1 数据库服务器连接
主要用于描述客户端是如何与数据库服务器建立连接,并且是如何执行sql语句的。
1.1 客户端连接处理(JDBCDriver)
1.1.1 描述
(1) 客户端通过JDBCDriver类创建JDBCConnection建立连接,最终通过ClientConnectionHttp、ClientConnction类创建一个与服务器类Server的socket连接。
(2) 上文已经描述了数据服务器启动的时候通过Server类、HsqlSoketFactory类创建了一个ServerSocket同时,通过accept方法循环监听客户端发起的请求。
(3) 当Server接收到客户端的请求时创建一个新的线程同时将该线程加入到线程组中,从代码看是创建了一个ServerConnection此类实现了runable接口。
此类的run方法实现了接收到请求后的具体处理。
核心类是ServerConnection类,下面是对ServerConnection类的详细描述。
1.1.2 JDBCDriver
public Connection connect(String url, Properties info) throws SQLException { if (url.regionMatches(true, 0, DatabaseURL.S_URL_INTERNAL, 0, DatabaseURL.S_URL_INTERNAL.length())) { JDBCConnection conn = (JDBCConnection) threadConnection.get(); if (conn == null) { returnnull; } return conn; } return getConnection(url, info);//此方法实现对JDBCConnectio类的创建 }
1.1.3 JDBCConnection
public JDBCConnection(HsqlProperties props) throws SQLException { String user = props.getProperty("user"); String password = props.getProperty("password"); String connType = props.getProperty("connection_type"); String host = props.getProperty("host"); int port = props.getIntegerProperty("port", 0); String path = props.getProperty("path"); String database = props.getProperty("database"); boolean isTLS = (connType == DatabaseURL.S_HSQLS || connType == DatabaseURL.S_HTTPS); boolean isTLSWrapper = props.isPropertyTrue(HsqlDatabaseProperties.url_tls_wrapper, false); isTLSWrapper &= isTLS; if (user == null) { user = "SA"; } if (password == null) { password = ""; } Calendar cal = Calendar.getInstance(); int zoneSeconds = HsqlDateTime.getZoneSeconds(cal); try { if (DatabaseURL.isInProcessDatabaseType(connType)) { /** * @todo- fredt - this should be the only static reference to * a core class (apart form references to the Type package) * from the jdbc package - we might make it dynamic */ sessionProxy = DatabaseManager.newSession(connType, database, user, password, props, null, zoneSeconds); } elseif (connType == DatabaseURL.S_HSQL || connType == DatabaseURL.S_HSQLS) { sessionProxy = new ClientConnection(host, port, path, database, isTLS, isTLSWrapper, user, password, zoneSeconds); isNetConn = true; } elseif (connType == DatabaseURL.S_HTTP || connType == DatabaseURL.S_HTTPS) { sessionProxy = new ClientConnectionHTTP(host, port, path, database, isTLS, isTLSWrapper, user, password, zoneSeconds); isNetConn = true; } else { // alias: type not yet implemented throw JDBCUtil.invalidArgument(connType); } sessionProxy.setJDBCConnection(this); connProperties = props; clientProperties = sessionProxy.getClientProperties(); if (connProperties != null) { isCloseResultSet = connProperties.isPropertyTrue( HsqlDatabaseProperties.url_close_result, false); isUseColumnName = connProperties.isPropertyTrue( HsqlDatabaseProperties.url_get_column_name, true); } } catch (HsqlException e) { throw JDBCUtil.sqlException(e); } }
1.1.4 ClientConnection->构造方法
/** * Establishes a connection to the server. */ public ClientConnection(String host, int port, String path, String database, boolean isTLS, boolean isTLSWrapper, String user, String password, int timeZoneSeconds) { this.host = host; this.port = port; this.path = path; this.database = database; this.isTLS = isTLS; this.isTLSWrapper = isTLSWrapper; this.zoneSeconds = timeZoneSeconds; this.zoneString = TimeZone.getDefault().getID(); initStructures(); Result login = Result.newConnectionAttemptRequest(user, password, database, zoneString, timeZoneSeconds); initConnection(host, port, isTLS); Result resultIn = execute(login); if (resultIn.isError()) { throw Error.error(resultIn); } sessionID = resultIn.getSessionId(); databaseID = resultIn.getDatabaseId(); databaseUniqueName = resultIn.getDatabaseName(); clientPropertiesString = resultIn.getMainString(); }
