Merge pull request #28 from rbayliss/regex_precompile_perf

Performance: Precompile the ImmutableComponentTrait::encode() regex
thephpleague · Dec 30, 2015 · d7d440b · d7d440b
2 parents 3d6f573 + 9501113
commit d7d440b
Show file tree

Hide file tree

Showing 3 changed files with 23 additions and 23 deletions.
diff --git a/src/Components/HierarchicalPath.php b/src/Components/HierarchicalPath.php
@@ -87,18 +87,15 @@ public function __construct($path = '')
      */
     protected function validate($data)
     {
-        $data = array_values(array_filter(explode(static::$separator, $data), function ($segment) {
-            return !is_null($segment);
-        }));
-
-        $reserved = implode('', array_map(function ($char) {
-            return preg_quote($char, '/');
-        }, static::$characters_set));
-        $regex = '/(?:[^'.$reserved.']+|%(?![A-Fa-f0-9]{2}))/';
-
-        return array_map(function ($segment) use ($regex) {
-            return preg_replace_callback($regex, [$this, 'decodeSegmentPart'], $segment);
-        }, $data);
+        $regex = $this->getReservedRegex();
+
+        // Run the regex on the entire string rather than exploding.
+        // The separator should always be a reserved character.
+        $data = preg_replace_callback($regex, [$this, 'decodeSegmentPart'], $data);
+
+        return array_filter(explode(static::$separator, $data), function ($segment) {
+            return isset($segment);
+        });
     }
 
     /**

diff --git a/src/Components/Path.php b/src/Components/Path.php
@@ -66,12 +66,8 @@ protected function validate($path)
     {
         $this->assertValidComponent($path);
 
-        $reserved = implode('', array_map(function ($char) {
-            return preg_quote($char, '/');
-        }, static::$characters_set));
-
         return preg_replace_callback(
-            '/(?:[^'.$reserved.']+|%(?![A-Fa-f0-9]{2}))/',
+            $this->getReservedRegex(),
             [$this, 'decodeSegmentPart'],
             $path
         );

diff --git a/src/Types/ImmutableComponentTrait.php b/src/Types/ImmutableComponentTrait.php
@@ -35,6 +35,8 @@ trait ImmutableComponentTrait
         '!', '$', '&', "'", '(', ')', '*', '+', ',', ';', '=', ':',
     ];
 
+    protected static $characters_set_compiled;
+
     /**
      * Encoded characters to conform to RFC3986 - http://tools.ietf.org/html/rfc3986#section-2
      *
@@ -83,6 +85,15 @@ abstract public function getUriComponent();
      */
     abstract public function __toString();
 
+    protected static function getReservedRegex()
+    {
+        if (!isset(static::$characters_set_compiled)) {
+            $reserved = preg_quote(implode('', static::$characters_set), '/');
+            static::$characters_set_compiled = '/(?:[^'.$reserved.']+|%(?![A-Fa-f0-9]{2}))/S';
+        }
+        return static::$characters_set_compiled;
+    }
+
     /**
      * Encoding string according to RFC3986
      *
@@ -92,19 +103,15 @@ abstract public function __toString();
      */
     protected static function encode($value)
     {
-        $reservedChars = implode('', array_map(function ($value) {
-            return preg_quote($value, '/');
-        }, static::$characters_set));
-
         $str = preg_replace_callback(
-            '/(?:[^'.$reservedChars.']+|%(?![A-Fa-f0-9]{2}))/',
+            self::getReservedRegex(),
             function (array $matches) {
                 return rawurlencode($matches[0]);
             },
             $value
         );
 
-        return preg_replace_callback(',(?<encode>%[0-9A-F]{2}),i', function (array $matches) {
+        return preg_replace_callback(',(?<encode>%[0-9a-f]{2}),', function (array $matches) {
             return strtoupper($matches['encode']);
         }, $str);
     }