#!/usr/bin/perl

### About This Script #########################################################
#
#	Title:	CSV Parser
#	Author:	Nathan Shaw
#______________________________________________________________________________
#
### Disclaimer ################################################################
#
# This script is not supported by its author.  It is provided AS IS without 
# warranty of any kind. The entire risk arising out of the use of this script 
# remains with you.  In no event shall the author of this script be liable
# for any damages arising out of the use of or inability to use the script.
#
# As with anything, you should test this script in a separate development
# environment before using it anywhere close to your production network.
#______________________________________________________________________________
#
### Description ###############################################################
#
# Take a file that *sometimes* uses a double-quote as a field boundary and 
# *sometimes does not* and output data with a different delimiter.
#  
# You can also use it to print only select fields from a file and even to 
# rearrange the order of the fields.
#
# Please see Version History for more information.
#______________________________________________________________________________
#
### Input/Output ##############################################################
#
# The script requires a file for input.
#______________________________________________________________________________
#
### Known Issues ##############################################################
#
# None yet...
#______________________________________________________________________________
#
### Version History ###########################################################
#
# v1.0.0 - 11/17/2009 - NShaw
#	Initial release.
# v1.0.1 - 12/14/2010 - NShaw
#	Added support for different input delimiters... So I guess this isn't 
#		just a CSV parser now.  :P  The -d | --delimiter option has been
#		changed to -o | --outputdelim to make it more clear.
#	Added a debug option along with a few debug commands.
#	Changed to use Text::CSV_XS instead of Text::CSV (only because it's part
#		of ActivePerl's default install and works the same as Text::CSV for 
#		what I'm doing).
# v1.0.2 - 10/06/2011
#	Will now print a message to STDERR if it can't parse a line in the file 
#		instead of exiting.
#______________________________________________________________________________
#
# Future ideas
#
# Add an --output option.  Make the debug messages go to the screen, but the
# delimited text file's output would go to the file specified.
#

use strict;
use Text::CSV_XS;
use Getopt::Long;

our $ERROR_INVALID_USAGE			=	1;
our $ERROR_MISSING_CSV			=	5;
our $ERROR_BAD_CSV				=	17;

our $opt_inputfile;
our $opt_output_delim;
our $opt_input_delim;
our $opt_fields;
our $opt_quotes;
our $opt_help;
our $opt_debug;

our $VERSION = "1.0.2";

GetOptions("c|csv=s" => \$opt_inputfile,
	"o|outputdelim=s" => \$opt_output_delim,
	"i|inputdelim=s" => \$opt_input_delim,
	"f|fields=s" => \$opt_fields,
	"q|quotes" => \$opt_quotes,
	"d|debug" => \$opt_debug,
	"h|help" => \$opt_help);

usageHelp() if $opt_help;

usageHelp($ERROR_INVALID_USAGE,"You must specify a file to parse!") if !$opt_inputfile;

$opt_output_delim = "," if !$opt_output_delim;
$opt_input_delim = "," if !$opt_input_delim;

print "Output Delimiter: $opt_output_delim\n" if $opt_debug;
print "Input Delimiter: $opt_input_delim\n" if $opt_debug;

#$opt_fields = 0 if !$opt_fields; # Makes sure that the first field always prints
my @printfields = split(",",$opt_fields);

if ($opt_debug) {
	print "\@printfields\n";
	for (@printfields){
		print "$_\n";
	}
	print "\n";
}

# Create CSV object
my $objCSV = Text::CSV_XS->new ({ sep_char => $opt_input_delim });

open INPUTFILE, $opt_inputfile or usageHelp($ERROR_MISSING_CSV,"Couldn't open input file: $opt_inputfile, for reading!",1);

my @csvline = (<INPUTFILE>);

my $linenum = 1;

foreach my $line (@csvline) {
	if (!$objCSV->parse($line)){
		# usageHelp($ERROR_BAD_CSV,"Encountered invalid delimited data on line $linenum: \'$line\'",1);
		print STDERR "ERROR parsing line $linenum: $line";
	} else {
		# Store the value of each of the fields from the file into the @csvfields array
		my @csvfields;
		@csvfields = $objCSV->fields();
		
		if ($opt_debug) {
			print "\@csvfields\n";
			for (@csvfields){
				print "$_\n";
			}
			print "\n";
		}
		
		my $data; # data to be printed
		my $index;
		my $i = 0;
		
		#print "\$#csvfields = $#csvfields\n";
		
		# If print fields were defined, only print them
		if (@printfields) {
			for (@printfields) {
				$index = $_;

				# Error if the field index is out of range
				if ($index >= 0 && $index <= $#csvfields) {
					$data = $data . "$csvfields[$index]" if !$opt_quotes;
					$data = $data . "\"$csvfields[$index]\"" if $opt_quotes;
					$data = $data . $opt_output_delim;
				} else {
					usageHelp($ERROR_INVALID_USAGE,"Field $index is out of range for line $linenum!",1);
				}
				$i++;
			}
		# No print fields were defined, so lets print them all
		} else {
			for (@csvfields) {
				$data = $data . "$_" if !$opt_quotes;
				$data = $data . "\"$_\"" if $opt_quotes;
				$data = $data . $opt_output_delim;
			}
		}
		
		# Strip any trailing delimeters - Probably a better way to do this but oh well..
		my $delen = length($opt_output_delim);
		$data =~ s/[$opt_output_delim]{$delen}$//;
		print "$data\n";
	}
	$linenum++;
}

sub usageHelp() {
	# 1st argument is the error code (required)
	# 2nd argument is error message (optional)
	# 3rd argument is bolean to hide the usage info (optional)
	my $code=$_[0];
	print $_[1]."\n" if $_[1];
	if (!$_[2]) {
		print "\n"; 
		print "CSV Parser v$VERSION - Parse data from a delimited text file.\n";
		print "\nUsage: $0 -c [file]\n\n";
		print "Required:\n";
		print "   -c | --csv [File]\t\tSpecify a file with the input data\n";
		print "\n";
		print "Options:\n";
		print "   -i | --inputdelim [char]\tInput file delimiter (Default: \",\")\n";
		print "   -o | --outputdelim [char]\tOutput using different delimiter (Default: \",\")\n";
		print "   -f | --fields [#,#,#,#]\tFields to print (base 0) - Order will be maintained. (Default: All)\n";
		print "   -q | --quotes\t\tInclude quotes around the fields (Default: Off)\n";
		print "   -h | --help\t\t\tHelp!\n";
		print "   -d | --deubg\t\t\tEnable debugging info\n";
		print "\n";
	} 
	print "\n";
	exit($code);
}