Lucene?索引刪除策略源碼解析

更新時(shí)間：2023年03月14日 10:46:22 作者：滄叔解碼

這篇文章主要為大家介紹了Lucene?索引刪除策略源碼解析，有需要的朋友可以借鑒參考下，希望能夠有所幫助，祝大家多多進(jìn)步，早日升職加薪

Lucene

從今天開(kāi)始，我們要開(kāi)始介紹Lucene中索引構(gòu)建的流程。因?yàn)樗饕龢?gòu)建的邏輯涉及到的東西非常多，如果從構(gòu)建入口IndexWriter來(lái)開(kāi)始介紹，是很難說(shuō)清楚的。所以接下來(lái)按化零為整的方式，從構(gòu)建相關(guān)的各個(gè)組件開(kāi)始介紹，盡量每一篇文章都是可以獨(dú)立閱讀，依賴的前置知識(shí)都是我已經(jīng)介紹的內(nèi)容。不管就算如此，還是會(huì)有部分內(nèi)容可能需要結(jié)合整體流程才能明白，對(duì)于這部分的內(nèi)容，大家可以先留個(gè)印象，以后介紹相關(guān)聯(lián)的內(nèi)容時(shí)，我會(huì)再重新指出。

今天我們一起來(lái)看看索引文件刪除相關(guān)的。

IndexCommit

Lucene中，需要持久化的索引信息都要進(jìn)行commit操作，然后會(huì)生成一個(gè)segments_N的索引文件記錄此次commit相關(guān)的索引信息。

一次commit生成segments_N之后，就對(duì)應(yīng)了一個(gè)IndexCommit，IndexCommit只是一個(gè)接口，它定義了可以從IndexCommit中獲取哪些信息：

public abstract class IndexCommit implements Comparable<IndexCommit> {
  // commit對(duì)應(yīng)的segments_N
  public abstract String getSegmentsFileName();
  // commit關(guān)聯(lián)的所有的索引文件
  public abstract Collection<String> getFileNames() throws IOException;
  // 索引所在的Directory
  public abstract Directory getDirectory();
  // 刪除commit，后面會(huì)看到，刪除其實(shí)減少commit關(guān)聯(lián)的索引文件的引用計(jì)數(shù)
  public abstract void delete();
  // commit是否被刪除了
  public abstract boolean isDeleted();
  // commit關(guān)聯(lián)了幾個(gè)segment
  public abstract int getSegmentCount();
  // segments_N文件中的N
  public abstract long getGeneration();
  // commit可以記錄一些用戶自定義的信息
  public abstract Map<String, String> getUserData() throws IOException;
  // 用來(lái)讀取commit對(duì)應(yīng)的索引數(shù)據(jù)
  StandardDirectoryReader getReader() {
    return null;
  }
}

IndexCommit有三個(gè)實(shí)現(xiàn)類：

CommitPoint
ReaderCommit
SnapshotCommitPoint

這個(gè)三個(gè)實(shí)現(xiàn)類都有對(duì)應(yīng)的使用場(chǎng)景，在用到的時(shí)候我會(huì)再詳細(xì)介紹，本文中會(huì)涉及到SnapshotCommitPoint，后面會(huì)詳細(xì)介紹它。

IndexDeletionPolicy

在索引的生命周期中，可以有多次的commit操作，因此也會(huì)生成多個(gè)segments_N文件，對(duì)于這些文件是否要保留還是刪除，lucene中是通過(guò)IndexDeletionPolicy來(lái)管理的。我們先來(lái)看下IndexDeletionPolicy的接口定義：

public abstract class IndexDeletionPolicy {
  protected IndexDeletionPolicy() {}
  // 重新打開(kāi)索引的時(shí)候，對(duì)所有commit的處理
  public abstract void onInit(List<? extends IndexCommit> commits) throws IOException;
  // 有新提交時(shí)對(duì)所有commit的處理
  public abstract void onCommit(List<? extends IndexCommit> commits) throws IOException;
}

從上面我可以看到，索引的刪除策略其實(shí)只在兩個(gè)地方進(jìn)行應(yīng)用，一個(gè)是加載索引的時(shí)候，打開(kāi)一個(gè)舊索引時(shí)，根據(jù)當(dāng)前設(shè)置的IndexDeletionPolicy進(jìn)行處理。另一個(gè)是有新的commit產(chǎn)生時(shí)，借這個(gè)機(jī)會(huì)處理所有的commit。Lucene中提供的索引刪除策略一共有四種，不過(guò)可以分為三類：

