试着手写了一个Http服务器

import java.io.*; import java.net.ServerSocket; import java.net.Socket; import java.util.HashMap; import java.util.Map; import java.util.concurrent.ExecutorService; import java.util.concurrent.Executors;

public class HttpServerV1 {
// HTTP 底层要基于 TCP 来实现. 需要按照 TCP 的基本格式来先进行开发.
private ServerSocket serverSocket = null;

public HttpServerV1(int port) throws IOException {
serverSocket = new ServerSocket(port);
}

public void start() throws IOException {
System.out.println("服务器启动");
ExecutorService executorService = Executors.newCachedThreadPool();
while (true) {
// 1. 获取连接
Socket clientSocket = serverSocket.accept();
// 2. 处理连接(使用短连接的方式实现)
executorService.execute(new Runnable() {
@Override
public void run() {
process(clientSocket);
}
});
}
}

private void process(Socket clientSocket) {
// 由于 HTTP 协议是文本协议, 所以仍然使用字符流来处理.
try (BufferedReader bufferedReader = new BufferedReader(new InputStreamReader(clientSocket.getInputStream()));
BufferedWriter bufferedWriter = new BufferedWriter(new OutputStreamWriter(clientSocket.getOutputStream()))) {
// 下面的操作都要严格按照 HTTP 协议来进行操作.
// 1. 读取请求并解析
// a) 解析首行, 三个部分使用空格切分
String firstLine = bufferedReader.readLine();
String[] firstLineTokens = firstLine.split(" ");
String method = firstLineTokens[0];
String url = firstLineTokens[1];
String version = firstLineTokens[2];
// b) 解析 header, 按行读取, 然后按照冒号空格来分割键值对
Map headers = new HashMap<>();
String line = "";
// readLine 读取的一行内容, 是会自动去掉换行符的. 对于空行来说, 去掉了换行符, 就变成空字符串
while ((line = bufferedReader.readLine()) != null && line.length() != 0) {
// 不能使用 : 来切分. 像 referer 字段, 里面的内容是可能包含 : .
String[] headerTokens = line.split(": ");
headers.put(headerTokens[0], headerTokens[1]);
}
// c) 解析 body (暂时先不考虑)
// 请求解析完毕, 加上一个日志, 观察请求的内容是否正确.
System.out.printf("%s %s %s\n", method, url, version);
for (Map.Entry entry : headers.entrySet()) {
System.out.println(entry.getKey() + ": " + entry.getValue());
}
System.out.println();
// 2. 根据请求计算响应
// 不管是啥样的请求, 咱们都返回一个 hello 这样的 html
String resp = "";
if (url.equals("/ok")) {
bufferedWriter.write(version + " 200 OK\n");
resp = "

hello

";
} else if (url.equals("/notfound")) {
bufferedWriter.write(version + " 404 Not Found\n");
resp = "

not found

";
} else if (url.equals("/seeother")) {
bufferedWriter.write(version + " 303 See Other\n");
bufferedWriter.write("Location: http://www.sogou.com\n");
resp = "";
} else {
bufferedWriter.write(version + " 200 OK\n");
resp = "

default

";
}
// 3. 把响应写回到客户端
bufferedWriter.write("Content-Type: text/html\n");
bufferedWriter.write("Content-Length: " + resp.getBytes().length + "\n"); // 此处的长度, 不能写成 resp.length(), 得到的是字符的数目, 而不是字节的数目
bufferedWriter.write("\n");
bufferedWriter.write(resp);
// 此处这个 flush 就算没有, 问题也不大. 紧接着
// bufferedWriter 对象就要被关闭了. close 时就会自动触发刷新操作.
bufferedWriter.flush();
} catch (IOException e) {
e.printStackTrace();
} finally {
try {
clientSocket.close();
} catch (IOException e) {
e.printStackTrace();
}
}
}

public static void main(String[] args) throws IOException {
HttpServerV1 server = new HttpServerV1(9090);
server.start();
} }