深刻Mybatis源碼——執行流程

前言

上一篇分析Mybatis是如何加載解析XML文件的,本篇緊接上文,分析Mybatis的剩餘兩個階段:代理封裝SQL執行java

正文

代理封裝

Mybatis有兩種方式調用Mapper接口:sql

private static SqlSessionFactory sqlMapper = new SqlSessionFactoryBuilder().build(reader);

// 第一種
try (SqlSession session = sqlMapper.openSession(TransactionIsolationLevel.SERIALIZABLE)) {
  Blog blog = session.selectOne("org.apache.ibatis.domain.blog.mappers.BlogMapper.selectBlogWithPostsUsingSubSelect", 1);
}

// 第二種
try (SqlSession session = sqlMapper.openSession()) {
  AuthorMapper mapper = session.getMapper(AuthorMapper.class);
  Author author = mapper.selectAuthor(101);
}

從上面代碼能夠看到不管是哪種首先都要建立SqlSessionFactory對象,而後經過這個對象拿到SqlSession對象。在早期版本中只能經過該對象的增刪改調用Mapper接口,很明顯這種方式可讀性不好,難以維護,寫起來也複雜,因此後面谷歌開始維護Mybatis後,從新封裝提供了第二種方式直接調用Mapper接口。不過本質上第二種是在第一種的基礎之上實現的,因此下面就以第二種爲主進行分析,進入到getMapper方法:數據庫

public <T> T getMapper(Class<T> type) {
    return configuration.<T>getMapper(type, this);
  }

  public <T> T getMapper(Class<T> type, SqlSession sqlSession) {
    return mapperRegistry.getMapper(type, sqlSession);
  }

  public <T> T getMapper(Class<T> type, SqlSession sqlSession) {
    final MapperProxyFactory<T> mapperProxyFactory = (MapperProxyFactory<T>) knownMappers.get(type);
    if (mapperProxyFactory == null) {
      throw new BindingException("Type " + type + " is not known to the MapperRegistry.");
    }
    try {
      return mapperProxyFactory.newInstance(sqlSession);
    } catch (Exception e) {
      throw new BindingException("Error getting mapper instance. Cause: " + e, e);
    }
  }

mapperRegistry對象在上一篇分析過,是在解析xml中的mapper節點時註冊進去的,而這個對象中緩存了Mapper接口和對應的代理工廠的映射,因此getMapper的核心就是經過這個工廠去建立代理對象:apache

public T newInstance(SqlSession sqlSession) {
	 //每次調用都會建立新的MapperProxy對象
    final MapperProxy<T> mapperProxy = new MapperProxy<>(sqlSession, mapperInterface, methodCache);
    return newInstance(mapperProxy);
  }

而後經過Mapper接口調用時首先就會調用到MapperProxyinvoke方法:數組

public Object invoke(Object proxy, Method method, Object[] args) throws Throwable {
    try {
      if (Object.class.equals(method.getDeclaringClass())) {//若是是Object自己的方法不加強
        return method.invoke(this, args);
      } else if (isDefaultMethod(method)) {
        return invokeDefaultMethod(proxy, method, args);
      }
    } catch (Throwable t) {
      throw ExceptionUtil.unwrapThrowable(t);
    }
    //從緩存中獲取mapperMethod對象,若是緩存中沒有,則建立一個,並添加到緩存中
    final MapperMethod mapperMethod = cachedMapperMethod(method);
    //調用execute方法執行sql
    return mapperMethod.execute(sqlSession, args);
  }

  private MapperMethod cachedMapperMethod(Method method) {
    return methodCache.computeIfAbsent(method, k -> new MapperMethod(mapperInterface, method, sqlSession.getConfiguration()));
  }

