Github地址,欢迎star和follow
新增android sqlite native 的代码
我们在使用android提供的SQLite存储数据的时候,就会用到SQLiteOpenHelper和SQLiteDataBase,但查询数据的时候会得到一个Cursor对象,这里我们将深入android提供的关于SQLite的封装以原理。
SQLiteOpenHelper
——封装管理数据库的创造和版本管理类
主要封装了数据库的创建和获取的方法,一般继承该类实现onCreate()、onUpdate()方法,在onCreate创建数据库,在onUpdate进行数据库升级操作。其中还有onConfigure()、onDowngrade()、onOpen()方法,将会在下面获取数据库对象分析进行解析
- 数据库的获取:
两个方法:getReadableDatabase()、getWritableDatabase()。需要注意的一点是这两个方法都加锁,是线程安全的。这两个方法最终调用getDatabaseLocked(boolean writable):
private SQLiteDatabase getDatabaseLocked(boolean writable) {
if (mDatabase != null) {
if (!mDatabase.isOpen()) { // 判断数据库是否已经关闭
// Darn! The user closed the database by calling mDatabase.close().
mDatabase = null;
} else if (!writable || !mDatabase.isReadOnly()) { //判断数据库是否符合要求,如果数据库可读可写则返回,即!mDatabase.isReadOnly()一直为true
// The database is already open for business.
return mDatabase;
}
}
// 正在初始化中
if (mIsInitializing) {
throw new IllegalStateException("getDatabase called recursively");
}
SQLiteDatabase db = mDatabase;
try {
mIsInitializing = true;
if (db != null) { // 数据库不为null,需要重新开启读写数据库使得符合要求
if (writable && db.isReadOnly()) {
db.reopenReadWrite();
}
} else if (mName == null) {
db = SQLiteDatabase.create(null);
} else {
try {
if (DEBUG_STRICT_READONLY && !writable) {
final String path = mContext.getDatabasePath(mName).getPath();
db = SQLiteDatabase.openDatabase(path, mFactory,
SQLiteDatabase.OPEN_READONLY, mErrorHandler);
} else {
// 通过mContext.openOrCreateDatabase创建数据库,其实还是调用SQLiteDatabase.openDatabase(..)创建数据库
db = mContext.openOrCreateDatabase(mName, mEnableWriteAheadLogging ?
Context.MODE_ENABLE_WRITE_AHEAD_LOGGING : 0,
mFactory, mErrorHandler);
}
} catch (SQLiteException ex) {
if (writable) {
throw ex;
}
Log.e(TAG, "Couldn't open " + mName
+ " for writing (will try read-only):", ex);
final String path = mContext.getDatabasePath(mName).getPath();
db = SQLiteDatabase.openDatabase(path, mFactory,
SQLiteDatabase.OPEN_READONLY, mErrorHandler);
}
}
// 调用onConfigure
onConfigure(db);
final int version = db.getVersion();
if (version != mNewVersion) {
if (db.isReadOnly()) {
throw new SQLiteException("Can't upgrade read-only database from version " +
db.getVersion() + " to " + mNewVersion + ": " + mName);
}
db.beginTransaction();
try {
// 当第一次创建数据库时DataBase的版本为0,会调用onCreate()方法
if (version == 0) {
onCreate(db);
} else { // 判断数据库版本升降级,调用相应方法
if (version > mNewVersion) {
onDowngrade(db, version, mNewVersion);
} else {
onUpgrade(db, version, mNewVersion);
}
}
db.setVersion(mNewVersion);
db.setTransactionSuccessful();
} finally {
db.endTransaction();
}
}
// 调用onOpen()方法
onOpen(db);
if (db.isReadOnly()) {
Log.w(TAG, "Opened " + mName + " in read-only mode");
}
mDatabase = db;
return db;
} finally {
mIsInitializing = false;
// 数据库创建失败时,进行close操作
if (db != null && db != mDatabase) {
db.close();
}
}
}
onCreate()、onUpdate()、onConfigure()、onDowngrade()、onOpen()方法的调用规则:
onConfigure():
当第一次调用getReadableDatabase()或者getWritableDatabase()会调用onConfigure(),如果第一是获取到只读形式的数据库,当转换成可写形式数据库时会再次调用onConfigure()。onCreate()
mDatabase第一次创建时会调用onCreate()onUpdate() / onDowngrade()
