#!/usr/bin/perl -w
# 
# t2h v0.03	--crb3 09feb01/20feb01/04mar02
# started out with fanfiction.net stream. then expanded switch-options 
# to allow for full pages, set up the title nicely, etc.; still have
# yet to get it to tear off mail-headers. 
# this tool doesn't do all the pretty tabling done by the various 
# tools on the Win32 side (ws2html etc), it just converts from
# a r.a.a.c-style text posting to something that can be fed to
# ff.net, or (full-page) posted on an internal server.
#
# --crb3 04mar04: if given filename doen't exist, try tacking ".txt" 
# onto it and look for that.
#
$debug = 0;		# get chatty, show internal results 
#
# these are command-line switches
#
$para=1;		# use <P></P> paragraphing?
$hardln=0;		# convert \n to <BR>?
$fullpage=0;		# generate full page?
$tpt=0;			# got a template for fullpage?
$template="";		# here's where its name goes

$title="title";
$author="author";

die "txt2html [options] infile [outfile]
  options:
  -f emit full HTML page, not just a story stream toggle
  -h keep hard linebreaks toggle
  -p use <P> paragraphs toggle (default)
  -a'AUTHOR' insert author's name\n" if($#ARGV < 0);

for($nm=$b=0;$b<=$#ARGV;$b++){
  if($ARGV[$b] =~ /^\-p/i){	    # P: use <P> paragraphs
    $para ^= 1;
  }elsif($ARGV[$b] =~ /^\-h/i){	    # H: keep hard linebreaks
    $hardln ^= 1;
  }elsif($ARGV[$b] =~ /^\-f/i){	    # F: emit full HTML page
    $fullpage ^= 1;
    $template = $ARGV[$b];
    chomp $template;
    $template =~ s/^\-f\=?//;
    if($template ne ""){ 
      $tpt=1;			    #  is a template filename 
      print "using template $template... ";
    }else{
      print "using internal template... ";
    }

  }elsif($ARGV[$b] =~ /^\-a/i){     # A: insert author's name
    $author = $ARGV[$b];
    $author =~ s/^\-a\=?//i;

    print "author = $author\n" if $debug;

  }elsif($nm==0){
    $infile = $ARGV[$b]; $nm++;

    print "infile = $infile\n" if $debug;

  }elsif($nm==1){
    $outfile = $ARGV[$b]; $nm++;

    print "outfile = $outfile\n" if $debug;

  }else{
    warn "Extra arg ignored: $ARGV[$b]\n";
  }
}

print "nm=$nm\n" if $debug;

if($nm<2){
  $outfile = $infile;
  $outfile =~ s/\.txt//i;
}

$infile .= ".txt" unless(-e $infile);

$outfile .= ".html" unless $outfile =~ /\.html?$/i;
die "oops, infile and outfile are both $infile\n" if $infile eq $outfile;

print "Converting $infile to $outfile";
print " full" if $fullpage;
print "\n";



open(IFIL,"<$infile") or die "Can't open infile $infile\n";
open(OFIL,">$outfile") or die "Can't make outfile $outfile\n";

print OFIL "<P>" if $para;
$firstln=1;
while(defined($inline=<IFIL>)){
  $blank=0;
  chomp $inline;

  if($firstln){
    $title = $inline;		# snag it in case it's used 
    $title =~ s/^\;//;
    $firstln=0;
$datacnt=0 if $debug;

  if($tpt){
    open(TFIL,"<$template") or die "can't open template $template\n";
  }
#
# #PAYLOAD# tag must be on a line by itself.
#
    if($fullpage){
      while(defined($pgline=&tfetch) and $pgline !~ /^\#PAYLOAD\#/i){
        $pgline =~ s/\#TITLE\#/$title/i; 
	$pgline =~ s/\#AUTHOR\#/$author/i;
        print OFIL $pgline;
        $datacnt++ if $debug;
      }
      next;			# don't print first line twice
    }
print "datacnt=$datacnt\n" if $debug;
  }

  if($inline =~ /^\;\</){
    substr($inline,0,1) = "";	# ;<tag> -> <tag>
  }else{
    $inline =~ s/\</&lt;/g;	# escape-out <> before adding
    $inline =~ s/\>/&gt;/g;	# in any tags
  }


  if($inline =~ /^\s*$/){			# blank line?
    if($para){
        $inline = "</P><P>";
    }else{
      $inline = "<BR>";
    }
    $blank=1;
  }

  $inline =~ s#(^|[\s,.!?"'])\_\*(\w)#$1<I><B>$2#g;		# _*bolditalic*_
  $inline =~ s#(\w[,.!?"']*)\*\_([\s.,!?"']|$)#$1</I></B>$2#g;

  $inline =~ s#(^|[\s.,!?"'])\_(\w)#$1<I>$2#g;	# _italic_
  $inline =~ s#(\w[,.!?"']*)\_([\s.,!?"']|$)#$1</I>$2#g;
#
# --crb3 02dec01: special coding for title...
#
  $inline =~ s#(^|[\s.,!?"'])\*\!\*(\w)#$1<H2>$2#g;	#*!*Title*!*
  $inline =~ s#(\w[\],.!?"']*)\*\!\*([\s,.!?"']|$)#$1</H2>$2#g;

  $inline =~ s#(^|[\s.,!?"'])\*(\w)#$1<B>$2#g;	# *bold*
  $inline =~ s#(\w[,.!?"']*)\*([\s,.!?"']|$)#$1</B>$2#g;

  $inline =~ s/\-{8,}/\<hr \/\>/g;

  $inline .= "<BR>" if ($hardln and not $blank);
  print OFIL "$inline\n";
}
print OFIL "</P>\n" if $para; 

if($fullpage){
  while(defined($pgline=&tfetch)){
    $pgline =~ s/\#TITLE\#/$title/i; 
    $pgline =~ s/\#AUTHOR\#/$author/i;
    print OFIL $pgline;
  }
}

close IFIL;
close OFIL;
close TFIL if $tpt;


#
# what's past here is a plain-jane page framework, just enough
# to put some margins on the page and get through picky browsers.
# I use it to post fics for intranet browsing at home.
#
#

sub tfetch {
  my($i);

  if($tpt){
    return($i=<TFIL>);    
  }else{
    return($i=<DATA>);
  }
}


__END__

<HTML>
<HEAD>
<META HTTP-EQUIV="Content-Type" CONTENT="text/html; charset=iso-8859-1">
<TITLE>
#TITLE#
</TITLE>
<META NAME="Author" CONTENT="
#AUTHOR#
">
<META NAME="GENERATOR" CONTENT="txt2html --crb3 09feb01">
</HEAD>
<BODY TEXT="#000000" BGCOLOR="#C0C0C0" LINK="#0000EE" VLINK="#CC0000" ALINK="#FFFF00">
  <CENTER>
  <H2>
#TITLE#
  </H2>
  <TABLE ALIGN="CENTER" WIDTH="90%">
    <TR><TD>
#PAYLOAD#
    </TD></TR>
  </TABLE>
  <HR>
  <CENTER>
</BODY>
</HTML>