def increase_worddetails_occured_num(self, word_text, wordtext):
word_details_query = WordDetails.objects.filter(pub_date=get_occurence(self.datetime), origin=get_source(self.source_name), word=wordtext)
word_query_length = len(word_details_query)
if word_query_length == 0:
worddetails = WordDetails.objects.create(pub_date=get_occurence(self.datetime), occured_num=1, origin=get_source(self.source_name), word=wordtext)
logging.debug(
'Created and saved a new WordDetails, for word {0}, datetime "{1}" and source {2} '\
.format(word_text, worddetails.pub_date.timestamp, worddetails.origin.source))
if word_query_length == 1:
worddetails = word_details_query[0]
worddetails.occured_num += 1
worddetails.save()
logging.debug(
'Retrieved a WordDetails, for word {0}, datetime "{1}" and source {2} and increased occured_num up to {3}'\
.format(word_text, worddetails.pub_date.timestamp, worddetails.origin.source, worddetails.occured_num))
if word_query_length > 1:
worddetails = word_details_query[0]
logging.error(
'QUERY LENGTH FOR WORD: {1} ID {0}. IT SHOULD NOT BE 1 OR 0. RETURNED THE FIRST ONE'\
.format(len(word_details_query), word_text))
def save_sentence_and_link(self, fixed_sentence, link, article_title):
words_collocated_to_link = []
# source = self.get_site_name()
words_text_arr = fixed_sentence.split(' ')
for word_text in words_text_arr:
wordtext = self.create_wordtext_record_if_there_is_no_one_and_return(word_text)
worddetails = self.increase_worddetails_occured_num(word_text, wordtext)
words_collocated_to_link.append(wordtext)
link_ = Link.objects.create(pub_date=get_occurence(self.datetime), link=link, origin=get_source(self.source_name), title=article_title)
logging.debug(
'Created and saved a new link, link id is {0}, link is for "{1}" sentence '.format(link_.id, article_title))
for word in words_collocated_to_link:
link_.words.add(word)
logging.debug(
'Collocated {0} words to link '.format(len(words_collocated_to_link)))
