#!/usr/bin/perl
#########################################################
# usage:
# merge.pl file1 file2
# merges two spreadsheets
#
# Example:
#
#  file1:
#
#  A 11 22 33
#  B 44 55
#  C 77 88 
#  D 99  
#
#  file2:
#  
#  A aa bb
#  C cc dd ee
#  B ff gg hh ii
#  E jj kk
#
#  merge file1 file2 produces the new matrix:
#
#  A 11 22 33 aa bb NA NA 
#  B 44 55 NA ff gg hh ii 
#  C 77 88 NA cc dd ee NA 
#  D 99 NA NA NA NA NA NA 
#
#  merge file2 file1 produces the new matrix:
#
#  A aa bb NA NA 11 22 33 
#  C cc dd ee NA 77 88 NA 
#  B ff gg hh ii 44 55 NA 
#  E jj kk NA NA NA NA NA 
#
# While this could be done more resourceful,
# the approach used here aims at clarity:
# one just builds two matrices, where entries can 
# be accessed like   &m1(3,2)  &m2(4,1)
# then prints out a newly merged matrix.  
#
# Oliver Knill, March, 2004
#########################################################

#########################################################
#   Configuration
#########################################################
#
# default delimiter and string for empty entries
$delimiter1 ="[ ]+";
$delimiter2 ="[ ]+";
$emptystr   = "NA"; 

#########################################################
#   Input part 
#########################################################

# read files
$file1      = $ARGV[0]; 
$file2      = $ARGV[1]; 
open (FILE1,"<$file1") || die "Can't open $file1 $!";
open (FILE2,"<$file2") || die "Can't open $file2 $!";
@matrix1 = <FILE1>;
@matrix2 = <FILE2>;

# find number of rows in each matrix
$num_of_rows1=$#matrix1+1;
$num_of_rows2=$#matrix2+1;

# count the maximal number of cols in first matrix
$num_of_cols1=0; 
for ($i=0; $i<$num_of_rows1; $i++) {
   my $line=@matrix1[$i];
   my @entries=split($delimiter1,$line);
   my $s=$#entries+1; 
   if ($s>$num_of_cols1) { $num_of_cols1=$s; }
}

# count the maximal number of cols in second matrix
$num_of_cols2=0; 
for ($i=0; $i<$num_of_rows2; $i++) {
   my $line=$matrix2[$i];
   my @entries=split($delimiter2,$line);
   my $s=$#entries+1;
   if ($s>$num_of_cols2) { $num_of_cols2=$s; }
}

#########################################################
##   Output part  
##########################################################

# print the new matrix
for ($i=1; $i<=$num_of_rows1; $i++) { 
  $found_entry="no"; 
  for ($j=1; $j<=$num_of_cols1; $j++) {print &m1($i,$j)," ";}
  for ($k=1; $k<=$num_of_rows2; $k++) {
    if (&m1($i,1) eq &m2($k,1)) {
      $found_entry="yes"; 
      for ($l=2; $l<=$num_of_cols2; $l++) {print &m2($k,$l)," ";}
    } 
  }
  if ($found_entry eq "no") {
    for ($l=2; $l<=$num_of_cols2; $l++) {print $emptystr," ";}
  }
  print "\n";
}

###########################################################
###   Procedures   
###########################################################

#  access matrix elements in first matrix
sub m1 {
   my @indices=@_;
   my $i=@indices[0]-1;
   my $j=@indices[1]-1;
   my $line=@matrix1[$i];
   my @entries=split($delimiter1,$line);
   my $entry=@entries[$j]; $entry =~ s/\n//; 
   if ($entry eq "") { $entry=$emptystr; }
   return $entry;
}

#  access matrix elements in second matrix
sub m2 {
   my @indices=@_;
   my $i=@indices[0]-1;
   my $j=@indices[1]-1;
   my $line=@matrix2[$i];
   my @entries=split($delimiter2,$line);
   my $entry=@entries[$j]; $entry =~ s/\n//; 
   if ($entry eq "") { $entry=$emptystr; }
   return $entry;
}