HTTPS URL的基本代理身份验证返回HTTP/1.0 407需要代理身份验证

Question

我想在Java中使用具有基本身份验证(用户名,密码)的代理进行连接(并且只有此连接).以下代码适用于HTTP URL(例如" http://www.google.com "):

URL url = new URL("http://www.google.com");
HttpURLConnection httpURLConnection = null;
InetSocketAddress proxyLocation = new InetSocketAddress(proxyHost, proxyPort);
Proxy proxy = new Proxy(Proxy.Type.HTTP, proxyLocation);
httpURLConnection = (HttpURLConnection) url.openConnection(proxy);
// Works for HTTP only! Doesn't work for HTTPS!
String encoded = new sun.misc.BASE64Encoder().encodeBuffer((proxyUserName + ":" + proxyPassword).getBytes()).replace("\r\n", "");
httpURLConnection.setRequestProperty("Proxy-Authorization", "Basic " + encoded);
InputStream is = httpURLConnection.getInputStream();
InputStreamReader isr = new InputStreamReader(is); 
int data = isr.read();
while(data != -1){
  char c = (char) data;
  data = isr.read();
  System.out.print(c);
}
isr.close();

但是,该代码不适用于HTTPS网址(例如" https://www.google.com ")!java.io.IOException: Unable to tunnel through proxy. Proxy returns "HTTP/1.0 407 Proxy Authentication Required"当我尝试访问HTTPS URL时,我得到了.

此代码适用于HTTP和HTTPS:

URL url = new URL("https://www.google.com");
HttpURLConnection httpURLConnection = null;
InetSocketAddress proxyLocation = new InetSocketAddress(proxyHost, proxyPort);
Proxy proxy = new Proxy(Proxy.Type.HTTP, proxyLocation);
httpURLConnection = (HttpURLConnection) url.openConnection(proxy);
// Works for HTTP and HTTPS, but sets a global default!
Authenticator.setDefault(new Authenticator() {
  protected PasswordAuthentication getPasswordAuthentication() {
    return new PasswordAuthentication(proxyUserName, proxyPassword.toCharArray());
  }
});
InputStream is = httpURLConnection.getInputStream();
InputStreamReader isr = new InputStreamReader(is); 
int data = isr.read();
while(data != -1){
  char c = (char) data;
  data = isr.read();
  System.out.print(c);
}
isr.close();

第二个代码的问题是它设置了一个新的默认值Authenticator,我不想这样做,因为这个代理仅由应用程序的一部分使用,而应用程序的不同部分可能使用不同的代理.我不想为整个应用程序设置全局默认值.有没有办法让第一个代码使用HTTPS或使用一种方法Authenticator而不将其设置为默认值？

我必须使用java.net.HttpURLConnection,因为我重写了一个必须返回一个类的方法HttpURLConnection,所以我不能使用Apache HttpClient.

Answer 1

您可以自己扩展ProxiedHttpsConnection和处理所有低级相关的东西。

要通过 HTTP 代理连接到 https 网站，需要执行以下步骤：

注意：与代理和 http 服务器的通信应该是ASCII7。

1. 发送CONNECT stackoverflow.com:443 HTTP/1.0\r\n给代理
2. 发送您的身份验证：Proxy-Authorization: Basic c2F5WW91SGF2ZVNlZW5UaGlzSW5UaGVDb21tZW50cw==\r\n。
3. 结束第一个请求： \r\n
4. 读取来自代理的响应，直到您看到组合“\r\n\r\n”。
5. 解析您从代理获得的响应的第一行，并检查它是否以HTTP/1.0 200.
6. 通过现有连接启动 SSL 会话。
7. 发送 http 请求的开始： GET /questions/3304006/persistent-httpurlconnection-in-java HTTP/1.0\r\n
8. 设置正确的主机头： Host: stackoverflow.com\r\n
9. 结束对http服务器的请求： \r\n
10. 阅读直到\r\n并将第一行解析为状态消息
11. 读取到请求正文的流结束

当我们要实现 HttpUrlConnection 类时，还需要考虑以下几点：

• 在构造类时，类应该存储数据以供将来连接使用，但不要直接存储
• 可以以任何顺序调用任何方法
• 关闭OutputStream意味着数据传输完成，而不是连接必须完成
• 每个 api 以不同的顺序使用这些方法
• HTTP 标头不区分大小写，java 映射区分大小写。

