In [1]:
import sys
sys.path.append('../src')
from numbers_scraper import NumbersScraper
SCRAPER = NumbersScraper()
In [2]:
df = SCRAPER.get_table_data()
df
Out[2]:
0
1
2
3
4
1
12/18/2009
Avatar
$425,000,000
$760,507,625
$2,783,918,982
2
12/18/2015
Star Wars Ep. VII: The Force Awakens
$306,000,000
$936,662,225
$2,058,662,225
3
5/24/2007
Pirates of the Caribbean: At Worldâs End
$300,000,000
$309,420,425
$963,420,425
4
11/6/2015
Spectre
$300,000,000
$200,074,175
$879,620,923
5
7/20/2012
The Dark Knight Rises
$275,000,000
$448,139,099
$1,084,439,099
6
7/2/2013
The Lone Ranger
$275,000,000
$89,302,115
$260,002,115
7
3/9/2012
John Carter
$275,000,000
$73,058,679
$282,778,100
8
11/24/2010
Tangled
$260,000,000
$200,821,936
$586,581,936
9
5/4/2007
Spider-Man 3
$258,000,000
$336,530,303
$894,860,230
10
5/1/2015
Avengers: Age of Ultron
$250,000,000
$459,005,868
$1,408,218,722
11
5/6/2016
Captain America: Civil War
$250,000,000
$408,084,349
$1,153,304,495
12
3/25/2016
Batman v Superman: Dawn of Justice
$250,000,000
$330,360,194
$868,160,194
13
12/14/2012
The Hobbit: An Unexpected Journey
$250,000,000
$303,003,568
$1,017,003,568
14
7/15/2009
Harry Potter and the Half-Blood Prince
$250,000,000
$301,959,197
$935,083,686
15
12/13/2013
The Hobbit: The Desolation of Smaug
$250,000,000
$258,366,855
$960,366,855
16
12/17/2014
The Hobbit: The Battle of the Five Armies
$250,000,000
$255,119,788
$955,119,788
17
5/20/2011
Pirates of the Caribbean: On Stranger Tides
$250,000,000
$241,063,875
$1,045,663,875
18
4/14/2017
The Fate of the Furious
$250,000,000
$225,764,765
$1,237,444,462
19
6/28/2006
Superman Returns
$232,000,000
$200,120,000
$374,085,065
20
5/26/2017
Pirates of the Caribbean: Dead Men Tell No Tales
$230,000,000
$171,832,628
$785,628,907
21
11/14/2008
Quantum of Solace
$230,000,000
$169,368,427
$591,692,078
22
5/4/2012
The Avengers
$225,000,000
$623,279,547
$1,519,479,547
23
7/7/2006
Pirates of the Caribbean: Dead Manâs Chest
$225,000,000
$423,315,812
$1,066,215,812
24
6/14/2013
Man of Steel
$225,000,000
$291,045,518
$667,999,518
25
5/16/2008
The Chronicles of Narnia: Prince Caspian
$225,000,000
$141,621,490
$417,341,288
26
7/3/2012
The Amazing Spider-Man
$220,000,000
$262,030,663
$757,890,267
27
6/21/2017
Transformers: The Last Knight
$217,000,000
$130,120,862
$601,120,862
28
6/12/2015
Jurassic World
$215,000,000
$652,198,010
$1,671,640,593
29
5/25/2012
Men in Black 3
$215,000,000
$179,020,854
$654,213,485
30
6/24/2009
Transformers: Revenge of the Fallen
$210,000,000
$402,111,870
$836,519,699
...
...
...
...
...
...
5367
12/1/2015
Dutch Kills
$25,000
$0
$0
5368
8/1/1991
Slacker
$23,000
$1,227,508
$1,227,508
5369
12/31/2014
Dry Spell
$22,000
$0
$0
5370
12/31/2003
Flywheel
$20,000
$0
$0
5371
1/4/2013
All Superheroes Must Die
$20,000
$0
$0
5372
4/21/2015
The Front Man
$20,000
$0
$0
5373
11/25/2011
The Ridges
$17,300
$0
$0
5374
6/2/2006
The Puffy Chair
$15,000
$194,523
$195,254
5375
4/2/2010
Breaking Upwards
$15,000
$115,592
$115,592
5376
12/31/2014
Stories of Our Lives
$15,000
$0
$0
5377
4/11/1997
Pink Flamingos
$12,000
$413,802
$413,802
5378
4/28/2006
Grip: A Criminal's Story
$12,000
$1,336
$1,336
5379
12/31/2007
Tin Can Man
$12,000
$0
$0
5380
3/9/2001
Dayereh
$10,000
$673,780
$673,780
5381
4/28/2006
Clean
$10,000
$138,711
$138,711
5382
7/6/2001
Cure
$10,000
$94,596
$94,596
5383
5/28/2004
On the Down Low
$10,000
$1,987
$1,987
5384
4/1/1996
Bang
$10,000
$527
$527
5385
8/14/2008
The Rise and Fall of Miss Thang
$10,000
$401
$401
5386
5/19/2015
Family Motocross
$10,000
$0
$0
5387
1/13/2012
Newlyweds
$9,000
$4,584
$4,584
5388
2/26/1993
El Mariachi
$7,000
$2,040,920
$2,041,928
5389
10/8/2004
Primer
$7,000
$424,760
$841,926
5390
5/26/2006
Cavite
$7,000
$70,071
$71,644
5391
1/1/2004
The Mongol King
$7,000
$900
$900
5392
4/2/1999
Following
$6,000
$48,482
$240,495
5393
7/13/2005
Return to the Land of Wonders
$5,000
$1,338
$1,338
5394
9/29/2015
