Spring JdbcTemplate实现通用的泛型dao三：构建动态sql

http://www.dexcoder.com/selfly/article/431

构建动态sql，其实说白了就是拼装sql语句，在这里我把传入的实体参数，属性有值的拼装进sql，为null的则忽略，要实现这个不用说，肯定要利用Java的反射功能。

来看一个具有代表性的insert语句的构建：

 
 
/**
 * 构建insert语句
 * 
 * @param entity 实体映射对象
 * @param nameHandler 名称转换处理器
 * @return
 */
public static SqlContext buildInsertSql(Object entity, NameHandler nameHandler) {
    Class<?> clazz = entity.getClass();
    String tableName = nameHandler.getTableName(clazz.getSimpleName());
    String primaryName = nameHandler.getPrimaryName(clazz.getSimpleName());
    StringBuilder sql = new StringBuilder("insert into ");
    List<Object> params = new ArrayList<Object>();
    sql.append(tableName);
    //获取属性信息
    BeanInfo beanInfo = ClassUtils.getSelfBeanInfo(clazz);
    PropertyDescriptor[] pds = beanInfo.getPropertyDescriptors();
    sql.append("(");
    StringBuilder args = new StringBuilder();
    args.append("(");
    for (PropertyDescriptor pd : pds) {
        Object value = getReadMethodValue(pd.getReadMethod(), entity);
        if (value == null) {
            continue;
        }
        sql.append(nameHandler.getColumnName(pd.getName()));
        args.append("?");
        params.add(value);
        sql.append(",");
        args.append(",");
    }
    sql.deleteCharAt(sql.length() - 1);
    args.deleteCharAt(args.length() - 1);
    args.append(")");
    sql.append(")");
    sql.append(" values ");
    sql.append(args);
    return new SqlContext(sql, primaryName, params);
}

众所周知，Java的反射是性能较低的，也有性能较好的第三方实现如cglib，这里并没有使用。在我的实测中两者差距不大。

但是注意这里并没有使用属性的操作方式，也就是没有使用jdk反射获取属性的getDeclaredFields()方法，而是使用了BeanInfo和PropertyDescriptor，因为后者的运行效率要远远高于前者。

在我的实测中，构建一个拥有12个属性的JavaBean的动态sql，十万次所耗时间为900毫秒左右，完全可以接受。当然，这里对JavaBean的信息进行了缓存，如果不缓存时间将多耗上几个数量级。

下面顺便贴上完整的代码：

 
 
/**
 * sql辅助为类
 * 
 * User: liyd
 * Date: 2/13/14
 * Time: 10:03 AM
 */
public class SqlUtils {
    /** 日志对象 */
    private static final Logger LOG = LoggerFactory.getLogger(SqlUtils.class);
    /**
     * 构建insert语句
     *
     * @param entity 实体映射对象
     * @param nameHandler 名称转换处理器
     * @return
     */
    public static SqlContext buildInsertSql(Object entity, NameHandler nameHandler) {
        Class<?> clazz = entity.getClass();
        String tableName = nameHandler.getTableName(clazz.getSimpleName());
        String primaryName = nameHandler.getPrimaryName(clazz.getSimpleName());
        StringBuilder sql = new StringBuilder("insert into ");
        List<Object> params = new ArrayList<Object>();
        sql.append(tableName);
        //获取属性信息
        BeanInfo beanInfo = ClassUtils.getSelfBeanInfo(clazz);
        PropertyDescriptor[] pds = beanInfo.getPropertyDescriptors();
        sql.append("(");
        StringBuilder args = new StringBuilder();
        args.append("(");
        for (PropertyDescriptor pd : pds) {
            Object value = getReadMethodValue(pd.getReadMethod(), entity);
            if (value == null) {
                continue;
            }
            sql.append(nameHandler.getColumnName(pd.getName()));
            args.append("?");
            params.add(value);
            sql.append(",");
            args.append(",");
        }
        sql.deleteCharAt(sql.length() - 1);
        args.deleteCharAt(args.length() - 1);
        args.append(")");
        sql.append(")");
        sql.append(" values ");
        sql.append(args);
        return new SqlContext(sql, primaryName, params);
    }
    /**
     * 构建更新sql
     * 
     * @param entity
     * @param nameHandler
     * @return
     */
    public static SqlContext buildUpdateSql(Object entity, NameHandler nameHandler) {
        Class<?> clazz = entity.getClass();
        StringBuilder sql = new StringBuilder();
        List<Object> params = new ArrayList<Object>();
        String tableName = nameHandler.getTableName(clazz.getSimpleName());
        String primaryName = nameHandler.getPrimaryName(clazz.getSimpleName());
        //获取属性信息
        BeanInfo beanInfo = ClassUtils.getSelfBeanInfo(clazz);
        PropertyDescriptor[] pds = beanInfo.getPropertyDescriptors();
        sql.append("update ");
        sql.append(tableName);
        sql.append(" set ");
        Object primaryValue = null;
        for (PropertyDescriptor pd : pds) {
            Object value = getReadMethodValue(pd.getReadMethod(), entity);
            if (value == null) {
                continue;
            }
            String columnName = nameHandler.getColumnName(pd.getName());
            if (primaryName.equalsIgnoreCase(columnName)) {
                primaryValue = value;
            }
            sql.append(columnName);
            sql.append(" = ");
            sql.append("?");
            params.add(value);
            sql.append(",");
        }
        sql.deleteCharAt(sql.length() - 1);
        sql.append(" where ");
        sql.append(primaryName);
        sql.append(" = ?");
        params.add(primaryValue);
        return new SqlContext(sql, primaryName, params);
    }
    /**
     * 构建查询条件
     * 
     * @param entity
     * @param nameHandler
     */
    public static SqlContext buildQueryCondition(Object entity, NameHandler nameHandler) {
        //获取属性信息
        BeanInfo beanInfo = ClassUtils.getSelfBeanInfo(entity.getClass());
        //        PropertyDescriptor[] pds = BeanUtils.getPropertyDescriptors(entityClass);
        PropertyDescriptor[] pds = beanInfo.getPropertyDescriptors();
        StringBuilder condition = new StringBuilder();
        List<Object> params = new ArrayList<Object>();
        int count = 0;
        for (PropertyDescriptor pd : pds) {
            Object value = getReadMethodValue(pd.getReadMethod(), entity);
            if (value == null) {
                continue;
            }
            if (count > 0) {
                condition.append(" and ");
            }
            condition.append(nameHandler.getColumnName(pd.getName()));
            condition.append(" = ?");
            params.add(value);
            count++;
        }
        return new SqlContext(condition, null, params);
    }
    /**
     * 获取属性值
     *
     * @param readMethod
     * @param entity
     * @return
     */
    private static Object getReadMethodValue(Method readMethod, Object entity) {
        if (readMethod == null) {
            return null;
        }
        try {
            if (!Modifier.isPublic(readMethod.getDeclaringClass().getModifiers())) {
                readMethod.setAccessible(true);
            }
            return readMethod.invoke(entity);
        } catch (Exception e) {
            LOG.error("获取属性值失败", e);
            throw new MincoderException(e);
        }
    }
}

