Merge branch 'master' of github.com:DocNow/twarc

utils/emojis.py

@@ -4,14 +4,30 @@
 import json
 import fileinput
 import collections
+import optparse
 
 import emoji
 
+opt_parser = optparse.OptionParser()
+
+opt_parser.add_option(
+    "-n",
+    "--number",
+    dest="number",
+    type="int",
+    default= 10
+)
+options, args = opt_parser.parse_args()
+tweets = args
+
+number_of_emojis = options.number
+tweets = tweets.pop()
+
 counts = collections.Counter()
 
 EMOJI_RE = emoji.get_emoji_regexp()
 
-for line in fileinput.input():
+for line in open(tweets):
     tweet = json.loads(line)
     if 'full_text' in tweet:
         text = tweet['full_text']
@@ -20,5 +36,5 @@
     for char in EMOJI_RE.findall(text):
         counts[char] += 1
 
-for char, count in counts.most_common():
+for char, count in counts.most_common(number_of_emojis):
     print("%s %5i" % (char, count))