1.1.5 ClientConnection-> initConnection
protectedvoid initConnection(String host, int port, boolean isTLS) { openConnection(host, port, isTLS); } protectedvoid openConnection(String host, int port, boolean isTLS) { try { if (isTLSWrapper) { socket = HsqlSocketFactory.getInstance(false).createSocket(host, port); } socket = HsqlSocketFactory.getInstance(isTLS).createSocket(socket, host, port); //建立一个soket连接 socket.setTcpNoDelay(true); dataOutput = new DataOutputStream(socket.getOutputStream()); dataInput = new DataInputStream( new BufferedInputStream(socket.getInputStream())); handshake(); } catch (Exception e) { // The details from "e" should not be thrown away here. This is // very useful info for end users to diagnose the runtime problem. thrownew HsqlException(e, Error.getStateString(ErrorCode.X_08001), -ErrorCode.X_08001); } }
1.1.6 ServerConnection –>Run方法
ServerConnection类是在服务器端接收到客户端请求后创建的一个线程类,此类用于实现对客户端请求的处理。
/** * Initializes this connection and runs the request handling * loop until closed. */ publicvoid run() { int msgType; init(); if (session != null) { try { while (keepAlive) { msgType = dataInput.readByte(); if (msgType < ResultConstants.MODE_UPPER_LIMIT) { receiveResult(msgType); } else { receiveOdbcPacket((char) msgType); } } } catch (CleanExit ce) { keepAlive = false; } catch (IOException e) { // fredt - is thrown when connection drops server.printWithThread(mThread + ":disconnected " + user); } catch (HsqlException e) { // fredt - is thrown in unforeseen circumstances if (keepAlive) { server.printStackTrace(e); } } catch (Throwable e) { // fredt - is thrown in unforeseen circumstances if (keepAlive) { server.printStackTrace(e); } } } close(); }
1.1.7 ServerConnection->init初始化
主要工作是获取客户端soket发起的请求数据,同时对请求数据进行转换包装,将数据包装Result,同时创建Session对象,Session对象是通过DatabaseMananger创建实现的。
/** * Initializes this connection. * <p> * Will return (not throw) if fail to initialize the connection. * </p> */ privatevoid init() { runnerThread = Thread.currentThread(); keepAlive = true; try { socket.setTcpNoDelay(true); dataInput = new DataInputStream( new BufferedInputStream(socket.getInputStream())); dataOutput = new DataOutputStream(socket.getOutputStream()); int firstInt = handshake(); switch (streamProtocol) { caseHSQL_STREAM_PROTOCOL : if (firstInt != ClientConnection .NETWORK_COMPATIBILITY_VERSION_INT) { if (firstInt == -1900000) { firstInt = -2000000; } String verString = ClientConnection.toNetCompVersionString(firstInt); throw Error.error( null, ErrorCode.SERVER_VERSIONS_INCOMPATIBLE, 0, new String[] { verString, HsqlDatabaseProperties.THIS_VERSION }); } Result resultIn = Result.newResult(dataInput, rowIn); resultIn.readAdditionalResults(session, dataInput, rowIn); Result resultOut; resultOut = setDatabase(resultIn); resultOut.write(session, dataOutput, rowOut); break; caseODBC_STREAM_PROTOCOL : odbcConnect(firstInt); break; default : // Protocol detection failures should already have been // handled. keepAlive = false; } } catch (Exception e) { // Only "unexpected" failures are caught here. // Expected failures will have been handled (by sending feedback // to user-- with an output Result for normal protocols), then // continuing. StringBuffer sb = new StringBuffer(mThread + ":Failed to connect client."); if (user != null) { sb.append(" User '" + user + "'."); } server.printWithThread(sb.toString() + " Stack trace follows."); server.printStackTrace(e); }
