def run(self, profile_path="/data/firefox_cache"):
"""
Faz a coleta dos metadados de grupos de Whatsapp de acordo
com os parâmetros fornecidos na criação do objeto de coleta.
Parâmetros
------------
profile_path : str
Caminho para um profile alternativo do navegador
utilizado na coleta.
"""
today = datetime.date.today().strftime('%Y-%m-%d')
all_groups_filename = '/data/all_grupos_%s.json' % (today)
with open(all_groups_filename, 'w') as json_file:
print('Collecting metadata for groups at %s' % (today))
if not os.path.exists(profile_path):
os.makedirs(profile_path)
driver = WhatsAPIDriver(loadstyles=True,
profile=profile_path,
client="remote",
command_executor=os.environ["SELENIUM"])
try:
print("Waiting for WhatsApp Web Login")
driver.wait_for_login()
print("Saving session")
driver.save_firefox_profile(remove_old=False)
print("Bot started")
pathlib.Path("/data/grupos").mkdir(parents=True, exist_ok=True)
print('>>>>>>>>>>> Loading chat ids')
chats = driver.get_all_chats()
for chat in (chats):
# Does not collect direct messages, only group chats
if not chat._js_obj['isGroup']:
continue
gid = chat.id
gid = gid.split('@')[0]
s_name = self._process_string(chat.name)
# Skip group if it is on blacklist (can be name or groupID)
if (s_name in self.group_blacklist
or gid in self.group_blacklist):
continue
group = dict()
_id = chat.id
creator = _id.split('-')[0]
timestamp = _id.split('-')[-1].split('@')[0]
date = convert_data_from_timestamp(float(timestamp))
str_date = date.strftime('%Y-%m-%d %H:%M:%S')
name = chat.name.strip().replace('\t', ' ')
kind = chat._js_obj["kind"]
participants = list()
for member in driver.group_get_participants(_id):
user = dict()
user['name'] = member.verified_name
user['short_name'] = member.short_name
user['nome_formatado'] = member.formatted_name
user['number'] = member.id
user['isBusiness'] = member.is_business
user['profile_pic'] = member.profile_pic
participants.append(user)
group['group_id'] = _id
group['creator'] = creator
group['kind'] = kind
group['creation'] = dict()
group['creation']['creation_date'] = str_date
group['creation']['creation_timestamp'] = timestamp
group['title'] = name
group['members'] = participants
path = '/data/grupos/'
filename = '%sgrupos_%s.json' % (path,
_id.split('@')[0].strip())
print(group)
with open(filename, 'w') as json_file:
json.dump(group, json_file)
print('', file=json_file)
with open(all_groups_filename, 'a') as json_file:
json.dump(group, json_file)
print('', file=json_file)
driver.close()
except Exception as e:
print(e)
driver.close()
raise Exception(e)
def run(self, profile_path="/data/firefox_cache"):
"""
Faz a coleta das mensagens de grupos de Whatsapp de acordo
com os parâmetros fornecidos na criação do objeto de coleta.
Parâmetros
------------
profile_path : str
Caminho para um profile alternativo do navegador
utilizado na coleta.
"""
if not os.path.exists(profile_path):
os.makedirs(profile_path)
driver = WhatsAPIDriver(loadstyles=True, profile=profile_path,
client="remote",
command_executor=os.environ["SELENIUM"])
pathlib.Path("/data/mensagens").mkdir(parents=True, exist_ok=True)
pathlib.Path("/data/image").mkdir(parents=True, exist_ok=True)
pathlib.Path("/data/audio").mkdir(parents=True, exist_ok=True)
pathlib.Path("/data/video").mkdir(parents=True, exist_ok=True)
pathlib.Path("/data/mensagens_grupo").mkdir(parents=True, exist_ok=True)
pathlib.Path("/data/notificacoes").mkdir(parents=True, exist_ok=True)
pathlib.Path("/data/mids").mkdir(parents=True, exist_ok=True)
min_date = self.start_date
max_date = self.end_date
include_notf = self.collect_notifications
looping = True
if (self.collection_mode == 'period') and (min_date < '2020-01-01'):
raise Exception("Can't start collection without a start and end"
" date.")
