HTMLify - Convert text to HTML paragraphs

use strict; use warnings; use diagnostics; use HTML::TreeBuilder; use HTML::Entities; use HTML::Element; sub traverse ; foreach my $file_name (@ARGV) { my $tree = HTML::TreeBuilder->new ; $tree->parse_file($file_name); $tree->dump ; print "\n\nWhere would you like to put the output file? " ; my $output = ; open OUTPUT_FILE, "> $output" or die $! ; select OUTPUT_FILE ; traverse ($tree); $tree = $tree->delete ; close OUTPUT_FILE or die $!; } sub traverse { foreach (@_) { if ($_) { if (ref $_) { print STDERR $_->tag(), "\n\n" ; if ($_->tag() ne "head" && $_->tag() ne "script" && $_->tag() ne "img" && $_->tag() ne "object" && $_->tag() ne "applet") { my @contents = $_->content_list() ; traverse (@contents) ; } if (!$_->parent) { my $s = $_->as_HTML ("",{}) ; $s =~ s/>\n/>/g ; $_ =~ s/'em\s/’em /g ; $_ =~ s/'tis\s/’tis /g ; $_ =~ s/'twas\s/’twas /g ; $_ =~ s/'Twas\s/’Twas /g ; $_ =~ s/'Tis\s/’Tis / ; $_ =~ s/'\s/’ /g ; $_ =~ s/^'/‘/g ; $_ =~ s/(\s)'/$1‘/g ; $_ =~ s/"'/“lsquo;/g ; $_ =~ s/'"/’”/g ; $_ =~ s/\s"/ “/g ; $_ =~ s/^'/‘/g ; $_ =~ s/^"/“/g ; $_ =~ s/"\s/” /g ; $_ =~ s/'$/’/g ; $_ =~ s/"$/”/g ; $_ =~ s/(,|\.)'/$1’/g ; $_ =~ s/(,|\.)"/$1”/g ; $_ =~ s/(\S)'(\S)/$1’$2/g ; print STDERR ($_ , "\n\n"); } } } } ##

##

foreach my $file_name (@ARGV) {
  my $tree = HTML::TreeBuilder->new ; 
  $tree->parse_file($file_name);
  print "\n\nWhere would you like to put the output file for $file_name? " ;
  my $output =  ;
  open OUTPUT_FILE, "> $output" or die $! ;
  $tree = traverse ($tree);
  print OUTPUT_FILE $tree->as_HTML ("","  ",{}) ;
  $tree = $tree->delete ;
  close OUTPUT_FILE or die $!;
}

##

##

sub traverse ($) {
  my $element = $_[0] ;
  if ($element) {
    if (ref $element) {
      print $element->tag(), "\n\n" ;
      if (go_ahead ($element))
      {
        for my $child ($element->content_list()) {
          $child = traverse ($child) ;
        }
      }
    }
    else {            
      print "Processing a string element...\n" ;
      $element = curly_quotes ($element) ;
      print ($element , "\n\n");
    }
  }
  return $element ;
}
##

##

C:\ ... \Programs\HTMLify>htmlify3.pl test.html


Where would you like to put the output file for test.html? output.txt
html

head

body

h1

Processing a string element...
Testing...’

p

Processing a string element...
“lsquo;I’d rather ne’er have been here,’she said,”
 I said. “So what?” I ‘wondered.’

pre

p

Processing a string element...
Jack & Jill went < the hill to > a pail of water. & <>

p

Processing a string element...
“I’m really happy!”
##

##



  
    
    
    HTMLify - Convert text to HTML paragraphs
  
  
    Testing...'
    "'I'd rather ne'er have been here,'she said," I said. "So what?" I 'wondered.'
Jack & Jill went < the hill to > a pail of water. & <>
    "I'm really happy!"
  


##

##

sub traverse ($) {
  my $element = $_[0] ;
  if ($element) {
    if (ref $element) {
      print $element->tag(), "\n\n" ;
      if (go_ahead ($element))
      {
        for my $child ($element->content_list()) {
          $child = traverse ($child) ;
        }
      }
    }
    else {            
      print "Processing a string element...\n" ;
      ${$element} = curly_quotes (${element}) ;
      print ($element , "\n\n");
    }
  }
  return $element ;
}
##

##

{
            my @contents = $_->content_list() ;
            print STDERR "before: @contents\n";
            traverse (@contents) ;
            print STDERR "after: @contents\n";
            $_->detach_content();
            $_->push_content(@contents);
        }

##

##

sub traverse {
  for my $element (@_) {
    if (Scalar::Util::blessed ($element)) {
      if (go_ahead($element)) {
        my @contents = $element->content_list() ;
        print "Before: ", @contents, "\n\n" ;
        traverse(@contents) ;
        print "After: ", @contents, "\n\n" ;
        $element->detach_content() ;
        $element->push_content (@contents) ;
      }
    }
    else {
      print "Processing a string: " ;
      $element = curly_quotes($element) ;
      print $element, "\n\n" ;
    }
  }
}

##

##

sub traverse {
  my $element = $_[0] ;
  if (Scalar::Util::blessed ($element)) {
    if (go_ahead($element)) {
      my @contents = $element->content_list() ;
      print "Before: ", @contents, "\n\n" ;
      for my $child (@contents) {
        traverse ($child) ;
        print "Middle: ", $child, "\n\n" ;
      }
      print "After: ", @contents, "\n\n" ;
    }
  }
  else {
    print "Processing a string: " ;
    $element = curly_quotes($element) ;
    print $element, "\n\n" ;
  }
}

##

##

sub traverse {
  my $element = $_[0] ;
  if (Scalar::Util::blessed (${element})) {
    if (go_ahead(${element})) {
      my @contents = ${element}->content_refs_list() ;
      print "Before: ", @{contents}, "\n\n" ;
      for my $child (@contents) {
        traverse (${${child}}) ;
        print "Middle: ", ${${child}}, "\n\n" ;
      }
      print "After: ", @{contents}, "\n\n" ;
    }
  }
  else {
    print "Processing a string: " ;
    ${element} = curly_quotes(${element}) ;
    print ${element}, "\n\n" ;
  }
}