package org.greenstone.server;

import java.io.BufferedReader;
import java.io.BufferedWriter;
import java.io.FileReader;
import java.io.FileWriter;
import java.io.Serializable;
import java.sql.Connection;
import java.sql.DriverManager;
import java.sql.Statement;
import java.util.ArrayList;
import java.util.HashMap;
import java.util.HashSet;

import com.sun.org.apache.bcel.internal.generic.NEW;

public class GazetteerTrieType5 implements Serializable
{
	private static final long serialVersionUID = -959184305931535981L;

	int _nameCount = 0;

	HashMap<String, Integer> _gazetteer = new HashMap<String, Integer>();

	/**
	 * Default constructor
	 */
	public GazetteerTrieType5()
	{
	}

	/**
	 * Contructor that takes a filename to generate the gazetteer
	 * 
	 * @param filename
	 *            is the name of the file to use to generate the gazetteer with
	 */
	public GazetteerTrieType5(String filename)
	{
		System.out.println("Loading " + filename + " as gazetteer");
		try
		{
			BufferedReader gazetteerFile = new BufferedReader(new FileReader(filename));
			String line = "";

			ArrayList<String> allPlaceNames = new ArrayList<String>();

			int count = 0;
			while ((line = gazetteerFile.readLine()) != null)
			{
				if (count++ % 10000 == 0)
				{
					System.out.println(count + " entries loaded");
				}
				// The file is tab seperated so split it by tabs
				String[] columns = line.split("\t");

				ArrayList<String> placeNames = new ArrayList<String>();

				String mainPlaceName = GazetteerHelper.getMainPlaceName(columns);
				if (mainPlaceName != null)
				{
					placeNames.add(mainPlaceName);
				}

				ArrayList<String> alternatePlaceNames = GazetteerHelper.getAlternativePlaceNames(columns);
				if (alternatePlaceNames != null)
				{
					placeNames.addAll(alternatePlaceNames);
				}

				allPlaceNames.addAll(placeNames);

				// Add the place names to the trie
				for (String placeName : placeNames)
				{
					this.addPlaceName(placeName);
				}
			}
		}
		catch (Exception ex)
		{
			ex.printStackTrace();
		}
		removeAmbiguousPlaceNames();
	}

	/**
	 * Removes a place name from the trie
	 * 
	 * @param placeName
	 *            is the place name to be removed
	 * @return true if sucessful and false if the place name did not exist
	 */

	public void removePlaceName(String placeName)
	{
		_gazetteer.remove(placeName);
	}

	/**
	 * Adds a place name to the trie
	 * 
	 * @param placeName
	 *            is the place name to add
	 */
	public void addPlaceName(String placeName)
	{
		String[] words = placeName.split(" ");

		for (int i = 0; i < (words.length - 1); i++)
		{
			_gazetteer.put(words[i], 0);
		}
		_gazetteer.put(words[words.length-1], 1);
	}

	/**
	 * Checks to see if a place name exists in the trie
	 * 
	 * @param placeName
	 *            is the place name to check
	 * @return 1 if the place name exists 0 if the place name does not exist but
	 *         there might be a match further down the trie -1 if the place name
	 *         does not exist and the trie is at a dead end
	 */
	public int checkPlaceName(String placeName)
	{
		String[] words = placeName.split(" ");
		
		for(int i = 0; i < (words.length - 1); i++)
		{
			if(_gazetteer.get(words[i]) == null)
			{
				return -1;
			}
		}
		
		if (_gazetteer.get(words[words.length-1]) == null)
		{
			return -1;
		}
		else
		{
			return _gazetteer.get(words[words.length-1]) ;
		}
	}