获取BeanInfo时写了一个ClassUtils来实现，里面对Bean信息进行了缓存。因为项目使用spring，本来想使用spring提供的BeanUtils.getPropertyDescriptor()方法的，里面同样拥有缓存，但是该方法会把实体类父类的属性信息也获取出来，而PropertyDescriptor中又没法判断，这将直接导致拼装sql时字段的错误，因为你不知道哪些字段是操作当前表所需要的。没办法，查看jdk本身的Introspector类，发现里面有如下方法定义：

 
 
public static BeanInfo getBeanInfo(Class<?> beanClass, Class<?> stopClass) throws IntrospectionException

即可以指定在哪个类停止获取属性，这正是我们需要的，可惜spring没有进行封装，只能自己实现了，参考了spring的实现，使用WeakHashMap来防止内存的溢出，及时清空Introspector本身的缓存：

 
 
/**
 * 类辅助
 *
 * User: liyd
 * Date: 2/12/14
 * Time: 10:08 PM
 */
public class ClassUtils {
    /** 日志对象 */
    private static final Logger               LOG        = LoggerFactory
                                                             .getLogger(ClassUtils.class);
    /**
     * Map keyed by class containing CachedIntrospectionResults.
     * Needs to be a WeakHashMap with WeakReferences as values to allow
     * for proper garbage collection in case of multiple class loaders.
     */
    private static final Map<Class, BeanInfo> classCache = Collections
                                                             .synchronizedMap(new WeakHashMap<Class, BeanInfo>());
    /**
     * 获取类本身的BeanInfo，不包含父类属性
     * 
     * @param clazz
     * @return
     */
    public static BeanInfo getSelfBeanInfo(Class<?> clazz) {
        try {
            BeanInfo beanInfo;
            if (classCache.get(clazz) == null) {
                beanInfo = Introspector.getBeanInfo(clazz, clazz.getSuperclass());
                classCache.put(clazz, beanInfo);
                // Immediately remove class from Introspector cache, to allow for proper
                // garbage collection on class loader shutdown - we cache it here anyway,
                // in a GC-friendly manner. In contrast to CachedIntrospectionResults,
                // Introspector does not use WeakReferences as values of its WeakHashMap!
                Class classToFlush = clazz;
                do {
                    Introspector.flushFromCaches(classToFlush);
                    classToFlush = classToFlush.getSuperclass();
                } while (classToFlush != null);
            } else {
                beanInfo = classCache.get(clazz);
            }
            return beanInfo;
        } catch (IntrospectionException e) {
            LOG.error("获取BeanInfo失败", e);
            throw new MincoderException(e);
        }
    }
    /**
     * 初始化实例
     * 
     * @param clazz
     * @return
     */
    public static Object newInstance(Class<?> clazz) {
        try {
            return clazz.newInstance();
        } catch (Exception e) {
            LOG.error("根据class创建实例失败", e);
            throw new MincoderException(e);
        }
    }
}

另外创建了对象SqlContext来保存构建后的sql和参数信息，定义如下：

 
 
/**
 * 执行sql的上下文内容
 * 
 * User: liyd
 * Date: 2/13/14
 * Time: 10:40 AM
 */
public class SqlContext {
    /** 执行的sql */
    private StringBuilder sql;
    /** 主键名称 */
    private String        primaryKey;
    /** 参数，对应sql中的?号 */
    private List<Object>  params;
    public SqlContext(StringBuilder sql, String primaryKey, List<Object> params) {
        this.sql = sql;
        this.primaryKey = primaryKey;
        this.params = params;
    }
	//getter setter 略
}