Java 利用 ByteArrayOutputStream 和 ByteArrayInputStream 避免重复读取配置文件

最近参与了github上的一个开源项目 Mycat,是一个mysql的分库分表的中间件。发现其中读取配置文件的代码,存在频繁多次重复打开,读取,关闭的问题,代码写的很初级,稍微看过一些框架源码的人,是不会犯这样的错误的。于是对其进行了一些优化。

优化之前的代码如下所示:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
private static Element loadRoot() {
    InputStream dtd = null;
    InputStream xml = null;
    Element root = null;
    try {
        dtd = ConfigFactory.class.getResourceAsStream("/mycat.dtd");
        xml = ConfigFactory.class.getResourceAsStream("/mycat.xml");
        root = ConfigUtil.getDocument(dtd, xml).getDocumentElement();
    catch (ConfigException e) {
        throw e;
    catch (Exception e) {
        throw new ConfigException(e);
    finally {
        if (dtd != null) {
            try {
                dtd.close();
            catch (IOException e) { }
        }
        if (xml != null) {
            try {
                xml.close();
            catch (IOException e) { }
        }
    }
    return root;
}

然后其它方法频繁调用 loadRoot():

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
@Override
public UserConfig getUserConfig(String user) {
    Element root = loadRoot();
    loadUsers(root);
    return this.users.get(user);
}
 
@Override
public Map<String, UserConfig> getUserConfigs() {
    Element root = loadRoot();
    loadUsers(root);
    return users;
}
 
@Override
public SystemConfig getSystemConfig() {
    Element root = loadRoot();
    loadSystem(root);
    return system;
}
// ... ...

ConfigUtil.getDocument(dtd, xml) 方法如下:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
    public static Document getDocument(final InputStream dtd, InputStream xml) throws ParserConfigurationException,
            SAXException, IOException {
        DocumentBuilderFactory factory = DocumentBuilderFactory.newInstance();
//factory.setValidating(false);
        factory.setNamespaceAware(false);
        DocumentBuilder builder = factory.newDocumentBuilder();
        builder.setEntityResolver(new EntityResolver() {
            @Override
            public InputSource resolveEntity(String publicId, String systemId) {
                return new InputSource(dtd);
            }
        });
        builder.setErrorHandler(new ErrorHandler() {
            @Override
            public void warning(SAXParseException e) {
            }
            @Override
            public void error(SAXParseException e) throws SAXException {
                throw e;
            }
            @Override
            public void fatalError(SAXParseException e) throws SAXException {
                throw e;
            }
        });
        return builder.parse(xml);
    }

显然这不是很好的处理方式。因为会多次重复读取配置文件。

1. 第一次优化:

为什么不读取一次,然后缓存起来呢?然后其它方法在调用 loadRoot() 时,就直接使用缓存中的就行了。但是遇到一个问题,InputStream 是不能被缓存,然后重复读取的,因为 InputStream 一旦被读取之后,其 pos 指针,等等都会发生变化,无法进行重复读取。所以只能将配置文件的内容读取处理,放入  byte[] 中缓存起来,然后配合 ByteArrayOutputStream,就可以重复读取 byte[] 缓存中的内容了。然后利用 ByteArrayOutputStream 来构造 InputStream 就达到了读取配置文件一次,然后重复构造 InputStream 进行重复读取,相关代码如下:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
// 为了避免原代码中频繁调用 loadRoot 去频繁读取 /mycat.dtd 和 /mycat.xml,所以将两个文件进行缓存,
// 注意这里并不会一直缓存在内存中,随着 LocalLoader 对象的回收,缓存占用的内存自然也会被回收。
private static byte[] xmlBuffer = null;
private static byte[] dtdBuffer = null;
private static  ByteArrayOutputStream xmlBaos = null;
private static  ByteArrayOutputStream dtdBaos = null;
 
static {
    InputStream input = ConfigFactory.class.getResourceAsStream("/mycat.dtd");
    if(input != null){
        dtdBuffer = new byte[1024 512];
        dtdBaos = new ByteArrayOutputStream();
        bufferFileStream(input, dtdBuffer, dtdBaos);
    }
     
    input = ConfigFactory.class.getResourceAsStream("/mycat.xml");
    if(input != null){
        xmlBuffer = new byte[1024 512];
        xmlBaos = new ByteArrayOutputStream();
        bufferFileStream(input, xmlBuffer, xmlBaos);
    }
}

bufferFileStream 方法:

1
2
3
4
5
6
7
8
9
10
11
12
private static void bufferFileStream(InputStream input, byte[] buffer, ByteArrayOutputStream baos){
    int len = -1;
    try {
        while ((len = input.read(buffer)) > -1 ) {
            baos.write(buffer, 0, len);
        }
        baos.flush();
    catch (IOException e) {
        e.printStackTrace();
        logger.error(" bufferFileStream error: " + e.getMessage());
    }
}

 loadRoat 优化之后如下:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
private static Element loadRoot() {
    Element root = null;
    InputStream mycatXml = null;
    InputStream mycatDtd = null;
     
    if(xmlBaos != null)
        mycatXml = new ByteArrayInputStream(xmlBaos.toByteArray());
    if(dtdBaos != null)
        mycatDtd = new ByteArrayInputStream(dtdBaos.toByteArray());
     
    try {
    root = ConfigUtil.getDocument(mycatDtd, mycatXml).getDocumentElement();
    catch (ParserConfigurationException | SAXException | IOException e1) {
        e1.printStackTrace();
        logger.error("loadRoot error: " + e1.getMessage());
    }finally{
        if(mycatXml != null){
            try { mycatXml.close(); } catch (IOException e) {}
        }
        if(mycatDtd != null){
            try { mycatDtd.close(); } catch (IOException e) {}
        }
    }
     
    return root;
}

这样优化之后,即使有很多方法频繁调用 loadRoot() 方法,也不会重复读取配置文件了,而是使用 byte[] 内容,重复构造 InputStream 而已。

