Tinyhttpd README

The project's README from github.com/EZLippi/Tinyhttpd. As one markdown file: README.md.

A mirror for tinyhttpd(Tinyhttpd非官方镜像,Fork自sourceForge,仅供学习)

测试CGI时需要本机安装PERL，同时安装perl-cgi

Prepare

Compile for Linux

 To compile for Linux:
  1) Comment out the #include <pthread.h> line.
  2) Comment out the line that defines the variable newthread.
  3) Comment out the two lines that run pthread_create().
  4) Uncomment the line that runs accept_request().
  5) Remove -lsocket from the Makefile.

     每个函数的作用：

     accept_request:  处理从套接字上监听到的一个 HTTP 请求，在这里可以很大一部分地体现服务器处理请求流程。

     bad_request: 返回给客户端这是个错误请求，HTTP 状态吗 400 BAD REQUEST.

     cat: 读取服务器上某个文件写到 socket 套接字。

     cannot_execute: 主要处理发生在执行 cgi 程序时出现的错误。

     error_die: 把错误信息写到 perror 并退出。

     execute_cgi: 运行 cgi 程序的处理，也是个主要函数。

     get_line: 读取套接字的一行，把回车换行等情况都统一为换行符结束。

     headers: 把 HTTP 响应的头部写到套接字。

     not_found: 主要处理找不到请求的文件时的情况。

     sever_file: 调用 cat 把服务器文件返回给浏览器。

     startup: 初始化 httpd 服务，包括建立套接字，绑定端口，进行监听等。

     unimplemented: 返回给浏览器表明收到的 HTTP 请求所用的 method 不被支持。


     建议源码阅读顺序： main -> startup -> accept_request -> execute_cgi, 通晓主要工作流程后再仔细把每个函数的源码看一看。


     工作流程

     （1） 服务器启动，在指定端口或随机选取端口绑定 httpd 服务。

     （2）收到一个 HTTP 请求时（其实就是 listen 的端口 accpet 的时候），派生一个线程运行 accept_request 函数。

     （3）取出 HTTP 请求中的 method (GET 或 POST) 和 url,。对于 GET 方法，如果有携带参数，则 query_string 指针指向 url 中 ？ 后面的 GET 参数。

     （4） 格式化 url 到 path 数组，表示浏览器请求的服务器文件路径，在 tinyhttpd 中服务器文件是在 htdocs 文件夹下。当 url 以 / 结尾，或 url 是个目录，则默认在 path 中加上 index.html，表示访问主页。

     （5）如果文件路径合法，对于无参数的 GET 请求，直接输出服务器文件到浏览器，即用 HTTP 格式写到套接字上，跳到（10）。其他情况（带参数 GET，POST 方式，url 为可执行文件），则调用 excute_cgi 函数执行 cgi 脚本。

    （6）读取整个 HTTP 请求并丢弃，如果是 POST 则找出 Content-Length. 把 HTTP 200  状态码写到套接字。

    （7） 建立两个管道，cgi_input 和 cgi_output, 并 fork 一个进程。

    （8） 在子进程中，把 STDOUT 重定向到 cgi_outputt 的写入端，把 STDIN 重定向到 cgi_input 的读取端，关闭 cgi_input 的写入端 和 cgi_output 的读取端，设置 request_method 的环境变量，GET 的话设置 query_string 的环境变量，POST 的话设置 content_length 的环境变量，这些环境变量都是为了给 cgi 脚本调用，接着用 execl 运行 cgi 程序。

    （9） 在父进程中，关闭 cgi_input 的读取端 和 cgi_output 的写入端，如果 POST 的话，把 POST 数据写入 cgi_input，已被重定向到 STDIN，读取 cgi_output 的管道输出到客户端，该管道输入是 STDOUT。接着关闭所有管道，等待子进程结束。这一部分比较乱，见下图说明：



图 1    管道初始状态


图 2  管道最终状态 


    （10） 关闭与浏览器的连接，完成了一次 HTTP 请求与回应，因为 HTTP 是无连接的。


以下内容来自源作者:

This software is copyright 1999 by J. David Blackstone. Permission is granted to redistribute and modify this software under the terms of the GNU General Public License, available at http://www.gnu.org/ .

