,
             **kwargs)
-    
+        upload_id = response.json()['UploadID']
+
+        # Split the file content into parts
+        parts_to_upload = self._split_into_parts(
+            data=file_content,
+            max_chunk_size=int(max_mb_per_part * 1024 * 1024)
+        )
+
+        part_numbers_and_etags = []
+
+        # helper function for uploading each part
+        def upload_part(part_index, bytes_part):
+            response = self._rest.POST(
+                url=content_url + f"/!uploads('{upload_id}')/Parts",
+                data=bytes_part,
+                headers={**self.binary_http_header, 'Accept': 'application/json,text/plain'},
+                async_requests_mode=False,
+                **kwargs)
+            return part_index, response.json()["PartNumber"], response.json()["@odata.etag"]
+
+        if max_workers > 1:
+            # upload parts concurrently
+            with concurrent.futures.ThreadPoolExecutor(max_workers=max_workers) as executor:
+
+                futures = {
+                    executor.submit(upload_part, i, part): i
+                    for i, part
+                    in enumerate(parts_to_upload)}
+
+                for future in concurrent.futures.as_completed(futures):
+                    part_index, part_number, etag = future.result()
+                    part_numbers_and_etags.append((part_index, part_number, etag))
+
+        else:
+            # Sequential upload
+            for i, bytes_part in enumerate(parts_to_upload):
+                part_index, part_number, etag = upload_part(i, bytes_part)
+                part_numbers_and_etags.append((part_index, part_number, etag))
+
+        # Complete the multipart upload
+        self._rest.POST(
+            url=content_url + f"/!uploads('{upload_id}')/mpu.Complete",
+            data=json.dumps(
+                {"Parts": [
+                    {"PartNumber": part_number, "ETag": etag}
+                    for _, part_number, etag in sorted(part_numbers_and_etags)
+                ]}
+            )
+        )
+
+    def _split_into_parts(self, data: Union[bytes, BytesIO], max_chunk_size: int = 200 * 1024 * 1024):
+        # Convert data to bytes if it's a BytesIO object
+        if isinstance(data, BytesIO):
+            data = data.getvalue()
+
+        # List to store chunks
+        parts = []
+
+        # Split data into chunks
+        for i in range(0, len(data), max_chunk_size):
+            part = data[i:i + max_chunk_size]
+            parts.append(part)
+
+        return parts
+
     @require_version(version="11.4")
-    def update(self, file_name: Union[str, Path], file_content: bytes, **kwargs):
+    def update(self, file_name: Union[str, Path], file_content: bytes, multi_part_upload: bool = False,
+               max_mb_per_part: float = 200, max_workers: int = 1, **kwargs):
         """ Update existing file
 
         :param file_name: file name in root or path to file
         :param file_content: file_content as bytes or BytesIO
+        :param multi_part_upload: boolean use multipart upload or not (only available from TM1 12 onwards)
+        :param max_mb_per_part: max megabyte per part in multipart upload (only available from TM1 12 onwards)
+        :param max_workers: max parallel workers for multipart upload (only available from TM1 12 onwards)
         """
         path = Path(file_name)
-        self._check_subfolder_support(path = path, function = "FileService.update")
+        self._check_subfolder_support(path=path, function="FileService.update")
 
-        url = self._construct_content_url(
-            path=path,
-            exclude_path_end=False,
-            extension="Content")
+        return self._upload_file_content(path, file_content, multi_part_upload, max_mb_per_part, max_workers, **kwargs)
 
-        return self._rest.PUT(
-            url=url,
-            data=file_content,
-            headers=self.binary_http_header,
-            **kwargs)
-    
     @require_version(version="11.4")
-    def update_or_create(self, file_name: Union[str, Path], file_content: bytes, **kwargs):
+    def update_or_create(self, file_name: Union[str, Path], file_content: bytes, multi_part_upload: bool = False,
+                         max_mb_per_part: float = 200, max_workers: int = 1, **kwargs):
         """ Create file or update file if it already exists
 
         :param file_name: file name in root or path to file
         :param file_content: file_content as bytes or BytesIO
+        :param multi_part_upload: boolean use multipart upload or not (only available from TM1 12 onwards)
+        :param max_mb_per_part: max megabyte per part in multipart upload (only available from TM1 12 onwards)
+        :param max_workers: max parallel workers for multipart upload (only available from TM1 12 onwards)
         """
         if self.exists(file_name, **kwargs):
-            return self.update(file_name, file_content, **kwargs)
+            return self.update(file_name, file_content, multi_part_upload, max_mb_per_part, max_workers, **kwargs)
+
+        return self.create(file_name, file_content, multi_part_upload, max_mb_per_part, max_workers, **kwargs)
 
