Reputation: 305
Trying to create new tweets to display in my index, but I don't wanna leave the page. My app's desired function is to track tweets from tweetstream, and automatically pass that info into my tweet object and save to db.
Controller:
class TweetsController < ApplicationController
TWITTER_COMSUMER_KEY = "GfqdzJKb5kIyEnYlQuNGlg"
TWITTER_CONSUMER_SECRET = "A3Fe0IvDbhlKgowCVmV1WVLlcdYgQ8w9clrDSegCQ"
TWITTER_OATH_TOKEN = "34012133-caUYq3eiNC7Z9L9KvTgG51VgyctqVxkXP0tKIXDk0"
TWITTER_OATH_TOKEN_SECRET = "DSLA3F8BPssyEeEP2wZgQ1OJRL5kIVPZfON4GYZFw"
TweetStream.configure do |config|
config.consumer_key = TWITTER_COMSUMER_KEY
config.consumer_secret = TWITTER_CONSUMER_SECRET
config.oauth_token = TWITTER_OATH_TOKEN
config.oauth_token_secret = TWITTER_OATH_TOKEN_SECRET
end
def index
@tweets = Tweet.all
end
def new
@tweet = Tweet.new
end
def create
TweetStream.track('bed', 'morning', 'breakfast') do |status|
temp = status.text
if(temp.include? "http")
@tweet = Tweet.new(status.text)
if @tweet.save
else
render "new"
end
end
end
end
def show
end
end
Index.html.erb
<h1>Tweet Tracker</h1>
<% @tweets.each do |tweet| %>
<p><%= tweet.content %></p>
<hr />
<% end %>
Upvotes: 2
Views: 160
Reputation:
You've got two big pieces of this:
The second bit can be done by some simple javascript polling, as in @AustinMullins's answer.
The first bit should not be done in a controller - they are for responding to requests and deviating from that kind of job may end up with unexpected behavior or performance issues.
For example, I found that on a site that's running on Phusion Passenger, the server would create a thread to handle a request, then kill it after a certain amount of time if it didn't finish by itself, which of course it would not if the controller starts listening to a neverending input stream.
Instead, you should get a separate script that you can start from the command line. Here's an example similar to what I'm using:
script/tracker.rb
#!/usr/bin/env ruby
ENV["RAILS_ENV"] ||= "production"
require File.dirname(__FILE__) + "/../config/environment"
TweetStream.configure do |config|
config.consumer_key = TWITTER_COMSUMER_KEY
config.consumer_secret = TWITTER_CONSUMER_SECRET
config.oauth_token = TWITTER_OATH_TOKEN
config.oauth_token_secret = TWITTER_OATH_TOKEN_SECRET
end
client = TweetStream::Client.new
client.track(*Keyword.pluck(:name)).each do |status|
Tweet.create(status.text)
end
script/tracker_ctl
#!/usr/bin/env ruby
require 'rubygems'
require 'daemons'
Daemon.new('tracker.rb')
Now, you could ssh into your server and run script/tracker_ctl start
or script/tracker_ctl stop
. But what you probably want to do is make a controller action issue those commands instead.
Upvotes: 1
Reputation: 7427
This is an interesting approach to the RESTful resource structure (index/show/create/edit/destroy), in that we're utilizing an automated tracker to do the creating rather than any user input. The problem is, your controller code is only run when accessed by a POST request (unless you've messed with routes.rb to make it respond to something else).
The cleanest solution in my mind is to write a coffee script to periodically post data to the controller and show the results.
$ ->
setInterval ->
$.post("/tweets", "", (html)->
$("#tweets").append html)
, 1000
<p><%= @tweet.content %></p>
<br/>
For this to work right, we need to wrap the tweets area in a div with id="tweets":
<h1>Tweet Tracker</h1>
<div id="tweets">
<% @tweets.each do |tweet| %>
<p><%= tweet.content %></p>
<br/>
<% end %>
</div>
You'll also want to replace render "new"
in the controller's create method with return
. This way, nothing is rendered when a save to the database fails, and everything continues with no errors.
Upvotes: 1