def create_index(self, url):
# url = "http://fapl.ru/"
htmlToTextConverter = HtmlToTextConverter()
title, content = htmlToTextConverter.transform_html_into_text(url)
buildIndex = BuildIndex(content)
index = buildIndex.get_index()
number_of_words = buildIndex.get_number_of_words()
try:
document = Page.objects.get(url=url)
for match in Match.objects.filter(page=document):
word = match.word
match.delete()
if not word.pages.all():
word.delete()
except ObjectDoesNotExist:
document = Page(url=url,
number_of_words=number_of_words,
title=title,
content=content)
document.save()
for word in index.keys():
positions = " ".join(str(x) for x in index[word])
try:
word = Word.objects.get(value=word)
except ObjectDoesNotExist:
word = Word(value=word)
word.save()
match = Match(word=word, page=document, positions=positions)
match.save()
