#!/usr/local/bin/perl

use Getopt::Std;
getopts ('hv', \%opts);

$me    = `basename $0`; chop $me;
$usage = "$me [-hv] [file ... fileN]
This program lists all the element and attribute names used in the XML input file(s).
Options:
  -h displays this help screen
  -v displays attribute values
Known bugs:
  attribute values enclosed in single quotes are not handled properly
";

die "$usage\n" if ($opts{h});
die "$usage\n" if (! @ARGV);

$/ = "<";

while (<>) {
  s/\/>/ \/>/;
  if (/^[^\/]*\/?>/) {
    ($element, $content) = split(/>/,$_,2);
    ($element, $attributes) = split(/\s+/,$element,2);

    $attributes =~ s/\s*\/$//;
    $attributes =~ s/([^=]+)=([\"\'][^\"\']*[\"\'])/PRATTSEP$1=$2/g;
    $attributes =~ s/"\s+/"/g;
    $attributes =~ s/\s+"/"/g;
    $attributes =~ s/'\s+/'/g;
    $attributes =~ s/\s+'/'/g;
    $attributes =~ s/^PRATTSEP//;
    $attributes =~ s/PRATTSEP /PRATTSEP/g;

    unless ($opts{v}) {
      # The following attributes are not displayed unless the -v option is supplied
      $attributes =~ s/orig="[^"]*"//;
      $attributes =~ s/reg="[^"]*"//;
      $attributes =~ s/PRATTSEPid="[^"]*"//;
      $attributes =~ s/^id="[^"]*"//;
      $attributes =~ s/entity="[^"]*"//;
      $attributes =~ s/sic="[^"]*"//;
      $attributes =~ s/corr="[^"]*"//;
      $attributes =~ s/entityref="[^"]*"//;
      $attributes =~ s/target="[^"]*"//;
      $attributes =~ s/PRATTSEPn="[^"]*"//;
      $attributes =~ s/^n="[^"]*"//;
      $attributes =~ s/PRATTSEPpid="[^"]*"//;
      $attributes =~ s/^pid="[^"]*"//;
    }
    $attributes =~ s/^PRATTSEP//;
    $attributes =~ s/\n/ /g;

    unless (($element =~ /ENTITY/)||($element =~ /NOTATION/)||
    ($element =~ /ATTLIST/)||($element =~ /\?xml/)||($element =~ /!--/)) {
      @elist{$element} = $element;
      @alist = split(/PRATTSEP/, $attributes);
      foreach $i (0..$#alist) {
        ($attname, $attvalue) = split(/=/, $alist[$i]);
        $attname  =~ s/^\s*//;
        $attname  =~ s/\s*$//;
        $attvalue =~ s/\n//g;
        $attvalue =~ s/^"//;
        $attvalue =~ s/"$//;
        $attvalue =~ s/^\s*//;
        $attvalue =~ s/\s*$//;
        $attvalue =~ s/\[/\\\[/g;
        $attvalue =~ s/\(/\\\(/g;
        $attvalue =~ s/\)/\\\)/g;
        $alistkey = $element."PRELATSEP".$attname;
        unless (@alist{$alistkey} =~ / $attvalue /) {
          @alist{$alistkey} = @alist{$alistkey}." ".$attvalue." ";
        }
      }
    }
  }
}

foreach $key (sort(keys(%elist))) {
  print "$elist{$key}\n";
  foreach $item (sort(keys(%alist))) {
    if ($item =~ /^$elist{$key}PRELATSEP/) {
      $label = $item;
      $label =~ s/^$elist{$key}PRELATSEP//;
      $value = $alist{$item};
      $value =~ s/  /, /g;
      $value =~ s/^ //;
      print "  $label=$value\n";
    }
  }
}