Compare commits
| Author | SHA1 | Date | |
|---|---|---|---|
|
|
0423913b3d
|
||
|
|
d60221de21
|
||
|
|
4c9954c1fd
|
||
|
|
914dd729aa
|
||
|
|
e5c28835b8
|
||
|
|
0cac30ccf8
|
||
|
|
1a4023ba47
|
||
|
|
8e03f7a7f4
|
+45
-17
@@ -2,6 +2,20 @@
|
||||
|
||||
All notable changes to this project will be documented in this file.
|
||||
|
||||
## [1.3.19] - 2026-09-04
|
||||
|
||||
### Fixed
|
||||
- Reported failed transfers instead of ending in success, so a run that lost files exits non-zero.
|
||||
- Left an existing local file intact when a download fails part way through.
|
||||
- Opened password-protected DOI folders, which previously failed outright.
|
||||
- Kept a session working across concurrent runs and long server queues.
|
||||
- Other minor fixes to `ls` output and remote URL handling.
|
||||
|
||||
### Changed
|
||||
- Moved packaging to PEP 621 metadata with flit as the build backend.
|
||||
- Updated dependencies.
|
||||
- **Breaking for embedders:** `CacheInterface` now requires `lock()` and `reload()`.
|
||||
|
||||
## [1.3.18] - 2026-07-02
|
||||
|
||||
### Added
|
||||
@@ -37,36 +51,41 @@ All notable changes to this project will be documented in this file.
|
||||
|
||||
## [1.3.15] - 2026-05-01
|
||||
|
||||
### Changed
|
||||
- Bumped package version to 1.3.15.
|
||||
### Fixed
|
||||
- Apply NFC normalization to filenames and folder names sent to the server.
|
||||
|
||||
## [1.3.14] - 2026-04-17
|
||||
|
||||
### Changed
|
||||
- Simplified `config list` command (removed `-l`/`--long` option, always display URL).
|
||||
- Renamed `--quick` option to `--quiet` for `ls` subcommand.
|
||||
|
||||
### Added
|
||||
- Added subcommand aliases for config commands (e.g. `ls` alias for list, `rm` alias for delete).
|
||||
- Added `version` command.
|
||||
|
||||
## [1.3.13] - 2025-07-02
|
||||
|
||||
### Changed
|
||||
- Bumped package version to 1.3.13.
|
||||
### Fixed
|
||||
- Fixed pagination logic for the `file.list` API.
|
||||
|
||||
## [1.3.12] - 2025-05-20
|
||||
|
||||
### Changed
|
||||
- Bumped package version to 1.3.12.
|
||||
### Fixed
|
||||
- Fixed bug where file downloading was skipped incorrectly when `-s`/`--skip-if-file-exists` option was present.
|
||||
|
||||
## [1.3.11] - 2025-01-21
|
||||
|
||||
### Changed
|
||||
- Bumped package version to 1.3.11.
|
||||
### Fixed
|
||||
- Follow-up fixes for User API specification changes.
|
||||
|
||||
## [1.3.10] - 2024-12-23
|
||||
|
||||
### Added
|
||||
- Delete broken files and show a summary when a file download fails.
|
||||
|
||||
### Changed
|
||||
- Bumped package version to 1.3.10.
|
||||
- Updated dependency libraries.
|
||||
|
||||
## [1.3.9] - 2024-10-23
|
||||
|
||||
@@ -75,8 +94,11 @@ All notable changes to this project will be documented in this file.
|
||||
|
||||
## [1.3.8] - 2024-09-18
|
||||
|
||||
### Changed
|
||||
- Bumped package version to 1.3.8.
|
||||
### Added
|
||||
- Implemented `-s`/`--skip-if-file-exists` option for `download` command.
|
||||
|
||||
### Fixed
|
||||
- Added exception handling for unexpected responses from the server.
|
||||
|
||||
## [1.3.7] - 2024-07-22
|
||||
|
||||
@@ -85,23 +107,29 @@ All notable changes to this project will be documented in this file.
|
||||
|
||||
## [1.3.6] - 2024-07-08
|
||||
|
||||
### Changed
|
||||
- Bumped package version to 1.3.6.
|
||||
### Added
|
||||
- Support cancelling recursive downloads if downloading some files fails.
|
||||
|
||||
## [1.3.5] - 2024-07-08
|
||||
|
||||
### Changed
|
||||
- Bumped package version to 1.3.5.
|
||||
### Added
|
||||
- Added authorization token validation checks for file download operations.
|
||||
|
||||
### Removed
|
||||
- Removed unnecessary debug code.
|
||||
|
||||
## [1.3.4] - 2024-07-04
|
||||
|
||||
### Added
|
||||
- Added some aliases for config sub command.
|
||||
|
||||
### Fixed
|
||||
- Fixed bug when uploading large files.
|
||||
|
||||
## [1.3.3] - 2024-02-13
|
||||
|
||||
### Changed
|
||||
- Bumped package version to 1.3.3.
|
||||
### Added
|
||||
- Implemented `-s`/`--skip-if-file-exists` option for `upload` command.
|
||||
|
||||
## [1.3.2] - 2024-02-09
|
||||
|
||||
|
||||
@@ -2,12 +2,53 @@
|
||||
|
||||
The mdrs-client-python is python library and a command-line client for up- and downloading files to and from MDRS based repository.
|
||||
|
||||
## Installing
|
||||
## Installation (Production)
|
||||
|
||||
It is recommended to use a virtual environment (`venv`) to keep your Python environment isolated.
|
||||
|
||||
### Option 1: Using standard `venv` + `pip`
|
||||
|
||||
```shell
|
||||
poetry install
|
||||
python3 -m venv .venv
|
||||
source .venv/bin/activate # On Windows: .venv\Scripts\activate
|
||||
pip install .
|
||||
```
|
||||
|
||||
### Option 2: Using `uv`
|
||||
|
||||
```shell
|
||||
uv venv
|
||||
source .venv/bin/activate # On Windows: .venv\Scripts\activate
|
||||
uv pip install .
|
||||
```
|
||||
|
||||
## Development Setup
|
||||
|
||||
To set up a local development environment with development tools (testing, formatting, linting):
|
||||
|
||||
### Option 1: Using standard `venv` + `pip`
|
||||
|
||||
```shell
|
||||
python3 -m venv .venv
|
||||
source .venv/bin/activate # On Windows: .venv\Scripts\activate
|
||||
pip install -e ".[dev]"
|
||||
```
|
||||
|
||||
### Option 2: Using `uv` (Recommended for developers)
|
||||
|
||||
```shell
|
||||
uv sync
|
||||
```
|
||||
|
||||
## Configuration
|
||||
|
||||
The following environment variables can be set, either in the shell or via a `.env` file placed in the working directory.
|
||||
|
||||
| Variable | Default | Description |
|
||||
|---|---|---|
|
||||
| `MDRS_CLIENT_CONFIG_DIRNAME` | `~/.mdrs-client` | Directory where config and login cache files are stored |
|
||||
| `MDRS_CLIENT_CONCURRENT` | `10` | Number of concurrent file transfers for upload and download |
|
||||
|
||||
## CLI Usage
|
||||
|
||||
### config create
|
||||
|
||||
+12
-1
@@ -5,7 +5,13 @@ from pydantic import TypeAdapter
|
||||
from requests import Response
|
||||
|
||||
from mdrsclient.connection import MDRSConnection
|
||||
from mdrsclient.exceptions import BadRequestException, ForbiddenException, UnauthorizedException, UnexpectedException
|
||||
from mdrsclient.exceptions import (
|
||||
BadRequestException,
|
||||
ForbiddenException,
|
||||
TokenExpiredException,
|
||||
UnauthorizedException,
|
||||
UnexpectedException,
|
||||
)
|
||||
from mdrsclient.models.error import DRFStandardizedErrors
|
||||
|
||||
|
||||
@@ -23,6 +29,11 @@ class BaseApi(ABC):
|
||||
if response.status_code == requests.codes.bad_request:
|
||||
raise BadRequestException(errors.errors[0].detail)
|
||||
elif response.status_code == requests.codes.unauthorized:
|
||||
# A request can wait in the server's queue for longer than the access
|
||||
# token it was sent with lives, so an expired token here does not mean
|
||||
# the session is over - it means this one request arrived too late.
|
||||
if any(e.code == "token_not_valid" for e in errors.errors):
|
||||
raise TokenExpiredException("Access token expired before the request was served.")
|
||||
raise UnauthorizedException("Login required.")
|
||||
elif response.status_code == requests.codes.forbidden:
|
||||
raise ForbiddenException("You do not have enough permissions. Access is denied.")
|
||||
|
||||
@@ -4,7 +4,7 @@ from pydantic import TypeAdapter
|
||||
from pydantic.dataclasses import dataclass
|
||||
|
||||
from mdrsclient.api.base import BaseApi
|
||||
from mdrsclient.api.utils import token_check
|
||||
from mdrsclient.api.utils import retry_on_expired_token, token_check
|
||||
from mdrsclient.models.doi import Doi
|
||||
|
||||
|
||||
@@ -23,6 +23,7 @@ class DoiRetrieveResponse:
|
||||
class DoiApi(BaseApi):
|
||||
ENTRYPOINT: Final[str] = "v3/doi/"
|
||||
|
||||
@retry_on_expired_token
|
||||
def retrieve(self, doi_id: str) -> Doi:
|
||||
"""Retrieve the folder associated with a DOI suffix ID (GET v3/doi/{id}/)."""
|
||||
url = self.ENTRYPOINT + doi_id + "/"
|
||||
|
||||
+54
-6
@@ -1,5 +1,6 @@
|
||||
import mimetypes
|
||||
import os
|
||||
import threading
|
||||
from typing import Any, Final
|
||||
from unicodedata import normalize
|
||||
|
||||
@@ -8,8 +9,8 @@ from pydantic.dataclasses import dataclass
|
||||
from requests_toolbelt.multipart.encoder import MultipartEncoder
|
||||
|
||||
from mdrsclient.api.base import BaseApi
|
||||
from mdrsclient.api.utils import token_check
|
||||
from mdrsclient.exceptions import UnexpectedException
|
||||
from mdrsclient.api.utils import retry_on_expired_token, token_check
|
||||
from mdrsclient.exceptions import MDRSException, UnexpectedException
|
||||
from mdrsclient.models import File
|
||||
|
||||
|
||||
@@ -30,6 +31,7 @@ class FilesApi(BaseApi):
|
||||
ENTRYPOINT: Final[str] = "v3/files/"
|
||||
FALLBACK_MIMETYPE: Final[str] = "application/octet-stream"
|
||||
|
||||
@retry_on_expired_token
|
||||
def list(self, folder_id: str, page_num: int) -> FilesApiListResponse:
|
||||
url = self.ENTRYPOINT
|
||||
token_check(self.connection)
|
||||
@@ -38,6 +40,7 @@ class FilesApi(BaseApi):
|
||||
self._raise_response_error(response)
|
||||
return TypeAdapter(FilesApiListResponse).validate_python(response.json())
|
||||
|
||||
@retry_on_expired_token
|
||||
def retrieve(self, id: str) -> File:
|
||||
# print(self.__class__.__name__ + "::" + sys._getframe().f_code.co_name)
|
||||
url = self.ENTRYPOINT + id + "/"
|
||||
@@ -46,6 +49,7 @@ class FilesApi(BaseApi):
|
||||
self._raise_response_error(response)
|
||||
return TypeAdapter(File).validate_python(response.json())
|
||||
|
||||
@retry_on_expired_token
|
||||
def create(self, folder_id: str, path: str) -> str:
|
||||
# print(self.__class__.__name__ + "::" + sys._getframe().f_code.co_name)
|
||||
url = self.ENTRYPOINT
|
||||
@@ -66,10 +70,14 @@ class FilesApi(BaseApi):
|
||||
raise UnexpectedException(f"Could not open `{path}` file.")
|
||||
except MemoryError:
|
||||
raise UnexpectedException("Out of memory.")
|
||||
except MDRSException:
|
||||
# Already says what went wrong, and the caller may want to act on the kind.
|
||||
raise
|
||||
except Exception as e:
|
||||
raise UnexpectedException("Unspecified error.") from e
|
||||
return ret.id
|
||||
|
||||
@retry_on_expired_token
|
||||
def update(self, file: File, path: str | None) -> bool:
|
||||
# print(self.__class__.__name__ + "::" + sys._getframe().f_code.co_name)
|
||||
url = self.ENTRYPOINT + file.id + "/"
|
||||
@@ -96,6 +104,7 @@ class FilesApi(BaseApi):
|
||||
self._raise_response_error(response)
|
||||
return True
|
||||
|
||||
@retry_on_expired_token
|
||||
def destroy(self, file: File) -> bool:
|
||||
# print(self.__class__.__name__ + "::" + sys._getframe().f_code.co_name)
|
||||
url = self.ENTRYPOINT + file.id + "/"
|
||||
@@ -104,6 +113,7 @@ class FilesApi(BaseApi):
|
||||
self._raise_response_error(response)
|
||||
return True
|
||||
|
||||
@retry_on_expired_token
|
||||
def move(self, file: File, folder_id: str, name: str) -> bool:
|
||||
# print(self.__class__.__name__ + "::" + sys._getframe().f_code.co_name)
|
||||
url = self.ENTRYPOINT + file.id + "/move/"
|
||||
@@ -113,6 +123,7 @@ class FilesApi(BaseApi):
|
||||
self._raise_response_error(response)
|
||||
return True
|
||||
|
||||
@retry_on_expired_token
|
||||
def copy(self, file: File, folder_id: str, name: str) -> bool:
|
||||
# print(self.__class__.__name__ + "::" + sys._getframe().f_code.co_name)
|
||||
url = self.ENTRYPOINT + file.id + "/copy/"
|
||||
@@ -122,6 +133,7 @@ class FilesApi(BaseApi):
|
||||
self._raise_response_error(response)
|
||||
return True
|
||||
|
||||
@retry_on_expired_token
|
||||
def metadata(self, file: File) -> dict[str, Any]:
|
||||
# print(self.__class__.__name__ + "::" + sys._getframe().f_code.co_name)
|
||||
url = self.ENTRYPOINT + file.id + "/metadata/"
|
||||
@@ -130,22 +142,58 @@ class FilesApi(BaseApi):
|
||||
self._raise_response_error(response)
|
||||
return response.json()
|
||||
|
||||
@retry_on_expired_token
|
||||
def download(self, file: File, path: str) -> bool:
|
||||
# print(self.__class__.__name__ + "::" + sys._getframe().f_code.co_name)
|
||||
url = file.download_url
|
||||
token_check(self.connection)
|
||||
# Refused before anything is fetched. The finished file is moved into place, and a
|
||||
# rename would replace a destination whose mode says it is protected.
|
||||
if os.path.exists(path):
|
||||
try:
|
||||
with open(path, "r+b"):
|
||||
pass
|
||||
except OSError as e:
|
||||
raise UnexpectedException(f"Cannot write `{path}`: {e}")
|
||||
response = self.connection.get(url, stream=True)
|
||||
self._raise_response_error(response)
|
||||
# Written beside the destination and moved in once the whole body has arrived, so
|
||||
# a transfer that fails part way leaves whatever was already there untouched and
|
||||
# never leaves a truncated file under the real name.
|
||||
fd, tmp_path = self._open_partial(path)
|
||||
try:
|
||||
with open(path, "wb") as f:
|
||||
with os.fdopen(fd, "wb") as f:
|
||||
for chunk in response.iter_content(chunk_size=4096):
|
||||
if chunk:
|
||||
f.write(chunk)
|
||||
f.flush()
|
||||
except PermissionError:
|
||||
print(f"Cannot create file `{path}`: Permission denied.")
|
||||
os.replace(tmp_path, path)
|
||||
except BaseException:
|
||||
# Only the scratch file goes: anything at the destination was not written here.
|
||||
if os.path.exists(tmp_path):
|
||||
os.unlink(tmp_path)
|
||||
raise
|
||||
return True
|
||||
|
||||
@staticmethod
|
||||
def _open_partial(path: str) -> tuple[int, str]:
|
||||
"""
|
||||
Create a scratch file beside `path` and return it open for writing.
|
||||
|
||||
Beside it, so moving the finished download into place is a rename within one
|
||||
directory. `0o666` rather than a private mode because the umask is what decided
|
||||
the permissions of a downloaded file before, and still should.
|
||||
"""
|
||||
base = f"{path}.{os.getpid()}-{threading.get_ident()}"
|
||||
for attempt in range(100):
|
||||
tmp_path = f"{base}-{attempt}.mdrspart"
|
||||
try:
|
||||
return os.open(tmp_path, os.O_CREAT | os.O_EXCL | os.O_WRONLY, 0o666), tmp_path
|
||||
except FileExistsError:
|
||||
continue
|
||||
except OSError as e:
|
||||
raise UnexpectedException(f"Cannot write `{path}`: {e}")
|
||||
raise UnexpectedException(f"Could not create a temporary file beside `{path}`.")