在版本改变时会调用相应的onUpdate()或onDowngrade()方法,onConfigure()
至于onOpen()的调用规则同onConfigure()。
那么onConfigure()和onOpen()方法可以干嘛呢,从api文档可以看到:
可以在onConfigure开启SQLite的WAL模式,以及设置外键的支持;
而onOpen()方法是说明数据库已经打开,可以进行一些自己的操作,但是需要通过SQLiteDatabase#isReadOnly方法检查数据库是否真正打开了
开启WAL方法:setWriteAheadLoggingEnabled(boolean enabled)
WAL支持读写并发,是通过将修改的数据单独写到一个wal文件中,默认在到达一个checkpoint时会将数据合并入主数据库中
至于关于WAL的详细介绍和分析可以参见SQLite3性能深入分析](http://blog.xcodev.com/posts/sqlite3-performance-indeep/)
SQLiteDatabase
open
获取SQLiteDatabase对象,从上面可以看到getReadableDatabase()、getWritableDatabase()是通过SQLiteDatabase.openDatabase(..)创建数据库,那么其中包含那些细节呢?
public static SQLiteDatabase openDatabase(String path, CursorFactory factory, int flags,
DatabaseErrorHandler errorHandler) {
SQLiteDatabase db = new SQLiteDatabase(path, flags, factory, errorHandler);
db.open();
return db;
}
可以看到new一个SQLiteDatabase对象,并调用open(),再返回该数据库对象,先看open()函数:
open():
private void open() {
try {
try {
openInner();
} catch (SQLiteDatabaseCorruptException ex) {
onCorruption();
openInner();
}
} catch (SQLiteException ex) {
// ....
}
}
private void openInner() {
synchronized (mLock) {
assert mConnectionPoolLocked == null;
mConnectionPoolLocked = SQLiteConnectionPool.open(mConfigurationLocked);
mCloseGuardLocked.open("close");
}
synchronized (sActiveDatabases) {
sActiveDatabases.put(this, null);
}
}
// 可以看到调用SQLiteConnectionPool.open(mConfigurationLocked):
public static SQLiteConnectionPool open(SQLiteDatabaseConfiguration configuration) {
if (configuration == null) {
throw new IllegalArgumentException("configuration must not be null.");
}
// Create the pool.
SQLiteConnectionPool pool = new SQLiteConnectionPool(configuration);
pool.open(); // might throw
return pool;
}
// 可以看到其中是创建一个SQLiteConnectionPool,并且调用open操作:
// Might throw
private void open() {
// Open the primary connection.
// This might throw if the database is corrupt.
mAvailablePrimaryConnection = openConnectionLocked(mConfiguration,
true /*primaryConnection*/); // might throw
// ...
}
// 可以看到创建了主连接mAvailablePrimaryConnection:
private SQLiteConnection openConnectionLocked(SQLiteDatabaseConfiguration configuration,
boolean primaryConnection) {
final int connectionId = mNextConnectionId++;
return SQLiteConnection.open(this, configuration,
connectionId, primaryConnection); // might throw
}
// 调用了SQLiteConnection.open()创建主连接:
static SQLiteConnection open(SQLiteConnectionPool pool,
SQLiteDatabaseConfiguration configuration,
int connectionId, boolean primaryConnection) {
SQLiteConnection connection = new SQLiteConnection(pool, configuration,
connectionId, primaryConnection);
try {
connection.open();
return connection;
} catch (SQLiteException ex) {
connection.dispose(false);
throw ex;
}
}
private void open() {
mConnectionPtr = nativeOpen(mConfiguration.path, mConfiguration.openFlags,
mConfiguration.label,
SQLiteDebug.DEBUG_SQL_STATEMENTS, SQLiteDebug.DEBUG_SQL_TIME);
setPageSize();
setForeignKeyModeFromConfiguration();
setWalModeFromConfiguration();
setJournalSizeLimit();
setAutoCheckpointInterval();
setLocaleFromConfiguration();
// Register custom functions.
