get_ganalytics_data.pl

#!/usr/bin/perl

use strict;
use warnings;
use Net::Google::Analytics;
use Net::Google::Analytics::OAuth2;
use DateTime::Format::Strptime;
use POSIX qw(ceil strftime);


# GLOBAL VARIABLES
# I KNOW THIS IS BAD...SO SHOOT ME
my $startDate; # Format should be '2015-07-01';
my $endDate; # Format should be '2015-08-01';
my $timestamp = `date +"%Y%m%d%H%M"`;
chomp($timestamp);

#
# VARIABLES TO LOG INTO YOUR GOOGLE ANALYTICS ACCOUNT
# NEED TO INSTALL THIS LIBRARY FROM CPAN: https://metacpan.org/pod/Net::Google::Analytics
# AND FOLLOW THE INSTRUCTIONS ON THAT PAGE TO GET YOUR REFRESH TOKEN
my $profileId = "INSERT_PROFILE_ID";
my $clientId = "INSERT_CLIENT_ID";
my $clientSecret = "INSERT_CLIENT_SECRET";
my $refreshToken = "INSERT_REFRESH_TOKEN";

#
# THIS FUNCTION GETS THE ARGUMENTS PASSED TO THE SCRIPT WHEN IT IS RUN
sub parse_args()
{
  my $arg;
  my @argv = @ARGV;
  
  if($#argv == "-1")
  {
    help();
  }
  
  while(@argv)
  {
    $arg = shift(@argv);
    if($arg eq "-s" || $arg eq "-S")
    {
      if($#argv == "-1")
      {
        die "Error: Missing start date after -s option\n";
      }
      $startDate = shift(@argv);
    }
    elsif($arg eq "-e" || $arg eq "-E")
    {
      if($#argv == "-1")
      {
        die "Error: Missing end date after -e option\n"; 
      }
      $endDate = shift(@argv);
    }
    elsif($arg eq "-h" || $arg eq "-H")
    {
      help();
    }
    else
    {
      die "Illegal option: $arg.\n";
    }
  }   
}    
  
sub help
{
  print "Usage:\n",
  "   get_shopify_data.pl <OPTIONS>\n",
  "     Required:\n",
  "       -s START_DATE      : The first day in the period you want to pull data for. In YYYY-MM-DD format, e.g. 2015-07-01 \n",
  "       -e END_DATE        : The last day in the period you want to pull data for. In YYYY-MM-DD format, e.g. 2015-07-31 \n",
  "       -h                 : See the help info \n";     
  die "\n";
}  
  


sub get_data
{
  #
  # CREATE THE LOG FILE NAMES AND ADD THE RANGE OF DAYS WE ARE GETTING DATA FOR TO IT
  my $parser = DateTime::Format::Strptime->new(pattern => '%Y-%m-%d');
  my $origStartDate = $parser->parse_datetime($startDate);
  my $origEndDate = $parser->parse_datetime($endDate);
  my $dateRange = $origStartDate->strftime('%m%d%Y') . "-" . $origEndDate->strftime('%m%d%Y');
  my $logfile = "GoogleAnalytics_raw_data_$dateRange\_$timestamp.csv"; #where the raw data is stored
  
  #
  # VARIABLES TO HOLD METRICS
  my $totalUsers = 0;
  my $newUsers = 0;
  my $returnUsers = 0;
  
  #
  # CREATE A TIMESTAMPED FILENAME FOR WRITING THE RAW DATA 
  # AND THEN OPEN THE FILE TO WRITE TO
  my $timestamp = strftime("%Y%m%d%H%M", localtime);
  
  #
  #
  # NOW THE CODE TO ACTUALLY DO THE PULL
  
  my $analytics = Net::Google::Analytics->new;
  
  #
  #Authenticate
  my $oauth = Net::Google::Analytics::OAuth2->new
  (
    client_id => $clientId,
    client_secret => $clientSecret,
  );
  # UNCOMMENT THIS THE FIRST TIME YOU RUN IT TO GO THROUGH THE PROCESS
  # OF GETTING YOUR REFRESH TOKEN
  #$oauth->interactive;
  my $token = $oauth->refresh_access_token($refreshToken);
  $analytics->token($token);
  
  #
  # BUILD GOOGLE ANALYTICS REQUEST TO GET TOTAL UNIQUE VISITORS TO THE WEBSITE
  my $req = $analytics->new_request
  (
    ids        => "ga:$profileId",
    metrics    => "ga:users",
    start_date => $startDate,
    end_date   => $endDate,
  );
  
  #
  # SEND REQUEST
  my $res = $analytics->retrieve($req);
  die("GA error: " . $res->error_message) if !$res->is_success;
  
  #
  # SET THE VARIABLE TO TOTAL USERS FOR THE SELECTED PERIOD     
  $totalUsers = $res->totals("users");
  
  #
  # BUILD GOOGLE ANALYTICS REQUEST TO GET NEW VISITORS AND RETURN VISITORS
  $req = $analytics->new_request
  (
    ids        => "ga:$profileId",
    dimensions => "ga:userType",
    metrics    => "ga:users",
    start_date => $startDate,
    end_date   => $endDate,
  );
  
  #
  # SEND REQUEST
  $res = $analytics->retrieve($req);
  die("GA error: " . $res->error_message) if !$res->is_success;
  
  #
  # ASSIGN THE NEW VISITORS AND RETURN VISITORS COUNT TO THE VARIABLES
  $newUsers = @{ $res->rows }[0]->get_users;
  $returnUsers = @{ $res->rows }[1]->get_users;
    
  #
  # PRINT THE DATA TO THE LOG FILE    
  open(my $log, '>', $logfile) or die "Could not open file '$logfile' $!";
  print $log "Period_Start,Period_End,Total_users,New_users,Return_users\n";
  print $log "$startDate,$endDate,$totalUsers,$newUsers,$returnUsers\n";
  close $log;
}


#################################################
#
# HERE IS WHERE THE EXECUTION HAPPENS
#

parse_args();
get_data();