def __init__(self, date_format = '%Y-%m-%d %H:%M:%S', *args, **kwargs ):
super(Instrument, self).__init__()
global SingletonInstrument
if SingletonInstrument and not kwargs.get('serial_emulator'):
raise ValueError("SingletonInstrument already set")
self.name = 'instrument'
self.uuid = os.environ["MQTT_UUID"]
self.mqtt_org = kwargs.get('mqtt_org')
self.mqtt_host = kwargs.get('mqtt_host')
self.mqtt_port = kwargs.get('mqtt_port')
self.mqtt_keep_alive = kwargs.get('mqtt_keep_alive')
self.mqtt_qos = kwargs.get('mqtt_qos')
self.serial_port_description = kwargs.get('serial_port_description')
self.serial_baudrate = kwargs.get('serial_baudrate')
self.serial_parity = kwargs.get('serial_parity')
self.serial_stopbits = kwargs.get('serial_stopbits')
self.serial_bytesize = kwargs.get('serial_bytesize')
self.serial_timeout = kwargs.get('serial_timeout')
self.storage_location = kwargs.get('storage_location')
init_models(self.storage_location)
# self.mqtt_publish_topic = ''
self.mqtt_publish_topic = self._get_topic(topic_type = MQTT_TYPE_READING)
self.mqtt_analysis_topic = self._get_topic(topic_type = MQTT_TYPE_ANALYSIS)
self._mqtt_connected = False
self._mqtt_clean_session = False
self._mqtt_retain = False
self._mqtt_messages_lost = False
# Depending on memory size this is approx 400 MB of memory
self._mqtt_max_queue = 100000
self._mqtt_client = self._setup_mqtt_client()
self._serial = None
self._imodules = {}
self.date_format = date_format
self._log_format = '%(asctime)s [%(levelname)s] %(message)s'
self._log_info = {}
self._logger = self._setup_logger(self.name)
self.scheduler = self._set_up_scheduler()
self._mqtt_client.enable_logger(logger=self._logger)
if not kwargs.get('serial_emulator'):
self._serial = self._set_up_serial()
self._modes = {}
SingletonInstrument = self
self.timezone = strftime("%z", gmtime())
self._countdown_current = 0
self._countdown_last = 0
def create_app():
app = Flask(__name__)
app.config.from_pyfile('settings.py')
init_models(app)
init_views(app)
init_loginmanager(app)
init_bootstrap(app)
return app
def init():
# modify the name of the database here
m.db.init('slack-archv-test.sqlite')
with m.db.atomic():
m.init_models()
# Add version info
m.Information.create_or_get(key='__version', value='1.0.0')
def init_app(app, test=False):
if test:
app.config['TESTING'] = True
from views import simple_page
app.register_blueprint(simple_page)
from models import init_models
init_models(app)
def flaskr_client(context, *args, **kwargs):
APP.config['DATABASE'] = {'engine': 'peewee.SqliteDatabase', 'name': ':memory:'}
# context.db, APP.config['DATABASE'] = tempfile.mkstemp()
APP.testing = True
context.client = APP.test_client()
with APP.app_context():
DB = Database(APP)
init_models(DB)
# pass
yield context.client
def create_app():
con = FlaskApp(__name__, specification_dir='vue-blog-api')
# con.add_api('api.yaml', resolver=MethodViewResolver('views'), validate_responses=True)
con.add_api('api.yaml', resolver=MethodViewResolver('views'))
app = con.app
app.config.from_pyfile('config.py')
CORS(app, resources='/api/*', supports_credentials=True)
with app.app_context():
models.init_models(app)
return app
def process():
data = load_json(PATH_STEP2_CLEAN)
kFold = MyKFold(10, shuffle=True)
print("MLE generating sentences using Language model.")
for train_tweets, _ in kFold(data):
models = init_models(3, MLE)
train_sents = compress(train_tweets)
# padded multiple models, returning a list of (ngram, vocab)
ngrams = padded_multiple_models(3, train_sents)
# train models using ngrams
fit_multiple_models(models, ngrams)
for i in range(3):
print(f"\nGenerating {Ngrams(i)} sentences:")
for j in range(10): # generate 10 sentences for each ngram.
