#!/usr/bin/perl -s

@multitags = ("A", "PRE", "B", "BLINK", "I", "U", "TT", "ADDRESS",
              "BLOCKQUOTE", "CITE", "CODE", "DFN", "EM", "KBD",
              "SAMP", "STRONG", "VAR", "FONT", "HEAD", "HTML",
              "TITLE", "BODY", "TABLE", "TR", "TD", "FORM",
              "FONT", "FRAMESET", "MAP", "CENTER", "SCRIPT",
              "NOSCRIPT","OL","UL","DIV","SPAN","SELECT");
@indenttags = ("PRE", "HEAD", "HTML", "BODY", "TABLE", "FORM",
               "FRAMESET", "MAP", "CENTER", "SCRIPT", "NOSCRIPT",
               "OL", "UL", "DIV", "SELECT");

@templatetags = (
             "foreach",
             "include",
             "include-sub",
             "print",
             "eval",
             "if",
             "ifnot",
             "ifeval",
             "else",
             "elseif",
             "elseifnot",
             "elseifeval",
           );
$templatetagreg = join '|', @templatetags;

$onelinenobr = 0;

foreach $file (@ARGV) {
  open(F,"<$file") or die "file not found";
  @textin=<F>;
  close(F);
  $textin=join('', @textin);
  $textin = &unwordify($textin);
  $textout = "";

  print STDERR "Preprocess Cleanup...\n";
  $textin =~ s!^( |\t)*(.*)( |\t)*$!$2!igm;
#  $textin =~ s|^ *(.*) *$|$1|igm;
#  $textin =~ s|^\t*(.*)\t*$|$1|igm;
  $textin =~ s|  +| |gis if $nodoublespace;
  $indent = "";
  @stack = ();