NoDeletionPolicy

NoDeletionPolicy索引刪除策略就是保留所有的commit信息，效果就是你有多少次commit就多少個(gè)segments_N文件，看個(gè)例子：

public class DeletionPolicyTest {
    private static final Random RANDOM = new Random();
    public static void main(String[] args) throws IOException {
        Directory directory = FSDirectory.open(new File("D:\\code\\lucene-9.1.0-learning\\data").toPath());
        WhitespaceAnalyzer analyzer = new WhitespaceAnalyzer();
        IndexWriterConfig indexWriterConfig = new IndexWriterConfig(analyzer);
        indexWriterConfig.setUseCompoundFile(true);
        indexWriterConfig.setIndexDeletionPolicy(NoDeletionPolicy.INSTANCE);
        IndexWriter indexWriter = new IndexWriter(directory, indexWriterConfig);
        indexWriter.addDocument(getDoc(RANDOM.nextInt(10000),RANDOM.nextInt(10000)));
        // 第一次commit，生成segments_1
        indexWriter.commit();
        indexWriter.addDocument(getDoc(RANDOM.nextInt(10000),RANDOM.nextInt(10000)));
        // 第二次commit，生成segments_2
        indexWriter.commit();
        indexWriter.close();
    }
    private static Document getDoc(int... point) {
        Document doc = new Document();
        IntPoint intPoint = new IntPoint("point", point);
        doc.add(intPoint);
        return doc;
    }
}

上面的例子中有兩次commit，下圖是NoDeletionPolicy策略進(jìn)行了兩次commit的索引目錄結(jié)構(gòu)，可以看到生成了兩個(gè)segments_N文件：

NoDeletionPolicy的代碼實(shí)現(xiàn)非常簡(jiǎn)單，單例實(shí)現(xiàn)，并且在onCommit和onInit的時(shí)候都是空操作：

public final class NoDeletionPolicy extends IndexDeletionPolicy {
  public static final IndexDeletionPolicy INSTANCE = new NoDeletionPolicy();
  private NoDeletionPolicy() {
  }
  public void onCommit(List<? extends IndexCommit> commits) {}
  public void onInit(List<? extends IndexCommit> commits) {}
}

KeepOnlyLastCommitDeletionPolicy

KeepOnlyLastCommitDeletionPolicy是Lucene默認(rèn)的索引刪除策略，只保留最新的一次commit，從索引目錄看不管執(zhí)行多少次commit只保留了N最大的segments_N文件，下圖是KeepOnlyLastCommitDeletionPolicy策略進(jìn)行了兩次commit的結(jié)果，KeepOnlyLastCommitDeletionPolicy刪除策略只保留了segments_2。把上面示例代碼中的刪除策略替換成KeepOnlyLastCommitDeletionPolicy，即可得到，注意需要先清空索引目錄：

KeepOnlyLastCommitDeletionPolicy代碼實(shí)現(xiàn)也比較簡(jiǎn)單，除了最后一個(gè)commit之外，其他的commit都刪除：

public final class KeepOnlyLastCommitDeletionPolicy extends IndexDeletionPolicy {
  public KeepOnlyLastCommitDeletionPolicy() {}
  public void onInit(List<? extends IndexCommit> commits) {
    onCommit(commits);
  }
  // commits是從舊到新排序的
  public void onCommit(List<? extends IndexCommit> commits) {
    // 只保留最新的一個(gè)
    int size = commits.size();
    for (int i = 0; i < size - 1; i++) {
      commits.get(i).delete();
    }
  }
}

兩個(gè)快照相關(guān)的刪除策略

快照相關(guān)的刪除策略有兩個(gè)，SnapshotDeletionPolicy和PersistentSnapshotDeletionPolicy，分別對(duì)應(yīng)了不可持久化和可持久化的模式。不管是SnapshotDeletionPolicy還是PersistentSnapshotDeletionPolicy，他們都封裝了其他的IndexDeletionPolicy來(lái)執(zhí)行刪除策略，他們只是提供了為當(dāng)前最新的commit生成快照的能力。只要快照存在，則跟快照相關(guān)的所有索引文件都會(huì)被無(wú)條件保留。