If you use this software or examine the code, I would appreciate knowing and would be overjoyed to hear about it at [email protected] .

This software is not production quality. It comes with no warranty of any kind, not even an implied warranty of fitness for a particular purpose. I am not responsible for the damage that will likely result if you use this software on your computer system.

I wrote this webserver for an assignment in my networking class in 1999. We were told that at a bare minimum the server had to serve pages, and told that we would get extra credit for doing "extras." Perl had introduced me to a whole lot of UNIX functionality (I learned sockets and fork from Perl!), and O'Reilly's lion book on UNIX system calls plus O'Reilly's books on CGI and writing web clients in Perl got me thinking and I realized I could make my webserver support CGI with little trouble.

Now, if you're a member of the Apache core group, you might not be impressed. But my professor was blown over. Try the color.cgi sample script and type in "chartreuse." Made me seem smarter than I am, at any rate. :)

Apache it's not. But I do hope that this program is a good educational tool for those interested in http/socket programming, as well as UNIX system calls. (There's some textbook uses of pipes, environment variables, forks, and so on.)

One last thing: if you look at my webserver or (are you out of mind?!?) use it, I would just be overjoyed to hear about it. Please email me. I probably won't really be releasing major updates, but if I help you learn something, I'd love to know!

Happy hacking!

                               J. David Blackstone

Tinyhttpd (EZLippi/Tinyhttpd)

Searching Database…

12.2K ·GitHub
Tinyhttpd
What is this?Tinyhttpd is a minimal HTTP server written in C, originally created by J. David Blackstone in 1999 as a class assignment; this repo is an unofficial mirror of the SourceForge project kept around purely for learning how HTTP servers work at the socket level.
Is this meant to be a real production web server?No - the original author explicitly warns it is not production quality and offers no warranty; it was written as a teaching tool, not something to compete with Apache.
How do I build and run it?Clone the repo and compile it for Linux following the 'Prepare / Compile for Linux' steps in the README, then run the resulting binary to start the server.
Do I need anything besides a C compiler to try the CGI feature?Yes - to test CGI you need Perl and the perl-cgi package installed locally, since the bundled example script (color.cgi) is a Perl CGI program.
Does the server keep connections alive between requests?No - after serving each HTTP request it closes the connection to the browser outright, treating HTTP as a connectionless protocol with no persistent/keep-alive sessions.
Can I build this on Windows, or is it Linux-only?The README only documents a Linux build path and Makefile; there's no Windows instructions, so it should be treated as Linux-only.
How does it handle CGI requests under the hood?It forks a child process, redirects STDIN/STDOUT to pipes connected to the client socket, sets environment variables like REQUESTMETHOD and QUERYSTRING, then execs the CGI script and streams its output back to the browser.
Does each incoming request get its own thread?Yes - startup() sets up the listening socket and binds the port, and every accepted HTTP request spawns a new thread running acceptrequest to process it.
What's a good order to read the source code in?The README suggests tracing main -> startup -> acceptrequest -> executecgi first to understand the overall flow, then reading the smaller helper functions like getline, headers, cat and notfound.
Why did the author originally build this?J. David Blackstone wrote it in 1999 for a university networking class assignment that only required serving static pages; he added CGI support himself to earn extra credit.
What are similar projects?h2o, TinyWebServer, caddy, newbee-mall, selfdb, goproxy, nginx, lytlnyblOS
Which topics is it in?socket, http-server, educational, web-server, systems-programming, http, beginner-friendly, c, learning, gplv3, networking, tutorial
Creator
EZLippi
Size
<0.1 MB
Stars
12.2K
Updated
CommitJan 20
Age
11y
Commits · 90 Days