The outputs were different in the terminal windows and txt file?

I am trying to merge two files by Perl.

Codes so far:

 my $hash_ref;  
 open (my $I_fh, "<", "File1.txt") or die $!;

 my $line = <$I_fh>;
 while ($line = <$I_fh>) {
 chomp $line;
 my @cols = split ("\t", $line);
 my $key = $cols[1];
 $hash_ref -> {$key} = \@cols;
 }
 close $I_fh;

 open (my $O_fh, "<", "File2.txt") or die $!;
 while ($line = <$O_fh>) {
 chomp $line;
 my @cols = split ("\t", $line);
 my $key = shift (@cols);
 push (@{$hash_ref -> {$key}}, @cols);

 }
 close $O_fh;


 open (my $out, ">", "merged.txt") or die $!;

 foreach my $key (sort keys %$hash_ref) {

 my $row = join ("\t", @{$hash_ref -> {$key}});

print $out "$key\t$row\n";
 }
close $out;

I am using print or Dumper function to check every steps. In the terminal windows, everything is fine. However, in my output file (merged txt), the format was changed. I would like to merge two files by adding more columns, not adding more rows. How can I fix codes?

  File 1.txt:  
  Index    Name    Column1   Column2  
   1        A1                  AB      
   2        A2                  CD   
   3        B1                  EF    
   4        B2                  GH   


    File 2.txt:   
    Name  Type  
     A1     1  
     A2     1   
     B1     2   
     B2     1    

   Merged file:  

   A1   1   AB    
        1     
   A2   2   CD  
        1      
   B1   3   EF  
        2      
   B2   4   GH   
        1      

Wanted file:  
Name  Type  Column2  

  A1   1   AB    
  A2   1   CD   
  B1   2   EF   
  B2   1   GH

Solution

Assuming the files are sorted based on the name column, this is really easy to do thanks to the join(1) program:

$ join --header -t $'\t' -o 2.1,2.2,1.4 -1 2 -2 1 file1.tsv file2.tsv
Name    Type    Column2
A1  1   AB
A2  1   CD
B1  2   EF
B2  1   GH

The --header option is a GNU extension that excludes the first lines of the two files from being joined and treats them as column titles instead. -t sets the column separator, -o controls what columns are included in the output (A list of FILE.COLUMN specifiers), and -1 and -2 choose the columns that are used to join the two files.

If they're not sorted, or if you're set on perl, your code looks very very close; besides all the typos and such, you're printing out every column, not just the ones your desired output suggest you care about. Consider:

#!/usr/bin/perl
use warnings;
use strict;
use feature qw/say/;
use autodie;

my %names;

sub read_file {
  my ($file, $idx) = @_;
  open my $in, "<", $file;
  my $header = <$in>;
  while (<$in>) {
    chomp;
    my @F = split /\t/;
    push @{$names{$F[$idx]}}, \@F;
  }
}

read_file "file1.tsv", 1;
read_file "file2.tsv", 0;

say "Name\tType\tColumn2";
for my $n (sort keys %names) {
  my $row = $names{$n};
  say "$n\t$row->[1][1]\t$row->[0][3]";
}

I also suspect your strange output might be explained by running your program on data files that use Windows-style line endings when your OS uses Unix-style line endings.