package com.swabunga.spell.engine;



import java.io.*;

import java.util.*;



/**

 * Another implementation of SpellDictionary that doesn't cache any words in memory. Avoids the huge

 * footprint of SpellDictionaryHashMap at the cost of relatively minor latency. A future version

 * of this class that implements some caching strategies might be a good idea in the future, if there's any

 * demand for it.

 *

 * This implementation requires a special dictionary file, with "code*word" lines sorted by code.

 * It's using a dichotomy algorithm to search for words in the dictionary

 *

 * @author Damien Guillaume

 * @version 0.1

 */

public class SpellDictionaryDichoDisk extends SpellDictionaryASpell {



  /** Holds the dictionary file for reading*/

  private RandomAccessFile dictFile = null;

  

  /** dictionary and phonetic file encoding */

  private String encoding = null;



  /**

   * Dictionary Convienence Constructor.

   */

  public SpellDictionaryDichoDisk(File wordList)

    throws FileNotFoundException, IOException {

    super((File) null);

    dictFile = new RandomAccessFile(wordList, "r");

  }



  /**

   * Dictionary Convienence Constructor.

   */

  public SpellDictionaryDichoDisk(File wordList, String encoding)

    throws FileNotFoundException, IOException {

    super((File) null);

    this.encoding = encoding;

    dictFile = new RandomAccessFile(wordList, "r");

  }



  /**

  * Dictionary constructor that uses an aspell phonetic file to

  * build the transformation table.

  */

  public SpellDictionaryDichoDisk(File wordList, File phonetic)

    throws FileNotFoundException, IOException {

    super(phonetic);

    dictFile = new RandomAccessFile(wordList, "r");

  }

  

  /**

  * Dictionary constructor that uses an aspell phonetic file to

  * build the transformation table.

  */

  public SpellDictionaryDichoDisk(File wordList, File phonetic, String encoding)

    throws FileNotFoundException, IOException {

    super(phonetic, encoding);

    this.encoding = encoding;

    dictFile = new RandomAccessFile(wordList, "r");

  }

  

  /**

   * Add a word permanantly to the dictionary (and the dictionary file).

   * not implemented !

   */

  public void addWord(String word) {

    System.err.println("error: addWord is not implemented for SpellDictionaryDichoDisk");

  }



  /**

    * Search the dictionary file for the words corresponding to the code

    * within positions p1 - p2

    */

   private LinkedList dichoFind(String code, long p1, long p2) throws IOException {

     //System.out.println("dichoFind("+code+","+p1+","+p2+")");

     long pm = (p1 + p2) / 2;

    dictFile.seek(pm);

    String l;

    if (encoding == null)

      l = dictFile.readLine();

    else

      l = dictReadLine();

    pm = dictFile.getFilePointer();

    if (encoding == null)

      l = dictFile.readLine();

    else

      l = dictReadLine();

    long pm2 = dictFile.getFilePointer();

    if (pm2 >= p2)

      return(seqFind(code, p1, p2));

    int istar = l.indexOf('*');

    if (istar == -1)

      throw new IOException("bad format: no * !");

    String testcode = l.substring(0, istar);

    int comp = code.compareTo(testcode);

    if (comp < 0)

      return(dichoFind(code, p1, pm-1));

    else if (comp > 0)

      return(dichoFind(code, pm2, p2));

    else {

      LinkedList l1 = dichoFind(code, p1, pm-1);

      LinkedList l2 = dichoFind(code, pm2, p2);

      String word = l.substring(istar+1);

      l1.add(word);

      l1.addAll(l2);

      return(l1);

    }

   }

   

   private LinkedList seqFind(String code, long p1, long p2) throws IOException {

     //System.out.println("seqFind("+code+","+p1+","+p2+")");

     LinkedList list = new LinkedList();

    dictFile.seek(p1);

    while (dictFile.getFilePointer() < p2) {

      String l;

      if (encoding == null)

        l = dictFile.readLine();

      else

        l = dictReadLine();

      int istar = l.indexOf('*');

      if (istar == -1)

        throw new IOException("bad format: no * !");

      String testcode = l.substring(0, istar);

      if (testcode.equals(code)) {

        String word = l.substring(istar+1);

        list.add(word);

      }

    }

    return(list);

   }

   

   /**

     * Read a line of dictFile with a specific encoding

     */

   private String dictReadLine() throws IOException {

     int max = 255;

     byte b=0;

    byte[] buf = new byte[max];

    int i=0;

     try {

       for (; b != '\n' && b != '\r' && i