thttpd源码分析-----http协议头部检测

最新推荐文章于 2020-07-20 14:36:02 发布

AshlingR

最新推荐文章于 2020-07-20 14:36:02 发布

阅读量4.1k

点赞数

CC 4.0 BY-SA版权

分类专栏： 4.源码分析

本文链接：https://blog.youkuaiyun.com/ashlingr/article/details/8242456

4.源码分析专栏收录该内容

0 篇文章

订阅专栏

本文分析了thttpd源码中的httpd_got_request函数，该函数主要负责HTTP头部的格式检测，包括分隔、空格和tab符的使用规则。它检查请求行是否包含三个单词，并且头部以空行结束，以此判断HTTP协议头部的合法性。

摘要生成于 C知道，由 DeepSeek-R1 满血版支持，前往体验 >

在源码中，用于http 头部检测的是 httpd_got_request 函数

该函数并没有检测头部的内容，只是检测头部的格式而已，最明显的格式检测就是 \r\n ，空格，tab符，即http协议中，头部的每一行都用 \r\n 结尾，在行中，每一词用空格或tab符分开，说明更简单一些，就是检查请求行是不是有三个words，头部结束是不是以空行结束。仅此而已

下面是用的几种状态来检测格式是否合法的

/* Checks hc->read_buf to see whether a complete request has been read so far;
 ** either the first line has two words (an HTTP/0.9 request), or the first
 ** line has three words and there's a blank line present.
 **
 ** hc->read_idx is how much has been read in; hc->checked_idx is how much we
 ** have checked so far; and hc->checked_state is the current state of the
 ** finite state machine.
 */
int httpd_got_request(httpd_conn* hc) {
	char c;

	for (; hc->checked_idx < hc->read_idx; ++hc->checked_idx) {
		c = hc->read_buf[hc->checked_idx];
		switch (hc->checked_state) {
		case CHST_FIRSTWORD://the first word "GET"
			switch (c) {
			case ' ':
			case '\t':
				hc->checked_state = CHST_FIRSTWS;
				break;
			case '\012':
			case '\015':
				hc->checked_state = CHST_BOGUS;
				return GR_BAD_REQUEST;
			}
			break;
		case CHST_FIRSTWS:
			switch (c) {
			case ' ':
			case '\t':
				break;
			case '\012':
			case '\015':
				hc->checked_state = CHST_BOGUS;
				return GR_BAD_REQUEST;
			default:
				hc->checked_state = CHST_SECONDWORD;
				break;
			}
			break;
		case CHST_SECONDWORD:
			switch (c) {
			case ' ':
			case '\t':
				hc->checked_state = CHST_SECONDWS;
				break;
			case '\012':
			case '\015':
				/* The first line has only two words - an HTTP/0.9 request. */
				return GR_GOT_REQUEST;
			}
			break;
		case CHST_SECONDWS:
			switch (c) {
			case ' ':
			case '\t':
				break;
			case '\012':
			case '\015':
				hc->checked_state = CHST_BOGUS;
				return GR_BAD_REQUEST;
			default:
				hc->checked_state = CHST_THIRDWORD;
				break;
			}
			break;
		case CHST_THIRDWORD:
			switch (c) {
			case ' ':
			case '\t':
				hc->checked_state = CHST_THIRDWS;
				break;
			case '\012':
				hc->checked_state = CHST_LF;
				break;
			case '\015':
				hc->checked_state = CHST_CR;
				break;
			}
			break;
		case CHST_THIRDWS:
			switch (c) {
			case ' ':
			case '\t':
				break;
			case '\012':
				hc->checked_state = CHST_LF;
				break;
			case '\015':
				hc->checked_state = CHST_CR;
				break;
			default:
				hc->checked_state = CHST_BOGUS;
				return GR_BAD_REQUEST;
			}
			break;
		case CHST_LINE:
			switch (c) {
			case '\012':
				hc->checked_state = CHST_LF;
				break;
			case '\015':
				hc->checked_state = CHST_CR;
				break;
			}
			break;
		case CHST_LF:
			switch (c) {
			case '\012':
				/* Two newlines in a row - a blank line - end of request. */
				return GR_GOT_REQUEST;
			case '\015':
				hc->checked_state = CHST_CR;
				break;
			default:
				hc->checked_state = CHST_LINE;
				break;
			}
			break;
		case CHST_CR:
			switch (c) {
			case '\012':
				hc->checked_state = CHST_CRLF;
				break;
			case '\015':
				/* Two returns in a row - end of request. */
				return GR_GOT_REQUEST;
			default:
				hc->checked_state = CHST_LINE;
				break;
			}
			break;
		case CHST_CRLF:
			switch (c) {
			case '\012':
				/* Two newlines in a row - end of request. */
				return GR_GOT_REQUEST;
			case '\015':
				hc->checked_state = CHST_CRLFCR;
				break;
			default:
				hc->checked_state = CHST_LINE;
				break;
			}
			break;
		case CHST_CRLFCR:
			switch (c) {
			case '\012':
			case '\015':
				/* Two CRLFs or two CRs in a row - end of request. */
				return GR_GOT_REQUEST;
			default:
				hc->checked_state = CHST_LINE;
				break;
			}
			break;
		case CHST_BOGUS:
			return GR_BAD_REQUEST;
		}
	}
	return GR_NO_REQUEST;
}