开发者

Why are cookies unrecognized when a link is clicked from an external source (i.e. Excel, Word, etc...)

开发者 https://www.devze.com 2022-12-27 16:57 出处:网络
I noticed that when a link is clicked externally from the web browser, such as from Excel or Word, that my session cookie is initially unrecognized, even if the link opens up in a new tab of the same

I noticed that when a link is clicked externally from the web browser, such as from Excel or Word, that my session cookie is initially unrecognized, even if the link opens up in a new tab of the same browser window.

The browser ends up recognizing its cookie eventually, but I am puzzled as to why that initial link from Excel or Word doesn't wo开发者_如何学Pythonrk. To make it even more challenging, clicking a link works fine from Outlook.

Does anybody know why this might be happening? I'm using the Zend Framework with PHP 5.3.


This is because MS Office is using Hlink.dll component to lookup if the link is Office document or something else. MS Office expect to open the document linked within documents without the aid of external browser (using Hlink.dll component of IE6).

If session cookie protects website Hlink naturally is being redirected to login page and having reached HTML page and not able to "understand" it opens it in external browser. Note that it opens not original URL (expected behavior) but the result of redirect, even if it was 302 redirect.

Microsoft has that bug in unsupported component (Hlink.dll), instead of recognizing the bug they turn it over to our head (trying to convince us that it is flaw of SSO system we use, i.e. session cookies) and refuses to upgrade it. It offers workaround that turns off the lookup functionality of MS Office:

HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\
  Office\9.0\Common\Internet\ForceShellExecute:DWORD=1

Or offer us to workaround serverside, to avoid HTTP redirects and change into Javascript redirects or META REFRESH redirects (i.e. to have Hlink get text/html page on original URL and make it run external browser to handle it).


Server side this worked for me in IIS (using a rewrite rule)

<rule name="WordBypass" enabled="true" stopProcessing="true">
    <match url=".*" />
    <conditions>
        <add input="{HTTP_USER_AGENT}" pattern="Word|Excel|PowerPoint|ms-office" />
    </conditions>
    <action type="CustomResponse" statusCode="200" statusReason="Refresh" statusDescription="Refresh" />
</rule>


We had this same problem and wrote an open source gem to help those using rails: https://github.com/spilliton/fix_microsoft_links

You can use the same approach we used on any framework though:

  1. Detect if the user agent is from a Microsoft product
  2. Render a blank html page with a meta refresh tag that will cause the browser to refresh the page with the correct cookies

Example code here: https://github.com/spilliton/fix_microsoft_links/blob/master/lib/fix_microsoft_links.rb


PHP solution:

This prevents the MS product recognising the redirect. MS therefore launches a browser from the required link.

if (isset($_SERVER['HTTP_USER_AGENT']))
{
    $http_user_agent = $_SERVER['HTTP_USER_AGENT']; 
    if (preg_match('/Word|Excel|PowerPoint|ms-office/i', $http_user_agent)) 
    {
        // Prevent MS office products detecting the upcoming re-direct .. forces them to launch the browser to this link
        die();
    }
}

.. redirect after this code


Fix for VB.NET:

Dim userAgent As String = System.Web.HttpContext.Current.Request.UserAgent

If userAgent.Contains("Word") Or userAgent.Contains("Excel") Or userAgent.Contains("PowerPoint") Or userAgent.Contains("ms-office") Then
       System.Web.HttpContext.Current.Response.Clear()
       System.Web.HttpContext.Current.Response.Write("<html><head><meta http-equiv='refresh' content='0'/></head><body></body></html>")
       System.Web.HttpContext.Current.Response.End()
End If

It basically forces the browser to refresh the page, so the request comes in with the user agent of the browser and all the correct cookies.


Here is a solution for C# ASP.NET based on spilliton's answer above. In Global.asax.cs, add the following:

    private static string MSUserAgentsRegex = @"[^\w](Word|Excel|PowerPoint|ms-office)([^\w]|\z)";
    protected void Application_OnPostAuthenticateRequest(object sender, EventArgs e)
    {
        if (System.Text.RegularExpressions.Regex.IsMatch(Request.UserAgent, MSUserAgentsRegex))
        {
            Response.Write("<html><head><meta http-equiv='refresh' content='0'/></head><body></body></html>");
            Response.End();
        }
    }


1.From excel/word point to http://example.com/from_excel.php

2.In "from_excel.php" redirect to page where you use session

<script>document.location.href = "http://example.com/page_with_session.php"; </script>


Here is how to workaround this with Java and Spring via a Filter:

/**
 * To see why this is necessary, check out this page:
 * https://support.microsoft.com/en-gb/help/899927.
 */
public class MicrosoftFilter extends OncePerRequestFilter {
  @Override
  protected void doFilterInternal(final HttpServletRequest request,
      final HttpServletResponse response,
      final FilterChain filterChain) throws ServletException, IOException {
    //Serve up a blank page to anything with a Microsoft Office user agent, forcing it to open the
    //URL in a browser instead of trying to pre-fetch it, getting redirected to SSO, and losing
    //the path of the original link.
    if (!request.getHeader("User-Agent").contains("ms-office")) {
      filterChain.doFilter(request, response);
    }
  }
}

/**
 * Security configuration.
 */
@Configuration
public class SecurityConfiguration {
  @Bean
  public FilterRegistrationBean microsoftFilterRegistrationBean() {
    FilterRegistrationBean<MicrosoftFilter> registrationBean = new FilterRegistrationBean<>();
    registrationBean.setFilter(new MicrosoftFilter());
    registrationBean.setOrder(Ordered.HIGHEST_PRECEDENCE);
    return registrationBean;
  }
}


We are seeing a problem that TWO Chrome tabs are opened when clicking an URL from MS Word, and the page to open has JavaScript redirection: window.location.href=blabla

By debugging from the servers side, we confirmed that there are requests sent from Office app, besides Chrome. This is so wierd.

