ads-make-keys

#!/usr/bin/perl 
########################################################################
# Simple perl script to create default format keys for BibTeX entries.
########################################################################
# Create input and output file names -> prefix.ads and prefix.bib
########################################################################
  $_   = $ARGV[0];
  $ext = ".ads";
  if (/$ext$/) {
    $last = rindex($_,$ext);
    $prefix = substr($_,0,$last);
  }
  else {
    $prefix  = $_;
  }
  $bibfile = $prefix . ".ads";
  $outfile = $prefix . ".bib";
  $extrafile = $prefix. ".nonads";
########################################################################
# See if we can read from and write to the respective files.
########################################################################
  open(INFILE_HANDLE,"<".$bibfile) ||
      die "\nCan't open file $bibfile for reading.\n\n";
  open(OUTFILE_HANDLE,">".$outfile) ||
      die "\nCan't open file $outfile for writing.\n\n";
########################################################################
# Look for @ which falls at start of each BibTeX entry.
# ADS puts year of publication at start of their key, e.g.
# @ARTICLE{1979SciN..115..371.,
# Extract the year and first authors last name to create a new key, in
# the format "first author last name":"year", e.g. Strobel:1998
########################################################################
# Initialize variables.
########################################################################
  $TRUE       = 1;
  $FALSE      = 0;
  $in_entry   = $FALSE;
  $in_author  = $FALSE;
  $in_journal = $FALSE;
  $in_title   = $FALSE;
  $publishstr = ' ';
  $editorstr  = ' ';
  $addressstr = ' ';
  $adsnotestr = ' ';
  $schoolstr  = ' ';
  $journalstr = ' ';
  @journalext = (' ',' ',' ',' ',' ');
  $njourext   = 0;
  $authorstr  = ' ';
  @authorext  = (' ',' ',' ',' ',' ',' ',' ');
  $nauthext   = 0;
  $titlestr   = ' ';
  @titleext   = (' ',' ',' ',' ',' ');
  $ntitlext   = 0;
  $volumestr  = ' ';
  $pagesstr   = ' ';
  $monthstr   = ' ';
  $yearstr    = ' ';
  $urlstr     = ' ';
  $title      = ' ';
  $year       = ' ';
  $first      = ' ';
  $keycount   = 0;
########################################################################
# Read from input file until we reach the end.
########################################################################
  while (<INFILE_HANDLE>) {
########################################################################
########################################################################
    if (/^\s*@/) {                    # Start of entry - Extract entry type
      ($type[$keycount],$rest) = split(/{/,$_);
      $in_entry   = $TRUE;
      $in_author  = $FALSE;
      $in_title   = $FALSE;
      $in_journal = $FALSE;
########################################################################
########################################################################
    }
    elsif (/^\s*}/) {               # End of entry - print it out
      $in_entry   = $FALSE;
      $in_author  = $FALSE;
      $in_title   = $FALSE;
      $in_journal = $FALSE;
      $entry[$keycount] = "";
      if ($first eq ' ') {
          # gsn - no colons
          $keyval[$keycount] = "$title:$year:$keycount";
#        $keyval[$keycount] = "$title$year$keycount";
      }
########################################################################
########################################################################
      else {
          # gsn - no colons
                  $keyval[$keycount] = "$first:$year:$keycount";
          #$keyval[$keycount] = "$first$year$keycount";
        $entry[$keycount] = join("",$entry[$keycount],"$authorstr");
        for ($i=0;$i<$nauthext;$i++) {
          $entry[$keycount] = join("",$entry[$keycount],"$authorext[$i]");
        }