其实其原理,就是利用 byte[] 作为一个中间容器,对byte进行缓存,ByteArrayOutputStream 将 InputStream 读取的 byte 存放如 byte[]容器,然后利用 ByteArrayInputStream 从 byte[]容器中读取内容,构造 InputStream,只要 byte[] 这个缓存容器存在,就可以多次重复构造出 InputStream。 于是达到了读取一次配置文件,而重复构造出InputStream,避免了每构造一次InputStream,就读取一次配置文件的问题。

2. 第二次优化:

可能你会想到更好的方法,比如:

为什么我们不将 private static Element root = null; 作为类属性,缓存起来,这样就不需要重复打开和关闭配置文件了,修改如下:

复制代码
public class LocalLoader implements ConfigLoader {
    private static final Logger logger = LoggerFactory.getLogger("LocalLoader");
    // ... ..    
    private static Element root = null;

    // 然后 loadRoot 方法改为:
    private static Element loadRoot() {
        InputStream dtd = null;
        InputStream xml = null;
//        Element root = null;
        if(root == null){
            try {
                dtd = ConfigFactory.class.getResourceAsStream("/mycat.dtd");
                xml = ConfigFactory.class.getResourceAsStream("/mycat.xml");
                root = ConfigUtil.getDocument(dtd, xml).getDocumentElement();
            } catch (ConfigException e) {
                throw e;
            } catch (Exception e) {
                throw new ConfigException(e);
            } finally {
                if (dtd != null) {
                    try {
                        dtd.close();
                    } catch (IOException e) { }
                }
                if (xml != null) {
                    try {
                        xml.close();
                    } catch (IOException e) { }
                }
            }
        }
        
        return root;
    }
复制代码

这样就不需要也不会重复 打开和关闭配置文件了。只要 root 属性没有被回收,那么 root 引入的 Document 对象也会在缓存中。这样显然比第一次优化要好很多,因为第一次优化,还是要从 byte[] 重复构造 InputStream, 然后重复 build 出 Document 对象。

3. 第三次优化

上面是将 private static Element root = null; 作为一个属性进行缓存,避免重复读取。那么我们干嘛不直接将 Document 对象作为一个属性,进行缓存呢。而且具有更好的语义,代码更好理解。代码如下:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
public class LocalLoader implements ConfigLoader {
    private static final Logger logger = LoggerFactory.getLogger("LocalLoader");
    // ... ...
    // 为了避免原代码中频繁调用 loadRoot 去频繁读取 /mycat.dtd 和 /mycat.xml,所以将 Document 进行缓存,
    private static Document document = null;
 
    private static Element loadRoot() {
        InputStream dtd = null;
        InputStream xml = null;       
        if(document == null){
           try {
                dtd = ConfigFactory.class.getResourceAsStream("/mycat.dtd");
                xml = ConfigFactory.class.getResourceAsStream("/mycat.xml");
                document = ConfigUtil.getDocument(dtd, xml);
                return document.getDocumentElement();
            catch (Exception e) {
                logger.error(" loadRoot error: " + e.getMessage());
                throw new ConfigException(e);
            finally {
                if (dtd != null) {
                    try { dtd.close(); } catch (IOException e) { }
                }
                if (xml != null) {
                    try { xml.close(); } catch (IOException e) { }
                }
            }
        }       
        return document.getDocumentElement();
    }

这样才是比较合格的实现。anyway, 第一种优化,学习到了 ByteArrayOutputStream 和 ByteArrayInputStream 同 byte[] 配合使用的方法。

---------------------分割线------------------------------------

参考文章:http://blog.csdn.net/it_magician/article/details/9240727 原文如下:

有时候我们需要对同一个InputStream对象使用多次。比如,客户端从服务器获取数据 ,利用HttpURLConnection的getInputStream()方法获得Stream对象,这时既要把数据显示到前台(第一次读取),又想把数据写进文件缓存到本地(第二次读取)。

但第一次读取InputStream对象后,第二次再读取时可能已经到Stream的结尾了(EOFException)或者Stream已经close掉了。

而InputStream对象本身不能复制,因为它没有实现Cloneable接口。此时,可以先把InputStream转化成ByteArrayOutputStream,后面要使用InputStream对象时,再从ByteArrayOutputStream转化回来就好了。代码实现如下:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
InputStream input =  httpconn.getInputStream();
                 
ByteArrayOutputStream baos = new ByteArrayOutputStream();
byte[] buffer = new byte[1024];
int len;
while ((len = input.read(buffer)) > -1 ) {
    baos.write(buffer, 0, len);
}
baos.flush();             
 
InputStream stream1 = new ByteArrayInputStream(baos.toByteArray());
 
//TODO:显示到前台
 
InputStream stream2 = new ByteArrayInputStream(baos.toByteArray());
 
//TODO:本地缓存
原文地址:https://www.cnblogs.com/jmsjh/p/7864821.html