=head1 NAME

order-footnote

=head1 SYNOPSIS

order-footnote.pl < [broken-page] > [fixed-page]

=head1 DESCRIPTION

This is a proof of concept program which reorders wikipedia automatic
footnotes.

=head1 BUGS

The program assumes that the {{ref}} macros all occur after the
{{note}} macros in one block, which they should.  Before bringing into
use, this should actually be checked.

We don't correctly handle duplicate references.  We haven't yet got
functionality to guess a good name and spit out the same footnote
twice.  In this case we output what looks like a missing footnote, but
links back to the earlier reference link forward to its matching
footnote.

=head1 COPYRIGHT

This program may be distributed under the terms of the GFDL or, more
appropriately under the terms of the GNU GPL.  Copyright 2005 "mozzerati"
of the wikipedia project.

=cut


#read through references till we hit a note

$reftemplate="ref";
$reftemplatere="(?:[Rr]ef|[a]n)";
$notetemplate="note";
$notetemplatere="(?:[Nn]ote|[Aa]nb)";


@reflist=();
%fixrefs=(); # mapping of references in original to new name for them
%notes=();
while ( <> ) {
  #check if we reach the footnotes section
  m/^([#\*])?\s*\{\{$notetemplatere\|/ && last;
  while ( s/(.*?)\{\{($reftemplatere)\|([^\{\}]+)(\})// ) {
    my $text=$1; my $template=$2; my $ref=$3; my $close=$4;
    my $baseref=$ref; my $differ=1;
    while (defined $fixrefs{$ref}) {
      $differ++;
      $ref=$baseref . $differ;
    }
    print $text, "{{", $template, "|", $ref, "$close";
    print STDERR "ref $ref\n";
    $fixrefs{$ref}=$baseref;
    push @reflist, $ref;
  }
  print $_; #remains of the line with no references
}

#read through the entire rest of the text finding notes; We print out
#notes all together at the position of the first note, but it's not
#totally obvious that that's correct.. perhaps we should really do two
#passes, read through from beginning to end, get all references, then
#position the notes at the position of the last note.

defined $_ or do {
  die "failed to find any footnotes" if @reflist;
  #apparently the article is empty of notes, so all should be okay.
  print STDERR "no notes found; article should be unchanged\n";
  exit;
};

$after ="";
$matched=0;

LOOP: {
  do{{
    defined $_ or last;
    $matched && m/^\#[#*:]|^\<\!\-\-/ && do {
      $notes{$name} .= $_;
      next;
    };
    $matched=0;
    # we should check for broken templates (e.g. not closed properly)
    if ( ($name)=m/^(?:[\#\*])?\s*\{\{$notetemplatere\|([^\{\}]+)\}/  ) {
      #we should check if the same note is already defined, and put at the end with a new name if it is...
      #in doing this we should prefer the fuller note, ignoring semi-empty ones
      $notes{$name} = $_;
      $matched=1;
    } else {
      $after .= $_;
    }
  } continue { $_ = <> } } while (defined $_);
}


%usednotes=();
while ( $name=shift @reflist ) {
  my $oldname=$fixrefs{$name};
  defined $notes{$oldname} or do {
    print "# {{$notetemplate|$name}} {{note-missing}}\n";
    next;
  };
  my $note=$notes{$oldname};
  $note =~ s/({{$notetemplatere|)$oldname/$1$name/;
  print $note;
  $usednotes{$name}=1;
}

# print out unused notes:
# the real version would probably separate them out
while (($key, $value) = each %notes) {
  print $value unless $usednotes{$key} ;
}

print $after;