  print STDERR "Processing...\n";
  $notif = 2000000000;
  $tilen = length($textin);
  while ($textin ne "") {
    while ($textin && substr($textin,0,1) eq "\n") {
      $textout = $textout . "\n" . $indent;
      $textin = substr($textin, 1);
    }
    if (($v=length($textin)) <= $notif) {
      print STDERR $v . " bytes left to process.\n";
      $notif = $v - 1000;
    }
    if (substr($textin,0,1) eq "<") {
#print STDERR "\$textin: '$textin'\n";
      if (substr($textin,1,1) ne "!") {
        while( $textin =~ s|^(<[^>]*)\n([^>]*>)|$1 $2|s ) {};
        $textin =~ m|^(<[^>]+>)(.*)|s;
        $tag = $1;
        $textin = $2;
        $tag =~ s|  | |g;
#        $tag = &rewritetag($tag) unless &in("script", @stack)!=-1;
        $tag = &rewritetag($tag);
        $textin = $tag . $textin;
      }
#print STDERR "\$textin: '$textin'\n";
      $textin =~ m|^<(/?)([^ >]+)|is;
      $endtag = ($1 eq "/") || 0;
      $thistag = $2;
#print STDERR "thistag: '$thistag'  endtag: $endtag\n";
#print STDERR ">Stack: (" . join(", ", @stack) . ")  \$stack[-1]: '$stack[-1]'  \$stack[-2]: '$stack[-2]'\n";
      if (((lc $thistag) eq "script" || &in("script", @stack)==-1) && &in($thistag, @multitags)>=0) {
#print STDERR "It's a multitag.\n";
        if (!$endtag) {
          push @stack, $thistag;
        } else {
          if ($thistag eq $stack[-1]) {
            pop stack;
          } else {
#          } elsif(&in("script", @stack)==-1) {
            print STDERR "Unmatched endtag '/" . $thistag . "' near byte " . ($tilen - length($textin) + 1) . "!\n";
            print STDERR "  Stack: ", join(", ", @stack) . "\n";
            local($contextlen) = 1050;
            local($context) = substr($textout, -$contextlen) . "==========> " . substr($textin,0,$contextlen);
            if($nocroak) {
              $context =~ s|    +|...|gs;
              $context =~ s|\n|\\n|gs;
              print STDERR "  Context: '$context'\n";
            } else {
              $context =~ s|\t|  |gs;
              $context =~ s|    +|...|gs;
              $context =~ s|\n+|\n|gs;
              print STDERR "  Context:\n$context\n";
              exit(1) unless $nocroak;
            }
          }
        }
      }
#print STDERR "<Stack: (" . join(", ", @stack) . ")  \$stack[-1]: '$stack[-1]'  \$stack[-2]: '$stack[-2]'\n";
      if (((lc $thistag) eq "script" || &in("script", @stack)==-1) && &in($thistag, @indenttags)>=0) {
#print STDERR "It's an indenttag.\n";
        if (!$endtag) {
          if ($textout !~ m|\n *$| ) {
            $textout = $textout . "\n" . $indent;
          }
          $indent = $indent . "  ";
          $textout .= substr($textin, 0, $i=(index($textin, ">")+1));
          $textout = $textout . "\n" . $indent;
          $textin = substr($textin, $i);
          $textin = substr($textin, 1) if substr($textin, 0, 1) eq "\n";
        } else {
#print STDERR "... and it's an endtag\n";
#print STDERR "] \$indent: '$indent'  \$tag: '$tag'\n";
#print STDERR "] \$textin[1..25]: '" . substr($textin, 0, 25) . "'\n";
#print STDERR "] \$textout[\$-24..\$]: '" . substr($textout, -25) . "'\n";
          $indent = substr($indent, 0, -2);
          if ($textout !~ m|\n *$| ) {
            $textout = $textout . "\n" . $indent;
          } elsif ($textout =~ m|\n +$| ) {
            $textout =~ s|\n +$|\n$indent|;
          }
          $textout .= substr($textin, 0, $i=(index($textin, ">")+1));
          $textout = $textout . "\n" . $indent;
          $textin = substr($textin, $i);
          $textin = substr($textin, 1) if substr($textin, 0, 1) eq "\n";
#print STDERR "[ \$indent: '$indent'  \$tag: '$tag'\n";
#print STDERR "[ \$textin[1..25]: '" . substr($textin, 0, 25) . "'\n";
#print STDERR "[ \$textout[\$-24..\$]: '" . substr($textout, -25) . "'\n";
        }
      } elsif (&in("script", @stack)==-1 && $thistag eq "TR") {
#print STDERR "TR tag\n";
        if (!$endtag) {
          if ($compact || $textin =~ m|^[^\n]+</TR>|i) {
            $onelinenobr++;
            $textout .= substr($textin,0,1);
            $textin = substr($textin, 1);
          } else {
            if ($textout !~ m|\n *$| ) {
              $textout = $textout . "\n" . $indent;
            }
            $indent = $indent . "  ";
            $textout .= substr($textin, 0, $i=(index($textin, ">")+1));
            $textout = $textout . "\n" . $indent;
            $textin = substr($textin, $i);
            $textin = substr($textin, 1) if substr($textin, 0, 1) eq "\n";
          }
        } else {
          if ($textout =~ m|<TR[^\n]+$|c) {
            $onelinenobr--;
            $textout .= substr($textin,0,1);
            $textin = substr($textin, 1);
          } else {
            $indent = substr($indent, 0, -2);
            if ($textout !~ m|\n *$| ) {
              $textout = $textout . "\n" . $indent;
            } elsif ($textout =~ m|\n +$| ) {
              $textout =~ s|\n +$|\n$indent|;
