php-gettext#282 Added support for extra escaping sequences

jonasraoni · Jul 22, 2022 · 078246a · 078246a
1 parent ef12806
commit 078246a
Showing 1 changed file with 46 additions and 12 deletions.
diff --git a/src/Loader/StrictPoLoader.php b/src/Loader/StrictPoLoader.php
@@ -8,7 +8,7 @@
 use Gettext\Translations;
 
 /**
- * Class to load a PO file following the same rules of the GNU tools.
+ * Class to load a PO file following the same rules of the GNU gettext tools.
  */
 final class StrictPoLoader extends Loader
 {
@@ -139,7 +139,19 @@ private function readNumber(): string
     }
 
     /**
-     * Attempts to read a standard comment string which ends on \n
+     * Read at least one character from the given character set
+     */
+    private function readCharset(string $charset, int $maxLength): string
+    {
+        for ($data = ''; ($char = $this->getChar()) !== null && is_int(strpos($charset, $char)) && --$maxLength >= 0; $data .= $this->nextChar());
+        if ($data === '') {
+            throw new Exception("Expected at least one occurrence of the characters \"{$charset}\" at byte {$this->position}");
+        }
+        return $data;
+    }
+
+    /**
+     * Attempts to read a standard comment string which ends on \n or \r
      */
     private function readCommentString(): string
     {
@@ -163,7 +175,9 @@ private function readQuotedString(): string
             't' => "\t",
             'v' => "\x0b",
             '"' => '"',
-        ];
+        ],
+        $octalDigits = '01234567',
+        $hexDigits = $octalDigits . '89abcdefABCDEF';
         $hasData = false;
         for ($data = '';;) {
             if (!$this->readChar('"')) {
@@ -176,15 +190,35 @@ private function readQuotedString(): string
             // Collects chars until the end of the data/file
             for (; ($char = $this->getChar() ?? '"') !== '"'; $data .= $char) {
                 $this->nextChar();
-                if ($char === '\\') {
-                    // Ensures the next char is a valid escape character
-                    if (($char = $aliases[$this->nextChar()] ?? null) === null) {
-                        throw new Exception("Invalid quoted character at byte {$this->position}");
-                    }
-                    continue;
-                }
-                if ($char === "\n" || $char === "\r") {
-                    throw new Exception("New line character must be encoded at byte {$this->position}");
+                $octalSequence = ctype_digit($char) ? $char : '';
+                switch ($char) {
+                    case '\\':
+                        // Ensures the next char is a valid escape character
+                        if (($char = $aliases[$this->nextChar()] ?? null) === null) {
+                            throw new Exception("Invalid quoted character at byte {$this->position}");
+                        }
+                        break;
+                    case "\n":
+                    case "\r":
+                        throw new Exception("New line character must be encoded at byte {$this->position}");
+                    case 'U':
+                    case 'u':
+                        // The GNU gettext is supposed to follow the escaping sequences of C, but curiously it doesn't support the u/U
+                        $data = $this->readCharset($hexDigits, $char === 'u' ? 4 : 8);
+                        $data = str_pad($data, strlen($data) + (strlen($data) & 1), '0', STR_PAD_LEFT);
+                        $char = json_decode("\"\\u{$data}\"");
+                        break;
+                    case $octalSequence:
+                        $data = $this->readCharset($octalDigits, 3);
+                        // GNU gettext fails with octals above the signed char range
+                        if (($decimal = octdec($data)) > 127) {
+                            throw new Exception("Octal value out of range [0, 0177] at byte {$this->position}");
+                        }
+                        $char = chr($decimal);
+                    case 'x':
+                        $data = $this->readCharset($hexDigits, PHP_INT_MAX);
+                        // GNU reads all valid hexadecimal chars, but only uses the last pair
+                        $char = hexdec(substr($data, -2));
                 }
             }
             if (!$this->readChar('"')) {