Session的创建,session首先通过databasemanager调用database,然后由database类调用SessionManager类调用session类创建一个session,Session是用于与数据库进行交互的基于JDBC的session级别的连接。
private Result setDatabase(Result resultIn) { try { String databaseName = resultIn.getDatabaseName(); dbIndex = server.getDBIndex(databaseName); dbID = server.dbID[dbIndex]; user = resultIn.getMainString(); if (!server.isSilent()) { server.printWithThread(mThread + ":Trying to connect user '" + user + "' to DB (" + databaseName + ')'); } session = DatabaseManager.newSession(dbID, user, resultIn.getSubString(), resultIn.getZoneString(), resultIn.getUpdateCount()); if (!server.isSilent()) { server.printWithThread(mThread + ":Connected user '" + user + "'"); } return Result.newConnectionAcknowledgeResponse( session.getDatabase(), session.getId(), session.getDatabase().getDatabaseID()); } catch (HsqlException e) { session = null; return Result.newErrorResult(e); } catch (RuntimeException e) { session = null; return Result.newErrorResult(e); } }
1.1.8 ServerConnection-> receiveResult接收结果集
主要的工作是输入信息进行类型的格式化,创建Result对象,然后对输入结果进行处理生成返回的结果集。
privatevoid receiveResult(int resultMode) throws CleanExit, IOException { boolean terminate = false; Result resultIn = Result.newResult(session, resultMode, dataInput, rowIn); resultIn.readLobResults(session, dataInput, rowIn); server.printRequest(mThread, resultIn); Result resultOut = null; switch (resultIn.getType()) { case ResultConstants.CONNECT : { resultOut = setDatabase(resultIn); break; } case ResultConstants.DISCONNECT : { resultOut = Result.updateZeroResult; terminate = true; break; } case ResultConstants.RESETSESSION : { session.resetSession(); resultOut = Result.updateZeroResult; break; } case ResultConstants.EXECUTE_INVALID : { resultOut = Result.newErrorResult(Error.error(ErrorCode.X_07502)); break; } default : { resultOut = session.execute(resultIn); break; } } resultOut.write(session, dataOutput, rowOut); rowOut.reset(mainBuffer); rowIn.resetRow(mainBuffer.length); if (terminate) { throwcleanExit; } }
1.1.9 Session ->execute返回结果构造
此操作由Session类实现,根据不同的请求类型,比如是建立连接,执行语句,执行预编译语句,执行更新,或者其他类型的语句,根据不同的类型执行不同的操作。
/** * Executes the command encapsulated by the cmd argument. * * @param cmd the command to execute * @return the result of executing the command */ publicsynchronized Result execute(Result cmd) { Object[] pvals = (Object[]) cmd.valueData; Result result = executeCompiledStatement(cs, pvals, cmd.queryTimeout); result = performPostExecute(cmd, result); return result; } }
1.1.10Session-> executeCompiledStatement语句的执行
最终的执行还是通过statement接口的实现类实现对结果的查询,更新,添加等操作的。
具体处理间statement章节,此步骤关联到2.2.5 章节。
1.2 数据库语句执行持久化流程(JDBCStatement)
1.2.1 JDBCStatement-> executeBatch
实现SQL语句的执行。
publicsynchronizedint[] executeBatch() throws SQLException { if (batchResultOut == null) { batchResultOut = Result.newBatchedExecuteRequest(); } try { resultIn = connection.sessionProxy.execute(batchResultOut); //此处的sessionProxy是session类的接口,session的初始化是在建立数据库连接的时候初始化的 performPostExecute(); } catch (HsqlException e) { batchResultOut.getNavigator().clear(); throw JDBCUtil.sqlException(e); } return updateCounts; }
1.2.2 JDBCStatement-> executeQuery
publicsynchronized ResultSet executeQuery( String sql) throws SQLException { fetchResult(sql, StatementTypes.RETURN_RESULT, JDBCStatementBase.NO_GENERATED_KEYS, null, null); return getResultSet(); }
1.2.3 JDBCStatement-> executeUpdate
