Ajax跨域、Json跨域、Socket跨域和Canvas跨域等同源策略限制的解决方法

时间：2017-05-15 19:59:15 阅读：324 评论：0 收藏：0 [点我收藏+]

标签：work ott 处理 nts cut safari engine XML iframe

同源是指同样的协议、域名、port，三者都同样才属于同域。不符合上述定义的请求，则称为跨域。

相信每一个开发者都曾遇到过跨域请求的情况，尽管情况不一样，但问题的本质都能够归为浏览器出于安全考虑下的同源策略的限制。

跨域的情形有非常多，最常见的有Ajax跨域、Socket跨域和Canvas跨域。以下列举一些我们常见的跨域情形下。某些浏览器控制台给出的错误提示：

FireFox下的提示：

已阻止交叉源请求：同源策略不同意读取***上的远程资源。能够将资源移动到同样的域名上或者启用 CORS 来解决问题。

Canvas跨域Chrome下的提示：

UncaughtSecurityError:Failed to execute‘getImageData‘ on ‘CanvasRenderingContext2D‘:The canvas has been taintedby cross-origin data.

或：

Imagefrom origin ‘http://js.xx.com‘ has been blocked from loading by Cross-OriginResource Sharing policy: No ‘Access-Control-Allow-Origin‘ header is present onthe requested resource. Origin ‘http://act.xx.com‘ is therefore not allowedaccess.

技术分享

网上有很多解决跨域的方法，大体上有这几种：

1）document.domain+iframe的设置

2）动态创建script

3）利用iframe和location.hash

4）window.name实现的跨域传输数据

5）使用HTML5 postMessage

6）利用flash

7）通过代理，js訪问代理，代理转到不同的域

http://developer.yahoo.com/javascript/howto-proxy.html

8）Jquery JSONP(不能成为真正的Ajax，本质上仍是动态创建script)

http://www.cnblogs.com/chopper/archive/2012/03/24/2403945.html

9）跨域资源共享（CORS）这是HTML5跨域问题的标准解决方式

说明：方案1~方案6见Rain Man所写的文章《JavaScript跨域总结与解决的方法》

http://www.cnblogs.com/rainman/archive/2011/02/20/1959325.html

以下主要就我总结的几种解决跨域的方法。展开说一下。

1）绕开跨域。

适用情形是：动静分离。

example1.com域名下的页面中跨域请求是以JavaScript内联方式实现的，而请求的目标静态资源是放在example2.com域名下，这时能够将运行跨域请求的JavaScript代码块独立出来，放到example2.com上，而example1.com页面通过外联方式引入该静态域名下的js文件。这样，js与请求的图片等资源都在example2.com下。就可以正常訪问到。这样的方法事实上是一种巧妙避开跨域的方法。

2）后台抓取克隆图片。

适用情形：动静不分离（两个域名均执行訪问静态资源）。

example1.com请求example2.com下的资源图片，能够使用PHP抓取图片并在example2.com下生成一份，这样就能够间接訪问到example1.com的静态资源。

html模板演示样例代码：

$("#scratchpad").wScratchPad({ //刮刮卡演示样例，当前域名http://act.xxx.com

width:283,

height:154,

//color: "#a9a9a7",

image2:"imgdata.php?url=http://js.xxx.com/static/activity/sq/guagua315/images/card_inner.jpg",

scratchMove:function() {

}

});

或：

xi=newXMLHttpRequest();

xi.open("GET","imgdata.php?url="+yourImageURL,true);

xi.send();

xi.onreadystatechange=function() {

if(xi.readyState==4 && xi.status==200) {

img=newImage;

img.onload=function(){

ctx.drawImage(img, 0, 0, canvas.width, canvas.height);

}

img.src=xi.responseText;

}

PHP处理代码：

<?php//imgdata.php

$url=$_GET[‘url‘];

$img =file_get_contents($url);

$imgname = substr(strrchr($url,"/"),1);

file_put_contents($fn,$img);

echo $imgname;

上述代码在当前php文件夹下生成了克隆生成了一张图片。

3）后台程序设置Access-Control-Allow-Origin

适用情形：Ajax获取跨域接口的JSON数据。

example1.com请求example2.com的数据接口，则须要在example2.com的数据接口加入跨域訪问授权。

PHP程序中開始出加入header(‘HeaderName:HeaderValue‘); 这种header标记：

header(‘Access-Control-Allow-Origin:*‘);

