allow users to pass neptune data directory as env var

CHANGELOG.md

@@ -1,9 +1,13 @@
-##  [UNRELEASED] neptune 1.5.0
+## [UNRELEASED] neptune 1.5.0
+
+### Features
+- Users can pass neptune data directory path by env variable ([#1409](https://github.com/neptune-ai/neptune-client/pull/1409))
 
 ### Fixes
 - Load CLI plug-ins in try..except block to avoid a failure in loading a plug-in to crash entire CLI ([#1392](https://github.com/neptune-ai/neptune-client/pull/1392))
 
 
+
 ##  neptune 1.4.1
 
 ### Fixes

src/neptune/internal/operation_processors/async_operation_processor.py

@@ -29,10 +29,7 @@
     Optional,
 )
 
-from neptune.constants import (
-    ASYNC_DIRECTORY,
-    NEPTUNE_DATA_DIRECTORY,
-)
+from neptune.constants import ASYNC_DIRECTORY
 from neptune.envs import NEPTUNE_SYNC_AFTER_STOP_TIMEOUT
 from neptune.exceptions import NeptuneSynchronizationAlreadyStoppedException
 from neptune.internal.backends.neptune_backend import NeptuneBackend
@@ -41,7 +38,10 @@
 from neptune.internal.id_formats import UniqueId
 from neptune.internal.operation import Operation
 from neptune.internal.operation_processors.operation_processor import OperationProcessor
-from neptune.internal.operation_processors.operation_storage import OperationStorage
+from neptune.internal.operation_processors.operation_storage import (
+    OperationStorage,
+    get_container_dir,
+)
 from neptune.internal.threading.daemon import Daemon
 from neptune.internal.utils.logger import logger
 
@@ -84,7 +84,7 @@ def __init__(
     @staticmethod
     def _init_data_path(container_id: UniqueId, container_type: ContainerType) -> Path:
         now = datetime.now()
-        container_dir = f"{NEPTUNE_DATA_DIRECTORY}/{ASYNC_DIRECTORY}/{container_type.create_dir_name(container_id)}"
+        container_dir = get_container_dir(ASYNC_DIRECTORY, container_id, container_type)
         data_path = f"{container_dir}/exec-{now.timestamp()}-{now.strftime('%Y-%m-%d_%H.%M.%S.%f')}-{os.getpid()}"
         data_path = data_path.replace(" ", "_").replace(":", ".")
         return Path(data_path)

src/neptune/internal/operation_processors/offline_operation_processor.py

@@ -19,16 +19,16 @@
 from pathlib import Path
 from typing import Optional
 
-from neptune.constants import (
-    NEPTUNE_DATA_DIRECTORY,
-    OFFLINE_DIRECTORY,
-)
+from neptune.constants import OFFLINE_DIRECTORY
 from neptune.internal.container_type import ContainerType
 from neptune.internal.disk_queue import DiskQueue
 from neptune.internal.id_formats import UniqueId
 from neptune.internal.operation import Operation
 from neptune.internal.operation_processors.operation_processor import OperationProcessor
-from neptune.internal.operation_processors.operation_storage import OperationStorage
+from neptune.internal.operation_processors.operation_storage import (
+    OperationStorage,
+    get_container_dir,
+)
 
 
 class OfflineOperationProcessor(OperationProcessor):
@@ -44,7 +44,8 @@ def __init__(self, container_id: UniqueId, container_type: ContainerType, lock:
 
     @staticmethod
     def _init_data_path(container_id: UniqueId, container_type: ContainerType) -> Path:
-        return Path(f"{NEPTUNE_DATA_DIRECTORY}/{OFFLINE_DIRECTORY}/{container_type.create_dir_name(container_id)}")
+        container_dir = get_container_dir(OFFLINE_DIRECTORY, container_id, container_type)
+        return Path(container_dir)
 
     def enqueue_operation(self, op: Operation, *, wait: bool) -> None:
         self._queue.put(op)

src/neptune/internal/operation_processors/operation_storage.py

@@ -13,9 +13,7 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 #
-__all__ = [
-    "OperationStorage",
-]
+__all__ = ["OperationStorage", "get_container_dir"]
 
 import os
 import shutil
@@ -27,6 +25,14 @@
 from neptune.internal.utils.logger import logger
 
 
+def get_container_dir(type_dir: str, container_id: UniqueId, container_type: ContainerType) -> str:
+    neptune_data_dir_env = os.getenv("NEPTUNE_DATA_DIRECTORY")
+
+    neptune_data_dir = neptune_data_dir_env if neptune_data_dir_env else NEPTUNE_DATA_DIRECTORY
+
+    return f"{neptune_data_dir}/{type_dir}/{container_type.create_dir_name(container_id)}"
+
+
 class OperationStorage:
     UPLOAD_PATH: str = "upload_path"
 
@@ -44,10 +50,6 @@ def data_path(self) -> Path:
     def upload_path(self) -> Path:
         return self.data_path / "upload_path"
 
-    @staticmethod
-    def _get_container_dir(type_dir: str, container_id: UniqueId, container_type: ContainerType):
-        return f"{NEPTUNE_DATA_DIRECTORY}/{type_dir}/{container_type.create_dir_name(container_id)}"
-
     def close(self):
         shutil.rmtree(self.data_path, ignore_errors=True)
 

src/neptune/internal/operation_processors/sync_operation_processor.py

@@ -19,16 +19,16 @@
 from pathlib import Path
 from typing import Optional
 
-from neptune.constants import (
-    NEPTUNE_DATA_DIRECTORY,
-    SYNC_DIRECTORY,
-)
+from neptune.constants import SYNC_DIRECTORY
 from neptune.internal.backends.neptune_backend import NeptuneBackend
 from neptune.internal.container_type import ContainerType
 from neptune.internal.id_formats import UniqueId
 from neptune.internal.operation import Operation
 from neptune.internal.operation_processors.operation_processor import OperationProcessor
-from neptune.internal.operation_processors.operation_storage import OperationStorage
+from neptune.internal.operation_processors.operation_storage import (
+    OperationStorage,
+    get_container_dir,
+)
 
 
 class SyncOperationProcessor(OperationProcessor):
@@ -41,7 +41,7 @@ def __init__(self, container_id: UniqueId, container_type: ContainerType, backen
     @staticmethod
     def _init_data_path(container_id: UniqueId, container_type: ContainerType) -> Path:
         now = datetime.now()
-        container_dir = f"{NEPTUNE_DATA_DIRECTORY}/{SYNC_DIRECTORY}/{container_type.create_dir_name(container_id)}"
+        container_dir = get_container_dir(SYNC_DIRECTORY, container_id, container_type)
         data_path = f"{container_dir}/exec-{now.timestamp()}-{now.strftime('%Y-%m-%d_%H.%M.%S.%f')}"
         return Path(data_path)