nextcloud-memento/lib/Controller

<?php
namespace OCA\Memento\Controller;

use DOMDocument;
use DateTime;

use OCP\IRequest;
use OCP\IServerContainer;
use OCP\AppFramework\Controller;
use OCP\AppFramework\Http\RedirectResponse;
use OCP\AppFramework\Http\DataDisplayResponse;

class TimeGateController extends Controller {
	private $userFolder;

	public function __construct(
		$AppName,
		IRequest $request,
		$UserId,
		IServerContainer $serverContainer
	) {
		parent::__construct($AppName, $request);
		$this->userFolder = $serverContainer->getUserFolder($UserId);
	}

	/**
	 * @PublicPage
	 * @NoAdminRequired
	 * @NoCSRFRequired
	 */
	public function timeGate($url) {
		// Get all HTML files the user owns.
		$files = $this->userFolder->searchByMime('text/html');

		// Filter them for pages that have a <link rel="original"> referring to the given URL.
		$matchingFiles = array();
		foreach ($files as $file) {
			$content = $file->getContent();
			try {
				$DOM = new DOMDocument;
				$DOM->loadHTML($content);
				$head = $DOM->documentElement->getElementsByTagName('head')[0];
				$originals = getOriginals($head);
				foreach ($originals as $original) {
					if (normaliseUrl($original) === normaliseUrl($url)) {
						// Found a match!
						$datetime = getDatetime($head);
						$matchingFiles[] = [
							'file' => $file,
							'original' => $original,
							'datetime' => $datetime
						];
					}
				}
			} catch (Exception $e) {
				continue;
			}
		}

		// Choose one of the matched files, if any.
		if (count($matchingFiles) === 0) {
			// No matches. :(
			$message = "<h1>No snapshots found for requested URL. :(</h1>";
			return new DataDisplayResponse($message, 404);
		} else if (count($matchingFiles) === 1) {
			// One match; no need to choose.
			$chosenFile = $matchingFiles[0];
		} else {
			// Multiple matches: choose based on requested date.
			$acceptDatetimeHeader = $this->request->getHeader('Accept-Datetime');
			if ($acceptDatetimeHeader) {
				try {
					$requestedDatetime = DateTime::createFromFormat(DateTime::RFC1123, $acceptDatetimeHeader)
						->getTimestamp();
				} catch (Exception $e) {
					return new DataDisplayResponse("Invalid Accept-Datetime header.", 400);
				}
			} else {
				// Not sending the header means requesting the most recent version.
				$requestedDatetime = time();
			}
			// Pick the one closest to the requested date (either before or after it).
			$chosenFile = minBy($matchingFiles, function ($matchingFile) use ($requestedDatetime) {
				return abs($matchingFile['datetime'] - $requestedDatetime);
			});
		}

		// Send a 302 Found redirect pointing to the chosen file.
		$absoluteFilePath = $chosenFile['file']->getPath();
		$relativeFilePath = $this->userFolder->getRelativePath($absoluteFilePath);
		$fileUrl = joinPaths("/apps/raw/files", $relativeFilePath); // XXX hardcoded dependency
		$originalUrl = $chosenFile['original'];
		$response = new RedirectResponse($fileUrl);
		$response->setStatus(302);
		$response->addHeader('Vary', 'accept-datetime');
		$response->addHeader('Link', "<$originalUrl>; rel=\"original\"");
		return $response;
	}
}

function joinPaths($piece1, $piece2) {
	$left = rtrim($piece1, '/');
	$right = ltrim($piece2, '/');
	return "$left/$right";
}

// Reads hrefs from any <link> with relation type "original".
// (note the plural: we also accept pages that claim to correspond to multiple original URLs)
function getOriginals($head) {
	$originals = [];
	$links = $head->getElementsByTagName('link');
	foreach ($links as $link) {
		$rels = explode(' ', $link->getAttribute('rel'));
		if (in_array('original', $rels)) {
			$href = $link->getAttribute('href');
			$href = filter_var($href, FILTER_VALIDATE_URL, FILTER_FLAG_SCHEME_REQUIRED);
			if ($href) {
				$originals[] = $href;
			}
		}
	}
	return $originals;
}

// Read the content of the first <meta http-equiv="Memento-Datetime">, if any.
function getDatetime($head) {
	$metas = $head->getElementsByTagName('meta');
	foreach($metas as $meta) {
		// Let's match case-insensitively, I guess?
		if (strtolower($meta->getAttribute('http-equiv')) === 'memento-datetime') {
			$datetime = $meta->getAttribute('content');
			$datetime = DateTime::createFromFormat(DateTime::RFC1123, $datetime)->getTimestamp();
			return $datetime; // Return directly at the first match
		}
	}
	return null;
}

function normaliseUrl($url) {
	// Ignore trailing slashes. Because everybody does.
	$url = rtrim($url, '/');

	// Replace multiple slashes with a single one. Because Nextcloud will have already done this to
	// the queried url (e.g. 'http://abc' arrives to us as 'http:/abc')
	$url = preg_replace('%/{2,}%', '/', $url);

	return $url;
}

function minBy($array, $iteratee) {
	// is there any simpler way for this in php?
	$values = array_map($iteratee, $array);
	$argmin = array_search(min($values), $values);
	return $array[$argmin];
}