-        return self.create(file_name, file_content, **kwargs)
-    
     @require_version(version="11.4")
     def exists(self, file_name: Union[str, Path], **kwargs):
         """ Check if file exists
@@ -193,23 +286,23 @@ def exists(self, file_name: Union[str, Path], **kwargs):
             extension="")
 
         return self._exists(url, **kwargs)
-    
+
     @require_version(version="11.4")
     def delete(self, file_name: Union[str, Path], **kwargs):
         """ Delete file
 
         :param file_name: file name in root or path to file
         """
         path = Path(file_name)
-        self._check_subfolder_support(path = path, function = "FileService.delete")
+        self._check_subfolder_support(path=path, function="FileService.delete")
 
         url = self._construct_content_url(
             path=path,
             exclude_path_end=False,
             extension="")
 
         return self._rest.DELETE(url, **kwargs)
-    
+
     @require_version(version="11.4")
     def search_string_in_name(self, name_startswith: str = None, name_contains: Iterable = None,
                               name_contains_operator: str = 'and', path: Union[Path, str] = "",
@@ -242,9 +335,9 @@ def search_string_in_name(self, name_startswith: str = None, name_contains: Iter
 
             else:
                 raise ValueError("'name_contains' must be str or iterable")
-            
+
         path = Path(path)
-        self._check_subfolder_support(path = path, function = "FileService.search_string_in_name")
+        self._check_subfolder_support(path=path, function="FileService.search_string_in_name")
 
         url = self._construct_content_url(
             path=Path(path),

Tests/FileService_test.py

@@ -41,16 +41,37 @@ def setUpV12(self):
         if self.tm1.files.exists(self.FILE_NAME2_IN_FOLDER):
             self.tm1.files.delete(self.FILE_NAME2_IN_FOLDER)
 
-    @skip_if_version_lower_than(version="11.4")
-    def test_create_get(self):
+
+    def run_create_get(self, mpu, max_mb_per_part=None, max_workers=1):
         with open(Path(__file__).parent.joinpath('resources', 'file.csv'), "rb") as original_file:
-            self.tm1.files.update_or_create(self.FILE_NAME1, original_file.read())
+            self.tm1.files.update_or_create(
+                self.FILE_NAME1,
+                original_file.read(),
+                multi_part_upload=mpu,
+                max_mb_per_part=max_mb_per_part,
+                max_workers=max_workers)
 
             created_file = self.tm1.files.get(self.FILE_NAME1)
-
         with open(Path(__file__).parent.joinpath('resources', 'file.csv'), "rb") as original_file:
             self.assertEqual(original_file.read(), created_file)
 
+    @skip_if_version_lower_than(version="11.4")
+    def test_create_get(self):
+        self.run_create_get(mpu=False)
+
+    @skip_if_version_lower_than(version="12")
+    def test_create_get_with_mpu_1_byte_per_part(self):
+        self.run_create_get(mpu=True, max_mb_per_part=1/(1024*1024))
+
+
+    @skip_if_version_lower_than(version="12")
+    def test_create_get_with_mpu_1_byte_per_part_10_max_workers(self):
+        self.run_create_get(mpu=True, max_mb_per_part=1/(1024*1024), max_workers=10)
+
+    @skip_if_version_lower_than(version="12")
+    def test_create_get_with_mpu_200_megabyte_per_part(self):
+        self.run_create_get(mpu=True, max_mb_per_part=200)
+
     @skip_if_version_lower_than(version="12")
     def test_create_get_in_folder(self):
         with open(Path(__file__).parent.joinpath('resources', 'file.csv'), "rb") as original_file:
@@ -61,6 +82,20 @@ def test_create_get_in_folder(self):
         with open(Path(__file__).parent.joinpath('resources', 'file.csv'), "rb") as original_file:
             self.assertEqual(original_file.read(), created_file)
 
+    @skip_if_version_lower_than(version="12")
+    def test_create_get_in_folder_with_mpu(self):
+        with open(Path(__file__).parent.joinpath('resources', 'file.csv'), "rb") as original_file:
+            self.tm1.files.update_or_create(
+                self.FILE_NAME1_IN_FOLDER,
+                original_file.read(),
+                mpu=True,
+                max_mb_per_part=1/(1024*1024))
+
+            created_file = self.tm1.files.get(self.FILE_NAME1_IN_FOLDER)
+
+        with open(Path(__file__).parent.joinpath('resources', 'file.csv'), "rb") as original_file:
+            self.assertEqual(original_file.read(), created_file)
+
     @skip_if_version_lower_than(version="11.4")
     @skip_if_version_higher_or_equal_than(version="12")
     def test_create_in_folder_exception(self):