But anyway, by checking the request header "User-Agent", and returning an empty page to Office apps, our TWO tabs issue got resolved. That's definitely the right thing to do!


Here is an example of the fix using a dotnet core middleware:

public class MicrosoftOfficeLinksHandlingMiddleware
{
    private static readonly Regex MsUserAgentsRegex = new Regex(@"[^\w](Word|Excel|PowerPoint|ms-office)([^\w]|\z)");
    private readonly RequestDelegate _next;

    public MicrosoftOfficeLinksHandlingMiddleware(RequestDelegate next)
    {
        _next = next;
    }

    public async Task Invoke(HttpContext context)
    {
        string userAgent = context.Request.Headers["User-Agent"].FirstOrDefault();

        if (userAgent != null && MsUserAgentsRegex.IsMatch(userAgent))
        {
            // just return an empty response to the office agent
            return;
        }

        await _next(context);
    }
}


Here is my solution for this in WordPress. Add this to functions.php in your theme or another plugin file.

This may be helpful if your system, like WP, sends logged out users to a login page with a redirect to the page they were trying to access. Word was sending users to this page, but then WP wasn't properly handling the case where a user was already logged in. This code checks if there is a current user and a redirect_to param passed. If so, it redirects to the redirect_to location.

function my_logged_in_redirect_to()
{
global $current_user;

if($current_user->ID && $_REQUEST['redirect_to'])
{           
    wp_redirect($_REQUEST['redirect_to']);
    exit;
}
}
add_action('wp', 'my_logged_in_redirect_to');


Here's a VBA fix, for Excel. The same concept can be applied for Microsoft Word. Basically, rather than firing off the link from within Excel, the code executes the link from within a shell. Here's the code:

Private Sub Worksheet_FollowHyperlink(ByVal objLink As Hyperlink)
    Application.EnableEvents = False
    Dim strAddress As String
    strAddress = "explorer " & objLink.TextToDisplay
    Dim dblReturn As Double
    dblReturn = Shell(strAddress)
    Application.EnableEvents = True
End Sub
  1. For the Excel sheet that contains the links, right-click the sheet tab and click View Code. The VBA editor appears.
  2. Paste the code into the window, and close the editor.
  3. Modify each link in the page so it simply points back to the cell that it is in. To do so:
  4. Right-click the link, and click Edit Hyperlink. An Edit Hyperlink window appears.
  5. Click Place In This Document.
  6. Click the sheet name.
  7. For Type the cell reference, enter a cell reference (e.g. A4).
  8. Click OK.

A couple of notes:

  • You will need to save the spreadsheet as a macro-enabled spreadsheet (.xlsm). When users open the spreadsheet, they will be asked to enable macros. If they answer No, the links will not work.
  • These instructions are based on Excel 2010. Presumably later versions are similar.


I can't believe they call this a feature. However, here's a featurefix for Apache:

RewriteEngine On

# Send a 200 to MS Office so it just hands over control to the browser
# It does not use existing session cookies and would be redirected to the login page otherwise
# https://www.wimpyprogrammer.com/microsoft-office-link-pre-fetching-and-single-sign-on/

RewriteCond %{HTTP_USER_AGENT} ;\sms-office(\)|;)
RewriteRule .* - [R=200,L]

Might not be best performance wise, as the whole page gets sent instead of an empty response, but I did not want to add another Apache modules just for fixing such an idio^H^H^H^H feature.


NGINX solution below:

if ($http_user_agent ~* Word|Excel|PowerPoint|ms-office) {
    return 200 '<html><head><meta http-equiv="refresh" content="0"/></head><body></body></html>';
}

You can put it in the server or location block. Works like charm.


I had to solve this issue for an ASP.NET site but I only wanted to use javascript/ jQuery:

var isCoBrowse = ('<%= Session["user"].ToString().ToLower() %>' != '0');
if (isCoBrowse && window.location.href.indexOf('ReturnUrl=') >= 0 && window.location.href.indexOf('dllCheq') == -1) {
    //redirect to the ReturnUrl & add dllCheq to the URI
    var toRedirect = decodeURIComponent(gup('ReturnUrl', window.location.href)) + '&dllCheq';
    window.location = toRedirect;
}

I got the gup function from: How to get the value from the URL parameter?


Use fix provided by microsoft given link below. https://support.microsoft.com/en-us/kb/218153


And in ColdFusion / Lucee

<cfif cgi.HTTP_USER_AGENT contains "Excel" OR cgi.HTTP_USER_AGENT contains "ms-office">
    <cfabort/>
</cfif>


I suspect this is a matter of how you are setting the cookie(s).

Due to the nature of how the web was created, example.com is not seen as the same domain as www.example.com; hence: you can be logged in at www.example.com and not logged in at example.com.

So in other words, check the URL in your word or excel file - is it the same domain as how you are logged in within your browser?

There are two fixes/solutions to this cookie inconsistency: 1. redirect anyone who tries to load your site without the www. to the same page with the www. (or vice versa), or 2. when you are setting the cookie, make sure to specify the domain argument as ".example.com". The leading dot indicates the cookie should be valid on all subdomains of that domain as well.

I suspect the reason the browser eventually recognizes it is because you probably eventually end up landing on a URL with the same domain structure as how you are logged in.

Hope this helps.

0

精彩评论

暂无评论...
验证码 换一张
取 消

关注公众号