new_sentence = []
word = models[i].generate(text_seed=['<s>'])
new_sentence.append(word)
while word != '</s>':
word = models[i].generate(text_seed=[word])
new_sentence.append(word)
print(f"#{j}: [{' '.join(new_sentence)}]")
break
def __init__(self, args, env):
self.args = args
assert (self.args.sync)
self.env = env
self.max_steps = self.args.max_steps
self.guides = generate_guide_grid(args.bin_divide)
self.bmanager = BufferManager(args) # spc buffer manager
self.amanager = ActionSampleManager(args,
self.guides) # action sampler
self.model, self.optim, self.epoch, self.exploration, self.num_steps = init_models(
self.args)
self.env.set_epoch(self.epoch)
# set logger, by default using wandb
self.logger = WandBLogger(
os.path.join(args.save_path, self.args.logger_path),
self.args.wandb)
# import some frequently used params
self.bsize = self.args.batch_size
self.pstep = self.args.pred_step
self.img_h = self.args.frame_height
self.img_w = self.args.frame_width
self.classes = self.args.classes # class number of pixel semantic labels
# define loss functions used
self.event_loss_func = nn.CrossEntropyLoss
self.guide_loss_func = nn.CrossEntropyLoss
self.speed_loss_func = nn.MSELoss
self.seg_loss_func = nn.NLLLoss
self.depth_loss_func = nn.L1Loss()
self.detect_loss_func = FocalLoss()
self.coll_with_loss_func = nn.CrossEntropyLoss
# figure out predictive task list
self.eventloss_weights = dict() # filed -> loss weight
self.speedloss_weight = 0.01
self.segloss_weight = 1.0
# if self.args.use_detection: self.eventloss_weights['detection'] = 1.0
# if self.args.use_colls_with: self.eventloss_weights['colls_with'] = 1.0
if self.args.use_collision: self.eventloss_weights['coll'] = 1.0
# if self.args.use_collision_other: self.eventloss_weights['coll_other'] = 0.5
if self.args.use_offroad: self.eventloss_weights['offroad'] = 1.0
if self.args.use_offlane: self.eventloss_weights['offlane'] = 0.2
self.timer = None
self.last_episode_step = 0
def process():
# 1. First load the data into memory
data = load_json(PATH_STEP2_CLEAN)
# 2. Making 10-Fold Cross Validation
kFold = MyKFold(10, False)
print("Starting 10-Fold CV training/test")
means = np.zeros((10, 3))
for idx, (train_tweets, test_tweets) in enumerate(kFold(data)):
print(f"Fold {idx}:")
models = init_models(3, KneserNeyInterpolated)
train_sents = compress(train_tweets)
test_sents = compress(test_tweets)
# padded multiple models, returning a list of (ngram, vocab)
ngrams = padded_multiple_models(3, train_sents)
# train models using ngrams
fit_multiple_models(models, ngrams)
test_ngrams = padded_multiple_models(3, test_sents)
for n in range(0, 3):
temp = [(models[n]).perplexity(i)
for i in tqdm((test_ngrams[n])[0],
desc=f"perplexity of {Ngrams(n)}",
total=len(test_sents))]
means[idx, n] = np.mean(temp)
print(
f"run {idx}, unigram: {means[idx, 0]}, bigram: {means[idx, 1]}, trigram: {means[idx, 2]}"
)
final_means = np.mean(means, axis=0)
print("Final mean of 10-Fold CV:")
print(
f"unigram: {final_means[0]}, bigram: {final_means[1]}, trigram: {final_means[2]}"
)
def setUp(self):
models.Base.metadata.create_all(bind=library.db.engine)
self.app = library.app.test_client()
models.init_models(library.db.session)
self.session = library.db.session
def setUp(self):
models.Base.metadata.create_all(bind=library.db.engine)
self.app = library.app.test_client()
models.init_models(library.db.session)
models.init_test_models(library.db.session)
self.session = library.db.session
white_medium_star='\U00002B50',
heavy_minus_sign='\U00002796',
cinema='\U0001F3A6',
top_hat='\U0001F3A9',
ticket='\U0001F3AB',
performing_arts='\U0001F3AD',
speech_balloon='\U0001F4AC',
heavy_dollar_sign='\U0001F4B2',
money_bag='\U0001F4B0',
floppy_disk='\U0001F4BE',
calendar='\U0001F4C5',
tear_off_calendar='\U0001F4C6',
pushpin='\U0001F4CC',
telephone_receiver='\U0001F4DE',
memo='\U0001F4DD',
thought_balloon='\U0001F4AD',
bicyclist='\U0001F6B4',
taxi='\U0001F695',
oncoming_taxi='\U0001F696'))
if __name__ == '__main__':
session = init_models(config.db_engine_url)
token = sys.argv[1]
bot = Bot(token, session, data)
bot.run()
print('Listening ...')
