Attachment #5457 for bug #7443

View | Details | Raw Unified | Return to bug 7443
Collapse All | Expand All





# elements that change text style
my %elements_text_style = map {; $_ => 1 }
  qw( body font table tr th td big small basefont marquee span p div a ),
;

# elements that insert whitespace

$ok_attributes{body}{$_} = 1 for qw( text bgcolor link alink vlink background );
$ok_attributes{font}{$_} = 1 for qw( color face size );
$ok_attributes{marquee}{$_} = 1 for qw( bgcolor background );
$ok_attributes{table}{$_} = 1 for qw( style bgcolor );
$ok_attributes{td}{$_} = 1 for qw( style bgcolor );
$ok_attributes{th}{$_} = 1 for qw( style bgcolor );
$ok_attributes{tr}{$_} = 1 for qw( style bgcolor );
$ok_attributes{span}{$_} = 1 for qw( style );
$ok_attributes{p}{$_} = 1 for qw( style );
$ok_attributes{a}{$_} = 1 for qw( style );
$ok_attributes{div}{$_} = 1 for qw( style );

sub new {

        $new{style} = $attr->{style};
	my @parts = split(/;/, $new{style});
	foreach (@parts) {
	  if (/^\s*(background)?-?color:\s*(.+)\s*$/i) {
	    my $whcolor = $1 ? 'bgcolor' : 'fgcolor';
	    my $value = lc $2;

	    # BLD: Protecting from the various word values that background-color may contain
	    next if(($whcolor =~ 'bgcolor') && ($value =~ /(?:inherit|initial|unset|transparent|currentColor)/i));

	    if ($value =~ /rgb/) {
	      $value =~ tr/0-9,//cd;
	      my @rgb = split(/,/, $value);
	      $new{$whcolor} = sprintf("#%02x%02x%02x",
                    	                   map { !$_ ? 0 : $_ > 255 ? 255 : $_ }
                                           @rgb[0..2]);
           }
	    else {
	      $new{$whcolor} = name_to_rgb($value);
	    }
	  } 
	  elsif (/^\s*background:\s*(.*)\s*$/i) {
		my $bgvalues = $1;
		# repetitive from above -- need to collapse logic
		if($bgvalues =~ /.*(rgb\(?[^)]+\)?).*/) {
		      $bgvalues =~ tr/0-9,//cd;
		      my @rgb = split(/,/, $bgvalues);
		      $new{'bgcolor'} = sprintf("#%02x%02x%02x",
					 map { !$_ ? 0 : $_ > 255 ? 255 : $_ }
                                           @rgb[0..2]);
		} else {
		     my @bitbkg = split(' ', $1);
		     foreach (@bitbkg) {
			next if ($_ =~ /(?:cover|url|repeat|right|top|bottom|center|left|transparent|initial|inherit|fixed|scroll)/);
	      		$new{'bgcolor'} = name_to_rgb($_);
		     }
		}
	  }
	  elsif (/^\s*([a-z_-]+)\s*:\s*(\S.*?)\s*$/i) {
	    # "display: none", "visibility: hidden", etc.

      }
      elsif ($name eq "bgcolor") {
	# overwrite with hex value, $new{bgcolor} is set below
        $new{'bgcolor'} = name_to_rgb($attr->{bgcolor});
      }
      else {
        # attribute is probably okay

	$self->{min_size} = $new{size};
      }
    }

    push @{ $self->{text_style} }, \%new;
  }
  # explicitly close a tag

sub html_font_invisible {
  my ($self, $text) = @_;


  my $fg = $self->{text_style}[-1]->{fgcolor};
  my $bg = $self->{text_style}[-1]->{bgcolor};
  my $size = $self->{text_style}[-1]->{size};

  # invisibility
  if (substr($fg,-6) eq substr($bg,-6)) {
    $self->put_results(font_low_contrast => 1);
#BLD
dbg("BLD: Low Contrast hit: fg %s, bg %s (%s)\n", $fg, $bg, $text);
    return 1;
  # near-invisibility
  } elsif ($fg =~ /^\#?([0-9a-f]{2})([0-9a-f]{2})([0-9a-f]{2})$/) {

      # 1.25% of HTML spam right now), but please test any changes first
      if ($distance < 12) {
	$self->put_results(font_low_contrast => 1);
#BLD
dbg("BLD: Low Contrast distance: fg %s, bg %s (%s)\n", $fg, $bg, $text);
	return 1;
      }
    }


  
  # invalid color
#BLD
dbg("BLD: checking color: %s bgcolor %s (%s)\n", $fg, $bg, $text);
  if ($fg eq 'invalid' or $bg eq 'invalid') {
    $self->put_results(font_invalid_color => 1);
    return 1;


  my $invisible_for_bayes = 0;