|
||||
|
||||
def _get_mime_type(self, path: str) -> str:
|
||||
mt = mimetypes.guess_type(path)
|
||||
if mt:
|
||||
|
||||
@@ -5,7 +5,7 @@ from pydantic import TypeAdapter
|
||||
from pydantic.dataclasses import dataclass
|
||||
|
||||
from mdrsclient.api.base import BaseApi
|
||||
from mdrsclient.api.utils import token_check
|
||||
from mdrsclient.api.utils import retry_on_expired_token, token_check
|
||||
from mdrsclient.exceptions import UnauthorizedException
|
||||
from mdrsclient.models import Folder, FolderSimple
|
||||
|
||||
@@ -18,6 +18,7 @@ class FoldersApiCreateResponse:
|
||||
class FoldersApi(BaseApi):
|
||||
ENTRYPOINT: Final[str] = "v3/folders/"
|
||||
|
||||
@retry_on_expired_token
|
||||
def list(self, laboratory_id: int, path: str) -> list[FolderSimple]:
|
||||
# print(self.__class__.__name__ + "::" + sys._getframe().f_code.co_name)
|
||||
url = self.ENTRYPOINT
|
||||
@@ -30,6 +31,7 @@ class FoldersApi(BaseApi):
|
||||
ret.append(TypeAdapter(FolderSimple).validate_python(data))
|
||||
return ret
|
||||
|
||||
@retry_on_expired_token
|
||||
def retrieve(self, id: str) -> Folder:
|
||||
# print(self.__class__.__name__ + "::" + sys._getframe().f_code.co_name)
|
||||
url = self.ENTRYPOINT + id + "/"
|
||||
@@ -39,6 +41,7 @@ class FoldersApi(BaseApi):
|
||||
ret = TypeAdapter(Folder).validate_python(response.json())
|
||||
return ret
|
||||
|
||||
@retry_on_expired_token
|
||||
def create(self, name: str, parent_id: str) -> str:
|
||||
# print(self.__class__.__name__ + "::" + sys._getframe().f_code.co_name)
|
||||
url = self.ENTRYPOINT
|
||||
@@ -49,6 +52,7 @@ class FoldersApi(BaseApi):
|
||||
ret = TypeAdapter(FoldersApiCreateResponse).validate_python(response.json())
|
||||
return ret.id
|
||||
|
||||
@retry_on_expired_token
|
||||
def update(self, folder: FolderSimple) -> bool:
|
||||
# print(self.__class__.__name__ + "::" + sys._getframe().f_code.co_name)
|
||||
url = self.ENTRYPOINT + folder.id + "/"
|
||||
@@ -61,6 +65,7 @@ class FoldersApi(BaseApi):
|
||||
self._raise_response_error(response)
|
||||
return True
|
||||
|
||||
@retry_on_expired_token
|
||||
def destroy(self, id: str, recursive: bool) -> bool:
|
||||
# print(self.__class__.__name__ + "::" + sys._getframe().f_code.co_name)
|
||||
url = self.ENTRYPOINT + id + "/"
|
||||
@@ -70,6 +75,7 @@ class FoldersApi(BaseApi):
|
||||
self._raise_response_error(response)
|
||||
return True
|
||||
|
||||
@retry_on_expired_token
|
||||
def auth(self, id: str, password: str) -> bool:
|
||||
# print(self.__class__.__name__ + "::" + sys._getframe().f_code.co_name)
|
||||
url = self.ENTRYPOINT + id + "/auth/"
|
||||
@@ -81,6 +87,7 @@ class FoldersApi(BaseApi):
|
||||
self._raise_response_error(response)
|
||||
return True
|
||||
|
||||
@retry_on_expired_token
|
||||
def acl(self, id: str, access_level: int, recursive: bool, password: str | None) -> bool:
|
||||
# print(self.__class__.__name__ + "::" + sys._getframe().f_code.co_name)
|
||||
url = self.ENTRYPOINT + id + "/acl/"
|
||||
@@ -94,6 +101,7 @@ class FoldersApi(BaseApi):
|
||||
self._raise_response_error(response)
|
||||
return True
|
||||
|
||||
@retry_on_expired_token
|
||||
def move(self, folder: FolderSimple, folder_id: str, name: str) -> bool:
|
||||
# print(self.__class__.__name__ + "::" + sys._getframe().f_code.co_name)
|
||||
url = self.ENTRYPOINT + folder.id + "/move/"
|
||||
@@ -103,6 +111,7 @@ class FoldersApi(BaseApi):
|
||||
self._raise_response_error(response)
|
||||
return True
|
||||
|
||||
@retry_on_expired_token
|
||||
def copy(self, folder: FolderSimple, folder_id: str, name: str) -> bool:
|
||||
# print(self.__class__.__name__ + "::" + sys._getframe().f_code.co_name)
|
||||
url = self.ENTRYPOINT + folder.id + "/copy/"
|
||||
@@ -112,6 +121,7 @@ class FoldersApi(BaseApi):
|
||||
self._raise_response_error(response)
|
||||
return True
|
||||
|
||||
@retry_on_expired_token
|
||||
def metadata(self, id: str) -> dict[str, Any]:
|
||||
# print(self.__class__.__name__ + "::" + sys._getframe().f_code.co_name)
|
||||
url = self.ENTRYPOINT + id + "/metadata/"
|
||||
|
||||
@@ -3,13 +3,14 @@ from typing import Final
|
||||
from pydantic import TypeAdapter
|
||||
|
||||
from mdrsclient.api.base import BaseApi
|
||||
from mdrsclient.api.utils import token_check
|
||||
from mdrsclient.api.utils import retry_on_expired_token, token_check
|
||||
from mdrsclient.models import Laboratories, Laboratory
|
||||
|
||||
|
||||
class LaboratoriesApi(BaseApi):
|
||||
ENTRYPOINT: Final[str] = "v3/laboratories/"
|
||||
|
||||
@retry_on_expired_token
|
||||
def list(self) -> Laboratories:
|
||||
# print(self.__class__.__name__ + "::" + sys._getframe().f_code.co_name)
|
||||
url = self.ENTRYPOINT
|
||||
|
||||
@@ -8,6 +8,10 @@ from mdrsclient.api.base import BaseApi
|
||||
from mdrsclient.exceptions import UnauthorizedException
|
||||
from mdrsclient.models import Token, User
|
||||
|
||||
# (connect, read) seconds for the token refresh. Uploads are served by a separate
|
||||
# instance, so this one is not queued behind them and has no reason to be slow.
|
||||
TOKEN_REFRESH_TIMEOUT: Final[tuple[float, float]] = (5.0, 30.0)
|
||||
|
||||
|
||||
@dataclass(frozen=True)
|
||||
class UsersCurrentResponseLaboratory:
|
||||
@@ -61,7 +65,10 @@ class UsersApi(BaseApi):
|
||||
# print(self.__class__.__name__ + "::" + sys._getframe().f_code.co_name)
|
||||
url = self.ENTRYPOINT + "token/refresh/"
|
||||
data: dict[str, str | int] = {"refresh": token.refresh}
|
||||
response = self.connection.post(url, data=data)
|
||||
# Bounded on its own: the caller holds a lock that spans processes while this runs,
|
||||
# so a provider that accepts the connection and then goes quiet would otherwise
|
||||
# stall every other request on this machine rather than just this one.
|
||||
response = self.connection.post(url, data=data, timeout=TOKEN_REFRESH_TIMEOUT)
|
||||
if response.status_code == requests.codes.unauthorized:
|
||||
raise UnauthorizedException("Token is invalid or expired.")
|
||||
self._raise_response_error(response)
|
||||
|
||||
+118
-13
@@ -1,19 +1,124 @@
|
||||
import functools
|
||||
from typing import Any, Callable, TypeVar, cast
|
||||
|
||||
from mdrsclient.api.users import UsersApi
|
||||
from mdrsclient.connection import MDRSConnection
|
||||
from mdrsclient.exceptions import UnauthorizedException
|
||||
from mdrsclient.exceptions import (
|
||||
MDRSException,
|
||||
ServerBusyException,
|
||||
TokenExpiredException,
|
||||
UnauthorizedException,
|
||||
)
|
||||
from mdrsclient.models import Token
|
||||
|
||||
F = TypeVar("F", bound=Callable[..., Any])
|
||||
|
||||
SERVER_BUSY_MESSAGE = (
|
||||
"The server took too long to start handling the request and may be overloaded. "
|
||||
"Try again, or reduce the number of parallel transfers."
|
||||
)
|
||||
|
||||
|
||||
def token_check(connection: MDRSConnection) -> None:
|
||||
try:
|
||||
connection.lock.acquire()
|
||||
if connection.token is not None:
|
||||
if connection.token.is_refresh_required:
|
||||
user_api = UsersApi(connection)
|
||||
try:
|
||||
connection.token = user_api.tokenRefresh(connection.token)
|
||||
except UnauthorizedException:
|
||||
connection.logout()
|
||||
elif connection.token.is_expired:
|
||||
"""
|
||||
Bring the access token up to date before a request goes out.
|
||||
|
||||
Refreshing is a read-modify-write over a cache shared with every other client
|
||||
process on this machine, and a rotating provider stops honouring the refresh token
|
||||
it replaces. Two processes reaching this at once would otherwise both send the same
|
||||
token, and the loser would be left holding one the server no longer accepts, so the
|
||||
whole sequence runs under a lock that spans processes and the cache is re-read
|
||||
inside it.
|
||||
"""
|
||||
with connection.lock:
|
||||
token = connection.token
|
||||
if token is None or not (token.is_refresh_required or token.is_expired):
|
||||
# Nothing to do, which is the answer for almost every request. The lock below
|
||||
# reaches across processes and is held for a round trip, so it is worth
|
||||
# knowing that before taking it.
|
||||
return
|
||||
with connection.cache_lock():
|
||||
connection.reload_cache()
|
||||
token = connection.token
|
||||
if token is None:
|
||||
return
|
||||
if token.is_expired:
|
||||
connection.logout()
|
||||
finally:
|
||||
connection.lock.release()
|
||||
return
|
||||
if not token.is_refresh_required:
|
||||
return
|
||||
user_api = UsersApi(connection)
|
||||
try:
|
||||
connection.token = user_api.tokenRefresh(token)
|
||||
except UnauthorizedException:
|
||||
connection.logout()
|
||||
|
||||
|
||||
def token_recover(connection: MDRSConnection, used: Token | None) -> bool:
|
||||
"""
|
||||
Get a usable access token after one was refused for having expired.
|
||||
|
||||
A request can sit in the server's queue for longer than its access token lives, and
|
||||
while it waited another thread or process may already have refreshed. Prefer what
|
||||
they left behind: refreshing again would spend a round trip, and with a provider
|
||||
that rotates refresh tokens it would retire one that is still in use.
|
||||
|
||||
Returns whether the caller now holds a token worth retrying with.
|
||||
"""
|
||||
with connection.lock, connection.cache_lock():
|
||||
connection.reload_cache()
|
||||
token = connection.token
|
||||
if token is None:
|
||||
return False
|
||||
if used is None or token != used:
|
||||
# Somebody else has been here since the request went out. Assign rather than
|
||||
# just read: the setter is what rewrites the session's Authorization header,
|
||||
# so a plain reload would leave the retry carrying the refused token.
|
||||
connection.token = token
|
||||
return True
|
||||
user_api = UsersApi(connection)
|
||||
try:
|
||||
connection.token = user_api.tokenRefresh(token)
|
||||
except UnauthorizedException:
|
||||
connection.logout()
|
||||
return False
|
||||
return True
|
||||
|
||||
|
||||
def retry_on_expired_token(func: F) -> F:
|
||||
"""
|
||||
Send a request again once when the server found its access token expired.
|
||||
|
||||
The token is checked before every request, so this only happens when the request was
|
||||
held long enough on the way in for a valid token to lapse - a queue behind uploads
|
||||
that take minutes, most often.
|
||||
"""
|
||||
|
||||
@functools.wraps(func)
|
||||
def wrapper(self: Any, *args: Any, **kwargs: Any) -> Any:
|
||||
connection: MDRSConnection = self.connection
|
||||
# Settle the token first so `used` is what the request actually carries. The
|
||||
# wrapped method checks it again, which costs nothing once it is already current.
|
||||
token_check(connection)
|
||||
used = connection.token
|
||||
try:
|
||||
return func(self, *args, **kwargs)
|
||||
except TokenExpiredException:
|
||||
try:
|
||||
recovered = token_recover(connection, used)
|
||||
except MDRSException:
|
||||
raise
|
||||
except Exception as e:
|
||||
# A busy server queues the refresh too, and failing here would report an
|
||||
# endpoint the user never asked for rather than the reason.
|
||||
raise ServerBusyException(SERVER_BUSY_MESSAGE) from e
|
||||
if not recovered:
|
||||
raise
|
||||
try:
|
||||
return func(self, *args, **kwargs)
|
||||
except TokenExpiredException as e:
|
||||
# A second refusal, with a token that was current when it left. The session is
|
||||
# fine; the server is not starting requests before their credentials lapse.
|
||||
raise ServerBusyException(SERVER_BUSY_MESSAGE) from e
|
||||
|
||||
return cast(F, wrapper)
|
||||
|
||||
+108
-20
@@ -1,8 +1,12 @@
|
||||
import contextlib
|
||||
import dataclasses
|
||||
import hashlib
|
||||
import json
|
||||
import os
|
||||
from typing import Protocol, runtime_checkable
|
||||
import tempfile
|
||||
import time
|
||||
from contextlib import AbstractContextManager
|
||||
from typing import Iterator, Protocol, runtime_checkable
|
||||
|
||||
from pydantic import TypeAdapter, ValidationError
|
||||
from pydantic.dataclasses import dataclass
|
||||
@@ -46,6 +50,19 @@ class CacheData:
|
||||
|
||||
@runtime_checkable
|
||||
class CacheInterface(Protocol):
|
||||
def lock(self) -> AbstractContextManager[None]:
|
||||
"""
|
||||
Hold exclusive access to the cache for the duration of the block.
|
||||
|
||||
Refreshing a token is a read-modify-write, and the cache is shared with every
|
||||
other client process using the same remote.
|
||||
"""
|
||||
...
|
||||
|
||||
def reload(self) -> None:
|
||||
"""Re-read the cache, discarding anything held from an earlier read."""
|
||||
...
|
||||
|
||||
@property
|
||||
def token(self) -> Token | None: ...
