Ejemplo n.º 1
0
	def parse(self):
		for list in self.lists:
			dt = self.get_date_from_tag(list.div.strong)
			if dt is None:
				continue

			movie_list = []
			for child in list.ul.find_all('li'):
				contents = child.dl.find_all('span', 'link_txt')
				movie = Movie()
				movie.title = self.get_title_from_tag(child)
				movie.genre = self.get_genre_list_from_tag(contents[0])
				movie.director = self.get_director_list_from_tag(contents[1])
				try:
					movie.actor = self.get_actor_list_from_tag(contents[2])
				except IndexError:
					movie.actor = ""
				movie_list.append(movie)

			self.result[dt] = movie_list
Ejemplo n.º 2
0
for movie_data in data.get('data'):
    movie_api_id = movie_data[column_pos.get('api_id')]
    title = movie_data[column_pos.get('title')]
    year = movie_data[column_pos.get('year')]
    director = movie_data[column_pos.get('director')]
    movie = Movie.query.filter(Movie.title == title,
                               Movie.year == year,
                               Movie.director == director).first()
    if not movie:
        movie = Movie()
        movie.api_id = movie_data[column_pos.get('api_id')]
        sys.stdout.write('\rNEW Movie %s' % movie.api_id)
        movie.title = movie_data[column_pos.get('title')]
        movie.year = movie_data[column_pos.get('year')]
        movie.director = movie_data[column_pos.get('director')]
        # Set actors
        actors = [
            movie_data[column_pos.get('actor_1')],
            movie_data[column_pos.get('actor_2')],
            movie_data[column_pos.get('actor_3')]
        ]
        for actor_name in [a for a in actors if a]:
            actor = Actor.query.filter(Actor.name == actor_name).first()
            if not actor:
                actor = Actor(actor_name)
            movie.actors.append(actor)
        # Set IMDB url
        movie.imdb_url = get_imdb_url(movie.title, movie.year, movie.director)
        db.session.add(movie)
    else:
Ejemplo n.º 3
0
    pos += 1

for movie_data in data.get('data'):
    movie_api_id = movie_data[column_pos.get('api_id')]
    title = movie_data[column_pos.get('title')]
    year = movie_data[column_pos.get('year')]
    director = movie_data[column_pos.get('director')]
    movie = Movie.query.filter(Movie.title == title, Movie.year == year,
                               Movie.director == director).first()
    if not movie:
        movie = Movie()
        movie.api_id = movie_data[column_pos.get('api_id')]
        sys.stdout.write('\rNEW Movie %s' % movie.api_id)
        movie.title = movie_data[column_pos.get('title')]
        movie.year = movie_data[column_pos.get('year')]
        movie.director = movie_data[column_pos.get('director')]
        # Set actors
        actors = [
            movie_data[column_pos.get('actor_1')],
            movie_data[column_pos.get('actor_2')],
            movie_data[column_pos.get('actor_3')]
        ]
        for actor_name in [a for a in actors if a]:
            actor = Actor.query.filter(Actor.name == actor_name).first()
            if not actor:
                actor = Actor(actor_name)
            movie.actors.append(actor)
        # Set IMDB url
        movie.imdb_url = get_imdb_url(movie.title, movie.year, movie.director)
        db.session.add(movie)
    else: