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)
new_term_found(word, 'word', item_id, item_date)
# check words set
for elem in 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:
new_term_found(word_set, 'set', item_id, item_date)
else:
time.sleep(5)
# refresh Tracked term
if last_refresh_word < Term.get_tracked_term_last_updated_by_type(
'word'):
list_tracked_words = Term.get_tracked_words_list()
last_refresh_word = time.time()
print('Tracked word refreshed')
if last_refresh_set < Term.get_tracked_term_last_updated_by_type(
'set'):
set_tracked_words_list = Term.get_set_tracked_words_list()
last_refresh_set = time.time()
print('Tracked set refreshed')
# Regex Frequency
while True:
item_id = p.get_from_set()
if item_id is not None:
item_date = Item.get_item_date(item_id)
item_content = Item.get_item_content(item_id)
for regex in dict_regex_tracked:
matched = regex_helper.regex_search(
module_name,
redis_cache_key,
dict_regex_tracked[regex],
item_id,
item_content,
max_time=max_execution_time)
if matched:
new_term_found(regex, 'regex', item_id, item_date)
else:
time.sleep(5)
# refresh Tracked term
if last_refresh < Term.get_tracked_term_last_updated_by_type('regex'):
dict_regex_tracked = Term.get_regex_tracked_words_dict()
last_refresh = time.time()
print('Tracked set refreshed')