快说，只是有很多陷阱

在我设计的类中，它使用布尔标志来记住connect方法和afterPostClosure方法是否被调用，它还支持 ifgetInputStream()在OutputStream关闭之前被调用。

此类还在套接字返回的流上使用尽可能少的包装，以防止变得非常复杂。

public class ProxiedHttpsConnection extends HttpURLConnection {

    private final String proxyHost;
    private final int proxyPort;
    private static final byte[] NEWLINE = "\r\n".getBytes();//should be "ASCII7"

    private Socket socket;
    private final Map<String, List<String>> headers = new TreeMap<>(String.CASE_INSENSITIVE_ORDER);
    private final Map<String, List<String>> sendheaders = new TreeMap<>(String.CASE_INSENSITIVE_ORDER);
    private final Map<String, List<String>> proxyheaders = new TreeMap<>(String.CASE_INSENSITIVE_ORDER);
    private final Map<String, List<String>> proxyreturnheaders = new TreeMap<>(String.CASE_INSENSITIVE_ORDER);
    private int statusCode;
    private String statusLine;
    private boolean isDoneWriting;

    public ProxiedHttpsConnection(URL url,
            String proxyHost, int proxyPort, String username, String password)
            throws IOException {
        super(url);
        socket = new Socket();
        this.proxyHost = proxyHost;
        this.proxyPort = proxyPort;
        String encoded = Base64.encode((username + ":" + password).getBytes())
                .replace("\r\n", "");
        proxyheaders.put("Proxy-Authorization", new ArrayList<>(Arrays.asList("Basic " + encoded)));
    }

    @Override
    public OutputStream getOutputStream() throws IOException {
        connect();
        afterWrite();
        return new FilterOutputStream(socket.getOutputStream()) {
            @Override
            public void write(byte[] b, int off, int len) throws IOException {
                out.write(String.valueOf(len).getBytes());
                out.write(NEWLINE);
                out.write(b, off, len);
                out.write(NEWLINE);
            }

            @Override
            public void write(byte[] b) throws IOException {
                out.write(String.valueOf(b.length).getBytes());
                out.write(NEWLINE);
                out.write(b);
                out.write(NEWLINE);
            }

            @Override
            public void write(int b) throws IOException {
                out.write(String.valueOf(1).getBytes());
                out.write(NEWLINE);
                out.write(b);
                out.write(NEWLINE);
            }

            @Override
            public void close() throws IOException {
                afterWrite();
            }

        };
    }

    private boolean afterwritten = false;

    @Override
    public InputStream getInputStream() throws IOException {
        connect();
        return socket.getInputStream();

    }

    @Override
    public void setRequestMethod(String method) throws ProtocolException {
        this.method = method;
    }

    @Override
    public void setRequestProperty(String key, String value) {
        sendheaders.put(key, new ArrayList<>(Arrays.asList(value)));
    }

    @Override
    public void addRequestProperty(String key, String value) {
        sendheaders.computeIfAbsent(key, l -> new ArrayList<>()).add(value);
    }

    @Override
    public Map<String, List<String>> getHeaderFields() {
        return headers;
    }

    @Override
    public void connect() throws IOException {
        if (connected) {
            return;
        }
        connected = true;
        socket.setSoTimeout(getReadTimeout());
        socket.connect(new InetSocketAddress(proxyHost, proxyPort), getConnectTimeout());
        StringBuilder msg = new StringBuilder();
        msg.append("CONNECT ");
        msg.append(url.getHost());
        msg.append(':');
        msg.append(url.getPort() == -1 ? 443 : url.getPort());
        msg.append(" HTTP/1.0\r\n");
        for (Map.Entry<String, List<String>> header : proxyheaders.entrySet()) {
            for (String l : header.getValue()) {
                msg.append(header.getKey()).append(": ").append(l);
                msg.append("\r\n");
            }
        }