|
||||
@token.setter
|
||||
@@ -70,6 +87,14 @@ class InMemoryCache(CacheInterface):
|
||||
def __init__(self) -> None:
|
||||
self.__data = CacheData()
|
||||
|
||||
@contextlib.contextmanager
|
||||
def lock(self) -> Iterator[None]:
|
||||
# Nothing else can reach this cache, so there is nothing to exclude.
|
||||
yield
|
||||
|
||||
def reload(self) -> None:
|
||||
pass
|
||||
|
||||
@property
|
||||
def token(self) -> Token | None:
|
||||
return self.__data.token
|
||||
@@ -109,14 +134,46 @@ class CacheFile(CacheInterface):
|
||||
__serial: int
|
||||
__cache_dir: str
|
||||
__cache_file: str
|
||||
__lock_file: str
|
||||
__lock_depth: int
|
||||
__data: CacheData
|
||||
|
||||
def __init__(self, remote: str) -> None:
|
||||
self.__serial = -1
|
||||
self.__cache_dir = os.path.join(CONFIG_DIRNAME, "cache")
|
||||
self.__cache_file = os.path.join(self.__cache_dir, remote + ".json")
|
||||
self.__lock_file = os.path.join(self.__cache_dir, remote + ".lock")
|
||||
self.__lock_depth = 0
|
||||
self.__data = CacheData()
|
||||
|
||||
@contextlib.contextmanager
|
||||
def lock(self) -> Iterator[None]:
|
||||
# Re-entrant, because every write takes it and a refresh is a write made while
|
||||
# already holding it. A second flock on the same file from the same process would
|
||||
# wait for a release that cannot come.
|
||||
if self.__lock_depth > 0:
|
||||
self.__lock_depth += 1
|
||||
try:
|
||||
yield
|
||||
finally:
|
||||
self.__lock_depth -= 1
|
||||
return
|
||||
# A separate file, so that replacing the cache cannot disturb the lock holders.
|
||||
self.__ensure_cache_dir()
|
||||
with open(self.__lock_file, "a") as f:
|
||||
FileLock.lock(f)
|
||||
self.__lock_depth = 1
|
||||
try:
|
||||
yield
|
||||
finally:
|
||||
self.__lock_depth = 0
|
||||
FileLock.unlock(f)
|
||||
os.chmod(self.__lock_file, 0o600)
|
||||
|
||||
def reload(self) -> None:
|
||||
self.__serial = -1
|
||||
self.__load()
|
||||
|
||||
@property
|
||||
def token(self) -> Token | None:
|
||||
self.__load()
|
||||
@@ -124,9 +181,10 @@ class CacheFile(CacheInterface):
|
||||
|
||||
@token.setter
|
||||
def token(self, token: Token) -> None:
|
||||
self.__load()
|
||||
self.__data.token = token
|
||||
self.__save()
|
||||
with self.lock():
|
||||
self.reload()
|
||||
self.__data.token = token
|
||||
self.__save()
|
||||
|
||||
@token.deleter
|
||||
def token(self) -> None:
|
||||
@@ -139,9 +197,10 @@ class CacheFile(CacheInterface):
|
||||
|
||||
@user.setter
|
||||
def user(self, user: User) -> None:
|
||||
self.__load()
|
||||
self.__data.user = user
|
||||
self.__save()
|
||||
with self.lock():
|
||||
self.reload()
|
||||
self.__data.user = user
|
||||
self.__save()
|
||||
|
||||
@user.deleter
|
||||
def user(self) -> None:
|
||||
@@ -154,9 +213,10 @@ class CacheFile(CacheInterface):
|
||||
|
||||
@laboratories.setter
|
||||
def laboratories(self, laboratories: Laboratories) -> None:
|
||||
self.__load()
|
||||
self.__data.laboratories = laboratories
|
||||
self.__save()
|
||||
with self.lock():
|
||||
self.reload()
|
||||
self.__data.laboratories = laboratories
|
||||
self.__save()
|
||||
|
||||
def __clear(self) -> None:
|
||||
self.__data.clear()
|
||||
@@ -164,8 +224,7 @@ class CacheFile(CacheInterface):
|
||||
|
||||
def __load(self) -> None:
|
||||
if os.path.isfile(self.__cache_file):
|
||||
stat = os.stat(self.__cache_file)
|
||||
serial = hash((stat.st_uid, stat.st_gid, stat.st_mode, stat.st_size, stat.st_mtime))
|
||||
serial = self.__stat_serial()
|
||||
if self.__serial != serial:
|
||||
try:
|
||||
with open(self.__cache_file) as f:
|
||||
@@ -185,15 +244,44 @@ class CacheFile(CacheInterface):
|
||||
|
||||
def __save(self) -> None:
|
||||
self.__ensure_cache_dir()
|
||||
with open(self.__cache_file, "w") as f:
|
||||
FileLock.lock(f)
|
||||
self.__data.update_digest()
|
||||
f.write(json.dumps(dataclasses.asdict(self.__data)))
|
||||
FileLock.unlock(f)
|
||||
self.__data.update_digest()
|
||||
payload = json.dumps(dataclasses.asdict(self.__data))
|
||||
# Written aside and moved into place: opening the cache for writing truncates it
|
||||
# first, and a reader landing in that window would find the file empty and treat
|
||||
# the session as broken.
|
||||
fd, tmp_file = tempfile.mkstemp(dir=self.__cache_dir, prefix=".tmp-")
|
||||
try:
|
||||
with os.fdopen(fd, "w") as f:
|
||||
f.write(payload)
|
||||
# ensure file is secure.
|
||||
os.chmod(tmp_file, 0o600)
|
||||
self.__replace(tmp_file, self.__cache_file)
|
||||
except BaseException:
|
||||
if os.path.exists(tmp_file):
|
||||
os.unlink(tmp_file)
|
||||
raise
|
||||
self.__serial = self.__stat_serial()
|
||||
|
||||
@staticmethod
|
||||
def __replace(source: str, destination: str) -> None:
|
||||
# Windows refuses the rename while another process still has the destination
|
||||
# open, which a reader briefly does, so give it a moment rather than failing the
|
||||
# save outright. On POSIX the rename always succeeds and the loop ends at once.
|
||||
deadline = time.monotonic() + 5.0
|
||||
while True:
|
||||
try:
|
||||
os.replace(source, destination)
|
||||
return
|
||||
except PermissionError:
|
||||
if time.monotonic() >= deadline:
|
||||
raise
|
||||
time.sleep(0.05)
|
||||
|
||||
def __stat_serial(self) -> int:
|
||||
stat = os.stat(self.__cache_file)
|
||||
self.__serial = hash((stat.st_uid, stat.st_gid, stat.st_mode, stat.st_size, stat.st_mtime))
|
||||
# ensure file is secure.
|
||||
os.chmod(self.__cache_file, 0o600)
|
||||
# st_ino and st_mtime_ns both move when the file is replaced, which a refresh
|
||||
# that happens to produce the same number of bytes otherwise would not show.
|
||||
return hash((stat.st_ino, stat.st_uid, stat.st_gid, stat.st_mode, stat.st_size, stat.st_mtime_ns))
|
||||
|
||||
def __ensure_cache_dir(self) -> None:
|
||||
if not os.path.exists(self.__cache_dir):
|
||||
|
||||
+94
-80
@@ -1,21 +1,40 @@
|
||||
import os
|
||||
from typing import Any
|
||||
from dataclasses import dataclass
|
||||
from unicodedata import normalize
|
||||
|
||||
from mdrsclient.api import DoiApi, FilesApi, FoldersApi, LaboratoriesApi, UsersApi
|
||||
from mdrsclient.api import FilesApi, FoldersApi
|
||||
from mdrsclient.cache import CacheInterface
|
||||
from mdrsclient.config import ConfigInterface
|
||||
from mdrsclient.connection import MDRSConnection
|
||||
from mdrsclient.exceptions import IllegalArgumentException, MDRSException, UnauthorizedException, UnexpectedException
|
||||
from mdrsclient.models import File, Folder, Laboratory, Token, User
|
||||
from mdrsclient.exceptions import IllegalArgumentException
|
||||
from mdrsclient.models import File, Folder, FolderSimple, Laboratory
|
||||
from mdrsclient.models.file import find_file
|
||||
from mdrsclient.services import MdrsService
|
||||
|
||||
|
||||
@dataclass(frozen=True)
|
||||
class TransferEndpoints:
|
||||
"""Both ends of a `cp` or `mv`, resolved: the parents, their contents, and the names."""
|
||||
|
||||
laboratory: Laboratory
|
||||
s_path: str
|
||||
s_basename: str
|
||||
s_parent_folder: Folder
|
||||
s_parent_files: list[File]
|
||||
d_path: str
|
||||
d_basename: str
|
||||
d_parent_folder: Folder
|
||||
d_parent_files: list[File]
|
||||
|
||||
def is_same_place(self) -> bool:
|
||||
"""Whether the destination names what the source already is, leaving nothing to do."""
|
||||
return self.s_parent_folder.id == self.d_parent_folder.id and self.s_basename == self.d_basename
|
||||
|
||||
|
||||
class MdrsClient(MdrsService):
|
||||
"""Service layer client for MDRS."""
|
||||
|
||||
def __init__(self, connection: MDRSConnection, config_class: type[ConfigInterface] | None = None):
|
||||
def __init__(self, connection: MDRSConnection | None, config_class: type[ConfigInterface] | None = None):
|
||||
super().__init__(connection, config_class)
|
||||
|
||||
@classmethod
|
||||
@@ -63,7 +82,8 @@ class MdrsClient(MdrsService):
|
||||
files = self.find_files(folder.id)
|
||||
return folder, files
|
||||
|
||||
def cp(self, src_path: str, dest_path: str, is_recursive: bool = False) -> None:
|
||||
def __resolve_transfer(self, src_path: str, dest_path: str) -> TransferEndpoints:
|
||||
"""Resolve both ends of a transfer, refusing one that crosses a remote or a laboratory."""
|
||||
s_remote, s_laboratory_name, s_path = self.parse_remote_host_with_path(src_path)
|
||||
d_remote, d_laboratory_name, d_path = self.parse_remote_host_with_path(dest_path)
|
||||
if s_remote != d_remote:
|
||||
@@ -71,8 +91,9 @@ class MdrsClient(MdrsService):
|
||||
if s_laboratory_name != d_laboratory_name:
|
||||
raise IllegalArgumentException("Laboratory mismatched.")
|
||||
s_path = s_path.rstrip("/")
|
||||
s_dirname = os.path.dirname(s_path)
|
||||
s_basename = os.path.basename(s_path)
|
||||
# A destination ending in a separator names a folder to put the source into, and
|
||||
# keeps the source's own name.
|
||||
if d_path.endswith("/"):
|
||||
d_dirname = d_path
|
||||
d_basename = s_basename
|
||||
@@ -80,84 +101,77 @@ class MdrsClient(MdrsService):
|
||||
d_dirname = os.path.dirname(d_path)
|
||||
d_basename = os.path.basename(d_path)
|
||||
laboratory = self.find_laboratory(s_laboratory_name)
|
||||
s_parent_folder = self.find_folder(laboratory, s_dirname)
|
||||
s_parent_files = self.find_files(s_parent_folder.id)
|
||||
s_parent_folder = self.find_folder(laboratory, os.path.dirname(s_path))
|
||||
d_parent_folder = self.find_folder(laboratory, d_dirname)
|
||||
d_parent_files = self.find_files(d_parent_folder.id)
|
||||
s_file = find_file(s_parent_files, s_basename)
|
||||
return TransferEndpoints(
|
||||
laboratory=laboratory,
|
||||
s_path=s_path,
|
||||
s_basename=s_basename,
|
||||
s_parent_folder=s_parent_folder,
|
||||
s_parent_files=self.find_files(s_parent_folder.id),
|
||||
d_path=d_path,
|
||||
d_basename=d_basename,
|
||||
d_parent_folder=d_parent_folder,
|
||||
d_parent_files=self.find_files(d_parent_folder.id),
|
||||
)
|
||||
|
||||
@staticmethod
|
||||
def __check_file_destination(ends: TransferEndpoints) -> None:
|
||||
"""Refuse a destination that a file cannot take the place of."""
|
||||
if find_file(ends.d_parent_files, ends.d_basename) is not None:
|
||||
raise IllegalArgumentException(f"File `{ends.d_basename}` already exists.")
|
||||
if ends.d_parent_folder.find_sub_folder(ends.d_basename) is not None:
|
||||
raise IllegalArgumentException(
|
||||
f"Cannot overwrite non-folder `{ends.d_basename}` with folder `{ends.d_path}`."
|
||||
)
|
||||
|
||||
@staticmethod
|
||||
def __check_folder_destination(ends: TransferEndpoints, s_folder: FolderSimple) -> None:
|
||||
"""Refuse a destination that a folder cannot take the place of."""
|
||||
if find_file(ends.d_parent_files, ends.d_basename) is not None:
|
||||
raise IllegalArgumentException(
|
||||
f"Cannot overwrite non-folder `{ends.d_basename}` with folder `{ends.s_path}`."
|
||||
)
|
||||
d_folder = ends.d_parent_folder.find_sub_folder(ends.d_basename)
|
||||
if d_folder is not None:
|
||||
if d_folder.id == s_folder.id:
|
||||
raise IllegalArgumentException(f"`{ends.s_path}` and `{ends.s_path}` are the same folder.")
|
||||
raise IllegalArgumentException(f"Cannot move `{ends.s_path}` to `{ends.d_path}`: Folder not empty.")
|
||||
|
||||
@staticmethod
|
||||
def __find_source_folder(ends: TransferEndpoints) -> FolderSimple:
|
||||
s_folder = ends.s_parent_folder.find_sub_folder(ends.s_basename)
|
||||
if s_folder is None:
|
||||
raise IllegalArgumentException(f"File or folder `{ends.s_basename}` not found.")
|
||||
return s_folder
|
||||
|
||||
def cp(self, src_path: str, dest_path: str, is_recursive: bool = False) -> None:
|
||||
ends = self.__resolve_transfer(src_path, dest_path)
|
||||
s_file = find_file(ends.s_parent_files, ends.s_basename)
|
||||
if s_file is not None:
|
||||
d_file = find_file(d_parent_files, d_basename)
|
||||
if d_file is not None:
|
||||
raise IllegalArgumentException(f"File `{d_basename}` already exists.")
|
||||
d_sub_folder = d_parent_folder.find_sub_folder(d_basename)
|
||||
if d_sub_folder is not None:
|
||||
raise IllegalArgumentException(f"Cannot overwrite non-folder `{d_basename}` with folder `{d_path}`.")
|
||||
file_api = FilesApi(self.connection)
|
||||
if s_parent_folder.id != d_parent_folder.id or d_basename != s_basename:
|
||||
file_api.copy(s_file, d_parent_folder.id, normalize("NFC", d_basename))
|
||||
else:
|
||||
s_folder = s_parent_folder.find_sub_folder(s_basename)
|
||||
if s_folder is None:
|
||||
raise IllegalArgumentException(f"File or folder `{s_basename}` not found.")
|
||||
if not is_recursive:
|
||||
raise IllegalArgumentException(f"Cannot copy `{s_path}`: Is a folder.")
|
||||
if find_file(d_parent_files, d_basename) is not None:
|
||||
raise IllegalArgumentException(f"Cannot overwrite non-folder `{d_basename}` with folder `{s_path}`.")
|
||||
d_folder = d_parent_folder.find_sub_folder(d_basename)
|
||||
if d_folder is not None:
|
||||
if d_folder.id == s_folder.id:
|
||||
raise IllegalArgumentException(f"`{s_path}` and `{s_path}` are the same folder.")
|
||||
raise IllegalArgumentException(f"Cannot move `{s_path}` to `{d_path}`: Folder not empty.")
|
||||
folder_api = FoldersApi(self.connection)
|
||||
if s_parent_folder.id != d_parent_folder.id or s_basename != d_basename:
|
||||
folder_api.copy(s_folder, d_parent_folder.id, normalize("NFC", d_basename))
|
||||
self.__check_file_destination(ends)
|
||||
if not ends.is_same_place():
|
||||
FilesApi(self.connection).copy(s_file, ends.d_parent_folder.id, normalize("NFC", ends.d_basename))
|
||||
return
|
||||
s_folder = self.__find_source_folder(ends)
|
||||
if not is_recursive:
|
||||
raise IllegalArgumentException(f"Cannot copy `{ends.s_path}`: Is a folder.")
|
||||
self.__check_folder_destination(ends, s_folder)
|
||||
if not ends.is_same_place():
|
||||
FoldersApi(self.connection).copy(s_folder, ends.d_parent_folder.id, normalize("NFC", ends.d_basename))
|
||||
|
||||
def mv(self, src_path: str, dest_path: str) -> None:
|
||||
s_remote, s_laboratory_name, s_path = self.parse_remote_host_with_path(src_path)
|
||||
d_remote, d_laboratory_name, d_path = self.parse_remote_host_with_path(dest_path)
|
||||
if s_remote != d_remote:
|
||||
raise IllegalArgumentException("Remote host mismatched.")
|
||||
if s_laboratory_name != d_laboratory_name:
|
||||
raise IllegalArgumentException("Laboratory mismatched.")
|
||||
s_path = s_path.rstrip("/")
|
||||
s_dirname = os.path.dirname(s_path)
|
||||
s_basename = os.path.basename(s_path)
|
||||
if d_path.endswith("/"):
|
||||
d_dirname = d_path
|
||||
d_basename = s_basename
|
||||
else:
|
||||
d_dirname = os.path.dirname(d_path)
|
||||
d_basename = os.path.basename(d_path)
|
||||
laboratory = self.find_laboratory(s_laboratory_name)
|
||||
s_parent_folder = self.find_folder(laboratory, s_dirname)
|
||||
s_parent_files = self.find_files(s_parent_folder.id)
|
||||
d_parent_folder = self.find_folder(laboratory, d_dirname)
|
||||
d_parent_files = self.find_files(d_parent_folder.id)
|
||||
s_file = find_file(s_parent_files, s_basename)
|
||||
ends = self.__resolve_transfer(src_path, dest_path)
|
||||
s_file = find_file(ends.s_parent_files, ends.s_basename)
|
||||
if s_file is not None:
|
||||
d_file = find_file(d_parent_files, d_basename)
|
||||
if d_file is not None:
|
||||
raise IllegalArgumentException(f"File `{d_basename}` already exists.")
|
||||
d_sub_folder = d_parent_folder.find_sub_folder(d_basename)
|
||||
if d_sub_folder is not None:
|
||||
raise IllegalArgumentException(f"Cannot overwrite non-folder `{d_basename}` with folder `{d_path}`.")
|
||||
file_api = FilesApi(self.connection)
|
||||
if s_parent_folder.id != d_parent_folder.id or d_basename != s_basename:
|
||||
file_api.move(s_file, d_parent_folder.id, normalize("NFC", d_basename))
|
||||
else:
|
||||
s_folder = s_parent_folder.find_sub_folder(s_basename)
|
||||
if s_folder is None:
|
||||
raise IllegalArgumentException(f"File or folder `{s_basename}` not found.")