Signed Sealed Delivered
$5,000
$0
$0
5395
9/29/2015
A Plague So Pleasant
$1,400
$0
$0
5396
8/5/2005
My Date With Drew
$1,100
$181,041
$181,041
5396 rows × 5 columns
In [3]:
df_clean = SCRAPER.clean_data()
df_clean
Out[3]:
release_date
title
production_budget
domestic_gross
world_gross
1
2009-12-18
Avatar
425000000
760507625
2783918982
2
2015-12-18
Star Wars Ep. VII: The Force Awakens
306000000
936662225
2058662225
3
2007-05-24
Pirates of the Caribbean: At Worlds End
300000000
309420425
963420425
4
2015-11-06
Spectre
300000000
200074175
879620923
5
2012-07-20
The Dark Knight Rises
275000000
448139099
1084439099
6
2013-07-02
The Lone Ranger
275000000
89302115
260002115
7
2012-03-09
John Carter
275000000
73058679
282778100
8
2010-11-24
Tangled
260000000
200821936
586581936
9
2007-05-04
Spider-Man 3
258000000
336530303
894860230
10
2015-05-01
Avengers: Age of Ultron
250000000
459005868
1408218722
11
2016-05-06
Captain America: Civil War
250000000
408084349
1153304495
12
2016-03-25
Batman v Superman: Dawn of Justice
250000000
330360194
868160194
13
2012-12-14
The Hobbit: An Unexpected Journey
250000000
303003568
1017003568
14
2009-07-15
Harry Potter and the Half-Blood Prince
250000000
301959197
935083686
15
2013-12-13
The Hobbit: The Desolation of Smaug
250000000
258366855
960366855
16
2014-12-17
The Hobbit: The Battle of the Five Armies
250000000
255119788
955119788
17
2011-05-20
Pirates of the Caribbean: On Stranger Tides
250000000
241063875
1045663875
18
2017-04-14
The Fate of the Furious
250000000
225764765
1237444462
19
2006-06-28
Superman Returns
232000000
200120000
374085065
20
2017-05-26
Pirates of the Caribbean: Dead Men Tell No Tales
230000000
171832628
785628907
21
2008-11-14
Quantum of Solace
230000000
169368427
591692078
22
2012-05-04
The Avengers
225000000
623279547
1519479547
23
2006-07-07
Pirates of the Caribbean: Dead Mans Chest
225000000
423315812
1066215812
24
2013-06-14
Man of Steel
225000000
291045518
667999518
25
2008-05-16
The Chronicles of Narnia: Prince Caspian
225000000
141621490
417341288
26
2012-07-03
The Amazing Spider-Man
220000000
262030663
757890267
27
2017-06-21
Transformers: The Last Knight
217000000
130120862
601120862
28
2015-06-12
Jurassic World
215000000
652198010
1671640593
29
2012-05-25
Men in Black 3
215000000
179020854
654213485
30
2009-06-24
Transformers: Revenge of the Fallen
210000000
402111870
836519699
...
...
...
...
...
...
5367
2015-12-01
Dutch Kills
25000
0
0
5368
1991-08-01
Slacker
23000
1227508
1227508
5369
2014-12-31
Dry Spell
22000
0
0
5370
2003-12-31
Flywheel
20000
0
0
5371
2013-01-04
All Superheroes Must Die
20000
0
0
5372
2015-04-21
The Front Man
20000
0
0
5373
2011-11-25
The Ridges
17300
0
0
5374
2006-06-02
The Puffy Chair
15000
194523
195254
5375
2010-04-02
Breaking Upwards
15000
115592
115592
5376
2014-12-31
Stories of Our Lives
15000
0
0
5377
1997-04-11
Pink Flamingos
12000
413802
413802
5378
2006-04-28
Grip: A Criminal's Story
12000
1336
1336
5379
2007-12-31
Tin Can Man
12000
0
0
5380
2001-03-09
Dayereh
10000
673780
673780
5381
2006-04-28
Clean
10000
138711
138711
5382
2001-07-06
Cure
10000
94596
94596
5383
2004-05-28
On the Down Low
10000
1987
1987
5384
1996-04-01
Bang
10000
527
527
5385
2008-08-14
The Rise and Fall of Miss Thang
10000
401
401
5386
2015-05-19
Family Motocross
10000
0
0
5387
2012-01-13
Newlyweds
9000
4584
4584
5388
1993-02-26
El Mariachi
7000
2040920
2041928
5389
2004-10-08
Primer
7000
424760
841926
5390
2006-05-26
Cavite
7000
70071
71644
5391
2004-01-01
The Mongol King
7000
900
900
5392
1999-04-02
Following
6000
48482
240495
5393
2005-07-13
Return to the Land of Wonders
5000
1338
1338
5394
2015-09-29
Signed Sealed Delivered
5000
0
0
5395
2015-09-29
A Plague So Pleasant
1400
0
0
5396
2005-08-05
My Date With Drew
1100
181041
181041
5396 rows × 5 columns
In [4]:
SCRAPER.upload_data_s3()
In [ ]:
Content source: bryantbiggs/movie_torrents
Similar notebooks: