package ILMT::TEL::HIN::WX2UTF;

use Dir::Self;
use Data::Dumper;
my $cwd = __DIR__;
use IPC::Run qw(run);
use File::Temp qw/ tempfile /;
use File::Slurp qw( slurp );
use ILMT::TEL::HIN::IndicCC;
use ILMT::TEL::HIN::SSFAPI::feature_filter;
use ILMT::TEL::HIN::SSFAPI::shakti_tree_api;



sub process {
    my %args = @_;
    utf8::encode($args{data});
    
    $args{'lang'} = $args{'src_lang'};
    $args{'src'} = 'wx';
    $args{'tgt'} = 'utf';
    $args{'type'} = 'ssf';
    $args{'data'} = $args{data};
    my $wxtoutf1 = convert_notation(%args);
    utf8::decode($wxtoutf1);
    return $wxtoutf1;
};


sub convert_notation {
    my %par = @_;
    my $data = $par{'data'};
    my $type = $par{'type'};
    my $lang = $par{'lang'};
    my $src  = $par{'src'};
    my $tgt  = $par{'tgt'};
    my $result = "";
    if( lc($src) eq "wx" and lc($tgt) eq "utf")
    {
        #open(STDIN,"<:utf8");
        open INFILE, '<', \$data or die $!;
        if( lc($type) eq "ssf" )
        {
		#while($line=<STDIN>)
		while($line=<INFILE>)
        	{
                	chomp ($line);
                	($num, $tkn, $pos, $fs) = split(/\t/,$line);
			if ($tkn !~ /^\^?\@.*/)
			{
			    $lex_out = &wx2utf($tkn, $lang);
			}
                        else
                        {
                            $lex_out = $tkn;
                        }
        	        if($fs ne "")
                	{
	                        @fss = split(/\|/, $fs);
        	                my $len = @fss;
                	        @string  = "";
                        	$newfs = "";
	                        my $i=0;
        	                foreach $af (@fss)
                	        {
	                                my $FSreference = &read_FS($af, $line);
        	                        my @lex_root = &get_values("lex", $FSreference);
        	                        my @cat_root = &get_values("cat", $FSreference);
                	                my @fs_vib = &get_values("vib", $FSreference);
                        	        my @fs_head = &get_values("head", $FSreference);
                                	my @fs_name = &get_values("name", $FSreference);
	                                foreach $field (@lex_root)
        	                        {
						# if lcat is punc then don't convert it
						if ($cat_root[0] ne "punc" and $lex_root[0] !~ /^\^?eng\~.*/)
                                                {
							$val_out = &wx2utf($field, $lang);
						} else {
							$field =~ s/eng~//;
							$val_out = $field;
						}
                                        	my @lex_arr=();
	                                        push @lex_arr,$val_out;
        	                                &update_attr_val("lex", \@lex_arr,$FSreference,$af);
                	                        $string[$i]=&make_string($FSreference,$af);
                                                
                        	        }
                                	foreach $field1 (@fs_vib)
	                                {
						if ($field1 !~ /^\^?eng\~.*/)
                                                {
							$vib_out = &wx2utf($field1, $lang);
						} else {
							$field1 =~ s/eng~//;
							$vib_out = $field1;
						}
                                        	my @fs_vib_arr=();
	                                        push @fs_vib_arr,$vib_out;
        	                                &update_attr_val("vib", \@fs_vib_arr,$FSreference,$af);
                	                        $string[$i]=&make_string($FSreference,$af);
                        	        }
                                	
	                                $i++;
        	                }
				foreach $string (@string)
				{	
					if(--$len)
					{	
        	                       		$newfs=$newfs.$string."|";
					}
					else
					{
						$newfs=$newfs.$string;
					}
				}
				$newfs =~s /af=',,,,,,,'//g; 
				delete @string[0..$#string];
				delete @lex_root[0..$#lex_root];
				delete @fss[0..$#fss];
				if($line =~ /\(\(/ or $line =~ /\)\)/)
				{
					($num,$lex,$pos,$fs) = split(/\t/,$line);
					binmode STDOUT, ":utf8";
					#binmode OUTFILE, ":utf8";
                                        
					$result .= $num."\t".$lex."\t".$pos."\t".$newfs."\n";
				}
				else
				{
					binmode STDOUT, ":utf8";
					$result .= $num."\t".$lex_out."\t".$pos."\t".$newfs."\n";
				}
			} # end if fs ne "" 
			else {  # try to understand this else block
				
				if($lex ne "((" and $lex ne "))")
				{
					binmode STDOUT, ":utf8";
                                        $result .= $num."\t".$lex_out."\t".$pos."\t".$fs."\n";
				}
				else {
                                        $result .= $line."\n";
				}
			}


        	} # end while loop
	} # if format SSF end
        #elsif(($type eq "TEXT") or ($type eq "text"))
        elsif( lc($type) eq "text" )
        {
            while(my $line=<INFILE>)
            {
                #chomp ($line);
                my $text_out = wx2utf($line, $lang);
                binmode STDOUT, ":utf8";
                $result .= $text_out;
                #print "under construction\n";

