def scrape_title(self):
_title=self.main_page_soup.findAll('title')
assert len(_title)==1
_title=_title[0]
_title=clean_unicode(_title.next)
f=re.match('(.+) \((\d+)\) - IMDb',_title)
groups=f.groups()
assert len(groups)==2
self.movie_name,release_year=groups
def scrape_title(self):
_title = self.main_page_soup.findAll('title')
assert len(_title) == 1
_title = _title[0]
_title = clean_unicode(_title.next)
f = re.match('(.+) \((\d+)\) - IMDb', _title)
groups = f.groups()
assert len(groups) == 2
self.movie_name, release_year = groups
def get_top_box_office_by_year(year, number, debug=False):
""" Pull out the 'number' highest-grosing
movies of the year.
"""
NUM_MOVIES_PER_PAGE = 50
sort = 'boxoffice_gross_us'
def get_website(start, year):
website = 'http://www.imdb.com/search/title?at=0&sort=%s&start=%s&title_type=feature&year=%s,%s' % (
sort, start, year, year)
return website
n = 1
ret_list = OrderedDict()
while n < number:
print 'n=%s/%s' % (n, number)
url_page = get_website(start=n, year=year)
print url_page
n += NUM_MOVIES_PER_PAGE
# I don't get why, but IMDB barfs when I specify a user agent???
soup = get_soup(url_page, no_user_agent=True)
# Match on <td class="number">, which refers to the ranking of the movie
all_movies = soup.findAll('td', **{'class': "number"})
for movie in all_movies:
title_part = movie.next.next.next.next.next.next.next.next.next.next.next.next.next
movie_name = clean_unicode(title_part.next)
link = str(title_part['href'])
m = re.match('/title/tt(\d+)/', link)
groups = m.groups()
assert len(groups) == 1
imdb_movie_id = int(groups[0])
_year = title_part.next.next.next.next
m = re.match(r'\((\d+)\)', _year)
groups = m.groups()
assert len(groups) == 1
year = int(groups[0])
ret_list[imdb_movie_id] = movie_name
# if only a few movies are requested
if len(ret_list) == number:
return ret_list
return ret_list
def get_top_box_office_by_year(year, number, debug=False):
""" Pull out the 'number' highest-grosing
movies of the year.
"""
NUM_MOVIES_PER_PAGE=50
sort='boxoffice_gross_us'
def get_website(start,year):
website='http://www.imdb.com/search/title?at=0&sort=%s&start=%s&title_type=feature&year=%s,%s' % (sort,start,year,year)
return website
n=1
ret_list=OrderedDict()
while n<number:
print 'n=%s/%s' % (n,number)
url_page = get_website(start=n,year=year)
print url_page
n+=NUM_MOVIES_PER_PAGE
# I don't get why, but IMDB barfs when I specify a user agent???
soup=get_soup(url_page,no_user_agent=True)
# Match on <td class="number">, which refers to the ranking of the movie
all_movies=soup.findAll('td',**{'class':"number"})
for movie in all_movies:
title_part=movie.next.next.next.next.next.next.next.next.next.next.next.next.next
movie_name=clean_unicode(title_part.next)
link=str(title_part['href'])
m=re.match('/title/tt(\d+)/',link)
groups=m.groups()
assert len(groups)==1
imdb_movie_id=int(groups[0])
_year=title_part.next.next.next.next
m=re.match(r'\((\d+)\)',_year)
groups=m.groups()
assert len(groups)==1
year=int(groups[0])
ret_list[imdb_movie_id]=movie_name
# if only a few movies are requested
if len(ret_list) == number:
return ret_list
return ret_list
def get_review_from_page(self,review_soup,imdb_review_url):
""" Pull out a single review form an IMDB
movie review page.
review is a soup object anchored on a reviewer's avatar.
"""
# Most reviews begin with the text
# > "XXX out of XXX found the following review useful:"
# we have to back up to find it, but sometimes it doesn't exist
_quality_of_review = review_soup.previous.previous.previous.previous
m=re.match('(\d+) out of (\d+) people found the following review useful:', str(_quality_of_review))
if m is not None and len(m.groups())==2:
groups=m.groups()
num_likes = int(groups[0])
num_dislikes = int(groups[1])-int(groups[0])
else:
num_likes = num_dislikes = None
_title=review_soup.next.next.next
review_title=clean_unicode(_title)
# the next thing to look for is the review score.
# Note that this doesn't not always exist:
review_image = _title.next.next
if review_image.name == 'img':
_review_score=_title.next.next.attrs['alt']
review_score=_review_score.split('/')
assert review_score[0].isdigit() and review_score[1].isdigit()
review_score=[int(review_score[0]),int(review_score[1])]
assert review_score[0] in range(1,11)
assert review_score[1]==10
review_score=review_score[0]
_reviewer=_title.next.next.next.contents[3].next
else:
# No user review, jump to reviewer
review_score=None
_reviewer=_title.next.next.next.next.next.next
reviewer_url=_reviewer.previous['href']
m=re.match('/user/ur(\d+)/',reviewer_url)
groups=m.groups()
assert len(groups)==1
imdb_reviewer_id = int(groups[0])
if _reviewer == ' ':
# for some reason, I think some reviewers don't have
# a reviewer name. I found this problem here:
# http://www.imdb.com/title/tt1408101/reviews?start=120
reviewer=None
_review_place=_reviewer.next.next
elif hasattr(_reviewer,'name') and _reviewer.name == 'br':
