def write_workflow_attachment(run_id: str, run_request: RunRequest,
files: Dict[str, FileStorage]) -> None:
exe_dir: Path = get_path(run_id, "exe_dir")
host = request.host_url.strip("/")
url_prefix = current_app.config['URL_PREFIX'].strip("/")
endpoint = f"{host}/{url_prefix}".strip("/")
for file in run_request["workflow_attachment"]:
if "file_name" in file and "file_url" in file:
file_name: str = file["file_name"]
file_url: str = file["file_url"]
parsed_url = parse.urlparse(file_url)
if parsed_url.scheme in ["http", "https"] and \
not file_url.startswith(endpoint):
file_path = \
exe_dir.joinpath(secure_filepath(file_name)).resolve()
file_path.parent.mkdir(parents=True, exist_ok=True)
response: Response = requests.get(file["file_url"])
with file_path.open(mode="wb") as f:
f.write(response.content)
if current_app.config["WORKFLOW_ATTACHMENT"]:
workflow_attachment = \
files.getlist("workflow_attachment[]") # type: ignore
for file in workflow_attachment:
file_name = secure_filepath(file.filename) # type: ignore
file_path = exe_dir.joinpath(file_name).resolve()
file_path.parent.mkdir(parents=True, exist_ok=True)
file.save(file_path) # type: ignore
def get_runs_id_data(run_id: str, subpath: str = "") -> Response:
"""
This provides a remote url to download a file or directory under the
`run_dir` of the sapporo-service.
- In the case of `path/to/file`, this returns the file.
- In the case of `path/to/dir`, this returns the list of files under
directory in JSON format.
- In the case of `path/to/dir?download=true`, this returns the directory
in zip format.
The path is relative to the base directory of each run.
See `README.md` in sapporo-service for the structure of `run_dir`.
For example, if you want to download the output `foo.txt`, specify
something like `outputs/foo.txt`.
`..` will be ignored.
"""
validate_run_id(run_id)
if Path(subpath).name[0] == ".":
requested_path = \
secure_filepath(str(Path(subpath).parent)
).joinpath(Path(subpath).name)
else:
requested_path = secure_filepath(subpath)
path = get_run_dir(run_id).joinpath(requested_path)
if not path.exists():
parent = Path(f"runs/{run_id}/data").joinpath(requested_path.parent)
abort(400,
f"The specified path: {requested_path} does not exist. "
f"Please make another request to `<endpoint>/{parent}/` again "
"and check the dir structure.")
if path.is_file():
return send_file(path, as_attachment=True)
else:
if str2bool(request.args.get("download", False)):
with NamedTemporaryFile() as f:
res = make_archive(f.name, "zip",
root_dir=path.parent, base_dir=path.name)
if "temp_files" not in g:
g.temp_files = [Path(f"{f.name}.zip")]
else:
g.temp_files.append(Path(f"{f.name}.zip"))
return send_file(res, as_attachment=True,
attachment_filename=f"{path.name}.zip")
else:
response: Response = jsonify(path_hierarchy(path, path))
response.status_code = GET_STATUS_CODE
return response
def validate_and_update_run_request(run_id: str,
run_request: RunRequest,
files: Dict[str, FileStorage]) \
-> RunRequest:
if current_app.config["REGISTERED_ONLY_MODE"] and \
"workflow_url" in run_request:
abort(
400, "Currently, sapporo-service is running with "
"registered_only_mode. "
"Therefore, you need to specify a workflow using "
"`workflow_name` field. A list of executable workflows can "
"be retrieved requesting `GET /service-info`")
if "workflow_name" in run_request:
wf: Workflow = get_workflow(run_request["workflow_name"])
run_request["workflow_url"] = wf["workflow_url"]
run_request["workflow_type"] = wf["workflow_type"]
run_request["workflow_type_version"] = wf["workflow_type_version"]
if "workflow_attachment" not in run_request:
run_request["workflow_attachment"] = wf["workflow_attachment"]
for field in [
"workflow_params", "workflow_type", "workflow_type_version",
"workflow_url", "workflow_engine_name"
]:
if field not in run_request:
abort(400,
f"{field} not included in the form data of the request.")
if "workflow_attachment" in run_request:
if type(run_request["workflow_attachment"]) is str: # type: ignore
run_request["workflow_attachment"] = \
json.loads(run_request["workflow_attachment"]) # type: ignore
else:
run_request["workflow_attachment"] = []
if "workflow_engine_parameters" not in run_request:
run_request["workflow_engine_parameters"] = "{}"
if "tags" not in run_request:
run_request["tags"] = "{}"
if "workflow_name" not in run_request:
tags = json.loads(run_request["tags"])
if "workflow_name" in tags:
run_request["workflow_name"] = tags["workflow_name"]
else:
run_request["workflow_name"] = \
parse.urlparse(run_request["workflow_url"]).path.split("/")[-1]
if current_app.config["WORKFLOW_ATTACHMENT"]:
workflow_attachment = \
files.getlist("workflow_attachment[]") # type: ignore
exe_dir: Path = get_path(run_id, "exe_dir")
host = request.host_url.strip("/")
url_prefix = current_app.config['URL_PREFIX'].strip("/")
endpoint = f"{host}/{url_prefix}".strip("/")
base_remote_url = f"{endpoint}/runs/{run_id}/data/"
for f in workflow_attachment:
file_name: Path = secure_filepath(f.filename)
file_path: Path = exe_dir.joinpath(file_name).resolve()
run_request["workflow_attachment"].append({
"file_name":
str(file_name),
"file_url":
base_remote_url + str(file_path.relative_to(exe_dir.parent))
})
validate_wf_type(run_request["workflow_type"],
run_request["workflow_type_version"])
validate_meta_charactors("workflow_url", run_request["workflow_url"])
validate_meta_charactors("workflow_engine_name",
run_request["workflow_engine_name"])
return run_request
def test_contain_space() -> None:
assert secure_filepath("My cool movie.mov") == Path("My_cool_movie.mov")
def test_hidden_file() -> None:
assert secure_filepath(".foo") == Path("foo")
def test_DS_STORE() -> None:
assert secure_filepath("._.DS_STORE") == Path("DS_STORE")
def test_only_double_dot() -> None:
assert secure_filepath("..") == Path("")
def test_only_root() -> None:
assert secure_filepath("/") == Path("")
def test_contain_ampersand() -> None:
assert secure_filepath("/&&/&foo/bar") == Path("foo/bar")
def test_contain_fullsize_ampersand() -> None:
assert secure_filepath("/＆foo/bar") == Path("foo/bar")
def test_contain_pipe() -> None:
assert secure_filepath("/||/|foo/bar") == Path("foo/bar")
def test_japanese_filename() -> None:
assert secure_filepath("/フーfoo/バーbar") == Path("foo/bar")
def test_contain_umlauts() -> None:
assert secure_filepath(
u"i contain cool \xfcml\xe4uts.txt") == \
Path("i_contain_cool_umlauts.txt")
def test_root_dir() -> None:
assert secure_filepath("/foo/bar") == Path("foo/bar")
def test_prev_dir() -> None:
assert secure_filepath("../../../etc/passwd") == Path("etc/passwd")