      }
      if ($titlestr   ne ' ') {
        $entry[$keycount] = join("",$entry[$keycount],"$titlestr");
      }
      for ($i=0;$i<$ntitlext;$i++) {
        $entry[$keycount] = join("",$entry[$keycount],"$titleext[$i]");
      }
      if ($schoolstr  ne ' ') {
        $entry[$keycount] = join("",$entry[$keycount],"$schoolstr");
      }
      if ($journalstr ne ' ') {
        $entry[$keycount] = join("",$entry[$keycount],"$journalstr");
      }
      for ($i=0;$i<$njourext;$i++) {
        $entry[$keycount] = join("",$entry[$keycount],"$journalext[$i]");
      }
      if ($yearstr    ne ' ') {
        $entry[$keycount] = join("",$entry[$keycount],"$yearstr");
      }
      if ($monthstr   ne ' ') {
        $entry[$keycount] = join("",$entry[$keycount],"$monthstr");
      }
      if ($volumestr  ne ' ') {
        $entry[$keycount] = join("",$entry[$keycount],"$volumestr");
      }
      if ($publishstr  ne ' ') {
        $entry[$keycount] = join("",$entry[$keycount],"$publishstr");
      }
      if ($editorstr  ne ' ') {
        $entry[$keycount] = join("",$entry[$keycount],"$editorstr");
      }
      if ($addressstr  ne ' ') {
        $entry[$keycount] = join("",$entry[$keycount],"$addressstr");
      }
      if ($pagesstr   ne ' ') {
        $entry[$keycount] = join("",$entry[$keycount],"$pagesstr");
      }
      # GSN - kill the freaking url field
      #if ($urlstr     ne ' ') {
      #  $entry[$keycount] = join("",$entry[$keycount],"$urlstr");
      #}
      $keycount++;
      $publishstr = ' ';
      $editorstr  = ' ';
      $addressstr = ' ';
      $adsnotestr = ' ';
      $schoolstr  = ' ';
      $volumestr  = ' ';
      $pagesstr   = ' ';
      $monthstr   = ' ';
      $yearstr    = ' ';
      $urlstr     = ' ';
      $title      = ' ';
      $year       = ' ';
      $first      = ' ';
      $journalstr = ' ';
      @journalext = (' ',' ',' ',' ',' ');
      $njourext   = 0;
      $authorstr  = ' ';
      @authorext  = (' ',' ',' ',' ',' ',' ',' ');
      $nauthext   = 0;
      $titlestr   = ' ';
      @titleext   = (' ',' ',' ',' ',' ');
      $ntitlext   = 0;
    }
########################################################################
# Match keywords to the appropriate strings.
########################################################################
    if (/^\s*booktitle/) {
      $journalstr=$_;$journalstr =~ s/ +/ $2/g;
      $in_title=$FALSE;$in_author=$FALSE;$in_journal = $TRUE;
    }
    if (/^\s*journal/  ) {
      $journalstr=$_;$journalstr =~ s/ +/ $2/g;
      $in_title=$FALSE;$in_author=$FALSE;$in_journal = $TRUE;
    }
    if (/^\s*publisher/  ) {
      $publishstr=$_;$publishstr =~ s/ +/ $2/g;
      $in_title=$FALSE;$in_author=$FALSE;$in_journal = $FALSE;
    }
    if (/^\s*editor/  ) {
      $editorstr=$_;$editorstr =~ s/ +/ $2/g;
      $in_title=$FALSE;$in_author=$FALSE;$in_journal = $FALSE;
    }
    if (/^\s*address/  ) {
      $addressstr=$_;$addressstr =~ s/ +/ $2/g;
      $in_title=$FALSE;$in_author=$FALSE;$in_journal = $FALSE;
    }
    if (/^\s*adsnote/  ) {
      $adsnotestr=$_;$adsnotestr =~ s/ +/ $2/g;
      $in_title=$FALSE;$in_author=$FALSE;$in_journal = $FALSE;
    }
    if (/^\s*volume/   ) {
      $volumestr =$_;$volumestr =~ s/ +/ $2/g;
      $in_title=$FALSE;$in_author=$FALSE;$in_journal = $FALSE;
    }
    if (/^\s*school/   ) {
      $schoolstr =$_;$schoolstr =~ s/ +/ $2/g;
      $in_title=$FALSE;$in_author=$FALSE;$in_journal = $FALSE;
    }
    if (/^\s*pages/    ) {
      $pagesstr  =$_;$pagesstr =~ s/ +/ $2/g;
      $in_title=$FALSE;$in_author=$FALSE;$in_journal = $FALSE;
    }
    if (/^\s*month/    ) {
      $monthstr  =$_;$monthstr =~ s/ +/ $2/g;
      $in_title=$FALSE;$in_author=$FALSE;$in_journal = $FALSE;
    }
    if (/^\s*url/      ) {
      $urlstr    =$_;$urlstr =~ s/ +/ $2/g;
      $in_title=$FALSE;$in_author=$FALSE;$in_journal = $FALSE;
