#!/usr/bin/perl -w
# NAME / AUTHOR
# ===================================
# htmlcleaner.pl by Eduardo Unda-Sanzana
# Antofagasta (CHILE)
#
# PURPOSE
# ===================================
# This script cleans and organize the HTML output from
# programs such as Open Office.
#

use Text::Wrap;

# Read arguments from the command line.
if (@ARGV != 2) {
	print "							  \n";
	print "htmlcleaner.pl - Eduardo Unda-Sanzana              \n";
	print "							  \n";
	print "Usage: perl htmlcleaner.pl in out                  \n";
	print "							  \n";
	print "in  : File to be processed.                        \n";
        print "out : File to store the cleaned HTML file.         \n";
	print "							  \n";
	die   "							  \n";
}

$in   = shift;     
$out  = shift;    

# Perform some control on the user's input.
(-e "$in")  || die "$in does not exist in this directory.\n";
(-f "$in")  || die "$in it is not a plain file.\n";
(-e "$out") && die "$out exists in this directory.\n";

# Open the file.
open(IN,"$in")   || die "Sorry, I couldn't open $in\n";
open(OUT,">$out") || die "Sorry, I couldn't open $out for writing.\n";
undef $/;  # This allows reading the whole file in one single string.
$content = <IN>;
$/ = "\n"; # Then I restore the undef variable to its normal behaviour. 
# $content =~ s/([0-9])([0-9])([0-9])([0-9])-([0-9])([0-9])-([0-9])([0-9])/$7$8\/$5$6\/$3$4/g;

# Remember to first eliminate everything useless. Only then
# format the output.

# Removes new lines.
$content =~ s/\n/ /g; 

# Remove everything prior to and after body. This might be optional in a later version.
$content =~ s/^(.*)<BODY(.*?)>//ig;
$content =~ s/<\/BODY(.*?)$//ig;

# Remove the header and footer div created by Open Office.
$content =~ s/<DIV TYPE=HEADER>(.*?)<\/DIV>//g;
$content =~ s/<DIV TYPE=FOOTER>(.*?)<\/DIV>//g;

# Remove useless blank space.
$content =~ s/\s+/ /g;
$content =~ s/>\s</></g;

# Eliminate font stuff.
$content =~ s/<FONT(.*?)>//ig;
$content =~ s/<\/FONT>//ig;

# Eliminate language stuff.
$content =~ s/<SPAN LANG(.*?)>//ig;
$content =~ s/<\/SPAN>//ig;

# Eliminate multicolumn stuff.
$content =~ s/<MULTICOL(.*?)>//ig;
$content =~ s/<\/MULTICOL>//ig;

# Eliminate column stuff.
$content =~ s/<COL(.*?)>//ig;

# Remove parameters from inside tags. Also capitalize.
$content =~ s/<[Hh]([1-6])(.*?)>/<h$1>/g;
$content =~ s/<\/[Hh]([1-6])>/<\/h$1>/g;
$content =~ s/<[Pp](.*?)>/<p>/g;
$content =~ s/<\/[Pp]>/<\/p>/g;
$content =~ s/<OL(.*?)>/<ol>/g;
$content =~ s/<\/OL>/<\/ol>/g;
$content =~ s/<UL(.*?)>/<ul>/g;
$content =~ s/<\/UL>/<\/ul>/g;
$content =~ s/<LI(.*?)>/<li>/g;
$content =~ s/<\/LI>/<\/li>/g;
$content =~ s/<B>/<b>/g;
$content =~ s/<\/B>/<\/b>/g;
$content =~ s/<I>/<i>/g;
$content =~ s/<\/I>/<\/i>/g;
$content =~ s/<U>/<u>/g;
$content =~ s/<\/U>/<\/u>/g;
$content =~ s/<IMG(.*?)>/<img$1>/g;
$content =~ s/<A(.*?)>/<a$1>/g;
$content =~ s/<\/A>/<\/a>/g;
$content =~ s/<SUP>/<sup>/g;
$content =~ s/<\/SUP>/<\/sup>/g;
$content =~ s/<SUB>/<sub>/g;
$content =~ s/<\/SUB>/<\/sub>/g;
$content =~ s/<TABLE(.*?)>/<table>/g;
$content =~ s/<\/TABLE>/<\/table>/g;
$content =~ s/<TR(.*?)>/<tr>/g;
$content =~ s/<\/TR>/<\/tr>/g;
$content =~ s/<TD(.*?)>/<td$1>/g; 
$content =~ s/<\/TD>/<\/td>/g;