	/**
	 * Removes place names that are unlikely to be meant as place names in a
	 * given text
	 * 
	 * @param gazetteer
	 *            is the gazetteer to remove the place names from
	 */
	public void removeAmbiguousPlaceNames()
	{
		removePlaceName("are");
		removePlaceName("is");
		removePlaceName("over");
		removePlaceName("at");
		removePlaceName("of");
		removePlaceName("to");
		removePlaceName("rule");
		removePlaceName("time");
		removePlaceName("real");
		removePlaceName("national");
		removePlaceName("early");
		removePlaceName("by");
		removePlaceName("as");
		removePlaceName("eastern");
		removePlaceName("western");
		removePlaceName("southern");
		removePlaceName("northern");
		removePlaceName("east");
		removePlaceName("west");
		removePlaceName("south");
		removePlaceName("north");
		removePlaceName("this");
		removePlaceName("between");
		removePlaceName("many");
		removePlaceName("strong");
		removePlaceName("economy");
		removePlaceName("mall");
		removePlaceName("they");
		removePlaceName("do");
		removePlaceName("image");
		removePlaceName("republic");
		removePlaceName("section");
		removePlaceName("dollar");
		removePlaceName("index");
		removePlaceName("day");
		removePlaceName("council");
		removePlaceName("use");
		removePlaceName("log");
		removePlaceName("logo");
		removePlaceName("best");
		removePlaceName("go");
		removePlaceName("portal");
		removePlaceName("list");
		removePlaceName("english");
		removePlaceName("page");
		removePlaceName("see");
		removePlaceName("ocean");
		removePlaceName("island");
		removePlaceName("x");
		removePlaceName("country");
		removePlaceName("colony");
		removePlaceName("christian");
		removePlaceName("black");
		removePlaceName("independence");
		removePlaceName("war");
		removePlaceName("no");
		removePlaceName("continental");
		removePlaceName("force");
		removePlaceName("reform");
		removePlaceName("rush");
		removePlaceName("read");
		removePlaceName("none");
		removePlaceName("justice");
		removePlaceName("font");
		removePlaceName("u");
		removePlaceName("y");
		removePlaceName("normal");
		removePlaceName("center");
		removePlaceName("date");
		removePlaceName("story");
		removePlaceName("union");
		removePlaceName("supreme");
		removePlaceName("house");
		removePlaceName("court");
		removePlaceName("data");
		removePlaceName("energy");
		removePlaceName("white");
		removePlaceName("universal");
		removePlaceName("protection");
		removePlaceName("great");
		removePlaceName("star");
		removePlaceName("banner");
		removePlaceName("capital");
		removePlaceName("much");
		removePlaceName("sidney");
		removePlaceName("media");
		removePlaceName("protection");

		addPlaceName("United States");
	}
}

// TreeMap<Character, Integer> charCount = new TreeMap<Character, Integer>();
// int upperCount = 0;
// int lowerCount = 0;
// int numbers = 0;
// int arabicCharacters = 0;
// int chineseCharacters = 0;

// if (!charCount.containsKey(placeName.charAt(placeName.length() - 1)))
// {
// charCount.put(placeName.charAt(placeName.length() - 1), 1);
// }
// else
// {
// charCount.put(placeName.charAt(placeName.length() - 1),
// charCount.get(placeName.charAt(placeName.length() - 1)) + 1);
// }

// for (Character c : charCount.keySet())
// {
// System.out.println(c + " (" + (int) c + ") -> " + charCount.get(c));
// }
//
// System.out.println("Upper -> " + upperCount);
// System.out.println("Lower -> " + lowerCount);
// System.out.println("Number -> " + numbers);
// System.out.println("Arabic -> " + arabicCharacters);
// System.out.println("Chinese -> " + chineseCharacters);

// if (placeName.charAt(i) >= 'A' && placeName.charAt(i) <= 'Z')
// {
// upperCount++;
// }
//
// if (placeName.charAt(i) >= 'a' && placeName.charAt(i) <= 'z')
// {
// lowerCount++;
// }
//
// if (placeName.charAt(i) >= '0' && placeName.charAt(i) <= '9')
// {
// numbers++;
// }
//
// if (placeName.charAt(i) >= 1569 && placeName.charAt(i) <= 1711)
// {
// arabicCharacters++;
// }
//
// if (placeName.charAt(i) >= 12293 && placeName.charAt(i) <= 64016)
// {
// chineseCharacters++;
// }
//
// if (!charCount.containsKey(placeName.charAt(i)))
// {
// charCount.put(placeName.charAt(i), 1);
// }
// else
// {
// charCount.put(placeName.charAt(i), charCount.get(placeName.charAt(i)) + 1);
// }