? Jakarta-Common-BeanUtils研究心得
一、概述
第一次看到BeanUtils包，是在Struts項目中，作為Struts一個工具來使用的，
估計功能越弄越強，就移到Common項目中了吧。

BeanUtils一共有四個package:
org.apache.commons.beanutils
org.apache.commons.beanutils.converters
org.apache.commons.beanutils.locale
org.apache.commons.beanutils.locale.converters
后三個包主要是用于數據的轉換，圍繞著一個Converter接口，該接口只有一個方法：
java.lang.Object convert(java.lang.Class type, java.lang.Object value) ，
用于將一個value轉換成另一個類型為type的Object。在一些自動化的應用中應該會有用。
這里不作評論，以后有興趣了，或者覺得有用了，再行研究。
這里只講第一個包。

二、測試用的Bean
在開始所有的測試之前，我寫了一個簡單的Bean，以便于測試，代碼如下：
package test.jakarta.commons.beanutils;

/**
* @author SonyMusic
*
*/
public class Month {
private int value;
private String name;
private int[] days={11,22,33,44,55};

public Month(int v, String n){
value=v;
name=n;
}

/**
* Returns the name.
* @return String
*/
public String getName() {
return name;
}

/**
* Returns the value.
* @return int
*/
public int getValue() {
return value;
}

/**
* Sets the name.
* @param name The name to set
*/
public void setName(String name) {
this.name = name;
}

/**
* Sets the value.
* @param value The value to set
*/
public void setValue(int value) {
this.value = value;
}

/**
* @see java.lang.Object#toString()
*/
public String toString() {
return value+"("+name+")";
}

public int[] getDays() {
return days;
}

public void setDays(int[] is) {
days = is;
}

}

三、BeanUtils
這是一個主要應用于Bean的Util（呵呵，這個解釋很絕吧），以下是其中幾個方法的例子

//static java.util.Map describe(java.lang.Object bean)
//這個方法返回一個Object中所有的可讀屬性，并將屬性名/屬性值放入一個Map中，另外還有
//一個名為class的屬性，屬性值是Object的類名，事實上class是java.lang.Object的一個屬性
Month month=new Month(1, "Jan");

try {
Map map=BeanUtils.describe(month);
Set keySet=map.keySet();
for (Iterator iter = keySet.iterator(); iter.hasNext();) {
Object element = (Object) iter.next();
System.out.println("KeyClass:"+element.getClass().getName());
System.out.println("ValueClass:"+map.get(element).getClass().getName());
System.out.print(element+"\t");
System.out.print(map.get(element));
System.out.println();
}
} catch (IllegalAccessException e) {
e.printStackTrace();
} catch (InvocationTargetException e) {
e.printStackTrace();
} catch (NoSuchMethodException e) {
e.printStackTrace();
}
輸出為：
KeyClass:java.lang.String
ValueClass:java.lang.String
value 1
KeyClass:java.lang.String
ValueClass:java.lang.String
class class test.jakarta.commons.beanutils.Month
KeyClass:java.lang.String
ValueClass:java.lang.String
name Jan
注意到所有Map中的key/value都是String，而不管object中實際的值是多少。
與此對應的還有static void populate(java.lang.Object bean, java.util.Map properties)
用于將剛才describe出的Map再裝配成一個對象。

再看這樣一段代碼
曹曉鋼也許還記得，為了取一個不確定對象的property，著實花了不少時間，
難度不大，但要做到100%的正確，仍然需要付出很大的精力。
//static java.lang.String getProperty(java.lang.Object bean, java.lang.String name)
Month month=new Month(1, "Jan");

try {
System.out.println(BeanUtils.getProperty(month,"value"));
} catch (Exception e) {
e.printStackTrace();
}
//輸出是：1

與getProperty類似的還有getIndexedProperty, getMappedProperty，
以getIndexedProperty為例：
Month month=new Month(1, "Jan");

try {
System.out.println(BeanUtils.getIndexedProperty(month,"days",1));
System.out.println(BeanUtils.getIndexedProperty(month,"days[1]"));
} catch (Exception e) {
e.printStackTrace();
}
這兩個調用是相同的。

BeanUtils中還有一個方法：
static void copyProperties(java.lang.Object dest, java.lang.Object orig)
它真是太有用，還記得struts中滿天飛的都是copyProperties，我甚至懷疑整個BeanUtils最初
是不是因為這個方法的需求才寫出來的。
它將對象orig中的屬性復制到dest中去。

