Originally published December 15, 2018 @ 9:14 pm

The ‘t’ is an excellent Ruby-based CLI utility for interacting with the Twitter API written by Erik Berlin. This is certainly not the only such tool available, but, in my estimation, it is the most full-featured and expertly-written. No amateur-hour coding here.

Here I am showing how to install and use t on CentOS 6. I am a professional Unix sysadmin and so getting me to upgrade anything takes a lot of incentive. Having said that, installing t on a more recent version of Linux is, obviously, a lot easier.

Installing Things

First, we’ll need to install some prerequisites. Mostly Ruby-related stuff for obvious reasons:

yum install ruby ruby-devel
yum install gcc g++ make automake autoconf curl-devel openssl-devel zlib-devel httpd-devel apr-devel apr-util-devel sqlite-devel
yum install ruby-rdoc ruby-devel
yum install rubygems

At this point I realized that I need Ruby 2.2+ and that’s not on tap for my OS version. So I had to get it the hard way:

yum install gcc-c++ patch readline readline-devel zlib zlib-devel
yum install libyaml-devel libffi-devel openssl-devel make
yum install bzip2 autoconf automake libtool bison iconv-devel sqlite-devel
curl -sSL https://rvm.io/mpapis.asc | gpg --import -
curl -L get.rvm.io | bash -s stable
source /etc/profile.d/rvm.sh
rvm reload
rvm requirements run
rvm install 2.2.4
rvm use 2.2.4 --default
ruby --version

And now to install t itself:

gem install t

Twitter Authorization

Since July 2018 Twitter tightened up requirements for a developer’s account. Now you need to submit an application and get approved before you can authorize your app. The authorization mechanism supplied with t does not function any longer, unfortunately.

So we have to jump through some hoops here. Apply for the dev account; create the application; and use twurl (https://github.com/twitter/twurl) to handle authorization.

twurl authorize --consumer-key ***** --consumer-secret *****
# You will need to open the displayed URL, approve the app access, and copy back the displayed PIN.
/bin/cp -p ~/.twurlrc ~/.trc
t accounts

For a complete overview of the available t commands, just hop over to it’s developer’s GitHub page.

Installing Optional Tools

When dealing with Twitter API, you’re likely to run into lots of CSV files. Dealing with CSV can be a pain, especially when the data is complex and multi-lingual. The csvfix tool will help you to fix your CSV files and massage them in various ways.

wget https://bitbucket.org/neilb/csvfix/get/version-1.6.zip
unzip version-1.6.zip
cd neilb-csvfix-e804a794d175
make lin
cd csvfix/bin
/bin/cp csvfix /usr/bin

The `timeline-csv-analysis.sh` script I forked from some dude’s coursework repo and modified it somewhat. What this script does now is it grabs the last whatever number of your tweets ad a list of common English words. And then it shows you most common hashtags (minus the most common words), likes, retweets, etc.

wget -O ~/timeline-csv-analysis.sh --no-check-certificate https://raw.githubusercontent.com/igoros777/compciv/master/homework/congress-twitter/timeline-csv-analysis.sh
chmod 750 ~/timeline-csv-analysis.sh
cd ~
mkdir ~/data-hold/
~/timeline-csv-analysis.sh igor_os777 1000 #last number shows how many tweets to download

And just in case you’re wondering, here’s what the output of this script looks like for me:

Sample output
[root@ncc1711:~] 1 # ~/timeline-csv-analysis.sh igor_os777 1000 
Analyzing 1001 tweets by igor_os777 since "2018-12-10 12:27:07 +0000"
Top 10 hashtags by igor_os777
     39 #tc
     19 #tech
     15 #science
     12 #culture
      9 #europe
      8 #prices
      5 #mobile
      5 #cloud
      4 #column
      3 #sec
Top 10 retweeted users by igor_os777
      1 @marmar9282
Top 10 mentioned users (not including retweets) by igor_os777
      1 @stewartbrewing
      1 @scottdevtweets
      1 @sanitariumone
      1 @rodriguezcommaj
      1 @nader_shalabi
      1 @kevscott
      1 @jwveldhuis
      1 @judyamazing
      1 @jremes84
      1 @ioanabarna2
Top tweeted 10 words with 5+ letters by igor_os777
     80 Google
     43 Apple
     17 Android
     15 Science
     13 Culture
     12 Could
     12 Cloud
     11 Future
     11 Europe
      8 Prices

