Twitter Interactive

Elvenware Logo

TOC

Twitter Query

The goal of this project is to query Twitter. We will ask for details about our account, but also about things that are happening on Twitter.

While working on this project, use your Delcious and Bitly projects as references. There is lots of code from those programs that will also be used in this project.

Step One

Make sure you have the most recent scripts:

cd $JSOBJECTS
git pull
cp $JSOBJECTS/Utilities/NodeInstall/CreateExpressProject ~/bin/.
cp $JSOBJECTS/Utilities/NodeInstall/TestReady ~/bin/.
cp $JSOBJECTS/Utilities/SetupLinuxBox/strip-triple-spaces ~/bin/.

If you don't have js-beautify installed globally, which you well might not, then run this command:

npm install -g js-beautify

Or use a different name for my scripts if you have custom versions with the same name.

I think most of you can then type i3 to get back to your repositories.

export TWITTER_PROJECT=Week07-Twitter
CreateExpressProject $TWITTER_PROJECT
cd $TWITTER_PROJECT
TestReady

Now run the following commands, one at a time, to confirm that all is well:

grunt check
grunt test

Step Two

To get started we need to get a consumer_key and access_token_key from Twitter:

  1. Go to: https://apps.twitter.com/
  2. Sign in.
  3. Make sure you have a mobile phone number set up on your account
  4. Generate an application
  5. On the application management page, choose Create New App
    1. name: bc-tweeter-lastname
    2. description: learning the twitter api
    3. website: http://www.example.com (or what pleases you)
    4. callback http://www.example.com/bc-tweeter-lastname
  6. Generate the consumer key (Button one of two)
  7. Then generate the access token. (Button two of two)

Step Three

Querying Twitter is perhaps unnecessarily complex. Fortunately, there is a node package that handles the details for us. To get twitter support, start by adding the package:

npm install twitter --save

This is node package, so it works on the server side. This is as it should be, since it keeps your keys hidden from users.

Our server side code can go in routes/index.js. First we want to include (require) the twitter package we downloaded. Then we create an elfTwitterClient from our Twitter package and use it when we query Twitter:

var Twitter = require('twitter');
var elfTwitterClient;

function getClient() {
    if (!elfTwitterClient) {
        elfTwitterClient = new Twitter({
            consumer_key: 'YOUR CONSUMER KEY HERE',
            consumer_secret: 'YOUR CONSUMER SECRET HERE',
            access_token_key: 'YOUR TOKEN KEY HERE',
            access_token_secret: 'YOUR TOKEN SECRET HERE'
        });
    }
    return elfTwitterClient;
}

Here is a request for information about #node.js:

router.get('/search', function (req, res, next) {
    var client = getClient();

    var params = {q: '#node.js'};
    client.get('search/tweets', params, function (error, tweets, response) {
        if (!error) {
            console.log(tweets);
            res.send(tweets);
        } else {
            console.log(error);
            res.send({fail: error});
        }
    });
});

We use client.get to retrieve information from Twitter, and we use res.send to send that information from the server to the browser.

We'll work on getting random searches a bit later.

Step Three

Get our mixins:

cp ~/Git/JsObjects/Utilities/Templates/JadeMixins/* views/.

Step Four

Notice that our search function calls router.get. The router object is built into express. The get function is a bit of middleware that is called when the browser sends a request to the server. In particular, it responds to URLs that contain the paramter search:

http://localhost:30025/search

Go ahead and paste that URL into your browser's address bar. Your server side method should get called and a lot of JSON should get send back and displayed in a big blob in your browser. To sort out that data, paste it into http://jsonlint.com/.

Of course, we want something a bit more fancy. But by now, you know how to transform data big blobs of JSON into code that someone can use.

The first step might be to declare a little jade:

+elfPanel("Bitly Links Table").elfDiv
    div.scroller
        table.table#tableLinks

+elfPanel("Debug", "panel-default").elfDiv
    button.btn.btn-default#getTweets(type='button') Get Tweets
    button.btn.btn-default#search(type='button') Search

    div
        ul#tweetList

    div
        pre#tweetData

The second step might be a call to getJSON:

function search() {
    $.getJSON('/search', function(result) {

        // STEP ONE:
        //    STRINGIFY THE RESULT AND DISPLAY IT IN THE PRE
        //    TAG WITH THE ID OF: #tweetData

        // STEP TWO:
        //   ITERATE OVER EACH ITEM IN THE RESULT'S STATUSES ARRAY
        //   AND DISPLY THE text PROPERTY IN THE #tweetList

    });
}

$(document).ready(function() {
    'use strict';

    $('#search').click(search)
});

At first, just display the text property from statuses array. Once that is working, you might want to get the first url from the urls array in the entities property and turn the text field into a hyperlink. More specifically:

Our code should add the tweet text:

If it does not have a url, it should add it:

It is very important that you test to make sure there are actually are URLs in the entities.urls array. If there are none, and you try to access one, then your code will throw a variable undefined error.

