Wednesday, August 12, 2015

Wired or weird Tweets? Doesn't matter to Bluemix, DB2, and Node-RED...

Robot with
a personality?
Someone to talk to, someone to take the meeting minutes. Well, I can offer that - at least partially. I took the Node-RED service on Bluemix for a test drive and
created a Twitter service. It automatically can respond to tweets and can also store the incoming messages in a DB2 table. In my example it is just a, more or less, silly service right now, but it could be expanded to react to and log Twitter-based customer feedback, provide some basic interim reaction or acknowledgement. Let me show you how easy it is to get started...

To build the Twitter service you need the "Node-RED Starter" boilerplate from the Bluemix catalog as well as the "SQL Database" service. Create the Node-RED boilerplate first, then add the database service and directly bind it to the new application. Once everything has been provisioned, it is time for wiring up the Twitter service in the flow editor. The following picture gives an overview of the two flows when completed. I made the flows available in the Node-RED library.

Flow in Node-RED for Twitter robot and reporting
The lower flow consists of a Twitter input node to receive incoming tweets for a user. The tweet is then analyzed for known hashtag commands (Process Twitter Input). If "#archive" has been specified, some data is handed over to the SQLDB output node (green, "tweet store"). In all cases a response is prepared and then a reply tweeted, in some cases also a delayed follow-up response.

The upper flow is used for Web-based reporting on stored tweets. An incoming http request is routed to a SQLDB query node. There the tweet data is fetched from DB2, handed over to a template to produce an HTML page with the result table which finally is returned as http response.

Without discussing all the details, I wanted to provide some information on how to work with DB2 from within Node-RED. There are input and output nodes for the SQLDB and the dashDB services. To store a single record, a "payload" with the column name/value pairs needs to be created. For the example I used a simple table "twitarchive" to hold the archived tweets. Here is the DDL:

create table twitarchive(
  id int generated always as identity,
  tstamp timestamp,
  tweet varchar(200),
  username varchar(100)
);

Because the value for the ID column is generated, I only need to pass on the values for the remaining columns. This is done in the "Process Twitter Input" node by using the structure "twitarchive":

    var twitarchive = { payload : {
        TSTAMP: 'TIMESTAMP',
        USERNAME: msg.user,
        TWEET: msg.tweet }
    };


It is important to use the same column names as in DB2, including upper casing. Else there won't be a match. In the input and output nodes for sqldb/dashDB access the same rule applies to specifying table names (remember: if no quotes are used for table names, names are converted to upper case).

I used a template node to format query results into a HTML table. The template and the resulting output are shown below.

This concludes the overview. Let me know if you are interested in details. And if you want to test the service, tweet to "@db2bm". The hashtag "#archive" in the tweet (when found :) causes the message to be store, "#autoanswer" produces a more elaborate random reply.




 <html>  
   <head>  
     <style>  
 table, th, td {  
   border: 1px solid black;  
 }  
 th {  
   text-align: left;  
 }  
 td {  
   padding: 15px;  
 }  
 th {  
   background-color: green;  
   color: white;  
 }  
   
 </style>  
     <title>Archived Tweets</title>  
   </head>  
   <body>  
     <h1>Archived Tweets</h1>  
     The following tweets have been archived in DB2 so far:  
     <table>  
       <tr><th>User</th><th>Tweet</th><th>Time</th></tr>  
       {{#payload}}<tr>  
         <td><a href="http://twitter.com/{{USERNAME}}">{{USERNAME}}</a></td><td>{{TWEET}}</td><td>{{TSTAMP}}</td>  
       </tr>{{/payload}}  
     </table>  
   </body>  
 </html>  


DB2-based result table with some archived tweets

No comments:

LinkWithin

Related Posts with Thumbnails