Creates a json file ready for _bulk ES API tested with strawberry-perl-5.18.1.1-64bit on window$

csv_tojson_bulk_es.pl

#!/usr/bin/env perl

## Creates a json file ready for _bulk ES API
## tested with strawberry-perl-5.18.1.1-64bit
## 2013-12-22 Author : FRN

use strict; use warnings;

use JSON; ## brings json_encode
use Tie::Array::CSV; ## brings @Array from csv file
use Tie::IxHash; ## brings ordered hashes
use Getopt::Long; ## brings GetOptions facilities

my $separator = " ";
my $show_help = 0;
my $index = "test_index";
my $type = "test_type";
GetOptions(
  "--separator|s=s" => \$separator,
  "--help|h!" => \$show_help,
  "--index=s" => \$index,
  "--type=s" => \$type,
) or show_help();
show_help() if $show_help;

## die if file name is not specified
my ($csvfile) = @ARGV;
unless (defined $csvfile) {
  die "input file name is required";
}

## Create array() from csv file
tie my @file, 'Tie::Array::CSV', $csvfile, sep_char => $separator;

## first line must contain fields name
my $header = \@{$file[0]};

## starting iterate array of rows from second row (offset[1]) to avoid header
## $#file = last @file element
for my $rownum (1 .. $#file) {
    my $row = \@{$file[$rownum]};	## getting the row
        my ($key, $value);
	tie my %r, "Tie::IxHash";	## create ordered hash
        @r{@$header} = @$row;		## magic trick : assigning each value to its column name
        $r{'message'} = "";		## empty message field
        
        ## get into each assigned field to provide little transformation
        while ( ($key, $value) = each %r ) {
	  
	  ## prevent some lines that does not contain all fields
	  ## TODO: better field number control
          if ($value) {
  	    ## special case : remove '[field name=value]' from value
            $r{$key} = $1 if $value =~ /\[$key=(.+)\]/;
            
            ## field message contains every value except himself
            ## TODO: trim leading space
            $r{'message'} = $r{'message'}." ".$value if $key !~ /message/;
          }
        }
        ## timestamp from date end time fields
        ## TODO: try to catch from any field without knowing it
        ## TODO: add eventtreatedTime
        $r{'@timestamp'} = $1."-".$2."-".$3."T".$4 if "$r{'date'} $r{'time'}" =~ /(\d{4})\/(\d{2})\/(\d{2})\s+(\d{2}:\d{2}:\d{2})/;

	## output bulk method
	## TODO: a lot to be more versatile
        print "{ \"index\" : { \"_index\" : \"$index\", \"_type\" : \"$type\"} })\n";

        ## output json line
        print encode_json(\%r)."\n";
	
	## testing pretty output
	#print JSON->new->utf8->pretty->encode(\%r)."\n"; 
}

sub show_help {
	print "Usage: $0 [--separator=?] [--index=<index>] [--type=<type>] [--help] <file>\n";
	exit(1);
}