nbcommunication
-
Content Count
44 -
Joined
-
Last visited
-
Hello, I thought it might be useful to post a CSP I've recently deployed using this module. Every site is different - there's no prescriptive policy and that's the main caveat here. This is for a site with an embedded Shopify store, an Issuu embed, a Google Tour embed and Google Maps implementation (JS API). It also uses Font Awesome 5 from their CDN, jQuery from CDNJS, and some Google Fonts. It also has TextformatterVideoEmbed installed alongside its extended options module. default-src 'none'; script-src 'self' cdnjs.cloudflare.com *.google.com *.gstatic.com *.googleapis.com www.google-analytics.com www.googletagmanager.com e.issuu.com sdks.shopifycdn.com; style-src 'self' 'unsafe-inline' cdnjs.cloudflare.com *.googleapis.com use.fontawesome.com; img-src 'self' data: *.google.com *.googleapis.com *.gstatic.com *.ggpht.com www.google-analytics.com www.googletagmanager.com brand.nbcommunication.com *.shopify.com sdks.shopifycdn.com; connect-src 'self' www.google-analytics.com ocean-kinetics.myshopify.com; font-src 'self' fonts.gstatic.com use.fontawesome.com; object-src 'self'; media-src 'self' data:; manifest-src 'self'; frame-src www.google.com www.youtube.com www.youtube-nocookie.com player.vimeo.com w.soundcloud.com e.issuu.com; form-action 'self'; base-uri 'self' The Shopify embed script and Google Analytics initialisation have been moved into script files so there are no inline scripts at all. The script-src 'unsafe-inline' directive is an obstacle to getting that A+ on Observatory. Google Analytics is also a bit of an impediment to getting a top-drawer score, as its script doesn't use SRI. However, there is a reason for that as I understand it - it is a script that just loads other scripts so SRI implementation would just be token, it wouldn't actually be improving security. Still, it is possible to get A+ without dealing with this. It would be great to get some discussion going on CSP implementation - I'm only a few weeks in myself, so have much to learn! Cheers, Chris NB
-
Update to 0.0.2 Deploy mode (only enable for superuser when off) Define params to exclude from reporting (e.g. originalPolicy) Filtering of selected false positive patterns Improvements to HTML page check
-
Please let me know how you find it - this module is really just a result of looking into it the past two weeks to try and get an A+ on Mozilla Observatory (did it! 😆) so I'd really appreciate any feedback!
-
Wondering how to get that A+ rating on Mozilla Observatory? Now you can with ⭐⭐⭐MarkupContentSecurityPolicy⭐⭐⭐ Of course, MarkupContentSecurityPolicy does not guarantee an A+ rating, but it does help you implement a Content Security Policy for your ProcessWire website. Markup Content Security Policy Configure and implement a Content Security Policy for all front-end HTML pages. This module should only be used in production once it has been fully tested in development. Implementing a Content Security Policy on a site without testing will almost certainly break something! Overview Website Security Auditing Tools such as Mozilla Observatory will only return a high score if a Content Security Policy is implemented. It is therefore desirable to implement one. A common way of adding the Content-Security-Policy header would be to add it to the .htaccess file in the site's root directory. However, this means the policy would also cover the ProcessWire admin, and this limits the level of security policy you can add. The solution is to use the <meta> element to configure a policy, for example: <meta http-equiv="Content-Security-Policy" content="default-src 'self'; img-src https://*; child-src 'none';">. MarkupContentSecurityPolicy places this element with your configured policy at the beginning of the <head> element on each HTML page of your site. There are some limitations to using the <meta> element: Not all directives are allowed. These include frame-ancestors, report-uri, and sandbox. The Content-Security-Policy-Report-Only header is not supported, so is not available for use by this module. Configuration To configure this module, go to Modules > Configure > MarkupContentSecurityPolicy. Directives The most commonly used directives are listed, with a field for each. The placeholder values given are examples, not suggestions, but they may provide a useful starting point. You will almost certainly need to use 'unsafe-inline' in the style-src directive as this is required by some modules (e.g. TextformatterVideoEmbed) or frameworks such as UIkit. Should you wish to add any other directives not listed, you can do so by adding them in Any other directives. Please refer to these links for more information on how to configure your policy: https://developer.mozilla.org/en-US/docs/Web/HTTP/CSP https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Content-Security-Policy https://scotthelme.co.uk/content-security-policy-an-introduction/ https://developers.google.com/web/fundamentals/security/csp/ Violation Reporting Because the report-uri directive is not available, when Violation Reporting is enabled a script is added to the <head>which listens for a