四、PropertyUtils
這個類和BeanUtils類很多的方法在參數上都是相同的，但返回值不同。
BeanUtils著重于"Bean"，返回值通常是String，而PropertyUtils著重于屬性，
它的返回值通常是Object。

五、ConstructorUtils
這個類中的方法主要分成兩種，一種是得到構造方法，一種是創建對象。
事實上多數時候得到構造方法的目的就是創建對象，這里只介紹一下創建對象。
//static java.lang.Object ConstructorUtils.invokeConstructor
//(java.lang.Class klass, java.lang.Object[] args)
//根據一個java.lang.Class以及相應的構造方法的參數，創建一個對象。
Object obj=ConstructorUtils.invokeConstructor(Month.class, {new Integer(1), "Jan"});
Month month=(Month)obj;
try {
System.out.println(BeanUtils.getProperty(month,"value"));
} catch (Exception e) {
e.printStackTrace();
}
輸出證明，構造方法的調用是成功的。
如果需要強制指定構造方法的參數類型，可以這樣調用：
Object[] args={new Integer(1), "Jan"};
Class[] argsType={int.class, String.class};
Object obj;
obj = ConstructorUtils.invokeExactConstructor(Month.class, args, argsType);
Month month=(Month)obj;
System.out.println(BeanUtils.getProperty(month,"value"));
argsType指定了參數的類型。

六、ConstructorUtils補遺
創建對象還有一個方法：invokeExactConstructor，該方法對參數要求
更加嚴格，傳遞進去的參數必須嚴格符合構造方法的參數列表。
例如：
Object[] args={new Integer(1), "Jan"};
Class[] argsType={int.class, String.class};
Object obj;
//下面這句調用將不會成功，因為args[0]的類型為Integer，而不是int
//obj = ConstructorUtils.invokeExactConstructor(Month.class, args);

//這一句就可以，因為argsType指定了類型。
obj = ConstructorUtils.invokeExactConstructor(Month.class, args, argsType);
Month month=(Month)obj;
System.out.println(BeanUtils.getProperty(month,"value"));

七、MethodUtils
與ConstructorUtils類似，不過調用的時候，通常需要再指定一個method name的參數。

八、DynaClass/DynaBean
這似乎是BeanUtils中最有趣的部分之一了，很簡單，簡單到光看這兩個接口中的方法會不明白
為什么要設計這兩個接口。不過看到ResultSetDynaClass后，就明白了。下面是java doc中的代碼：
?? ResultSet rs = ...;
?? ResultSetDynaClass rsdc = new ResultSetDynaClass(rs);
?? Iterator rows = rsdc.iterator();
?? while (rows.hasNext())? {
???? DynaBean row = (DynaBean) rows.next();
???? ... process this row ...
?? }
?? rs.close();
原來這是一個ResultSet的包裝器，ResultSetDynaClass實現了DynaClass，它的iterator方法返回一個
ResultSetIterator，則是實現了DynaBean接口。
在獲得一個DynaBean之后，我們就可以用
???? DynaBean row = (DynaBean) rows.next();
???? System.out.println(row.get("field1")); //field1是其中一個字段的名字

再看另一個類RowSetDynaClass的用法，代碼如下：
String driver="com.mysql.jdbc.Driver";
String url="jdbc:mysql://localhost/2hu?useUnicode=true&characterEncoding=GBK";
String username="root";
String password="";

java.sql.Connection con=null;
PreparedStatement ps=null;
ResultSet rs=null;
try {
Class.forName(driver).newInstance();
con = DriverManager.getConnection(url);
ps=con.prepareStatement("select * from forumlist");
rs=ps.executeQuery();
//先打印一下，用于檢驗后面的結果。
while(rs.next()){
System.out.println(rs.getString("name"));
}
rs.beforeFirst();//這里必須用beforeFirst，因為RowSetDynaClass只從當前位置向前滾動

RowSetDynaClass rsdc = new RowSetDynaClass(rs);
rs.close();
ps.close();
List rows = rsdc.getRows();//返回一個標準的List，存放的是DynaBean
for (int i = 0; i <rows.size(); i++) {
DynaBean b=(DynaBean)rows.get(i);
System.out.println(b.get("name"));
}
} catch (Exception e) {
e.printStackTrace();
}
finally{
try {
con.close();
} catch (Exception e) {
}
}