        msg.append("Connection: close\r\n");
        msg.append("\r\n");
        byte[] bytes;
        try {
            bytes = msg.toString().getBytes("ASCII7");
        } catch (UnsupportedEncodingException ignored) {
            bytes = msg.toString().getBytes();
        }
        socket.getOutputStream().write(bytes);
        socket.getOutputStream().flush();
        byte reply[] = new byte[200];
        byte header[] = new byte[200];
        int replyLen = 0;
        int headerLen = 0;
        int newlinesSeen = 0;
        boolean headerDone = false;
        /* Done on first newline */
        InputStream in = socket.getInputStream();
        while (newlinesSeen < 2) {
            int i = in.read();
            if (i < 0) {
                throw new IOException("Unexpected EOF from remote server");
            }
            if (i == '\n') {
                if (newlinesSeen != 0) {
                    String h = new String(header, 0, headerLen);
                    String[] split = h.split(": ");
                    if (split.length != 1) {
                        proxyreturnheaders.computeIfAbsent(split[0], l -> new ArrayList<>()).add(split[1]);
                    }
                }
                headerDone = true;
                ++newlinesSeen;
                headerLen = 0;
            } else if (i != '\r') {
                newlinesSeen = 0;
                if (!headerDone && replyLen < reply.length) {
                    reply[replyLen++] = (byte) i;
                } else if (headerLen < reply.length) {
                    header[headerLen++] = (byte) i;
                }
            }
        }

        String replyStr;
        try {
            replyStr = new String(reply, 0, replyLen, "ASCII7");
        } catch (UnsupportedEncodingException ignored) {
            replyStr = new String(reply, 0, replyLen);
        }

        // Some proxies return http/1.1, some http/1.0 even we asked for 1.0
        if (!replyStr.startsWith("HTTP/1.0 200") && !replyStr.startsWith("HTTP/1.1 200")) {
            throw new IOException("Unable to tunnel. Proxy returns \"" + replyStr + "\"");
        }
        SSLSocket s = (SSLSocket) ((SSLSocketFactory) SSLSocketFactory.getDefault())
                .createSocket(socket, url.getHost(), url.getPort(), true);
        s.startHandshake();
        socket = s;
        msg.setLength(0);
        msg.append(method);
        msg.append(" ");
        msg.append(url.toExternalForm().split(String.valueOf(url.getPort()), -2)[1]);
        msg.append(" HTTP/1.0\r\n");
        for (Map.Entry<String, List<String>> h : sendheaders.entrySet()) {
            for (String l : h.getValue()) {
                msg.append(h.getKey()).append(": ").append(l);
                msg.append("\r\n");
            }
        }
        if (method.equals("POST") || method.equals("PUT")) {
            msg.append("Transfer-Encoding: Chunked\r\n");
        }
        msg.append("Host: ").append(url.getHost()).append("\r\n");
        msg.append("Connection: close\r\n");
        msg.append("\r\n");
        try {
            bytes = msg.toString().getBytes("ASCII7");
        } catch (UnsupportedEncodingException ignored) {
            bytes = msg.toString().getBytes();
        }
        socket.getOutputStream().write(bytes);
        socket.getOutputStream().flush();
    }

    private void afterWrite() throws IOException {
        if (afterwritten) {
            return;
        }
        afterwritten = true;
        socket.getOutputStream().write(String.valueOf(0).getBytes());
        socket.getOutputStream().write(NEWLINE);
        socket.getOutputStream().write(NEWLINE);
        byte reply[] = new byte[200];
        byte header[] = new byte[200];
        int replyLen = 0;
        int headerLen = 0;
        int newlinesSeen = 0;
        boolean headerDone = false;
        /* Done on first newline */
        InputStream in = socket.getInputStream();
        while (newlinesSeen < 2) {
            int i = in.read();
            if (i < 0) {
                throw new IOException("Unexpected EOF from remote server");
            }
            if (i == '\n') {
                if (headerDone) {
                    String h = new String(header, 0, headerLen);
                    String[] split = h.split(": ");
                    if (split.length != 1) {
                        headers.computeIfAbsent(split[0], l -> new ArrayList<>()).add(split[1]);
                    }
                }
                headerDone = true;
                ++newlinesSeen;
                headerLen = 0;
            } else if (i != '\r') {
                newlinesSeen = 0;
                if (!headerDone && replyLen < reply.length) {
                    reply[replyLen++] = (byte) i;
                } else if (headerLen < header.length) {
                    header[headerLen++] = (byte) i;
                }
            }
        }

        String replyStr;
        try {
            replyStr = new String(reply, 0, replyLen, "ASCII7");
        } catch (UnsupportedEncodingException ignored) {
            replyStr = new String(reply, 0, replyLen);
        }

