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

package org.greenstone.gsdl3.gs3build.util;

import java.io.*;
import java.util.*;

import org.greenstone.gsdl3.gs3build.doctypes.DocumentLoader;

public class HTMLEntity
{ String  name;
  int     code;
  static Vector entities = null;
  static Vector nmapping = null;

  public HTMLEntity(String name, int code)
  { int start, end, at;
    HTMLEntity compareEntity;

    if (HTMLEntity.entities == null)
    { HTMLEntity.entities = new Vector();
    }
    if (HTMLEntity.nmapping == null)
    { HTMLEntity.nmapping = new Vector();
    }

    this.name   = name;
    this.code   = code;

    entities.addElement(this);
    start = 0;
    end   = nmapping.size();
    while (start != end)
    { at = (start + end) / 2;
    
      compareEntity = 
    (HTMLEntity) entities.elementAt(((Integer) nmapping.elementAt(at)).intValue());
      if (compareEntity.code > code)
      { end = at;
      }
      else
      { start = at + 1;
      }
    }
    if (end == nmapping.size())
    { nmapping.addElement(new Integer(nmapping.size()));
    }
    else
    { nmapping.insertElementAt(new Integer(nmapping.size()), start);
    }
  }

  public String toString()
  { Character c;

    c = new Character((char) this.code);
    return c.toString();
  }

  public static String codeString(char c)
  { StringBuffer    reply;
    String        codestring;

    if (c < 128 || c == '&' || c == '<' || c == '>')
    { return null;
    }
    codestring = Integer.toString((int) c);
    reply      = new StringBuffer("&#");
    reply.append(codestring);
    reply.append(";");

    return reply.toString();
  }

  public static String nameCodeString(char c)
  { int start, end, at, compare;
    HTMLEntity compareEntity;

    start = 0;
    end     = nmapping.size();
    while (start != end)
    { at    = (start + end) / 2;
      
      compareEntity = 
    (HTMLEntity) entities.elementAt(((Integer) nmapping.elementAt(at)).intValue());
      if (compareEntity.code == (int) c)
      { StringBuffer reply;

        reply = new StringBuffer("&");
    reply.append(compareEntity.name);
    reply.append(";");
    return reply.toString();
      }
      else if (compareEntity.code < c)
      { end = at;
      }
      else
      { start   = at + 1;
      }
    }
    return codeString(c);
  }

  public static void getEntities(String source)
  { HTMLEntity  newentity;
    int         pos;
    int                 start;
    String          name;
    int                 code;
    
    if (source == null)
    { return;
    }

    start = 0;
    while (start < source.length())
    { pos = start;
      while (source.charAt(pos) != ':')
      { pos ++;
      }
      name = source.substring(start, pos);

      // skip colon and reset
      pos ++;
      start = pos;

      while(pos < source.length() && source.charAt(pos) > ' ')
      { pos ++;
      }
      code = Integer.parseInt(source.substring(start, pos));

      newentity = new HTMLEntity(name, code);
      
      while ( pos < source.length() && source.charAt(pos) < ' ')
      { pos ++;
      }
      start = pos;
    }
  }

  public static void getEntities(InputStream in)
  { getEntities(DocumentLoader.getAsString(in));
  }

  public static void getEntities(File file)
  { FileInputStream in;

    try
    { in = new FileInputStream(file);
      getEntities(in);
      in.close();
    }
    catch (IOException io)
    { System.out.println("No entities");
    }
  }

  public static void getEntities()
  { ClassLoader loader;
    InputStream in;

    in = ClassLoader.getSystemResourceAsStream("org\\greenstone\\gsdl3\\gs3build\\util\\Entities.map");
    if (in == null)
    { System.out.println("Unable to load Entities.map from org.greenstone.gsdl3.gs3build.util");
    }
    getEntities(in);
  }

  public static String encodeText(String source, boolean named)
  { StringBuffer reply;
    int          start,at;

    if (source == null)
    { return null;
    }

    if (HTMLEntity.entities == null)
    {   getEntities();
    }

    reply   = new StringBuffer();
    at      = 0;
    start   = 0;
    while (at < source.length())
    { if (source.charAt(at) > 128)
      { if (at != start)
    { reply.append(source.substring(start, at));
    }

