From 7c4f019f4ebf97a51b84837c61cba4094f06ea65 Mon Sep 17 00:00:00 2001
From: Isabel <isabel-andrea.uffinger@stud.hs-hannover.de>
Date: Sat, 4 May 2024 13:14:47 +0200
Subject: [PATCH] checking concrete encodings and added error mesage
---
externallib.php | 3 +++
locallib.php | 13 ++++++++-----
2 files changed, 11 insertions(+), 5 deletions(-)
diff --git a/externallib.php b/externallib.php
index e6dd854..0a69499 100644
--- a/externallib.php
+++ b/externallib.php
@@ -195,6 +195,9 @@ class qtype_moopt_external extends external_api {
$encoding = $child->attributes->getNamedItem('encoding')->nodeValue;
$filecontent = get_text_content_from_file($usercontext, $draftid, $keepfilename,
$pathinfo['dirname'] . '/', $pathinfo['basename'], true, $encoding);
+ if($filecontent === null){
+ $returnval['error'] = "Das Encoding der Datei ".$pathinfo['basename']." konnte nicht erkannt werden";
+ }
$defaultfilename = basename($child->nodeValue);
$fileid = $file->attributes->getNamedItem('id')->nodeValue;
$enablefreetext = true;
diff --git a/locallib.php b/locallib.php
index 1d2201b..3750080 100644
--- a/locallib.php
+++ b/locallib.php
@@ -294,18 +294,21 @@ function get_text_content_from_file($usercontext, $draftareaid, $keepfilename, $
// TODO: make sure the mimetype is plain text
// even task.xmls may contain mistakes (eg PDF )
- //check if encoding of attached file is utf-8 else convert
+ //check if encoding of attached file is utf-8 else convert if encoding could be detectet
$content = $file->get_content();
+ $test_encodings = array('UTF-8', 'ISO-8859-1', 'ISO-8859-15', 'Windows-1252', 'UTF-16');
if($attached){
if($encoding!=null){
$enc=$encoding;
} else {
- $enc = mb_detect_encoding($content, null, true);
- if($enc===false){
- throw new invalid_parameter_exception('Encoding of attached file ' . $filepath . $filename . ' could\'nt be detectet.');
+ $enc = mb_detect_encoding($content, null, false);
+ if ($enc==false){
+ $enc = mb_detect_encoding($content, $test_encodings, true);
}
}
- if($enc!=='UTF-8'){
+ if($enc==false){
+ return null;
+ } else if($enc!=='UTF-8'){
$content = mb_convert_encoding($content, 'UTF-8', $enc);
}
}
--
GitLab