是不是很有趣？封裝了ResultSet的數據，代價是占用內存。如果一個表有10萬條記錄，rsdc.getRows()
就會返回10萬個記錄。@_@

需要注意的是ResultSetDynaClass和RowSetDynaClass的不同之處：
1，ResultSetDynaClass是基于Iterator的，一次只返回一條記錄，而RowSetDynaClass是基于
List的，一次性返回全部記錄。直接影響是在數據比較多時ResultSetDynaClass會比較的快速，
而RowSetDynaClass需要將ResultSet中的全部數據都讀出來（并存儲在其內部），會占用過多的
內存，并且速度也會比較慢。
2，ResultSetDynaClass一次只處理一條記錄，在處理完成之前，ResultSet不可以關閉。
3，ResultSetIterator的next()方法返回的DynaBean其實是指向其內部的一個固定
對象，在每次next()之后，內部的值都會被改變。這樣做的目的是節約內存，如果你需要保存每
次生成的DynaBean，就需要創建另一個DynaBean，并將數據復制過去，下面也是java doc中的代碼：
?? ArrayList results = new ArrayList(); // To hold copied list
?? ResultSetDynaClass rsdc = ...;
?? DynaProperty properties[] = rsdc.getDynaProperties();
?? BasicDynaClass bdc =
???? new BasicDynaClass("foo", BasicDynaBean.class,
??????????????????????? rsdc.getDynaProperties());
?? Iterator rows = rsdc.iterator();
?? while (rows.hasNext()) {
???? DynaBean oldRow = (DynaBean) rows.next();
???? DynaBean newRow = bdc.newInstance();
???? PropertyUtils.copyProperties(newRow, oldRow);
???? results.add(newRow);
?? }

事實上DynaClass/DynaBean可以用于很多地方，存儲各種類型的數據。自己想吧。嘿嘿。

九、自定義的CustomRowSetDynaClass
兩年前寫過一個與RowSetDynaClass目標相同的類，不過多一個功能，就是分頁，只取需要的數據，
這樣內存占用就會減少。

先看一段代碼：
String driver="com.mysql.jdbc.Driver";
String url="jdbc:mysql://localhost/2hu?useUnicode=true&characterEncoding=GBK";
String username="root";
String password="";

java.sql.Connection con=null;
PreparedStatement ps=null;
ResultSet rs=null;
try {
Class.forName(driver).newInstance();
con = DriverManager.getConnection(url);
ps=con.prepareStatement("select * from forumlist order by name");
rs=ps.executeQuery();
/*
while(rs.next()){
System.out.println(rs.getString("name"));
}
rs.beforeFirst();
*/

//第二個參數表示第幾頁，第三個參數表示頁的大小
CustomRowSetDynaClass rsdc = new CustomRowSetDynaClass(rs, 2, 5);
//RowSetDynaClass rsdc = new RowSetDynaClass(rs);
rs.close();
ps.close();
List rows = rsdc.getRows();
for (int i = 0; i <rows.size(); i++) {
DynaBean b=(DynaBean)rows.get(i);
System.out.println(b.get("name"));
}
} catch (Exception e) {
e.printStackTrace();
}
finally{
try {
con.close();
} catch (Exception e) {
}
}
在這里用到了一個CustomRowSetDynaClass類，構造方法中增加了page和pageSize兩個參數，
這樣，不管數據庫里有多少條記錄，最多只取pageSize條記錄，若pageSize==-1，則功能和
RowSetDynaClass一樣。這在大多數情況下是適用的。該類的代碼如下：

package test.jakarta.commons.beanutils;

import java.io.*;
import java.sql.*;
import java.util.*;

import org.apache.commons.beanutils.*;

/**
* @author SonyMusic
*
* To change this generated comment edit the template variable "typecomment":
* Window>Preferences>Java>Templates.
* To enable and disable the creation of type comments go to
* Window>Preferences>Java>Code Generation.
*/
public class CustomRowSetDynaClass implements DynaClass, Serializable {

