Nginx “邪恶” rewrite

概述

本文主要针对nginx rewrite指令困惑的地方进行讲解,中间会涉及部分原理部分,我尽量用通俗易懂的语言来形容

功能讲解

大致流程

    The ngx_http_rewrite_module module directives are processed in the following order:

    the directives of this module specified on the server level are executed sequentially;

    repeatedly:

    • location is searched based on a request URI;

    • the directives of this module specified inside the found location are executed sequentially;

    • the loop is repeated if a request URI was rewritten, but not more than 10 times.

涉及phase
  •     NGX_HTTP_SERVER_REWRITE_PHASE,

  •     NGX_HTTP_FIND_CONFIG_PHASE,

  •     NGX_HTTP_REWRITE_PHASE,

  •     NGX_HTTP_POST_REWRITE_PHASE,

环境:
  • /aaa内容aaa

  • /bbb内容bbb

Example 1:
server {
    listen   80;
    location /aaa {
 
    if ($http_user_agent ~ Mozilla) {
            rewrite /aaa /bbb; 
 
    }
    return 403;
 
    }
 
    location /bbb {
    return 402
    }
 
}

在上述nginx.conf情况下,使用firefox浏览器访问nginx,if匹配成功,执行rewrite重写r->uri(/aaa转换为/bbb),

然后继续执行rewrite模块其他指令,执行到return 403后,该request在nginx中的处理完毕,返回403;

因此浏览器收到应答码为403

Example 2:
    server {
        listen   80;
        location /aaa {
 
        if ($http_user_agent ~ Mozilla) {
                rewrite /aaa /bbb; 
 
        }
#        return 403;
 
        }
 
        location /bbb {
        return 402
        }
 
    }

在上述nginx.conf情况下,使用firefox浏览器访问nginx,if匹配成功,执行rewrite重写r->uri(/aaa转换为/bbb)

rewrite模块执行完毕。因为所有rewrite模块的指令都执行完毕,进入POST_REWRITE_PHASE所在的checker函数ngx_http_core_post_rewrite_phase

中,由于执行了rewrite指令,在函数ngx_http_script_regex_start_code中

if (code->uri) {
    r->internal = 1;
    r->valid_unparsed_uri = 0;
    if (code->break_cycle) {//rewrite ....break 指令分支
        r->valid_location = 0;
        r->uri_changed = 0;
    else {
        r->uri_changed = 1;
    }
}

可以得知r->uri_changed=1,于是在函数ngx_http_core_post_rewrite_phase中执行如下流程

r->uri_changes--;
if (r->uri_changes == 0) {
    ngx_log_error(NGX_LOG_ERR, r->connection->log, 0,
                  "rewrite or internal redirection cycle "
                  "while processing "%V"", &r->uri);
    ngx_http_finalize_request(r, NGX_HTTP_INTERNAL_SERVER_ERROR);
    return NGX_OK;
}
r->phase_handler = ph->next;//ph->next为find_config_index,在ngx_http_init_phase_handlers中可见
 
 
cscf = ngx_http_get_module_srv_conf(r, ngx_http_core_module);
r->loc_conf = cscf->ctx->loc_conf;

由于重新进行server和location匹配,因此在上述例子中,浏览器收到的应答状态吗为402

Example 3:
server {
    listen   80;
    location /aaa {
 
    if ($http_user_agent ~ Mozilla) {
            rewrite /aaa /bbb break
 
    }
    return 403;
 
    }
 
    location /bbb {
        return 402
    }
 
}

在上述nginx.conf情况下,使用firefox浏览器访问nginx,if匹配成功,执行rewrite重写r->uri(/aaa转换为/bbb),

由于rewrite break标记符存在,从rewrite指令的配置解析函数ngx_http_rewrite和rewrite模块的handler函数ngx_http_rewrite_handler中

如下部分可以得知,将停止执行rewrite模块的其他一切指令;

ngx_http_rewrite {
....
....
    if (cf->args->nelts == 4) {
        if (ngx_strcmp(value[3].data, "last") == 0) {
            last = 1;
        else if (ngx_strcmp(value[3].data, "break") == 0) {
            regex->break_cycle = 1;
            last = 1;
        else if (ngx_strcmp(value[3].data, "redirect") == 0) {
            regex->status = NGX_HTTP_MOVED_TEMPORARILY;
            regex->redirect = 1;
            last = 1;
        else if (ngx_strcmp(value[3].data, "permanent") == 0) {
            regex->status = NGX_HTTP_MOVED_PERMANENTLY;
            regex->redirect = 1;
            last = 1;
        else {
            ngx_conf_log_error(NGX_LOG_EMERG, cf, 0,
                               "invalid parameter "%V"", &value[3]);
            return NGX_CONF_ERROR;
        }
.....
.....
.....
    if (last) {
        code = ngx_http_script_add_code(lcf->codes, sizeof(uintptr_t), &regex);
        if (code == NULL) {
            return NGX_CONF_ERROR;
        }
        *code = NULL;
    }
}
 
 
ngx_http_rewrite_handler
{
.....
.....
    while (*(uintptr_t *) e->ip) {
        code = *(ngx_http_script_code_pt *) e->ip;
        code(e);
    }
    if (e->status < NGX_HTTP_BAD_REQUEST) {//如果rewrite指令后不是break和last标记,而是其他两个重定向标记,则从此处直接结束该request处理流程
        return e->status;
    }
    if (r->err_status == 0) {
        return e->status;
    }
    return r->err_status;
.....
.....
}

进入到POST_REWRITE_PHASE所在的checker函数ngx_http_core_post_rewrite_phase中,由example 2 可知r->uri_changed = 0;

在ngx_http_core_post_rewrite_phase中会执行如下流程,因此会继续执行当前server 和 location 上下文中后续的handler,因此浏览器收到内容为bbb

if (!r->uri_changed) {
    r->phase_handler++;
    return NGX_AGAIN;
}
Example 4:
server {
    listen   80;
    location /aaa {
 
    if ($http_user_agent ~ Mozilla) {
            rewrite /aaa /bbb last; 
 
    }
    return 403;
 
    }
 
    location /bbb {
        return 402
    }
 
}

在上述nginx.conf情况下,使用firefox浏览器访问nginx,if匹配成功,执行rewrite重写r->uri(/aaa转换为/bbb),

由于rewrite last标记符存在,停止执行rewrite模块的其他一切指令,参考example 3。

进入到rewrite模块所在的checker函数ngx_http_core_post_rewrite_phase中,由example 2可知r->uri_changed = 1;

于是重新进行server和location匹配,因此在上述例子中,浏览器收到的内容的状态码为402

原文地址:https://www.cnblogs.com/scottieyuyang/p/5722604.html