def find_network(network, user_id):
result = {network: user_id}
friends = handler[network](user_id)
if network == 'vk':
query = Q(vk__in=friends)
elif network == 'instagram':
query = Q(instagram__in=friends)
elif network == 'twitter':
query = Q(twitter__in=friends)
else:
return result
data = {'vk': list(), 'instagram': list(), 'twitter': list()}
for profile in Profile.objects(query):
if network != 'vk' and profile.vk:
for id in profile.vk:
data['vk'].append(id)
if network != 'instagram' and profile.instagram:
for id in profile.instagram:
data['instagram'].append(id)
if network != 'twitter' and profile.twitter:
for id in profile.twitter:
data['twitter'].append(id)
print("Profiles from database is finded") # DEBUG
friends = {'vk': list(), 'instagram': list(), 'twitter': list()}
with ThreadPoolExecutor(max_workers=12) as executor:
for key in data:
for item in executor.map(lambda x: handler[key](x), data[key]):
for uid in item:
friends[key].append(uid)
print("Parsing complete " + key)
print("Friend list is generated") # DEBUG
temp = {'vk': list(), 'instagram': list(), 'twitter': list()}
for key in data:
for id in data[key]:
response = run([
'dotnet', 'trainer/Bindex.Trainer.dll', network,
str(user_id), key,
str(id)
],
stdout=PIPE,
stderr=PIPE,
universal_newlines=True)
#try:
vector = [
float(item) for item in response.stdout.strip().split(' ')
]
with graph.as_default():
factor = model.predict(np.array([vector]))[0][0]
if factor >= 0.5:
temp[key].append((id, factor))
#except:
#pass
print("Neural network is used") # DEBUG
for key in temp:
if not temp[key]:
continue
result[key] = int(max(temp[key], key=lambda x: x[1])[0])
print("Profiles from other networks is finded") # DEBUG
return result