while True:
time.sleep(10)
def train(real, opt):
""" Wrapper function for training. Calculates necessary scales then calls train_single_scale on each. """
generators = []
noise_maps = []
noise_amplitudes = []
if opt.game == 'mario':
token_group = MARIO_TOKEN_GROUPS
else: # if opt.game == 'mariokart':
token_group = MARIOKART_TOKEN_GROUPS
scales = [[x, x] for x in opt.scales]
opt.num_scales = len(scales)
if opt.game == 'mario':
scaled_list = special_mario_downsampling(opt.num_scales, scales, real,
opt.token_list)
else: # if opt.game == 'mariokart':
scaled_list = special_mariokart_downsampling(opt.num_scales, scales,
real, opt.token_list)
reals = [*scaled_list, real]
# If (experimental) token grouping feature is used:
if opt.token_insert >= 0:
reals = [(token_to_group(r, opt.token_list, token_group)
if i < opt.token_insert else r) for i, r in enumerate(reals)]
reals.insert(
opt.token_insert,
token_to_group(reals[opt.token_insert], opt.token_list,
token_group))
input_from_prev_scale = torch.zeros_like(reals[0])
stop_scale = len(reals)
opt.stop_scale = stop_scale
# Log the original input level as an image
img = opt.ImgGen.render(one_hot_to_ascii_level(real, opt.token_list))
wandb.log({"real": wandb.Image(img)}, commit=False)
os.makedirs("%s/state_dicts" % (opt.out_), exist_ok=True)
# Training Loop
divergences = []
for current_scale in range(0, stop_scale):
opt.outf = "%s/%d" % (opt.out_, current_scale)
try:
os.makedirs(opt.outf)
except OSError:
pass
# If we are seeding, we need to adjust the number of channels
if current_scale < (opt.token_insert + 1): # (stop_scale - 1):
opt.nc_current = len(token_group)
# Initialize models
D, G = init_models(opt)
# If we are seeding, the weights after the seed need to be adjusted
if current_scale == (opt.token_insert + 1): # (stop_scale - 1):
D, G = restore_weights(D, G, current_scale, opt)
# Actually train the current scale
z_opt, input_from_prev_scale, G, divs = train_single_scale(
D, G, reals, generators, noise_maps, input_from_prev_scale,
noise_amplitudes, opt)
# Reset grads and save current scale
G = reset_grads(G, False)
G.eval()
D = reset_grads(D, False)
D.eval()
generators.append(G)
noise_maps.append(z_opt)
noise_amplitudes.append(opt.noise_amp)
divergences.append(divs)
torch.save(noise_maps, "%s/noise_maps.pth" % (opt.out_))
torch.save(generators, "%s/generators.pth" % (opt.out_))
torch.save(reals, "%s/reals.pth" % (opt.out_))
torch.save(noise_amplitudes, "%s/noise_amplitudes.pth" % (opt.out_))
torch.save(opt.num_layer, "%s/num_layer.pth" % (opt.out_))
torch.save(opt.token_list, "%s/token_list.pth" % (opt.out_))
wandb.save("%s/*.pth" % opt.out_)
torch.save(G.state_dict(),
"%s/state_dicts/G_%d.pth" % (opt.out_, current_scale))
wandb.save("%s/state_dicts/*.pth" % opt.out_)
del D, G
torch.save(torch.tensor(divergences), "%s/divergences.pth" % opt.out_)
return generators, noise_maps, reals, noise_amplitudes
def train_policy(args, env, max_steps=40000000):
guides = generate_guide_grid(args.bin_divide)
train_net, net, optimizer, epoch, exploration, num_steps = init_models(args)
buffer_manager = BufferManager(args)
action_manager = ActionSampleManager(args, guides)
action_var = Variable(torch.from_numpy(np.array([-1.0, 0.0])).repeat(1, args.frame_history_len - 1, 1), requires_grad=False).float()
# prepare video recording
if args.recording:
video_folder = os.path.join(args.video_folder, "%d" % num_steps)
os.makedirs(video_folder, exist_ok=True)
if args.sync:
video = cv2.VideoWriter(os.path.join(video_folder, 'video.avi'),
cv2.VideoWriter_fourcc(*'MJPG'),
24.0, (args.frame_width, args.frame_height), True)
else:
video = None
signal = mp.Value('i', 1)
p = mp.Process(target=record_screen,
args=(signal,
os.path.join(video_folder, 'video.avi'),
1280, 800, 24))
p.start()
# initialize environment
obs, info = env.reset()
if args.recording:
log_frame(obs, buffer_manager.prev_act, video_folder, video)
num_episode = 1
print('Start training...')