SnapshotDeletionPolicy

例子

public class SnapshotDeletionPolicyTest {
    private static final Random RANDOM = new Random();
    public static void main(String[] args) throws IOException, InterruptedException {
        Directory directory = FSDirectory.open(new File("D:\\code\\lucene-9.1.0-learning\\data").toPath());
        WhitespaceAnalyzer analyzer = new WhitespaceAnalyzer();
        IndexWriterConfig indexWriterConfig = new IndexWriterConfig(analyzer);
        indexWriterConfig.setUseCompoundFile(true);
        SnapshotDeletionPolicy snapshotDeletionPolicy = new SnapshotDeletionPolicy(new KeepOnlyLastCommitDeletionPolicy());
        indexWriterConfig.setIndexDeletionPolicy(snapshotDeletionPolicy);
        IndexWriter indexWriter = new IndexWriter(directory, indexWriterConfig);
        indexWriter.addDocument(getDoc(RANDOM.nextInt(10000),RANDOM.nextInt(10000)));
        // 第一次commit，生成segments_1
        indexWriter.commit();
        indexWriter.addDocument(getDoc(RANDOM.nextInt(10000),RANDOM.nextInt(10000)));
        // 第二次commit，生成segments_2
        indexWriter.commit();
        // segments_2當(dāng)做快照，無(wú)條件保留
        snapshotDeletionPolicy.snapshot();
        indexWriter.addDocument(getDoc(RANDOM.nextInt(10000),RANDOM.nextInt(10000)));
        // 第三次commit，生成segments_3
        indexWriter.commit();
        indexWriter.close();
    }
    private static Document getDoc(int... point) {
        Document doc = new Document();
        IntPoint intPoint = new IntPoint("point", point);
        doc.add(intPoint);
        return doc;
    }
}

在上面的例子中，我們使用SnapshotDeletionPolicy，SnapshotDeletionPolicy底層封裝的是KeepOnlyLastCommitDeletionPolicy，我們進(jìn)行了三次commit，理論上KeepOnlyLastCommitDeletionPolicy只會(huì)保留最后一次，但是因?yàn)槲覀儗?duì)第一次的commit進(jìn)行了快照，所以第一次commit也被保留了：

接下來(lái)我們看看SnapshotDeletionPolicy是怎么實(shí)現(xiàn)。SnapshotDeletionPolicy保證生成快照的commit不會(huì)被刪除的原理就是引用計(jì)數(shù)，SnapshotDeletionPolicy會(huì)記錄每個(gè)commit生成快照的次數(shù)，在刪除的時(shí)候，只會(huì)刪除引用計(jì)數(shù)為0的commit。

成員變量

  // key是IndexCommit的generation，value是對(duì)應(yīng)的IndexCommit有多少個(gè)快照
  // 需要注意的是，有被快照引用的才會(huì)記錄在refCounts中，也就是只要被記錄在refCounts中，引用次數(shù)至少是1
  protected final Map<Long, Integer> refCounts = new HashMap<>();
  // key是IndexCommit的generation，value是對(duì)應(yīng)的IndexCommit
  protected final Map<Long, IndexCommit> indexCommits = new HashMap<>();
  // SnapshotDeletionPolicy只是增加了支持快照的功能，刪除的邏輯是由primary參數(shù)對(duì)應(yīng)的刪除策略提供的
  private final IndexDeletionPolicy primary;
  // 最近一次提交的commit，只會(huì)對(duì)這個(gè)IndexCommit生成快照
  protected IndexCommit lastCommit;
  // 是否初始化的標(biāo)記，實(shí)例化后，必須先調(diào)用onInit方法
  private boolean initCalled;

生成快照

生成快照只會(huì)對(duì)當(dāng)前最新的一個(gè)commit進(jìn)行快照：

  public synchronized IndexCommit snapshot() throws IOException {
    if (!initCalled) {
      throw new IllegalStateException(
          "this instance is not being used by IndexWriter; be sure to use the instance returned from writer.getConfig().getIndexDeletionPolicy()");
    }
    if (lastCommit == null) {
      throw new IllegalStateException("No index commit to snapshot");
    }
    // 新增lastCommit的引用計(jì)數(shù)
    incRef(lastCommit);
    return lastCommit;
  }
  protected synchronized void incRef(IndexCommit ic) {
    long gen = ic.getGeneration();
    Integer refCount = refCounts.get(gen);
    int refCountInt;
    if (refCount == null) { // 第一次被引用
      indexCommits.put(gen, lastCommit);
      refCountInt = 0;
    } else {
      refCountInt = refCount.intValue();
    }
    // 引用計(jì)數(shù)加+1  
    refCounts.put(gen, refCountInt + 1);
  }