    }
########################################################################
# Process keywords which are used to construct the BibTeX keys.
# Extract year of publication
########################################################################
    if (/^\s*year/) {
      $in_title   = $FALSE;
      $in_author  = $FALSE;
      $in_journal = $FALSE;
      $yearstr    = $_;
      $yearstr    =~ s/( +)/ $2/g;
      ($junk,$year,$rest)=split(/=/,$yearstr); # get stuff to right of =
      $year =~ s/,//g;                         # strip commas
      for ($year) {s/^\s+//; s/\s+$//;}        # strip white space at ends
      # Gsn- get rid of 19-- and 20--
      $year =~ s/^19//;
      $year =~ s/^20//;
    }
########################################################################
# Extract first author last name
########################################################################
    elsif (/^\s*author/) {
      $authorstr  = $_;
      $authorstr  =~ s/( +)/ $2/g;
      $in_author  = $TRUE;
      $in_title   = $FALSE;
      $in_journal = $FALSE;
      ($junk,$rest)=split(/=/,$authorstr); # get stuff to right of =
      ($first,$rest)=split(/\}/,$rest);     # get stuff to left of }
      $first =~ s/\{//g;                    # strip outer brackets
      $first =~ s/\"//g;                    # strip outer quotes
      for ($first) {    # strip all white space
          s/\s+//g;
      }
      $first =~ tr/A-Z/a-z/; # GSN - everyone's lower case
    }
########################################################################
# Extract title for use in key when no author info is given.
########################################################################
    elsif (/^\s*title/) {
      $titlestr   = $_;
      $titlestr   =~ s/ +/ $2/g;
      $in_title   = $TRUE;
      $in_author  = $FALSE;
      $in_journal = $FALSE;
      ($junk,$rest)=split(/=/,$titlestr); # get stuff to right of =
      ($title,$rest)=split(/}/,$rest);    # get stuff to left of }
      $title =~ s/{//g;                   # strip outer brackets
      for ($title) {s/\s+//g;}            # strip all white space
      $title =~ s/,//g;                   # strip commas
      $title =~ s/.//g;                   # strip periods
    }
    elsif ($in_entry && $in_author && !/\s+=/ && !/^$/){#Additional author entry
      $authorext[$nauthext++] = $_;
    }
    elsif ($in_entry && $in_title && !/\s+=/ && !/^$/) {#Additional title entry
      $titleext[$ntitlext++] = $_;
    }
    elsif ($in_entry && $in_journal && !/\s+=/ && !/^$/) {#Additional jnl entry
      $journalext[$njourext++] = $_;
    }
  }
########################################################################
# Sort keys, adding a,b,c, ... as necessary to ensure uniqueness
########################################################################
   @keyval  = sort @keyval;
   @keyauth = @keyval;
   @keyyear = @keyval;
   @keypos  = @keyval;
   @keyextn = @keyval;
   for ($i=0;$i<$keycount;$i++) {
     ($keyauth[$i],$keyyear[$i],$keypos[$i]) = split(/:/,$keyval[$i]);
     $keyextn[$i] = "";
   }
   @extn = ('a','b','c','d','e','f','g','h','i');
   $eind = 0;
   for ($i=0;$i<$keycount-1;$i++) {
     if (($keyauth[$i] eq $keyauth[$i+1]) &&
         ($keyyear[$i] eq $keyyear[$i+1])) {
       $keyextn[$i]   = $extn[$eind++];
       $keyextn[$i+1] = $extn[$eind];
       print STDOUT "Identical Keys - Modifying ...\n";
       print STDOUT $keyauth[$i],":",$keyyear[$i],$keyextn[$i],"\n";
       print STDOUT $keyauth[$i+1],":",$keyyear[$i+1],$keyextn[$i+1],"\n";
     }
     else {$eind = 0}
   }
########################################################################
# Print out sorted keys and associated BIBTEX entries.
########################################################################
  for ($i=0;$i<$keycount;$i++) {
    print OUTFILE_HANDLE "$type[$keypos[$i]]\{$keyauth[$i]:";
    print OUTFILE_HANDLE "$keyyear[$i]$keyextn[$i],\n";
    print OUTFILE_HANDLE "$entry[$keypos[$i]]";
    print OUTFILE_HANDLE "\}\n\n";
  }
  $isextrafile=1;
  open(EXTRAFILE_HANDLE,"<".$extrafile) ||
    ($isextrafile = 0);
print $isextrafile."\n";
  if ($isextrafile==1) {
    while (<EXTRAFILE_HANDLE>) {
        print OUTFILE_HANDLE;
    }
  }
########################################################################
# Clean Up and Exit
########################################################################
  close(INFILE_HANDLE);
  close(OUTFILE_HANDLE);
  close(EXTRAFILE_HANDLE);
  print STDOUT "\n  Default BibTeX keys created for file $outfile\n\n";
########################################################################
########################################################################