Bot spams encouragement with multiple hashtags about 100daysofcode-twitter-bot

[spences10]

@robbawebba this was quite puzzling for me so I used my own twitter accounts for testing which worked gloriously well.

@amandeepmittal I think this is the way forward, just have contributors use the code base but use their own keys on either a throw away account or one they will be using for their own development.

Also 🎉 🎆 🍾

quite happy with myself at figuring that one out 😎

[spences10]

If no one else is looking at this I'm thinking about using level db add a user ID, something like.

db.put('twitterId_encourage', 'true')
db.put('twitterId_congratulate', 'true')

Then destroy the db every 24 hrs so no one is given too many inspirational quotes 😄

https://github.com/Level/level

[amandeepmittal]

@spences10 why don't you create a prototype of suggested with leveldb, and then we both can test it?

[amandeepmittal]

@spences10 Is it good to go? No breaks, I guess?

[spences10]

Ok, for some reason I couldn't trigger sentiment on my test bot on the #100DaysOfCode tag but the production bot found them just fine

I switched to my @DroidScott twitter account for the testing and got the result I wanted on #someTestHashTag and got the result I was after

So this is g2g, no changes needed to the PR #30

[spences10]

Ok, so I set up this bot on a different account and different hashtag and repeated the test so there are three accounts I'm testing with, so first I test with my personal Twitter account @spences10 output here:

I get the output we get on the @_100DaysOfCode bot production console, same but there is only one response from the bot account @DroidScott, so I test with my @ScottDevTweets account

I check the twitter account output:

So there is one tweet per user that has tweeted negative sentiment more than once:

So, it looks like there must be another instance of the @_100DaysOfCode twitter bot running

Solution?

Regenerate the Twitter keys, @amandeepmittal can you do this?

[amandeepmittal]

@spences10 Done

[robbawebba]

Lol I wish there was an easier way to test twitter bots locally instead of having to use the same auth tokens as production. 😅 This problem is also noted in #22

[spences10]

I have been getting multiple encouraging tweets, can we get the id of the tweet that triggered the response and then reply to that so there's an audit trail?

[spences10]

Hi @amandeepmittal I have and am currently educating myself with LevelDB, something I thought was quite straight forward has consumed quite a bit of time now.

I have db.js for the database

var level = require('level');  
var path = require('path');

var dbPath = process.env.DB_PATH || path.join(__dirname, 'mydb');  
var db = level(dbPath);

module.exports = db;

And I want to be able to query the db, so in app.js

var db = require('./db', {
  valueEncoding: 'json'
})

db.put('name', 'ID001')
db.put('name', 'ID002')
db.put('name', 'ID003')
db.put('name', 'ID004')
db.put('name', 'ID005')
db.put('name', 'ID006')
db.put('name', 'ID007')

db.createReadStream()
  .on('data', function (entry) {
    console.log(entry.value);
  })

Output is

spences10:~/workspace/level-db $ node app.js
ID006
spences10:~/workspace/level-db $ node app.js
ID007
spences10:~/workspace/level-db $ node app.js
ID007
spences10:~/workspace/level-db $ node app.js
ID005
spences10:~/workspace/level-db $ node app.js
ID007
spences10:~/workspace/level-db $ node app.js
ID007
spences10:~/workspace/level-db $ node app.js
ID006
spences10:~/workspace/level-db $ node app.js
ID005
spences10:~/workspace/level-db $ node app.js
ID003
spences10:~/workspace/level-db $ node app.js
ID007

That's from mashing the node app.js in the terminal

So, I'm trying to work out how to iterate or just get data from the db

[spences10]

Added PR #30 for this 👍 @amandeepmittal you can test from my branch if you like

https://github.com/spences10/100DaysOfCode-twitter-bot/tree/add-user-blacklist

or if you prefer I can give you access to my c9 environment as a collaborator, mail me: [email protected]

[spences10]

@amandeepmittal yis g2g 😎

[spences10]

@amandeepmittal maybe I should explain the changes before you merge?

Added logic to the sentiment detection, and add user name as the key and encourage as the value

There's a check to see if the screen_name is already in the db before the .put

      // if sentiment is Negative and the confidence is above 75%
      if (sentim == 'Negative' && confidence >= 75) {
        // get a random quote
        var phrase = sentiment.randomQuote()
        var screen_name = tweet.user.screen_name
        // Check key isn't in db already, key being the screen_name
        db.get(screen_name, function(err, value) {
          if (typeof(value) !== 'undefined') {
            console.log('ALREADY IN DB USER ', screen_name);
          }
          else {
            // Put a user name and that they have been encouraged 
            db.put(screen_name, 'encourage', function(err) {
              if (err) return console.log('Ooops!', err) // some kind of I/O error
              console.log('LOGGED USER ', screen_name)
              // tweet a random encouragement phrase
              tweetNow('@' + screen_name + ' ' + phrase)
            })
          }
        })
      }

I also added in fs-extra so that the database could be rebuilt each day

var refreshDB = function() {
  var fs = require('fs-extra')

  fs.remove('./blacklistUsersDb', function(err) {
    if (err) return console.error(err)

    console.log('success!')
  })
}

refreshDB()
  // retweet every 24 hrs
setInterval(refreshDB, 60000 * 1440)

Maybe I could add something a bit more detailed other than success for the deletion of the database though.

[spences10]

Ok, this is still an issue @Ilyes-Hammadi are you able to add anything to this?

The logging for the database functions as expected, see the output here:

That has my twitter handle ScottDevTweets added to the database, but the output from the bot is double as per the original issue:

So the next time I tweet with negative sentiment I get this output:

So the maximum any user is going to get currently is two every 24 hours, this should be one.

Technically it shouldn't post the two tweets as the second tweet should be handled

        // Check key isn't in db already, key being the screen_name
        db.get(screen_name, function(err, value) {

          if (typeof(value) !== 'undefined') {
            console.log('ALREADY IN DB USER ', screen_name);
          }
          else {
            // Put a user name and that they have been encouraged 
            db.put(screen_name, 'encourage', function(err) {
              if (err) return console.log('Ooops!', err) // some kind of I/O error

              console.log('LOGGED USER ', screen_name)

              // tweet a random encouragement phrase
              tweetNow('@' + screen_name + ' ' + phrase)

            })
          }

        })

If the user key is in the database then tweetNow isn't called, @amandeepmittal is that your understanding too?

I'm going to do some more testing on the #someTestHashTag I'd like to hear peoples thoughts 😄 ❓

[spences10]

Anyone going to own up to leaving the bot running?

[amandeepmittal]

@robbawebba The purpose of giving access to tokens is that user can test themselves. But I guess, the purpose is no longer fulfilled.

[amandeepmittal]

@spences10 you were absolutely right. Someone was running another instance of the bot. This issue can be closed once again!