From be6b26745a9317b045b424f4d8ab33b8668bc85e Mon Sep 17 00:00:00 2001
From: Erik Hanson <erik@erikhanson.dev>
Date: Fri, 15 Nov 2024 10:01:37 -0800
Subject: [PATCH] pkp/jatsTemplate#59 PDFs not prioritized for full body text

---
 JatsTemplatePlugin.php | 24 +++++++++++++++++++-----
 1 file changed, 19 insertions(+), 5 deletions(-)

diff --git a/JatsTemplatePlugin.php b/JatsTemplatePlugin.php
index 33795fd..52b0608 100644
--- a/JatsTemplatePlugin.php
+++ b/JatsTemplatePlugin.php
@@ -20,6 +20,7 @@
 use HTMLPurifier_Config;
 use PKP\core\PKPString;
 use PKP\db\DAORegistry;
+use PKP\galley\Galley;
 use PKP\plugins\GenericPlugin;
 use PKP\plugins\Hook;
 use PKP\plugins\PluginRegistry;
@@ -325,11 +326,24 @@ function toXml(&$record, $format = null) {
 		$text = '';
 		$galleys = $article->getGalleys();
 
-		// Give precedence to HTML galleys, as they're quickest to parse
-		usort($galleys, function($a, $b) {
-			return $a->getFileType() == 'text/html'?-1:1;
-		});
-
+        // Get HTML galleys for top of list, as they're quickest to parse
+        // PDFs have second-highest priority over other file types
+        $items = array_reduce($galleys, function(array $carry, Galley $galley) {
+            $fileType = $galley->getFileType();
+
+            switch ($fileType) {
+                case 'text/html':
+                    $carry['html'][] = $galley;
+                    break;
+                case 'application/pdf':
+                    $carry['pdf'][] = $galley;
+                    break;
+                default:
+                    $carry['other'][] = $galley;
+            }
+            return $carry;
+        }, ['html' => [], 'pdf' => [], 'other' => []]);
+        $galleys = array_merge($items['html'], $items['pdf'], $items['other']);
 		// Provide the full-text.
 		$fileService = Services::get('file');
 		foreach ($galleys as $galley) {