#!/usr/bin/perl

# med2ace.pl
# Converts a medline file to a .ace file for the MalDB database.
# The medline fields used are signified by: -
# UI  - A medline record number. Should be first field entry
#     	because this program uses it to seperate records and
#	initiate translating last record from medline format 
#	to .ace format.
# TI  - Title of the paper.
# AU  - Author
# SO  - Journal, year, volume, and page.
# AB  - Abstract of paper
# All other fields are ignored and, apart from UI, the fields 
# may appear in any order.


($#ARGV>=0) || die
   "usage: med2ace.pl [-D] <filename> > <outputfilename>
    if -D is presented, the ace file created will have \"-D <Class>: id\" line
    proceeded to each object \"<Class>: id\", as the \"-D\" flag of ACeDB;
    use - for standard input;
    the default output is on the screen;
    options must be given in the order shown above\n";

# If -D option used, set $VAR_D and increment $ARGV pointer.
if ($ARGV[0] =~ /^-D$/)
{       $VAR_D = 1;
        shift
}
else {  $VAR_D = 0 };




#Open medline file.
open (INPUT, "<$ARGV[0]");

#Initialise fields to "".
&init;

#Ignore first UI.
$start = 1;

while (<INPUT>){
        chop;  
	# If line is not the beginning of a new field,
	# append line to current field, 
	# else check if previous field is relevant to .ace output file.
        if ($_ !~ /^[A-Z]/){
                $_ =~ s/^     //;
                $section = $section . "\n" . $_
        } else {
	      # If previous field is AU,  store author in $auth array.
              if ($section =~ /^AU/){
		$author = $section;
		&author;
	      }

	      # Store Title field.
	      if ($section =~ /^TI/){
		$title = $section;
	      }

	      # Store Journal field.
	      if ($section =~ /^SO/){
		$journalstats = $section;

	      # Store Abstract field.
	      } 
	      if ($section =~ /^AB/){
		$abstract = $section;
	      }
	      # If line is beginning of a new record (but not 1st record),
	      # extract .ace file data from stored medline data
	      # and print to stdout in .ace format.
	      if (($_ =~ /^UI  - /) && ($start == 0)) {
			&title;
			&journalstats;
			&abstract;
			&printout;
			&init; 
	      } else {
		 if ($start == 1){
                        $start = 0;
                 } else {
			$section = $_;
		 }
	      }
        }
}

# Repeat process for final record.
if ($section =~ /^AU/){
		$author = $section;
		&author;
}
if ($section =~ /^TI/){
		$title = $section;
}
if ($section =~ /^SO/){
		$journalstats = $section;
}
if ($section =~ /^AB/){
		$abstract = $section;
}
&title;
&journalstats;
&abstract;
&printout;

#Initalise variables for storing relevent medline data.
sub init {
	$title = "";
	$author = "";
	$journalstats = "";
	$abstract = "";
	$section = "";
	$n = 0;
	$Pap = "";
	$tit = "";
	$Jour = "";
	$Year = "";
	$Vol = "";
	$Pag= "";
	$Abs = "";
	$LongText = "";
}

# Store author in $auth array.
sub author {
	$author =~ s/\n//g;
        $auth[$n] = $author; 
        $auth[$n] =~ s/^AU  - /Author \"/;
        $auth[$n] .= "\"\n";
        $n++;
        $section = "";
}

# Store title in $tit. 
sub title {
	$title =~ s/\n//g;
        $tit = $title;   
        $tit =~ s/\"/\'/g;
        $tit =~ s/^TI  - /Title \"/;
        $tit .= "\"\n";
        $section = "";
}

# Store journal, year, volume and page no. in $Jour, $Year, $Vol, and $Pag
# and define the paper as the concatonation of these variables, for the
# .ace Paper entry.
sub journalstats {
	$journalstats =~ s/\n//g;
        $journalstats =~ s/ 19/:19/;
        $journalstats =~ s/;/:/;
        ($Jour, $Year, $Vol, $Pag) = split(":", $journalstats);
        $Pap = "Paper : \"" . $Jour . "_" . $Year . "_" . $Vol .  "_" . $Pag . "\"\n";
        $Pap =~ s/^Paper : "SO  - /Paper : "/;
        $Jour =~ s/"/'/g;
        $Jour =~ s/^SO  - /Journal "/;
        $Jour = $Jour . "\"\n";
        $Year =~ s/^19\(\d\d\)\s.*/19\1\2/g;
        $Year =~ s/\D//g;
        $Year =~ s/^19\d\d/$&/g;
        $Year = "Year " . $Year . "\n";
        $Vol =~ s/\s/,/g;
        $Vol = "Volume " . $Vol . "\n";
        $Pag =~ s/\s/,/g;
        $Pag = "Page " . $Pag . "\n";
        $section = "";
}

# Store abstract in $LongText and use $Pap as the header for
# the abstract.
sub abstract {
	$Abs = $Pap;
        $Abs =~ s/^Paper : /Abstract /;
        $LongTit = $Abs;
        $LongTit =~ s/^Abstract //;
        $LongText = $abstract;
        $LongText =~ s/^AB  - //;
        $LongText = "LongText " . $LongTit . $LongText . "\n***LongTextEnd***\n";
        $LongText =~ s/\n /\n/g;
        $Abs = $Abs . "\n";
        $section = "";
}

# Print record to stdout in .ace format.
sub printout {
	print "\n-D $Pap\n\n" if ($VAR_D);
	print $Pap;
        for ($i = 0; $i < $n; $i++){
               print $auth[$i];
        }
#       @auth = ();
        print $tit;
        print $Jour;
        print $Year;
        print $Vol;
        print $Pag;
        print $Abs;
        print $LongText;
        print "\n\n\n";
}