The data you get back is shaped a bit like this:

{
  "statuses": [
        {
            "metadata": {
                "iso_language_code": "en",
                "result_type": "recent"
            },
            "created_at": "Tue Nov 03 19:25:02 +0000 2015",
            "id": 661625149184213000,
            "id_str": "661625149184212992",
            "text": "Understanding Components in Ember 2 https://t.co/iaKG6y6wIt #dev #node #js",
            // CODE OMITTED HERE
            "user": {
                "id": 3011007497,
                "id_str": "3011007497",
                "name": "Codi",
                "screen_name": "codi_ar",
                "location": "",
                "description": "",
                "url": null,
                "entities": {
                    "description": {
                        "urls": []
                    }
                },
                "protected": false,
                "followers_count": 11,
                // CODE OMITTED HERE
            },
            "geo": null,
            // CODE OMITTED HERE
            "entities": {
                "hashtags": [
                    { "text": "dev", "indices": [60, 64] },
                    { "text": "node", "indices": [65,70] },
                    { "text": "js", "indices": [71,74] }
                ],
                "symbols": [],
                "user_mentions": [],
                "urls": [
                    {
                        "url": "https://t.co/iaKG6y6wIt",
                        "expanded_url": "http://buff.ly/1Q3yPje",
                        "display_url": "buff.ly/1Q3yPje",
                        "indices": [36,59]
                    }
                ]
            },
            // CODE OMITTED HERE
        },
        // MANY SIMILAR OBJECTS
   ],
   search_metadata: {
        completed_in: 0.026,
        max_id: 661625223423385600,
        max_id_str: '661625223423385602',
        next_results: '?max_id=661623659165937663&q=%23node.js&include_entities=1',
        query: '%23node.js',
        refresh_url: '?since_id=661625223423385602&q=%23node.js&include_entities=1',
        count: 15,
        since_id: 0,
        since_id_str: '0'
     }
 }

Here's a couple utility methods you can use if you don't want to write it yourself:

function appendUrl(selector, index, text, url) {
    var anchor = '<a href="' + url + '" target="_blank">' + text + '</a>';
    var details = '<a onclick="detailDelicious(' + index + ')">Details</a>';
    $(selector).append('<li>' + anchor + ' - ' + details + '</li>');
}

Step Five

The final step will be to use our scrolling table for items without links, and our list for any items that have links. Our vice-versa if you prefer. I'm not sure which is best. The point for now is just to get practice with both ways of displaying data.

Here is a method that you might find useful:

var renderTable = function (text, user) {
    'use strict';
    var title = '<td>' + text + '</td>';
    var keyword = '<td>' + user + '</td>';
    var tableRowStart = '<tr>';
    $('#tableLinks').append(tableRowStart + title + keyword + '<tr>');
};

You would want to call this from your loop in the success callback for getJSON. The user should be the user name: result.user.name (or something like that).

Step 5.5 Startup

NOTE: No one has to follow the structure defined in this section. But if you need help building the application, this section might help.

Normally, we would create a method like this that would initilize our buttons:

$(document).ready(function() {
    $('#search').click(twitterRefine.search);
});

Let's move it into its own method:

var init: function() {
    $('#search').click(twitterRefine.search);
}

$(document).ready(function() {
    'use strict';
    init();
});

Then, as we refine the code further, let's move it into an object that holds all the main methods for the program:

var twitterInteractive = {

    init: function() {
        $('#search').click(twitterRefine.search);
    },

    appendUrl: function (selector, index, text, url) {
        // CODE OMITTED HERE
    },

    // LOTS OF CODE OMITTED HERE

};

$(document).ready(function() {
    'use strict';
    twitterInteractive.init();
});

There is a unit test called test-twitter-core.js and a file called tweets.js in $ELF_TEMPLATES/UnitTest/TwitterInteractive that can help you define this object. Just copy the test into your spec folder. Don't forget to load tweets.js in karma.conf.js

Turn it in

Per usual.