From 46394865a1d4b801dd0925537f4fd19fd132d45e Mon Sep 17 00:00:00 2001
From: Kurt Thiemann <kurt@aternos.org>
Date: Tue, 29 Nov 2022 18:50:00 +0100
Subject: [PATCH] added support for decoding Java's weird modified UTF-8
 encoding

---
 README.md                                |  12 ++
 composer.json                            |   3 +-
 src/String/JavaEncoding.php              | 139 +++++++++++++++++++++++
 src/String/StringDataFormatException.php |   8 ++
 src/Tag/StringTag.php                    |  23 ++++
 5 files changed, 184 insertions(+), 1 deletion(-)
 create mode 100644 src/String/JavaEncoding.php
 create mode 100644 src/String/StringDataFormatException.php

diff --git a/README.md b/README.md
index 2c11a51..d603bde 100644
--- a/README.md
+++ b/README.md
@@ -49,6 +49,18 @@ $myInt->setValue(42);
 echo $myInt->getValue(); // 42
 ```
 
+On String tags, `getValue()` and `setValue()` will use the raw string data, which uses Java's modified
+UTF-8 encoding. To use different encodings,
+use `getDecodedValue($encoding = "UTF-8")` and `setDecodedValue($value, $encoding = "UTF-8")` instead.
+A list of supported encodings is returned by the `mb_list_encodings()` function.
+
+```php
+$myString new \Aternos\Nbt\Tag\StringTag();
+
+$myString->setDecodedValue("Hello world!");
+echo $myString->getDecodedValue(); // Hello world!
+```
+
 Compound tags, list tags, and array tags implement the `ArrayAccess`, `Countable`,
 and `Iterator` interfaces and can therefore be accessed as arrays.
 ```php
diff --git a/composer.json b/composer.json
index e3ebc55..d81e0fe 100644
--- a/composer.json
+++ b/composer.json
@@ -19,6 +19,7 @@
     "php": ">=8.0",
     "php-64bit": "*",
     "ext-zlib": "*",
-    "ext-json": "*"
+    "ext-json": "*",
+    "ext-mbstring": "*"
   }
 }
diff --git a/src/String/JavaEncoding.php b/src/String/JavaEncoding.php
new file mode 100644
index 0000000..6fcaa5a
--- /dev/null
+++ b/src/String/JavaEncoding.php
@@ -0,0 +1,139 @@
+<?php
+
+namespace Aternos\Nbt\String;
+
+/**
+ * https://py2jdbc.readthedocs.io/en/latest/mutf8.html
+ * https://docs.oracle.com/javase/8/docs/api/java/io/DataInput.html#modified-utf-8
+ * https://docs.oracle.com/javase/6/docs/api/java/io/DataInput.html#readUTF%28%29
+ * Good luck
+ */
+class JavaEncoding
+{
+    protected static ?JavaEncoding $instance = null;
+
+    /**
+     * @return static
+     */
+    static function getInstance(): static
+    {
+        if (static::$instance === null) {
+            static::$instance = new static();
+        }
+        return static::$instance;
+    }
+
+    /**
+     * @param string $string
+     * @param string $sourceEncoding
+     * @return string
+     */
+    public function encode(string $string, string $sourceEncoding = "UTF-8"): string
+    {
+        $result = "";
+
+        $chars = mb_str_split($string, 1, $sourceEncoding);
+        foreach ($chars as $char) {
+            $c = mb_ord($char, $sourceEncoding);
+
+            if($c === 0) {
+                $result .= "\xC0\x80";
+                continue;
+            }
+
+            if($c <= 0x7F) {
+                $result .= chr($c);
+                continue;
+            }
+
+            if($c <= 0x7FF) {
+                $result .= chr(0xC0 | (0x1F & ($c >> 0x06)));
+                $result .= chr(0x80 | (0x3F & $c));
+                continue;
+            }
+
+            if($c <= 0xFFFF) {
+                $result .= chr(0xE0 | (0x0F & ($c >> 0x0C)));
+                $result .= chr(0x80 | (0x3F & ($c >> 0x06)));
+                $result .= chr(0x80 | (0x3F & $c));
+                continue;
+            }
+
+            $result .= chr(0xED);
+            $result .= chr(0xA0 | (($c >> 0x10) & 0x0F));
+            $result .= chr(0x80 | (($c >> 0x0A) & 0x3f));
+            $result .= chr(0xED);
+            $result .= chr(0xb0 | (($c >> 0x06) & 0x0f));
+            $result .= chr(0x80 | ($c & 0x3f));
+        }
+
+        return $result;
+    }
+
+    /**
+     * @throws StringDataFormatException
+     */