            }
        } # type text end
        else
        {
            #print "type mismatch\n";
            print "format type $type is NOT supported.\n";
            print "Try $0 --help for more information\n";
            exit(0);
        }

    } # end if statement src=wx tgt=utf
    elsif( lc($src) eq "utf" and lc($tgt) eq "wx")
    {
        #open(STDIN,"<:utf8");
        #open INFILE, '<', \$data or die $!;
	my @temparray = split('\n', $data);
        if( lc($type) eq "ssf" )
        {
            #print "ssf conversion start\n";
            #while(my $line=<INFILE>)
	    while(my $line=shift(@temparray))
            {
                chomp ($line);
                	($num, $tkn, $pos, $fs) = split(/\t/,$line);
			#$lex_out = &utf2wx($tkn, $lang);
			if ($tkn !~ /[a-zA-Z]/ ){
				$lex_out = &utf2wx($tkn, $lang);
			} else {
				$lex_out = "eng~".$tkn;
			}

        	        if($fs ne "")
                	{
	                        @fss = split(/\|/, $fs);
        	                my $len = @fss;
                	        @string  = "";
                        	$newfs = "";
	                        my $i=0;
        	                foreach $af (@fss)
                	        {
					if ($af !~ /^\^<fs  *af=.*$/)
                                        {
                                                $af =~s /<fs /<fs af=',,,,,,,' /g;
					}
	                                my $FSreference = &read_FS($af, $line);
        	                        my @lex_root = &get_values("lex", $FSreference);
        	                        my @cat_root = &get_values("cat", $FSreference);
                	                my @fs_vib = &get_values("vib", $FSreference);
                        	        my @fs_head = &get_values("head", $FSreference);
                                	my @fs_name = &get_values("name", $FSreference);
	                                foreach $field (@lex_root)
        	                        {
						# if lcat is punc then don't convert it
                                                if ($cat_root[0] ne "punc" and $lex_root[0] !~ /[a-zA-Z]/)
                                                {
							$val_out = &utf2wx($field, $lang);
						} else {
							$val_out = "eng~".$field;
						}
                                        	my @lex_arr=();
	                                        push @lex_arr,$val_out;
        	                                &update_attr_val("lex", \@lex_arr,$FSreference,$af);
                	                        $string[$i]=&make_string($FSreference,$af);
                        	        }
                                	foreach $field1 (@fs_vib)
	                                {
						if ($field1 !~ /[a-zA-Z]/) {
							$vib_out = &utf2wx($field1, $lang);
						} else {
							$vib_out = "eng~".$field1;
						}
                                        	my @fs_vib_arr=();
	                                        push @fs_vib_arr,$vib_out;
        	                                &update_attr_val("vib", \@fs_vib_arr,$FSreference,$af);
                	                        $string[$i]=&make_string($FSreference,$af);
                        	        }
                                	
	                                $i++;
        	                }
				foreach $string (@string)
				{	
					if(--$len)
					{	
        	                       		$newfs=$newfs.$string."|";
					}
					else
					{
						$newfs=$newfs.$string;
					}
				}
				$newfs =~s /af=',,,,,,,'/ /g; 
				delete @string[0..$#string];
				delete @lex_root[0..$#lex_root];
				delete @fss[0..$#fss];
				if($line =~ /\(\(/ or $line =~ /\)\)/)
				{
					($num,$lex,$pos,$fs) = split(/\t/,$line);
					#print $num,"\t",$lex,"\t",$pos,"\t",$newfs,"\n";
					binmode OUTFILE, ":utf8";
                                        $result .= $num."\t".$lex."\t".$pos."\t".$newfs."\n";
				}
				else
				{
					binmode STDOUT, ":utf8";
                                        $result .= $num."\t".$lex_out."\t".$pos."\t".$newfs."\n";
					#print $num,"\t",$lex_out,"\t",$pos,"\t",$newfs,"\n";
				}
			} # end if fs ne "" 
			else {  # try to understand this else block
				if($lex ne "((" and $lex ne "))")
				{
					binmode STDOUT, ":utf8";
                                        $result .= $num."\t".$lex_out."\t".$pos."\t".$fs."\n";
					#print $num,"\t",$lex_out,"\t",$pos,"\t",$fs,"\n";
				}
				else{
                                        $result .= $line."\n";
				}
			}



            } # end while loop

        } # type SSF end
        elsif( lc($type) eq "text")
        {
            #print "text conversion block called\n";
            while(my $line=<INFILE>)
            {
                #chomp ($line);
                #print "before convert call $line\n";
                my $text_out = utf2wx($line, $lang);
                binmode STDOUT, ":utf8";
                $result .= $text_out;
                #print "under construction\n";
            }
        } # end format type text
        else
        {
            print "format type $type is NOT supported.\n";
            print "Try $0 --help for more information\n";
            exit(0);
        }

    } # end elsif statement src=utf tgt=wx
    else
    {
        print "$0: source and target encoding mismatch\n";
        print "Try $0 --help for more information\n";
        exit(0);
    }
    return $result;
};

1;