while looping:
if self.collection_mode == 'continuous':
looping = True
else:
looping = False
try:
print("Waiting for WhatsApp Web Login")
driver.wait_for_login()
print("Saving session")
driver.save_firefox_profile(remove_old=False)
print("Bot started")
print('>>>>>>>>>>> Loading previous saved Messages')
messagesID = self._get_load_messages()
notificationsID = self._get_load_notifications()
today_date = datetime.date.today().strftime("%Y-%m-%d")
date_format = "%Y-%m-%d"
file_name = "/data/mensagens/mensagens_" + today_date + ".json"
start_date = min_date
print('>>>>>>>>>>>>Getting Groups Messages...', end=' ')
chats = driver.get_all_chats()
count = 0
all_chats = list(chats)
print(' DONE! %d chats loaded!' % (len(all_chats)))
random.shuffle(all_chats)
for chat in (all_chats):
# Does not collect direct messages, only group chats
if not chat._js_obj['isGroup']:
continue
gid = chat.id
gid = gid.split('@')[0]
s_name = self._process_string(chat.name)
# Skip group if it is on blacklist (can be name or groupID)
if (s_name in self.group_blacklist or
gid in self.group_blacklist):
continue
# PRINT CHAT INFORMATION
members = chat._js_obj['groupMetadata']['participants']
timestamp = gid.split('-')[-1]
date = convert_data_from_timestamp(float(timestamp))
str_date = date.strftime('%Y-%m-%d %H:%M:%S')
chat_print = "<Group chat - {name}: {id}, {participants} " \
"participants - at {time}!!>".format(
name=s_name, id=gid, participants=len(members),
time=str_date)
print('>>>>>Loading messages from', chat_print)
if gid not in messagesID:
messagesID[gid] = dict()
messagesID[gid]['messages'] = set()
messagesID[gid]['date'] = '2000-01-01'
# PROCESS PREVIOUS LOADED MESSAGES ID AND LAST DATE
if self.collection_mode == 'continuous':
if messagesID[gid]['date'] > max_date:
continue
if messagesID[gid]['date'] > min_date:
start_date = messagesID[gid]['date']
till_date = datetime.datetime.strptime(start_date,
date_format)
else:
start_date = min_date
till_date = datetime.datetime.strptime(start_date,
date_format)
# LOAD MESSAGES FROM WHATSAPP SINCE MIN_DATE
messages = chat.load_earlier_messages_till(till_date)
messages = driver.get_all_message_ids_in_chat(
chat, include_notifications=include_notf)
elif self.collection_mode == 'period':
till_date = datetime.datetime.strptime(start_date,
date_format)
# LOAD MESSAGES FROM WHATSAPP SINCE MIN_DATE
messages = chat.load_earlier_messages_till(till_date)
messages = driver.get_all_message_ids_in_chat(
chat, include_notifications=include_notf)
elif self.collection_mode == 'unread':
# LOAD UNREAD MESSAGES FROM WHATSAPP
messages = chat.get_unread_messages(
include_me=False,
include_notifications=include_notf)
print('>>>>>Total messages %d' % (len(messages)))
count += 1
for msg in messages:
count += 1
gid = gid.split('@')[0]
mid = msg
if self._is_notification(mid):
if gid not in notificationsID.keys():
notificationsID[gid] = set()
if mid.strip() in notificationsID[gid]:
continue
j = driver.get_message_by_id(mid)
self._save_notification_(j, gid)
continue
if mid.strip() in messagesID[gid]['messages']:
print('Message: %d >>> %s from %s was CHECKED' %
(count, mid, gid))
continue
else:
try:
j = driver.get_message_by_id(mid)
except Exception as e:
print('Error getting a message >>', e)
continue
if not j:
continue
sender = j.sender.id
sender = sender.replace(' ', '').strip()
sender = sender.split('@')[0]
if (sender in self.user_blacklist or
'+' + sender in self.user_blacklist):
continue
try:
date = self._get_date_from_message(j)
except Exception:
continue
if (date > max_date) and (self.collection_mode == 'period'):
break
if (date < start_date):
continue
# Update day
if today_date != date:
today_date = date
file_name = "/data/mensagens/mensagens_" + today_date + ".json"
if self.collect_images:
try:
self._get_image_from_message(j)
except Exception as ei:
print('!!!!Error getting image!!!! ', ei)
if self.collect_videos:
try:
self._get_video_from_message(j)
except Exception as ev:
print('!!!!Error getting video!!!! ', ev)
if self.collect_audios:
try:
self._get_audio_from_message(j)
except Exception as ea:
print('!!!!Error getting audio!!!! ', ea)
if self.collect_messages:
self._save_message(j, s_name, gid, mid, file_name)
driver.close()
except Exception as e:
print(e)
driver.close()
raise Exception(e)
if looping:
print('Waiting code to start again...')