SecurityPolicyViolationEvent. This script is based on https://developer.mozilla.org/en-US/docs/Web/API/SecurityPolicyViolationEvent and POSTs the generated report to ?csp-violations=1. The module then logs the violation report to csp-violations. Unfortunately, most of the violations that are reported are false positives, and not actual attempts to violate the policy. These are most likely from browser extensions and are not easy to determine and filter. For this reason, there is no option for the report to be emailed when a policy is violated. Instead, you can specify an endpoint for the report to be sent to. This allows you to handle additional reporting in a way that meets your needs. For example, you may want to log all reports in a central location and send out an email once a day to an administrator notifying them of all sites with violations since the last email. Retrieving the Report To retrieve the report at your endpoint, the following can be used: $report = file_get_contents("php://input"); if(!empty($report)) { $report = json_decode($report, 1); if(isset($report) && is_array($report) && isset($report["documentURI"])) { // Do something } } Debug Mode When this is enabled, a range of information is logged to markup-content-security-policy. This is probably most useful when debugging a reporting endpoint. Additional .htaccess Rules To get an A+ score on Mozilla Observatory, besides using HTTPS and enabling the HSTS header, you can also place the following prior to ProcessWire's htaccess directives: Header set Content-Security-Policy "frame-ancestors 'self'" Header set Referrer-Policy "no-referrer-when-downgrade" Installation Download the zip file at Github or clone the repo into your site/modules directory. If you downloaded the zip file, extract it in your sites/modules directory. In your admin, go to Modules > Refresh, then Modules > New, then click on the Install button for this module. ProcessWire >= 3.0.123 is required to use this module.
- 4 replies
-
- 11
-
-
I've been doing a little rejigging of our default htaccess file after yesterday's dev release, and trying to understand Section 18 a bit better too... In section 18 we use a rule based on what is outlined here: https://processwire.com/blog/posts/optimizing-404s-in-processwire/ RewriteCond %{REQUEST_URI} !\.(jpg|jpeg|gif|png|ico|webp|svg|php|cgi|pl|asp|rar|zip)$ [NC] This is entered after the existing section 18 rules, which are left commented out. After the end of ProcessWire's htaccess rules we have: <FilesMatch "\.(jpg|jpeg|gif|png|ico|webp|svg|php|cgi|pl|asp|rar|zip)$"> ErrorDocument 404 "The requested file was not found. </FilesMatch> Without that rule, I get the ProcessWire 404 page when accessing a non-existent file matching one of those types. On a different htaccess-related note, I recommend the 6G htaccess Firewall https://perishablepress.com/6g/. We have this at the start of our default .htaccess file, followed by: ErrorDocument 403 "Sorry, you are not permitted to access this resource. The only issue I've come across after a few years of use is with the autocomplete Page field, when using OR selectors (e.g. template=home|default). I wrote this hook as a remedy: <?php // Replace pipes (|) with %7C in PageAutocomplete data-url attribute // This gets around 6G htaccess rules which disallows pipes in urls $wire->addHookAfter("InputfieldPageAutocomplete::render", function(HookEvent $event) { $out = $event->return; if(strpos($out, "data-url") === false) return; $url = explode("'", ltrim(explode("data-url=", $out)[1], "'"))[0]; $event->return = str_replace($url, str_replace(" ", "+", str_replace("|", "%7C", $url)), $out); }); Cheers, Chris
-
Hey, Launched a site today with a form that sends attachments and it turned out that our production server doesn't have mime_content_type() 🙁 https://github.com/chriswthomson/WireMailMailgun/pull/17 I've added a private method getMimeType() which checks first for mime_content_type(), then finfo_open(), then falls back to inferring based on file extension if neither exist. This is based on https://www.php.net/manual/en/function.mime-content-type.php#87856 and https://gist.github.com/Erutan409/8e774dfb2b343fe78b14#file-mimetype-php Also, a WireException is now thrown if an attachment has been added using WireMail::attachment() but curl_file_create() doesn't exist. Cheers, Chris
-
Hi @Schwab - I'm not sure how this can be done, although given that there are a number of WireMail extension modules out there, I'm sure it must be possible and discussed previously. Might be worth having a look through the threads for WireMail SMTP etc to see if there is any info there. I'm assuming this is @Macrura's version on the module repo - these methods came from the original "plauclair" version. I removed them entirely from my version as they were unnecessary, and didn't look like they would work anyway, given that trackClicks and trackOpens should be booleans (technically integers, but evaluate as either true or false). I updated chriswthomson/WireMailMailgun last week with fallbacks etc, tested on 2.7.2 and above. Ended up moving straight onto another task and completely forgot to post!