            }
            $textout .= substr($textin, 0, $i=(index($textin, ">")+1));
            $textout = $textout . "\n" . $indent;
            $textin = substr($textin, $i);
            $textin = substr($textin, 1) if substr($textin, 0, 1) eq "\n";
          }
        }
      } elsif (&in("script", @stack)==-1 && $thistag eq "TD") {
#print STDERR "TD tag\n";
        if (!$endtag) {
          if ($compact || $textin =~ m|^[^\n]+</TD>|i) {
            $onelinenobr++;
            $textout .= substr($textin,0,1);
            $textin = substr($textin, 1);
          } else {
            if ($textout !~ m|\n *$| ) {
              $textout = $textout . "\n" . $indent;
            }
            $indent = $indent . "  ";
            $textout .= substr($textin, 0, $i=(index($textin, ">")+1));
            $textout = $textout . "\n" . $indent;
            $textin = substr($textin, $i);
            $textin = substr($textin, 1) if substr($textin, 0, 1) eq "\n";
          }
        } else {
          if ($textout =~ m|<TD[^\n]+$|c) {
            $onelinenobr--;
            $textout .= substr($textin,0,1);
            $textin = substr($textin, 1);
          } else {
            $indent = substr($indent, 0, -2);
            if ($textout !~ m|\n *$| ) {
              $textout = $textout . "\n" . $indent;
            } elsif ($textout =~ m|\n +$| ) {
              $textout =~ s|\n +$|\n$indent|;
            }
            $textout .= substr($textin, 0, $i=(index($textin, ">")+1));
            $textout = $textout . "\n" . $indent;
            $textin = substr($textin, $i);
            $textin = substr($textin, 1) if substr($textin, 0, 1) eq "\n";
          }
        }
      } elsif (&in("script", @stack)==-1 && $thistag eq "FONT") {
#print STDERR "Font tag\n";
        if (!$endtag) {
          if ($compact || $stack[-2] eq "A" || $textin =~ m|^[^\n]+</FONT>|i) {
            $onelinenobr++;
            $textout .= substr($textin,0,1);
            $textin = substr($textin, 1);
          } else {
            if ($textout !~ m|\n *$| ) {
              $textout = $textout . "\n" . $indent;
            }
            $indent = $indent . "  ";
            $textout .= substr($textin, 0, $i=(index($textin, ">")+1));
            $textout = $textout . "\n" . $indent;
            $textin = substr($textin, $i);
            $textin = substr($textin, 1) if substr($textin, 0, 1) eq "\n";
          }
        } else {
          if ($stack[-1] eq "A" || $textout =~ m|<FONT[^\n]+$|c) {
            $onelinenobr--;
            $textout .= substr($textin,0,1);
            $textin = substr($textin, 1);
          } else {
            $indent = substr($indent, 0, -2);
            if ($textout !~ m|\n *$| ) {
              $textout = $textout . "\n" . $indent;
            } elsif ($textout =~ m|\n +$| ) {
              $textout =~ s|\n +$|\n$indent|;
            }
            $textout .= substr($textin, 0, $i=(index($textin, ">")+1));
            $textout = $textout . "\n" . $indent;
            $textin = substr($textin, $i);
            $textin = substr($textin, 1) if substr($textin, 0, 1) eq "\n";
          }
        }
      } elsif (!$onelinenobr && !$nobr && &in("script", @stack)==-1 && $thistag eq "BR" && $textin !~ m|^<BR><BR|i) {
#print STDERR "\$textin: '$textin'\n";
#print STDERR "\$i=" . (index($textin, ">")+1) . "\n";
#print STDERR "substr='" . substr($textin, 0, $i=(index($textin, ">")+1)) . "'\n";
        $textout .= substr($textin, 0, $i=(index($textin, ">")+1));
        $textout = $textout . "\n" . $indent;
        $textin = substr($textin, $i);
        $textin = substr($textin, 1) if substr($textin, 0, 1) eq "\n";
      } else {
        $textout .= substr($textin,0,1);
        $textin = substr($textin, 1);
      }
    } elsif ($textin =~ m|^\{| && (&in("script", @stack)>=0)) {
#print STDERR "\$textin: '$textin'\n";
#print STDERR "found { in script.\n";
      $indent = $indent . "  ";
      if ($textout !~ m|[ \n]$| ) {
        $textout .= " ";
      }
      $textout .= substr($textin,0,1);
      $textout = $textout . "\n" . $indent;
      $textin = substr($textin, 1);
      $textin = substr($textin, 1) if substr($textin, 0, 1) eq "\n";
    } elsif ($textin =~ m|^\}| && (&in("script", @stack)>=0)) {
      $indent = substr($indent, 0, -2);
      if ($textout !~ m|\n *$| ) {
        $textout = $textout . "\n" . $indent;
      } elsif ($textout =~ m|\n +$| ) {
        $textout =~ s|\n +$|\n$indent|;
      }
      $textout .= substr($textin,0,1);
      if ($textin !~ m|^\} *[^\n]|) {
        $textout = $textout . "\n" . $indent;
      }
      $textin = substr($textin, 1);
      $textin = substr($textin, 1) if substr($textin, 0, 1) eq "\n";
    } else {
      $textout .= substr($textin,0,1);
      $textin = substr($textin, 1);
    }
    while ($textin ne "" && substr($textin,0,1) eq "\n") {
      $textout = $textout . "\n" . $indent;
      $textin = substr($textin, 1);
    }
  }

