From dbbdf4f230e6d76d6f7774397545e9275fd9ffbe Mon Sep 17 00:00:00 2001
From: Joshua Ramon Enslin <joshua@museum-digital.de>
Date: Thu, 1 Jul 2021 15:34:46 +0200
Subject: [PATCH] Add function to ensure an input string is UTF-8 encoded

---
 src/MD_STD_IN.php | 51 +++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 51 insertions(+)

diff --git a/src/MD_STD_IN.php b/src/MD_STD_IN.php
index 0bae377..b9578b6 100644
--- a/src/MD_STD_IN.php
+++ b/src/MD_STD_IN.php
@@ -266,4 +266,55 @@ final class MD_STD_IN {
         throw new MDgenericInvalidInputsException("ISBNs must be either 10 or 13 characters long.");
 
     }
+
+    /**
+     * Returns an UTF8 version of a string.
+     *
+     * @param string $input Input string.
+     *
+     * @return string
+     */
+    public static function ensureStringIsUtf8(string $input):string {
+
+        // If the input is valid UTF8 from the start, it is simply returned in its
+        // original form.
+        if (\mb_check_encoding($input, 'UTF-8')) {
+            return $input;
+        }
+
+        // To detect and convert the encoding for non-UTF8 strings, the list of
+        // encodings known to PHP's mbstring functions is checked against the input string.
+        // If any encoding matches the string, it will be converted to UTF8 accordingly.
+        $suitableEncodings = [];
+        $encodings = \mb_list_encodings();
+        foreach ($encodings as $encoding) {
+            if (\mb_detect_encoding($input, $encoding, true) !== false) {
+                $suitableEncodings[] = $encoding;
+            }
+        }
+
+        // If ISO-8859-1 is in the list of suitable encodings, try to convert with that.
+        if (\in_array('ISO-8859-1', $suitableEncodings, true)) {
+            if (($converted = \iconv('ISO-8859-1', "UTF-8//TRANSLIT", $input)) !== false) {
+                return $converted;
+            }
+        }
+
+        // If a conversion from ISO-8859-1 doesn't work, just take any of the other ones.
+        $suitableEncodings = \array_reverse($suitableEncodings);
+        foreach ($suitableEncodings as $encoding) {
+            if (($converted = \iconv($encoding, "UTF-8//TRANSLIT", $input)) !== false) {
+                return $converted;
+            }
+        }
+
+        /*
+        if (count($suitableEncodings) === 1) {
+            return mb_convert_encoding($input, 'UTF-8', );
+        }
+         */
+
+        return $input;
+
+    }
 }