publicsynchronizedint executeUpdate(String sql) throws SQLException { fetchResult(sql, StatementTypes.RETURN_COUNT, JDBCStatementBase.NO_GENERATED_KEYS, null, null); returnresultIn.getUpdateCount(); }
1.2.4 JDBCStatement->fetchResult
privatevoid fetchResult(String sql, int statementRetType, int generatedKeys, int[] generatedIndexes, String[] generatedNames) throws SQLException { if (isEscapeProcessing) { sql = connection.nativeSQL(sql); } resultOut.setPrepareOrExecuteProperties(sql, maxRows, fetchSize, statementRetType, queryTimeout, rsProperties, generatedKeys, generatedIndexes, generatedNames); //将sql转换为Result标准对象 try { resultIn = connection.sessionProxy.execute(resultOut); //此处的sessionProxy是session类的接口,session的初始化是在建立数据库连接的时候初始化的。 performPostExecute(); } catch (HsqlException e) { throw JDBCUtil.sqlException(e); } }
1.2.5 Session
实现对一个数据库连接上操作的管理。
/** * Executes the command encapsulated by the cmd argument. * * @param cmd the command to execute * @return the result of executing the command */ publicsynchronized Result execute(Result cmd) { Object[] pvals = (Object[]) cmd.valueData; Result result = executeCompiledStatement(cs, pvals, cmd.queryTimeout); result = performPostExecute(cmd, result); } public Result executeCompiledStatement(Statement cs, Object[] pvals, int timeout) { Result r; r = cs.execute(this); }
1.2.6 StatementCommand
SQL命令Statement接口的实现类,这里的execute方法就是一般我们JDBC直接调用的方法,此方法的关键调用类是logger.java类,此类实现了对数据库涉及的所有文件的操作,包括文件的读、写、备份、新建、删除等。
public Result execute(Session session) { Result result; try { result = getResult(session); } catch (Throwable t) { result = Result.newErrorResult(t, null); } if (result.isError()) { result.getException().setStatementType(group, type); return result; } try { if (isLogged) { session.database.logger.writeOtherStatement(session, sql); } } catch (Throwable e) { return Result.newErrorResult(e, sql); } return result; }
------具体处理结果的方法
Result getResult(Session session) { try { session.checkAdmin(); session.checkDDLWrite(); session.database.logger.checkpoint(defrag); return Result.updateZeroResult; } catch (HsqlException e) { return Result.newErrorResult(e, sql); } finally { session.database.lobManager.unlock(); } }
1.2.7 Logger.java
用于实现打开数据库文件信息
/** * Performs checkpoint including pre and post operations. Returns to the * same state as before the checkpoint. */ void checkpoint(boolean defrag) { if (filesReadOnly) { return; } if (cache == null) { defrag = false; } elseif (forceDefrag()) { defrag = true; } // test code /* if (database.logger.isStoredFileAccess) { defrag = false; } */ if (defrag) { defrag(); } else { checkpoint(); } } /** * Writes out all the rows to a new file without fragmentation. */ publicvoid defrag() { database.logger.logInfoEvent("defrag start"); try { synchLog(); database.lobManager.synch(); deleteOldDataFiles(); DataFileDefrag dfd = cache.defrag(); database.persistentStoreCollection.setNewTableSpaces(); database.sessionManager.resetLoggedSchemas(); } catch (HsqlException e) { throw e; } catch (Throwable e) { database.logger.logSevereEvent("defrag failure", e); throw Error.error(ErrorCode.DATA_FILE_ERROR, e); }
1.2.8 DataFileCache
DataFileCache作为将缓存数据持久化到文件中的管理类