|
||||
if find_file(d_parent_files, d_basename) is not None:
|
||||
raise IllegalArgumentException(f"Cannot overwrite non-folder `{d_basename}` with folder `{s_path}`.")
|
||||
d_folder = d_parent_folder.find_sub_folder(d_basename)
|
||||
if d_folder is not None:
|
||||
if d_folder.id == s_folder.id:
|
||||
raise IllegalArgumentException(f"`{s_path}` and `{s_path}` are the same folder.")
|
||||
raise IllegalArgumentException(f"Cannot move `{s_path}` to `{d_path}`: Folder not empty.")
|
||||
folder_api = FoldersApi(self.connection)
|
||||
if s_parent_folder.id != d_parent_folder.id or d_basename != s_basename:
|
||||
folder_api.move(s_folder, d_parent_folder.id, normalize("NFC", d_basename))
|
||||
self.__check_file_destination(ends)
|
||||
if not ends.is_same_place():
|
||||
FilesApi(self.connection).move(s_file, ends.d_parent_folder.id, normalize("NFC", ends.d_basename))
|
||||
return
|
||||
s_folder = self.__find_source_folder(ends)
|
||||
self.__check_folder_destination(ends, s_folder)
|
||||
if not ends.is_same_place():
|
||||
FoldersApi(self.connection).move(s_folder, ends.d_parent_folder.id, normalize("NFC", ends.d_basename))
|
||||
|
||||
def chacl(
|
||||
self, remote_path: str, access_level: int, is_recursive: bool = False, password: str | None = None
|
||||
|
||||
@@ -1,7 +1,6 @@
|
||||
from argparse import Namespace
|
||||
from typing import Any
|
||||
|
||||
from mdrsclient.api import FoldersApi
|
||||
from mdrsclient.commands.base import BaseCommand
|
||||
from mdrsclient.exceptions import IllegalArgumentException
|
||||
from mdrsclient.models import FolderAccessLevel
|
||||
|
||||
@@ -1,12 +1,7 @@
|
||||
import os
|
||||
from argparse import Namespace
|
||||
from typing import Any
|
||||
from unicodedata import normalize
|
||||
|
||||
from mdrsclient.api import FilesApi, FoldersApi
|
||||
from mdrsclient.commands.base import BaseCommand
|
||||
from mdrsclient.exceptions import IllegalArgumentException
|
||||
from mdrsclient.models.file import find_file
|
||||
|
||||
|
||||
class CpCommand(BaseCommand):
|
||||
|
||||
@@ -1,12 +1,8 @@
|
||||
import json
|
||||
import os
|
||||
from argparse import Namespace
|
||||
from typing import Any
|
||||
|
||||
from mdrsclient.api import FilesApi
|
||||
from mdrsclient.commands.base import BaseCommand
|
||||
from mdrsclient.exceptions import IllegalArgumentException
|
||||
from mdrsclient.models.file import find_file
|
||||
|
||||
|
||||
class FileMetadataCommand(BaseCommand):
|
||||
|
||||
@@ -1,7 +1,6 @@
|
||||
from argparse import Namespace
|
||||
from typing import Any
|
||||
|
||||
from mdrsclient.api import LaboratoriesApi
|
||||
from mdrsclient.commands.base import BaseCommand
|
||||
|
||||
|
||||
|
||||
+40
-29
@@ -1,12 +1,13 @@
|
||||
import json
|
||||
from argparse import Namespace
|
||||
from typing import Any
|
||||
from typing import Any, Final
|
||||
|
||||
from pydantic.dataclasses import dataclass
|
||||
|
||||
from mdrsclient.api import FilesApi, FoldersApi
|
||||
from mdrsclient.api import FoldersApi
|
||||
from mdrsclient.client import MdrsClient
|
||||
from mdrsclient.commands.base import BaseCommand
|
||||
from mdrsclient.config import build_download_url
|
||||
from mdrsclient.exceptions import UnauthorizedException
|
||||
from mdrsclient.models import File, Folder, FolderSimple, Laboratory
|
||||
|
||||
@@ -94,25 +95,24 @@ class LsCommand(BaseCommand):
|
||||
def _ls_json(cls, context: LsCommandContext, folder: Folder, files: list[File]) -> None:
|
||||
print(json.dumps(cls._folder2dict(context, folder, files), ensure_ascii=False))
|
||||
|
||||
LABELS: Final[dict[str, str]] = {
|
||||
"type": "Type",
|
||||
"acl": "Access",
|
||||
"laboratory": "Laboratory",
|
||||
"size": "Size",
|
||||
"date": "Date",
|
||||
"name": "Name",
|
||||
}
|
||||
|
||||
@classmethod
|
||||
def _ls_plain(cls, context: LsCommandContext, folder: Folder, files: list[File]) -> None:
|
||||
label = {
|
||||
"type": "Type",
|
||||
"acl": "Access",
|
||||
"laboratory": "Laboratory",
|
||||
"size": "Size",
|
||||
"date": "Date",
|
||||
"name": "Name",
|
||||
}
|
||||
length: dict[str, int] = {}
|
||||
for key in label.keys():
|
||||
length[key] = len(label[key]) if not context.is_quiet else 0
|
||||
def _column_widths(cls, context: LsCommandContext, folder: Folder, files: list[File]) -> dict[str, int]:
|
||||
"""Width of each column: the widest of its heading and everything printed under it."""
|
||||
length = {key: len(label) if not context.is_quiet else 0 for key, label in cls.LABELS.items()}
|
||||
for sub_folder in folder.sub_folders:
|
||||
sub_laboratory = context.client.connection.laboratories.find_by_id(sub_folder.laboratory_id)
|
||||
sub_laboratory_name = sub_laboratory.name if sub_laboratory is not None else "(invalid)"
|
||||
sub_laboratory_name = cls._laboratory_name(context, sub_folder.laboratory_id)
|
||||
length["acl"] = max(length["acl"], len(sub_folder.access_level_name))
|
||||
length["laboratory"] = max(length["laboratory"], len(sub_laboratory_name))
|
||||
length["size"] = max(length["size"], len(str(folder.size)))
|
||||
length["size"] = max(length["size"], len(str(sub_folder.size)))
|
||||
length["date"] = max(length["date"], len(sub_folder.updated_at_name))
|
||||
length["name"] = max(length["name"], len(sub_folder.name))
|
||||
for file in files:
|
||||
@@ -121,6 +121,27 @@ class LsCommand(BaseCommand):
|
||||
length["name"] = max(length["name"], len(file.name))
|
||||
length["acl"] = max(length["acl"], len(folder.access_level_name))
|
||||
length["laboratory"] = max(length["laboratory"], len(context.laboratory.name))
|
||||
return length
|
||||
|
||||
@classmethod
|
||||
def _ls_plain_children(cls, context: LsCommandContext, folder: Folder) -> None:
|
||||
"""List each sub-folder in turn, passing over the ones the caller cannot open."""
|
||||
print("")
|
||||
folder_api = FoldersApi(context.client.connection)
|
||||
for sub_folder in sorted(folder.sub_folders, key=lambda x: x.name):
|
||||
try:
|
||||
if sub_folder.lock:
|
||||
folder_api.auth(sub_folder.id, context.password)
|
||||
sub_detail = folder_api.retrieve(sub_folder.id)
|
||||
sub_files = context.client.find_files(sub_folder.id)
|
||||
except UnauthorizedException:
|
||||
continue
|
||||
cls._ls_plain(context, sub_detail, sub_files)
|
||||
|
||||
@classmethod
|
||||
def _ls_plain(cls, context: LsCommandContext, folder: Folder, files: list[File]) -> None:
|
||||
label = cls.LABELS
|
||||
length = cls._column_widths(context, folder, files)
|
||||
header = (
|
||||
f"{label['type']:{length['type']}}\t{label['acl']:{length['acl']}}\t"
|
||||
f"{label['laboratory']:{length['laboratory']}}\t{label['size']:{length['size']}}\t"
|
||||
@@ -151,17 +172,7 @@ class LsCommand(BaseCommand):
|
||||
)
|
||||
|
||||
if context.is_recursive:
|
||||
print("")
|
||||
for sub_folder in sorted(folder.sub_folders, key=lambda x: x.name):
|
||||
folder_api = FoldersApi(context.client.connection)
|
||||
try:
|
||||
if sub_folder.lock:
|
||||
folder_api.auth(sub_folder.id, context.password)
|
||||
folder = folder_api.retrieve(sub_folder.id)
|
||||
files = context.client.find_files(sub_folder.id)
|
||||
cls._ls_plain(context, folder, files)
|
||||
except UnauthorizedException:
|
||||
pass
|
||||
cls._ls_plain_children(context, folder)
|
||||
|
||||
@classmethod
|
||||
def _folder2dict(
|
||||
@@ -211,7 +222,7 @@ class LsCommand(BaseCommand):
|
||||
# "thumbnail": file.thumbnail,
|
||||
"description": file.description,
|
||||
"metadata": file.metadata,
|
||||
"download_url": f"{context.client.connection.url}/{file.download_url}",
|
||||
"download_url": build_download_url(context.client.connection.url, file.download_url),
|
||||
"created_at": file.created_at,
|
||||
"updated_at": file.updated_at,
|
||||
}
|
||||
|
||||
@@ -2,7 +2,6 @@ import json
|
||||
from argparse import Namespace
|
||||
from typing import Any
|
||||
|
||||
from mdrsclient.api import FoldersApi
|
||||
from mdrsclient.commands.base import BaseCommand
|
||||
|
||||
|
||||
|
||||
@@ -1,12 +1,7 @@
|
||||
import os
|
||||
from argparse import Namespace
|
||||
from typing import Any
|
||||
from unicodedata import normalize
|
||||
|
||||
from mdrsclient.api import FoldersApi
|
||||
from mdrsclient.commands.base import BaseCommand
|
||||
from mdrsclient.exceptions import IllegalArgumentException
|
||||
from mdrsclient.models.file import find_file
|
||||
|
||||
|
||||
class MkdirCommand(BaseCommand):
|
||||
|
||||
@@ -1,12 +1,7 @@
|
||||
import os
|
||||
from argparse import Namespace
|
||||
from typing import Any
|
||||
from unicodedata import normalize
|
||||
|
||||
from mdrsclient.api import FilesApi, FoldersApi
|
||||
from mdrsclient.commands.base import BaseCommand
|
||||
from mdrsclient.exceptions import IllegalArgumentException
|
||||
from mdrsclient.models.file import find_file
|
||||
|
||||
|
||||
class MvCommand(BaseCommand):
|
||||
|
||||
@@ -1,11 +1,7 @@
|
||||
import os
|
||||
from argparse import Namespace
|
||||
from typing import Any
|
||||
|
||||
from mdrsclient.api import FilesApi, FoldersApi
|
||||
from mdrsclient.commands.base import BaseCommand
|
||||
from mdrsclient.exceptions import IllegalArgumentException
|
||||
from mdrsclient.models.file import find_file
|
||||
|
||||
|
||||
class RmCommand(BaseCommand):
|
||||
|
||||
@@ -1,7 +1,6 @@
|
||||
from argparse import Namespace
|
||||
from typing import Any
|
||||
|
||||
from mdrsclient.__version__ import __version__
|
||||
from mdrsclient.commands.base import BaseCommand
|
||||
|
||||
|
||||
|
||||
+36
-5
@@ -1,7 +1,7 @@
|
||||
import configparser
|
||||
import os
|
||||
import threading
|
||||
from typing import Final, Protocol, runtime_checkable
|
||||
from typing import TYPE_CHECKING, Final, Protocol, runtime_checkable
|
||||
|
||||
import validators
|
||||
|
||||
@@ -10,10 +10,43 @@ from mdrsclient.settings import CONFIG_DIRNAME
|
||||
from mdrsclient.utils import FileLock
|
||||
|
||||
|
||||
def normalize_url(url: str) -> str:
|
||||
"""
|
||||
Check a remote URL and put it in the one form every client agrees on.
|
||||
|
||||
`simple_host` is what lets a bare hostname through, so a development server on
|
||||
`localhost` is as acceptable as a deployment behind a domain name. The trailing
|
||||
slash goes because the URL is joined with a path that brings its own.
|
||||
"""
|
||||
if not validators.url(url, simple_host=True, validate_scheme=lambda scheme: scheme in ("http", "https")):
|
||||
raise IllegalArgumentException("malformed URI sequence")
|
||||
return url.rstrip("/")
|
||||
|
||||
|
||||
def build_download_url(base_url: str | None, path: str) -> str:
|
||||
"""
|
||||
Join a remote base URL with a path the API returned.
|
||||
|
||||
The API answers with a relative path and no leading separator, and a configuration
|
||||
written before the URL was normalised may still carry a trailing one.
|
||||
"""
|
||||
if path.startswith(("http://", "https://")):
|
||||
return path
|
||||
return f"{(base_url or '').rstrip('/')}/{path.lstrip('/')}"
|
||||
|
||||
|
||||
@runtime_checkable
|
||||
class ConfigInterface(Protocol):
|
||||
remote: str
|
||||
|
||||
if TYPE_CHECKING:
|
||||
# Declared for the type checker, because the class itself is passed around and
|
||||
# called: without this it describes something that takes no arguments to build.
|
||||
# Kept out of the running class, because a protocol that defines `__init__` loses
|
||||
# the guard that stops it being instantiated, and hands that empty `__init__` to
|
||||
# any implementation that does not write its own.
|
||||
def __init__(self, remote: str) -> None: ...
|
||||
|
||||
def list(self) -> list[tuple[str, str]]: ...
|
||||
@property
|
||||
def url(self) -> str | None: ...
|
||||
@@ -42,8 +75,7 @@ class InMemoryConfig(ConfigInterface):
|
||||
|
||||
@url.setter
|
||||
def url(self, url: str) -> None:
|
||||
if not validators.url(url):
|
||||
raise IllegalArgumentException("malformed URI sequence")
|
||||
url = normalize_url(url)
|
||||
with self.__lock:
|
||||
self.__configs[self.remote] = url
|
||||
|
||||
@@ -91,8 +123,7 @@ class ConfigFile(ConfigInterface):
|
||||
|
||||
@url.setter
|
||||
def url(self, url: str) -> None:
|
||||
if not validators.url(url):
|
||||
raise IllegalArgumentException("malformed URI sequence")
|
||||
url = normalize_url(url)
|
||||
self.__load()
|
||||
if self.__config.has_section(self.remote):
|
||||
self.__config.remove_section(self.remote)
|
||||
|
||||
@@ -1,5 +1,6 @@
|
||||
import platform
|
||||
import threading
|
||||
from contextlib import AbstractContextManager
|
||||
from typing import TypedDict
|
||||
|
||||
from requests import Response, Session
|
||||
@@ -23,6 +24,7 @@ class _KwArgsMDRSConnectionPost(TypedDict, total=False):
|
||||
params: dict[str, str | int]
|
||||
data: dict[str, str | int] | MultipartEncoder
|
||||
headers: dict[str, str]
|
||||
timeout: float | tuple[float, float]
|
||||
|
||||
|
||||
class _KwArgsMDRSConnectionPut(TypedDict, total=False):
|
||||
@@ -61,6 +63,14 @@ class MDRSConnection:
|
||||
def delete(self, url: str, **kwargs: Unpack[_KwArgsMDRSConnectionDelete]) -> Response:
|
||||
return self.session.delete(self.__build_url(url), **kwargs)
|
||||
|
||||
def cache_lock(self) -> AbstractContextManager[None]:
|
||||
"""Hold exclusive access to the login cache across every process using it."""
|
||||
return self.__cache.lock()
|
||||
|
||||
def reload_cache(self) -> None:
|
||||
"""Re-read the login cache, discarding anything held from an earlier read."""
