#!/usr/bin/perl -w
# htmo_to_text
#    VERSION: 2 (13 September 2004)
#    PURPOSE:  Convert HTML files to text files
#              Very simple minded and not very effective
#    INPUT FILES:  Any file in HTML format
#    OUTPUT FILES: Same file in text format

############## LIBRARIES AND PRAGMAS ################
  use strict;

#################### VARIABLES ######################

my $line;                         # explanation of variable 
my $text;                         # explanation of variable 

###################### FILES ########################

   my $html_file = 'C:\temp\Program5\Coordinate system transformation.html';
   open HTML_FILE, "<$html_file" or die "Can't open $html_file: $!\n";

   my $text_file = 'C:\temp\Program5\Coordinate system transformation.txt';
   open TEXT_FILE, ">$text_file" or die "Can't open $text_file: $!\n";

################### MAIN PROGRAM ####################

   $text = Read_html_file();
   Convert_formatting();
   Remove_remaining_tags();
   Write_text_file();

#################### SUBROUTINES ####################

#### READ_HTML_FILE
# Concatenates file into a single string
sub Read_html_file {
   my $text = "";
   $line = <HTML_FILE>;
   while (defined $line) {
      chomp $line;
      $text .= $line;
      $line = <HTML_FILE>;
   }
   return $text;
}

#### CONVERT_FORMATTING
# Converts <p> into two LFs
# Converts <br> into one line feeds
# One could do more, but that's it for now
sub Convert_formatting {
   $text =~ s/<p>/\n\n\n/g;
   $text =~ s/<br>/\n\n/g;
}

#### REMOVE_REMAINING_TAGS
# Delete <...>
sub Remove_remaining_tags {
   $text =~ s/<.+>//g;
}

#### WRITE_TEXT_FILE 
# Write what remains to a text file
# Use max line length
sub Write_text_file {
   print TEXT_FILE $text;
}