sub html_tests()

in lib/Mail/SpamAssassin/HTML.pm [844:920]

• 61 lines of code
• 37 McCabe index (conditional complexity)

sub html_tests {
  my ($self, $tag, $attr, $num) = @_;

  if ($tag eq "font" && exists $attr->{face}) {
    # Fixes from Bug 5956, 7312
    # Examples seen in ham:
    #  "Tahoma", Verdana, Arial, sans-serif
    #  'Montserrat', sans-serif
    #  Arial,Helvetica,Sans-Serif;
    #  .SFUIDisplay
    #  hirakakupro-w3
    # TODO: There's still the problem completely foreign unicode strings,
    # probably this rule should be deprecated.
    if ($attr->{face} !~ /^\s*["'.]?[a-z ][a-z -]*[a-z]\d?["']?(?:,\s*["']?[a-z][a-z -]*[a-z]\d?["']?)*;?$/i) {
      $self->put_results(font_face_bad => 1);
    }
  }
  if ($tag eq "img" && exists $self->{inside}{a} && $self->{inside}{a} > 0) {
    my $uri = $self->{anchor_last};
    utf8::encode($uri) if $self->{SA_encode_results};
    $self->{uri}->{$uri}->{anchor_text}->[-1] .= "<img>\n";
    $self->{anchor}->[-1] .= "<img>\n";
  }

  if ($tag eq "img" && exists $attr->{width} && exists $attr->{height}) {
    my $width = 0;
    my $height = 0;
    my $area = 0;

    # assume 800x600 screen for percentage values
    if ($attr->{width} =~ /^(\d+)(\%)?$/) {
      $width = $1;
      $width *= 8 if (defined $2 && $2 eq "%");
    }
    if ($attr->{height} =~ /^(\d+)(\%)?$/) {
      $height = $1;
      $height *= 6 if (defined $2 && $2 eq "%");
    }
    # guess size
    $width = 200 if $width <= 0;
    $height = 200 if $height <= 0;
    if ($width > 0 && $height > 0) {
      $area = $width * $height;
      $self->{image_area} += $area;
    }
  }
  if ($tag eq "form" && exists $attr->{action}) {
    $self->put_results(form_action_mailto => 1) if $attr->{action} =~ /mailto:/i
  }
  if ($tag eq "object" || $tag eq "embed") {
    $self->put_results(embeds => 1);
  }

  # special text delimiters - <a> and <title>
  if ($tag eq "a") {
    my $uri = $self->{anchor_last} =
      (exists $attr->{href} ? $self->canon_uri($attr->{href}) : "");
    utf8::encode($uri) if $self->{SA_encode_results};
    push(@{$self->{uri}->{$uri}->{anchor_text}}, '');
    push(@{$self->{anchor}}, '');
  }
  if ($tag eq "title") {
    $self->{title_index}++;
    $self->{title}->[$self->{title_index}] = "";
  }

  if ($tag eq "meta" &&
      exists $attr->{'http-equiv'} &&
      exists $attr->{content} &&
      $attr->{'http-equiv'} =~ /Content-Type/i &&
      $attr->{content} =~ /\bcharset\s*=\s*["']?([^"']+)/i)
  {
    $self->{charsets} .= exists $self->{charsets} ? " $1" : $1;
  }

  # todo: capture URI from meta refresh tag
}