  # NBSP:  UTF-8: C2 A0, ISO-8859-*: A0
  if ($text !~ /^(?:[ \t\n\r\f\x0b]|\xc2\xa0)*\z/s) {
    $invisible_for_bayes = $self->html_font_invisible($text);

Return to bug 7443

Lines 55-61 Link Here

(-)HTML.pm (-10 / +40 lines)
55		55
56	# elements that change text style	56	# elements that change text style
57	my %elements_text_style = map {; $_ => 1 }	57	my %elements_text_style = map {; $_ => 1 }
58	qw( body font table tr th td big small basefont marquee span p div ),	58	qw( body font table tr th td big small basefont marquee span p div a ),
59	;	59	;
60		60
61	# elements that insert whitespace	61	# elements that insert whitespace
Lines 80-91 Link Here
80	$ok_attributes{body}{$_} = 1 for qw( text bgcolor link alink vlink background );	80	$ok_attributes{body}{$_} = 1 for qw( text bgcolor link alink vlink background );
81	$ok_attributes{font}{$_} = 1 for qw( color face size );	81	$ok_attributes{font}{$_} = 1 for qw( color face size );
82	$ok_attributes{marquee}{$_} = 1 for qw( bgcolor background );	82	$ok_attributes{marquee}{$_} = 1 for qw( bgcolor background );
83	$ok_attributes{table}{$_} = 1 for qw( bgcolor );	83	$ok_attributes{table}{$_} = 1 for qw( style bgcolor );
84	$ok_attributes{td}{$_} = 1 for qw( bgcolor );	84	$ok_attributes{td}{$_} = 1 for qw( style bgcolor );
85	$ok_attributes{th}{$_} = 1 for qw( bgcolor );	85	$ok_attributes{th}{$_} = 1 for qw( style bgcolor );
86	$ok_attributes{tr}{$_} = 1 for qw( bgcolor );	86	$ok_attributes{tr}{$_} = 1 for qw( style bgcolor );
87	$ok_attributes{span}{$_} = 1 for qw( style );	87	$ok_attributes{span}{$_} = 1 for qw( style );
88	$ok_attributes{p}{$_} = 1 for qw( style );	88	$ok_attributes{p}{$_} = 1 for qw( style );
		89	$ok_attributes{a}{$_} = 1 for qw( style );
89	$ok_attributes{div}{$_} = 1 for qw( style );	90	$ok_attributes{div}{$_} = 1 for qw( style );
90		91
91	sub new {	92	sub new {
Lines 493-512 Link Here
493	$new{style} = $attr->{style};	494	$new{style} = $attr->{style};
494	my @parts = split(/;/, $new{style});	495	my @parts = split(/;/, $new{style});
495	foreach (@parts) {	496	foreach (@parts) {
496	if (/^\s(background-)?color:\s(.+)\s*$/i) {	497	if (/^\s(background)?-?color:\s(.+)\s*$/i) {
497	my $whcolor = $1 ? 'bgcolor' : 'fgcolor';	498	my $whcolor = $1 ? 'bgcolor' : 'fgcolor';
498	my $value = lc $2;	499	my $value = lc $2;
499		500
		501	# BLD: Protecting from the various word values that background-color may contain
		502	next if(($whcolor =~ 'bgcolor') && ($value =~ /(?:inherit\|initial\|unset\|transparent\|currentColor)/i));
		503
500	if ($value =~ /rgb/) {	504	if ($value =~ /rgb/) {
501	$value =~ tr/0-9,//cd;	505	$value =~ tr/0-9,//cd;
502	my @rgb = split(/,/, $value);	506	my @rgb = split(/,/, $value);
503	$new{$whcolor} = sprintf("#%02x%02x%02x",	507	$new{$whcolor} = sprintf("#%02x%02x%02x",
504	map { !$_ ? 0 : $_ > 255 ? 255 : $_ }	508	map { !$_ ? 0 : $_ > 255 ? 255 : $_ }
505	@rgb[0..2]);	509	@rgb[0..2]);
506	}	510	}
507	else {	511	else {
508	$new{$whcolor} = name_to_rgb($value);	512	$new{$whcolor} = name_to_rgb($value);
509	}	513	}
		514	}
		515	elsif (/^\sbackground:\s(.)\s$/i) {
		516	my $bgvalues = $1;
		517	# repetitive from above -- need to collapse logic
		518	if($bgvalues =~ /.(rgb\(?[^)]+\)?)./) {
		519	$bgvalues =~ tr/0-9,//cd;
		520	my @rgb = split(/,/, $bgvalues);
		521	$new{'bgcolor'} = sprintf("#%02x%02x%02x",
		522	map { !$_ ? 0 : $_ > 255 ? 255 : $_ }
		523	@rgb[0..2]);
		524	} else {