首先從緩存中拿到MapperMethod對象,這個對象封裝了SQL語句的類型、命名空間、入參、返回類型等信息,而後經過它的execute方法調用SqlSession的增刪查改方法:緩存

public Object execute(SqlSession sqlSession, Object[] args) {
    Object result;
    //根據sql語句類型以及接口返回的參數選擇調用不一樣的
    switch (command.getType()) {
      case INSERT: {
    	Object param = method.convertArgsToSqlCommandParam(args);
        result = rowCountResult(sqlSession.insert(command.getName(), param));
        break;
      }
      case UPDATE: {
        Object param = method.convertArgsToSqlCommandParam(args);
        result = rowCountResult(sqlSession.update(command.getName(), param));
        break;
      }
      case DELETE: {
        Object param = method.convertArgsToSqlCommandParam(args);
        result = rowCountResult(sqlSession.delete(command.getName(), param));
        break;
      }
      case SELECT:
        if (method.returnsVoid() && method.hasResultHandler()) {//返回值爲void
          executeWithResultHandler(sqlSession, args);
          result = null;
        } else if (method.returnsMany()) {//返回值爲集合或者數組
          result = executeForMany(sqlSession, args);
        } else if (method.returnsMap()) {//返回值爲map
          result = executeForMap(sqlSession, args);
        } else if (method.returnsCursor()) {//返回值爲遊標
          result = executeForCursor(sqlSession, args);
        } else {//處理返回爲單一對象的狀況
          //經過參數解析器解析解析參數
          Object param = method.convertArgsToSqlCommandParam(args);
          result = sqlSession.selectOne(command.getName(), param);
          if (method.returnsOptional() &&
              (result == null || !method.getReturnType().equals(result.getClass()))) {
            result = OptionalUtil.ofNullable(result);
          }
        }
        break;
      case FLUSH:
        result = sqlSession.flushStatements();
        break;
      default:
        throw new BindingException("Unknown execution method for: " + command.getName());
    }
    if (result == null && method.getReturnType().isPrimitive() && !method.returnsVoid()) {
      throw new BindingException("Mapper method '" + command.getName() 
          + " attempted to return null from a method with a primitive return type (" + method.getReturnType() + ").");
    }
    return result;
  }

上文說過SqlSession本質上是門面模式的體現,其本質上是經過Executor執行器組件實現的,在該組件中定義了全部訪問數據庫的方法:session

public <E> List<E> selectList(String statement, Object parameter, RowBounds rowBounds) {
    try {
      //從configuration中獲取要執行的sql語句的配置信息
      MappedStatement ms = configuration.getMappedStatement(statement);
      //經過executor執行語句,並返回指定的結果集
      return executor.query(ms, wrapCollection(parameter), rowBounds, Executor.NO_RESULT_HANDLER);
    } catch (Exception e) {
      throw ExceptionFactory.wrapException("Error querying database.  Cause: " + e, e);
    } finally {
      ErrorContext.instance().reset();
    }
  }

Executor對象是在獲取SqlSession時建立的:mybatis

public SqlSession openSession() {
    return openSessionFromDataSource(configuration.getDefaultExecutorType(), null, false);
  }

  private SqlSession openSessionFromDataSource(ExecutorType execType, TransactionIsolationLevel level, boolean autoCommit) {
    Transaction tx = null;
    try {
    	//獲取mybatis配置文件中的environment對象
      final Environment environment = configuration.getEnvironment();
      //從environment獲取transactionFactory對象
      final TransactionFactory transactionFactory = getTransactionFactoryFromEnvironment(environment);
      //建立事務對象
      tx = transactionFactory.newTransaction(environment.getDataSource(), level, autoCommit);
      //根據配置建立executor
      final Executor executor = configuration.newExecutor(tx, execType);
      //建立DefaultSqlSession
      return new DefaultSqlSession(configuration, executor, autoCommit);
    } catch (Exception e) {
      closeTransaction(tx); // may have fetched a connection so lets call close()
      throw ExceptionFactory.wrapException("Error opening session.  Cause: " + e, e);
    } finally {
      ErrorContext.instance().reset();
    }
  }

TransactionFactory是咱們在xml中配置的transactionManager屬性,可選的屬性有JDBC和Managed,而後根據咱們的配置建立事務對象,以後纔是建立Executor對象。app

public Executor newExecutor(Transaction transaction, ExecutorType executorType) {
    executorType = executorType == null ? defaultExecutorType : executorType;
    executorType = executorType == null ? ExecutorType.SIMPLE : executorType;
    Executor executor;
    if (ExecutorType.BATCH == executorType) {
      executor = new BatchExecutor(this, transaction);
    } else if (ExecutorType.REUSE == executorType) {
      executor = new ReuseExecutor(this, transaction);
    } else {
      executor = new SimpleExecutor(this, transaction);
    }
    //若是有<cache>節點,經過裝飾器,添加二級緩存的能力
    if (cacheEnabled) {
      executor = new CachingExecutor(executor);
    }
    //經過interceptorChain遍歷全部的插件爲executor加強,添加插件的功能
    executor = (Executor) interceptorChain.pluginAll(executor);
    return executor;
  }

