All links in cached page appended with ?stage=Live in SS3.7.3

[TomDownard]

Any links in pages generated through the Sitetree Link() function are being appended with "?stage=Live" in the cached files. This is an issue for SEO reasons, as the same content is being served on two urls.

For example, when looping through dataobjects in a template:

<% loop $Apartments %>
	<a href="$Link">$Title</a>
<% end_loop %>

This would result in a URL like: /apartmtent_title?stage=Live being output.

I have tried overriding the Link function in Page.php:

public function Link($action = null) {
	$relativeLink = $this->RelativeLink($action);
	$link = Controller::join_links(Director::baseURL(), $relativeLink);
	$this->extend('updateLink', $link, $action, $relativeLink);
	if(strpos($link, '?stage=Live') !== false){
		$link = str_replace("stage=Live&", "", $link);
		$link = str_replace("stage=Live", "", $link);
	}
	return $link;
}

However this results in staged content being included in the cache in loops such as above.

Has anybody come across this issue or a solution for it?

[mediabeastnz]

+1 also having this issue

[mediabeastnz]

In the file URLArrayObject.php and function sanitize_url() there are a few $_GET variables unset. Should stage not be added to this e.g.

unset($getVars['_ID'], $getVars['_ClassName'], $getVars['stage']);

Applying this patch seems to resolve the issue for me but doesn't really touch on why the stage is in the URL in the first place.

[NightJar]

This is probably a hang over from the alteration to not store global state between requests - it was causing confusion with editors viewing draft content without realising, as there was no indication.

Remember the old days when ?Stage=draft would set a session var, then every request regardless of GET param would always be draft? This might be caused by the job being queued from within the CMS, or is a hangover from ensuring that the live site is cached.

This is all out loud thoughts, in case anyone wants to take a look and maybe submit a PR for HacktoberFest ;)
If so, please see https://docs.silverstripe.org/en/4/contributing/code/ about the SilverStripe process about branching and making PRs against the correct branches :)

I say this because even though the module is marked as supported, this is the SilverStripe 3 compatible version - which is in security support only.
Alternatively if you can reproduce this issue with SilverStripe 4, this is more likely to be prioritised to be looked at.

[emteknetnz]

Related #100

[quamsta]

If anyone is still having an issue with this, I had to implement a hacky fix on an old site we needed to statically generate due to some unexpectedly high traffic to it. In FilesystemPublisher.php, around line 305, I added some code to strip out ?stage=Live in the to-be-saved .html cached file after the if-statment:

"if (Config::inst()->get('FilesystemPublisher', 'include_caching_metadata')) {":

$content = str_replace( '?stage=Live', '',$content );

[dhensby]

I'm not sure if this is fixed, but I'm afraid v3 is out of support now