共计 5413 个字符,预计需要花费 14 分钟才能阅读完成。
Filter 可以对请求进行预处理,因此,我们可以把很多公共预处理逻辑放到 Filter 中完成。
考察这样一种需求:我们在 Web 应用中经常需要处理用户上传文件,例如,一个 UploadServlet 可以简单地编写如下:
@WebServlet(urlPatterns = "/upload/file")
public class UploadServlet extends HttpServlet {protected void doPost(HttpServletRequest req, HttpServletResponse resp) throws ServletException, IOException {// 读取 Request Body:
InputStream input = req.getInputStream();
ByteArrayOutputStream output = new ByteArrayOutputStream();
byte[] buffer = new byte[1024];
for (;;) {int len = input.read(buffer);
if (len == -1) {break;
}
output.write(buffer, 0, len);
}
// TODO: 写入文件:
// 显示上传结果:
String uploadedText = output.toString(StandardCharsets.UTF_8);
PrintWriter pw = resp.getWriter();
pw.write("<h1>Uploaded:</h1>");
pw.write("<pre><code>");
pw.write(uploadedText);
pw.write("</code></pre>");
pw.flush();}
}
但是要保证文件上传的完整性怎么办?在哈希算法一节中,我们知道,如果在上传文件的同时,把文件的哈希也传过来,服务器端做一个验证,就可以确保用户上传的文件一定是完整的。
这个验证逻辑非常适合写在 ValidateUploadFilter
中,因为它可以复用。
我们先写一个简单的版本,快速实现 ValidateUploadFilter
的逻辑:
@WebFilter("/upload/*")
public class ValidateUploadFilter implements Filter {@Override
public void doFilter(ServletRequest request, ServletResponse response, FilterChain chain)
throws IOException, ServletException {HttpServletRequest req = (HttpServletRequest) request;
HttpServletResponse resp = (HttpServletResponse) response;
// 获取客户端传入的签名方法和签名:
String digest = req.getHeader("Signature-Method");
String signature = req.getHeader("Signature");
if (digest == null || digest.isEmpty() || signature == null || signature.isEmpty()) {sendErrorPage(resp, "Missing signature.");
return;
}
// 读取 Request 的 Body 并验证签名:
MessageDigest md = getMessageDigest(digest);
InputStream input = new DigestInputStream(request.getInputStream(), md);
byte[] buffer = new byte[1024];
for (;;) {int len = input.read(buffer);
if (len == -1) {break;
}
}
String actual = toHexString(md.digest());
if (!signature.equals(actual)) {sendErrorPage(resp, "Invalid signature.");
return;
}
// 验证成功后继续处理:
chain.doFilter(request, response);
}
// 将 byte[]转换为 hex string:
private String toHexString(byte[] digest) {StringBuilder sb = new StringBuilder();
for (byte b : digest) {sb.append(String.format("%02x", b));
}
return sb.toString();}
// 根据名称创建 MessageDigest:
private MessageDigest getMessageDigest(String name) throws ServletException {try {return MessageDigest.getInstance(name);
} catch (NoSuchAlgorithmException e) {throw new ServletException(e);
}
}
// 发送一个错误响应:
private void sendErrorPage(HttpServletResponse resp, String errorMessage) throws IOException {resp.setStatus(HttpServletResponse.SC_BAD_REQUEST);
PrintWriter pw = resp.getWriter();
pw.write("<html><body><h1>");
pw.write(errorMessage);
pw.write("</h1></body></html>");
pw.flush();}
}
这个 ValidateUploadFilter
的逻辑似乎没有问题,我们可以用 curl 命令测试:
$ curl http://localhost:8080/upload/file -v -d 'test-data' \
-H 'Signature-Method: SHA-1' \
-H 'Signature: 7115e9890f5b5cc6914bdfa3b7c011db1cdafedb' \
-H 'Content-Type: application/octet-stream'
* Trying ::1...