DataFileDefrag defrag() { writeLock.lock(); try { cache.saveAll(); DataFileDefrag dfd = new DataFileDefrag(database, this, dataFileName); dfd.process(); close(); cache.clear(); if (!database.logger.propIncrementBackup) { backupNewDataFile(true); } database.schemaManager.setTempIndexRoots(dfd.getIndexRoots()); try { database.logger.log.writeScript(false); } finally { database.schemaManager.setTempIndexRoots(null); } database.getProperties().setProperty( HsqlDatabaseProperties.hsqldb_script_format, database.logger.propScriptFormat); database.getProperties().setDBModified( HsqlDatabaseProperties.FILES_MODIFIED_NEW); database.logger.log.closeLog(); database.logger.log.deleteLog(); database.logger.log.renameNewScript(); renameBackupFile(); renameDataFile(); database.getProperties().setDBModified( HsqlDatabaseProperties.FILES_NOT_MODIFIED); open(false); database.schemaManager.setIndexRoots(dfd.getIndexRoots()); if (database.logger.log.dbLogWriter != null) { database.logger.log.openLog(); } return dfd; } finally { writeLock.unlock(); } }
1.2.9 DataFileDefrag
实现根据索引查找指定的数据块,同时将数据持久化到文件中。
--- 索引的处理
void process() { Throwable error = null; database.logger.logDetailEvent("Defrag process begins"); HsqlArrayList allTables = database.schemaManager.getAllTables(true); rootsList = newlong[allTables.size()][]; long maxSize = 0; for (int i = 0, tSize = allTables.size(); i < tSize; i++) { Table table = (Table) allTables.get(i); if (table.getTableType() == TableBase.CACHED_TABLE) { PersistentStore store = database.persistentStoreCollection.getStore(table); long size = store.elementCount(); if (size > maxSize) { maxSize = size; } } } if (maxSize > Integer.MAX_VALUE) { throw Error.error(ErrorCode.X_2200T); } try { pointerLookup = new DoubleIntIndex((int) maxSize, false); dataFileOut = new DataFileCache(database, dataFileName, true); pointerLookup.setKeysSearchTarget(); for (int i = 0, tSize = allTables.size(); i < tSize; i++) { Table t = (Table) allTables.get(i); if (t.getTableType() == TableBase.CACHED_TABLE) { long[] rootsArray = writeTableToDataFile(t); rootsList[i] = rootsArray; } else { rootsList[i] = null; } database.logger.logDetailEvent("table complete " + t.getName().name); } dataFileOut.fileModified = true; dataFileOut.close(); dataFileOut = null; for (int i = 0, size = rootsList.length; i < size; i++) { long[] roots = rootsList[i]; if (roots != null) { database.logger.logDetailEvent("roots: " + StringUtil.getList(roots, ",", "")); } } } catch (OutOfMemoryError e) { error = e; throw Error.error(ErrorCode.OUT_OF_MEMORY, e); } catch (Throwable t) { error = t; throw Error.error(ErrorCode.GENERAL_ERROR, t); } finally { try { if (dataFileOut != null) { dataFileOut.release(); } } catch (Throwable t) {} if (error instanceof OutOfMemoryError) { database.logger.logInfoEvent( "defrag failed - out of memory - required: " + maxSize * 8); } if (error == null) { database.logger.logDetailEvent("Defrag transfer complete: " + stopw.elapsedTime()); } else { database.logger.logSevereEvent("defrag failed ", error); database.logger.getFileAccess().removeElement(dataFileName + Logger.newFileExtension); } } }
-----数据的持久化
long[] writeTableToDataFile(Table table) { PersistentStore store = table.database.persistentStoreCollection.getStore(table); TableSpaceManager space = dataFileOut.spaceManager.getTableSpace(table.getSpaceID()); long[] rootsArray = table.getIndexRootsArray(); pointerLookup.clear(); database.logger.logDetailEvent("lookup begins " + table.getName().name + " " + stopw.elapsedTime()); RowStoreAVLDisk.moveDataToSpace(store, dataFileOut, space, pointerLookup); for (int i = 0; i < table.getIndexCount(); i++) { if (rootsArray[i] == -1) { continue; } long pos = pointerLookup.lookup(rootsArray[i], -1); if (pos == -1) { throw Error.error(ErrorCode.DATA_FILE_ERROR); } rootsArray[i] = pos; } // log any discrepency in row count long count = rootsArray[table.getIndexCount() * 2]; if (count != pointerLookup.size()) { database.logger.logSevereEvent("discrepency in row count " + table.getName().name + " " + count + " " + pointerLookup.size(), null); } rootsArray[table.getIndexCount()] = 0; rootsArray[table.getIndexCount() * 2] = pointerLookup.size(); database.logger.logDetailEvent("table written " + table.getName().name); return rootsArray; }
