Skip to content

Latest commit

 

History

History
229 lines (156 loc) · 13 KB

Headers.md

File metadata and controls

229 lines (156 loc) · 13 KB

Headers

Functions that send/handle different HTTP headers.

I hope that at some point this will become popular. While some frameworks do seem to implement some similar functionality it seems to be generally ignored, while HTTP headers can make user experience better and more secure. If you start using Headers (especially security) it will (by default) force you into following best practices when designing your website. If you need to loosen up your security for some reason, it is possible in a relatively convenient way. If you find that it isn't (like you need to use unsafe directives, for example), most likely you are trying to save a security hole, that you should not be saving.

\Simbiat\HTTP20\Headers::nameOfFunction();

cacheControl

cacheControl(string $string, string $cacheStrat = '', bool $exit = false, string $postfix ='');

Allows you to send appropriate Cache-Control headers (refer to https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Cache-Control for explanation on parameters):

switch (mb_strtolower($cacheStrat, 'UTF-8')) {
	case 'aggressive':
	    header('Cache-Control: max-age=31536000, immutable, no-transform');
	    break;
	case 'private':
	    header('Cache-Control: private, no-cache, no-store, no-transform');
	    break;
	case 'live':
	    header('Cache-Control: no-cache, no-transform');
	    break;
	case 'month':
	    #28 days to be more precise
	    header('Cache-Control: max-age=2419200, must-revalidate, stale-while-revalidate=86400, stale-if-error=86400, no-transform');
	    break;
	case 'week':
	    header('Cache-Control: max-age=604800, must-revalidate, stale-while-revalidate=86400, stale-if-error=86400, no-transform');
	    break;
	case 'day':
	    header('Cache-Control: max-age=86400, must-revalidate, stale-while-revalidate=86400, stale-if-error=86400, no-transform');
	    break;
	case 'hour':
	    header('Cache-Control: max-age=3600, must-revalidate, stale-while-revalidate=86400, stale-if-error=86400, no-transform');
	    break;
}

ETag processing using eTag() will happen regardless (unless string is empty: then it will simply make no sense).

$exit if set to true will exit the script right after HTTP 304 is sent (that is we hit the cache).

$postfix is an optional string to add to eTag string and used mainly for zEcho, to comply with recommendations when using compression.

eTag

eTag(string $etag, bool $exit = false);

Sends ETag header and handles its validation depending on requesting headers (If-Match, If-None-Match).

$exit if set to true will exit the script right after HTTP 304 is sent (that is we hit the cache).

lastModified

lastModified(int $modTime = 0, bool $exit = false);

Sends Last-Modified header based on either provided parameter, or the freshest date of all the script files used to generate a page. Also handles HTTP_IF_MODIFIED_SINCE header from client, if it was sent, allowing for some performance improvement if cache can be used.

$exit if set to true will exit the script right after HTTP 304 is sent (that is we hit the cache).

performance

performance(int $keepalive = 0, array $clientHints = []);

Sends some headers that may improve performance on client side.

$keepalive is used for Keep-Alive header governing how long the connection should stay up. Header will be sent only if server is using HTTP version other than 2.0.

$clientHints instructs clients, that your server supports Client Hints (https://developer.mozilla.org/en-US/docs/Glossary/Client_hints) like DPR, Width, Viewport-Width, Downlink, etc. and client should cache the output accordingly, in order to increase allow cache hitting and thus improve performance.

security

security(string $strat = 'strict', array $allowOrigins = [], array $exposeHeaders = [], array $allowHeaders = [], array $allowMethods = []);

Sends headers that can improve security of your page.

$strat allows to select one fo 3 strategies for CORS headers (https://developer.mozilla.org/ru/docs/Web/HTTP/CORS). strict is default, because security is king.

case 'mild':
	header('Cross-Origin-Embedder-Policy: unsafe-none');
	header('Cross-Origin-Embedder-Policy: same-origin-allow-popups');
	header('Cross-Origin-Resource-Policy: same-site');
	header('Referrer-Policy: strict-origin');
	break;
case 'loose':
	header('Cross-Origin-Embedder-Policy: unsafe-none');
	header('Cross-Origin-Opener-Policy: unsafe-none');
	header('Cross-Origin-Resource-Policy: cross-origin');
	header('Referrer-Policy: strict-origin-when-cross-origin');
	break;
case 'strict':
default:
	header('Cross-Origin-Embedder-Policy: require-corp');
	header('Cross-Origin-Opener-Policy: same-origin');
	header('Cross-Origin-Resource-Policy: same-origin');
	header('Referrer-Policy: no-referrer');
	break;

$allowOrigins (https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Access-Control-Allow-Origin) allows you to set a list of allowed Origin values, that can access your page. If empty - will allow access to all (*).

Access-Control-Allow-Origin allows only 1 values by specification, but $allowOrigins allows to overcome it by doing validation against the list you've provided. Then, if the origin is allowed - access will be granted, if not - 403 will be sent and code will exit.

$exposeHeaders (https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Access-Control-Expose-Headers) allows you to set a list of headers, that you are ok to expose to client. Headers, that are provided by this class, will always be exposed.

$allowHeaders (https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Access-Control-Allow-Headers) lets you add headers, that you are willing to accept and use to change states in your code.