for step in range(num_steps, max_steps):
obs_var = buffer_manager.store_frame(obs, info)
action, guide_action = action_manager.sample_action(net=net,
obs=obs,
obs_var=obs_var,
action_var=action_var,
exploration=exploration,
step=step,
explore=num_episode % 2)
obs, reward, done, info = env.step(action)
print("action [{0:.2f}, {1:.2f}]".format(action[0], action[1]) + " " +
"collision {}".format(str(bool(info['collision']))) + " " +
"off-road {}".format(str(bool(info['offroad']))) + " " +
"speed {0:.2f}".format(info['speed']) + " " +
"reward {0:.2f}".format(reward) + " " +
"explore {0:.2f}".format(exploration.value(step))
)
action_var = buffer_manager.store_effect(guide_action=guide_action,
action=action,
reward=reward,
done=done,
collision=info['collision'],
offroad=info['offroad'])
if args.recording:
log_frame(obs, action, video_folder, video)
if done:
print('Episode {} finished'.format(num_episode))
if not args.sync and args.recording:
signal.value = 0
p.join()
del p
# train SPN
if buffer_manager.spc_buffer.can_sample(args.batch_size) and ((not args.sync and done) or (args.sync and step % args.learning_freq == 0)):
# train model
for ep in range(args.num_train_steps):
optimizer.zero_grad()
loss = train_model(args=args,
net=train_net,
spc_buffer=buffer_manager.spc_buffer)
if args.use_guidance:
loss += train_guide_action(args=args,
net=train_net,
spc_buffer=buffer_manager.spc_buffer,
guides=guides)
print('loss = %0.4f\n' % loss.data.cpu().numpy())
loss.backward()
optimizer.step()
epoch += 1
net.load_state_dict(train_net.state_dict())
# save model
if epoch % args.save_freq == 0:
print(color_text('Saving models ...', 'green'))
torch.save(train_net.module.state_dict(),
os.path.join(args.save_path, 'model', 'pred_model_%09d.pt' % step))
torch.save(optimizer.state_dict(),
os.path.join(args.save_path, 'optimizer', 'optimizer.pt'))
with open(os.path.join(args.save_path, 'epoch.pkl'), 'wb') as f:
pkl.dump(epoch, f)
buffer_manager.save_spc_buffer()
print(color_text('Model saved successfully!', 'green'))
if done:
# reset video recording
if args.recording:
if args.sync:
video.release()
if sys.platform == 'linux': # save memory
os.system('ffmpeg -y -i {0} {1}'.format(
os.path.join(video_folder, 'video.avi'),
os.path.join(video_folder, 'video.mp4')
))
if os.path.exists(os.path.join(video_folder, 'video.mp4')):
os.remove(os.path.join(video_folder, 'video.avi'))
video_folder = os.path.join(args.video_folder, "%d" % step)
os.makedirs(video_folder, exist_ok=True)
video = cv2.VideoWriter(os.path.join(video_folder, 'video.avi'),
cv2.VideoWriter_fourcc(*'MJPG'),
24.0, (args.frame_width, args.frame_height), True)
else:
video_folder = os.path.join(args.video_folder, "%d" % step)
os.makedirs(video_folder, exist_ok=True)
signal.value = 1
p = mp.Process(target=record_screen,
args=(signal, os.path.join(video_folder, 'obs.avi'), 1280, 800, 24))
p.start()
num_episode += 1
obs, info = env.reset()
buffer_manager.reset(step)
action_manager.reset()
if args.recording:
log_frame(obs, buffer_manager.prev_act, video_folder, video)
import os
from meme_bot import MemeBot
from models import init_models
BOT_TOKEN = os.getenv("BOT_TOKEN")
bot = MemeBot(BOT_TOKEN)
@bot.message_handler(content_types=['text', 'video', 'photo', 'document'])
def message_handler(message):
bot.meme_handler(message)
@bot.callback_query_handler(func=lambda call: True)
def callback_handler(call):
bot.vote_handler(call)
if __name__ == '__main__':
print('Start', flush=True)
init_models()
bot.run_pooling()
bot.run_scheduler()
def create_app():
app = Flask(__name__) # Создаем экземпляр класса Flask-приложения
app.url_map.strict_slashes = local_settings.TRAILING_SLASH # Указываем игнорирововать слеша в конце url
app.config.from_object(
local_settings) # Передаём остальные настройки в приложение
return app
APP = create_app() # Инициируем приложение
DB = Database(
APP
) # Инициируем работу с БД. Тут же создаюётся таблицы, если их нет в БД.