4）改动server配置启用CORS

适用情形：跨域訪问静态资源。

Access-Control-Allow-Origin是什么作用呢？用于授权资源的跨站訪问。比方,静态资源图片都放在example2.com 域名下, 假设在返回的头中没有设置 Access-Control-Allow-Origin , 那么别的域是没有权限外链你的图片的。

要实现CORS跨域，服务端须要这个一个流程，图片引自html5rocks，附图例如以下

技术分享

a. 对于简单请求，如GET。仅仅须要在HTTP Response后加入Access-Control-Allow-Origin。

b. 对于非简单请求，比方POST、PUT、DELETE等，浏览器会分两次应答。

第一次preflight（method: OPTIONS）。主要验证来源是否合法。并返回同意的Header等。第二次才是真正的HTTP应答。所以server必须处理OPTIONS应答。

这里是一个nginx启用CORS的參考配置演示样例http://enable-cors.org/server_nginx.html。代码：

#
# A CORS (Cross-Origin Resouce Sharing) config for nginx
#
# == Purpose
#
# This nginx configuration enables CORS requests in the following way:
# - enables CORS just for origins on a whitelist specified by a regular expression
# - CORS preflight request (OPTIONS) are responded immediately
# - Access-Control-Allow-Credentials=true for GET and POST requests
# - Access-Control-Max-Age=20days, to minimize repetitive OPTIONS requests
# - various superluous settings to accommodate nonconformant browsers
#
# == Comment on echoing Access-Control-Allow-Origin
# 
# How do you allow CORS requests only from certain domains? The last
# published W3C candidate recommendation states that the
# Access-Control-Allow-Origin header can include a list of origins.
# (See: http://www.w3.org/TR/2013/CR-cors-20130129/#access-control-allow-origin-response-header )
# However, browsers do not support this well and it likely will be
# dropped from the spec (see, http://www.rfc-editor.org/errata_search.php?rfc=6454&eid=3249 ).
# 
# The usual workaround is for the server to keep a whitelist of
# acceptable origins (as a regular expression), match the request‘s
# Origin header against the list, and echo back the matched value.
#
# (Yes you can use ‘*‘ to accept all origins but this is too open and
# prevents using ‘Access-Control-Allow-Credentials: true‘, which is
# needed for HTTP Basic Access authentication.)
#
# == Comment on  spec
#
# Comments below are all based on my reading of the CORS spec as of
# 2013-Jan-29 ( http://www.w3.org/TR/2013/CR-cors-20130129/ ), the
# XMLHttpRequest spec (
# http://www.w3.org/TR/2012/WD-XMLHttpRequest-20121206/ ), and
# experimentation with latest versions of Firefox, Chrome, Safari at
# that point in time.
#
# == Changelog
#
# shared at: https://gist.github.com/algal/5480916
# based on: https://gist.github.com/alexjs/4165271
#

location / {

    # if the request included an Origin: header with an origin on the whitelist,
    # then it is some kind of CORS request.

    # specifically, this example allow CORS requests from
    #  scheme    : http or https
    #  authority : any authority ending in ".mckinsey.com"
    #  port      : nothing, or :
    if ($http_origin ~* (https?://[^/]*\.mckinsey\.com(:[0-9]+)?)$) {
        set $cors "true";
    }

    # Nginx doesn‘t support nested If statements, so we use string
    # concatenation to create a flag for compound conditions

    # OPTIONS indicates a CORS pre-flight request
    if ($request_method = ‘OPTIONS‘) {
        set $cors "${cors}options";  
    }

    # non-OPTIONS indicates a normal CORS request
    if ($request_method = ‘GET‘) {
        set $cors "${cors}get";  
    }
    if ($request_method = ‘POST‘) {
        set $cors "${cors}post";
    }

    # if it‘s a GET or POST, set the standard CORS responses header
    if ($cors = "trueget") {
        # Tells the browser this origin may make cross-origin requests
        # (Here, we echo the requesting origin, which matched the whitelist.)
        add_header ‘Access-Control-Allow-Origin‘ "$http_origin";
        # Tells the browser it may show the response, when XmlHttpRequest.withCredentials=true.
        add_header ‘Access-Control-Allow-Credentials‘ ‘true‘;
        # # Tell the browser which response headers the JS can see, besides the "simple response headers"
        # add_header ‘Access-Control-Expose-Headers‘ ‘myresponseheader‘;
    }

    if ($cors = "truepost") {
        # Tells the browser this origin may make cross-origin requests
        # (Here, we echo the requesting origin, which matched the whitelist.)
        add_header ‘Access-Control-Allow-Origin‘ "$http_origin";
        # Tells the browser it may show the response, when XmlHttpRequest.withCredentials=true.
        add_header ‘Access-Control-Allow-Credentials‘ ‘true‘;
        # # Tell the browser which response headers the JS can see, besides the "simple response headers"
        # add_header ‘Access-Control-Expose-Headers‘ ‘myresponseheader‘;
    }

