resolve PREG_JIT_STACKLIMIT_ERROR error for larger queries

it could happen that large emulated queries fail because the regular expression that checked if a question mark is outside a string used a capture group that went though the rest of the query character by character. I could optimize it to read larger batches at once.
Nemo64 · Jan 6, 2021 · 802c500 · 802c500
1 parent 4d83370
commit 802c500
Show file tree

Hide file tree

Showing 2 changed files with 70 additions and 22 deletions.
diff --git a/src/RdsDataParameterBag.php b/src/RdsDataParameterBag.php
@@ -11,7 +11,7 @@ class RdsDataParameterBag
      * This expression can be used to find numeric parameters in an sql statement.
      * It understands strings and prevents matching within them.
      */
-    private const NUMERIC_PARAMETER_EXPRESSION = '/\?(?=([^\'"`]|\'([^\']|\\\\\')*\'|"([^"]|\\\\")*"|`([^`]|\\\\`)*`)*$)/';
+    private const NUMERIC_PARAMETER_EXPRESSION = '/\?(?=([^\'"`]+|\'([^\']|\\\\\')*\'|"([^"]|\\\\")*"|`([^`]|\\\\`)*`)*$)/';
 
     /**
      * @var array
@@ -89,20 +89,27 @@ public function getParameters(): array
     public function prepareSqlStatement(string $sql): string
     {
         $numericParameters = array_filter(array_keys($this->parameters), 'is_int');
+        if (count($numericParameters) <= 0) {
+            return $sql;
+        }
 
-        if (count($numericParameters) > 0) {
-
-            // it is valid to start numeric parameters 0 and 1
-            $index = min($numericParameters);
-            if ($index !== 0 && $index !== 1) {
-                throw new \LogicException("Numeric parameters must start with 0 or 1.");
-            }
-
-            $createParameter = function () use (&$index) {
-                return ':' . $index++;
-            };
+        // it is valid to start numeric parameters 0 and 1
+        $index = min($numericParameters);
+        if ($index !== 0 && $index !== 1) {
+            throw new \LogicException("Numeric parameters must start with 0 or 1.");
+        }
 
-            $sql = preg_replace_callback(self::NUMERIC_PARAMETER_EXPRESSION, $createParameter, $sql);
+        $createParameter = static function () use (&$index) {
+            return ':' . $index++;
+        };
+
+        $sql = preg_replace_callback(self::NUMERIC_PARAMETER_EXPRESSION, $createParameter, $sql);
+        if (!is_string($sql)) {
+            // snipped from https://www.php.net/manual/de/function.preg-last-error.php#124124
+            $pregError = array_flip(array_filter(get_defined_constants(true)['pcre'], function ($value) {
+                return substr($value, -6) === '_ERROR';
+            }, ARRAY_FILTER_USE_KEY))[preg_last_error()] ?? 'unknown error';
+            throw new \RuntimeException("sql param replacement failed: $pregError");
         }
 
         return $sql;

diff --git a/tests/RdsDataParameterBagTest.php b/tests/RdsDataParameterBagTest.php
@@ -11,15 +11,56 @@ class RdsDataParameterBagTest extends TestCase
     public static function sqlPreparation()
     {
         return [
-            [[0 => 1], 'SELECT * FROM x WHERE y = ?', 'SELECT * FROM x WHERE y = :0'],
-            [[1 => 1], 'SELECT * FROM x WHERE y = ?', 'SELECT * FROM x WHERE y = :1'],
-            [[1 => 1], "SELECT * FROM x WHERE x = '?' AND y = ?", "SELECT * FROM x WHERE x = '?' AND y = :1"],
-            [[1 => 1], "SELECT * FROM x WHERE x = ? AND y = '?'", "SELECT * FROM x WHERE x = :1 AND y = '?'"],
-            [[1 => 1], "SELECT * FROM x WHERE x = ? AND y = `?`", "SELECT * FROM x WHERE x = :1 AND y = `?`"],
-            [[1 => 1], 'SELECT * FROM x WHERE x = ? AND y = "?"', 'SELECT * FROM x WHERE x = :1 AND y = "?"'],
-            [[1 => 1], "SELECT * FROM x WHERE x = ? AND y = '\\'?'", "SELECT * FROM x WHERE x = :1 AND y = '\\'?'"],
-            [[1 => 1], "SELECT * FROM x WHERE x = '\\\\' AND y = ? AND z = '\\\\'", "SELECT * FROM x WHERE x = '\\\\' AND y = :1 AND z = '\\\\'"],
-            [['foo' => 1], 'SELECT * FROM x WHERE x = ? AND y = "?"', 'SELECT * FROM x WHERE x = ? AND y = "?"'],
+            [
+                [0 => 1],
+                'SELECT * FROM x WHERE y = ?',
+                'SELECT * FROM x WHERE y = :0'
+            ],
+            [
+                [1 => 1],
+                'SELECT * FROM x WHERE y = ?',
+                'SELECT * FROM x WHERE y = :1'
+            ],
+            [
+                [1 => 1],
+                "SELECT * FROM x WHERE x = '?' AND y = ?",
+                "SELECT * FROM x WHERE x = '?' AND y = :1"
+            ],
+            [
+                [1 => 1],
+                "SELECT * FROM x WHERE x = ? AND y = '?'",
+                "SELECT * FROM x WHERE x = :1 AND y = '?'"
+            ],
+            [
+                [1 => 1],
+                "SELECT * FROM x WHERE x = ? AND y = `?`",
+                "SELECT * FROM x WHERE x = :1 AND y = `?`"
+            ],
+            [
+                [1 => 1],
+                'SELECT * FROM x WHERE x = ? AND y = "?"',
+                'SELECT * FROM x WHERE x = :1 AND y = "?"'
+            ],
+            [
+                [1 => 1],
+                "SELECT * FROM x WHERE x = ? AND y = '\\'?'",
+                "SELECT * FROM x WHERE x = :1 AND y = '\\'?'"
+            ],
+            [
+                [1 => 1],
+                "SELECT * FROM x WHERE x = '\\\\' AND y = ? AND z = '\\\\'",
+                "SELECT * FROM x WHERE x = '\\\\' AND y = :1 AND z = '\\\\'"
+            ],
+            [
+                ['foo' => 1],
+                'SELECT * FROM x WHERE x = ? AND y = "?"',
+                'SELECT * FROM x WHERE x = ? AND y = "?"'
+            ],
+            [
+                [1 => 1],
+                'SELECT * FROM x WHERE x = ? AND y IN (' . implode(',', range(0, 1500)) . ')',
+                'SELECT * FROM x WHERE x = :1 AND y IN (' . implode(',', range(0, 1500)) . ')'
+            ],
         ];
     }