-
@Macrura - Thanks for the info. I wasn't thinking outwith explicit calls to WireMailMailgun, so that makes sense. I'll probably still add the setSender() and setRegion() functions though, might be useful to have. Cheers, Chris
-
@jens.martsch - thinking through it a little more, I think the addition of h:sender should only happen if the "from" domain is different to $this->domain. I'll pop that in place soon. I'd wondered whether using this would effect the DKIM auth etc but a few quick tests suggest that it doesn't 🙂. @Macrura - I've been thinking about the Dynamic Domain setting, which I'd implemented based on your version but didn't give it much thought. Does it actually work without specifying the API key for the different domain? I wonder whether this could be removed and the following be the way to send from a different domain? <?php $mg = $mail->new(); $sent = $mg->to($to) ->setDomainName("mg.differentdomain.com") ->setApiKey($differentApiKey) ->setRegion("eu") // If the region is different //->setSender("mg.differentdomain.com", $differentApiKey, "eu") // Alternative, region optional ->subject($subject) ->bodyHTML($message) ->send(); Cheers, Chris
-
Hi @jens.martsch - I've merged your request. I'd always thought that was the correct behaviour because if you send from a domain that isn't what you have in Mailgun (e.g. @test.com when @mg.test.com is your verified domain) then the "on behalf of" is the correct result. However, if one line fixes that, that's great! Cheers, Chris
-
New blog post: Lots of module updates
nbcommunication replied to ryan's topic in News & Announcements
Much love to you and your family Ryan, haven't had that experience before, and not looking forward to having it in future - cats make a bigger difference to your life than you'd think! Thanks again for all you do. -
Hi, I've added an addInlineImage() method to https://github.com/chriswthomson/WireMailMailgun <?php $img = $pages->get(1)->images->getRandom(); // An example $customFilename = "test.$img->ext"; // Don't want to use the basename for this example $mg = $mail->new(); $mg->to("your@email.com") ->subject("Test Inline Image") ->bodyHTML("<p>Testing, 123, is there an image for you to see?</p><img src='cid:$customFilename'>") ->addInlineImage($img->filename, $customFilename) // Second argument is optional ->send(); If the image isn't referenced in the HTML it just appears as a standard attachment from what I can see. Cheers, Chris
-
Having a look at https://documentation.mailgun.com/en/latest/api-sending.html#sending inline is a separate parameter that would need to be implemented. I’ll have a look at this on Monday when I’m back at the coalface. 🙂 Chris
-
Agreed. I think that perhaps the best solution is for you to maintain the module, integrating anything applicable from my version. I’d be happy to help with this process.
-
I’m not really sure to be honest - I haven’t added anything to the directory before so will need to have a look through the process. A couple of points: - is it acceptable/desirable to have a module that’s only compatible with the lastest master (123) and beyond? - ProMailer: I’ve not looked at this myself, and don’t expect to in the near future. My assumption is that Ryan has developed this to be compatible with the initial WireMailMailgun, which has differences in batchMode for cc/bcc - maybe this wouldn’t be an issue at all, I don’t know. I guess what I’m trying to say is that I’d be delighted to add it to the directory, but feel that the original should be retained. Perhaps the new version should be renamed? WireMailMailgun3? Cheers, Chris