Executor有三個基本的實現類:dom

  • BatchExecutor:批處理執行器,執行批量更新、插入等操做。
  • ReuseExecutor:可重用執行器,緩存並重用Statement(Statement、PreparedStatement、CallableStatement)。
  • SimpleExecutor:默認使用的執行器,每次執行都會建立 新的Statement

這三個執行器都繼承了自抽象的BaseExecutor,同時若是開啓了二級緩存功能,在這裏還會裝飾一個CachingExecutor爲其添加二級緩存的能力。另外還要注意在這段代碼的最後還有攔截器進行了包裝,也就是擴展插件的實現 ,關於這部份內容在一篇進行分析。

SQL執行

二級緩存的代碼很簡單,這裏直接略過,因此直接進入到BaseExecutor.query方法:

public <E> List<E> query(MappedStatement ms, Object parameter, RowBounds rowBounds, ResultHandler resultHandler) throws SQLException {
	//獲取sql語句信息,包括佔位符,參數等信息
    BoundSql boundSql = ms.getBoundSql(parameter);
    //拼裝緩存的key值
    CacheKey key = createCacheKey(ms, parameter, rowBounds, boundSql);
    return query(ms, parameter, rowBounds, resultHandler, key, boundSql);
 }

  public <E> List<E> query(MappedStatement ms, Object parameter, RowBounds rowBounds, ResultHandler resultHandler, CacheKey key, BoundSql boundSql) throws SQLException {
    ErrorContext.instance().resource(ms.getResource()).activity("executing a query").object(ms.getId());
    if (closed) {//檢查當前executor是否關閉
      throw new ExecutorException("Executor was closed.");
    }
    if (queryStack == 0 && ms.isFlushCacheRequired()) {//非嵌套查詢,而且FlushCache配置爲true,則須要清空一級緩存
      clearLocalCache();
    }
    List<E> list;
    try {
      queryStack++;//查詢層次加一
      list = resultHandler == null ? (List<E>) localCache.getObject(key) : null;//查詢以及緩存
      if (list != null) {
    	 //針對調用存儲過程的結果處理
        handleLocallyCachedOutputParameters(ms, key, parameter, boundSql);
      } else {
    	 //緩存未命中,從數據庫加載數據
        list = queryFromDatabase(ms, parameter, rowBounds, resultHandler, key, boundSql);
      }
    } finally {
      queryStack--;
    }
    
    
    if (queryStack == 0) {
      for (DeferredLoad deferredLoad : deferredLoads) {//延遲加載處理
        deferredLoad.load();
      }
      // issue #601
      deferredLoads.clear();
      if (configuration.getLocalCacheScope() == LocalCacheScope.STATEMENT) {//若是當前sql的一級緩存配置爲STATEMENT,查詢完既清空一集緩存
        // issue #482
        clearLocalCache();
      }
    }
    return list;
  }

首先從一級緩存localCache裏面拿,若是沒有,才真正地訪問數據庫,並將返回結果存入到一級緩存中。

private <E> List<E> queryFromDatabase(MappedStatement ms, Object parameter, RowBounds rowBounds, ResultHandler resultHandler, CacheKey key, BoundSql boundSql) throws SQLException {
    List<E> list;
    localCache.putObject(key, EXECUTION_PLACEHOLDER);//在緩存中添加佔位符
    try {
      //調用抽象方法doQuery,方法查詢數據庫並返回結果,可選的實現包括:simple、reuse、batch
      list = doQuery(ms, parameter, rowBounds, resultHandler, boundSql);
    } finally {
      localCache.removeObject(key);//在緩存中刪除佔位符
    }
    localCache.putObject(key, list);//將真正的結果對象添加到一級緩存
    if (ms.getStatementType() == StatementType.CALLABLE) {//若是是調用存儲過程
      localOutputParameterCache.putObject(key, parameter);//緩存輸出類型結果參數
    }
    return list;
  }

