Changing usage so that multiple users can have their tweets scraped into one file

Author: Tmonster

.gitignore

@@ -5,3 +5,5 @@
 *.json
 *.ignore
 tweet.py
+saved_tweets/*
+authentication_files/*

getTweets.py

@@ -11,6 +11,7 @@
 # "access_token"        : "value",
 # "access_token_secret" : "value"
 # }
+
 def get_credentials(filename):
     with open(filename) as file:
         cfg = json.load(file)
@@ -23,19 +24,39 @@ def get_api(cfg):
 
 def main():
     try:
-        username = sys.argv[1]
-        amount = int(sys.argv[2])
+        users = 1
+        usernames = []
+        amount = 0
+        place_in_named_file = False
+        for i in range(len(sys.argv)):
+            try:
+                # if I can cast to integer, interpret
+                # as amount
+                amount = int(sys.argv[users])
+                break;
+            except:
+                # otherwise, interpret as twitterhandle
+                usernames.append(sys.argv[users])
+                users+=1
+        if sys.argv[-2] == "-f":
+            named_file = sys.argv[-1]
+            place_in_named_file = True
+        if len(usernames) == 0 or amount == 0:
+            raise IndexError
     except IndexError as e:
-      print "Usage: python getTweets.py username amount [filename]"
+      print "Usage: python getTweets.py username [username...] amount [-f filename]"
       sys.exit(1)
+
     # Fill in the values noted in previous step here
-    cfg = get_credentials("the_Tmonster.json")
+    cfg = get_credentials("./authentication_files/the_Tmonster.json")
     api = get_api(cfg)
-    user = api.get_user(username)
-    try:
-      tweet_dumper.get_all_tweets4args(username, api, amount, sys.argv[3])
-    except IndexError as e:
-      tweet_dumper.get_all_tweets3args(username, api, amount)
+
+    for handle in usernames:
+        user = api.get_user(handle)
+        if place_in_named_file:
+            tweet_dumper.get_all_tweets4args(handle, api, amount, named_file)
+        else:
+            tweet_dumper.get_all_tweets3args(handle, api, amount)
 
 if __name__ == "__main__":
     main()

tweet_dumper.py

@@ -55,8 +55,18 @@ def get_all_tweets4args(screen_name, api, amount, filename):
 
 
 def writeTweets(filename, mode, tweets):
-    f = open(filename, mode)
-    # print len(alltweets)
+    if filename.find("saved_tweets") == -1:
+        if os.path.isdir("./saved_tweets"):
+            f = open("./saved_tweets/" + filename, mode)
+        else:
+            os.makedirs("./saved_tweets")
+            f = open("./saved_tweets/" + filename, mode)
+    elif os.path.isdir("./saved_tweets"):
+        f = open(filename, mode)
+    else:
+        os.makedirs("./saved_tweets")
+        f = open(filename, mode)
+
     for tweet in tweets:
         if tweet != "\n" or tweet != "":
             f.write(tweet)
@@ -73,7 +83,7 @@ def get_all_tweets5args(screen_name, api, amount, filename, lastTweetScrapedFile
     new_tweets = api.user_timeline(screen_name = screen_name,count=200, include_rts = True)
     numTweets = len(new_tweets)
     MostRecentTweetPulled = getMostRecentTweet(lastTweetScrapedFile, screen_name)
-    print "Most Recent Tweet Pulled =", MostRecentTweetPulled
+    print "Most Recent Tweet ID Pulled for", screen_name, "=", MostRecentTweetPulled
 
     # record most recent tweet id.
     # The id of the last tweet the user tweeted.
@@ -104,7 +114,7 @@ def get_all_tweets5args(screen_name, api, amount, filename, lastTweetScrapedFile
                         strippedTweet+=character
                 tweet = strippedTweet
             # strip URLS, Ampersands, retweets, and newlines
-            tweet = re.sub(r'(?:www|https?)[^\s]+\s', '', tweet, flags=re.MULTILINE)
+            tweet = re.sub(r'(www|https?)[^\s]+', '', tweet)
             tweet = re.sub(r'&', '&', tweet, flags=re.MULTILINE)
             tweet = re.sub(r'^RT.*:+ ', '', tweet, flags=re.MULTILINE)
             tweet = tweet.replace('\n', ' ')
@@ -121,5 +131,5 @@ def get_all_tweets5args(screen_name, api, amount, filename, lastTweetScrapedFile
         numTweets += len(new_tweets)
 
     writeTweets(filename, 'a', alltweets)
-    # write to txt file
+    # update Last Tweet
     updateLastTweet(lastTweetScrapedFile, screen_name, most_recent_tweet)