釋放指定的快照

public synchronized void release(IndexCommit commit) throws IOException {
  long gen = commit.getGeneration();
  releaseGen(gen);
}
protected void releaseGen(long gen) throws IOException {
  if (!initCalled) {
    throw new IllegalStateException(
        "this instance is not being used by IndexWriter; be sure to use the instance returned from writer.getConfig().getIndexDeletionPolicy()");
  }
  Integer refCount = refCounts.get(gen);
  if (refCount == null) {
    throw new IllegalArgumentException("commit gen=" + gen + " is not currently snapshotted");
  }
  int refCountInt = refCount.intValue();
  assert refCountInt > 0;
  refCountInt--;
  if (refCountInt == 0) { // 引用計(jì)數(shù)為0，直接從refCounts中移除
    refCounts.remove(gen);
    indexCommits.remove(gen);
  } else {
    refCounts.put(gen, refCountInt);
  }
}

刪除commit

  public synchronized void onCommit(List<? extends IndexCommit> commits) throws IOException {
    // 把commits中的所有IndexCommit都封裝成SnapshotCommitPoint，再使用primary執(zhí)行onCommit方法  
    primary.onCommit(wrapCommits(commits));
    // 更新最新的commit  
    lastCommit = commits.get(commits.size() - 1);
  }
  @Override
  public synchronized void onInit(List<? extends IndexCommit> commits) throws IOException {
    // 設(shè)置初始化的標(biāo)記  
    initCalled = true;
    primary.onInit(wrapCommits(commits));
    for (IndexCommit commit : commits) { 
      if (refCounts.containsKey(commit.getGeneration())) {
        indexCommits.put(commit.getGeneration(), commit);
      }
    }
    if (!commits.isEmpty()) {
      lastCommit = commits.get(commits.size() - 1);
    }
  }
  private List<IndexCommit> wrapCommits(List<? extends IndexCommit> commits) {
    List<IndexCommit> wrappedCommits = new ArrayList<>(commits.size());
    for (IndexCommit ic : commits) {
      // 把IndexCommit都封裝成 SnapshotCommitPoint
      wrappedCommits.add(new SnapshotCommitPoint(ic));
    }
    return wrappedCommits;
  }

前面我們列出了SnapshotCommitPoint是IndexCommit的一個(gè)實(shí)現(xiàn)類，但是沒(méi)有詳細(xì)介紹，SnapshotCommitPoint除了能夠提供IndexCommit接口所提供的信息之外，最核心的是在刪除的時(shí)候，會(huì)先判斷IndexCommit是否被快照引用，只有沒(méi)有任何快照引用的IndexCommit才能刪除：

public void delete() {
    synchronized (SnapshotDeletionPolicy.this) {
        if (!refCounts.containsKey(cp.getGeneration())) {
            cp.delete();
        }
    }
}

存在的問(wèn)題

需要注意的是SnapshotDeletionPolicy的快照信息是沒(méi)有持久化，我們重新打開(kāi)SnapshotDeletionPolicyTest例子中生成的索引：

public class SnapshotDeletionPolicyTest2 {
    public static void main(String[] args) throws IOException, InterruptedException {
        Directory directory = FSDirectory.open(new File("D:\\code\\lucene-9.1.0-learning\\data").toPath());
        WhitespaceAnalyzer analyzer = new WhitespaceAnalyzer();
        IndexWriterConfig indexWriterConfig = new IndexWriterConfig(analyzer);
        indexWriterConfig.setUseCompoundFile(true);
        SnapshotDeletionPolicy snapshotDeletionPolicy = new SnapshotDeletionPolicy(new KeepOnlyLastCommitDeletionPolicy());
        indexWriterConfig.setIndexDeletionPolicy(snapshotDeletionPolicy);
        // 重新打開(kāi)索引
        IndexWriter indexWriter = new IndexWriter(directory, indexWriterConfig);
        indexWriter.close();
    }
}

可以發(fā)現(xiàn)segments_1被刪除了，因?yàn)闆](méi)有持久化快照信息，所以根據(jù)KeepOnlyLastCommitDeletionPolicy的刪除策略，只保留了最新的一個(gè)commit：