|
||||
self.__cache.reload()
|
||||
|
||||
def logout(self) -> None:
|
||||
del self.__cache.user
|
||||
del self.__cache.token
|
||||
@@ -96,7 +106,9 @@ class MDRSConnection:
|
||||
return path
|
||||
if self.url == "":
|
||||
raise MissingConfigurationException("remote host is not configured")
|
||||
return f"{self.url}/{path}"
|
||||
# The path brings its own separator, and a configuration written before the URL
|
||||
# was normalised may still carry a trailing slash of its own.
|
||||
return f"{self.url.rstrip('/')}/{path}"
|
||||
|
||||
def __prepare_headers(self) -> None:
|
||||
self.session.headers.update(
|
||||
|
||||
@@ -28,6 +28,18 @@ class UnauthorizedException(MDRSException):
|
||||
pass
|
||||
|
||||
|
||||
class TokenExpiredException(UnauthorizedException):
|
||||
"""Thrown when the access token was still valid when sent but had expired on arrival"""
|
||||
|
||||
pass
|
||||
|
||||
|
||||
class ServerBusyException(MDRSException):
|
||||
"""Thrown when the server did not start handling a request before its token lapsed"""
|
||||
|
||||
pass
|
||||
|
||||
|
||||
class ForbiddenException(MDRSException):
|
||||
"""Thrown when the current user does not have enough privileges to access the resource"""
|
||||
|
||||
|
||||
+22
-4
@@ -1,6 +1,5 @@
|
||||
import os
|
||||
import re
|
||||
from typing import Any
|
||||
from unicodedata import normalize
|
||||
|
||||
from mdrsclient.api import DoiApi, FilesApi, FoldersApi, LaboratoriesApi, UsersApi
|
||||
@@ -20,11 +19,30 @@ from mdrsclient.utils import page_num_from_url
|
||||
class MdrsService:
|
||||
config_class: type[ConfigInterface] = ConfigFile
|
||||
|
||||
def __init__(self, connection: MDRSConnection, config_class: type[ConfigInterface] | None = None):
|
||||
self.connection = connection
|
||||
def __init__(self, connection: MDRSConnection | None, config_class: type[ConfigInterface] | None = None):
|
||||
self.__connection = connection
|
||||
if config_class is not None:
|
||||
self.config_class = config_class
|
||||
|
||||
@property
|
||||
def connection(self) -> MDRSConnection:
|
||||
"""
|
||||
The connection every remote operation goes through.
|
||||
|
||||
Optional to supply, because `config` and `version` do their work without a remote
|
||||
and are reached through the same client. Asking for it when none was given is a
|
||||
mistake in the caller, and says so rather than failing later on `None`.
|
||||
"""
|
||||
if self.__connection is None:
|
||||
raise MissingConfigurationException("This operation requires a remote host.")
|
||||
return self.__connection
|
||||
|
||||
@connection.setter
|
||||
def connection(self, connection: MDRSConnection | None) -> None:
|
||||
# Assignable as it always was: this is a property to check for absence on read,
|
||||
# not to make the connection fixed once the client is built.
|
||||
self.__connection = connection
|
||||
|
||||
@classmethod
|
||||
def create_connection(
|
||||
cls, remote: str, cache: CacheInterface | None = None, config: ConfigInterface | None = None
|
||||
@@ -182,7 +200,7 @@ class MdrsService:
|
||||
if folder.lock:
|
||||
if password is None:
|
||||
raise UnauthorizedException(f"Folder for DOI `{doi_clean}` is locked.")
|
||||
folder_api.auth(doi_resp.folder.id, password)
|
||||
folder_api.auth(doi_resp.folder_id, password)
|
||||
|
||||
lab_api = LaboratoriesApi(self.connection)
|
||||
labs = lab_api.list()
|
||||
|
||||
+197
-139
@@ -6,7 +6,7 @@ from unicodedata import normalize
|
||||
from pydantic.dataclasses import dataclass
|
||||
|
||||
from mdrsclient.api import FilesApi, FoldersApi
|
||||
from mdrsclient.exceptions import IllegalArgumentException, MDRSException, UnexpectedException
|
||||
from mdrsclient.exceptions import IllegalArgumentException, UnexpectedException
|
||||
from mdrsclient.models import File, Folder, Laboratory
|
||||
from mdrsclient.models.file import find_file
|
||||
from mdrsclient.settings import CONCURRENT
|
||||
@@ -27,7 +27,6 @@ class DownloadFileInfo:
|
||||
|
||||
@dataclass
|
||||
class DownloadContext:
|
||||
hasError: bool
|
||||
isSkipIfExists: bool
|
||||
files: list[DownloadFileInfo]
|
||||
|
||||
@@ -47,54 +46,67 @@ class Uploader:
|
||||
laboratory = self.client.find_laboratory(laboratory_name)
|
||||
folder = self.client.find_folder(laboratory, r_path)
|
||||
files = self.client.find_files(folder.id)
|
||||
infos: list[UploadFileInfo] = []
|
||||
if os.path.isdir(l_path):
|
||||
if not is_recursive:
|
||||
raise IllegalArgumentException(f"Cannot upload `{local_path}`: Is a directory.")
|
||||
folder_api = FoldersApi(self.client.connection)
|
||||
folder_map: dict[str, Folder] = {}
|
||||
folder_map[r_path] = folder
|
||||
files_map: dict[str, list[File]] = {}
|
||||
files_map[r_path] = files
|
||||
l_basename = os.path.basename(l_path)
|
||||
for dirpath, _, filenames in os.walk(l_path, followlinks=True):
|
||||
sub = l_basename if dirpath == l_path else os.path.join(l_basename, os.path.relpath(dirpath, l_path))
|
||||
d_dirname = os.path.join(r_path, sub)
|
||||
d_basename = os.path.basename(d_dirname)
|
||||
# prepare destination parent path
|
||||
d_parent_dirname = os.path.dirname(d_dirname)
|
||||
if folder_map.get(d_parent_dirname) is None:
|
||||
parent_folder = self.client.find_folder(laboratory, d_parent_dirname)
|
||||
folder_map[d_parent_dirname] = parent_folder
|
||||
parent_files = self.client.find_files(parent_folder.id)
|
||||
files_map[d_parent_dirname] = parent_files
|
||||
# prepare destination path
|
||||
if folder_map.get(d_dirname) is None:
|
||||
d_folder = folder_map[d_parent_dirname].find_sub_folder(d_basename)
|
||||
if d_folder is None:
|
||||
d_folder_id = folder_api.create(normalize("NFC", d_basename), folder_map[d_parent_dirname].id)
|
||||
else:
|
||||
d_folder_id = d_folder.id
|
||||
print(d_dirname)
|
||||
folder_map[d_dirname] = folder_api.retrieve(d_folder_id)
|
||||
files_map[d_dirname] = self.client.find_files(d_folder_id)
|
||||
if d_folder is None:
|
||||
folder_map[d_parent_dirname].sub_folders.append(folder_map[d_dirname])
|
||||
# register upload file list
|
||||
for filename in filenames:
|
||||
infos.append(
|
||||
UploadFileInfo(folder_map[d_dirname], files_map[d_dirname], os.path.join(dirpath, filename))
|
||||
)
|
||||
infos = self.__collect_directory_uploads(laboratory, r_path, l_path, folder, files)
|
||||
else:
|
||||
infos.append(UploadFileInfo(folder, files, l_path))
|
||||
self.__multiple_upload(infos, is_skip_if_exists)
|
||||
infos = [UploadFileInfo(folder, files, l_path)]
|
||||
if not self.__multiple_upload(infos, is_skip_if_exists):
|
||||
# One file failing is worth reporting on its own line, and worth the caller
|
||||
# hearing about: a batch that lost files is not a batch that succeeded.
|
||||
raise UnexpectedException("Some files failed to upload.")
|
||||
|
||||
def __multiple_upload(self, infos: list[UploadFileInfo], is_skip_if_exists: bool) -> None:
|
||||
def __collect_directory_uploads(
|
||||
self, laboratory: Laboratory, r_path: str, l_path: str, folder: Folder, files: list[File]
|
||||
) -> list[UploadFileInfo]:
|
||||
"""Mirror a local directory tree on the remote, and list the files to send into it."""
|
||||
infos: list[UploadFileInfo] = []
|
||||
folder_api = FoldersApi(self.client.connection)
|
||||
folder_map: dict[str, Folder] = {}
|
||||
folder_map[r_path] = folder
|
||||
files_map: dict[str, list[File]] = {}
|
||||
files_map[r_path] = files
|
||||
l_basename = os.path.basename(l_path)
|
||||
for dirpath, _, filenames in os.walk(l_path, followlinks=True):
|
||||
sub = l_basename if dirpath == l_path else os.path.join(l_basename, os.path.relpath(dirpath, l_path))
|
||||
d_dirname = os.path.join(r_path, sub)
|
||||
d_basename = os.path.basename(d_dirname)
|
||||
# prepare destination parent path
|
||||
d_parent_dirname = os.path.dirname(d_dirname)
|
||||
if folder_map.get(d_parent_dirname) is None:
|
||||
parent_folder = self.client.find_folder(laboratory, d_parent_dirname)
|
||||
folder_map[d_parent_dirname] = parent_folder
|
||||
parent_files = self.client.find_files(parent_folder.id)
|
||||
files_map[d_parent_dirname] = parent_files
|
||||
# prepare destination path
|
||||
if folder_map.get(d_dirname) is None:
|
||||
d_folder = folder_map[d_parent_dirname].find_sub_folder(d_basename)
|
||||
if d_folder is None:
|
||||
d_folder_id = folder_api.create(normalize("NFC", d_basename), folder_map[d_parent_dirname].id)
|
||||
else:
|
||||
d_folder_id = d_folder.id
|
||||
print(d_dirname)
|
||||
folder_map[d_dirname] = folder_api.retrieve(d_folder_id)
|
||||
files_map[d_dirname] = self.client.find_files(d_folder_id)
|
||||
if d_folder is None:
|
||||
folder_map[d_parent_dirname].sub_folders.append(folder_map[d_dirname])
|
||||
# register upload file list
|
||||
for filename in filenames:
|
||||
infos.append(
|
||||
UploadFileInfo(folder_map[d_dirname], files_map[d_dirname], os.path.join(dirpath, filename))
|
||||
)
|
||||
return infos
|
||||
|
||||
def __multiple_upload(self, infos: list[UploadFileInfo], is_skip_if_exists: bool) -> bool:
|
||||
"""Send every file, and report whether all of them arrived."""
|
||||
file_api = FilesApi(self.client.connection)
|
||||
with ThreadPoolExecutor(max_workers=CONCURRENT) as pool:
|
||||
pool.map(lambda x: self.__multiple_upload_worker(file_api, x, is_skip_if_exists), infos)
|
||||
results = pool.map(lambda x: self.__multiple_upload_worker(file_api, x, is_skip_if_exists), infos)
|
||||
# Consumed inside the block: the results are what carry each worker's verdict.
|
||||
return all(list(results))
|
||||
|
||||
def __multiple_upload_worker(self, file_api: FilesApi, info: UploadFileInfo, is_skip_if_exists: bool) -> None:
|
||||
def __multiple_upload_worker(self, file_api: FilesApi, info: UploadFileInfo, is_skip_if_exists: bool) -> bool:
|
||||
basename = os.path.basename(info.path)
|
||||
file = find_file(info.files, basename)
|
||||
try:
|
||||
@@ -103,8 +115,14 @@ class Uploader:
|
||||
elif not is_skip_if_exists or file.size != os.path.getsize(info.path):
|
||||
file_api.update(file, info.path)
|
||||
print(os.path.join(info.folder.path, basename))
|
||||
except MDRSException as e:
|
||||
print(f"Error: {e}")
|
||||
except Exception as e:
|
||||
# Everything, not just the exceptions the API layer raises: the batch verdict
|
||||
# is read now that the results are consumed, and a file vanishing between the
|
||||
# walk and the upload would otherwise end the whole run with a traceback and
|
||||
# throw away what every other file did.
|
||||
print(f"Failed: {info.path}: {e}")
|
||||
return False
|
||||
return True
|
||||
|
||||
|
||||
class Downloader:
|
||||
@@ -120,90 +138,117 @@ class Downloader:
|
||||
password: str | None = None,
|
||||
excludes: list[str] | None = None,
|
||||
) -> None:
|
||||
if not self.__download(remote_path, local_path, is_recursive, is_skip_if_exists, password, excludes):
|
||||
# Every failure has already been printed against the file it belongs to.
|
||||
# This is what makes the command as a whole end in failure.
|
||||
raise UnexpectedException("Some files failed to download.")
|
||||
|
||||
def __download(
|
||||
self,
|
||||
remote_path: str,
|
||||
local_path: str,
|
||||
is_recursive: bool,
|
||||
is_skip_if_exists: bool,
|
||||
password: str | None,
|
||||
excludes: list[str] | None,
|
||||
) -> bool:
|
||||
"""Fetch what the remote path names, and report whether every file arrived."""
|
||||
excludes_clean = excludes or []
|
||||
# Detect DOI path: "remote:10.xxxx/prefix.ID[/optional/sub/path]"
|
||||
path_component = remote_path.split(":", 1)[1] if ":" in remote_path else ""
|
||||
if self.client.is_doi(path_component):
|
||||
remote, doi, subpath = self.client.parse_doi_remote_host(remote_path)
|
||||
|
||||
l_dirname = os.path.realpath(local_path)
|
||||
if not os.path.isdir(l_dirname):
|
||||
raise IllegalArgumentException(f"Local directory `{local_path}` not found.")
|
||||
doi_folder, laboratory = self.client.find_folder_by_doi(doi, password)
|
||||
|
||||
subpath_clean = subpath.rstrip("/")
|
||||
if not subpath_clean:
|
||||
folder = doi_folder
|
||||
is_folder = True
|
||||
else:
|
||||
r_dirname = os.path.dirname(subpath_clean)
|
||||
r_basename = os.path.basename(subpath_clean)
|
||||
abs_path = doi_folder.path.rstrip("/") + r_dirname
|
||||
r_parent_folder = self.client.find_folder(laboratory, abs_path, password)
|
||||
r_parent_files = self.client.find_files(r_parent_folder.id)
|
||||
file = find_file(r_parent_files, r_basename)
|
||||
if file is not None:
|
||||
if self.__check_excludes(excludes_clean, laboratory, r_parent_folder, file):
|
||||
return
|
||||
context = DownloadContext(False, is_skip_if_exists, [])
|
||||
l_path = os.path.join(l_dirname, r_basename)
|
||||
context.files.append(DownloadFileInfo(file, l_path))
|
||||
self.__multiple_download(context)
|
||||
return
|
||||
else:
|
||||
folder_simple = r_parent_folder.find_sub_folder(r_basename)
|
||||
if folder_simple is None:
|
||||
raise IllegalArgumentException(f"File or folder `{subpath_clean}` not found.")
|
||||
folder = FoldersApi(self.client.connection).retrieve(folder_simple.id)
|
||||
is_folder = True
|
||||
|
||||
# For a DOI target the whole folder is the download target.
|
||||
if not is_recursive:
|
||||
# Non-recursive: download only the files at the top level of the DOI folder.
|
||||
files = self.client.find_files(folder.id)
|
||||
context = DownloadContext(False, is_skip_if_exists, [])
|
||||
for file in files:
|
||||
if self.__check_excludes(excludes_clean, laboratory, folder, file):
|
||||
continue
|
||||
l_path = os.path.join(l_dirname, file.name)
|
||||
context.files.append(DownloadFileInfo(file, l_path))
|
||||
self.__multiple_download(context)
|
||||
return
|
||||
folder_api = FoldersApi(self.client.connection)
|
||||
self.__multiple_download_pickup_recursive_files(
|
||||
folder_api, laboratory, folder.id, l_dirname, excludes_clean, is_skip_if_exists
|
||||
)
|
||||
return
|
||||
|
||||
remote, laboratory_name, r_path = self.client.parse_remote_host_with_path(remote_path)
|
||||
r_path = r_path.rstrip("/")
|
||||
r_dirname = os.path.dirname(r_path)
|
||||
r_basename = os.path.basename(r_path)
|
||||
|
||||
l_dirname = os.path.realpath(local_path)
|
||||
if not os.path.isdir(l_dirname):
|
||||
raise IllegalArgumentException(f"Local directory `{local_path}` not found.")
|
||||
|
||||
# "remote:10.xxxx/prefix.ID[/optional/sub/path]" names a published dataset rather
|
||||
# than a path within a laboratory, and is resolved through the DOI instead.
|
||||
path_component = remote_path.split(":", 1)[1] if ":" in remote_path else ""
|
||||
if self.client.is_doi(path_component):
|
||||
return self.__download_doi(
|
||||
remote_path, l_dirname, is_recursive, is_skip_if_exists, password, excludes_clean
|
||||
)
|
||||
|
||||
remote, laboratory_name, r_path = self.client.parse_remote_host_with_path(remote_path)
|
||||
r_path = r_path.rstrip("/")
|
||||
r_basename = os.path.basename(r_path)
|
||||
laboratory = self.client.find_laboratory(laboratory_name)
|
||||
r_parent_folder = self.client.find_folder(laboratory, r_dirname, password)
|
||||
r_parent_folder = self.client.find_folder(laboratory, os.path.dirname(r_path), password)
|
||||
r_parent_files = self.client.find_files(r_parent_folder.id)
|
||||
file = find_file(r_parent_files, r_basename)
|
||||
if file is not None:
|
||||
if self.__check_excludes(excludes_clean, laboratory, r_parent_folder, file):
|
||||
return
|
||||
context = DownloadContext(False, is_skip_if_exists, [])
|
||||
l_path = os.path.join(l_dirname, r_basename)
|
||||
context.files.append(DownloadFileInfo(file, l_path))
|
||||
self.__multiple_download(context)
|
||||
else:
|
||||
folder = r_parent_folder.find_sub_folder(r_basename)
|
||||
if folder is None:
|
||||
raise IllegalArgumentException(f"File or folder `{r_path}` not found.")
