def compute(self, item_id):
# refresh Tracked term
if self.last_refresh_word < Term.get_tracked_term_last_updated_by_type(
'word'):
self.list_tracked_words = Term.get_tracked_words_list()
self.last_refresh_word = time.time()
self.redis_logger.debug('Tracked word refreshed')
print('Tracked word refreshed')
if self.last_refresh_set < Term.get_tracked_term_last_updated_by_type(
'set'):
self.set_tracked_words_list = Term.get_set_tracked_words_list()
self.last_refresh_set = time.time()
self.redis_logger.debug('Tracked set refreshed')
print('Tracked set refreshed')
# Cast message as Item
item = Item(item_id)
item_date = item.get_date()
item_content = item.get_content()
signal.alarm(self.max_execution_time)
dict_words_freq = None
try:
dict_words_freq = Term.get_text_word_frequency(item_content)
except TimeoutException:
self.redis_logger.warning(f"{item.get_id()} processing timeout")
else:
signal.alarm(0)
if dict_words_freq:
# create token statistics
# for word in dict_words_freq:
# Term.create_token_statistics(item_date, word, dict_words_freq[word])
item_source = item.get_source()
# check solo words
####### # TODO: check if source needed #######
for word in self.list_tracked_words:
if word in dict_words_freq:
self.new_term_found(word, 'word', item.get_id(),
item_source)
# check words set
for elem in self.set_tracked_words_list:
list_words = elem[0]
nb_words_threshold = elem[1]
word_set = elem[2]
nb_uniq_word = 0
for word in list_words:
if word in dict_words_freq:
nb_uniq_word += 1
if nb_uniq_word >= nb_words_threshold:
self.new_term_found(word_set, 'set', item.get_id(),
item_source)
def __init__(self):
super(Tracker_Term, self).__init__()
self.pending_seconds = 5
self.max_execution_time = self.process.config.getint(
'Tracker_Term', "max_execution_time")
self.full_item_url = self.process.config.get(
"Notifications", "ail_domain") + "/object/item?id="
# loads tracked words
self.list_tracked_words = Term.get_tracked_words_list()
self.last_refresh_word = time.time()
self.set_tracked_words_list = Term.get_set_tracked_words_list()
self.last_refresh_set = time.time()
self.redis_logger.info(f"Module: {self.module_name} Launched")
def __init__(self):
super(TermTrackerMod, self).__init__()
self.pending_seconds = 5
self.max_execution_time = self.process.config.getint(
'TermTrackerMod', "max_execution_time")
self.full_item_url = self.process.config.get(
"Notifications", "ail_domain") + "/object/item?id="
# loads tracked words
self.list_tracked_words = Term.get_tracked_words_list()
self.last_refresh_word = time.time()
self.set_tracked_words_list = Term.get_set_tracked_words_list()
self.last_refresh_set = time.time()
# Send module state to logs
self.redis_logger.info("Module %s initialized" % (self._module_name()))
import NotificationHelper
from packages import Item
from packages import Term
sys.path.append(os.path.join(os.environ['AIL_FLASK'], 'modules'))
import Flask_config
full_item_url = "/showsavedpaste/?paste="
mail_body_template = "AIL Framework,\nNew occurrence for term tracked term: {}\nitem id: {}\nurl: {}{}"
# loads tracked words
list_tracked_words = Term.get_tracked_words_list()
last_refresh_word = time.time()
set_tracked_words_list = Term.get_set_tracked_words_list()
last_refresh_set = time.time()
class TimeoutException(Exception):
pass
def timeout_handler(signum, frame):
raise TimeoutException
signal.signal(signal.SIGALRM, timeout_handler)
def new_term_found(term, term_type, item_id, item_date):