// ----------------------------------------------------------- Constructors

/**
* <p>Construct a new {@link RowSetDynaClass} for the specified
* <code>ResultSet</code>.? The property names corresponding
* to column names in the result set will be lower cased.</p>
*
* @param resultSet The result set to be wrapped
*
* @exception NullPointerException if <code>resultSet</code>
*? is <code>null</code>
* @exception SQLException if the metadata for this result set
*? cannot be introspected
*/
public CustomRowSetDynaClass(ResultSet resultSet) throws SQLException {

this(resultSet, true);

}

/**
* <p>Construct a new {@link RowSetDynaClass} for the specified
* <code>ResultSet</code>.? The property names corresponding
* to the column names in the result set will be lower cased or not,
* depending on the specified <code>lowerCase</code> value.</p>
*
* <p><strong>WARNING</strong> - If you specify <code>false</code>
* for <code>lowerCase</code>, the returned property names will
* exactly match the column names returned by your JDBC driver.
* Because different drivers might return column names in different
* cases, the property names seen by your application will vary
* depending on which JDBC driver you are using.</p>
*
* @param resultSet The result set to be wrapped
* @param lowerCase Should property names be lower cased?
*
* @exception NullPointerException if <code>resultSet</code>
*? is <code>null</code>
* @exception SQLException if the metadata for this result set
*? cannot be introspected
*/
public CustomRowSetDynaClass(ResultSet resultSet, boolean lowerCase)
throws SQLException {

this(resultSet, 1, -1, lowerCase);

}

public CustomRowSetDynaClass(
ResultSet resultSet,
int page,
int pageSize,
boolean lowerCase)
throws SQLException {

if (resultSet == null) {
throw new NullPointerException();
}
this.lowerCase = lowerCase;
this.page = page;
this.pageSize = pageSize;

introspect(resultSet);
copy(resultSet);

}

public CustomRowSetDynaClass(ResultSet resultSet, int page, int pageSize)
throws SQLException {
this(resultSet, page, pageSize, true);
}

// ----------------------------------------------------- Instance Variables

/**
* <p>Flag defining whether column names should be lower cased when
* converted to property names.</p>
*/
protected boolean lowerCase = true;

protected int page = 1;
protected int pageSize = -1;

/**
* <p>The set of dynamic properties that are part of this
* {@link DynaClass}.</p>
*/
protected DynaProperty properties[] = null;

/**
* <p>The set of dynamic properties that are part of this
* {@link DynaClass}, keyed by the property name.? Individual descriptor
* instances will be the same instances as those in the
* <code>properties</code> list.</p>
*/
protected Map propertiesMap = new HashMap();

/**
* <p>The list of {@link DynaBean}s representing the contents of
* the original <code>ResultSet</code> on which this
* {@link RowSetDynaClass} was based.</p>
*/
protected List rows = new ArrayList();

// ------------------------------------------------------ DynaClass Methods

/**
* <p>Return the name of this DynaClass (analogous to the
* <code>getName()</code> method of <code>java.lang.Class</code), which
* allows the same <code>DynaClass</code> implementation class to support
* different dynamic classes, with different sets of properties.</p>
*/
public String getName() {

return (this.getClass().getName());

}

/**
* <p>Return a property descriptor for the specified property, if it
* exists; otherwise, return <code>null</code>.</p>
*
* @param name Name of the dynamic property for which a descriptor
*? is requested
*
* @exception IllegalArgumentException if no property name is specified
*/
public DynaProperty getDynaProperty(String name) {

if (name == null) {
throw new IllegalArgumentException("No property name specified");
}
return ((DynaProperty) propertiesMap.get(name));

}

/**
* <p>Return an array of <code>ProperyDescriptors</code> for the properties
* currently defined in this DynaClass.? If no properties are defined, a
* zero-length array will be returned.</p>
*/
public DynaProperty[] getDynaProperties() {

return (properties);

}

/**
* <p>Instantiate and return a new DynaBean instance, associated
* with this DynaClass.? <strong>NOTE</strong> - This operation is not
* supported, and throws an exception.</p>
*
* @exception IllegalAccessException if the Class or the appropriate
*? constructor is not accessible
* @exception InstantiationException if this Class represents an abstract
*? class, an array class, a primitive type, or void; or if instantiation
*? fails for some other reason
*/
public DynaBean newInstance()
throws IllegalAccessException, InstantiationException {

throw new UnsupportedOperationException("newInstance() not supported");