$allowMethods (https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Access-Control-Allow-Methods) allows you to restrict accepted methods. If request to page is done by a method not in the list - it will be rejected (405). By default, GET, POST and HEAD are the only allowed.

contentPolicy

contentPolicy(array $cspDirectives = [], bool $reportOnly = false, bool $reportUri = false);

Sends Content-Security-Policy header, that improves your page security. It's done separately from other security stuff, because unlike the rest of the headers this is usable only for HTML.

$cspDirectives (https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Content-Security-Policy) allows you to provide a list of directives and their settings (with validation) to control CSP headers. By default, essentially everything is either disabled or allowed only from self, which give you a solid base in terms of restricting access.

$reportOnly allows you to control, whether you only report (Content-Security-Policy-Report-Only) CPS violations or report and block them. Be default it's set as false for security enforcement. Note, that if it's set to true, but you do not provide report-to directive no CSP header will be sent, reducing your security. For that reason, if you do want to report, I can suggest using https://rapidsec.com/ which is free. Also note, that while report-uri is temporary added until report-to is supported by all browsers, report-uri will be discarded if it's provided without report-to to encourage the use of a modern directive.

$reportUri will add report-uri in the headers as well. It is deprecated, thus defaults to false, but if you want to support still - you can.

features

features(array $features = [], bool $forceCheck = true, bool $permissions = false);

Allows controlling different features through Feature-Policy header. It should only be used, when sending HTML.

$features expects associative array, where each key is name of the policy in lower case and value - expected allow list. If an empty array is sent, default values will be applied (most features are disabled).

$forceCheck is added for futureproofing, but is enabled by default. If set to true will check if the feature is "supported" (present in default array) and value complies with the standard. Setting it to false will allow you to utilize a feature or value not yet supported by the library.

$permissions is a flag to toggle Permissions-Policy, which is replacement for Feature-Policy header. Alternatively you can use features(array $features = [], bool $forceCheck = true) signature, which will call features internally.

secFetch

secFetch(array $site = [], array $mode = [], array $user = [], array $dest = [], bool $strict = true);

Allows validation of Sec-Fetch-* headers from client against the provided list of allowed values. Each of the 4 array values represent a list of values of respective Sec-Fetch-* header, which you allow to be processed. For more information refer https://www.w3.org/TR/fetch-metadata/

$strict allows enforcing compliance with supported values only. Current W3C allows ignoring headers, but we want to be stricter so setting this to true by default. You may want to set to false to in situations, when your site is parsed by certain bots (including those from search engines), since they usually do not set this value.

Be mindful: unlike security, which, essentially, attempts to be as secure as possible by default, this may be too lax for some use-cases. It is recommended, that you call it with different parameters depending on what is calling what on your server. For example, you may want to restrict certain code getting called with Sec-Fetch-Destination: image, especially, if it's a POST request, let alone DELETE. Thus, the best way to use this is in some switch or if-elseif-else scenario, rather than universally.

clientReturn

clientReturn(string $code = '500', bool $exit = true);

Returns a selected HTTP status code (defaults to 500) with option to forcibly close HTTP connection ($exit = true). This is mostly useful for returning error codes, especially, when you want to close the connection, even if the client is still sending something, thus the default values are 500 and true.

redirect

redirect(string $newURI, bool $permanent = true, bool $preserveMethod = true, bool $forceGET = false)

Function to allow redirects.

$newURI is the URL, that will be used in Location header. If it fails validation as URI, instead of redirect a 500 error will be sent to client.

$permanent governs whether this is a permanent redirect or a temporary one. Permanent redirects tell browser to always use the new address, while temporary - only this time. true by default.

$preserveMethod governs whether method is allowed to be changed when redirecting. Historically 301 and 302 redirects does not restrict client in this regard, which may cause some issues sometimes. If set to true (default) 307 and 308 codes will be used.

$forceGET is a flag to use 303 code. If you want client to specifically change method to GET when redirecting, you can use this one. Useful, if after POST or PUT you want to show output, that is handled by some other page.

notaccept

notAccept(array $supported = ['text/html'], bool $exit = true)

This is quite niche, but still may be useful in some cases. If you provide several formats of the data (for example, JSON and XML) and you want client to negotiate the format that the client will receive, you can use this function.

Client will send standard Accept HTTP header with list of acceptable MIME types, this function will check if any of the MIME types your backend provides is in the list and the use the one with the highest priority. That is, if client will accept both JSON and XML, but JSON will have priority 0.9 and XML - 0.8, function will return application/json, to let you know, that you should provide the data in JSON format.

If none of your supported MIME types match Accept header, 406 header will be returned to client. If header is not provided by client, function will return true. It will also return true, if client provides */* MIME type. This is why it will be niche: most browsers are sending it in their Accept headers. If you are using some custom API - this may be useful, still.

$supported - array of MIME types you support.

$exit if set to true will exit the script right after HTTP 406 is sent, otherwise will return false.

multiPartFormParse

multiPartFormParse()

This function parses multipart/form-data data for PUT, DELETE and PATCH methods and dumps the result as associative array to respective static variables $_PUT, $_DELETE and $_PATCH (accessed as \Simbiat\HTTP20\Headers::$_PUT).

The same logic can technically be used for POST as well, but PHP already parses it into native $_POST, so need to cover it.

Other HTTP verbs are not supposed to be accompanied by form data like this.