這裏的doQuery是子類實現的,即模板模式,以SimpleExecutor爲例:

public <E> List<E> doQuery(MappedStatement ms, Object parameter, RowBounds rowBounds, ResultHandler resultHandler, BoundSql boundSql) throws SQLException {
    Statement stmt = null;
    try {
      Configuration configuration = ms.getConfiguration();//獲取configuration對象
      //建立StatementHandler對象,
      StatementHandler handler = configuration.newStatementHandler(wrapper, ms, parameter, rowBounds, resultHandler, boundSql);
      //StatementHandler對象建立stmt,並使用parameterHandler對佔位符進行處理
      stmt = prepareStatement(handler, ms.getStatementLog());
      //經過statementHandler對象調用ResultSetHandler將結果集轉化爲指定對象返回
      return handler.<E>query(stmt, resultHandler);
    } finally {
      closeStatement(stmt);
    }
  }

通讀這裏的代碼咱們能夠發現,Executor自己是不會訪問到數據庫,而是做爲指揮官,指揮三個小弟幹事:

  • StatementHandler:建立PreparedStatementStatementCallableStatement對象。
  • ParameterHandler:在StatementHandler構造函數中建立,對預編譯的 SQL 語句進行參數設置。
  • ResultSetHandler:在StatementHandler構造函數中建立,對數據庫返回的結果集(ResultSet)進行封裝,返回用戶指定的實體類型。

上面三個對象都是在configuration.newStatementHandler方法中建立的,而後調用prepareStatement拿到合適的Statement,若是是預編譯的還會進行參數設置:

private Statement prepareStatement(StatementHandler handler, Log statementLog) throws SQLException {
    Statement stmt;
    //獲取connection對象的動態代理,添加日誌能力;
    Connection connection = getConnection(statementLog);
    //經過不一樣的StatementHandler,利用connection建立(prepare)Statement
    stmt = handler.prepare(connection, transaction.getTimeout());
    //使用parameterHandler處理佔位符
    handler.parameterize(stmt);
    return stmt;
  }

若是在DEBUG模式下拿到的Connection對象是ConnectionLogger,這就和第一篇的內容串聯起來了。以後再經過query方法調用execute執行SQL語句,並使用ResultSetHandler處理結果集:

public List<Object> handleResultSets(Statement stmt) throws SQLException {
    ErrorContext.instance().activity("handling results").object(mappedStatement.getId());
    //用於保存結果集對象
    final List<Object> multipleResults = new ArrayList<>();

    int resultSetCount = 0;
    //statment可能返回多個結果集對象,這裏先取出第一個結果集
    ResultSetWrapper rsw = getFirstResultSet(stmt);
    //獲取結果集對應resultMap,本質就是獲取字段與java屬性的映射規則
    List<ResultMap> resultMaps = mappedStatement.getResultMaps();
    int resultMapCount = resultMaps.size();
    validateResultMapsCount(rsw, resultMapCount);//結果集和resultMap不能爲空,爲空拋出異常
    while (rsw != null && resultMapCount > resultSetCount) {
     //獲取當前結果集對應的resultMap
      ResultMap resultMap = resultMaps.get(resultSetCount);
      //根據映射規則(resultMap)對結果集進行轉化,轉換成目標對象之後放入multipleResults中
      handleResultSet(rsw, resultMap, multipleResults, null);
      rsw = getNextResultSet(stmt);//獲取下一個結果集
      cleanUpAfterHandlingResultSet();//清空nestedResultObjects對象
      resultSetCount++;
    }
    //獲取多結果集。多結果集通常出如今存儲過程的執行,存儲過程返回多個resultset,
    //mappedStatement.resultSets屬性列出多個結果集的名稱,用逗號分割;
    //多結果集的處理不是重點,暫時不分析
    String[] resultSets = mappedStatement.getResultSets();
    if (resultSets != null) {
      while (rsw != null && resultSetCount < resultSets.length) {
        ResultMapping parentMapping = nextResultMaps.get(resultSets[resultSetCount]);
        if (parentMapping != null) {
          String nestedResultMapId = parentMapping.getNestedResultMapId();
          ResultMap resultMap = configuration.getResultMap(nestedResultMapId);
          handleResultSet(rsw, resultMap, null, parentMapping);
        }
        rsw = getNextResultSet(stmt);
        cleanUpAfterHandlingResultSet();
        resultSetCount++;
      }
    }

