hebing1.java

采用JAVA实现两个词库的合并

import java.io.*;

public class heBing{

	/**
	 * @param args
	 */
	public static void main(String[] args) {
		// TODO 自动生成方法存根
		ciku ck = new ciku();
		ck.readword();
	}
}

class ciku{
	String s1 = null;//第一个词库的词
	String s2 = null;//第二个词库的词
	int i = 1;	
	int i2 = 0;
	int j = 1;  //6747个首字
	int k = 1;  //以某个首字开头的词的个数
	boolean mod = false ; //标志
	boolean cunzai = false; //标志
	String [] f = new String [6748]; //存放所有首字  
	String [] str = new String [548390];  //存放所有词
	String [][] w = new String [6748][];  //存放某一首字下的词
	int [] wordNum = new int [6748] ;  //存放某一首字下词的个数
	File file1 = new File("紫光输入法词库(548387词).txt");
	File file2 = new File("11.txt");
	String [] str2 = new String [98447];
	int y = 0;

	int min(int m,int y)
	{
		if(m>y)
			return y;
		else
			return m;
	}
	
	void comWord(int x)
	{
		cunzai = false;
		for(int q = 1;q <= wordNum[x];q++)//某一首字下的各词查找
		{
			if(s2.equals(w[x][q]))//查找到相同的词，返回
			{
				cunzai = true;
				//System.out .println(i2 + ":" + s2 + "same") ;
				break;
			}
		}
		if(!cunzai) //不存在，需要插入的词
		{
			y++;
			str2[y] = s2;
			
				
////////////////查找插入的位置
			String [] oldw = new String [w[x].length ];
			int ix =0;
			boolean sma = false;
			for(int q =1 ;q<= wordNum[x];q++)
			{ 
				if(!sma) 
				{
					oldw[q] = w[x][q];
					int bjfw = min(w[x][q].length(),s2.length());
					//System.out.println(w[x][q].length() + " " + s2.length() + " " + bjfw);
					for(int wor = 1; wor < bjfw; wor ++)
					{
						if(s2.substring(wor, wor+1).equals(w[x][q].substring(wor, wor+1)))
						{
							//sam = true;
							if((wor + 1) == w[x][q].length())
							{
								ix = q;
								break;
							}
						}
						else 
						{
							String pys2 = GB2Alpha.getFullSpell(s2.substring(wor,wor+1));
							String pyw = GB2Alpha.getFullSpell(w[x][q].substring(wor, wor+1));
							byte[] bytes = (String.valueOf(pyw.substring(0, 1))).getBytes();
							if(bytes.length ==1)
							{
								if(pys2.substring(0, 1).compareTo(pyw.substring(0, 1)) < 0)
								{
									//System.out.print("aaaaa");
									ix = q - 1;sma = true;
									break ;
									
								}
								else
								{
									if(q == wordNum[x])
									{
										ix = q;
									}
									break;
								}
							}
							break;
						}
					}
					}		
				}
				
				System.out.println(x + " " + ix + ":" + w[x][ix]);
///////////插入				
				w[x]= new String [w[x].length +1];
				wordNum[x]++;
				for(int a =1 ;a<= ix;a++)
				{
					w[x][a] = oldw[a];
				}
				w[x][ix+1 ]= s2;
				for(int a= ix+2;a<w[x].length ;a++)
				{
					w[x][a]= oldw[a-1];
				}
			/*	
				for(int a=1 ;a<w[x].length ;a++)
				{
					System.out.println(w[x][a]);
				}
			*/	
				System.out.println(y + "-" + i2 + "=" + s2) ;
			}
	}
	
	
	/////从第二个词库读取词，以同首字为范围查找，不存在则插入
	void comFirWord()
	{
		try{
			FileReader fileIn2 = new FileReader(file2);
			BufferedReader in2 = new BufferedReader(fileIn2);
			
			while((s2 = in2.readLine()) != null)
			{
				i2++;
				mod = false;
				for(int p = 1;p<=j;p++)//逐个首字查找
				{
					if(s2.substring( 0,1).equals( f[p]))//查找到同首字
					{
						mod = true;
						
						comWord(p);//参数p为行数
						break;
					}
				}
				if(!mod )//首字不存在，词即不存在
				{
					//System.out.println(i2 + ":" + "none") ;
					y++;
					str2[y] = s2;
					System.out.println(y + "-" + i2 + ":" + str2[y]) ;
				}
			}
			in2.close();
		}
		catch(IOException ee)
		{
			System.out .println("error:" + ee) ;
		}
	}
	
	
	////////从一个词库中读取所有词并存入数组
	void readword()
	{
		try{
			FileReader fileIn1 = new FileReader(file1);
			BufferedReader in1 = new BufferedReader(fileIn1);
			
			s1 = in1.readLine();
			/////////////////////////////第一个词
			f[j] = s1.substring(0,1);
			str[i] = s1;
		//	System.out .println(j + ":" + f[j]) ;
			
			s1 = in1.readLine();
			while(s1!= null)
				{
					i++;
					str[i] = s1.trim(); 
					if(str[i].substring( 0,1) .equals( str[i-1].substring(0,1)) )
					{
						k++;
					}
					else
					{
						wordNum[j] = k;
					//	System.out.println(j +":"+ wordNum[j]) ;
						w[j] = new String [k+1];
						for(int t=k;t>0;t--)
						{
							w[j][k-t+1] = str[i-t];
							//System.out .println(j + "-" + (k-t+1) + ":" +w[j][k-t+1]);
						}
						j++;
						k = 1;
						f[j] = s1.substring( 0,1);
						//System.out .println(j + ":" + f[j]);
						
					}
					s1 = in1.readLine();
					
				}
			if(s1==null)
			{
				wordNum[j] = k;
				w[j] = new String [k+1];
				for(int t=k-1;t>=0;t--)
				{
					w[j][k-t] = str[i-t];
			//		System.out .println(j + "-" + (k-t) + ":" +w[j][k-t]);
				}
			}
			in1.close();
			
			comFirWord();
			/*								
/////////将合并结果写入
			File file = new File("1.txt");
			FileWriter fileOut = new FileWriter(file);
			BufferedWriter out = new BufferedWriter(fileOut);
			
			for(int fir=1;fir <= j;fir++)
			{
				//System.out.println(fir + " " + f[fir]);
				for(int word = 1;word < w[fir].length ;word++)
				{
					System.out.println(fir + " " + word + ":" + w[fir][word]);
					out.write(w[fir][word],0,w[fir][word].length());
					out.newLine();
				}
			}
			
			
						
			for(int a =1;a<str2.length ;a++)
			{
				out.write( str2[a],0,str2[a].length() );
				out.newLine() ;
			}
			out.close();
				
	*/		
			
		}
		catch(IOException e)
		{
			System.out .println("error is :" + e) ;
		}
		
	
	}
}