def python_widget_execute(output_type):
# retrieve input parameters
try:
request_body = request.get_json(force=True)
token, output_file = utils.get_widget_config(request_body)
isAuthenticated, script = security.jwt_token_to_python_script(token)
dataset_name, datastore = utils.get_dataset(request_body)
drivers = utils.get_analytical_drivers(request_body)
except Exception as e:
return raise_error("Error during request decoding: {}".format(e), e)
if not isAuthenticated:
logging.error("Unauthorized access")
return "Unauthorized", 401
# resolve analytical drivers
for d in drivers:
script = script.replace("$P{" + d + "}", "drivers_.get(\'" + d + "\')")
# resolve dataset references
if dataset_name != None:
script = script.replace(dataset_name, "df_")
# convert to dataframe
df = {}
try:
if datastore != None:
df = utils.datastore_to_dataframe(datastore['metaData']['fields'],
datastore['rows'])
except Exception as e:
return raise_error("Error during dataframe conversion: {}".format(e),
e)
# execute script
try:
namespace = {"df_": df, "drivers_": drivers}
exec(script, namespace)
except Exception as e:
return raise_error("Error during script execution: {}".format(e), e)
# collect script result
with open(output_file, "rb") as f:
output_file_content = f.read()
#delete temp files
try:
os.remove(output_file)
except Exception:
pass
if output_type == "img":
encoded_img = base64.b64encode(output_file_content)
to_return = "<img src=\"data:image/;base64, " + encoded_img.decode(
'utf-8') + "\" style=\"width:100%;height:100%;\">"
else:
to_return = output_file_content
return to_return, 200
def python_img():
# retrieve input parameters
script, img_file = utils.get_widget_config(request.get_json())
user_id = utils.get_knowage_token(request.headers)
dataset_name, datastore_request = utils.get_dataset(request.get_json())
document_id, widget_id = utils.get_widget_info(request.get_json())
drivers = utils.get_analytical_drivers(request.get_json())
python_widget = PythonWidgetExecution(analytical_drivers=drivers,
script=script,
output_variable=img_file,
user_id=user_id,
dataset_name=dataset_name,
datastore_request=datastore_request)
# check authentication
if not security.is_user_authorized_for_functionality(
python_widget, constants.EDIT_PYTHON_SCRIPTS):
return "Error: authentication failed", 401
# resolve analytical drivers
for d in drivers:
python_widget.script = python_widget.script.replace(
"$P{" + d + "}", "drivers_.get(\'" + d + "\')")
try:
# retrieve dataset
if python_widget.dataset_name != None:
dataset_file = constants.TMP_FOLDER + python_widget.dataset_name + ".pckl"
df = utils.load_data_as_dataframe(python_widget)
df.to_pickle(dataset_file)
python_widget.script = "import pandas as pd\n" + python_widget.dataset_name + " = pd.read_pickle(\"" + dataset_file + "\")\n" + python_widget.script
except Exception as e:
logging.error("Error during dataframe conversion: {}".format(e))
return str(e), 400
try:
# execute script
namespace = {"drivers_": drivers}
exec(python_widget.script, namespace)
except Exception as e:
logging.error("Error during script execution: {}".format(e))
return str(e), 400
# collect script result
with open(img_file, "rb") as f:
encoded_img = base64.b64encode(f.read())
#delete temp files
try:
os.remove(img_file)
os.remove(dataset_file)
except Exception:
pass
return "<img src=\"data:image/;base64, " + encoded_img.decode(
'utf-8') + "\" style=\"width:100%;height:100%;\">", 200
def python_html():
#retrieve input parameters
script, output_variable = utils.get_widget_config(request.get_json())
user_id = utils.get_knowage_token(request.headers)
dataset_name, datastore_request = utils.get_dataset(request.get_json())
drivers = utils.get_analytical_drivers(request.get_json())
python_widget = PythonWidgetExecution(analytical_drivers=drivers,
script=script,
output_variable=output_variable,
user_id=user_id,
dataset_name=dataset_name,
datastore_request=datastore_request)
# check authentication
if not security.is_user_authorized_for_functionality(
python_widget, constants.EDIT_PYTHON_SCRIPTS):