time.sleep(3600)
# Deactivating reload
with db_conn.cursor() as cur:
cur.execute(deactivate_reload, (
process_note,
reload_contact_row_id,
))
db_conn.commit()
logging.info("Reloading deactivated for " +
str(reload_contact_row_sender))
chat.send_seen()
else:
logging.debug("Nothing to reload, continue")
except Exception as e:
logging.exception(e)
if 'driver' in locals() and driver is not None:
driver.close()
logging.info('Selenium driver connection closed')
if 'db_conn' in locals() and db_conn is not None:
db_conn.close()
logging.info('Database connection closed')
raise
finally:
if 'driver' in locals() and driver is not None:
driver.close()
logging.info('Selenium driver connection closed')
if 'db_conn' in locals() and db_conn is not None:
db_conn.close()
logging.info('Database connection closed')
class WPChannelBot():
def __init__(self):
self.model = WPChannelBotModel()
self.data = self.model.get_all()
self.convs = self.model.get_convs()
self.convs_state = self.model.get_convs_state()
self.simple_steps = True
self.log_file = "log/chatbot.log"
self.cmd_wait_from = None
self.cmd_wait = False
self.profile = "profile"
self.driver = None
def start(self):
print("Iniciando bot...")
self.driver = WhatsAPIDriver(profile=self.profile)
time.sleep(3)
if not self.driver.get_status() == "LoggedIn":
print("Carregando QRCode")
self.driver.get_qr("qrcode.png")
print("Escaneie o QRCode no arquivo qrcode.png")
self.driver.wait_for_login()
print("Bot iniciado")
self.driver.save_firefox_profile()
while True:
time.sleep(1)
for contact in self.driver.get_unread(include_me=False,
include_notifications=True,
use_unread_count=True):
if len(contact.messages) == 1:
for message in contact.messages:
if isinstance(message, Message):
self.new_message(message.content, contact)
self.driver.chat_send_seen(contact.chat.id)
time.sleep(3)
else:
contact.chat.send_message(
"Fico confuso com muitas mensagens :S Por favor, envie uma de cada vez e espere eu responder tá?"
)
contact.chat.send_message(CHANNEL_ASK_KEYWORD)
def new_message(self, message, contact):
if not self._is_cmd(message):
if self.cmd_wait and contact.chat.id == self.cmd_wait_from:
self._cmd_envio(message, contact.chat)
elif not contact.chat.id in self.convs:
self._proc_etapa(contact.chat.id, message, contact.chat, 2)
else:
for conv in self.convs_state:
if conv['id'] == contact.chat.id:
e = self._proc_etapa(contact.chat.id, message,
contact.chat, conv['etapa'])
conv['etapa'] = e
self.model.conv_update(contact.chat.id, e)
else:
print("ADMINISTRADOR")
self._run_cmd(message, contact.chat)
def shutdown(self):
print("Desconectando...")
self.driver.close()
time.sleep(3)
print("Desconectado")
def _already_user(self, id, chat):
if isinstance(self.model.get(id), dict):
chat.send_message(
"Olá, você já está cadastrado neste canal. Assim que tiver novidade você vai receber!"
