Change token map lookup to avoid creating associative array

WordPress · May 1, 2024 · bb6be71 · bb6be71
1 parent e8b2cff
commit bb6be71
Show file tree

Hide file tree

Showing 2 changed files with 665 additions and 646 deletions.
diff --git a/src/wp-includes/class-wp-token-map.php b/src/wp-includes/class-wp-token-map.php
@@ -98,6 +98,8 @@ class WP_Token_Map {
 	 */
 	private $large_words = array();
 
+	private $groups = '';
+
 	/**
 	 * Stores an optimized row of short words, where every entry is
 	 * `$this->key_size + 1` bytes long and zero-extended.
@@ -195,16 +197,23 @@ public static function from_array( $mappings, $key_length = 2 ) {
 			$map->small_mappings[] = $mappings[ $word ];
 		}
 
-		foreach ( $groups as $group => $group_words ) {
+		$group_keys = array_keys( $groups );
+		sort( $group_keys );
+
+		foreach ( $group_keys as $group ) {
+			$map->groups .= "{$group}\x00";
+
 			$group_string = '';
 
-			foreach ( $group_words as $group_word ) {
+			foreach ( $groups[ $group ] as $group_word ) {
 				list( $word, $mapping ) = $group_word;
 
-				$group_string .= pack( 'C', strlen( $word ) ) . $word . pack( 'C', strlen( $mapping ) ) . $mapping;
+				$word_length    = pack( 'C', strlen( $word ) );
+				$mapping_length = pack( 'C', strlen( $mapping ) );
+				$group_string  .= "{$word_length}{$word}{$mapping_length}{$mapping}";
 			}
 
-			$map->large_words[ $group ] = $group_string;
+			$map->large_words[] = $group_string;
 		}
 
 		return $map;
@@ -226,10 +235,11 @@ public static function from_array( $mappings, $key_length = 2 ) {
 	 *
 	 * @return WP_Token_Map Map with precomputed data loaded.
 	 */
-	public static function from_precomputed_table( $key_length, $large_words, $small_words, $small_mappings ) {
+	public static function from_precomputed_table( $key_length, $groups, $large_words, $small_words, $small_mappings ) {
 		$map = new WP_Token_Map();
 
 		$map->key_length     = $key_length;
+		$map->groups         = $groups;
 		$map->large_words    = $large_words;
 		$map->small_words    = $small_words;
 		$map->small_mappings = $small_mappings;
@@ -261,11 +271,11 @@ public function contains( $word ) {
 		}
 
 		$group_key = substr( $word, 0, $this->key_length );
-		if ( ! isset( $this->large_words[ $group_key ] ) ) {
+		$group_at  = strpos( $this->groups, $group_key );
+		if ( false === $group_at ) {
 			return false;
 		}
-
-		$group        = $this->large_words[ $group_key ];
+		$group        = $this->large_words[ $group_at / ( $this->key_length + 1 ) ];
 		$group_length = strlen( $group );
 		$slug         = substr( $word, $this->key_length );
 		$length       = strlen( $slug );
@@ -336,7 +346,8 @@ public function read_token( $text, $offset = 0, &$skip_bytes = null ) {
 		if ( $text_length > $this->key_length ) {
 			$group_key = substr( $text, $offset, $this->key_length );
 
-			if ( ! isset( $this->large_words[ $group_key ] ) ) {
+			$group_at = strpos( $this->groups, $group_key );
+			if ( false === $group_at ) {
 				// Perhaps a short word then.
 				$small_text = str_pad( substr( $text, $offset, $this->key_length ), $this->key_length + 1, "\x00", STR_PAD_RIGHT );
 				$at         = strpos( $this->small_words, $small_text );
@@ -349,7 +360,7 @@ public function read_token( $text, $offset = 0, &$skip_bytes = null ) {
 				return $this->small_mappings[ $at / ( $this->key_length + 1 ) ];
 			}
 
-			$group        = $this->large_words[ $group_key ];
+			$group        = $this->large_words[ $group_at / ( $this->key_length + 1 ) ];
 			$group_length = strlen( $group );
 			$at           = 0;
 			while ( $at < $group_length ) {
@@ -408,7 +419,8 @@ public function to_array() {
 			$at += $this->key_length + 1;
 		}
 
-		foreach ( $this->large_words as $prefix => $group ) {
+		foreach ( $this->large_words as $index => $group ) {
+			$prefix       = substr( $this->groups, $index * ( $this->key_length + 1 ), 2 );
 			$group_length = strlen( $group );
 			$at           = 0;
 			while ( $at < $group_length ) {
@@ -457,15 +469,21 @@ public function precomputed_php_source_table( $indent = "\t" ) {
 
 		$output  = self::class . "::from_precomputed_table(\n";
 		$output .= "{$i1}{$this->key_length},\n";
+
+		$group_line = str_replace( "\x00", "\\x00", $this->groups );
+		$output    .= "{$i1}\"{$group_line}\",\n";
+
 		$output .= "{$i1}array(\n";
 
-		$prefixes = array_keys( $this->large_words );
-		sort( $prefixes );
-		foreach ( $prefixes as $prefix ) {
-			$group        = $this->large_words[ $prefix ];
+		$prefixes = explode( "\x00", $this->groups );
+		foreach ( $prefixes as $index => $prefix ) {
+			if ( '' === $prefix ) {
+				break;
+			}
+			$group        = $this->large_words[ $index ];
 			$group_length = strlen( $group );
 			$comment_line = "{$i2}//";
-			$data_line    = "{$i2}'{$prefix}' => \"";
+			$data_line    = "{$i2}\"";
 			$at           = 0;
 			while ( $at < $group_length ) {
 				$token_length   = unpack( 'C', $group[ $at++ ] )[1];