  if(@stack) {
    print STDERR "Unclosed tags remaining!\n";
    print STDERR "  Stack: ", join(", ", @stack) . "\n";
    exit(1) unless $nocroak;
  }

  print STDERR "Postprocess Cleanup...\n";
  $textout =~ s|^\n+||igs;
  $textout =~ s|\n+$|\n|igs;

  print STDERR "Postpostprocess Cleanup...\n";
  $textout =~ s|\n +\n|\n|igs;
  $textout =~ s|\n+|\n|igs;
  $textout =~ s| +\n|\n|igs;

  print STDERR "Postpostpostprocess Cleanup...\n";
  if(!$xml) {
    $textout =~ s|\"(#[0-9a-f]+)\"|uc($1)|egis;
    $textout =~ s|\\\"(#[0-9a-f]+)\\\"|uc($1)|egis;
  }

  print STDERR "Postpostpostpostprocess Cleanup...\n";
  if(!$xml) {
    $textout =~ s| align=\"left\"| ALIGN=LEFT|gis;
    $textout =~ s| align=\"right\"| ALIGN=RIGHT|gis;
    $textout =~ s| align=\"center\"| ALIGN=CENTER|gis;
    $textout =~ s| align=\"middle\"| ALIGN=CENTER|gis;
    $textout =~ s| align=middle| ALIGN=CENTER|gis;
    $textout =~ s| valign=\"top\"| VALIGN=TOP|gis;
    $textout =~ s| valign=\"bottom\"| VALIGN=BOTTOM|gis;
    $textout =~ s| valign=\"center\"| VALIGN=CENTER|gis;
    $textout =~ s| valign=\"middle\"| VALIGN=CENTER|gis;
    $textout =~ s| valign=middle| VALIGN=CENTER|gis;
  }

  if($inplace) {
    rename($file,"$file.bak") unless $nobak;
    open(F,">$file") or die "$!";
    print F $textout;
    close(F);
  } else {
    print $textout;
  }
}

