Tweepy: get old tweets now possible with Twitter s

Question

Accoring to http://www.theverge.com/2014/11/18/7242477/twitter-search-now-lets-you-find-any-tweet-ever-sent Twitter search now lets you find any tweet ever sent.

But when i am trying to get tweets from 2014 to 2015 using tweepy it gets only recent:

    query = 'Nivea'
    max_tweets = 1000
    searched_tweets = [json.loads(status.json) for status in tweepy.Cursor(api.search,
                                                                           q=query,
                                                                           count=100,
                                                                           #since_id="24012619984051000",
                                                                           since="2014-02-01",
                                                                           until="2015-02-01",
                                                                           result_type="mixed",
                                                                           lang="en"
                                                                           ).items(max_tweets)]

I tried since="2014-02-01", and since_id but no matter.

Answer 1

I use my own piece of code which uses a HttpURLConnection and a twitter search url. I then use a regular expression to pull out the last 20 matching tweets... Luckily as I'm deleting the tweets I can simply search again until I can't find anymore tweets. I'm including the code although it's in Java but the same would apply for any language. First I use a class to actually search for tweets and record their details:

public class ReadSearch{
    private String startURL = "https://twitter.com/search?f=realtime&q=from%3A";
    private String middleURL = "%20%40";
    private String endURL = "&src=typd";

    public ArrayList<Tweet> getTweets(String user, String troll) {
        ArrayList<Tweet> tweets = new ArrayList<Tweet>();
        String expr = "small.class=\"time\".*?href=\"/"
                + "([^/]+)"
                + ".*?status/"
                + "([^\"]+)"
                + ".*?title=\""
                + "([^\"]+)";
        Pattern patt = Pattern.compile(expr, Pattern.DOTALL | Pattern.UNIX_LINES);
        try {
            Matcher m = patt.matcher(getData(startURL+user+middleURL+troll+endURL));
            while (m.find()) {
                if(user.equals(m.group(1).trim())){
                    Tweet tw = new Tweet();
                    tw.setUser(m.group(1).trim());
                    tw.setTweetid(Long.parseLong(m.group(2).trim()));
                    tw.setDate(m.group(3).trim());
                    tweets.add(tw);
                }
            }
        } catch (Exception e) {
            e.printStackTrace();
            System.out.println("Exception " + e);
        }
        return tweets;
    }

    private StringBuilder getData(String dataurl) throws MalformedURLException, IOException{
        URL url = new URL(dataurl);
        HttpURLConnection httpcon = (HttpURLConnection) url.openConnection();
        httpcon.addRequestProperty("User-Agent", "Mozilla/4.76");
        StringBuilder sb = new StringBuilder(16384);
        BufferedReader br = new BufferedReader(new InputStreamReader(httpcon.getInputStream(), "ISO-8859-1"));
        String line;
        while ((line = br.readLine()) != null){
            sb.append(line);
            sb.append('\n');
        }
        httpcon.disconnect();
        br.close();
        return sb;
    }

    public static void main(String [] args){
        //testing
        ReadSearch rs = new ReadSearch();
        ArrayList<Tweet> tweets = rs.getTweets("Tony_Kennah", "PickLuckier");
        for(Tweet t : tweets){
            System.out.println("TWEET: " + t.toString());
        }
    }
}

We then need the Tweet class itself so we can group Tweets up and do things with them, it's just a bean like this:

public class Tweet{ 
    private String user;
    private long tweetid;
    private String date;

    public String getUser(){
        return user;
    }
    public void setUser(String user){
        this.user = user;
    }
    public long getTweetid(){
        return tweetid;
    }
    public void setTweetid(long tweetid){
        this.tweetid = tweetid;
    }
    public String getDate(){
        return date;
    }
    public void setDate(String date){
        this.date = date;
    }
    public String toString(){
        return this.tweetid + " " + this.user + " " + this.date;
    }
}

... and so that was all just standard java. To make use of the above code I use the Twitter4J API and do this:

public class DeleteTweets
{
    public static void main(String args[]) throws Exception
    {
        Twitter twitter = TwitterFactory.getSingleton();
        ArrayList<Tweet> tweets = new ArrayList<Tweet>();
        String [] people = { "PickLuckier" };
        for(String s : people){
            do{
                ReadSearch rs = new ReadSearch();
                tweets = rs.getTweets(twitter.getScreenName(), s);
                for(Tweet tw : tweets){
                    twitter.destroyStatus(tw.getTweetid());
                }
            } while(tweets.size()!=0);
        }
    }
}

That's it. I don't use comments but I hope it's easy to see what's going on and that this helps you out.

Answer 2

Unfortunately, you cannot access past data from Twitter. Is not a problem of what library you're using: Tweepy, Twitter4J, whatever, is just that Twitter won't provide any data that is older than more or less 2 weeks.

To get historical data you'll need access to firehose, directly through Twitter or third-party resellers like GNIP.