def movie_launcher():
"""[summary: launch the script to scrap all information of imb movie platform]
"""
entity = Movie() # instantiation of the movie objet
for step in range(1, 13117,
50): # we loop into 13117 movies that contains the imb
# build the url of each page
url = 'https://www.imdb.com/search/title/?at=0&num_votes=5000,&sort=user_rating,desc&start=' + str(
step) + '&title_type=feature'
res = requests.get(url)
if res.ok:
soup = BeautifulSoup(res.content.decode('utf-8', 'ignore'),
'html.parser') # we get the soup content
tags = soup.find_all('div', attrs={'class': 'lister-item-content'})
for tag in tags:
r = requests.get(
_link(getlink(tag))
) # 'https://www.imdb.com/title/tt0111161/?ref_=adv_li_tt'
if r.ok:
print('i')
content = BeautifulSoup(
r.text, 'html.parser') # we get the soup content
manager = Manager(
entity,
[content, tag]) # initialization of Manager of movie
manager.parse_json(
entity) # convert the current objet to dictionnary
manager.to_csv(
) # register the current entity of movie into the csv file
time.sleep(4)
def test_getMovieById(self):
if self.controller.getMovieById('11111111') is False:
self.controller.addMovie('11111111', 'sfsdfsf', 'sdfdsfsfdsfdsfs',
'action')
self.assertEqual(
self.controller.getMovieById('11111111'),
Movie('11111111', 'sfsdfsf', 'sdfdsfsfdsfdsfs', 'action'))
def updateMovieId(self, movieId, newId):
'''
A functions that updates a movie's ID and also provides data validation
:param movieId: integer
:param newId: intege
:return: True if the movie was updated, False, otherwise
'''
try:
movieId = int(movieId)
except ValueError:
raise ValueError("Error: \n Movie Id cannot be string")
try:
newId = int(newId)
except ValueError:
raise ValueError("Error: \n New Movie Id cannot be string")
item = self.getItemById(movieId)
newItem = self.getItemById(newId)
if newItem:
raise ValueError("Error:\n The given id is already taken")
if item is not False:
item = Movie(*item)
item.id = newId
return self.updateItemById(movieId, item)
def test_strMovie(self):
self.movie.id = 3
self.movie.genre = 'action'
self.movie.title = 'Titanic4'
self.movie.description = 'supers'
m = Movie(3, 'sd', 'sds', 'action')
self.assertNotEqual(m, self.movie)
m = self.movie
self.assertEqual(m, self.movie)
self.assertEqual(str(self.movie), '3, Titanic4, supers, action')
def test_search(self):
c = MovieController(Repository("testSearch", "d"))
c.addItem(Movie(3, "action", "sdf", "family"))
self.assertEqual(c.searchMovieByGenre("action"), [])
c.removeMovie(3)
self.assertEqual(c.searchMovieByTitle("d"), [])
self.assertEqual(c.searchMovieByDescription("descriere"), [])
self.assertRaises(ValueError, c.searchMovieByDescription, "")
self.assertRaises(ValueError, c.searchMovieByGenre, "")
self.assertRaises(ValueError, c.searchMovieByTitle, "")
def addMovie(self, movieId, title, description, genre):
'''
A function that validates and adds a movie to the repository
:param movieId: integer
:param title: string / cannot be empty
:param description: string / cannot be empty
:param genre: one of MOVIE_GENRE
:return: True if the movie was added with success, raises errors if there are any validation problems
'''
if MovieValidator.validate(movieId, title, description, genre):
movieId = int(movieId)
movie = Movie(movieId, title, description, genre)
if self.repo.getItemById(int(movieId)) is not False:
raise ValueError("Error:\n The given id is already taken")
self.addItem(movie)
return True
def updateMovieGenre(self, movieId, newGenre):
'''
A functions that updates a movie's genre and also provides data validation
:param movieId: integer
:param newGenre: string // cannot be empty
:return: True if the movie was updated, False, otherwise
'''
if ',' in newGenre:
raise ValueError(
"Error:\n Parameters should not contain the ',' character")
try:
movieId = int(movieId)
except ValueError:
raise ValueError("Error: \n Movie Id cannot be string")
if newGenre not in MOVIE_GENRE:
raise ValueError("Error: \n Movie genre isn't valid")
item = self.getItemById(movieId)
if item is not False:
item = Movie(*item)
item.genre = newGenre
return self.repo.updateItemById(movieId, item)
def setUp(self):
self.movie = Movie(3, 'Titanic4', 'supers', 'action')
self.repo = Repository('testMovies', 'testMovies')
self.controller = MovieController(self.repo)
if self.controller.getMovieById(77) is not False:
self.controller.removeMovie(77)
if self.controller.getMovieById(772) is not False:
self.controller.removeMovie(772)
if self.controller.getMovieById(773) is not False:
self.controller.removeMovie(773)
if self.controller.getMovieById(23) is False:
self.controller.addMovie("23", "title", "description", "action")
if self.controller.getMovieById('233333') is not False:
self.controller.removeMovie('233333')
if self.controller.getMovieById(20) is False:
self.controller.addMovie("20", "title", "description", "action")
if self.controller.getMovieById(200) is False:
self.controller.addMovie("200", "title", "description", "action")
if self.controller.getMovieById(90) is False:
self.controller.addMovie("90", "title", "description", "action")
if self.controller.getMovieById(123) is False:
self.controller.addMovie("123", "title", "description", "action")
if self.controller.getMovieById(234) is False:
self.controller.addMovie("234", "title", "description", "action")
if self.controller.getMovieById(33) is False:
self.controller.addMovie("33", "title", "description", "action")
if self.controller.getMovieById(90) is not False:
self.controller.removeMovie(99)
if self.controller.getMovieById(323) is False:
self.controller.addMovie("323", "title", "description", "action")
if self.controller.getMovieById(929) is not False:
self.controller.removeMovie(929)
if self.controller.getMovieById(33) is False:
self.controller.addMovie("333", "title", "description", "action")
if self.controller.getMovieById(939) is not False:
self.controller.removeMovie(939)
def updateMovieName(self, movieId, movieTitle):
'''
A functions that updates a movie's title and also provides data validation
:param movieId: integer
:param movieTitle: string // cannot be empty
:return: True if the movie was updated, False, otherwise
'''
if ',' in movieTitle:
raise ValueError(
"Error:\n Parameters should not contain the ',' character")
try:
movieId = int(movieId)
except ValueError:
raise ValueError("Error:\n Movie id must be integer")
if len(movieTitle.strip()) == 0:
raise ValueError("Error: \n Movie Title should not be empty")
item = self.getItemById(movieId)
if item is not False:
item = Movie(*item)
if len(movieTitle.strip()) > 0:
item.title = movieTitle
return self.updateItemById(int(movieId), item)
else:
raise ValueError("Error:\n Movie Id cannot be find")
def test_UpdateMovie5(self):
self.controller.addItem(Movie(20000000000, "sds", "sdfs", "action"))
def test_UpdateMovie4(self):
self.controller.addItem(Movie(20000000000, "sds", "sdfs", "action"))
self.assertRaises(ValueError, self.controller.updateMovie, '23',
'233333', "", "sdfsdfssfs", "action")
self.controller.removeMovie(20000000000)
def test_GetAllMovies(self):
repo2 = Repository('movieUpdateTest', "test2")
movie = Movie("2", "223", '22320', 'action')
controller2 = MovieController(repo2)
l = controller2.getAllMovies()
self.assertEqual(controller2.getAllMovies(), [movie])