* TCP_NODELAY set
* Connected to localhost (::1) port 8080 (#0)
> POST /upload/file HTTP/1.1
> Host: localhost:8080
> User-Agent: curl/7.64.1
> Accept: */*
> Signature-Method: SHA-1
> Signature: 7115e9890f5b5cc6914bdfa3b7c011db1cdafedb
> Content-Type: application/octet-stream
> Content-Length: 9
>
* upload completely sent off: 9 out of 9 bytes
< HTTP/1.1 200
< Transfer-Encoding: chunked
< Date: Thu, 30 Jan 2020 13:56:39 GMT
<
* Connection #0 to host localhost left intact
<h1>Uploaded:</h1><pre><code></code></pre>
* Closing connection 0
ValidateUploadFilter
对签名进行验证的逻辑是没有问题的,但是,细心的童鞋注意到,UploadServlet
并未读取到任何数据!
这里的原因是对 HttpServletRequest
进行读取时,只能读取一次。如果 Filter 调用 getInputStream()
读取了一次数据,后续 Servlet 处理时,再次读取,将无法读到任何数据。怎么办?
这个时候,我们需要一个“伪造”的 HttpServletRequest
,具体做法是使用代理模式,对getInputStream()
和getReader()
返回一个新的流:
class ReReadableHttpServletRequest extends HttpServletRequestWrapper {private byte[] body;
private boolean open = false;
public ReReadableHttpServletRequest(HttpServletRequest request, byte[] body) {super(request);
this.body = body;
}
// 返回 InputStream:
public ServletInputStream getInputStream() throws IOException {if (open) {throw new IllegalStateException("Cannot re-open input stream!");
}
open = true;
return new ServletInputStream() {private int offset = 0;
public boolean isFinished() {return offset >= body.length;
}
public boolean isReady() {return true;
}
public void setReadListener(ReadListener listener) { }
public int read() throws IOException {if (offset >= body.length) {return -1;
}
int n = body[offset] & 0xff;
offset++;
return n;
}
};
}
// 返回 Reader:
public BufferedReader getReader() throws IOException {if (open) {throw new IllegalStateException("Cannot re-open reader!");
}
open = true;
return new BufferedReader(new InputStreamReader(new ByteArrayInputStream(body), "UTF-8"));
}
}
注意观察 ReReadableHttpServletRequest
的构造方法,它保存了 ValidateUploadFilter
读取的 byte[]
内容,并在调用 getInputStream()
时通过 byte[]
构造了一个新的ServletInputStream
。
然后,我们在 ValidateUploadFilter
中,把 doFilter()
调用时传给下一个处理者的 HttpServletRequest
替换为我们自己“伪造”的ReReadableHttpServletRequest
:
public void doFilter(ServletRequest request, ServletResponse response, FilterChain chain)
throws IOException, ServletException {
...
chain.doFilter(new ReReadableHttpServletRequest(req, output.toByteArray()), response);
}
再注意到我们编写 ReReadableHttpServletRequest
时,是从 HttpServletRequestWrapper
继承,而不是直接实现 HttpServletRequest
接口。这是因为,Servlet 的每个新版本都会对接口增加一些新方法,从 HttpServletRequestWrapper
继承可以确保新方法被正确地覆写了,因为 HttpServletRequestWrapper
是由 Servlet 的 jar 包提供的,目的就是为了让我们方便地实现对 HttpServletRequest
接口的代理。
我们总结一下对 HttpServletRequest
接口进行代理的步骤:
- 从
HttpServletRequestWrapper
继承一个XxxHttpServletRequest
,需要传入原始的HttpServletRequest
实例; - 覆写某些方法,使得新的
XxxHttpServletRequest
实例看上去“改变”了原始的HttpServletRequest
实例; - 在
doFilter()
中传入新的XxxHttpServletRequest
实例。
虽然整个 Filter 的代码比较复杂,但它的好处在于:这个 Filter 在整个处理链中实现了灵活的“可插拔”特性,即是否启用对 Web 应用程序的其他组件(Filter、Servlet)完全没有影响。
练习
使用 Filter 修改 HttpServletRequest
请求。
下载练习
小结
借助 HttpServletRequestWrapper
,我们可以在 Filter 中实现对原始HttpServletRequest
的修改。