def add_feature(feature: str,
UserIdentifier: str = Form(...),
ValidateFieldAnswer: str = Form(...)):
"""
This endpoint gets called when the user answers a question from the self screening collect,
the path param must match the symptoms on the feature_mappings and the answer must match
the ones in the response_mapping. If successful, adds the feature to the Endless Medical
current session.
:param: feature: that must match the mapping at features_mapping
:param: ValidateFieldAnswer: answer from the user, must match reponse_mappings to be valid
"""
session_id = UserDocument.get_by_phone(
UserIdentifier).endless_medical_token
try:
logger.debug("Adding {feature} with {value}".format(
feature=features_mapping[feature],
value=reponse_mappings[ValidateFieldAnswer.lower()],
))
res = endless_medical_api.add_feature(
session_id,
features_mapping[feature],
reponse_mappings[ValidateFieldAnswer.lower()],
)
logger.debug(res)
# Result as per the Twilio Docs
if not res:
return {"valid": False}
return {"valid": True}
except BaseHTTPError as err:
capture_message(err)
return {"valid": False}
except KeyError as err:
capture_message(err)
return {"valid": False}
def self_screening_lives_in_area(UserIdentifier: str = Form(...),
Memory: str = Form(...)):
"""
Checks if the user lives in a COVID-19 affected area, leaved here for legacy reasons; probably
does not make sense anymore due to above implementation
:param: UserIdentifier: Phone number from Twilio
:param: Memory: JSON Stringified object from Twilio
"""
memory = json.loads(Memory)
twilio = memory.pop("twilio")
start_screening = twilio["collected_data"]["q1"]["answers"][
"lives-in-area"]["answer"]
if start_screening == "No":
return screening_not_in_danger
try:
accept_tos_and_store_session_id(UserIdentifier)
return {"actions": [{"redirect": "task://self-screening-q-rest"}]}
except faunadb.errors.BadRequest as err:
capture_message(err)
return {"actions": [{"redirect": "task://fallback"}]}
except BaseHTTPError as err:
capture_message(err)
return {"actions": [{"redirect": "task://fallback"}]}
def error_fallback_action(_, exc: Exception):
"""
This function catches 500 code exceptions and returns a fallback action for the bot to say.
"""
capture_message(traceback.print_exception(None, exc, exc.__traceback__))
return JSONResponse(
content={
"actions": {
"actions": [
{
"say":
"Oops, looks like ~the hive mind can't come to an agreement~ I can fulfill your task; "
"let's try again "
},
{
"redirect": "task://menu-description"
},
]
}
})
def self_screening_start(UserIdentifier: str = Form(...),
Memory: str = Form(...)):
"""
Starts the screening process
:param: UserIdentifier: Phone number from Twilio
:param: Memory: JSON Stringified object from Twilio
"""
memory = json.loads(Memory)
twilio = memory.pop("twilio")
start_screening = twilio["collected_data"]["accepts-test"]["answers"][
"start-screening"]["answer"]
user = UserDocument.get_by_phone(UserIdentifier)
try:
if start_screening == "No":
return {
"actions": [
{
"say": "Ok no problem! Let's go back to the menu"
},
{
"redirect": "task://menu-description"
},
]
}
# Check if user is in db, otherwise we cannot continue
if start_screening == "Yes" and not user:
return {
"actions": [
{
"say":
"Sorry, I cannot continue until you give me your name \U00012639"
},
{
"redirect": "task://can-have-name"
},
]
}
# Check if the user lives in a COVID-19 affected area
lives_in_risky_area = novelcovid_api.lives_in_risky_zone(user.country)
# Said yes but is not in risky area
if start_screening == "Yes" and not lives_in_risky_area:
return screening_not_in_danger
# Said yes and IS in risky area, so we can mark the first question as "yes"
if start_screening == "Yes" and lives_in_risky_area:
accept_tos_and_store_session_id(UserIdentifier)
return {
"actions": [
{
"say":
"Your country already has cases of COVID-19, so we will skip that question; let's go "
"with the rest "
},
{
"redirect": "task://self-screening-q-rest"
},
]
}
# Continue if user accepts
if start_screening == "Yes":
return {
"actions": [
{
"say": "Alright, let's start"
},
{
"redirect": "task://self-screening-lives-in-area"
},
]
}
return {
"actions": [
{
"say": "Cool! No problem, let's get back to the menu"
},
{
"redirect": "task://menu-description"
},
]
}
except faunadb.errors.BadRequest as err:
capture_message(err)
return {"actions": [{"redirect": "task://fallback"}]}
except BaseHTTPError as err:
capture_message(err)
return {"actions": [{"redirect": "task://fallback"}]}
except Exception as err:
capture_message(err)
return {"actions": [{"redirect": "task://fallback"}]}
def analyze_answers(UserIdentifier: str = Form(...)):
"""
Analyzes the answers given by the user, see that we are not parsing the Memory object,
that's because on every questions the answers were validated by the endpoint defined below
and added to the Endless Medical API session
"""
# We don't need to access collect since arriving here means
# all answers have been added to the Endless Medical Session
try:
session_id = UserDocument.get_by_phone(
UserIdentifier).endless_medical_token
except AttributeError as err:
capture_message(err)
return {"actions": {"redirect": "task://fallback"}}
# Just an inline function to clean the user
def reset_session_id_token():
user = UserDocument.get_by_phone(UserIdentifier)
user.endless_medical_token = None
user.update()
try:
# Analyse data and get only the posible diseases
outcomes: List[Dict[str, str]] = endless_medical_api.analyse(
session_id)["Diseases"]
logger.debug(outcomes)
# Parsing the diseases, the API returns a confidence from 0.0 to 1.0 for every predicted disease
# if COVID-19 has >= 50 % we recommend calling the doctor; else we check if any other diseases has at least
# 50 % chance, if so, COVID-19 it's discarded but we recommend calling the doctor anyway
# Check if COVID-19 passes threshold
if check_outcomes(outcomes, outcomes_mapping.get("covid-19")):
return {
"actions":
list(
chain(
[{
"say":
"You should seek medical attention as soon as possible"
}],
screening_pre_results_warning,
))
}
# COVID-19 did not pass threshold, check now if any other outcome does
if check_outcomes(outcomes):
return {
"actions":
list(
chain(
[{
"say":
"You probably do not have COVID-19, but, according to your symptoms, "
"you should seek medical attention anyways "
}],
self_screening_lives_in_area,
))
}
# User is probably ok
return {
"actions": [
{
"say":
"Great news! You don not have anything to worry about 🥳\U0001f973"
},
{
"redirect": "task://menu-description"
},
]
}
except BaseHTTPError as err:
capture_message(err)
return {
"actions": [
{
"say":
"Sorry, our super AI doctor had a problem analysing the results; please try again."
},
{
"redirect": "task://menu-description"
},
]
}
finally:
reset_session_id_token()