        if (named)
    { reply.append(nameCodeString(source.charAt(at)));
    }
    else
    { reply.append(codeString(source.charAt(at)));
    }
    start   = at + 1;
      }
      at ++;
    }
        
    return reply.toString();
  }

  public static String decodeText(String source)
  { int start, end, loff, roff;
    int sstart, send, test;
    int startoff, endoff, baseoff;
    boolean matched;
    String  tail;
    StringBuffer reply;

    if (source == null)
    { return null;
    }

    if (HTMLEntity.entities == null)
    {   getEntities();
    }

    reply   = new StringBuffer();
    baseoff = 0;

    tail = source;
    do
    { // get next ampersand
        startoff = tail.indexOf('&');
      if (startoff < 0 || startoff == tail.length() - 1)
      { break;
      }

      // if it's followed by a hash, evaluate as a number
      if (tail.charAt(startoff + 1) == '#')
      { if (startoff < tail.length() - 2 &&
                tail.charAt(startoff+2) >= '0' &&
            tail.charAt(startoff+2) <= '9')
        { int code;

            loff = startoff + 2;
            while (tail.charAt(loff) >= '0' &&
                        tail.charAt(loff) <= '9')
            { loff ++;
          }

          code = Integer.parseInt(tail.substring(startoff+2, loff));
            reply.append(tail.substring(0, startoff));
                    reply.append(new Character((char) code));

          if (tail.charAt(loff) == ';')
          { loff ++;
          }
          loff = loff - startoff - 1;
        }
        else
        { loff = 0;
            reply.append(tail.substring(0, startoff+1));
        }
      }
      else
      { start = 0;
        end   = HTMLEntity.entities.size();
        loff  = 0;

    while (startoff+1+loff < tail.length() && start != end)
    { sstart = start;
      send   = end;
          test   = (sstart + send) >> 1;

      matched   = false;
              
      roff  = startoff + 1 + loff;

      while (sstart != send)
      { test = (sstart + send) >> 1;

        if (tail.charAt(roff) < ((HTMLEntity) HTMLEntity.entities.elementAt(test)).name.charAt(loff))
        { send = test;
        }
        else if (tail.charAt(roff) > ((HTMLEntity) HTMLEntity.entities.elementAt(test)).name.charAt(loff))
        { sstart    = test + 1;
        }
        else
        { break;
        }
      }

      //                    System.out.println(sstart+">"+send+":"+loff);

      if (sstart != send) // found a match
      { sstart = test;
        while (sstart >= start &&
           tail.charAt(roff) ==
           ((HTMLEntity) HTMLEntity.entities.elementAt(sstart)).name.charAt(loff))
        { sstart --;
        }
        sstart ++;

        send = test + 1;
        while (send < end &&
           tail.charAt(roff) ==
           ((HTMLEntity) HTMLEntity.entities.elementAt(send)).name.charAt(loff))
        { send ++;
        }

//                      System.out.println(sstart+"!"+send+":"+loff);

        start = sstart;
        end = send;
      }
      else
      { loff    = 0;
        reply.append(tail.substring(0, startoff+1));
        break;
      }

//                  System.out.println(start+">"+end+":"+loff);
      
      loff ++;
      if (end == start + 1 &&
          ((HTMLEntity) HTMLEntity.entities.elementAt(test)).name.length() == loff)
      { reply.append(tail.substring(0, startoff));
        reply.append(((HTMLEntity) HTMLEntity.entities.elementAt(test)).toString());
        if (roff < tail.length() - 1 && tail.charAt(roff+1) == ';')
        { loff ++;
        }
        break;
      }
    }
      }

      tail = tail.substring(startoff + 1 + loff);
    } while (tail != null && tail.length() > 0);
    
    if (tail != null)
    { reply.append(tail);
    }
    return reply.toString();
  }

  public static void main(String args[])
  { String reply;

    getEntities();
    
    reply = decodeText(args[0]);
    System.out.println(reply);
  }
}