def worker(news_text):
key_word_list = key_words.get_key_word(news_text.text,
news_text.news.title)
for word in key_word_list:
key_word, is_created = KeyWord.objects.get_or_create(word=word)
key_word.news.add(news_text.news)
news_text.is_keywords_extracted = True
news_text.save()
return news_text.pk, len(key_word_list)
def key_words_create():
for newstext in NewsText.objects.filter(is_keywords_extracted=False).iterator():
if newstext.text:
key_word_list = key_words.get_key_word(newstext.text, newstext.news.title)
print(key_word_list)
for word in key_word_list:
try:
key_word = KeyWord.objects.get(word=word)
except:
key_word = KeyWord(word=word)
key_word.save()
key_word.news.add(newstext.news)
def worker(work_obj):
news_pk, news_title, news_text = work_obj
key_word_list = key_words.get_key_word(news_text, news_title)
print("news_id {}\t num of keywords {}".format(news_pk,
len(key_word_list)))
return news_pk, key_word_list
