Clean up Word documents that have been translated to HTML

#/usr/local/bin/perl    -w
use strict;
 
#############################################################
#                                                           #
#                                                           #
#                                                           #
#                      NOAH SUSSMAN                         #
#                                                           #
#                     clean up word                         #
#                                                           #
#                Created 5/16/01 at 02:33 PM                #
#                                                           #
# Clean up Word documents that have been translated to HTML #    
#                                                           #
#                                                           #
#############################################################
 
@ARGV[0] = "Macintosh HD:NOAH:2001:05-MAY 2001:3-May 15-21:3-Revisions to Corp Site:large number of Word docs:1.2 Services.html" ;
 
$^I=".bk";
 
undef $/ ;           # slurp the whole file into $_
 
while (<>) {
 
	s{<(?!/?(a|b|img|center|p|ul|ol|li|table|td|tr|html|body|head|title))\s*[^>]*>\s*}{}gi;    # Destroy all tags except A, B, IMG, CENTER, P, UL, OL, LI, TABLE, TD, TR, HTML, BODY, HEAD and TITLE
 
	s{<(\w+)>(.*?)<([^$1])>(.*?)<(/$1)>(.*?)<(/$2)>}{<$3>$2<$1>$4<$5>$6<$7>}gi;    #  Fix mis-nested tags, if any.
 
	print $_ ;
 
}