final int functionCount = mConfiguration.customFunctions.size();
for (int i = 0; i < functionCount; i++) {
SQLiteCustomFunction function = mConfiguration.customFunctions.get(i);
nativeRegisterCustomFunction(mConnectionPtr, function);
}
}
// 可以看到最终调用了nativeOpen打开一个主数据库连接,并且设置各自sqlite的属性。
创建流程:
可以看出,创建一个数据库对象,会创建一个数据库连接池,并且会创建出一个主连接
数据库连接池用于管理数据库连接对象
而数据库连接SQLiteConnection则在其中包装了native的sqlite3对象,数据库sql语句最终会通过sqlite3对象执行可以看出,创建一个数据库对象,会创建一个数据库连接池,并且会创建出一个主连接
数据库连接池用于管理数据库连接对象
而数据库连接SQLiteConnection则在其中包装了native的sqlite3对象,数据库sql语句最终会通过sqlite3对象执行
insert
那么接下来就可以对数据库进行一些CRUD操作
先分析一下insert()和insertOrThrow()插入函数:
// 最终会调用insertWithOnConflict
public long insertWithOnConflict(String table, String nullColumnHack,
ContentValues initialValues, int conflictAlgorithm) {
acquireReference();
try {
StringBuilder sql = new StringBuilder();
// 构造insert SQL语句
// 创建SQLiteStatement对象,并调用executeInsert()
SQLiteStatement statement = new SQLiteStatement(this, sql.toString(), bindArgs);
try {
return statement.executeInsert();
} finally {
statement.close();
}
} finally {
releaseReference();
}
}
// SQLiteStatement::executeInsert():
public long executeInsert() {
acquireReference();
try {
return getSession().executeForLastInsertedRowId(
getSql(), getBindArgs(), getConnectionFlags(), null);
} catch (SQLiteDatabaseCorruptException ex) {
onCorruption();
throw ex;
} finally {
releaseReference();
}
}
// getSession()调用的是mDatabase.getThreadSession(),获取到SQLiteSession对象:
// SQLiteSession::executeForLastInsertedRowId():
public long executeForLastInsertedRowId(String sql, Object[] bindArgs, int connectionFlags,
CancellationSignal cancellationSignal) {
// 验证判断
// 获取一个数据库连接
acquireConnection(sql, connectionFlags, cancellationSignal); // might throw
try {
// 执行sql语句
return mConnection.executeForLastInsertedRowId(sql, bindArgs,
cancellationSignal); // might throw
} finally {
releaseConnection(); // might throw
}
}
// SQLiteConnection::executeForLastInsertedRowId():
public long executeForLastInsertedRowId(String sql, Object[] bindArgs,
CancellationSignal cancellationSignal) {
if (sql == null) {
throw new IllegalArgumentException("sql must not be null.");
}
final int cookie = mRecentOperations.beginOperation("executeForLastInsertedRowId",
sql, bindArgs);
try {
final PreparedStatement statement = acquirePreparedStatement(sql);
try {
throwIfStatementForbidden(statement);
// 绑定数据参数
bindArguments(statement, bindArgs);
applyBlockGuardPolicy(statement);
attachCancellationSignal(cancellationSignal);
try {
// 调用native执行sql语句
return nativeExecuteForLastInsertedRowId(
mConnectionPtr, statement.mStatementPtr);
} finally {
detachCancellationSignal(cancellationSignal);
}
} finally {
releasePreparedStatement(statement);
}
} catch (RuntimeException ex) {
mRecentOperations.failOperation(cookie, ex);
throw ex;
} finally {
mRecentOperations.endOperation(cookie);
}
}
流程图:
这里有几个需要注意一下:
- SQLiteSession:
private final ThreadLocal<SQLiteSession> mThreadSession = new ThreadLocal<SQLiteSession>() {
@Override
protected SQLiteSession initialValue() {
return createSession();
}
};
每个线程都拥有自己的SQLiteSession对象。多个线程进行数据操作的时候需要注意和处理保持数据的原子性
- SQLiteStatement
SQLiteStatement类代表一个sql语句,其父类为SQLiteProgram,从上面可以看到,insert操作会先构造出SQLiteStatement,其构造方法:
SQLiteProgram(SQLiteDatabase db, String sql, Object[] bindArgs,
CancellationSignal cancellationSignalForPrepare) {
mDatabase = db;
mSql = sql.trim();
int n = DatabaseUtils.getSqlStatementType(mSql);
switch (n) {
case DatabaseUtils.STATEMENT_BEGIN:
case DatabaseUtils.STATEMENT_COMMIT:
case DatabaseUtils.STATEMENT_ABORT:
mReadOnly = false;
mColumnNames = EMPTY_STRING_ARRAY;
mNumParameters = 0;
break;
default:
boolean assumeReadOnly = (n == DatabaseUtils.STATEMENT_SELECT);
SQLiteStatementInfo info = new SQLiteStatementInfo();
db.getThreadSession().prepare(mSql,
db.getThreadDefaultConnectionFlags(assumeReadOnly),
cancellationSignalForPrepare, info);
mReadOnly = info.readOnly;
mColumnNames = info.columnNames;
mNumParameters = info.numParameters;
break;
}
// 参数初始化操作
}
可以看到其会调用SQLiteSession::prepare()操作,又是转发到SQLiteConnection::prepare()操作,进行SQL语法预编译,并会返回行列信息到SQLiteStatementInfo中。
再看下插入函数public long executeForLastInsertedRowId(String sql, Object[] bindArgs, CancellationSignal cancellationSignal)
通过前面的SQLiteStatement
将sql语句和参数组成sql并传递进来,通过PreparedStatement acquirePreparedStatement(String sql)
获取PreparedStatement
对象,再通过nativeExecuteForLastInsertedRowId( mConnectionPtr, statement.mStatementPtr)
native方法执行sql语句。
在获取PreparedStatement
的时候,可以看到PreparedStatement通过一个mPreparedStatementCache来进行缓存操作,具体是一个LruCache<String, PreparedStatement>
来完成sql的缓存
replace、delete
同理的操作有replace()、replaceOrThrow、delete、updateupdateWithOnConflict、execSQL等函数。
读者可按照前面思路分析
query
现在重点分析一下SQLiteDatabase的查询操作:
从源码可以看出查询操作最终会调用rawQueryWithFactory():
public Cursor rawQueryWithFactory(
CursorFactory cursorFactory, String sql, String[] selectionArgs,
String editTable, CancellationSignal cancellationSignal) {
acquireReference();
try {
SQLiteCursorDriver driver = new SQLiteDirectCursorDriver(this, sql, editTable,
cancellationSignal);
return driver.query(cursorFactory != null ? cursorFactory : mCursorFactory,
selectionArgs);
} finally {
releaseReference();
}
}
可以看出先构造出SQLiteDirectCursorDriver,再调用其query操作:
// SQLiteDirectCursorDriver::query():
public Cursor query(CursorFactory factory, String[] selectionArgs) {
final SQLiteQuery query = new SQLiteQuery(mDatabase, mSql, mCancellationSignal);
final Cursor cursor;
try {
query.bindAllArgsAsStrings(selectionArgs);
if (factory == null) {
cursor = new SQLiteCursor(this, mEditTable, query);
} else {
cursor = factory.newCursor(mDatabase, this, mEditTable, query);
}
} catch (RuntimeException ex) {
query.close();
throw ex;
}
mQuery = query;
return cursor;
}
流程图:
可以看出先构造出SQLiteQuery,在构造出SQLiteCursor,并返回SQLiteCursor对象。
所以得到的Cursor的原型是SQLiteCursor类,你会发现没有其他操作,那么查询数据是在哪里呢?
