feat: Async and queued page archival

4 files changed, 142 insertions, 4 deletions

diff --git a/controllers/archive.php b/controllers/archive.php
index 367acc7..4e4ac77 100644
--- a/controllers/archive.php
+++ b/controllers/archive.php
@@ -3,24 +3,76 @@ namespace Controller;
 use Database;
 use DOMDocument;
 use Exception;
+use TypeError;
+use ValueError;
 
 function on_post() {
+    if (!array_key_exists('async', $_POST) || $_POST['async'] !== 'true') {
+        return;
+    }
+
+    session_start();
+
     global $TOKEN;
 
     $WEBSITE_CATEGORY = 'url';
     $DOWNLOADS_FOLDER = getenv('ARCHIVES_DIR');
     $website_url = $_POST[$WEBSITE_CATEGORY];
     $uid = 1;
+    $authorized = false;
     if ($TOKEN !== "") {
         try {
-            $uid = Database\Cookie::fromDB($TOKEN)->UID;
+            $user = Database\Cookie::fromDB($TOKEN);
+            $uid = $user->UID;
+            $authorized = $user->Role === 'Admin';
         }
         catch (Exception $e) {}
     }
-    $currentPage = new DownloadPage($website_url, $DOWNLOADS_FOLDER, $uid);
 
-    header('Location: /archive/?url=' . $website_url);
-    exit();
+    $manual_start = $authorized
+                    && array_key_exists('manual', $_POST)
+                    && $_POST['manual'] === 'true';
+    // The first request to archive a page becomes a "worker", which will archive
+    // the requested page and any other which might be requested in the meantime
+    $start_worker = !array_key_exists('archive_queue', $_SESSION)
+                    || count($_SESSION['archive_queue']) === 0
+                    || $manual_start;
+
+    if (!array_key_exists('archive_queue', $_SESSION)) {
+        $_SESSION['archive_queue'] = array();
+        $_SESSION['archive_current'] = 0;
+    }
+    else if ($start_worker) {
+        $_SESSION['archive_current'] = 0;
+    }
+
+    $current = $_SESSION['archive_current'] + count($_SESSION['archive_queue']);
+    if (!$manual_start) {
+        array_push(
+            $_SESSION['archive_queue'],
+            new DownloadInfo($website_url, $DOWNLOADS_FOLDER, $uid)
+        );
+    }
+
+    if ($start_worker) {
+        while (count($_SESSION['archive_queue']) > 0) {
+            $downloadInfo = $_SESSION['archive_queue'][0];
+            session_write_close();
+
+            try {
+                $downloadInfo->download();
+            }
+            catch(Exception $e) { }
+            catch(TypeError $e) { }
+            catch(ValueError $e) { }
+
+            session_start();
+            array_shift($_SESSION['archive_queue']);
+            $_SESSION['archive_current']++;
+        }
+    }
+    echo $current;
+    exit;
 }
 
 function on_delete() {
@@ -57,6 +109,22 @@ function on_delete() {
     exit();
 }
 
+class DownloadInfo {
+    public $page_url;
+    private $folder_location;
+    private $requester_uid;
+
+    function __construct(string $page_url, string $folder_location, string $requester_uid) {
+        $this->page_url = $page_url;
+        $this->folder_location = $folder_location;
+        $this->requester_uid = $requester_uid;
+    }
+
+    function download() : DownloadPage {
+        return new DownloadPage($this->page_url, $this->folder_location, $this->requester_uid);
+    }
+}
+
 class DownloadPage {
     private $folder_location;
     private $folder_name;
diff --git a/views/admin/index.php b/views/admin/index.php
index 750a246..e633d32 100644
--- a/views/admin/index.php
+++ b/views/admin/index.php
@@ -35,6 +35,24 @@
         <input type="submit" value="Delete">
     </form>
 
+    <h2>Archive queue</h2>
+
+    <button id="manual-start">Start worker manually</button>
+    <script type="text/javascript">
+        document.getElementById('manual-start').onclick = function() {
+            var request = new XMLHttpRequest();
+            request.onreadystatechange = function() {
+                if (request.readyState < 4) return;
+
+                console.log(request.responseText);
+            }
+            request.open("POST", "/archive/create", true);
+            request.setRequestHeader('Content-type', 'application/x-www-form-urlencoded');
+            request.withCredentials = true;
+            request.send('async=true&url=localhost&manual=true');
+        }
+    </script>
+
 <?php else: ?>
     <h2>Permission denied, you're not an admin!</h2>
 
diff --git a/views/archive/create/index.php b/views/archive/create/index.php
index 378d2de..b435e4d 100644
--- a/views/archive/create/index.php
+++ b/views/archive/create/index.php
@@ -1 +1,48 @@
 <h2>Archiving <?= $url ?>...</h2>
+
+<p>Current position in queue: <span id="position">unavailable</span></p>
+
+<script type="text/javascript">
+// If our request is the one that starts the worker, it's output will be 0
+// and it won't update the variable soon (or ever)
+// But if it's not the worker, the variable will be updated (quickly)
+var queuePos = 0;
+
+function requestDownload(url) {
+    var request = new XMLHttpRequest();
+    request.onreadystatechange = function() {
+        if (request.readyState < 4) return;
+
+        console.log(request.responseText);
+        queuePos = request.responseText;
+    }
+    request.open("POST", "#", true);
+    request.setRequestHeader('Content-type', 'application/x-www-form-urlencoded');
+    request.withCredentials = true;
+    request.send('async=true&url=' +
+            url +
+            '<?php if (array_key_exists("manual", $_POST)) { echo "&manual=" . $_POST["manual"]; } ?>');
+}
+requestDownload('<?= $url ?>');
+
+const position = document.getElementById('position');
+function updatePosition(url) {
+    var request = new XMLHttpRequest();
+    request.onreadystatechange = function() {
+        if (request.readyState < 4) return;
+
+        if (queuePos < request.responseText) {
+            window.location.href = '/archive/?url=' + url;
+            return;
+        }
+
+        position.innerText = queuePos - request.responseText;
+        setTimeout(updatePosition, 1000, url);
+    }
+    request.open("POST", "/archive/create/status.php", true);
+    request.setRequestHeader('Content-type', 'application/x-www-form-urlencoded');
+    request.send('url=' + url);
+}
+
+updatePosition('<?= $url ?>');
+</script>
diff --git a/views/archive/create/status.php b/views/archive/create/status.php
new file mode 100644
index 0000000..d020c4a
--- /dev/null
+++ b/views/archive/create/status.php
@@ -0,0 +1,5 @@
+<?php
+
+session_start();
+
+echo $_SESSION['archive_current'];