return "Error: authentication failed", 401
# resolve analytical drivers
for d in drivers:
python_widget.script = python_widget.script.replace(
"$P{" + d + "}", "drivers_.get(\'" + d + "\')")
try:
#retrieve dataset
if python_widget.dataset_name != None:
dataset_file = constants.TMP_FOLDER + python_widget.dataset_name + ".pckl"
df = utils.load_data_as_dataframe(python_widget)
df.to_pickle(dataset_file)
python_widget.script = "import pandas as pd\n" + python_widget.dataset_name + " = pd.read_pickle(\"" + dataset_file + "\")\n" + python_widget.script
except Exception as e:
logging.error("Error during dataframe conversion: {}".format(e))
return str(e), 400
try:
#execute script
namespace = {python_widget.output_variable: "", "drivers_": drivers}
exec(python_widget.script, namespace)
except Exception as e:
logging.error("Error during script execution: {}".format(e))
return str(e), 400
#collect script result
html = namespace[python_widget.output_variable]
#remove dataset tmp file
try:
os.remove(dataset_file)
except Exception:
pass
return html, 200
def python_bokeh():
utils.bokeh_garbage_collector()
# retrieve input parameters
document_id, widget_id = utils.get_widget_info(request.get_json())
script_file_name = constants.TMP_FOLDER + "bokeh_script_" + str(
widget_id) + ".txt"
user_id = utils.get_knowage_token(request.headers)
dataset_name, datastore_request = utils.get_dataset(request.get_json())
drivers = utils.get_analytical_drivers(request.get_json())
python_widget = PythonWidgetExecution(analytical_drivers=drivers,
user_id=user_id,
document_id=document_id,
widget_id=widget_id,
dataset_name=dataset_name,
datastore_request=datastore_request)
python_widget.script = security.load_script_from_template(python_widget)
#destroy old bokeh server
if utils.server_exists(python_widget.widget_id):
utils.destroy_server(python_widget.widget_id)
# resolve analytical drivers
for d in drivers:
python_widget.script = python_widget.script.replace(
"$P{" + d + "}", "drivers_.get(\'" + d + "\')")
# retrieve dataset
if python_widget.dataset_name != None:
dataset_file = constants.TMP_FOLDER + python_widget.dataset_name + ".pckl"
df = utils.load_data_as_dataframe(python_widget)
df.to_pickle(dataset_file)
python_widget.script = "import pandas as pd\n" + python_widget.dataset_name + " = pd.read_pickle(\"" + dataset_file + "\")\n" + python_widget.script
#function executed by bokeh server
def modify_doc(doc):
# retrieve script from file
with open(script_file_name, "r") as bk_file:
bk_script = bk_file.read()
#replace curdoc() with keyword "curdoc_"
bk_script = bk_script.replace("curdoc()", "curdoc_")
# execute script
namespace = {'curdoc_': doc, "drivers_": drivers}
exec(bk_script, namespace)
#secondary thread function (bokeh server)
def bk_worker():
server = Server({'/bkapp' + str(python_widget.widget_id): modify_doc},
io_loop=IOLoop(),
allow_websocket_origin=["*"],
port=cuncurrency_manager.bokeh_resources[
python_widget.widget_id].port)
with cuncurrency_manager.lck:
cuncurrency_manager.active_servers.update(
{python_widget.widget_id: server}) #{widget_id : bokeh_server}
server.start()
server.io_loop.start()
#flush script content to file so that modify_doc() can retrieve the code to be executed
with open(script_file_name, "w") as bokeh_file:
bokeh_file.write(python_widget.script)
#instance a bokeh server for the widget if not instanciated yet
if not utils.server_exists(
python_widget.widget_id): #allocate bokeh server
t = Thread(target=bk_worker) #thread that hosts bokeh server
bk_res = BokehResourceList(thread=t,
timestamp=datetime.now(),
port=utils.find_free_port(),
dataset_name=dataset_name)
with cuncurrency_manager.lck:
cuncurrency_manager.bokeh_resources.update(
{python_widget.widget_id:
bk_res}) #{widget_id : BokehResourceList}
t.start()
#serve plot
jscript = server_document(utils.get_python_engine_address() + ":" + str(
cuncurrency_manager.bokeh_resources[python_widget.widget_id].port) +
"/bkapp" + str(python_widget.widget_id))
return render_template("embed.html", script=jscript)