|
||||
if not is_recursive:
|
||||
raise IllegalArgumentException(f"Cannot download `{r_path}`: Is a folder.")
|
||||
folder_api = FoldersApi(self.client.connection)
|
||||
self.__multiple_download_pickup_recursive_files(
|
||||
folder_api, laboratory, folder.id, l_dirname, excludes_clean, is_skip_if_exists
|
||||
return self.__download_one(
|
||||
excludes_clean, laboratory, r_parent_folder, file, l_dirname, r_basename, is_skip_if_exists
|
||||
)
|
||||
folder = r_parent_folder.find_sub_folder(r_basename)
|
||||
if folder is None:
|
||||
raise IllegalArgumentException(f"File or folder `{r_path}` not found.")
|
||||
if not is_recursive:
|
||||
raise IllegalArgumentException(f"Cannot download `{r_path}`: Is a folder.")
|
||||
return self.__multiple_download_pickup_recursive_files(
|
||||
FoldersApi(self.client.connection), laboratory, folder.id, l_dirname, excludes_clean, is_skip_if_exists
|
||||
)
|
||||
|
||||
def __download_doi(
|
||||
self,
|
||||
remote_path: str,
|
||||
l_dirname: str,
|
||||
is_recursive: bool,
|
||||
is_skip_if_exists: bool,
|
||||
password: str | None,
|
||||
excludes: list[str],
|
||||
) -> bool:
|
||||
"""Fetch what a DOI names: the dataset's folder, or something inside it."""
|
||||
remote, doi, subpath = self.client.parse_doi_remote_host(remote_path)
|
||||
doi_folder, laboratory = self.client.find_folder_by_doi(doi, password)
|
||||
|
||||
subpath_clean = subpath.rstrip("/")
|
||||
if not subpath_clean:
|
||||
folder = doi_folder
|
||||
else:
|
||||
r_basename = os.path.basename(subpath_clean)
|
||||
abs_path = doi_folder.path.rstrip("/") + os.path.dirname(subpath_clean)
|
||||
r_parent_folder = self.client.find_folder(laboratory, abs_path, password)
|
||||
file = find_file(self.client.find_files(r_parent_folder.id), r_basename)
|
||||
if file is not None:
|
||||
return self.__download_one(
|
||||
excludes, laboratory, r_parent_folder, file, l_dirname, r_basename, is_skip_if_exists
|
||||
)
|
||||
folder_simple = r_parent_folder.find_sub_folder(r_basename)
|
||||
if folder_simple is None:
|
||||
raise IllegalArgumentException(f"File or folder `{subpath_clean}` not found.")
|
||||
folder = FoldersApi(self.client.connection).retrieve(folder_simple.id)
|
||||
|
||||
if is_recursive:
|
||||
return self.__multiple_download_pickup_recursive_files(
|
||||
FoldersApi(self.client.connection), laboratory, folder.id, l_dirname, excludes, is_skip_if_exists
|
||||
)
|
||||
# Without -r the dataset's own files are fetched, and its sub-folders are not.
|
||||
context = DownloadContext(is_skip_if_exists, [])
|
||||
for file in self.client.find_files(folder.id):
|
||||
if self.__check_excludes(excludes, laboratory, folder, file):
|
||||
continue
|
||||
context.files.append(DownloadFileInfo(file, os.path.join(l_dirname, file.name)))
|
||||
return self.__multiple_download(context)
|
||||
|
||||
def __download_one(
|
||||
self,
|
||||
excludes: list[str],
|
||||
laboratory: Laboratory,
|
||||
folder: Folder,
|
||||
file: File,
|
||||
l_dirname: str,
|
||||
local_name: str,
|
||||
is_skip_if_exists: bool,
|
||||
) -> bool:
|
||||
"""
|
||||
Fetch a single named file into the local directory.
|
||||
|
||||
Saved under the name the caller asked for rather than the one the server holds:
|
||||
the two are matched case-insensitively, so they need not be spelled alike.
|
||||
"""
|
||||
if self.__check_excludes(excludes, laboratory, folder, file):
|
||||
return True
|
||||
context = DownloadContext(is_skip_if_exists, [])
|
||||
context.files.append(DownloadFileInfo(file, os.path.join(l_dirname, local_name)))
|
||||
return self.__multiple_download(context)
|
||||
|
||||
def __multiple_download_pickup_recursive_files(
|
||||
self,
|
||||
@@ -213,47 +258,60 @@ class Downloader:
|
||||
basedir: str,
|
||||
excludes: list[str],
|
||||
is_skip_if_exists: bool,
|
||||
) -> None:
|
||||
context = DownloadContext(False, is_skip_if_exists, [])
|
||||
folder = folder_api.retrieve(folder_id)
|
||||
files = self.client.find_files(folder.id)
|
||||
) -> bool:
|
||||
context = DownloadContext(is_skip_if_exists, [])
|
||||
try:
|
||||
folder = folder_api.retrieve(folder_id)
|
||||
files = self.client.find_files(folder.id)
|
||||
except Exception as e:
|
||||
print(f"Failed: {basedir}: {e}")
|
||||
return False
|
||||
dirname = os.path.join(basedir, folder.name)
|
||||
if self.__check_excludes(excludes, laboratory, folder, None):
|
||||
return
|
||||
if not os.path.exists(dirname):
|
||||
os.makedirs(dirname)
|
||||
return True
|
||||
try:
|
||||
# `exist_ok` rather than a prior check: two workers can reach the same parent.
|
||||
os.makedirs(dirname, exist_ok=True)
|
||||
except OSError as e:
|
||||
# One folder the client cannot make locally is not a reason to abandon its
|
||||
# siblings, which is what this walk now promises.
|
||||
print(f"Failed: {dirname}: {e}")
|
||||
return False
|
||||
print(dirname)
|
||||
for file in files:
|
||||
if self.__check_excludes(excludes, laboratory, folder, file):
|
||||
continue
|
||||
path = os.path.join(dirname, file.name)
|
||||
context.files.append(DownloadFileInfo(file, path))
|
||||
self.__multiple_download(context)
|
||||
if context.hasError:
|
||||
raise UnexpectedException("Some files failed to download.")
|
||||
succeeded = self.__multiple_download(context)
|
||||
# A folder that lost a file is still a folder whose sub-folders the user asked
|
||||
# for, so the walk carries on and the verdict is collected for the caller.
|
||||
for sub_folder in folder.sub_folders:
|
||||
self.__multiple_download_pickup_recursive_files(
|
||||
if not self.__multiple_download_pickup_recursive_files(
|
||||
folder_api, laboratory, sub_folder.id, dirname, excludes, is_skip_if_exists
|
||||
)
|
||||
):
|
||||
succeeded = False
|
||||
return succeeded
|
||||
|
||||
def __multiple_download(self, context: DownloadContext) -> None:
|
||||
def __multiple_download(self, context: DownloadContext) -> bool:
|
||||
"""Fetch every file in the batch, and report whether all of them arrived."""
|
||||
file_api = FilesApi(self.client.connection)
|
||||
with ThreadPoolExecutor(max_workers=CONCURRENT) as pool:
|
||||
results = pool.map(
|
||||
lambda x: self.__multiple_download_worker(file_api, x, context.isSkipIfExists), context.files
|
||||
)
|
||||
hasError = next(filter(lambda x: x is False, results), None)
|
||||
if hasError is not None:
|
||||
context.hasError = True
|
||||
# Consumed inside the block, and in full: every worker's verdict counts, not
|
||||
# just the first refusal.
|
||||
return all(list(results))
|
||||
|
||||
def __multiple_download_worker(self, file_api: FilesApi, info: DownloadFileInfo, is_skip_if_exists: bool) -> bool:
|
||||
if not is_skip_if_exists or not os.path.exists(info.path) or info.file.size != os.path.getsize(info.path):
|
||||
try:
|
||||
file_api.download(info.file, info.path)
|
||||
except Exception:
|
||||
print(f"Failed: {info.path}")
|
||||
if os.path.isfile(info.path):
|
||||
os.remove(info.path)
|
||||
except Exception as e:
|
||||
# Nothing to clear up: a failed transfer writes only to its own scratch
|
||||
# file beside the destination, and removes that itself.
|
||||
print(f"Failed: {info.path}: {e}")
|
||||
return False
|
||||
print(info.path)
|
||||
return True
|
||||
|
||||
+16
-2
@@ -1,5 +1,6 @@
|
||||
import os
|
||||
from typing import IO, Any
|
||||
import time
|
||||
from typing import IO, Any, Final
|
||||
from urllib.parse import parse_qs, urlparse
|
||||
|
||||
if os.name == "nt":
|
||||
@@ -9,10 +10,23 @@ elif os.name == "posix":
|
||||
|
||||
|
||||
class FileLock:
|
||||
# Long enough to outlast a token refresh, which is what the lock is held across.
|
||||
WAIT_SECONDS: Final[float] = 60.0
|
||||
|
||||
@staticmethod
|
||||
def lock(file: IO[Any]) -> None:
|
||||
if os.name == "nt":
|
||||
msvcrt.locking(file.fileno(), msvcrt.LK_LOCK, 1)
|
||||
# msvcrt.LK_LOCK gives up after ten one-second attempts, which is shorter
|
||||
# than the refresh it now has to wait for, so do the waiting here instead.
|
||||
deadline = time.monotonic() + FileLock.WAIT_SECONDS
|
||||
while True:
|
||||
try:
|
||||
msvcrt.locking(file.fileno(), msvcrt.LK_NBLCK, 1)
|
||||
return
|
||||
except OSError:
|
||||
if time.monotonic() >= deadline:
|
||||
raise
|
||||
time.sleep(0.1)
|
||||
elif os.name == "posix":
|
||||
fcntl.flock(file.fileno(), fcntl.LOCK_EX)
|
||||
|
||||
|
||||
+43
-29
@@ -1,11 +1,21 @@
|
||||
[tool.poetry]
|
||||
[build-system]
|
||||
requires = ["flit_core>=4.0.2,<5.0.0"]
|
||||
build-backend = "flit_core.buildapi"
|
||||
|
||||
[tool.flit.module]
|
||||
name = "mdrsclient"
|
||||
|
||||
[project]
|
||||
name = "mdrs-client-python"
|
||||
version = "1.3.18"
|
||||
version = "1.3.19"
|
||||
description = "The mdrs-client-python is python library and a command-line client for up- and downloading files to and from MDRS based repository."
|
||||
authors = ["Yoshihiro OKUMURA <yoshihiro.okumura@riken.jp>"]
|
||||
license = "MIT"
|
||||
authors = [
|
||||
{ name = "Yoshihiro OKUMURA", email = "yoshihiro.okumura@riken.jp" }
|
||||
]
|
||||
license = { text = "MIT" }
|
||||
readme = "README.md"
|
||||
classifiers=[
|
||||
requires-python = ">=3.10"
|
||||
classifiers = [
|
||||
"Development Status :: 4 - Beta",
|
||||
"Environment :: Console",
|
||||
"Intended Audience :: Developers",
|
||||
@@ -18,33 +28,28 @@ classifiers=[
|
||||
"OSI Approved :: MIT License",
|
||||
"Topic :: Utilities",
|
||||
]
|
||||
packages = [
|
||||
{ include = "mdrsclient" }
|
||||
|
||||
dependencies = [
|
||||
"requests>=2.34.2",
|
||||
"requests-toolbelt>=1.0.0",
|
||||
"python-dotenv>=1.2.3",
|
||||
"pydantic>=2.13.5",
|
||||
"pydantic-settings>=2.15.0",
|
||||
"PyJWT>=2.13.0",
|
||||
"validators>=0.35.0",
|
||||
]
|
||||
|
||||
[tool.poetry.dependencies]
|
||||
python = "^3.10"
|
||||
requests = "^2.34.2"
|
||||
requests-toolbelt = "^1.0.0"
|
||||
python-dotenv = "^1.1.0"
|
||||
pydantic = "^2.13.4"
|
||||
pydantic-settings = "^2.14.2"
|
||||
PyJWT = "^2.13.0"
|
||||
validators = "^0.35.0"
|
||||
[project.optional-dependencies]
|
||||
dev = [
|
||||
"black>=26.5.1",
|
||||
"flake8>=7.3.0",
|
||||
"Flake8-pyproject>=1.2.4",
|
||||
"isort>=9.0.1",
|
||||
"pyright>=1.1.411",
|
||||
]
|
||||
|
||||
[tool.poetry.group.dev.dependencies]
|
||||
black = "^26.5.1"
|
||||
flake8 = "^7.2.0"
|
||||
Flake8-pyproject = "^1.2.3"
|
||||
isort = "^8.0.1"
|
||||
pyright = "^1.1.411"
|
||||
|
||||
[tool.poetry.scripts]
|
||||
mdrs = 'mdrsclient.__main__:main'
|
||||
|
||||
[build-system]
|
||||
requires = ["poetry-core"]
|
||||
build-backend = "poetry.core.masonry.api"
|
||||
[project.scripts]
|
||||
mdrs = "mdrsclient.__main__:main"
|
||||
|
||||
[tool.black]
|
||||
line-length = 120
|
||||
@@ -63,6 +68,9 @@ exclude = '''
|
||||
exclude = ".git, .venv, __pycache__, dist"
|
||||
max-complexity = 10
|
||||
max-line-length = 120
|
||||
# E203 flags the space black puts before a slice colon. Black formats this project, so
|
||||
# its output is the standard and flake8 has to accept it.
|
||||
extend-ignore = ["E203"]
|
||||
|
||||
[tool.isort]
|
||||
profile = "black"
|
||||
@@ -72,6 +80,12 @@ line_length = 120
|
||||
|
||||
[tool.pyright]
|
||||
typeCheckingMode = "basic"
|
||||
# This project keeps its environment in `./.venv`. Named explicitly because without it
|
||||
# pyright falls back to the system interpreter, cannot resolve pydantic, and reports every
|
||||
# model field as an unknown argument. An environment kept elsewhere gets that same fallback
|
||||
# and can override it with a local `pyrightconfig.json` or `pyright --pythonpath`.
|
||||
venvPath = "."
|
||||
venv = ".venv"
|
||||
exclude = ["**/__pycache__", "**/.*", "dist"]
|
||||
#reportUnknownMemberType = "warning"
|
||||
#reportUnknownVariableType = "warning"
|
||||
|
||||
@@ -0,0 +1,165 @@
|
||||
import unittest
|
||||
from unittest.mock import MagicMock, patch
|
||||
|
||||
from mdrsclient.client import MdrsClient
|
||||
from mdrsclient.exceptions import IllegalArgumentException
|
||||
from mdrsclient.models import File, Folder, FolderSimple, Laboratories, Laboratory
|
||||
|
||||
TIMESTAMP = "2026-01-01T00:00:00+09:00"
|
||||
LABORATORY = Laboratory(id=1, name="mylab", pi_name="PI", full_name="My Laboratory")
|
||||
|
||||
|
||||
def make_file(id: str, name: str) -> File:
|
||||
return File(
|
||||
id=id,
|
||||
name=name,
|
||||
type="text/plain",
|
||||
size=1,
|
||||
thumbnail=None,
|
||||
description="",
|
||||
metadata={},
|
||||
download_url=f"v3/files/{id}/download/",
|
||||
created_at=TIMESTAMP,
|
||||
updated_at=TIMESTAMP,
|
||||
)
|
||||
|
||||
|
||||
def make_sub_folder(id: str, name: str) -> FolderSimple:
|
||||
return FolderSimple(
|
||||
id=id,
|
||||
pid=None,
|
||||
name=name,
|
||||
access_level=1,
|
||||
lock=False,
|
||||
size=0,
|
||||
laboratory_id=1,
|
||||
description="",
|
||||
created_at=TIMESTAMP,
|
||||
updated_at=TIMESTAMP,
|
||||
restrict_opened_at=None,
|
||||
)
|
||||
|
||||
|
||||
def make_folder(id: str, name: str, path: str, sub_folders: list[FolderSimple] | None = None) -> Folder:
|
||||
return Folder(
|
||||
id=id,
|
||||
pid=None,
|
||||
name=name,
|
||||
access_level=1,
|
||||
lock=False,
|
||||
size=0,
|
||||
laboratory_id=1,
|
||||
description="",
|
||||
created_at=TIMESTAMP,
|
||||
updated_at=TIMESTAMP,
|
||||
restrict_opened_at=None,
|
||||
metadata=[],
|
||||
sub_folders=sub_folders if sub_folders is not None else [],
|
||||
path=path,
|
||||
)
|
||||
|
||||
|
||||
class TestLsColumnWidths(unittest.TestCase):
|
||||
"""Each column is as wide as the widest thing printed under it."""
