TweetStream¶ ↑

TweetStream provides simple Ruby access to Twitter’s Streaming API (apiwiki.twitter.com/Streaming-API-Documentation).

This fork has been modified to support the site_streams API currently in Beta.

We have also added support for location-based queries to the streaming API.

Installation¶ ↑

To install from Gemcutter:

gem install tweetstream

Usage¶ ↑

Using TweetStream is quite simple:

require 'rubygems'
require 'tweetstream'

# This will pull a sample of all tweets based on
# your Twitter account's Streaming API role.
TweetStream::Client.new('username','password').sample do |status|
  # The status object is a special Hash with
  # method access to its keys.
  puts "#{status.text}"
end

You can also use it to track keywords or follow a given set of user ids:

# Use 'track' to track a list of single-word keywords
TweetStream::Client.new('username','password').track('term1', 'term2') do |status|
  puts "#{status.text}"
end

# Use 'follow' to follow a group of user ids (integers, not screen names)
TweetStream::Client.new('username','password').follow(14252, 53235) do |status|
  puts "#{status.text}"
end

The methods available to TweetStream::Client will be kept in parity with the methods available on the Streaming API wiki page.

Swappable JSON Parsing¶ ↑

As of version 1.0, TweetStream supports swappable JSON backends for parsing the Tweets. These are specified when you initialize the client or daemon by passing it in as the last argument:

# Parse tweets using Yajl-Ruby
TweetStream::Client.new('abc','def',:yajl) # ...

Available options are :yajl, :json_gem (default), :json_pure, and :active_support.

Handling Deletes and Rate Limitations¶ ↑

Sometimes the Streaming API will send messages other than statuses. Specifically, it does so when a status is deleted or rate limitations have caused some tweets not to appear in the stream. To handle these, you can use the on_delete and on_limit methods. Example:

@client = TweetStream::Client.new('user','pass')

@client.on_delete do |status_id, user_id|
  Tweet.delete(status_id)
end

@client.on_limit do |skip_count|
  # do something
end  

@client.track('intridea')

The on_delete and on_limit methods can also be chained, like so:

TweetStream::Client.new('user','pass').on_delete{ |status_id, user_id|
  Tweet.delete(status_id)
}.on_limit { |skip_count|
  # do something
}.track('intridea') do |status|
  # do something with the status like normal
end

You can also provide :delete and/or :limit options when you make your method call:

TweetStream::Client.new('user','pass').track('intridea',
  :delete => Proc.new{ |status_id, user_id| # do something },
  :limit => Proc.new{ |skip_count| # do something }
) do |status|
  # do something with the status like normal
end

Twitter recommends honoring deletions as quickly as possible, and you would likely be wise to integrate this functionality into your application.

Errors and Reconnecting¶ ↑

TweetStream uses EventMachine to connect to the Twitter Streaming API, and attempts to honor Twitter’s guidelines in terms of automatic reconnection. When Twitter becomes unavailable, the block specified by you in on_error will be called. Note that this does not indicate something is actually wrong, just that Twitter is momentarily down. It could be for routine maintenance, etc.

TweetStream::Client.new('abc','def').on_error do |message|
  # Log your error message somewhere
end.track('term') do |status|
  # Do things when nothing's wrong
end

However, if the maximum number of reconnect attempts has been reached, TweetStream will raise a TweetStream::ReconnectError with information about the timeout and number of retries attempted.

Terminating a TweetStream¶ ↑

It is often the case that you will need to change the parameters of your track or follow tweet streams. In the case that you need to terminate a stream, you may add a second argument to your block that will yield the client itself:

# Stop after collecting 10 statuses
@statuses = []
TweetStream::Client.new('username','password').sample do |status, client|
  @statuses << status
  client.stop if @statuses.size >= 10
end

When stop is called, TweetStream will return from the block the last successfully yielded status, allowing you to make note of it in your application as necessary.

Daemonizing¶ ↑

It is also possible to create a daemonized script quite easily using the TweetStream library:

# The third argument is an optional process name
TweetStream::Daemon.new('username','password', 'tracker').track('term1', 'term2') do |status|
  # do something in the background
end

If you put the above into a script and run the script with ruby scriptname.rb, you will see a list of daemonization commands such as start, stop, and run.

Note on Patches/Pull Requests¶ ↑

Fork the project.
Make your feature addition or bug fix.
Add tests for it. This is important so I don’t break it in a future version unintentionally.
Commit, do not mess with rakefile, version, or history. (if you want to have your own version, that is fine but bump version in a commit by itself I can ignore when I pull)
Send me a pull request. Bonus points for topic branches.

Contributors¶ ↑

Michael Bleigh (initial gem)

erichurst/tweetstream