Fix json_encode failure for src containing single byte chars with val…

…ues greater than 127
joernio · Apr 12, 2023 · 79424c9 · 79424c9
1 parent 0ffddce
commit 79424c9
Show file tree

Hide file tree

Showing 2 changed files with 19 additions and 2 deletions.
diff --git a/bin/php-parse b/bin/php-parse
@@ -8,6 +8,14 @@ foreach ([__DIR__ . '/../../../autoload.php', __DIR__ . '/../vendor/autoload.php
     }
 }
 
+spl_autoload_register(function ($class_name) {
+    $file = __DIR__ . '/../lib/' . $class_name . '.php';
+    $file = str_replace(array('/', '\\'), DIRECTORY_SEPARATOR, $file);
+    if (file_exists($file)) {
+        require $file;
+    }
+});
+
 ini_set('xdebug.max_nesting_level', 3000);
 
 // Disable Xdebug var_dump() output truncation
@@ -85,7 +93,7 @@ foreach ($files as $file) {
             echo $prettyPrinter->prettyPrintFile($stmts), "\n";
         } elseif ('json-dump' === $operation) {
             fwrite(STDERR, "==> JSON dump:\n");
-            echo json_encode($stmts, JSON_PRETTY_PRINT), "\n";
+            echo json_encode($stmts, JSON_PRETTY_PRINT | JSON_INVALID_UTF8_SUBSTITUTE), "\n";
         } elseif ('var-dump' === $operation) {
             fwrite(STDERR, "==> var_dump():\n");
             var_dump($stmts);

diff --git a/lib/PhpParser/Node/Scalar/String_.php b/lib/PhpParser/Node/Scalar/String_.php
@@ -116,7 +116,7 @@ function($matches) {
                 if (isset(self::$replacements[$str])) {
                     return self::$replacements[$str];
                 } elseif ('x' === $str[0] || 'X' === $str[0]) {
-                    return chr(hexdec(substr($str, 1)));
+                    return self::singleByteHexReplacement(substr($str, 1));
                 } elseif ('u' === $str[0]) {
                     return self::codePointToUtf8(hexdec($matches[2]));
                 } else {
@@ -127,6 +127,15 @@ function($matches) {
         );
     }
 
+    private static function singleByteHexReplacement(string $hexstr) : string {
+        $num = hexdec($hexstr);
+        if ($num <= 0x7F) {
+            return chr($num);
+        } else {
+            return "\x" . $hexstr;
+        }
+    }
+
     /**
      * Converts a Unicode code point to its UTF-8 encoded representation.
      *