|
||||
|
||||
def test_the_size_column_fits_a_sub_folder_larger_than_its_parent(self):
|
||||
from unittest.mock import MagicMock as _MagicMock
|
||||
|
||||
from mdrsclient.commands.ls import LsCommand, LsCommandContext
|
||||
|
||||
laboratories = Laboratories()
|
||||
laboratories.append(LABORATORY)
|
||||
connection = _MagicMock()
|
||||
connection.laboratories = laboratories
|
||||
client = MdrsClient(connection)
|
||||
|
||||
big = make_sub_folder("s1", "big")
|
||||
object.__setattr__(big, "size", 123456789)
|
||||
parent = make_folder("f1", "root", "/root/", [big])
|
||||
context = LsCommandContext("remote:/mylab", client, LABORATORY, "", False, False, False)
|
||||
|
||||
widths = LsCommand._column_widths(context, parent, [])
|
||||
|
||||
self.assertEqual(widths["size"], len("123456789"))
|
||||
|
||||
|
||||
class TestCopyAndMove(unittest.TestCase):
|
||||
"""`cp` and `mv` resolve both ends the same way and differ only in what they call."""
|
||||
|
||||
def make_client(self, folders_by_path: dict, files_by_folder: dict) -> MdrsClient:
|
||||
client = MdrsClient(MagicMock())
|
||||
client.find_laboratory = MagicMock(return_value=LABORATORY)
|
||||
client.find_folder = MagicMock(side_effect=lambda lab, path, password=None: folders_by_path[path])
|
||||
client.find_files = MagicMock(side_effect=lambda folder_id: files_by_folder.get(folder_id, []))
|
||||
return client
|
||||
|
||||
def make_tree(self, source_has: str) -> tuple[MdrsClient, dict, dict]:
|
||||
sub = make_sub_folder("fsub", "sub")
|
||||
folder_a = make_folder("fa", "a", "/a/", [sub] if source_has == "folder" else [])
|
||||
folder_b = make_folder("fb", "b", "/b/")
|
||||
folders = {"/a": folder_a, "/b": folder_b, "/b/": folder_b}
|
||||
files = {"fa": [make_file("x1", "data.txt")] if source_has == "file" else [], "fb": []}
|
||||
return self.make_client(folders, files), folders, files
|
||||
|
||||
def test_a_file_is_copied_into_the_destination_folder(self):
|
||||
client, _, files = self.make_tree("file")
|
||||
with patch("mdrsclient.client.FilesApi") as files_api:
|
||||
client.cp("myremote:/mylab/a/data.txt", "myremote:/mylab/b/data.txt")
|
||||
files_api.return_value.copy.assert_called_once_with(files["fa"][0], "fb", "data.txt")
|
||||
|
||||
def test_a_file_is_moved_into_the_destination_folder(self):
|
||||
client, _, files = self.make_tree("file")
|
||||
with patch("mdrsclient.client.FilesApi") as files_api:
|
||||
client.mv("myremote:/mylab/a/data.txt", "myremote:/mylab/b/data.txt")
|
||||
files_api.return_value.move.assert_called_once_with(files["fa"][0], "fb", "data.txt")
|
||||
|
||||
def test_a_trailing_separator_keeps_the_source_name(self):
|
||||
client, _, files = self.make_tree("file")
|
||||
with patch("mdrsclient.client.FilesApi") as files_api:
|
||||
client.mv("myremote:/mylab/a/data.txt", "myremote:/mylab/b/")
|
||||
files_api.return_value.move.assert_called_once_with(files["fa"][0], "fb", "data.txt")
|
||||
|
||||
def test_copying_a_folder_needs_the_recursive_flag(self):
|
||||
client, _, _ = self.make_tree("folder")
|
||||
with patch("mdrsclient.client.FoldersApi") as folders_api:
|
||||
with self.assertRaises(IllegalArgumentException) as caught:
|
||||
client.cp("myremote:/mylab/a/sub", "myremote:/mylab/b/sub")
|
||||
self.assertIn("Is a folder", str(caught.exception))
|
||||
folders_api.return_value.copy.assert_not_called()
|
||||
|
||||
def test_a_folder_is_copied_when_recursive(self):
|
||||
client, folders, _ = self.make_tree("folder")
|
||||
with patch("mdrsclient.client.FoldersApi") as folders_api:
|
||||
client.cp("myremote:/mylab/a/sub", "myremote:/mylab/b/sub", is_recursive=True)
|
||||
folders_api.return_value.copy.assert_called_once_with(folders["/a"].sub_folders[0], "fb", "sub")
|
||||
|
||||
def test_a_folder_is_moved_without_the_recursive_flag(self):
|
||||
client, folders, _ = self.make_tree("folder")
|
||||
with patch("mdrsclient.client.FoldersApi") as folders_api:
|
||||
client.mv("myremote:/mylab/a/sub", "myremote:/mylab/b/sub")
|
||||
folders_api.return_value.move.assert_called_once_with(folders["/a"].sub_folders[0], "fb", "sub")
|
||||
|
||||
def test_a_destination_that_already_holds_the_name_is_refused(self):
|
||||
sub = make_sub_folder("fsub", "sub")
|
||||
folders = {"/a": make_folder("fa", "a", "/a/", [sub]), "/b": make_folder("fb", "b", "/b/")}
|
||||
files = {"fa": [], "fb": [make_file("x2", "sub")]}
|
||||
client = self.make_client(folders, files)
|
||||
with self.assertRaises(IllegalArgumentException) as caught:
|
||||
client.mv("myremote:/mylab/a/sub", "myremote:/mylab/b/sub")
|
||||
self.assertIn("Cannot overwrite non-folder", str(caught.exception))
|
||||
|
||||
def test_a_transfer_across_laboratories_is_refused(self):
|
||||
client, _, _ = self.make_tree("file")
|
||||
with self.assertRaises(IllegalArgumentException) as caught:
|
||||
client.cp("myremote:/mylab/a/data.txt", "myremote:/otherlab/b/data.txt")
|
||||
self.assertIn("Laboratory mismatched", str(caught.exception))
|
||||
|
||||
def test_a_source_that_does_not_exist_is_refused(self):
|
||||
client, _, _ = self.make_tree("file")
|
||||
with self.assertRaises(IllegalArgumentException) as caught:
|
||||
client.mv("myremote:/mylab/a/missing.txt", "myremote:/mylab/b/missing.txt")
|
||||
self.assertIn("not found", str(caught.exception))
|
||||
|
||||
|
||||
if __name__ == "__main__":
|
||||
unittest.main()
|
||||
@@ -0,0 +1,55 @@
|
||||
import unittest
|
||||
from unittest.mock import MagicMock, patch
|
||||
|
||||
from mdrsclient.models import Doi, Folder, Laboratories, Laboratory
|
||||
from mdrsclient.services import MdrsService
|
||||
|
||||
TIMESTAMP = "2026-01-01T00:00:00+09:00"
|
||||
DOI = "10.60178/cbs.20260429-001"
|
||||
LABORATORY = Laboratory(id=1, name="mylab", pi_name="PI", full_name="My Laboratory")
|
||||
|
||||
|
||||
def make_locked_folder() -> Folder:
|
||||
return Folder(
|
||||
id="f1",
|
||||
pid=None,
|
||||
name="root",
|
||||
access_level=4,
|
||||
lock=True,
|
||||
size=0,
|
||||
laboratory_id=1,
|
||||
description="",
|
||||
created_at=TIMESTAMP,
|
||||
updated_at=TIMESTAMP,
|
||||
restrict_opened_at=None,
|
||||
metadata=[],
|
||||
sub_folders=[],
|
||||
path="/root/",
|
||||
)
|
||||
|
||||
|
||||
class TestFindFolderByDoi(unittest.TestCase):
|
||||
"""The DOI response carries the folder id as a field, not as a nested object."""
|
||||
|
||||
def test_a_locked_doi_folder_is_unlocked_with_its_folder_id(self):
|
||||
laboratories = Laboratories()
|
||||
laboratories.append(LABORATORY)
|
||||
|
||||
with (
|
||||
patch("mdrsclient.services.DoiApi") as doi_api,
|
||||
patch("mdrsclient.services.FoldersApi") as folders_api,
|
||||
patch("mdrsclient.services.LaboratoriesApi") as laboratories_api,
|
||||
):
|
||||
doi_api.return_value.retrieve.return_value = Doi(id="20260429-001", doi=DOI, folder_id="f1")
|
||||
folders_api.return_value.retrieve.return_value = make_locked_folder()
|
||||
laboratories_api.return_value.list.return_value = laboratories
|
||||
|
||||
folder, laboratory = MdrsService(MagicMock()).find_folder_by_doi(DOI, "secret")
|
||||
|
||||
folders_api.return_value.auth.assert_called_once_with("f1", "secret")
|
||||
self.assertEqual(folder.id, "f1")
|
||||
self.assertEqual(laboratory.name, "mylab")
|
||||
|
||||
|
||||
if __name__ == "__main__":
|
||||
unittest.main()
|
||||
@@ -0,0 +1,198 @@
|
||||
import tempfile
|
||||
import time
|
||||
import unittest
|
||||
from unittest.mock import patch
|
||||
|
||||
import jwt
|
||||
|
||||
from mdrsclient.api.utils import retry_on_expired_token, token_check, token_recover
|
||||
from mdrsclient.cache import CacheFile
|
||||
from mdrsclient.config import normalize_url
|
||||
from mdrsclient.connection import MDRSConnection
|
||||
from mdrsclient.exceptions import IllegalArgumentException, ServerBusyException, TokenExpiredException
|
||||
from mdrsclient.models import Token
|
||||
|
||||
REMOTE = "unittest"
|
||||
|
||||
|
||||
def make_token(access_offset: int, refresh_offset: int, label: str) -> Token:
|
||||
now = int(time.time())
|
||||
|
||||
def encode(token_type: str, offset: int) -> str:
|
||||
return jwt.encode(
|
||||
{
|
||||
"token_type": token_type,
|
||||
"exp": now + offset,
|
||||
"iat": now,
|
||||
"jti": f"{label}-{token_type}",
|
||||
"user_id": 1,
|
||||
},
|
||||
"unittest-signing-key-not-verified-anywhere",
|
||||
)
|
||||
|
||||
return Token(access=encode("access", access_offset), refresh=encode("refresh", refresh_offset))
|
||||
|
||||
|
||||
class TestTokenCache(unittest.TestCase):
|
||||
def setUp(self) -> None:
|
||||
self.tmp = tempfile.TemporaryDirectory()
|
||||
patcher = patch("mdrsclient.cache.CONFIG_DIRNAME", self.tmp.name)
|
||||
patcher.start()
|
||||
self.addCleanup(patcher.stop)
|
||||
self.addCleanup(self.tmp.cleanup)
|
||||
|
||||
def test_reload_picks_up_a_write_from_another_holder(self) -> None:
|
||||
"""The cache is shared, so a value read earlier can already be out of date."""
|
||||
reader = CacheFile(REMOTE)
|
||||
writer = CacheFile(REMOTE)
|
||||
|
||||
first = make_token(3600, 86400, "first")
|
||||
writer.token = first
|
||||
self.assertEqual(reader.token, first)
|
||||
|
||||
second = make_token(3600, 86400, "second")
|
||||
writer.token = second
|
||||
|
||||
reader.reload()
|
||||
self.assertEqual(reader.token, second)
|
||||
|
||||
def test_token_check_uses_the_token_another_holder_just_wrote(self) -> None:
|
||||
"""A rotating provider drops the token it replaces, so the refresh must not
|
||||
be sent again once someone else has already made the round trip."""
|
||||
connection = MDRSConnection(REMOTE, "http://localhost:8000/api/")
|
||||
connection.token = make_token(-60, 86400, "stale")
|
||||
|
||||
# Another process refreshes while this one is between requests.
|
||||
other = CacheFile(REMOTE)
|
||||
rotated = make_token(3600, 86400, "rotated")
|
||||
other.token = rotated
|
||||
|
||||
with patch("mdrsclient.api.utils.UsersApi") as users_api:
|
||||
token_check(connection)
|
||||
|
||||
users_api.assert_not_called()
|
||||
self.assertEqual(connection.token, rotated)
|
||||
|
||||
def test_token_check_refreshes_when_nothing_else_has(self) -> None:
|
||||
connection = MDRSConnection(REMOTE, "http://localhost:8000/api/")
|
||||
connection.token = make_token(-60, 86400, "stale")
|
||||
rotated = make_token(3600, 86400, "rotated")
|
||||
|
||||
with patch("mdrsclient.api.utils.UsersApi") as users_api:
|
||||
users_api.return_value.tokenRefresh.return_value = rotated
|
||||
token_check(connection)
|
||||
|
||||
users_api.return_value.tokenRefresh.assert_called_once()
|
||||
self.assertEqual(connection.token, rotated)
|
||||
|
||||
def test_token_check_logs_out_once_the_refresh_token_expires(self) -> None:
|
||||
connection = MDRSConnection(REMOTE, "http://localhost:8000/api/")
|
||||
connection.token = make_token(-3600, -60, "dead")
|
||||
|
||||
with patch("mdrsclient.api.utils.UsersApi") as users_api:
|
||||
token_check(connection)
|
||||
|
||||
users_api.assert_not_called()
|
||||
self.assertIsNone(connection.token)
|
||||
|
||||
|
||||
class TestUrlNormalization(unittest.TestCase):
|
||||
"""Both clients share config.ini, so they have to agree on what a remote URL is."""
|
||||
|
||||
def test_trailing_slash_is_dropped(self) -> None:
|
||||
self.assertEqual(normalize_url("http://127.0.0.1:8000/api/"), "http://127.0.0.1:8000/api")
|
||||
self.assertEqual(normalize_url("https://neurodata.riken.jp/api/"), "https://neurodata.riken.jp/api")
|
||||
|
||||
def test_bare_hostname_is_accepted(self) -> None:
|
||||
self.assertEqual(normalize_url("http://localhost:8000/api"), "http://localhost:8000/api")
|
||||
|
||||
def test_only_http_schemes_are_accepted(self) -> None:
|
||||
for url in ("ftp://x.example.com/", "file:///etc/passwd", "not-a-url", "http://"):
|
||||
with self.subTest(url=url), self.assertRaises(IllegalArgumentException):
|
||||
normalize_url(url)
|
||||
|
||||
|
||||
class TestExpiredTokenRetry(unittest.TestCase):
|
||||
"""A request can wait in the server's queue for longer than its access token lives."""
