Print emojis usage by user (issue #37)

visualize.py

@@ -49,6 +49,14 @@ def cloud(self):
         args = parser.parse_args(sys.argv[2:])
         main(args)
 
+    def emoji(self):
+        from visualizers.emoji import main
+        parser = ArgParseDefault(description='Visualize emoji usage')
+        parser = add_load_data_args(parser)
+        parser.add_argument('--top-n-users', dest='top_n_users', type=int, default=10, help='Show top n users')
+        parser.add_argument('--top-n-emojis', dest='top_n_emojis', type=int, default=20, help='Show top n emojis per user')
+        args = parser.parse_args(sys.argv[2:])
+        main(args)
 
 if __name__ == '__main__':
     ArgParse()

visualizers/emoji.py

@@ -0,0 +1,47 @@
+import re
+import logging
+from visualizers.utils import save_fig
+from utils import load_data
+from collections import Counter
+import itertools
+import numpy as np
+
+log = logging.getLogger(__name__)
+EMOJI_PATTERN = re.compile("["
+        u"\U0001F600-\U0001F64F"  # emoticons
+        u"\U0001F300-\U0001F5FF"  # symbols & pictographs
+        u"\U0001F680-\U0001F6FF"  # transport & map symbols
+        u"\U0001F1E0-\U0001F1FF"  # flags (iOS)
+                           "]+", flags=re.UNICODE)
+
+def main(args):
+    # load data
+    df = load_data(args)
+    log.info(f'Computing emoji frequencies on {len(df):,} messages... 😤')
+    # Compute emoji frequencies
+    freqs = compute_emoji_freqs(df)
+    print_emoji_freqs(freqs, top_n_users=args.top_n_users, top_n_emojis=args.top_n_emojis)
+
+def compute_emoji_freqs(df):
+    df.loc[:, 'emojis'] = df.text.str.findall(EMOJI_PATTERN)
+    df.dropna(subset=['emojis'], inplace=True)
+    df = df.loc[df.emojis.apply(lambda s: len(s)) > 0, ['emojis', 'senderName', 'outgoing']]
+    df.loc[:, 'emojis'] = df.emojis.apply(lambda s: list(''.join(s)))
+    num_emojis = df.emojis.apply(lambda s: len(s)).sum()
+    log.info(f'Found {len(df):,} messages containing {num_emojis:,} emojis! 🤗🤗')
+    df.loc[df['outgoing'], 'senderName'] = 'You'
+    freqs = {}
+    for name, group in df.groupby('senderName'):
+        emojis = group.emojis.tolist()
+        freqs[name] = Counter(itertools.chain.from_iterable(emojis))
+    return freqs
+
+def print_emoji_freqs(freqs, top_n_users=10, top_n_emojis=10):
+    # selecting top n emoji users
+    by_user = {k: sum(c.values()) for k, c in freqs.items()}
+    top_users = [c[0] for c in Counter(by_user).most_common(top_n_users)]
+    # print out top users emoji usage
+    for user in top_users:
+        print(f'Favorite emojis for {user}')
+        emojis = ' '.join([emoji for emoji, count in freqs[user].most_common(top_n_emojis)])
+        print(emojis)