Upload app.py

app.py

@@ -122,18 +122,6 @@ def authenticate_kaggle():
     api.authenticate()
 
 
-def map_year(x):
-    a = ''
-    if isinstance(x, str):
-        try:
-            a = re.search('\(([0-9_]+)\)', x[-7:]).group(1)
-        except:
-            a = ''
-    else:
-        a = ''
-    return a
-
-
 @st.experimental_memo(persist=True, show_spinner=False, suppress_st_warning=True, max_entries=1)
 def load_dataset():
     """
@@ -155,17 +143,17 @@ def load_dataset():
     ## ----------------------------- ##
 
     # Downloading Movies dataset
-    api.dataset_download_file('garymk/movielens-25m-dataset', 'ml-25m/movies.csv')
+    api.dataset_download_file('rounakbanik/the-movies-dataset', 'movies_metadata.csv')
 
     # Extract data
-    zf = ZipFile('movies.csv.zip')
+    zf = ZipFile('movies_metadata.csv.zip')
     zf.extractall() 
     zf.close()
 
     # Create dataframe
-    data = pd.read_csv('movies.csv', low_memory=False)
-    #data['year'] = data['release_date'].map(lambda x: x.split('-')[0] if isinstance(x, str) else '0')
-    data['year'] = data['title'].map(map_year)
+    data = pd.read_csv('movies_metadata.csv', low_memory=False)
+    data['year'] = data["release_date"].map(lambda x: x.split('-')[0] if isinstance(x, str) else '0')
+    data['title_year'] = data['title'] + ' (' + data['year'] + ')'
 
     return data
 
@@ -223,7 +211,9 @@ def generate_poster(movie_data):
     """
 
     # Get movie metadata
-    genres_string = movie_data['genres'].values[0].replace('|', ', ')
+    genres = [i['name'] for i in eval(movie_data['genres'].values[0])]
+    genres_string = ', '.join(genres)
+
     year = movie_data['year'].values[0]
 
 
@@ -304,7 +294,7 @@ if __name__ == "__main__":
     st.text("")
 
     ## Select box with all the movies as choices
-    session.selected_movie = st.selectbox(label="Select a movie to generate alternative poster", options=data.title)
+    session.selected_movie = st.selectbox(label="Select a movie to generate alternative poster", options=data.title_year)
 
     st.text("")
     st.text("")
@@ -315,7 +305,7 @@ if __name__ == "__main__":
 
     ## Generate poster
     if is_clicked:
-        poster = generate_poster(data[data.title==session.selected_movie])
+        poster = generate_poster(data[data.title_year==session.selected_movie])
         #generate_poster.clear()
         #st.runtime.legacy_caching.clear_cache()
 
@@ -331,6 +321,6 @@ if __name__ == "__main__":
         is_clicked_rerun = col1.button(label="Rerun with same movie!")
 
     if is_clicked_rerun:
-        poster = generate_poster(data[data.title==selected_movie])
+        poster = generate_poster(data[data.title_year==selected_movie])
     """