|
||||
|
||||
def setUp(self) -> None:
|
||||
self.tmp = tempfile.TemporaryDirectory()
|
||||
patcher = patch("mdrsclient.cache.CONFIG_DIRNAME", self.tmp.name)
|
||||
patcher.start()
|
||||
self.addCleanup(patcher.stop)
|
||||
self.addCleanup(self.tmp.cleanup)
|
||||
self.connection = MDRSConnection(REMOTE, "http://localhost:8000/api")
|
||||
self.connection.token = make_token(3600, 86400, "sent")
|
||||
|
||||
def test_a_token_someone_else_refreshed_is_reused_rather_than_replaced(self) -> None:
|
||||
used = self.connection.token
|
||||
rotated = make_token(3600, 86400, "rotated")
|
||||
CacheFile(REMOTE).token = rotated # another process, while the request waited
|
||||
|
||||
with patch("mdrsclient.api.utils.UsersApi") as users_api:
|
||||
recovered = token_recover(self.connection, used)
|
||||
|
||||
self.assertTrue(recovered)
|
||||
users_api.assert_not_called()
|
||||
self.assertEqual(self.connection.token, rotated)
|
||||
# The retry carries whatever the session header says, not what the cache holds,
|
||||
# so reading the new token through is not enough on its own.
|
||||
self.assertEqual(self.connection.session.headers["Authorization"], f"Bearer {rotated.access}")
|
||||
|
||||
def test_the_token_is_refreshed_when_nobody_else_has(self) -> None:
|
||||
used = self.connection.token
|
||||
rotated = make_token(3600, 86400, "rotated")
|
||||
|
||||
with patch("mdrsclient.api.utils.UsersApi") as users_api:
|
||||
users_api.return_value.tokenRefresh.return_value = rotated
|
||||
recovered = token_recover(self.connection, used)
|
||||
|
||||
self.assertTrue(recovered)
|
||||
users_api.return_value.tokenRefresh.assert_called_once()
|
||||
self.assertEqual(self.connection.token, rotated)
|
||||
|
||||
def test_the_request_is_sent_once_more_and_only_once(self) -> None:
|
||||
calls: list[str] = []
|
||||
|
||||
class Api:
|
||||
def __init__(self, connection: MDRSConnection) -> None:
|
||||
self.connection = connection
|
||||
|
||||
@retry_on_expired_token
|
||||
def send(self) -> str:
|
||||
calls.append("sent")
|
||||
if len(calls) == 1:
|
||||
raise TokenExpiredException("expired on arrival")
|
||||
return "ok"
|
||||
|
||||
with patch("mdrsclient.api.utils.UsersApi") as users_api:
|
||||
users_api.return_value.tokenRefresh.return_value = make_token(3600, 86400, "rotated")
|
||||
self.assertEqual(Api(self.connection).send(), "ok")
|
||||
|
||||
self.assertEqual(len(calls), 2)
|
||||
|
||||
def test_a_second_refusal_is_reported_as_an_overloaded_server(self) -> None:
|
||||
calls: list[str] = []
|
||||
|
||||
class Api:
|
||||
def __init__(self, connection: MDRSConnection) -> None:
|
||||
self.connection = connection
|
||||
|
||||
@retry_on_expired_token
|
||||
def send(self) -> str:
|
||||
calls.append("sent")
|
||||
raise TokenExpiredException("expired on arrival")
|
||||
|
||||
with patch("mdrsclient.api.utils.UsersApi") as users_api:
|
||||
users_api.return_value.tokenRefresh.return_value = make_token(3600, 86400, "rotated")
|
||||
with self.assertRaises(ServerBusyException) as caught:
|
||||
Api(self.connection).send()
|
||||
|
||||
# Sent twice and no more, and the message names the cause the user can act on
|
||||
# rather than an authentication failure that would send them to the login form.
|
||||
self.assertEqual(len(calls), 2)
|
||||
self.assertIn("overloaded", str(caught.exception))
|
||||
|
||||
|
||||
if __name__ == "__main__":
|
||||
unittest.main()
|
||||
@@ -0,0 +1,297 @@
|
||||
import os
|
||||
import tempfile
|
||||
import unittest
|
||||
from io import StringIO
|
||||
from unittest.mock import MagicMock, patch
|
||||
|
||||
from mdrsclient.api import FilesApi
|
||||
from mdrsclient.exceptions import MDRSException, UnexpectedException
|
||||
from mdrsclient.models import File, Folder, FolderSimple, Laboratory
|
||||
from mdrsclient.transfer import Downloader, Uploader
|
||||
|
||||
TIMESTAMP = "2026-01-01T00:00:00+09:00"
|
||||
|
||||
|
||||
def make_file(id: str, name: str, size: int = 1) -> File:
|
||||
return File(
|
||||
id=id,
|
||||
name=name,
|
||||
type="text/plain",
|
||||
size=size,
|
||||
thumbnail=None,
|
||||
description="",
|
||||
metadata={},
|
||||
download_url=f"v3/files/{id}/download/",
|
||||
created_at=TIMESTAMP,
|
||||
updated_at=TIMESTAMP,
|
||||
)
|
||||
|
||||
|
||||
def make_folder_simple(id: str, name: str) -> FolderSimple:
|
||||
return FolderSimple(
|
||||
id=id,
|
||||
pid=None,
|
||||
name=name,
|
||||
access_level=1,
|
||||
lock=False,
|
||||
size=0,
|
||||
laboratory_id=1,
|
||||
description="",
|
||||
created_at=TIMESTAMP,
|
||||
updated_at=TIMESTAMP,
|
||||
restrict_opened_at=None,
|
||||
)
|
||||
|
||||
|
||||
def make_folder(id: str, name: str, path: str, sub_folders: list[FolderSimple] | None = None) -> Folder:
|
||||
return Folder(
|
||||
id=id,
|
||||
pid=None,
|
||||
name=name,
|
||||
access_level=1,
|
||||
lock=False,
|
||||
size=0,
|
||||
laboratory_id=1,
|
||||
description="",
|
||||
created_at=TIMESTAMP,
|
||||
updated_at=TIMESTAMP,
|
||||
restrict_opened_at=None,
|
||||
metadata=[],
|
||||
sub_folders=sub_folders if sub_folders is not None else [],
|
||||
path=path,
|
||||
)
|
||||
|
||||
|
||||
LABORATORY = Laboratory(id=1, name="mylab", pi_name="PI", full_name="My Laboratory")
|
||||
|
||||
|
||||
class TestUploadReportsFailure(unittest.TestCase):
|
||||
"""A file that never reached the server must not leave the command reporting success."""
|
||||
|
||||
def make_client(self, folder: Folder, files: list[File]) -> MagicMock:
|
||||
client = MagicMock()
|
||||
client.parse_remote_host_with_path.return_value = ("myremote", "mylab", "/")
|
||||
client.find_laboratory.return_value = LABORATORY
|
||||
client.find_folder.return_value = folder
|
||||
client.find_files.return_value = files
|
||||
return client
|
||||
|
||||
def test_a_refused_upload_is_raised_to_the_caller(self):
|
||||
client = self.make_client(make_folder("f1", "root", "/"), [])
|
||||
with tempfile.TemporaryDirectory() as tmp:
|
||||
local = os.path.join(tmp, "data.txt")
|
||||
with open(local, "w") as f:
|
||||
f.write("x")
|
||||
with patch("mdrsclient.transfer.FilesApi") as files_api_class:
|
||||
files_api_class.return_value.create.side_effect = MDRSException("Access is denied.")
|
||||
with self.assertRaises(UnexpectedException):
|
||||
Uploader(client).upload(local, "myremote:/mylab/", False, False)
|
||||
|
||||
def test_an_error_the_api_layer_did_not_raise_is_still_counted(self):
|
||||
"""A file that vanished between the walk and the upload must not end the run."""
|
||||
client = self.make_client(make_folder("f1", "root", "/"), [])
|
||||
with tempfile.TemporaryDirectory() as tmp:
|
||||
local = os.path.join(tmp, "data.txt")
|
||||
with open(local, "w") as f:
|
||||
f.write("x")
|
||||
with patch("mdrsclient.transfer.FilesApi") as files_api_class:
|
||||
files_api_class.return_value.create.side_effect = FileNotFoundError(local)
|
||||
with self.assertRaises(UnexpectedException):
|
||||
Uploader(client).upload(local, "myremote:/mylab/", False, False)
|
||||
|
||||
def test_a_failed_upload_names_the_file(self):
|
||||
client = self.make_client(make_folder("f1", "root", "/"), [])
|
||||
with tempfile.TemporaryDirectory() as tmp:
|
||||
local = os.path.join(tmp, "data.txt")
|
||||
with open(local, "w") as f:
|
||||
f.write("x")
|
||||
with patch("mdrsclient.transfer.FilesApi") as files_api_class:
|
||||
files_api_class.return_value.create.side_effect = MDRSException("Access is denied.")
|
||||
with patch("sys.stdout", new=StringIO()) as fake_out:
|
||||
with self.assertRaises(UnexpectedException):
|
||||
Uploader(client).upload(local, "myremote:/mylab/", False, False)
|
||||
self.assertIn(local, fake_out.getvalue())
|
||||
|
||||
def test_an_upload_that_worked_stays_quiet(self):
|
||||
client = self.make_client(make_folder("f1", "root", "/"), [])
|
||||
with tempfile.TemporaryDirectory() as tmp:
|
||||
local = os.path.join(tmp, "data.txt")
|
||||
with open(local, "w") as f:
|
||||
f.write("x")
|
||||
with patch("mdrsclient.transfer.FilesApi") as files_api_class:
|
||||
files_api_class.return_value.create.return_value = "new-id"
|
||||
Uploader(client).upload(local, "myremote:/mylab/", False, False)
|
||||
files_api_class.return_value.create.assert_called_once()
|
||||
|
||||
|
||||
class TestDownloadReportsFailure(unittest.TestCase):
|
||||
def make_client(self, remote_path_parts, folder: Folder, files_by_folder: dict[str, list[File]]) -> MagicMock:
|
||||
client = MagicMock()
|
||||
client.is_doi.return_value = False
|
||||
client.parse_remote_host_with_path.return_value = remote_path_parts
|
||||
client.find_laboratory.return_value = LABORATORY
|
||||
client.find_folder.return_value = folder
|
||||
client.find_files.side_effect = lambda folder_id: files_by_folder.get(folder_id, [])
|
||||
return client
|
||||
|
||||
def test_a_single_file_that_failed_is_raised_to_the_caller(self):
|
||||
parent = make_folder("p1", "root", "/")
|
||||
client = self.make_client(("myremote", "mylab", "/data.txt"), parent, {"p1": [make_file("x1", "data.txt")]})
|
||||
with tempfile.TemporaryDirectory() as tmp:
|
||||
with patch("mdrsclient.transfer.FilesApi") as files_api_class:
|
||||
files_api_class.return_value.download.side_effect = OSError("Permission denied.")
|
||||
with self.assertRaises(UnexpectedException):
|
||||
Downloader(client).download("myremote:/mylab/data.txt", tmp)
|
||||
|
||||
def test_a_single_file_that_arrived_stays_quiet(self):
|
||||
parent = make_folder("p1", "root", "/")
|
||||
client = self.make_client(("myremote", "mylab", "/data.txt"), parent, {"p1": [make_file("x1", "data.txt")]})
|
||||
with tempfile.TemporaryDirectory() as tmp:
|
||||
with patch("mdrsclient.transfer.FilesApi") as files_api_class:
|
||||
files_api_class.return_value.download.return_value = True
|
||||
Downloader(client).download("myremote:/mylab/data.txt", tmp)
|
||||
files_api_class.return_value.download.assert_called_once()
|
||||
|
||||
def test_a_failed_file_does_not_abandon_the_remaining_sub_folders(self):
|
||||
parent = make_folder("p1", "lab", "/", [make_folder_simple("f1", "root")])
|
||||
folders = {
|
||||
"f1": make_folder("f1", "root", "/root/", [make_folder_simple("fa", "a"), make_folder_simple("fb", "b")]),
|
||||
"fa": make_folder("fa", "a", "/root/a/"),
|
||||
"fb": make_folder("fb", "b", "/root/b/"),
|
||||
}
|
||||
files_by_folder = {
|
||||
"p1": [],
|
||||
"f1": [make_file("bad", "bad.txt")],
|
||||
"fa": [make_file("good1", "good1.txt")],
|
||||
"fb": [make_file("good2", "good2.txt")],
|
||||
}
|
||||
client = self.make_client(("myremote", "mylab", "/root"), parent, files_by_folder)
|
||||
attempted: list[str] = []
|
||||
|
||||
def download(file: File, path: str) -> bool:
|
||||
attempted.append(file.name)
|
||||
if file.name == "bad.txt":
|
||||
raise OSError("Permission denied.")
|
||||
return True
|
||||
|
||||
with tempfile.TemporaryDirectory() as tmp:
|
||||
with (
|
||||
patch("mdrsclient.transfer.FoldersApi") as folders_api_class,
|
||||
patch("mdrsclient.transfer.FilesApi") as files_api_class,
|
||||
):
|
||||
folders_api_class.return_value.retrieve.side_effect = lambda folder_id: folders[folder_id]
|
||||
files_api_class.return_value.download.side_effect = download
|
||||
with self.assertRaises(UnexpectedException):
|
||||
Downloader(client).download("myremote:/mylab/root", tmp, is_recursive=True)
|
||||
|
||||
self.assertEqual(sorted(attempted), ["bad.txt", "good1.txt", "good2.txt"])
|
||||
|
||||
|
||||
class TestRecursiveDownloadResilience(unittest.TestCase):
|
||||
"""A folder the client cannot prepare locally must not abandon its siblings."""
|
||||
|
||||
def test_a_folder_that_cannot_be_created_does_not_stop_the_walk(self):
|
||||
parent = make_folder("p1", "lab", "/", [make_folder_simple("f1", "root")])
|
||||
folders = {
|
||||
"f1": make_folder("f1", "root", "/root/", [make_folder_simple("fa", "a"), make_folder_simple("fb", "b")]),
|
||||
"fa": make_folder("fa", "a", "/root/a/"),
|
||||
"fb": make_folder("fb", "b", "/root/b/"),
|
||||
}
|
||||
files_by_folder = {"p1": [], "f1": [], "fa": [make_file("good1", "good1.txt")], "fb": []}
|
||||
client = MagicMock()
|
||||
client.is_doi.return_value = False
|
||||
client.parse_remote_host_with_path.return_value = ("myremote", "mylab", "/root")
|
||||
client.find_laboratory.return_value = LABORATORY
|
||||
client.find_folder.return_value = parent
|
||||
client.find_files.side_effect = lambda folder_id: files_by_folder.get(folder_id, [])
|
||||
|
||||
real_makedirs = os.makedirs
|
||||
|
||||
def makedirs(path, *args, **kwargs):
|
||||
if os.path.basename(path) == "b":
|
||||
raise PermissionError("Permission denied")
|
||||
return real_makedirs(path, *args, **kwargs)
|
||||
|
||||
attempted: list[str] = []
|
||||
|
||||
with tempfile.TemporaryDirectory() as tmp:
|
||||
with (
|
||||
patch("mdrsclient.transfer.FoldersApi") as folders_api_class,
|
||||
patch("mdrsclient.transfer.FilesApi") as files_api_class,
|
||||
patch("mdrsclient.transfer.os.makedirs", side_effect=makedirs),
|
||||
):
|
||||
folders_api_class.return_value.retrieve.side_effect = lambda folder_id: folders[folder_id]
|
||||
files_api_class.return_value.download.side_effect = lambda file, path: attempted.append(file.name)
|
||||
with self.assertRaises(UnexpectedException):
|
||||
Downloader(client).download("myremote:/mylab/root", tmp, is_recursive=True)
|
||||
|
||||
self.assertEqual(attempted, ["good1.txt"])
|
||||
|
||||
|
||||
class TestFileDownloadPermission(unittest.TestCase):
|
||||
"""A file the client could not write is a failure, not a line of successful output."""
|
||||
|
||||
def make_connection(self, chunks) -> MagicMock:
|
||||
connection = MagicMock()
|
||||
connection.token = None
|
||||
response = MagicMock()
|
||||
response.status_code = 200
|
||||
response.iter_content.return_value = chunks
|
||||
connection.get.return_value = response
|
||||
return connection
|
||||
|
||||
def test_a_download_that_cannot_be_written_is_reported(self):
|
||||
connection = self.make_connection([b"payload"])
|
||||
with tempfile.TemporaryDirectory() as tmp:
|
||||
path = os.path.join(tmp, "precious.dat")
|
||||
with open(path, "wb") as f:
|
||||
f.write(b"do not touch")
|
||||
os.chmod(path, 0o444)
|
||||
try:
|
||||
with open(path, "r+b"):
|
||||
self.skipTest("running with rights that ignore the file mode")
|
||||
except OSError:
|
||||
pass
|
||||
|
||||
with self.assertRaises(UnexpectedException):
|
||||
FilesApi(connection).download(make_file("x1", "precious.dat"), path)
|
||||
|
||||
os.chmod(path, 0o644)
|
||||
with open(path, "rb") as f:
|
||||
self.assertEqual(f.read(), b"do not touch")
|
||||
connection.get.assert_not_called()
|
||||
|
||||
def test_an_interrupted_download_leaves_the_existing_file_intact(self):
|
||||
def chunks():
|
||||
yield b"half a file"
|
||||
raise ConnectionError("connection reset")
|
||||
|
||||
connection = self.make_connection(chunks())
|
||||
with tempfile.TemporaryDirectory() as tmp:
|
||||
path = os.path.join(tmp, "existing.dat")
|
||||
with open(path, "wb") as f:
|
||||
f.write(b"the copy already here")
|
||||
|
||||
with self.assertRaises(ConnectionError):
|
||||
FilesApi(connection).download(make_file("x1", "existing.dat"), path)
|
||||
|
||||
with open(path, "rb") as f:
|
||||
self.assertEqual(f.read(), b"the copy already here")
|
||||
self.assertEqual(os.listdir(tmp), ["existing.dat"])
|
||||
|
||||
def test_a_finished_download_replaces_the_destination(self):
|
||||
connection = self.make_connection([b"new ", b"contents"])
|
||||
with tempfile.TemporaryDirectory() as tmp:
|
||||
path = os.path.join(tmp, "existing.dat")
|
||||
with open(path, "wb") as f:
|
||||
f.write(b"old")
|
||||
|
||||
FilesApi(connection).download(make_file("x1", "existing.dat"), path)
|
||||
|
||||
with open(path, "rb") as f:
|
||||
self.assertEqual(f.read(), b"new contents")
|
||||
self.assertEqual(os.listdir(tmp), ["existing.dat"])
|
||||
|
||||
|
||||
if __name__ == "__main__":
|
||||
unittest.main()
|
||||
Reference in New Issue
Block a user