From 54ecc337276f1854b209e4a258a198e5374091cb Mon Sep 17 00:00:00 2001
From: Christos Sidiropoulos <csidirop@runbox.com>
Date: Thu, 14 Dec 2023 10:50:39 +0000
Subject: [PATCH] Refactor server URL generation

Signed-off-by: Christos Sidiropoulos <csidirop@runbox.com>
---
 Classes/Controller/PageViewController.php | 28 ++++++++++++++++-------
 Classes/Plugin/FullTextGenerator.php      |  4 ++--
 Classes/Plugin/FullTextXMLtools.php       |  2 +-
 3 files changed, 23 insertions(+), 11 deletions(-)

diff --git a/Classes/Controller/PageViewController.php b/Classes/Controller/PageViewController.php
index 17d44d66b..26fa4f1d0 100644
--- a/Classes/Controller/PageViewController.php
+++ b/Classes/Controller/PageViewController.php
@@ -176,14 +176,7 @@ protected function getFulltext($page)
         //check if local fulltext exists:
         if (PageViewController::getOCRengine(Doc::$extKey) != "originalremote") {
             if (FullTextGenerator::checkLocal(Doc::$extKey, $this->document, $page)) { //fulltext is locally present
-                //check server protocol (https://stackoverflow.com/a/14270161):
-                if ( isset($_SERVER['HTTPS'])  &&  ($_SERVER['HTTPS'] == 'on' || $_SERVER['HTTPS'] == 1)
-                    ||  isset($_SERVER['HTTP_X_FORWARDED_PROTO'])  &&  $_SERVER['HTTP_X_FORWARDED_PROTO'] == 'https') {
-                    $protocol = 'https://';
-                } else {
-                    $protocol = 'http://';
-                }
-                $fulltext['url'] = $protocol . $_SERVER['HTTP_HOST'] . "/" . FullTextGenerator::getPageLocalPath(Doc::$extKey, $this->document, $page);
+                $fulltext['url'] = PageViewController::getServerUrl() . "/" . FullTextGenerator::getPageLocalPath(Doc::$extKey, $this->document, $page);
                 $fulltext['mimetype'] = "text/xml";
             }
         }
@@ -435,6 +428,25 @@ protected function generateFullText():void {
         FullTextGenerator::createPageFullText(Doc::$extKey, $this->document, $this->getImage($this->requestData['page'])["url"], $this->requestData['page'], $engine);
     }
 
+    /**
+     * Returns the server URL (including networkprotocol: http or https)
+     * eg. https://www.example.com
+     * 
+     * @access public
+     * 
+     * @return string The server URL
+     */
+    public static function getServerUrl():string {
+        //check server protocol (parts from https://stackoverflow.com/a/14270161):
+        if (GeneralUtility::getIndpEnv('TYPO3_SSL') == true
+            || isset($_SERVER['HTTPS'])  &&  ($_SERVER['HTTPS'] == 'on'  ||  $_SERVER['HTTPS'] == 1)
+            || isset($_SERVER['HTTP_X_FORWARDED_PROTO'])  &&  $_SERVER['HTTP_X_FORWARDED_PROTO'] == 'https') {
+            return 'https://' . $_SERVER['HTTP_HOST'];
+        } else {
+            return 'http://'. $_SERVER['HTTP_HOST'];
+        }
+    }
+
     /**
      * This function is a workaround to circumvent TYPO3s disturbing caching.
      * It clears the stored page cache (for presentations viewer only!) on calling.
diff --git a/Classes/Plugin/FullTextGenerator.php b/Classes/Plugin/FullTextGenerator.php
index 11d7f27eb..6b676bfcc 100644
--- a/Classes/Plugin/FullTextGenerator.php
+++ b/Classes/Plugin/FullTextGenerator.php
@@ -254,10 +254,10 @@ protected static function generatePageOCR(string $extKey, array $conf, Document
     //Determine if the image should be downloaded. Than use remote URL ($imageUrl) or local PATH ($tmpImagePath):
     if ($conf['ocrDwnlTempImage']){ //download image
       $imageDownloadCommand = "wget $imageUrl -O $tmpImagePath"; //wget image and save to $tmpImagePath
-      $ocrShellCommand .= self::genShellCommand($ocrEnginePath, $tmpImagePath, $tmpOutputPath, $outputPath, $tmpImagePath, $pageId, $conf['ocrPlaceholderText'], "http://".$_SERVER['HTTP_HOST']."/".$outputPath, $conf['ocrUpdateMets'], $conf['ocrIndexMets']);
+      $ocrShellCommand .= self::genShellCommand($ocrEnginePath, $tmpImagePath, $tmpOutputPath, $outputPath, $tmpImagePath, $pageId, $conf['ocrPlaceholderText'], PageViewController::getServerUrl()."/".$outputPath, $conf['ocrUpdateMets'], $conf['ocrIndexMets']);
       $ocrShellCommand .= " && rm $tmpImagePath";  // Remove used image
     } else { //do not download image, pass URL to the engine
-      $ocrShellCommand .= self::genShellCommand($ocrEnginePath, $imageUrl, $tmpOutputPath, $outputPath, $tmpImagePath, $pageId, $pageNum, $conf['ocrPlaceholderText'], "http://".$_SERVER['HTTP_HOST']."/".$outputPath, $conf['ocrUpdateMets'], $conf['ocrIndexMets']);
+      $ocrShellCommand .= self::genShellCommand($ocrEnginePath, $imageUrl, $tmpOutputPath, $outputPath, $tmpImagePath, $pageId, $pageNum, $conf['ocrPlaceholderText'], PageViewController::getServerUrl()."/".$outputPath, $conf['ocrUpdateMets'], $conf['ocrIndexMets']);
     }
 
     /* DEBUG */ if($conf['ocrDebug']) echo '<script>alert("'.$ocrShellCommand.'")</script>'; //DEBUG
diff --git a/Classes/Plugin/FullTextXMLtools.php b/Classes/Plugin/FullTextXMLtools.php
index 44c0e0238..d7e78909c 100644
--- a/Classes/Plugin/FullTextXMLtools.php
+++ b/Classes/Plugin/FullTextXMLtools.php
@@ -179,7 +179,7 @@ protected static function updateMetsNode(XMLWriter $writer, string $alto_path, s
       $writer->writeAttribute('SOFTWARE', "DFG-Viewer-5-OCR-$ocr_script");
       $writer->startElement('mets:FLocat'); // <mets:FLocat LOCTYPE="URL" xlink:href="https://digi.bib.uni-mannheim.de/fileadmin/digi/log59088/alto/log59088_431.xml"/>
         $writer->writeAttribute('LOCTYPE', 'URL');
-        $writer->writeAttribute('xlink:href', "http://".$_SERVER['HTTP_HOST']."/".$alto_path);
+        $writer->writeAttribute('xlink:href', PageViewController::getServerUrl()."/".$alto_path);
       $writer->endElement();
     $writer->endElement();
   }