java

package com.chenlb.mmseg4j;

import java.io.BufferedInputStream;
import java.io.BufferedReader;
import java.io.File;
import java.io.FileInputStream;
import java.io.FilenameFilter;
import java.io.IOException;
import java.io.InputStream;
import java.io.InputStreamReader;
import java.net.URL;
import java.util.ArrayList;
import java.util.HashMap;
import java.util.Map;
import java.util.Map.Entry;
import java.util.concurrent.ConcurrentHashMap;
import java.util.logging.Level;
import java.util.logging.Logger;

/**
* 词典类. 词库目录单例模式.<br/>
* 保存单字与其频率,还有词库.<br/>
* 有检测词典变更的接口，外部程序可以使用 {@link #wordsFileIsChange()} 和 {@link #reload()} 来完成检测与加载的工作.
*
* @author chenlb 2009-2-20 下午11:34:29
*/
public class Dictionary {

     private static final Logger log = Logger.getLogger(Dictionary.class.getName());

     private File dicPath;     //词库目录
     private volatile Map<Character, CharNode> dict;
     private volatile Map<Character, Object> unit;     //单个字的单位

     /** 记录 word 文件的最后修改时间 */
     private Map<File, Long> wordsLastTime = null;
     private long lastLoadTime = 0;

     /** 不要直接使用, 通过 {@link #getDefalutPath()} 使用*/
     private static File defalutPath = null;
     private static final ConcurrentHashMap<File, Dictionary> dics = new ConcurrentHashMap<File, Dictionary>();

     protected void finalize() throws Throwable {
          /*
          * 使 class reload 的时也可以释放词库
          */
          destroy();
     }

     /**
     * 从默认目录加载词库文件.<p/>
     * 查找默认目录顺序:
     *

     *
1. 从系统属性mmseg.dic.path指定的目录中加载

     *
2. 从classpath/data目录

     *
3. 从user.dir/data目录

     *

     * @see #getDefalutPath()
     */
     public static Dictionary getInstance() {
          File path = getDefalutPath();
          return getInstance(path);
     }

     /**
     * @param path 词典的目录
     */
     public static Dictionary getInstance(String path) {
          return getInstance(new File(path));
     }

     /**
     * @param path 词典的目录
     */
     public static Dictionary getInstance(File path) {
          log.info("try to load dir="+path);
          File normalizeDir = normalizeFile(path);
        System.out.println(normalizeDir);
          Dictionary dic = dics.get(normalizeDir);
          if(dic == null) {
               dic = new Dictionary(normalizeDir);
               dics.put(normalizeDir, dic);
          }
          return dic;
     }

     public static File normalizeFile(File file) {
          if(file == defalutPath) {
               return defalutPath;
          }
          try {
               return file.getCanonicalFile();
          } catch (IOException e) {
               throw new RuntimeException("normalize file=["+file+"] fail", e);
          }
     }

     /**
     * 销毁, 释放资源. 此后此对像不再可用.
     */
     void destroy() {
          clear(dicPath);

          dicPath = null;
          dict = null;
          unit = null;
     }

     /**
     * @see Dictionary#clear(File)
     */
     public static Dictionary clear(String path) {
          return clear(new File(path));
     }

     /**
     * 从单例缓存中去除
     * @param path
     * @return 没有返回 null
     */
     public static Dictionary clear(File path) {
          File normalizeDir = normalizeFile(path);
          return dics.remove(normalizeDir);
     }

     /**
     * 词典的目录
     */
     private Dictionary(File path) {
          init(path);
     }

     private void init(File path) {
          dicPath = path;
          wordsLastTime = new HashMap<File, Long>();

          reload();     //加载词典
     }

     private static long now() {
          return System.currentTimeMillis();
     }

     /**
     * 只要 wordsXXX.dic的文件
     * @return
     */
     protected File[] listWordsFiles() {
          return dicPath.listFiles(new FilenameFilter() {

               public boolean accept(File dir, String name) {

                    return name.startsWith("words") && name.endsWith(".dic");
               }

          });
     }

     private Map<Character, CharNode> loadDic(File wordsPath) throws IOException {
          InputStream charsIn = null;
          File charsFile = new File(wordsPath, "chars.dic");
          if(charsFile.exists()) {
               charsIn = new FileInputStream(charsFile);
               addLastTime(charsFile);     //chars.dic 也检测是否变更
          } else {     //从 jar 里加载
               charsIn = this.getClass().getResourceAsStream("/data/chars.dic");
               charsFile = new File(this.getClass().getResource("/data/chars.dic").getFile());     //only for log
          }
          final Map<Character, CharNode> dic = new HashMap<Character, CharNode>();
          int lineNum = 0;
          long s = now();
          long ss = s;
          lineNum = load(charsIn, new FileLoading() {     //单个字的

               public void row(String line, int n) {
                    if(line.length() < 1) {
                         return;
                    }
                    String[] w = line.split(" ");
                    CharNode cn = new CharNode();
                    switch(w.length) {
                    case 2:
                         try {
                              cn.setFreq((int)(Math.log(Integer.parseInt(w[1]))*100));//字频计算出自由度
                         } catch(NumberFormatException e) {
                              //eat...
                         }
                    case 1:

                         dic.put(w[0].charAt(0), cn);
                    }
               }
          });
          log.info("chars loaded time="+(now()-s)+"ms, line="+lineNum+", on file="+charsFile);

          //try load words.dic in jar
          InputStream wordsDicIn = this.getClass().getResourceAsStream("/data/words.dic");
          if(wordsDicIn != null) {
               File wordsDic = new File(this.getClass().getResource("/data/words.dic").getFile());
               loadWord(wordsDicIn, dic, wordsDic);
          }

          File[] words = listWordsFiles();     //只要 wordsXXX.dic的文件
          if(words != null) {     //扩展词库目录
               for(File wordsFile : words) {
                    loadWord(new FileInputStream(wordsFile), dic, wordsFile);

                    addLastTime(wordsFile);     //用于检测是否修改
               }
          }

          log.info("load all dic use time="+(now()-ss)+"ms");
          return dic;
     }

     /**
     * @param is 词库文件流
     * @param dic 加载的词保存在结构中
     * @param wordsFile     日志用
     * @throws IOException from {@link #load(InputStream, FileLoading)}
     */
     private void loadWord(InputStream is, Map<Character, CharNode> dic, File wordsFile) throws IOException {
          long s = now();
          int lineNum = load(is, new WordsFileLoading(dic)); //正常的词库
          log.info("words loaded time="+(now()-s)+"ms, line="+lineNum+", on file="+wordsFile);
     }

     private Map<Character, Object> loadUnit(File path) throws IOException {
          InputStream fin = null;
          File unitFile = new File(path, "units.dic");
          if(unitFile.exists()) {
               fin = new FileInputStream(unitFile);
               addLastTime(unitFile);
          } else {     //在jar包里的/data/unit.dic
               fin = Dictionary.class.getResourceAsStream("/data/units.dic");
               unitFile = new File(Dictionary.class.getResource("/data/units.dic").getFile());
          }

          final Map<Character, Object> unit = new HashMap<Character, Object>();

          long s = now();
          int lineNum = load(fin, new FileLoading() {

               public void row(String line, int n) {
                    if(line.length() != 1) {
                         return;
                    }
                    unit.put(line.charAt(0), Dictionary.class);
               }
          });
          log.info("unit loaded time="+(now()-s)+"ms, line="+lineNum+", on file="+unitFile);

          return unit;
     }

     /**
     * 加载 wordsXXX.dic 文件类。
     *
     * @author chenlb 2009-10-15 下午02:12:55
     */
     private static class WordsFileLoading implements FileLoading {
          final Map<Character, CharNode> dic;

          /**
          * @param dic 加载的词，保存在此结构中。
          */
          public WordsFileLoading(Map<Character, CharNode> dic) {
               this.dic = dic;
          }

          public void row(String line, int n) {
               if(line.length() < 2) {
                    return;
               }
               CharNode cn = dic.get(line.charAt(0));
               if(cn == null) {
                    cn = new CharNode();
                    dic.put(line.charAt(0), cn);
               }
               cn.addWordTail(tail(line));
          }
     }

     /**
     * 加载词文件的模板
     * @return 文件总行数
     */
     public static int load(InputStream fin, FileLoading loading) throws IOException {
          BufferedReader br = new BufferedReader(
                    new InputStreamReader(new BufferedInputStream(fin), "UTF-8"));
          String line = null;
          int n = 0;
          while((line = br.readLine()) != null) {
               if(line == null || line.startsWith("#")) {
                    continue;
               }
               n++;
               loading.row(line, n);
          }
          return n;
     }

     /**
     * 取得 str 除去第一个char的部分
     * @author chenlb 2009-3-3 下午10:05:26
     */
     private static char[] tail(String str) {
          char[] cs = new char[str.length()-1];
          str.getChars(1, str.length(), cs, 0);
          return cs;
     }

     public static interface FileLoading {
          /**
          * @param line 读出的一行
          * @param n 当前第几行
          * @author chenlb 2009-3-3 下午09:55:54
          */
          void row(String line, int n);
     }