PersistentSnapshotDeletionPolicy

例子

PersistentSnapshotDeletionPolicy主要是為了解決SnapshotDeletionPolicy無(wú)法持久化的問(wèn)題。PersistentSnapshotDeletionPolicy持久化的時(shí)候會(huì)生成snapshots_N的索引文件，我們看個(gè)例子：

public class PersistentSnapshotDeletionPolicyTest {
    private static final Random RANDOM = new Random();
    public static void main(String[] args) throws IOException, InterruptedException {
        Directory directory = FSDirectory.open(new File("D:\\code\\lucene-9.1.0-learning\\data").toPath());
        WhitespaceAnalyzer analyzer = new WhitespaceAnalyzer();
        IndexWriterConfig indexWriterConfig = new IndexWriterConfig(analyzer);
        indexWriterConfig.setUseCompoundFile(true);
        PersistentSnapshotDeletionPolicy persistentSnapshotDeletionPolicy = new PersistentSnapshotDeletionPolicy(new KeepOnlyLastCommitDeletionPolicy(), directory);
        indexWriterConfig.setIndexDeletionPolicy(persistentSnapshotDeletionPolicy);
        IndexWriter indexWriter = new IndexWriter(directory, indexWriterConfig);
        indexWriter.addDocument(getDoc(RANDOM.nextInt(10000),RANDOM.nextInt(10000)));
        // 第一次commit，生成segments_1
        indexWriter.commit();
        // segments_1當(dāng)做快照，無(wú)條件保留
        persistentSnapshotDeletionPolicy.snapshot();
        indexWriter.addDocument(getDoc(RANDOM.nextInt(10000),RANDOM.nextInt(10000)));
        // 第二次commit，生成segments_2
        indexWriter.commit();
        indexWriter.addDocument(getDoc(RANDOM.nextInt(10000),RANDOM.nextInt(10000)));
        // 第三次commit，生成segments_3
        indexWriter.commit();
        indexWriter.close();
    }
    private static Document getDoc(int... point) {
        Document doc = new Document();
        IntPoint intPoint = new IntPoint("point", point);
        doc.add(intPoint);
        return doc;
    }
}

上面的例子和我們?cè)诮榻BSnapshotDeletionPolicy的時(shí)候邏輯一樣，只是把SnapshotDeletionPolicy換成了PersistentSnapshotDeletionPolicy，我們看結(jié)果：

從上面結(jié)果圖中可以看到，segments_1和segments_3同樣被保留了，但是多了一個(gè)持久化的快照信息的文件snapshots_0，有了這個(gè)文件，索引重新打開(kāi)的時(shí)候就可以恢復(fù)快照信息，segments_1還是會(huì)被保留，用下面的例子我們重新打開(kāi)索引，可以發(fā)現(xiàn)segments_1還是被保留了：

public class PersistentSnapshotDeletionPolicyTest2 {
    public static void main(String[] args) throws IOException, InterruptedException {
        Directory directory = FSDirectory.open(new File("D:\\code\\lucene-9.1.0-learning\\data").toPath());
        WhitespaceAnalyzer analyzer = new WhitespaceAnalyzer();
        IndexWriterConfig indexWriterConfig = new IndexWriterConfig(analyzer);
        indexWriterConfig.setUseCompoundFile(true);
        PersistentSnapshotDeletionPolicy persistentSnapshotDeletionPolicy = new PersistentSnapshotDeletionPolicy(new KeepOnlyLastCommitDeletionPolicy(), directory);
        indexWriterConfig.setIndexDeletionPolicy(persistentSnapshotDeletionPolicy);
        IndexWriter indexWriter = new IndexWriter(directory, indexWriterConfig);
        indexWriter.close();
    }
}

接下來(lái)我們看看PersistentSnapshotDeletionPolicy的實(shí)現(xiàn)，主要就是持久化和恢復(fù)快照信息的邏輯。

成員變量

  // 持久化快照信息的文件名snapshots_N中的N，從0開(kāi)始
  private long nextWriteGen;
  // 持久化的文件所在的目錄
  private final Directory dir;

構(gòu)造函數(shù)

  public PersistentSnapshotDeletionPolicy(IndexDeletionPolicy primary, Directory dir)
      throws IOException {
    this(primary, dir, OpenMode.CREATE_OR_APPEND);
  }
  public PersistentSnapshotDeletionPolicy(IndexDeletionPolicy primary, Directory dir, OpenMode mode)
      throws IOException {
    super(primary);
    this.dir = dir;
    if (mode == OpenMode.CREATE) { // 新建索引的模式，則需要清除所有的快照信息，索引模式以后再介紹
      clearPriorSnapshots();
    }
    // 加載快照信息
    loadPriorSnapshots();
    if (mode == OpenMode.APPEND && nextWriteGen == 0) {
      throw new IllegalStateException("no snapshots stored in this directory");
    }
  }