SQLiteCursor分析:
public final boolean moveToFirst() {
return moveToPosition(0);
}
public final boolean moveToPosition(int position) {
// Make sure position isn't past the end of the cursor
final int count = getCount();
if (position >= count) {
mPos = count;
return false;
}
// Make sure position isn't before the beginning of the cursor
if (position < 0) {
mPos = -1;
return false;
}
// Check for no-op moves, and skip the rest of the work for them
if (position == mPos) {
return true;
}
boolean result = onMove(mPos, position);
if (result == false) {
mPos = -1;
} else {
mPos = position;
}
return result;
}
public int getCount() {
if (mCount == NO_COUNT) {
fillWindow(0);
}
return mCount;
}
private void fillWindow(int requiredPos) {
clearOrCreateWindow(getDatabase().getPath());
try {
if (mCount == NO_COUNT) {
int startPos = DatabaseUtils.cursorPickFillWindowStartPosition(requiredPos, 0);
mCount = mQuery.fillWindow(mWindow, startPos, requiredPos, true);
mCursorWindowCapacity = mWindow.getNumRows();
if (Log.isLoggable(TAG, Log.DEBUG)) {
Log.d(TAG, "received count(*) from native_fill_window: " + mCount);
}
} else {
int startPos = DatabaseUtils.cursorPickFillWindowStartPosition(requiredPos,
mCursorWindowCapacity);
mQuery.fillWindow(mWindow, startPos, requiredPos, false);
}
} catch (RuntimeException ex) {
// Close the cursor window if the query failed and therefore will
// not produce any results. This helps to avoid accidentally leaking
// the cursor window if the client does not correctly handle exceptions
// and fails to close the cursor.
closeWindow();
throw ex;
}
}
protected void clearOrCreateWindow(String name) {
if (mWindow == null) {
mWindow = new CursorWindow(name);
} else {
mWindow.clear();
}
}
到这里你会发现CursorWindow,那这个对象是干嘛的呢?从文档上看可以发现其保存查询数据库的缓存,那么数据是缓存在哪的呢?先看器构造器:
public CursorWindow(String name) {
// ...
mWindowPtr = nativeCreate(mName, sCursorWindowSize);
// ..
}
nativeCreate通过JNI调用CursorWindow.cpp的create():
status_t CursorWindow::create(const String8& name, size_t size, CursorWindow** outCursorWindow) {
String8 ashmemName("CursorWindow: ");
ashmemName.append(name);
status_t result;
// 创建共享内存
int ashmemFd = ashmem_create_region(ashmemName.string(), size);
if (ashmemFd < 0) {
result = -errno;
} else {
result = ashmem_set_prot_region(ashmemFd, PROT_READ | PROT_WRITE);
if (result >= 0) {
// 内存映射
void* data = ::mmap(NULL, size, PROT_READ | PROT_WRITE, MAP_SHARED, ashmemFd, 0);
// ...
}
*outCursorWindow = NULL;
return result;
}
可以看到查询数据是通过创建共享内存来保存的,但是数据在哪里被保存了呢?
继续分析上面SQLiteCursor:: fillWindow()函数:
mQuery.fillWindow(mWindow, startPos, requiredPos, true);
其最终会调用SQLiteConnection::executeForCursorWindow,也是通过JNI调用cpp文件将查询数据保存到共享内存中。
至于共享内存的知识点,可以参考 Android系统匿名共享内存Ashmem
总结
经过上面分析,关于数据库的操作应该有了大致的了解:
当然里面也有些地方也是可以加以改善,取得更好的效率。
而且你会发现SQLiteConnection里面包含许多native方法,通过jni与sqlite进行交互,除了在android提供的sqlite库的基础上优化之外,也可以基于SQLiteConnection,甚至是完全使用c++来实现数据库的封装也是可以的
最后,如果本文有哪些地方不足或者错误,还请指出,谢谢。