    return collapseSingleResultList(multipleResults);
  }

這裏最終就是經過反射模塊以及Configuration類中的result相關配置進行結果映射:

private void handleResultSet(ResultSetWrapper rsw, ResultMap resultMap, List<Object> multipleResults, ResultMapping parentMapping) throws SQLException {
    try {
      if (parentMapping != null) {//處理多結果集的嵌套映射
        handleRowValues(rsw, resultMap, null, RowBounds.DEFAULT, parentMapping);
      } else {
        if (resultHandler == null) {//若是resultHandler爲空,實例化一我的默認的resultHandler
          DefaultResultHandler defaultResultHandler = new DefaultResultHandler(objectFactory);
          //對ResultSet進行映射,映射結果暫存在resultHandler中
          handleRowValues(rsw, resultMap, defaultResultHandler, rowBounds, null);
          //將暫存在resultHandler中的映射結果,填充到multipleResults
          multipleResults.add(defaultResultHandler.getResultList());
        } else {
          //使用指定的rusultHandler進行轉換
          handleRowValues(rsw, resultMap, resultHandler, rowBounds, null);
        }
      }
    } finally {
      // issue #228 (close resultsets)
      //調用resultset.close()關閉結果集
      closeResultSet(rsw.getResultSet());
    }
  }

  public void handleRowValues(ResultSetWrapper rsw, ResultMap resultMap, ResultHandler<?> resultHandler, RowBounds rowBounds, ResultMapping parentMapping) throws SQLException {
    if (resultMap.hasNestedResultMaps()) {//處理有嵌套resultmap的狀況
      ensureNoRowBounds();
      checkResultHandler();
      handleRowValuesForNestedResultMap(rsw, resultMap, resultHandler, rowBounds, parentMapping);
    } else {//處理沒有嵌套resultmap的狀況
      handleRowValuesForSimpleResultMap(rsw, resultMap, resultHandler, rowBounds, parentMapping);
    }
  }

  private void handleRowValuesForSimpleResultMap(ResultSetWrapper rsw, ResultMap resultMap, ResultHandler<?> resultHandler, RowBounds rowBounds, ResultMapping parentMapping)
      throws SQLException {
	//建立結果上下文,所謂的上下文就是專門在循環中緩存結果對象的
    DefaultResultContext<Object> resultContext = new DefaultResultContext<>();
    //1.根據分頁信息,定位到指定的記錄
    skipRows(rsw.getResultSet(), rowBounds);
    //2.shouldProcessMoreRows判斷是否須要映射後續的結果,實際仍是翻頁處理,避免超過limit
    while (shouldProcessMoreRows(resultContext, rowBounds) && rsw.getResultSet().next()) {
      //3.進一步完善resultMap信息,主要是處理鑑別器的信息
      ResultMap discriminatedResultMap = resolveDiscriminatedResultMap(rsw.getResultSet(), resultMap, null);
      //4.讀取resultSet中的一行記錄並進行映射,轉化並返回目標對象
      Object rowValue = getRowValue(rsw, discriminatedResultMap);
      //5.保存映射結果對象
      storeObject(resultHandler, resultContext, rowValue, parentMapping, rsw.getResultSet());
    }
  }