+    public function decode(string $string, string $outputEncoding = "UTF-8"): string
+    {
+        $result = "";
+        for ($i = 0; $i < strlen($string); $i++) {
+            $a = ord($string[$i]);
+
+            if ($a === 0) {
+                throw new StringDataFormatException("Invalid NULL byte in string");
+            }
+
+            // Single byte character
+            if (($a & 0b10000000) === 0b0) {
+                $result .= mb_chr($a, $outputEncoding);
+                continue;
+            }
+
+            $b = ord($string[++$i] ?? "\0");
+
+            // Two byte character
+            if (($a & 0b11100000) === 0b11000000) {
+                if (($b & 0b11000000) !== 0b10000000) {
+                    throw new StringDataFormatException("Invalid \"UTF-8\" sequence");
+                }
+
+                $result .= mb_chr((($a & 0x1F) << 6) | ($b & 0x3F), $outputEncoding);
+                continue;
+            }
+
+            $c = ord($string[++$i] ?? "\0");
+
+            // Maybe six byte character
+            if ($a === 0b11101101 && ($b & 0b11110000) === 0b10100000 && ($c & 0b11000000) === 0b10000000) {
+                $d = ord($string[$i + 1] ?? "\0");
+                $e = ord($string[$i + 2] ?? "\0");
+                $f = ord($string[$i + 3] ?? "\0");
+
+                // Six byte character
+                if ($d === 0b11101101 && ($e & 0b11110000) === 0b10110000 && ($f & 0b11000000) === 0b10000000) {
+                    $result .= mb_chr(0x10000 |
+                        ($b & 0x0F) << 0x10 |
+                        ($c & 0x3F) << 0x0A |
+                        ($e & 0x0F) << 0x06 |
+                        ($f & 0x3F), $outputEncoding);
+
+                    $i += 3;
+                    continue;
+                }
+            }
+
+            // Three byte character
+            if (($a & 0b11110000) === 0b11100000) {
+                if (($b & 0b11000000) !== 0b10000000 || ($c & 0b11000000) !== 0b10000000) {
+                    throw new StringDataFormatException("Invalid \"UTF-8\" sequence");
+                }
+
+                $result .= mb_chr((($a & 0x0F) << 12) | (($b & 0x3F) << 6) | ($c & 0x3F), $outputEncoding);
+                continue;
+            }
+
+            throw new StringDataFormatException("Invalid \"UTF-8\" sequence");
+        }
+        return $result;
+    }
+}
\ No newline at end of file
diff --git a/src/String/StringDataFormatException.php b/src/String/StringDataFormatException.php
new file mode 100644
index 0000000..65e779e
--- /dev/null
+++ b/src/String/StringDataFormatException.php
@@ -0,0 +1,8 @@
+<?php
+
+namespace Aternos\Nbt\String;
+
+class StringDataFormatException extends \Exception
+{
+
+}
\ No newline at end of file
diff --git a/src/Tag/StringTag.php b/src/Tag/StringTag.php
index 7f08dea..10bc686 100644
--- a/src/Tag/StringTag.php
+++ b/src/Tag/StringTag.php
@@ -4,6 +4,8 @@
 
 use Aternos\Nbt\IO\Reader\Reader;
 use Aternos\Nbt\IO\Writer\Writer;
+use Aternos\Nbt\String\JavaEncoding;
+use Aternos\Nbt\String\StringDataFormatException;
 use Exception;
 
 class StringTag extends Tag
@@ -20,6 +22,16 @@ public function getValue(): string
         return $this->value;
     }
 
+    /**
+     * @param string $encoding
+     * @return string
+     * @throws StringDataFormatException
+     */
+    public function getDecodedValue(string $encoding = "UTF-8"): string
+    {
+        return JavaEncoding::getInstance()->decode($this->value, $encoding);
+    }
+
     /**
      * @param string $value
      * @return StringTag
@@ -30,6 +42,17 @@ public function setValue(string $value): StringTag
         return $this;
     }
 
+    /**
+     * @param string $value
+     * @param string $encoding
+     * @return StringTag
+     */
+    public function setDecodedValue(string $value, string $encoding = "UTF-8"): StringTag
+    {
+        $this->value = JavaEncoding::getInstance()->encode($value, $encoding);
+        return $this;
+    }
+
     /**
      * @return int
      */