    # if it‘s OPTIONS, then it‘s a CORS preflight request so respond immediately with no response body
    if ($cors = "trueoptions") {
        # Tells the browser this origin may make cross-origin requests
        # (Here, we echo the requesting origin, which matched the whitelist.)
        add_header ‘Access-Control-Allow-Origin‘ "$http_origin";
        # in a preflight response, tells browser the subsequent actual request can include user credentials (e.g., cookies)
        add_header ‘Access-Control-Allow-Credentials‘ ‘true‘;

        #
        # Return special preflight info
        #
        
        # Tell browser to cache this pre-flight info for 20 days
        add_header ‘Access-Control-Max-Age‘ 1728000;

        # Tell browser we respond to GET,POST,OPTIONS in normal CORS requests.
        #
        # Not officially needed but still included to help non-conforming browsers.
        #
        # OPTIONS should not be needed here, since the field is used
        # to indicate methods allowed for "actual request" not the
        # preflight request.
        #
        # GET,POST also should not be needed, since the "simple
        # methods" GET,POST,HEAD are included by default.
        #
        # We should only need this header for non-simple requests
        # methods (e.g., DELETE), or custom request methods (e.g., XMODIFY)
        add_header ‘Access-Control-Allow-Methods‘ ‘GET, POST, OPTIONS‘;
        
        # Tell browser we accept these headers in the actual request
        #
        # A dynamic, wide-open config would just echo back all the headers
        # listed in the preflight request‘s
        # Access-Control-Request-Headers.
        #
        # A dynamic, restrictive config, would just echo back the
        # subset of Access-Control-Request-Headers headers which are
        # allowed for this resource.
        #
        # This static, fairly open config just returns a hardcoded set of
        # headers that covers many cases, including some headers that
        # are officially unnecessary but actually needed to support
        # non-conforming browsers
        # 
        # Comment on some particular headers below:
        #
        # Authorization -- practically and officially needed to support
        # requests using HTTP Basic Access authentication. Browser JS
        # can use HTTP BA authentication with an XmlHttpRequest object
        # req by calling
        # 
        #   req.withCredentials=true,  and
        #   req.setRequestHeader(‘Authorization‘,‘Basic ‘ + window.btoa(theusername + ‘:‘ + thepassword))
        #
        # Counterintuitively, the username and password fields on
        # XmlHttpRequest#open cannot be used to set the authorization
        # field automatically for CORS requests.
        #
        # Content-Type -- this is a "simple header" only when it‘s
        # value is either application/x-www-form-urlencoded,
        # multipart/form-data, or text/plain; and in that case it does
        # not officially need to be included. But, if your browser
        # code sets the content type as application/json, for example,
        # then that makes the header non-simple, and then your server
        # must declare that it allows the Content-Type header.
        # 
        # Accept,Accept-Language,Content-Language -- these are the
        # "simple headers" and they are officially never
        # required. Practically, possibly required.
        #
        # Origin -- logically, should not need to be explicitly
        # required, since it‘s implicitly required by all of
        # CORS. officially, it is unclear if it is required or
        # forbidden! practically, probably required by existing
        # browsers (Gecko does not request it but WebKit does, so
        # WebKit might choke if it‘s not returned back).
        #
        # User-Agent,DNT -- officially, should not be required, as
        # they cannot be set as "author request headers". practically,
        # may be required.
        # 
        # My Comment:
        #
        # The specs are contradictory, or else just confusing to me,
        # in how they describe certain headers as required by CORS but
        # forbidden by XmlHttpRequest. The CORS spec says the browser
        # is supposed to set Access-Control-Request-Headers to include
        # only "author request headers" (section 7.1.5). And then the
        # server is supposed to use Access-Control-Allow-Headers to
        # echo back the subset of those which is allowed, telling the
        # browser that it should not continue and perform the actual
        # request if it includes additional headers (section 7.1.5,
        # step 8). So this implies the browser client code must take
        # care to include all necessary headers as author request
        # headers.
        # 
        # However, the spec for XmlHttpRequest#setRequestHeader
        # (section 4.6.2) provides a long list of headers which the
        # the browser client code is forbidden to set, including for
        # instance Origin, DNT (do not track), User-Agent, etc.. This
        # is understandable: these are all headers that we want the
        # browser itself to control, so that malicious browser client
        # code cannot spoof them and for instance pretend to be from a
        # different origin, etc..
        #
        # But if XmlHttpRequest forbids the browser client code from
        # setting these (as per the XmlHttpRequest spec), then they
        # are not author request headers. And if they are not author
        # request headers, then the browser should not include them in
        # the preflight request‘s Access-Control-Request-Headers. And
        # if they are not included in Access-Control-Request-Headers,
        # then they should not be echoed by
        # Access-Control-Allow-Headers. And if they are not echoed by
        # Access-Control-Allow-Headers, then the browser should not
        # continue and execute actual request. So this seems to imply
        # that the CORS and XmlHttpRequest specs forbid certain
        # widely-used fields in CORS requests, including the Origin
        # field, which they also require for CORS requests.
        #
        # The bottom line: it seems there are headers needed for the
        # web and CORS to work, which at the moment you should
        # hard-code into Access-Control-Allow-Headers, although
        # official specs imply this should not be necessary.
        # 
        add_header ‘Access-Control-Allow-Headers‘ ‘Authorization,Content-Type,Accept,Origin,User-Agent,DNT,Cache-Control,X-Mx-ReqToken,Keep-Alive,X-Requested-With,If-Modified-Since‘;