sub rewritetag {
  local($tag) = @_;
  return $tag if $template && ($tag =~ /^(.*?)(<(\/?)($templatetagreg)( ([^>]*?))?(\/?)>)(\n?)(.*)$/si);
#print STDERR "rewritetag(\"$tag\")\n";
  $tag =~ s|^<(.*)>$|$1|is;
  $tag =~ m|^([^ ]+) ?(.*)$|is;
  $tagmeat = $2;
  $tag = uc($1);
  $newtag = "<$tag";
  $key = "";
  $value = "";
  $find = "=";
  while ($tagmeat ne "") {
#print STDERR "> tagmeat: '$tagmeat'\n";
#print STDERR "\$find: '$find' \$key: '$key' \$value: '$value'\n";
    if (substr($tagmeat,0, length($find)) eq $find) {
#print STDERR "found '$find'\n";
      if ($find eq "=") {
        $find = (substr($tagmeat,1,2) eq "\\\"") ? "\\\"" : ((substr($tagmeat,1,1) eq "\"") ? "\"" : ((substr($tagmeat,1,1) eq "'") ? "'" : " "));
        $value = substr($tagmeat,1, length($find));
        $tagmeat = substr($tagmeat, length($find));
      } else {
        if ($value !~ m|\\$| ) {
          if ($find ne " ") {
            $value = $value . substr($tagmeat,0, length($find));
          }
#print STDERR "-Stack: (" . join(", ", @stack) . ")  \$stack[-1]: '$stack[-1]'  \$stack[-2]: '$stack[-2]'\n";
#print STDERR ">\$key: '$key' \$value: '$value'\n";
          if ($value =~ m|^'(.*)'$|) {
            if(&in("script", @stack)!=-1) {
              $value = "\\\"" . $1 . "\\\"";
            } else {
              $value = "\"" . $1 . "\"";
            }
          } elsif ( ($xml || ($value !~ m|^-?[0-9]+$|)) && ($value !~ m|^\".*\"$|) && ($value !~ m|^\\\".*\\\"$|)) {
            if(&in("script", @stack)!=-1) {
              $value = "\\\"" . $value . "\\\"";
            } else {
              $value = "\"" . $value . "\"";
            }
          } elsif ($value =~ m|^\"([0-9]+)\"$|) {
            $value = $1;
          }
#print STDERR "<\$key: '$key' \$value: '$value'\n";
          $newtag = $newtag . " " . uc($key) . "=" . $value;
          $key = "";
          $value = "";
          $tagmeat = substr($tagmeat, length($find));
          $find = "=";
          while ($tagmeat ne "" && (substr($tagmeat,0,1) eq " ")) {
            $tagmeat = substr($tagmeat, 1);
          }
          $tagmeat = " " . $tagmeat;
        } else {
          $value .= substr($tagmeat, 0, length($find));
        }
      }
    } elsif ($find eq "=" && substr($tagmeat,0, length($find)) eq " ") {
      $newtag = $newtag . " " . uc($key);
      $key = "";
      $value = "";
      $tagmeat = substr($tagmeat, length($find));
      $find = "=";
      while ($tagmeat ne "" && (substr($tagmeat,0,1) eq " ")) {
        $tagmeat = substr($tagmeat, 1);
      }
      $tagmeat = " " . $tagmeat;
    } elsif ($find eq "=") {
      $key .= substr($tagmeat,0,1);
    } else {
      $value .= substr($tagmeat,0,1);
    }
    $tagmeat = substr($tagmeat, 1);
#print STDERR "< tagmeat: '$tagmeat'\n";
  }
  if($key ne "" && $value ne "") {
#print STDERR "-Stack: (" . join(", ", @stack) . ")  \$stack[-1]: '$stack[-1]'  \$stack[-2]: '$stack[-2]'\n";
#print STDERR ">\$key: '$key' \$value: '$value'\n";
    if ($value =~ m|^'(.*)'$|) {
      if(&in("script", @stack)!=-1) {
        $value = "\\\"" . $1 . "\\\"";
      } else {
        $value = "\"" . $1 . "\"";
      }
    } elsif ( ($xml || ($value !~ m|^-?[0-9]+$|)) && ($value !~ m|^\".*\"$|) && ($value !~ m|^\\\".*\\\"$|)) {
      if(&in("script", @stack)!=-1) {
        $value = "\\\"" . $value . "\\\"";
      } else {
        $value = "\"" . $value . "\"";
      }
    }
#print STDERR "<\$key: '$key' \$value: '$value'\n";
    $newtag = $newtag . " " . uc($key) . "=" . $value;
  } elsif($key ne "" && $find eq "=") {
    $newtag = $newtag . " " . uc($key);
  }
  $newtag .= ">";
#print STDERR "rewritetag => '$newtag'\n";
  return $newtag;
}

sub in {
  local($val, @list) = @_;
  for($i=0; $i<=$#list; $i++) {
    return $i if (uc($list[$i]) eq uc($val));
  }
  return -1;
}

# Translate stupid word characters
sub unwordify {
    my($todecode) = @_;
    $todecode =~ s/\xAE/\&reg\;/gs;
    $todecode =~ s/\xA9/\&copy\;/gs;
    $todecode =~ s/\x99/\&\#153\;/gs;
    $todecode =~ s/\x91/'/gs;
    $todecode =~ s/\x92/'/gs;
    $todecode =~ s/\xB4/'/gs;
    $todecode =~ s/\x93/\"/gs;
    $todecode =~ s/\x94/\"/gs;
    $todecode =~ s/\&quot\;/\"/gs;
    $todecode =~ s/\xBA/\&deg\;/gs;
    $todecode =~ s/\x82/,/gs;
    $todecode =~ s/\x85/.../gs;
    $todecode =~ s/\x84/,,/gs;
    $todecode =~ s/\x96/--/gs;
    $todecode =~ s/\x97/---/gs;
    return $todecode;
}