Factor out snapshot search from timegate logic.

lib/Controller/TimeGateController.php

@@ -1,7 +1,8 @@
 <?php
 namespace OCA\Memento\Controller;
 
-use DOMDocument;
+require_once __DIR__ . '/findMementos.php';
+
 use DateTime;
 
 use OCP\IRequest;
@@ -10,6 +11,8 @@ use OCP\AppFramework\Controller;
 use OCP\AppFramework\Http\RedirectResponse;
 use OCP\AppFramework\Http\DataDisplayResponse;
 
+use findMementos;
+
 class TimeGateController extends Controller {
 	private $userFolder;
 
@@ -29,42 +32,16 @@ class TimeGateController extends Controller {
 	 * @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;
-			}
-		}
+		$matchingMementos = findMementos($this->userFolder, $url);
 
-		// Choose one of the matched files, if any.
-		if (count($matchingFiles) === 0) {
+		// Choose one of the matched mementos, if any.
+		if (count($matchingMementos) === 0) {
 			// No matches. :(
 			$message = "<h1>No snapshots found for requested URL. :(</h1>";
 			return new DataDisplayResponse($message, 404);
-		} else if (count($matchingFiles) === 1) {
+		} else if (count($matchingMementos) === 1) {
 			// One match; no need to choose.
-			$chosenFile = $matchingFiles[0];
+			$chosenMemento = $matchingMementos[0];
 		} else {
 			// Multiple matches: choose based on requested date.
 			$acceptDatetimeHeader = $this->request->getHeader('Accept-Datetime');
@@ -80,73 +57,22 @@ class TimeGateController extends Controller {
 				$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);
-			});
+			$chosenMemento = minBy($matchingMementos,
+				function ($matchingMemento) use ($requestedDatetime) {
+					return abs($matchingMemento['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);
+		// Send a 302 Found redirect pointing to the chosen memento.
+		$response = new RedirectResponse($chosenMemento['mementoUrl']);
 		$response->setStatus(302);
 		$response->addHeader('Vary', 'accept-datetime');
-		$response->addHeader('Link', "<$originalUrl>; rel=\"original\"");
+		$response->addHeader('Link', "<{$chosenMemento['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);

lib/Controller/findMementos.php

@@ -0,0 +1,93 @@
+<?php
+namespace OCA\Memento\Controller;
+
+use DOMDocument;
+use DateTime;
+
+// Finds HTML files that claim to be a snapshot of the given URL;
+// Returns an array, each item being a file's url + original url + snapshot datetime.
+function findMementos($folder, $url) {
+	// Get all HTML files the user owns.
+	$files = $folder->searchByMime('text/html');
+
+	// Filter them for pages that have a <link rel="original"> referring to the given URL.
+	$matchingMementos = array();
+	foreach ($files as $file) {
+		$content = $file->getContent();
+		try {
+			$DOM = new DOMDocument;
+			$DOM->loadHTML($content);
+			$headElement = $DOM->documentElement->getElementsByTagName('head')[0];
+			$originalUrls = getOriginalUrls($headElement);
+			foreach ($originalUrls as $originalUrl) {
+				if (normaliseUrl($originalUrl) === normaliseUrl($url)) {
+					// Found a match!
+					// Read its datetime
+					$datetime = getDatetime($headElement);
+					// Construct its URL.
+					$absoluteFilePath = $file->getPath();
+					$relativeFilePath = $folder->getRelativePath($absoluteFilePath);
+					$mementoUrl = joinPaths("/apps/raw/files", $relativeFilePath); // XXX hardcoded dependency
+
+					$matchingMementos[] = [
+						'mementoUrl' =>  $mementoUrl,
+						'originalUrl' => $originalUrl,
+						'datetime' => $datetime
+					];
+				}
+			}
+		} catch (Exception $e) {
+			continue;
+		}
+	}
+	return $matchingMementos;
+}
+
+function joinPaths($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 getOriginalUrls($headElement) {
+	$originalUrls = [];
+	$links = $headElement->getElementsByTagName('link');
+	foreach ($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) {
+				$originalUrls[] = $href;
+			}
+		}
+	}
+	return $originalUrls;
+}
+
+// Read the content of the first <meta http-equiv="Memento-Datetime">, if any.
+function getDatetime($headElement) {
+	$metas = $headElement->getElementsByTagName('meta');
+	foreach($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, '/');
+
+	// HACK. 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;
+}