        # build entire response to the preflight request
        # no body in this response
        add_header ‘Content-Length‘ 0;
        # (should not be necessary, but included for non-conforming browsers)
        add_header ‘Content-Type‘ ‘text/plain charset=UTF-8‘;
        # indicate successful return with no content
        return 204;
    }
    # --PUT YOUR REGULAR NGINX CODE HERE--
}

server解析流程例如以下：

a.首先查看http头部有无origin字段；

b.假设没有，或者不同意，直接当成普通请求处理，结束；

c.假设有而且是同意的。那么再看是否是preflight(method=OPTIONS)；

d.假设是preflight，就返回Allow-Headers、Allow-Methods等，内容为空；

e.假设不是preflight。就返回Allow-Origin、Allow-Credentials等，并返回正常内容。

若server为nginx。能够在nginx的conf文件里增加下面内容：

location / {
  add_header Access-Control-Allow-Origin *;
}

若server为Apache，则能够依照例如以下配置：

<IfModule mod_setenvif.c>
    <IfModule mod_headers.c>
        <FilesMatch "\.(cur|gif|ico|jpe?g|png|svgz?|webp)$">
            SetEnvIf Origin ":" IS_CORS
            Header set Access-Control-Allow-Origin "*" env=IS_CORS
        </FilesMatch>
    </IfModule>
</IfModule>

为安全起见，Access-Control-Allow-Origin也可设为特定域名的方式。

在HTML5中，有些HTML元素为CORS提供了支持，如img、video新增了crossOrigin属性。属性值能够为anonymous或use-credentials。

比方，canvas画图要用到跨域图片，在JavaScript中要设置img.crossOrigin="Anonymous";

var img = new Image,
    canvas = document.createElement("canvas"),
    ctx = canvas.getContext("2d"),
    src = "http://example.com/image"; // insert image url here

img.crossOrigin = "Anonymous";

img.onload = function() {
    canvas.width = img.width;
    canvas.height = img.height;
    ctx.drawImage( img, 0, 0 );
    localStorage.setItem( "savedImageData", canvas.toDataURL("image/png") );
}
img.src = src;
// make sure the load event fires for cached images too
if ( img.complete || img.complete === undefined ) {
    img.src = "data:image/gif;base64,R0lGODlhAQABAIAAAAAAAP///ywAAAAAAQABAAACAUwAOw==";
    img.src = src;
}

上述配置完毕后，重新启动server。CORS启用。

然后我们再刷新页面，查询请求头的參数，能够发现多出一个：Access-Control-Allow-Origin:*

，到此证明server配置已经生效。

同一时候我们的canvas画图也能够正常使用了。

刷新页面返回请求响应结果后，HTTP Request Headers的内容：

Remote Address:222.132.18.xx:80

Request URL:http://js.xx.com/static/activity/sq/guagua315/images/card_inner.jpg

Request Method:GET

Status Code:200 OK

Request Headersview source

Accept:image/webp,*/*;q=0.8

Accept-Encoding:gzip, deflate, sdch