     /**
     * 把 wordsFile 文件的最后更新时间加记录下来.
     * @param wordsFile 非 null
     */
     private synchronized void addLastTime(File wordsFile) {
          if(wordsFile != null) {
               wordsLastTime.put(wordsFile, wordsFile.lastModified());
          }
     }

     /**
     * 词典文件是否有修改过
     * @return
     */
     public synchronized boolean wordsFileIsChange() {
          //检查是否有修改文件,包括删除的
          for(Entry<File, Long> flt : wordsLastTime.entrySet()) {
               File words = flt.getKey();
               if(!words.canRead()) {     //可能是删除了
                    return true;
               }
               if(words.lastModified() > flt.getValue()) {     //更新了文件
                    return true;
               }
          }
          //检查是否有新文件
          File[] words = listWordsFiles();
          if(words != null) {
               for(File wordsFile : words) {
                    if(!wordsLastTime.containsKey(wordsFile)) {     //有新词典文件
                         return true;
                    }
               }
          }
          return false;
     }

     /**
     * 全新加载词库，没有成功加载会回滚。<P/>
     * 注意：重新加载时，务必有两倍的词库树结构的内存，默认词库是 50M/个 左右。否则抛出 OOM。
     * @return 是否成功加载
     */
     public synchronized boolean reload() {
          Map<File, Long> oldWordsLastTime = new HashMap<File, Long>(wordsLastTime);
          Map<Character, CharNode> oldDict = dict;
          Map<Character, Object> oldUnit = unit;

          try {
               wordsLastTime.clear();
               dict = loadDic(dicPath);
               unit = loadUnit(dicPath);
               lastLoadTime = System.currentTimeMillis();
          } catch (IOException e) {
               //rollback
               wordsLastTime.putAll(oldWordsLastTime);
               dict = oldDict;
               unit = oldUnit;

               if(log.isLoggable(Level.WARNING)) {
                    log.log(Level.WARNING, "reload dic error! dic="+dicPath+", and rollbacked.", e);
               }

               return false;
          }
          return true;
     }

     /**
     * word 能否在词库里找到
     * @author chenlb 2009-3-3 下午11:10:45
     */
     public boolean match(String word) {
          if(word == null || word.length() < 2) {
               return false;
          }
          CharNode cn = dict.get(word.charAt(0));
          return search(cn, word.toCharArray(), 0, word.length()-1) >= 0;
     }

     public CharNode head(char ch) {
          return dict.get(ch);
     }

     /**
     * sen[offset] 后 tailLen 长的词是否存在.
     * @see CharNode#indexOf(char[], int, int)
     * @author chenlb 2009-4-8 下午11:13:49
     */
     public int search(CharNode node, char[] sen, int offset, int tailLen) {
          if(node != null) {
               return node.indexOf(sen, offset, tailLen);
          }
          return -1;
     }

     public int maxMatch(char[] sen, int offset) {
          CharNode node = dict.get(sen[offset]);
          return maxMatch(node, sen, offset);
     }

     public int maxMatch(CharNode node, char[] sen, int offset) {
          if(node != null) {
               return node.maxMatch(sen, offset+1);
          }
          return 0;
     }

     public ArrayList<Integer> maxMatch(CharNode node, ArrayList<Integer> tailLens, char[] sen, int offset) {
          tailLens.clear();
          tailLens.add(0);
          if(node != null) {
               return node.maxMatch(tailLens, sen, offset+1);
          }
          return tailLens;
     }

     public boolean isUnit(Character ch) {
          return unit.containsKey(ch);
     }

     /**
     * 当 words.dic 是从 jar 里加载时, 可能 defalut 不存在
     */
     public static File getDefalutPath() {
          if(defalutPath == null) {
               String defPath = System.getProperty("mmseg.dic.path");
               log.info("look up in mmseg.dic.path="+defPath);
               if(defPath == null) {
                    URL url = Dictionary.class.getClassLoader().getResource("data");
                    if(url != null) {
                         defPath = url.getFile();
                         log.info("look up in classpath="+defPath);
                    } else {
                         defPath = System.getProperty("user.dir")+"/data";
                         log.info("look up in user.dir="+defPath);
                    }

               }

               defalutPath = new File(defPath);
               if(!defalutPath.exists()) {
                    log.warning("defalut dic path="+defalutPath+" not exist");
               }
          }
          return defalutPath;
     }

     /**
     * 仅仅用来观察词库.
     */
     public Map<Character, CharNode> getDict() {
          return dict;
     }

     /**
     * 注意：当 words.dic 是从 jar 里加载时，此时 File 可能是不存在的。
     */
     public File getDicPath() {
          return dicPath;
     }

     /** 最后加载词库的时间 */
     public long getLastLoadTime() {
          return lastLoadTime;
     }
}