# this happens when there is no reviewer and no place!
# This happend at: http://www.imdb.com/title/tt1392170/reviews?start=1340
# If so, move the '_place' up the "<small>8 April 2012</small>"
# html so that it will get caught at the next condition
reviewer=None
_review_place=_reviewer.next.next
else:
reviewer=clean_unicode(_reviewer)
_review_place=_reviewer.next.next.next
if hasattr(_review_place,'name') and _review_place.name == 'small':
# this happens when there is no place.
# If so, skip on to date
# For an example of this ...
review_place = None
_date = _review_place.next
else:
m = re.match('from (.+)', _review_place)
groups=m.groups()
assert len(groups)==1
review_place = groups[0]
review_place=review_place
_date=_review_place.next.contents[1].next
date=str(_date)
date=datetime.datetime.strptime(date,'%d %B %Y')
_review_text=_date.next.next.next.next
imdb_review_text=get_html_text(_review_text)
if imdb_review_text=='*** This review may contain spoilers ***.':
spoilers=True
_review_text=_review_text.next.next.next.next
imdb_review_text=get_html_text(_review_text)
else:
spoilers=False
d=dict(review_title=review_title,
date=date,
review_score=review_score,
reviewer=reviewer,
review_place=review_place,
imdb_review_text=imdb_review_text,
spoilers=spoilers,
num_likes = num_likes,
num_dislikes = num_dislikes,
imdb_movie_id=self.imdb_movie_id,
imdb_reviewer_id=imdb_reviewer_id,
imdb_review_ranking=self.imdb_review_ranking_counter,
imdb_review_url=imdb_review_url)
self.imdb_review_ranking_counter+=1
return d
def get_review_from_page(self, review_soup, imdb_review_url):
""" Pull out a single review form an IMDB
movie review page.
review is a soup object anchored on a reviewer's avatar.
"""
# Most reviews begin with the text
# > "XXX out of XXX found the following review useful:"
# we have to back up to find it, but sometimes it doesn't exist
_quality_of_review = review_soup.previous.previous.previous.previous
m = re.match(
'(\d+) out of (\d+) people found the following review useful:',
str(_quality_of_review))
if m is not None and len(m.groups()) == 2:
groups = m.groups()
num_likes = int(groups[0])
num_dislikes = int(groups[1]) - int(groups[0])
else:
num_likes = num_dislikes = None
_title = review_soup.next.next.next
review_title = clean_unicode(_title)
# the next thing to look for is the review score.
# Note that this doesn't not always exist:
review_image = _title.next.next
if review_image.name == 'img':
_review_score = _title.next.next.attrs['alt']
review_score = _review_score.split('/')
assert review_score[0].isdigit() and review_score[1].isdigit()
review_score = [int(review_score[0]), int(review_score[1])]
assert review_score[0] in range(1, 11)
assert review_score[1] == 10
review_score = review_score[0]
_reviewer = _title.next.next.next.contents[3].next
else:
# No user review, jump to reviewer
review_score = None
_reviewer = _title.next.next.next.next.next.next
reviewer_url = _reviewer.previous['href']
m = re.match('/user/ur(\d+)/', reviewer_url)
groups = m.groups()
assert len(groups) == 1
imdb_reviewer_id = int(groups[0])
if _reviewer == ' ':
# for some reason, I think some reviewers don't have
# a reviewer name. I found this problem here:
# http://www.imdb.com/title/tt1408101/reviews?start=120
reviewer = None
_review_place = _reviewer.next.next
elif hasattr(_reviewer, 'name') and _reviewer.name == 'br':
# this happens when there is no reviewer and no place!
# This happend at: http://www.imdb.com/title/tt1392170/reviews?start=1340
# If so, move the '_place' up the "<small>8 April 2012</small>"
# html so that it will get caught at the next condition
reviewer = None
_review_place = _reviewer.next.next
else:
reviewer = clean_unicode(_reviewer)
_review_place = _reviewer.next.next.next
if hasattr(_review_place, 'name') and _review_place.name == 'small':
# this happens when there is no place.
# If so, skip on to date
# For an example of this ...
review_place = None
_date = _review_place.next
else:
m = re.match('from (.+)', _review_place)
groups = m.groups()
assert len(groups) == 1
review_place = groups[0]
review_place = review_place
_date = _review_place.next.contents[1].next
date = str(_date)
date = datetime.datetime.strptime(date, '%d %B %Y')
_review_text = _date.next.next.next.next
imdb_review_text = get_html_text(_review_text)
if imdb_review_text == '*** This review may contain spoilers ***.':
spoilers = True
_review_text = _review_text.next.next.next.next
imdb_review_text = get_html_text(_review_text)
else:
spoilers = False
d = dict(review_title=review_title,
date=date,
review_score=review_score,
reviewer=reviewer,
review_place=review_place,
imdb_review_text=imdb_review_text,
spoilers=spoilers,
num_likes=num_likes,
num_dislikes=num_dislikes,
imdb_movie_id=self.imdb_movie_id,
imdb_reviewer_id=imdb_reviewer_id,
imdb_review_ranking=self.imdb_review_ranking_counter,
imdb_review_url=imdb_review_url)
self.imdb_review_ranking_counter += 1
return d