)
return True
else:
return False
def _is_keyword(self, content, chat):
if content.lower() == CHANNEL_KEYWORD:
return True
else:
chat.send_message(CHANNEL_ASK_KEYWORD)
return False
def _proc_etapa(self, id, content, chat, etapa):
if etapa == 2:
if not self._already_user(id, chat) and self._is_keyword(
content, chat):
# Efetua registros
self.convs.append(id)
self.convs_state.append({"id": id, "etapa": 4})
self.model.conv_add(id, 4)
# Introdução do canal - Solicita nome
chat.send_message(CHANNEL_INTRO)
chat.send_message(CHANNEL_MSGS[0])
self._to_log("Iniciando cadastro: %s" % id)
elif etapa == 4:
# Armazena nome - Solicita cidade
if self.simple_steps:
self.data.append({"id": id, "nome": content})
# Salva no banco de dados
self.model.add(id, content)
chat.send_message((CHANNEL_MSGS[3] % content))
self._remove_convs(id)
self._to_log("Finalizado cadastro: %s - %s" % (id, content))
else:
self.data.append({
"id": id,
"nome": content,
"cidade": "",
"bairro": ""
})
chat.send_message(CHANNEL_MSGS[1])
# Salva no banco de dados
self.model.add(id, content)
self._to_log("Registrado nome: %s - %s" % (id, content))
return 6
elif etapa == 6:
# Implementar veficação de validade de cidade
# Verifica cidade - volta ao 5 : armazena cidade - solicita bairro ou passo
for obj in self.data:
if obj["id"] == id:
obj["cidade"] = content
self.model.update(id=id, cidade=content)
chat.send_message(CHANNEL_MSGS[2])
self._to_log("Registrado cidade: %s - %s" % (id, content))
return 7
elif etapa == 7:
# Implementar veficação de validade de bairro
if content == "passo":
# Finaliza caso não seja informado bairro
chat.send_message((CHANNEL_MSGS[3] % self._get_conv_nome(id)))
self._remove_convs(id)
self._to_log("Finalizado cadastro: %s - %s" % (id, content))
else:
# Armazena bairro - Finaliza cadastro
for obj in self.data:
if obj["id"] == id:
obj["bairro"] = content
self.model.update(id=id, bairro=content)
chat.send_message(
(CHANNEL_MSGS[3] % self._get_conv_nome(id)))
self._remove_convs(id)
self._to_log("Finalizado cadastro: %s - %s" %
(id, content))
def _to_log(self, log):
file = open(self.log_file, "a")
file.write("\n>> %s " % log)
file.close()
return
def _get_conv_nome(self, id):
for obj in self.data:
if obj["id"] == id:
return obj["nome"]
def _remove_convs(self, id):
self.convs.remove(id)
for conv in self.convs_state:
if conv["id"] == id:
self.convs_state.remove(conv)
self.model.conv_delete(id)
def _is_cmd(self, content):
if content[:4] == "/cmd":
return True
else:
return False
def _run_cmd(self, content, chat):
cmd = content[5:]
if not self.model.check_admin(chat.id) == False:
if cmd == "usuarios":
self._cmd_usuarios(chat)
elif cmd == "envio":
self.cmd_wait = True
self.cmd_wait_from = chat.id
chat.send_message(
"*ENVIE A SEGUIR A MENSAGEM A SER ENVIADA PARA O CANAL*")
else:
chat.send_message("*COMANDO NÃO RECONHECIDO*")
elif self.model.check_admin(id=None,
all=True) == False and cmd[:5] == "admin":
print("Cadastrando novo admin")
self.model.add_admin(chat.id, content[11:])
chat.send_message("*ADMINISTRADOR CADASTRADO*")
else:
chat.send_message(CHANNEL_ASK_KEYWORD)
def _cmd_usuarios(self, chat):
response = "*USUÁRIOS CADASTRADOS*\n\n"
i = 0
users = self.model.get_all()
for user in users:
i += 1
response += "\n%d) %s - %s" % (i, user['id'], user['nome'])
chat.send_message(response)
def _cmd_envio(self, content, chat):
i = 0
users = self.model.get_all()
for user in users:
i += 1
self.driver.send_message_to_id(user['id'], content)
self.cmd_wait_from = None
self.cmd_wait = False
chat.send_message("*MENSAGEM ENVIADA PARA %d USUÁRIOS DO CANAL*" % i)