1.2.10 DataFileCache
实现缓存对象的处理,是对缓存表持久化处理的管理类。
public DataFileCache(Database db, String baseFileName, boolean defrag) { initParams(db, baseFileName, true); cache = new Cache(this); try { if (database.logger.isStoredFileAccess()) { dataFile = RAFile.newScaledRAFile(database, dataFileName, false, RAFile.DATA_FILE_STORED); } else { dataFile = new RAFileSimple(database, dataFileName, "rw"); } } catch (Throwable t) { throw Error.error(ErrorCode.FILE_IO_ERROR, t); } initNewFile(); initBuffers(); if (database.logger.isDataFileSpaces()) { spaceManager = new DataSpaceManagerBlocks(this); } else { spaceManager = new DataSpaceManagerSimple(this); } }
1.2.1 RowStoreAVLDiskData
获取文件指针,持久化数据
publicstaticvoid moveDataToSpace(PersistentStore store, DataFileCache cache, TableSpaceManager tableSpace, LongLookup pointerLookup) { RowIterator it = store.rowIterator(); while (it.hasNext()) { CachedObject row = it.getNextRow(); long newPos = tableSpace.getFilePosition(row.getStorageSize(), false); pointerLookup.addUnsorted(row.getPos(), newPos); } it = store.rowIterator(); while (it.hasNext()) { CachedObject row = it.getNextRow(); cache.rowOut.reset(); row.write(cache.rowOut, pointerLookup); long pos = pointerLookup.lookup(row.getPos()); cache.saveRowOutput(pos); } }
1.2.2 DataFileCache
将数据写到文件中
publicvoid saveRowOutput(long pos) { try { dataFile.seek(pos * dataFileScale); dataFile.write(rowOut.getOutputStream().getBuffer(), 0, rowOut.getOutputStream().size()); } catch (Throwable t) { logSevereEvent("DataFileCache.saveRowOutput", t, pos); throw Error.error(ErrorCode.DATA_FILE_ERROR, t); } }
1.2.3 RAFileSimple
此类是用于实现对文件进行随机访问的包装类,此类为一个抽象类,此类包含的实现类有
RAFileHybrid.java,RAFileInJar.java,RAFileNIO.java, RAFileSimple RandomAccessFile file = new RandomAccessFile(name, openMode); publicvoid write(byte[] b, int off, int len) throws IOException { file.write(b, off, len); }
相关推荐
例如,如果项目中存在`HSQLDB`的jar,Spring Boot会自动配置一个内存数据库。 3. **嵌入式Web服务器**:Spring Boot支持内嵌Tomcat、Jetty或Undertow等Web服务器,无需额外部署,简化了开发流程。 4. **Actuator**...
SpringBoot经典学习笔记是针对Java开发者的宝贵资源,它涵盖了SpringBoot的核心概念、特性以及实践应用。SpringBoot是由Pivotal团队开发的框架,旨在简化Spring应用的初始搭建以及开发过程,通过“约定优于配置”的...
在"SpringBoot经典学习笔记"中,你可能会了解到以下关键知识点: 1. **起步依赖(Starter Dependencies)**:SpringBoot通过starter依赖来简化构建配置,比如`spring-boot-starter-web`用于Web应用,`spring-boot-...
**Spring Boot核心技术详解** Spring Boot是由Pivotal团队提供的全新框架,其设计目标是为了简化Spring应用的...希望这份学习笔记能帮助你深入理解和掌握Spring Boot的精髓,祝你在Spring Boot的学习之路上一帆风顺!
HSQL(HyperSQL)是一个开源的关系型数据库管理系统,它支持内存模式、文件模式以及服务器模式,能够处理各种类型的数据,包括文本数据库。这篇笔记将探讨如何使用HSQL处理文本数据库,结合标签"源码"和"工具",我们...
### jBoss + Tomcat 学习笔记大全及帮助文档概览 #### 一、环境搭建与配置 在开始深入探讨jBoss与Tomcat的集成之前,我们先来了解如何搭建基本的开发环境。 ##### 1. Java 环境配置 - **安装 J2SDK1.4+ 和 J2...
在"springboot-guides"的学习笔记中,我们可以深入探讨以下几个核心知识点: 1. **起步依赖(Starters)**: SpringBoot通过起步依赖来简化Maven或Gradle的配置,每个起步依赖都是一个包含相关依赖的模块集合,如...
6. **Cloud Connectors**:Spring Boot Cloud Connectors帮助在云环境中快速连接到数据库、消息队列等服务,使得应用能在不同云平台上无缝迁移。 7. **YAML/Properties配置**:Spring Boot支持使用YAML或Properties...
例如,如果在类路径下发现了`HSQLDB`数据库,Spring Boot将自动配置一个内存数据库连接。 3. **内嵌Web服务器(Embedded Web Server)** Spring Boot支持内嵌Tomcat、Jetty或Undertow等Web服务器,无需额外安装和...