		525	my @bitbkg = split(' ', $1);
		526	foreach (@bitbkg) {
		527	next if ($_ =~ /(?:cover\|url\|repeat\|right\|top\|bottom\|center\|left\|transparent\|initial\|inherit\|fixed\|scroll)/);
		528	$new{'bgcolor'} = name_to_rgb($_);
		529	}
		530	}
510	}	531	}
511	elsif (/^\s([a-z_-]+)\s:\s(\S.?)\s*$/i) {	532	elsif (/^\s([a-z_-]+)\s:\s(\S.?)\s*$/i) {
512	# "display: none", "visibility: hidden", etc.	533	# "display: none", "visibility: hidden", etc.
Lines 516-522 Link Here
516	}	537	}
517	elsif ($name eq "bgcolor") {	538	elsif ($name eq "bgcolor") {
518	# overwrite with hex value, $new{bgcolor} is set below	539	# overwrite with hex value, $new{bgcolor} is set below
519	$attr->{bgcolor} = name_to_rgb($attr->{bgcolor});	540	$new{'bgcolor'} = name_to_rgb($attr->{bgcolor});
520	}	541	}
521	else {	542	else {
522	# attribute is probably okay	543	# attribute is probably okay
Lines 530-535 Link Here
530	$self->{min_size} = $new{size};	551	$self->{min_size} = $new{size};
531	}	552	}
532	}	553	}
		554
533	push @{ $self->{text_style} }, \%new;	555	push @{ $self->{text_style} }, \%new;
534	}	556	}
535	# explicitly close a tag	557	# explicitly close a tag
Lines 544-549 Link Here
544	sub html_font_invisible {	566	sub html_font_invisible {
545	my ($self, $text) = @_;	567	my ($self, $text) = @_;
546		568
		569
547	my $fg = $self->{text_style}[-1]->{fgcolor};	570	my $fg = $self->{text_style}[-1]->{fgcolor};
548	my $bg = $self->{text_style}[-1]->{bgcolor};	571	my $bg = $self->{text_style}[-1]->{bgcolor};
549	my $size = $self->{text_style}[-1]->{size};	572	my $size = $self->{text_style}[-1]->{size};
Lines 553-558 Link Here
553	# invisibility	576	# invisibility
554	if (substr($fg,-6) eq substr($bg,-6)) {	577	if (substr($fg,-6) eq substr($bg,-6)) {
555	$self->put_results(font_low_contrast => 1);	578	$self->put_results(font_low_contrast => 1);
		579	#BLD
		580	dbg("BLD: Low Contrast hit: fg %s, bg %s (%s)\n", $fg, $bg, $text);
556	return 1;	581	return 1;
557	# near-invisibility	582	# near-invisibility
558	} elsif ($fg =~ /^\#?([0-9a-f]{2})([0-9a-f]{2})([0-9a-f]{2})$/) {	583	} elsif ($fg =~ /^\#?([0-9a-f]{2})([0-9a-f]{2})([0-9a-f]{2})$/) {
Lines 575-580 Link Here
575	# 1.25% of HTML spam right now), but please test any changes first	600	# 1.25% of HTML spam right now), but please test any changes first
576	if ($distance < 12) {	601	if ($distance < 12) {
577	$self->put_results(font_low_contrast => 1);	602	$self->put_results(font_low_contrast => 1);
		603	#BLD
		604	dbg("BLD: Low Contrast distance: fg %s, bg %s (%s)\n", $fg, $bg, $text);
578	return 1;	605	return 1;
579	}	606	}
580	}	607	}
Lines 582-587 Link Here
582		609
583		610
584	# invalid color	611	# invalid color
		612	#BLD
		613	dbg("BLD: checking color: %s bgcolor %s (%s)\n", $fg, $bg, $text);
585	if ($fg eq 'invalid' or $bg eq 'invalid') {	614	if ($fg eq 'invalid' or $bg eq 'invalid') {
586	$self->put_results(font_invalid_color => 1);	615	$self->put_results(font_invalid_color => 1);
587	return 1;	616	return 1;
Lines 741-746 Link Here
741		770
742	my $invisible_for_bayes = 0;	771	my $invisible_for_bayes = 0;
743		772
		773
744	# NBSP: UTF-8: C2 A0, ISO-8859-*: A0	774	# NBSP: UTF-8: C2 A0, ISO-8859-*: A0
745	if ($text !~ /^(?:[ \t\n\r\f\x0b]\|\xc2\xa0)*\z/s) {	775	if ($text !~ /^(?:[ \t\n\r\f\x0b]\|\xc2\xa0)*\z/s) {
746	$invisible_for_bayes = $self->html_font_invisible($text);	776	$invisible_for_bayes = $self->html_font_invisible($text);