backfill movie data

flasky.py

@@ -7,6 +7,7 @@ import bson
 
 from datetime import timedelta
 from functools import update_wrapper
+from settings import DATABASE
 
 
 def crossdomain(origin=None, methods=None, headers=None,
@@ -52,7 +53,7 @@ def crossdomain(origin=None, methods=None, headers=None,
 
 
 app = Flask(__name__)
-client = MongoClient(**{'host': 'localhost', 'port': 27017})
+client = MongoClient(**DATABASE)
 db = client.showtimes
 
 miscObjHandler = lambda obj: (

movies.py

@@ -0,0 +1,146 @@
+from __future__ import absolute_import, print_function
+from pymongo import MongoClient
+from settings import DATABASE, TMDB_APIKEY
+from datetime import date
+import tmdbsimple as tmdb
+import sys
+import time
+
+
+client = MongoClient(**DATABASE)
+db = client.showtimes
+tmdb.API_KEY = TMDB_APIKEY
+
+
+def update_showtimes(_id, movie):
+    key = {'_id': _id}
+    update_dict = {
+        'movie_id': movie['id'],
+        'poster_path': movie['poster_path'],
+    }
+    result = db.showtimes.update(key, {'$set': update_dict}, True)
+    ## try adding to movie if not exist
+    add_movie(movie)
+    return result
+
+
+def add_movie(movie):
+    movie_collection = db.movies
+    q = {
+        'id': movie['id']
+    }
+    _m = movie_collection.find_one(q)
+    if _m:
+        return _m
+    '''
+    ## adding new one with full info
+    {
+        u'poster_path': u'/otcv3wWKz0vJyicOZgz2AheJ5UV.jpg',
+        u'production_countries': [
+            {u'iso_3166_1': u'US', u'name': u'United States of America'}
+        ],
+        u'revenue': 0,
+        u'overview': u'........................',
+        u'id': 245891,
+        u'genres': [
+            {u'id': 28, u'name': u'Action'},
+            {u'id': 53, u'name': u'Thriller'}],
+        u'title': u'John Wick',
+        u'tagline': u'',
+        u'vote_count': 5,
+        u'homepage': u'',
+        u'belongs_to_collection': None,
+        u'status': u'Released',
+        u'spoken_languages': [],
+        u'imdb_id': u'tt2911666',
+        u'adult': False,
+        u'backdrop_path': u'/umC04Cozevu8nn3JTDJ1pc7PVTn.jpg',
+        u'production_companies': [],
+        u'release_date': u'2014-10-24',
+        u'popularity': 3.044875809,
+        u'original_title': u'John Wick',
+        u'budget': 0,
+        u'vote_average': 8.4,
+        u'runtime': 0}
+    '''
+    identity = tmdb.Movies(movie['id'])
+    m_info = identity.info()
+    new_movie = movie_collection.insert(m_info)
+    return new_movie
+
+
+def find_movie(qword):
+    '''
+    find in db.movies first, then fallback to tmdb
+    '''
+    q = {
+        'title': qword
+    }
+    _mx = db.movies.find_one(q)
+    if _mx:
+        print('FOUND')
+        print(_mx)
+    else:
+        print('NOT FOUND')
+
+    return []
+
+    '''
+    return something like this if found
+    {
+        u'poster_path': u'/otcv3wWKz0vJyicOZgz2AheJ5UV.jpg',
+        u'title': u'John Wick',
+        u'release_date': u'2014-10-24',
+        u'popularity': 3.044875809,
+        u'original_title': u'John Wick',
+        u'backdrop_path': u'/umC04Cozevu8nn3JTDJ1pc7PVTn.jpg',
+        u'vote_count': 5,
+        u'adult': False,
+        u'vote_average': 8.4,
+        u'id': 245891
+    }
+    '''
+    search = tmdb.Search()
+    response = search.movie(query=qword)
+    if not search.results:
+        return []
+    '''
+    dang we pick the first match..
+    since I don't know how to deal with multiple result yet
+    '''
+    return search.results[0]
+
+
+def main(argv):
+    '''
+    Mainly we are getting movie name from db.showtimes
+    & check if there is any info on db.movies.
+    If not, we are trying to fetch for new info
+    '''
+    print('Fetching Movie without information')
+    q = {
+        '$or': [
+            {'movie_id': {'$exists': False}},
+            {'movie_id': ''},
+            {'poster_path': ''},
+        ]
+    }
+    result = db.showtimes.find(q)
+    cannot_find = []
+    for _ in result:
+        # skip if we know that we aren't going get any info anyway
+        if _['movie'] in cannot_find:
+            print('Nah -- you will not find this: ', _['movie'])
+            continue
+
+        time.sleep(1)
+        found = find_movie(_['movie'])
+
+        if found:
+            update_showtimes(_['_id'], found)
+        else:
+            cannot_find.append(_['movie'])
+
+
+if __name__ == '__main__':
+    main(sys.argv[1:])

pip_requirements.txt

@@ -1,4 +1,5 @@
 requests==2.4.0
 tornado>=4.0
 flask
-python-dateutil
+python-dateutil
+tmdbsimple

settings.py

@@ -5,7 +5,8 @@ from tornado.options import define, options
 import os
 import logconfig
 
-
+TMDB_APIKEY = '4885bf6b6de87bc56599ad0147a0818f'
+DATABASE = {'host': 'localhost', 'port': 27017}
 # Make filepaths relative to settings.
 path = lambda root, *a: os.path.join(root, *a)
 ROOT = os.path.dirname(os.path.abspath(__file__))