init_models(DB)
API = RestAPI(APP) # Инициируем RestAPI от peewee
init_api(API)
ADMIN = init_admin(APP, DB) # Инициируем Админку
import ui.root # Импортируем view для главной страницы
# Api на flask_restful и роуты для API
from flask_restful import Api
api = Api(APP)
from services.product import GetProducts, AddProduct, DeleteProduct, UpdateProduct
api.add_resource(GetProducts, '/product/get')
def evaluate_policy(args, env):
guides = generate_guide_grid(args.bin_divide)
args.checkpoint = args.checkpoint
net, optimizer, epoch, exploration, num_steps = init_models(args)
output_path = args.output_path
for episode in range(100):
buffer_manager = BufferManager(args)
action_manager = ActionSampleManager(args, guides)
action_var = torch.from_numpy(np.array([-1.0, 0.0])).repeat(
1, args.frame_history_len - 1, 1).float()
# initialize environment
obs_ori, info = env.reset()
obs = obs_ori.reshape(
(args.frame_height, 4, args.frame_width, 4, 3)).max(3).max(1)
info['seg'].resize(args.frame_height, args.frame_width)
encoder = DataEncoder()
print('Start episode...')
for step in range(args.max_eval_step):
obs_var = buffer_manager.store_frame(obs, info)
action, guide_action, p = action_manager.sample_action(
net=net,
obs=obs,
obs_var=obs_var,
action_var=action_var,
exploration=exploration,
step=step,
explore=False,
testing=True)
# in the test mode, sample_action outputs the guide_action and action
# for future pred_step steps, while we only take those for the next frame
# to execute and store into buffer
output = net_infer(args, obs_var, action, net, action_var)
bboxes, labels, scores = [], [], []
if args.use_detection:
loc_preds, cls_preds = output['loc_pred'][0].cpu(
), output['cls_pred'][0].cpu()
for find in range(loc_preds.size(0)):
print(cls_preds[find].sigmoid().max())
frame_loc_pred = loc_preds[find]
frame_cls_pred = cls_preds[find]
pred_bboxes, pred_labels, pred_scores = encoder.decode(
frame_loc_pred,
frame_cls_pred,
input_size=(args.frame_width, args.frame_height))
bboxes.append(pred_bboxes)
labels.append(pred_labels)
scores.append(pred_scores)
draw_prediction(step, args, output, bboxes, scores, 'outcome',
'{}/{}'.format(output_path, episode))
# in the testing mode, $action and $guide_action are for multiple samples, we only pick up the first one
guide_action = guide_action[0]
action = action[0]
obs_ori, reward, done, info = env.step(action)
obs = obs_ori.reshape(
(args.frame_height, 4, args.frame_width, 4, 3)).max(3).max(1)
info['seg'].resize(args.frame_height, args.frame_width)
draw_current_frame(
args, action, obs_ori, p, None, bboxes, scores,
os.path.join(output_path, str(episode), str(step)), step)
print(
"step: {0} | action [{1:.2f}, {2:.2f}] coll {3} offroad {4} offlane {5} speed {6:.2f} reward {7:.2f}"
.format(step, action[0], action[1], info['collision'],
info['offroad'], info['offlane'], info['speed'],
reward))
action_var = buffer_manager.store_effect(guide_action=guide_action,
action=action,
reward=reward,
done=done,
info=info)
if done:
break
'MONGODB_PORT': 27017,
'MONGODB_DB_NAME': 'eta_app',
'MONGODB_SEGMENT_COLLECTION': 'segments',
'MONGODB_VEHICLE_COLLECTION': 'vehicles',
'MONGOENGINE_DB_ALIAS': 'default',
'JSON_ROUTE_DATA': 'ikot_route_test.geojson',
'ZOMBIE_THRESHOLD_TIMEDELTA': 5
}
# configure timezone
os.environ['TZ'] = "Asia/Manila"
me_connection = me.register_connection(db=config['MONGODB_DB_NAME'],
alias=config['MONGOENGINE_DB_ALIAS'])
models = models.init_models(config)
# current_folder = Path(os.getcwd())
# outer_folder = current_folder.parent
# update_segment_folder = Path(str(outer_folder), "metrics/data/network_computation/update_segment")
app_directory = os.getcwd()
update_segment_folder = os.path.join(
app_directory, 'metrics/data/network_computation/update_segment')
# print(update_segment_folder)
# Updates a Segment (given seg_id) in the database from traffic data in NIMPA
# Flow:
# 1. fetch the desired Segment
from flask import Flask, request, jsonify, abort
from models import init_app as init_models, Note, db
from datetime import datetime, timedelta
app = Flask(__name__)
init_models(app)
def note_to_json(note: Note):
return {
'id': note.id,
'kind': note.kind,
'content': note.content,
'lat': note.lat,
'lon': note.long,
'altitude': note.altitude,
'expires': note.expires,
'color': note.colour
}
@app.route('/notes', methods=['GET'])
def note_list():
long = float(request.args['long'])
lat = float(request.args['lat'])
resolution = float(request.args.get('resolution', '0.001'))
# TODO: this will fail opposite the prime meridian and/or at the poles
notes = Note.query.filter(Note.long >= long - resolution,
Note.long <= long + resolution,
Note.lat >= lat - resolution,
Note.lat <= Note.lat + resolution).all()
def evaluate_policy(args, env):
guides = generate_guide_grid(args.bin_divide)
train_net, net, optimizer, epoch, exploration, num_steps = init_models(
args)
buffer_manager = BufferManager(args)
action_manager = ActionSampleManager(args, guides)
action_var = Variable(torch.from_numpy(np.array([-1.0, 0.0])).repeat(
1, args.frame_history_len - 1, 1),
requires_grad=False).float()
# prepare video recording
if args.recording:
video_folder = os.path.join(args.video_folder, "%d" % num_steps)
os.makedirs(video_folder, exist_ok=True)
if args.sync:
video = cv2.VideoWriter(os.path.join(video_folder, 'video.avi'),
cv2.VideoWriter_fourcc(*'MJPG'), 24.0,
(args.frame_width, args.frame_height),
True)
else:
video = None
signal = mp.Value('i', 1)
p = mp.Process(target=record_screen,
args=(signal, os.path.join(video_folder,
'video.avi'), 1280, 800,
24))
p.start()
# initialize environment
obs, info = env.reset()
if args.recording:
log_frame(obs, buffer_manager.prev_act, video_folder, video)
print('Start episode...')
for step in range(1000):
obs_var = buffer_manager.store_frame(obs, info)
action, guide_action = action_manager.sample_action(
net=net,
obs=obs,
obs_var=obs_var,
action_var=action_var,
exploration=exploration,
step=step,
explore=False,
testing=True)
draw(action, step, obs_var, net, args, action_var, 'outcome')
cv2.imwrite(os.path.join('demo', str(step), 'obs.png'),
cv2.cvtColor(obs, cv2.COLOR_BGR2RGB))
obs, reward, done, info = env.step(action[0])
action_var = buffer_manager.store_effect(guide_action=guide_action,
action=action,
reward=reward,
done=done,
collision=info['collision'],
offroad=info['offroad'])
if args.recording:
log_frame(obs, action, video_folder, video)
if done:
print('Episode finished ...')
if args.recording:
if args.sync:
video.release()
if sys.platform == 'linux': # save memory
os.system('ffmpeg -y -i {0} {1}'.format(
os.path.join(video_folder, 'video.avi'),
os.path.join(video_folder, 'video.mp4')))
if os.path.exists(
os.path.join(video_folder, 'video.mp4')):
os.remove(os.path.join(video_folder, 'video.avi'))
else:
signal.value = 0
p.join()
del p
break
import asyncio
from config import Config
from parsers import SerialsParser, UpdateSerialParser
from models import Serial, LastUpdateHash, User, init_models
if __name__ == "__main__":
loop = asyncio.get_event_loop()
init_models(Config.RDB, loop)
cours = asyncio.gather(
# SerialsParser(Config).parse(),
UpdateSerialParser(Config).parse()
)
try:
loop.run_until_complete(cours)
finally:
loop.close()