生成快照

public synchronized IndexCommit snapshot() throws IOException {
  // 使用SnapshotDeletionPolicy來(lái)生成快照  
  IndexCommit ic = super.snapshot();
  // 標(biāo)記持久化是否成功，不成功的話需要?jiǎng)h除快照  
  boolean success = false;
  try {
    // 持久化最新的快照信息
    persist();
    success = true;
  } finally {
    if (!success) { // 持久化失敗，刪除快照
      try {
        super.release(ic);
      } catch (
          @SuppressWarnings("unused")
          Exception e) {
        // Suppress so we keep throwing original exception
      }
    }
  }
  return ic;
}

釋放快照

public synchronized void release(IndexCommit commit) throws IOException {
  // 使用SnapshotDeletionPolicy來(lái)釋放快照  
  super.release(commit);
  // 持久化快照信息是否成功  
  boolean success = false;
  try {
    // 持久化最新的快照信息  
    persist();
    success = true;
  } finally {
    if (!success) { // 持久化失敗，重新加回快照信息
      try {
        incRef(commit);
      } catch (
          @SuppressWarnings("unused")
          Exception e) {
        // Suppress so we keep throwing original exception
      }
    }
  }
}

持久化快照信息

private synchronized void persist() throws IOException {
  // 快照文件名  
  String fileName = SNAPSHOTS_PREFIX + nextWriteGen;
  boolean success = false;
  try (IndexOutput out = dir.createOutput(fileName, IOContext.DEFAULT)) {
    CodecUtil.writeHeader(out, CODEC_NAME, VERSION_CURRENT);
    out.writeVInt(refCounts.size());
    for (Entry<Long, Integer> ent : refCounts.entrySet()) { // 持久化所有的引用信息
      out.writeVLong(ent.getKey());
      out.writeVInt(ent.getValue());
    }
    success = true;
  } finally {
    if (!success) {
      IOUtils.deleteFilesIgnoringExceptions(dir, fileName);
    }
  }
  dir.sync(Collections.singletonList(fileName));
  if (nextWriteGen > 0) {
    String lastSaveFile = SNAPSHOTS_PREFIX + (nextWriteGen - 1);
    // 刪除前一個(gè)快照文件，因?yàn)槊看纬志没际前旬?dāng)前的快照信息全量持久化，所以只需要保留最新的一個(gè)就可以
    // 這里有可能刪除失敗，所以在啟動(dòng)加載的時(shí)候會(huì)再次嘗試把舊版本的文件都刪掉  
    IOUtils.deleteFilesIgnoringExceptions(dir, lastSaveFile);
  }
  nextWriteGen++;
}

加載快照信息

private synchronized void loadPriorSnapshots() throws IOException {
  long genLoaded = -1;
  IOException ioe = null;
  List<String> snapshotFiles = new ArrayList<>();
  for (String file : dir.listAll()) {
    if (file.startsWith(SNAPSHOTS_PREFIX)) { // 找到快照文件
      long gen = Long.parseLong(file.substring(SNAPSHOTS_PREFIX.length()));
      if (genLoaded == -1 || gen > genLoaded) { // 找到gen最大的快照文件
        snapshotFiles.add(file);
        Map<Long, Integer> m = new HashMap<>();
        IndexInput in = dir.openInput(file, IOContext.DEFAULT);
        try {
          CodecUtil.checkHeader(in, CODEC_NAME, VERSION_START, VERSION_START);
          int count = in.readVInt();
          for (int i = 0; i < count; i++) {
            long commitGen = in.readVLong();
            int refCount = in.readVInt();
            m.put(commitGen, refCount);
          }
        } catch (IOException ioe2) {
          // 保存第一個(gè)捕獲到的異常
          if (ioe == null) {
            ioe = ioe2;
          }
        } finally {
          in.close();
        }
        genLoaded = gen;
        // 清除舊數(shù)據(jù)  
        refCounts.clear();
        // 保留最新的  
        refCounts.putAll(m);
      }
    }
  }
  if (genLoaded == -1) { // 沒(méi)有加載快照文件
    if (ioe != null) { // 加載過(guò)程中捕獲到異常了，直接拋出
      throw ioe;
    }
  } else { // 把舊版本的快照文件都刪掉
    if (snapshotFiles.size() > 1) {
      String curFileName = SNAPSHOTS_PREFIX + genLoaded;
      for (String file : snapshotFiles) {
        if (!curFileName.equals(file)) {
          IOUtils.deleteFilesIgnoringExceptions(dir, file);
        }
      }
    }
    nextWriteGen = 1 + genLoaded;
  }
}