}

// --------------------------------------------------------- Public Methods

/**
* <p>Return a <code>List</code> containing the {@link DynaBean}s that
* represent the contents of each <code>Row</code> from the
* <code>ResultSet</code> that was the basis of this
* {@link RowSetDynaClass} instance.? These {@link DynaBean}s are
* disconnected from the database itself, so there is no problem with
* modifying the contents of the list, or the values of the properties
* of these {@link DynaBean}s.? However, it is the application's
* responsibility to persist any such changes back to the database,
* if it so desires.</p>
*/
public List getRows() {

return (this.rows);

}

// ------------------------------------------------------ Protected Methods

/**
* <p>Copy the column values for each row in the specified
* <code>ResultSet</code> into a newly created {@link DynaBean}, and add
* this bean to the list of {@link DynaBean}s that will later by
* returned by a call to <code>getRows()</code>.</p>
*
* @param resultSet The <code>ResultSet</code> whose data is to be
*? copied
*
* @exception SQLException if an error is encountered copying the data
*/
protected void copy(ResultSet resultSet) throws SQLException {
int abs = 0;
int rowsCount = 0;
int currentPageRows = 0;
resultSet.last();
rowsCount = resultSet.getRow();
if (pageSize != -1) {
int totalPages = (int) Math.ceil(((double) rowsCount) / pageSize);
if (page > totalPages)
page = totalPages;
if (page < 1)
page = 1;
abs = (page - 1) * pageSize;

//currentPageRows=(page==totalPages?rowsCount-pageSize*(totalPages-1):pageSize);
} else
pageSize = rowsCount;
if (abs == 0)
resultSet.beforeFirst();
else
resultSet.absolute(abs);
//int
while (resultSet.next() && ++currentPageRows <= pageSize) {
DynaBean bean = new BasicDynaBean(this);
for (int i = 0; i < properties.length; i++) {
String name = properties[i].getName();
bean.set(name, resultSet.getObject(name));
}
rows.add(bean);
}

}

/**
* <p>Introspect the metadata associated with our result set, and populate
* the <code>properties</code> and <code>propertiesMap</code> instance
* variables.</p>
*
* @param resultSet The <code>resultSet</code> whose metadata is to
*? be introspected
*
* @exception SQLException if an error is encountered processing the
*? result set metadata
*/
protected void introspect(ResultSet resultSet) throws SQLException {

// Accumulate an ordered list of DynaProperties
ArrayList list = new ArrayList();
ResultSetMetaData metadata = resultSet.getMetaData();
int n = metadata.getColumnCount();
for (int i = 1; i <= n; i++) { // JDBC is one-relative!
DynaProperty dynaProperty = createDynaProperty(metadata, i);
if (dynaProperty != null) {
list.add(dynaProperty);
}
}

// Convert this list into the internal data structures we need
properties =
(DynaProperty[]) list.toArray(new DynaProperty[list.size()]);
for (int i = 0; i < properties.length; i++) {
propertiesMap.put(properties[i].getName(), properties[i]);
}

}

/**
* <p>Factory method to create a new DynaProperty for the given index
* into the result set metadata.</p>
*
* @param metadata is the result set metadata
* @param i is the column index in the metadata
* @return the newly created DynaProperty instance
*/
protected DynaProperty createDynaProperty(
ResultSetMetaData metadata,
int i)
throws SQLException {

String name = null;
if (lowerCase) {
name = metadata.getColumnName(i).toLowerCase();
} else {
name = metadata.getColumnName(i);
}
String className = null;
try {
className = metadata.getColumnClassName(i);
} catch (SQLException e) {
// this is a patch for HsqlDb to ignore exceptions
// thrown by its metadata implementation
}

// Default to Object type if no class name could be retrieved
// from the metadata
Class clazz = Object.class;
if (className != null) {
clazz = loadClass(className);
}
return new DynaProperty(name, clazz);

}

/**
* <p>Loads and returns the <code>Class</code> of the given name.
* By default, a load from the thread context class loader is attempted.
* If there is no such class loader, the class loader used to load this
* class will be utilized.</p>
*
* @exception SQLException if an exception was thrown trying to load
*? the specified class
*/
protected Class loadClass(String className) throws SQLException {

try {
ClassLoader cl = Thread.currentThread().getContextClassLoader();
if (cl == null) {
cl = this.getClass().getClassLoader();
}
return (cl.loadClass(className));
} catch (Exception e) {
throw new SQLException(
"Cannot load column class '" + className + "': " + e);
}

}

}