# Format table tags.



# Delete empty structures.
$content =~ s/<p>\s*<BR>\s*<\/p>//g;
$content =~ s/<p>\s*<br>\s*<\/p>//g;
$content =~ s/<p>\s*<BR \/>\s*<\/p>//g;
$content =~ s/<p>\s*<br \/>\s*<\/p>//g;
$content =~ s/&shy;//g;

# Check headers for img tags.
$content =~ s/<h([1-6])>(.*?)(<img.*?>)(.*?)<\/h([1-6])>/<h$1>$2$4<\/h$5>$3/g;

# Check <li> structures. Don't mess with the order in which I'm checking
# this. It took me days to figure out that the first one must be indeed
# the first one... By the way, the first test includes a nifty piece of
# code. Note that the weird structure in the middle of the pattern instructs
# the s operator to *not* match the pattern <p>. In this way I can avoid
# a problem arising from the combination of left-to-right replacement plus
# greediness. So that you can remember what I'm talking about, consider
# this bit of uncleaned HTML code:
#
# <ol><li><p>One.</p><li><p>Two.</p></ol>
#
# How do you tell the s operator to pick:
#
# <li><p>Two.</p></ol>
#
# and do something with it? It is not trivial because something like
#
# s/<li><p>(.+?)<\/p><\/ol>/do something/g
#
# will take everything after the *first* <li>.
#
$content =~ s/<li><p>((?:(?!<p>).)*)<\/p><\/ol>/<li>$1<\/li><\/ol>/sg;
$content =~ s/<ol><li><p>(.+?)<\/p><\/ol>/<ol><li>$1<\/li><\/ol>/g;
$content =~ s/<ol><li><p>(.+?)<\/p><li>/<ol><li>$1<\/li><li>/g;
$content =~ s/<li><p>((?:(?!<p>).)*)<\/p><\/ul>/<li>$1<\/li><\/ul>/sg;
$content =~ s/<ul><li><p>(.+?)<\/p><\/ul>/<ul><li>$1<\/li><\/ul>/g;
$content =~ s/<ul><li><p>(.+?)<\/p><li>/<ul><li>$1<\/li><li>/g;
1 while ($content =~ s/<li><p>(.+?)<\/p><li>/<li>$1<\/li><li>/g);

# Format break tags.
$content =~ s/<BR>/<br \/>/g;
$content =~ s/<br>/<br \/>/g;

# Trust local encoding...
$content =~ s/&aacute\;/á/g;
$content =~ s/&eacute\;/é/g;
$content =~ s/&iacute\;/í/g;
$content =~ s/&oacute\;/ó/g;
$content =~ s/&uacute\;/ú/g;
$content =~ s/&Aacute\;/Á/g;
$content =~ s/&Eacute\;/É/g;
$content =~ s/&Iacute\;/Í/g;
$content =~ s/&Oacute\;/Ó/g;
$content =~ s/&Uacute\;/Ú/g;
$content =~ s/&ntilde\;/ñ/g;
$content =~ s/&Ntilde\;/Ñ/g;
$content =~ s/&iquest\;/¿/g;
$content =~ s/&ndash\;/-/g;
$content =~ s/&ldquo\;/"/g;
$content =~ s/&rdquo\;/"/g;
$content =~ s/&iexcl\;/¡/g;

# Format headers.
$content =~ s/(<h[1-6]>)(.*?)(<\/h[1-6]>)/\n$1\n$2\n$3\n/g;

# Format image tags.
$content =~ s/<img(.*?)SRC=\"(.*?)\"(.*?)>/<img src=\"$2\"$1$3>/g;
$content =~ s/<img(.*?)ALIGN=\w+(.*?)>/<img$1$2>/g;
$content =~ s/<img(.*?)NAME=\"\w+\"(.*?)>/<img$1$2>/g;
$content =~ s/<img(.*?)HSPACE=\w+(.*?)>/<img$1$2>/g;
$content =~ s/<img(.*?)VSPACE=\w+(.*?)>/<img$1$2>/g;
$content =~ s/<img(.*?)WIDTH=\w+(.*?)>/<img$1$2>/g;
$content =~ s/<img(.*?)HEIGHT=\w+(.*?)>/<img$1$2>/g;
$content =~ s/<img(.*?)BORDER=\w+(.*?)>/<img$1$2>/g;
$content =~ s/<img(.*?)ALT=(\".+\")(.*?)>/<img alt=$2$1$3>/g;
$content =~ s/<img(.*?)(alt\=\"\.+\"){0}(.*?)>/<img alt=\"Imagen.\"$1$3>/g;
$content =~ s/<img(.*?)(\s+?)>/<img$1>/g;
$content =~ s/<img (alt=\".+\")(.\s*)(src=\".+\")(.*)>/<img $3\n     $1 \/>/g;
$content =~ s/(<img.*?>)/\n$1\n/gs;

