view release on metacpan or  search on metacpan

lib/Alvis/HTML.pm  view on Meta::CPAN

    #           3. Pick the leftmost '>' before the start of the next
    #              tag as the end of the tag.
    #           4. Remove all tags.               
    #
    if ($self->{alvisKeep})
    { 
	$html=~s/<\/?(?:(?i)a|frame|iframe|h[1-6]|p|div|dl|ul|ol|table|li|dd|dt|th|td|caption)(?=\W)/\0/sgo;
    }
    if ($self->{alvisRemove})
    {
	$html=~s/<\/?(?:(?i)tr|blockquote|hr|br|dir|menu|form|fieldset|legend|label|input|select|option|textarea|isindex|noframes|frameset|tfoot|body|tbody|html|head|abbr|acronym|address|applet|area|b|base|basefont|bdo|big|button|center|cite|code|col|colgro...
    }
    if ($self->{obsolete})
    {
	$html=~s/<\/?(?:(?i)header|nextid|section|listing|xmp|plaintext)(?=\W)/\0/sgo;
    }
    if ($self->{proprietary})
    {
	$html=~s/<\/?(?:(?i)align|blink|embed|ilayer|keygen|layer|multicol|noembed|nolayer|nosave|spacer|inlineinput|sound|audioscope|blackface|animate|bgsound|comment|marquee|xml|o:p|csaction|csactions|csactiondict|csscriptdict|csactionitem|csobj|wbr|nobr|...
    }
    if ($self->{xhtml})

lib/Alvis/HTML.pm  view on Meta::CPAN

    $html=~s/(?<=\0).*?>//sgo;
    $html=~s/\0/ /go;
    
    # We have removed those tags we wanted to now
    
    # If we have some tags left, do some fixing 
    if (!$self->{alvisKeep}||!$self->{alvisRemove}||!$self->{obsolete}||
	!$self->{proprietary}||!$self->{xhtml}||!$self->{wml})
    {
	# Often we have <TAG ... </TAG>. Fix that.
	$html=~s/(<\/?(?:(?i)a|frame|iframe|h[1-6]|p|div|dl|ul|ol|table|li|dd|dt|th|td|caption|tr|blockquote|hr|br|dir|menu|form|fieldset|legend|label|input|select|option|textarea|isindex|noframes|frameset|tfoot|body|tbody|html|head|abbr|acronym|address|app...
	$html=~s/(?<=\0)([^>]*?)(?=\0)/$1>/sgo;
	$html=~s/(?<=\0)([^\0>]*?)$/$1>/sgo;
	$html=~s/\0/ /go;
    }

    # Alvis needs some finer tuning
    if (!$self->{alvisKeep})
    { 
	# Fix attributes of interest
	$html=~s/(<a\W[^>]*?href\s*=\s*)([\"\'])(\S*?)(\s.*?)?>/$self->_fix_attr($1,$2,$3,$4)/isgoe;