Some Examples

The function below will search the latest tweets for keywords picked at random from your keyword list. And then it will follow a few authors of those tweets. Some of the variables here:

max_keywords – the number of words to pick at random from the keyword array

max_count – the maximum number of people to follow

logfile – the log file used to store script output and make sure you don’t try to follow people you already following

function func_follow_keyword() {
  echo "Following by keyword"
  declare -a follow_keywords=('linux' 'gnulinux' 'sysadmin' 'opensource' 'ubuntu' 'kalilinux' 'gnu' 'kernel' 'devicetree' 'network_security' 'computer_security' 'blockchain' 'bitcoin' 'machine_learning' 'artificial_intelligence' 'Android' 'AppDev' 'BigData' 'CloudComputing' 'Datamining' 'Hadoop' 'Microsoft' 'OpenFlow' 'OpenCloud' 'OpenSource' 'PredictiveAnalytics' 'Techfail' 'VMWare' 'WebDev' 'WebDesign' 'ZeroDay')
  for i in $(printf '%s\n' ${follow_keywords[@]} | shuf | head -${max_keywords})
    k="$(echo ${i} | sed -e 's/^/#/g' -e 's/_/ #/g')"
    q="lang:en ${k}"
    echo "Following based on keyword: ${q}"
     t follow $(t search all "${q}" -n ${max_count} | grep -vP "\bRT\b" | grep -v "${u}" | grep -v -P "@.*http" | \
     grep -P -B1 http | grep -P '^[ ]{1,}@\S*' | grep -vP "$(grep "t unfollow " "${logfile}" | \
     grep -oP '@\S*' | sed -s 's/[`]//g' | sort -u | xargs | sed 's/ /|/g')" | xargs)
    echo ""

Here’s another function that will retweet a certain number of tweets based on keyword. The variables here are some of the same from the previous example. The one extra touch here is the favorite_exclude variable that is set by scanning the logfile for users you have unfollowed in the past.

function func_retweet_by_keyword() {
  echo "Retweeting by keyword"
  declare -a favorite_keywords=('linux' 'gnulinux' 'sysadmin' 'opensource' 'ubuntu' 'kalilinux' 'gnu' 'kernel' 'devicetree' 'network_security' 'computer_security' 'blockchain' 'bitcoin' 'machine_learning' 'artificial_intelligence' 'Android' 'AppDev' 'BigData' 'CloudComputing' 'Datamining' 'Hadoop' 'Microsoft' 'OpenFlow' 'OpenCloud' 'OpenSource' 'PredictiveAnalytics' 'Techfail' 'VMWare' 'WebDev' 'WebDesign' 'ZeroDay')
  favorite_exclude="$(grep "delete favorite" "${logfile}" | grep -oP "\b[0-9]{19,}\b" | sort -u | xargs | sed 's/ /|/g')"
  for i in $(printf '%s\n' ${favorite_keywords[@]} | shuf | head -3)
    k="$(echo ${i} | sed -e 's/^/#/g' -e 's/_/ #/g')"
    q="lang:en ${k}"
    echo "Retweeting based on keyword: ${q}"
     t retweet $(t search all "${q}" -n ${max_count} --csv | grep -P "^[0-9]{19}," | grep -vP "(RT)\s" | awk -F, '!($3 in a){a[$3];print}' | awk -F, '!($4 in a){a[$4];print}' | grep -P "http(s)?:" | awk -F, '{print $1}' | grep -oP "^[0-9]{19}" | grep -vP "${favorite_exclude}" | shuf | head -n ${max_count_retweet} | xargs)
    echo ""

The function below will delete your tweets that contain specified keywords. This can be useful if you’re using a bot to post some of your tweets and sometime undesired material sneaks in.

function func_delete_keyword() {
  echo "Deleting by keyword"
  declare -a delete_keywords=('best_deals' 'black_friday' 'kinjadeals' 'amazondeals')
  for i in $(printf '%s\n' ${delete_keywords[@]})
    k="$(echo ${i} | sed 's/_/ /g')"
    echo "Deleting based on keyword: ${k}"
    for id in $(t search timeline @${u} "${k}" --csv | grep -v ^ID | grep ",${u}," | awk -F, '{print $1}')
      echo y |  t delete status ${id}
    echo ""

Similar to the previous example, the function below will delete your tweets that tell you can buy something for a price. Very simple but surprisingly effective.

function func_delete_ads() {
  echo "Deleting ads"
  echo y |  t delete status $(awk -F, '{print $1}' <(t search timeline --csv | grep -v ^ID | \
  grep ",${u}," | grep -P '$[0-9]{1,}(,)?(\.[0-9]{1,})?\s' | grep -v illion) | xargs)