        /* We asked for HTTP/1.0, so we should get that back */
        if ((!replyStr.startsWith("HTTP/1.0 200")) && !replyStr.startsWith("HTTP/1.1 200")) {
            throw new IOException("Server returns \"" + replyStr + "\"");
        }
    }

    @Override
    public void disconnect() {
        try {
            socket.close();
        } catch (IOException ex) {
            Logger.getLogger(ProxiedHttpsConnection.class.getName()).log(Level.SEVERE, null, ex);
        }
    }

    @Override
    public boolean usingProxy() {
        return true;
    }
}

上述代码的当前错误：

• 发布期间不会因错误而关闭流
• 在与代理的初始联系发生错误期间不会关闭流
• 它不支持 http 重定向
• 它不支持像分块和 gzip 编码这样的 http 1.1 东西，但这没有问题，因为我们宣布自己是 http1.0 客户端。

上面的代码可以像这样使用：

    ProxiedHttpsConnection n = new ProxiedHttpsConnection(
            new URL("https://stackoverflow.com:443/questions/3304006/persistent-httpurlconnection-in-java"), 
            "proxy.example.com", 8080, "root", "flg83yvem#");
    n.setRequestMethod("GET");
    n.addRequestProperty("User-Agent", "Java test /sf/users/107990641/");
    //try (OutputStream out = n.getOutputStream()) {
    //  out.write("Hello?".getBytes());
    //}
    try (InputStream in = n.getInputStream()) {
        byte[] buff = new byte[1024];
        int length;
        while ((length = in.read(buff)) >= 0) {
            System.out.write(buff, 0, length);
        }
    }

如果您打算将它与一种代理选择器一起使用，您应该检查 url 的协议，看看它是 http 还是 https，如果是 http，不要使用此类，而是手动附加标头，如：

httpURLConnection.setRequestProperty("Proxy-Authorization", "Basic " + encoded);

为什么不使用httpsUrlConnection.setSSLSocketFactory

虽然 java 有这个方法，尝试使用它会告诉你为什么它不起作用，java 只是createSocket(Socket s, String host, int port, boolean autoClose)用一个已经打开的连接调用，因此无法手动执行代理操作。

Answer 2

不幸的是,没有简单的解决方案可以实现您的目标.您的第一个代码不适用于HTTPS,因为您直接设置了身份验证标头.由于客户端加密所有数据,因此代理服务器无法从请求中提取任何信息.

事实上,HTTPS和代理服务器以相反的方式工作.代理服务器希望查看在客户端和最终服务器之间流动的所有数据,并根据其看到的内容采取措施.另一方面,HTTPS协议对所有数据进行加密,这样在到达最终目的地之前,任何人都无法看到数据.加密算法在客户端和最终目的地之间协商,以便代理服务器不能解密任何信息,实际上它甚至不知道客户端正在使用哪种协议.

要在HTTPS连接上使用代理服务器,客户端必须建立隧道.为此,它必须直接向代理发出CONNECT命令,例如:

CONNECT www.google.com:443 HTTP/1.0

并发送凭据以使用代理服务器进行身份验证.

如果连接成功,则客户端可以通过连接发送和接收数据.代理服务器对数据完全失明.数据仅在客户端和服务器之间的路径上传递.

当您url.openConnection(proxy)在HTTP URL上执行时,它返回一个实例HttpURLConnection,当在HTTPS URL上运行时,就像在第二个代码中一样,它返回一个实例HttpsURLConnection.

您收到407错误代码,因为代理服务器无法从您发送的标头中提取身份验证信息.查看异常堆栈,我们可以看到抛出异常sun.net.www.protocol.http.HttpURLConnection.doTunneling(),发出CONNECT命令以通过代理建立HTTPS隧道.在源代码中sun.net.www.protocol.http.HttpURLConnection我们可以看到:

/* We only have a single static authenticator for now.
 * REMIND:  backwards compatibility with JDK 1.1.  Should be
 * eliminated for JDK 2.0.
 */
private static HttpAuthenticator defaultAuth;

因此,默认身份验证器似乎是提供代理凭据的唯一方法.

要做你想做的事,你必须自己去连接级别并处理HTTP协议,因为你必须与代理服务器直接对话,不要直接与谷歌服务器通信.