source: trunk/greenstone3-extensions/gs3build/src/org/greenstone/gsdl3/gs3build/util/HTMLTag.java@ 12188

package org.greenstone.gsdl3.gs3build.util;

public class HTMLTag implements HTMLObject
{ String tagtext;
  String name;
  int    docpos;
  int    endpos;

  public static final String HTML_ELEMENT_TYPE = "Element";

  public HTMLTag(String tagtext, int docpos, int endpos)
  { this.tagtext = tagtext;
    this.name    = null;
    this.docpos  = docpos;
    if (this.docpos < 0)
    { this.endpos = -1;
    }
    else
    { this.endpos = endpos;
    }
  }

  public HTMLTag(String tagtext)
  { this.tagtext    = tagtext;
    this.docpos     = -1;
    this.name           = null;
  }

  public String getHTMLType()
  { return HTML_ELEMENT_TYPE;
  }

    public HTMLTag endTag()
    { HTMLTag end;
        String  endtext;

        endtext = "</"+this.xtagName()+">";

        end = new HTMLTag(endtext);
        return end;
    }

    public boolean tagIsOpening()
    { return (this.tagtext.charAt(1) != '/');
    }

    /**
     *  Tells you if a tag is a singleton tag - TODO: complete list
     */
    public boolean tagIsSingleton()
    {   String tagName = this.tagName();
        
        return (tagName.equals("img"));
    }

    /**
     *  @return: the name of a tag including any leading '/'
     */
    public String tagName()
    { int offset, start, end;
        String reply;

        if (this.name != null)
        {   return this.name;
        }
    
        // NB: a starting '<' and an ending '>' are guaranteed
        offset = 1; // skip <
        while (this.tagtext.charAt(offset) <= 32 ||
                     this.tagtext.charAt(offset) == 127)
        { offset ++;
        }
        start = offset;
        while (this.tagtext.charAt(offset) > 32 &&
                     this.tagtext.charAt(offset) != '>' &&
                     this.tagtext.charAt(offset) != 127)
        { offset ++;
        }
        end = offset;

        reply = this.tagtext.substring(start, end);
        reply = reply.toLowerCase();

        this.name   = reply;

        return reply;
    }

    // Returns the name of the tag *minus* any leading / for
    // closing tags
    public String xtagName()
    { String reply;

        reply = this.tagName();
        if (reply == null || reply.length() == 0)
        {   return null;
        }

        if (reply.charAt(0) == '/')
        { if (reply.length() > 1)
            {   reply = reply.substring(1, reply.length());
            }
            else
            { reply = "";
            }
        }
        return reply;
    }

    public String endTagName()
    {   String reply;

        reply = "/" + xtagName();
        return reply;
    }

    public boolean isClosing()
    { String name;

        name    = this.tagName();
        if (name.charAt(0) == '/')
        {   return true;
        }
        return false;
    }

    public int tagLevel()
    { String    name;
        int         level;

        level = 7;

        name = this.xtagName();
        if (name.charAt(0) == 'h')
        { level = Integer.parseInt(name.substring(1, 2));
        }
        
        return level;
    }

    public int startPos()
    {   return this.docpos;
    }

    public int endPos()
    {   return this.endpos;
    }

    static public boolean isTag(String text)
    { if (text.charAt(0) == '<')
        { return true;
        }
        return false;
    }

    public String toString()
    {   return this.tagtext;
    }

    //
    // Returns the value of the identifier given as a string
    //
    public String idValue(String idName)
    {   int offset, start, end;
        int idstart, idend, valuestart, valueend;
        boolean isvalue, isquoted, isstopped;
        char        quotechar;
        String reply;

        // NB: a starting '<' and an ending '>' are guaranteed
        offset = 1; // skip <
        while (this.tagtext.charAt(offset) <= 32 ||
                     this.tagtext.charAt(offset) == 127)
        { offset ++;
        }
        start = offset;
        while (this.tagtext.charAt(offset) > 32 &&
                     this.tagtext.charAt(offset) != '>' &&
                     this.tagtext.charAt(offset) != 127)
        { offset ++;
        }
        end = offset;
        // End of tag name

        if (this.tagtext.charAt(offset) == '>')
        { return null;
        }

        isstopped = false;
        while (!isstopped)
        {   isvalue     = false;
            isquoted    = false;
            valuestart = -1;
            valueend = -1;

            while (this.tagtext.charAt(offset) <= 32 ||
                        this.tagtext.charAt(offset) == 127)
            { offset ++;
            }
            idstart = offset;

            if (this.tagtext.charAt(offset) == '>')
            { break;
            }

            while (this.tagtext.charAt(offset) > 32 &&
                         this.tagtext.charAt(offset) != '=' &&
                         this.tagtext.charAt(offset) != '>' &&
                         this.tagtext.charAt(offset) != 127)
            { offset ++;
            }
            idend = offset;

            while (this.tagtext.charAt(offset) <= 32 ||
                         this.tagtext.charAt(offset) == 127)
            { offset ++;
            }

            if (this.tagtext.charAt(offset) == '=')
            { isvalue = true;
                quotechar = ' ';
                offset ++;              

                while (this.tagtext.charAt(offset) <= 32 ||
                             this.tagtext.charAt(offset) == 127)
                { offset ++;
                }
                if (this.tagtext.charAt(offset) == '"' ||
                        this.tagtext.charAt(offset) == '\'')
                { quotechar = this.tagtext.charAt(offset);
                    isquoted = true;
                    offset ++;
                }
                valuestart = offset;

                while (offset < this.tagtext.length() &&
                             (isquoted && this.tagtext.charAt(offset) != quotechar) ||
                             (!isquoted && this.tagtext.charAt(offset) > 32 &&
                               this.tagtext.charAt(offset) != '>'))
                { offset ++;
                }
                valueend = offset;
            }

            if (offset >= this.tagtext.length() ||
                    this.tagtext.charAt(offset) == '>')
            { isstopped = true;
            }
            offset ++;

            reply = this.tagtext.substring(idstart, idend);
            reply = reply.toLowerCase();

            if (reply.equalsIgnoreCase(idName))
            { if (isvalue)
                { return this.tagtext.substring(valuestart, valueend);
                }
                else
                { return reply;
                }
            }
        }
        return null;
    }
}