def get_preview():
"""Get files preview
Returns
-------
json
previewFiles: preview of selected files
error: True if error, else False
errorMessage: the error message of error, else an empty string
"""
data = request.get_json()
try:
files_handler = FilesHandler(current_app, session)
files_handler.handle_files(data['filesId'])
results = []
for file in files_handler.files:
file.set_preview()
res = file.get_preview()
results.append(res)
except Exception as e:
traceback.print_exc(file=sys.stdout)
return jsonify({
'previewFiles': [],
'error': True,
'errorMessage': str(e)
}), 500
return jsonify({
'previewFiles': results,
'error': False,
'errorMessage': ''
})
def integrate():
"""Integrate a file
Returns
-------
json
task_id: celery task id
error: True if error, else False
errorMessage: the error message of error, else an empty string
"""
data = request.get_json()
session_dict = {'user': session['user']}
task = None
try:
files_handler = FilesHandler(current_app, session, host_url=request.host_url)
files_handler.handle_files([data["fileId"], ])
for file in files_handler.files:
data["externalEndpoint"] = data["externalEndpoint"] if data["externalEndpoint"] else None
data["customUri"] = data["customUri"] if data["customUri"] else None
dataset_info = {
"celery_id": None,
"file_id": file.id,
"name": file.human_name,
"graph_name": file.file_graph,
"public": data["public"] if session["user"]["admin"] else False
}
dataset = Dataset(current_app, session, dataset_info)
dataset.save_in_db()
data["dataset_id"] = dataset.id
task = current_app.celery.send_task('integrate', (session_dict, data, request.host_url))
dataset.update_celery(task.id)
except Exception as e:
traceback.print_exc(file=sys.stdout)
return jsonify({
'error': True,
'errorMessage': str(e),
'task_id': ''
}), 500
return jsonify({
'error': False,
'errorMessage': '',
'task_id': task.id if task else ''
})
def integrate_file(self, info, public=False):
"""Summary
Parameters
----------
info : TYPE
Description
Returns
-------
TYPE
Description
"""
files_handler = FilesHandler(self.app, self.session)
files_handler.handle_files([
info["id"],
])
for file in files_handler.files:
dataset_info = {
"celery_id": "000000000",
"file_id": file.id,
"name": file.name,
"graph_name": file.file_graph,
"public": public
}
dataset = Dataset(self.app, self.session, dataset_info)
dataset.save_in_db()
if file.type == "csv/tsv":
file.integrate(dataset.id, info["columns_type"], public=public)
elif file.type == "gff/gff3":
file.integrate(dataset.id, info["entities"], public=public)
elif file.type == "bed":
file.integrate(dataset.id, info["entity_name"], public=public)
# done
dataset.update_in_db("success")
dataset.set_info_from_db()
return {
"timestamp": file.timestamp,
"start": dataset.start,
"end": dataset.end
}
def edit_file():
"""Edit file name
Returns
-------
json
files: list of all files of current user
error: True if error, else False
errorMessage: the error message of error, else an empty string
"""
data = request.get_json()
current_app.logger.debug(data)
if not (data and data.get("id") and data.get("newName")):
return jsonify({
'files': [],
'diskSpace': 0,
'error': True,
'errorMessage': "Missing parameters"
}), 400
files_id = [data["id"]]
new_name = data["newName"]
try:
files_handler = FilesHandler(current_app, session)
files_handler.handle_files(files_id)
for file in files_handler.files:
file.edit_name_in_db(new_name)
files = files_handler.get_files_infos()
except Exception as e:
traceback.print_exc(file=sys.stdout)
return jsonify({
'files': [],
'diskSpace': 0,
'error': True,
'errorMessage': str(e)
}), 500
return jsonify({
'files': files,
'diskSpace': files_handler.get_size_occupied_by_user(),
'error': False,
'errorMessage': ''
})
def integrate():
"""Integrate a file
Returns
-------
json
datasets_id: dataset ids
error: True if error, else False
errorMessage: the error message of error, else an empty string
"""
data = request.get_json()
if not (data and data.get("fileId")):
return jsonify({
'error': True,
'errorMessage': "Missing fileId parameter",
'dataset_ids': None
}), 400
session_dict = {'user': session['user']}
task = None
dataset_ids = []
try:
files_handler = FilesHandler(current_app,
session,
host_url=request.host_url)
files_handler.handle_files([
data["fileId"],
])
for file in files_handler.files:
data["externalEndpoint"] = data["externalEndpoint"] if (
data.get("externalEndpoint")
and isinstance(file, RdfFile)) else None
data["customUri"] = data["customUri"] if (
data.get("customUri")
and not isinstance(file, RdfFile)) else None
dataset_info = {
"celery_id": None,
"file_id": file.id,
"name": file.human_name,
"graph_name": file.file_graph,
"public":
data.get("public") if session["user"]["admin"] else False
}
dataset = Dataset(current_app, session, dataset_info)
dataset.save_in_db()
data["dataset_id"] = dataset.id
dataset_ids.append(dataset.id)
task = current_app.celery.send_task(
'integrate', (session_dict, data, request.host_url))
dataset.update_celery(task.id)
except Exception as e:
traceback.print_exc(file=sys.stdout)
return jsonify({
'error': True,
'errorMessage': str(e),
'dataset_ids': None
}), 500
return jsonify({
'error': False,
'errorMessage': '',
'dataset_ids': dataset_ids
})
def integrate(self, session, data, host_url):
"""Integrate a file into the triplestore
Parameters
----------
session : dict
AskOmics session
data : dict
fileId: file to integrate
public: integrate as public or private data
host_url : string
AskOmics host url
Returns
-------
dict
error: True if error, else False
errorMessage: the error message of error, else an empty string
"""
files_handler = FilesHandler(app,
session,
host_url=host_url,
external_endpoint=data["externalEndpoint"],
custom_uri=data["customUri"])
files_handler.handle_files([
data["fileId"],
])
public = data["public"] if session["user"]["admin"] else False
for file in files_handler.files:
try:
dataset_info = {
"celery_id": self.request.id,
"id": data["dataset_id"],
"file_id": file.id,
"name": file.human_name,
"graph_name": file.file_graph,
"public": public
}
dataset = Dataset(app, session, dataset_info)
dataset.update_in_db("started",
update_date=True,
update_graph=True)
if file.type == "csv/tsv":
file.integrate(data["dataset_id"],
data['columns_type'],
data['header_names'],
public=public)
elif file.type == "gff/gff3":
file.integrate(data["dataset_id"],
data["entities"],
public=public)
elif file.type == "turtle":
file.integrate(public=public)
elif file.type == "bed":
file.integrate(data["dataset_id"],
data["entity_name"],
public=public)
# done
dataset.update_in_db("success", ntriples=file.ntriples)
except Exception as e:
traceback.print_exc(file=sys.stdout)
trace = traceback.format_exc()
dataset.update_in_db("failure",
error=True,
error_message=str(e),
traceback=trace)
# Rollback
file.rollback()
raise e
return {'error': True, 'errorMessage': str(e)}
return {'error': False, 'errorMessage': ''}