Here’s another function – and, I promise, the last one – it will unfollow people who haven’t tweeted in a long time. The threshold is specified in months.

function func_unfollow_stale() {
  echo "Unfollowing stale"
  (( threshold = threshold_months * 30 * 24 * 60 * 60 ))
   t followings -l --sort=tweeted --csv | grep -v "^ID," | while read i
    last_tweeted="$(date -d"$(echo "${i}" | awk -F, '{print $3}')" +'%s' 2>/dev/null)"
    if [[ ${last_tweeted} =~ ${re} ]] ; then
      screen_name="$(echo "${i}" | awk -F, '{print $9}')"
      (( time_diff = this_time_epoch - last_tweeted ))
      if [[ ${time_diff} =~ ${re} ]] ; then
        if [ "${time_diff}" -ge "${threshold}" ]
          echo "@${screen_name}" >> "${tmpfile}"
   t unfollow $(cat "${tmpfile}" | xargs)
  /bin/rm -f "${tmpfile}"

So there you go: Twitter is pretty much a bunch of cron jobs conversing with each other.

Added bonus: deleting your old tweets

The first step is to download your tweet archive. Tweeter’s own documentation for this process is outdated. You do need to go into “Account Settings”, but then you go to “Your Twitter Data”, scroll all the way to the bottom, and under “Download your Twitter data” select “Twitter”.

You will see something like the screenshot below and the process will take a long while, so go grab a coffee or, better yet, go to sleep.

Early in the morning check your email for the download link from Twitter. A word to the wise: check your spam folder, for this is where Twitter emails usually end up (or should end up). Depending on the severity of your social media disorder, the archive file may be large: mine was 2GB+ compressed.

The file you eventually download will be twitter-YYYY-mm-dd-*.zip. The only file you need from this archive is tweet.js, so you can save yourself some time by just unzipping what you need:

unzip -p twitter*.zip data/tweet.js > tweet.js

After extracting the file, remove this string from it: window.YTD.tweet.part0 = . Make sure the first line of the file now begins with a left square bracket [

There are many elements associated with each tweet in this file, but our aim here is simple: find your tweets older than, say, six months, and delete them, if they don’t have any likes or re-tweets. The script below is not the most efficient, but it will get the job done eventually. You can also download it here.

# Parse Twitter's 'tweet.js' data file and delete old posts that have no likes or retweets
# For details see https://www.igoroseledko.com/installing-t-cli-power-tool-for-twitter/
# Extract tweet.js from Twitter data archive and remove this string from it:
# window.YTD.tweet.part0 =
# Make sure the first line of the file now begins with a left square bracket [

if [ -z "" ]; then
  echo "Specify the location of tweet.js"
  exit 1

if [ -z "" ]; then
  echo "Specify the username without '@'"
  exit 1

if [ ! -f "${infile}" ]; then
  echo "Input file ${infile} not found. Exiting..."
  exit 1

if [ ! -x "${T}" ]; then
  echo "Unable to access ${T}"
  exit 1
read -r ux <<<$(${T} accounts | sed 'N;s/\n/ /' | grep "${u}" | awk '{print $2}')
${T} set active ${u} ${ux}

mt="$(date -d '3 months ago' +'%s')"

echo "Writing Tweet IDs to ${f}"

id_check() {
  if [ ! -z "${id}" ] && [ ! -z "${fc}" ] && [ ! -z "${rc}" ] && [ ! -z "${ct}" ]
    if [ ${fc} -eq 0 ] && [ ${rc} -eq 0 ] && [ ${ct} -lt ${mt} ]
      echo "${id}" | tee -a "${f}"

line_parse() {
  read id <<<"$(cut -d@ -f1 <<<"${line}")"
  read fc <<<"$(cut -d@ -f2 <<<"${line}")"
  read rc <<<"$(cut -d@ -f3 <<<"${line}")"
  ct="$(date -d "$(cut -d@ -f4 <<<"${line}")" +'%s')"
  id_check &

jq -r '.[] | .[] | .id + "@" + .favorite_count + "@" + .retweet_count + "@" + .created_at' 2>/dev/null <"${infile}" | while read line
  line_parse &
${T} delete status -f $(sort -u "${f}" | xargs -n100 -P$(grep -c proc /proc/cpuinfo)) 2>/dev/null
/bin/rm -f "${f}"