Abbreviates a String which can contain html tags. Html tags are not counted in String length. It also try to handle open tags and html entities. Adopted from http://www.displaytag.org/1.2/displaytag/apidocs/org/displaytag/util/HtmlTagUtil.html

abbreviateHtmlString.vm

#*
Abbreviates a String which can contain html tags. Html tags are not counted in String length. It also try to handle open tags and html entities.

Adopted from http://www.displaytag.org/1.2/displaytag/apidocs/org/displaytag/util/HtmlTagUtil.html

@param str full String. null is handled by returning null
@param maxLength maximum number of characters (excluding tags)
@param byNumberOfWords if true maxLength will be the number of words returned, elsewhere will represent the number of characters.
@return abbreviated String
*#
#macro (abbreviateHtmlString $str, $maxLength, $byNumberOfWords)
  #if ($str == "" || $str.length() <= $maxLength )
    $str
  #else
    #set ($sz = $str.length())
    #set ($szMinusOne = $sz - 1)
    #set ($newStr = "")
    #set ($inTag = false)
    #set ($inTagName = false)
    #set ($endingTag = false)
    #set ($count = 0)
    #set ($chopped = false)
    #set ($choppedPos = 0) 
    #set ($entityChars = 0)

    #set ($currentTag = "")

    #set ($openTags = [])

    #foreach ($i in [0..$szMinusOne])
      #if ($count >= $maxLength) 
        #set ($chopped = true) 
        #set ($choppedPos = $i) 
        #break 
      #end

      #set ($c = $str.charAt($i))

      #if ($c == "<")
        #set ($inTag = true)
        #set ($inTagName = true)
      #elseif ($inTag)
        #if ($inTagName && $c == "/")        
          #if ($currentTag.length() == 0)
            ## Ending tag found
            #set ($endingTag = true)
          #else
            ## Empty tag, reset and don't save
            #set ($inTagName = false)            
          #end
          #set ($currentTag = "")
        #elseif ($inTagName && ($c == " " || $c == ">"))
          #set ($inTagName = false)

          #if (!$endingTag)
            #set ($_void = $openTags.add(0, $currentTag))
          #else
            #set ($_void = $openTags.remove($currentTag))
          #end

          #set ($currentTag = "")          
          #set ($endingTag = false)
          
          #if ($c == ">")
            #set ($inTag = false)            
          #end
        #elseif ($c == ">")
          #set ($inTag = false)
        #elseif ($inTagName)
          #set ($currentTag = "${currentTag}${c}")
        #end
      #else
        #if ($byNumberOfWords)
          #if ($c.matches("\s"))
            #set ($count = $count + 1)
          #end
        #else
          ## Handle entities
          #if ($c == "&")
            #set ($entityChars = 1)
          #elseif ($entityChars == 0)
            #set ($count = $count + 1)
          #else
            ## End entity
            #if ($entityChars > 0 && $c == ";")
              #set ($entityChars = 0)
              #set ($count = $count + 1)
            #else
              #set ($entityChars = $entityChars + 1)
            #end

            ## Assume an unescaped & if entity doesn't close after max 5 chars
            #if ($entityChars > 5)
              #set ($count = $count + $entityChars)
              #set ($entityChars = 0)
            #end
          #end
        #end
      #end

      #if ($inTag || (!$byNumberOfWords || $count < $maxLength))
        #set ($newStr = "${newStr}${c}")
      #end      
    #end

    #if ($chopped)
      #set ($newStr = "${newStr}...")
    #end

    #if ($openTags.size() > 0)
      ## Quickly fixes closed tags
      #set ($remainingToken = $str.substring($choppedPos))
      #set ($openTagsSize = $openTags.size())
      #set ($openTagsSizeMinusOne = $openTagsSize - 1)

      #foreach ($j in [0..$openTagsSizeMinusOne])
        #set ($closingTag = $openTags.get($j))
        #set ($closingTag = "</${closingTag}>")

        ## We only add closing tags that exists in the original String, so we don't have to understand
        ## html/xhtml differences and keep a list of html unclosed tags
        #if ($remainingToken.indexOf($closingTag) > -1)        
          #set ($newStr = "${newStr}${closingTag}")
        #end
      #end
    #end

    $newStr
  #end
#end