# Format anchor tags.
$content =~ s/<a(.*?)NAME=\"(.*?)\"(.*?)>/<a name=\"$2\"$1$3>/g;
$content =~ s/<a(.*?)HREF=\"(.*?)\"(.*?)>/<a href=\"$2\"$1$3>/g;
$content =~ s/<a(.*?)CLASS=\".+?\"(.*?)>/<a$1$2>/g;
$content =~ s/<a(.*?)(\s*)>/<a$1>/g;

# Format paragraph tags.
$content =~ s/<p>(.+?)<\/p>/format_pa($1)/ge;
sub format_pa {
    my($text) = @_;
    my $string;
    $initial_tab    = "    "; # Space before first lines.
    $subsequent_tab = "    "; # And the same for the other lines.
    $string = "\n<p>\n";
    $Text::Wrap::columns = 72;
    $string .= wrap($initial_tab, $subsequent_tab,$1);
    $string .= "\n</p>\n";
    return $string;
}

# Format lists.
$content =~ s/<ol><li>/\n<ol>\n<li>/g;
$content =~ s/<\/li><\/ol>/<\/li>\n<\/ol>\n/g;
$content =~ s/<ul><li>/\n<ul>\n<li>/g;
$content =~ s/<\/li><\/ul>/<\/li>\n<\/ul>\n/g;
$content =~ s/<li>(.*?)<\/li>/format_li($1)/ge;
sub format_li {
    my($text) = @_;
    my $string; 
    $initial_tab    = "        "; # Space before first lines.
    $subsequent_tab = "        "; # And the same for the other lines.
    $string = "    <li>\n";
    $Text::Wrap::columns = 72;
    $string .= wrap($initial_tab, $subsequent_tab,$1);
    $string .= "\n    </li>\n";
    return $string;
}

# Format tables.
$content =~ s/<td(.*?)WIDTH=\w+(.*?)>/<td$1$2>/g;
$content =~ s/<td(.*?)COLSPAN=(\w+?)(.*?)>/<td colspan=\"$2\">/g;
$content =~ s/<td(.*?)(\s+?)>/<td$1>/g;
$content =~ s/<table><tr>/\n<table border=\"1\">\n<tr>/g;
$content =~ s/<\/tr><\/table>/<\/tr><\/table>\n/g;
$content =~ s/<tr><td(.*?)>/    <tr>\n<td$1>/g;
$content =~ s/<\/td><\/tr>/<\/td>\n    <\/tr>\n/g;
$content =~ s/<\/td><td(.*?)>/<\/td>\n<td$1>/g;
$content =~ s/<td(.*?)><\/td>/<td$1>\n<\/td>/g;
$content =~ s/<td(.*?)>(.*?)<\/td>/format_td($1,$2)/ges;
sub format_td {
    my($parameters,$text) = @_;
    my $string;
    $initial_tab    = "            "; # Space before first lines.
    $subsequent_tab = "            "; # And the same for the other lines.
    $string = "        <td$1>";
    $Text::Wrap::columns = 72;
    $string .= wrap($initial_tab, $subsequent_tab,$2);
    $string .= "\n        </td>";
    return $string;
}

# Preserve content from "Sections" but do away with the tags.
$content =~ s/<DIV ID=\"Section\d+\".*?>(.*?)<\/DIV>/$1/isg;

# Respect but format footnotes. They have part of the text!
$content =~ s/<DIV ID=\"sdfootnote1\">(.*?)<\/DIV>/\n<h2>\nNotas al pie\n<\/h2>\n\n$1\n/gs;
$content =~ s/<DIV ID=\"sdfootnote[2-9]\d*\">(.*?)<\/DIV>/\n\n$1\n/gs;

# Remove redundant new lines. Ugly, I know, but it works :-)
$content =~ s/<\/p>\n\n+<p>/<\/p>\n\n<p>/sg;
$content =~ s/<\/li>\n+<\/ol>/<\/li>\n<\/ol>/sg;
$content =~ s/<\/li>\n+<\/ul>/<\/li>\n<\/ul>/sg;

# Customize the file for my own use, adding standard header and footer.
# You may change this by whatever you need.
print OUT <<'END';
<?php
include($_SERVER["DOCUMENT_ROOT"] . "/blocks/header.php");
?>
<!-- NO EDITAR ANTES DE ESTA LÍNEA -->

END
print OUT $content;
print OUT <<'END';

<!-- NO EDITAR MÁS ALLÁ DE ESTA LÍNEA -->
<?php
include($_SERVER["DOCUMENT_ROOT"] . "/blocks/footer.php");
?>
END
close(OUT);
close(IN);