  private Object getRowValue(ResultSetWrapper rsw, ResultMap resultMap) throws SQLException {
    final ResultLoaderMap lazyLoader = new ResultLoaderMap();
    //4.1 根據resultMap的type屬性,實例化目標對象
    Object rowValue = createResultObject(rsw, resultMap, lazyLoader, null);
    if (rowValue != null && !hasTypeHandlerForResultObject(rsw, resultMap.getType())) {
      //4.2 對目標對象進行封裝獲得metaObjcect,爲後續的賦值操做作好準備
      final MetaObject metaObject = configuration.newMetaObject(rowValue);
      boolean foundValues = this.useConstructorMappings;//取得是否使用構造函數初始化屬性值
      if (shouldApplyAutomaticMappings(resultMap, false)) {//是否使用自動映射
    	 //4.3通常狀況下 autoMappingBehavior默認值爲PARTIAL,對未明確指定映射規則的字段進行自動映射
        foundValues = applyAutomaticMappings(rsw, resultMap, metaObject, null) || foundValues;
      }
       //4.4 映射resultMap中明確指定須要映射的列
      foundValues = applyPropertyMappings(rsw, resultMap, metaObject, lazyLoader, null) || foundValues;
      foundValues = lazyLoader.size() > 0 || foundValues;
      //4.5 若是沒有一個映射成功的屬性,則根據<returnInstanceForEmptyRow>的配置返回null或者結果對象
      rowValue = foundValues || configuration.isReturnInstanceForEmptyRow() ? rowValue : null;
    }
    return rowValue;
  }
  • 自動映射
private boolean applyAutomaticMappings(ResultSetWrapper rsw, ResultMap resultMap, MetaObject metaObject, String columnPrefix) throws SQLException {
	//獲取resultSet中存在的,可是ResultMap中沒有明確映射的列,填充至autoMapping中
    List<UnMappedColumnAutoMapping> autoMapping = createAutomaticMappings(rsw, resultMap, metaObject, columnPrefix);
    boolean foundValues = false;
    if (!autoMapping.isEmpty()) {
      //遍歷autoMapping,經過自動匹配的方式爲屬性複製
      for (UnMappedColumnAutoMapping mapping : autoMapping) {
    	//經過typeHandler從resultset中拿值
        final Object value = mapping.typeHandler.getResult(rsw.getResultSet(), mapping.column);
        if (value != null) {
          foundValues = true;
        }
        if (value != null || (configuration.isCallSettersOnNulls() && !mapping.primitive)) {
          // gcode issue #377, call setter on nulls (value is not 'found')
          //經過metaObject給屬性賦值
          metaObject.setValue(mapping.property, value);
        }
      }
    }
    return foundValues;
  }
  • 指定映射
private boolean applyPropertyMappings(ResultSetWrapper rsw, ResultMap resultMap, MetaObject metaObject, ResultLoaderMap lazyLoader, String columnPrefix)
      throws SQLException {
	//從resultMap中獲取明確須要轉換的列名集合
    final List<String> mappedColumnNames = rsw.getMappedColumnNames(resultMap, columnPrefix);
    boolean foundValues = false;
    //獲取ResultMapping集合
    final List<ResultMapping> propertyMappings = resultMap.getPropertyResultMappings();
    for (ResultMapping propertyMapping : propertyMappings) {
      String column = prependPrefix(propertyMapping.getColumn(), columnPrefix);//得到列名,注意前綴的處理
      if (propertyMapping.getNestedResultMapId() != null) {
        // the user added a column attribute to a nested result map, ignore it
    	//若是屬性經過另一個resultMap映射,則忽略
        column = null;
      }
      if (propertyMapping.isCompositeResult()//若是是嵌套查詢,column={prop1=col1,prop2=col2}
          || (column != null && mappedColumnNames.contains(column.toUpperCase(Locale.ENGLISH)))//基本類型映射
          || propertyMapping.getResultSet() != null) {//嵌套查詢的結果
    	//得到屬性值
        Object value = getPropertyMappingValue(rsw.getResultSet(), metaObject, propertyMapping, lazyLoader, columnPrefix);
        // issue #541 make property optional
        //得到屬性名稱
        final String property = propertyMapping.getProperty();
        if (property == null) {//屬性名爲空跳出循環
          continue;
        } else if (value == DEFERED) {//屬性名爲DEFERED,延遲加載的處理
          foundValues = true;
          continue;
        }
        if (value != null) {
          foundValues = true;
        }
        if (value != null || (configuration.isCallSettersOnNulls() && !metaObject.getSetterType(property).isPrimitive())) {
          // gcode issue #377, call setter on nulls (value is not 'found')
          //經過metaObject爲目標對象設置屬性值
          metaObject.setValue(property, value);
        }
      }
    }
    return foundValues;
  }

反射實例化對象的代碼比較長,但邏輯都比較清晰,上面的關鍵流程代碼也都加上了註釋,讀者可自行參照源碼閱讀。

總結

Mybatis核心原理就分析完了,相比較Spring源碼簡單了不少,但代碼的優雅度和優秀的設計思想一點也不亞於Spring,也是很是值得咱們好好學習掌握的。不過這3篇只是分析了Mybaits的核心執行原理,另外還有插件怎麼擴展、攔截器會攔截哪些方法以及Mybatis和Spring的整合又是怎麼實現的呢?讀者們能夠好好思考下,答案將在下一篇揭曉。

相關文章
相關標籤/搜索