Work on indexer (#12253)

* Merge some unset() * Lots of work on the indexer itself: - Type safe comparisons - a few simplifications - a few performance optimizations * Some Elvis's and a missing @SInCE annotation * One more StringHelper... * Codestyle... * Not to potentially break BC, reversed back to `!empty($var)` operation but with the non inverted style `(bool) $var` * Formatting * Fix formatting * Revert this change for now. Look into it later again. * Codesniffer... * Removed repeated `use` * Some extra work according to @andrepereiradasilva's notes * Additional changes according to reviewer's comments * Code style * Fix misbehavior due to not correctly handling the single object case of mixed variable (array|object) * Pass float value to db->escape as float may produce problems for some locales. * "Make values float again" - (pun intended)
joomla · Sep 1, 2018 · 8ac2ff3 · 8ac2ff3
1 parent d0798c6
commit 8ac2ff3
Show file tree

Hide file tree

Showing 16 changed files with 178 additions and 158 deletions.
diff --git a/administrator/components/com_finder/helpers/indexer/adapter.php b/administrator/components/com_finder/helpers/indexer/adapter.php
@@ -160,7 +160,7 @@ public function __construct(&$subject, $config)
 	/**
 	 * Method to get the adapter state and push it into the indexer.
 	 *
-	 * @return  boolean  True on success.
+	 * @return  void
 	 *
 	 * @since   2.5
 	 * @throws  Exception on error.
@@ -312,7 +312,7 @@ abstract protected function index(FinderIndexerResult $item);
 	 *
 	 * @param   integer  $id  The ID of the item to reindex.
 	 *
-	 * @return  boolean  True on success.
+	 * @return  void
 	 *
 	 * @since   2.5
 	 * @throws  Exception on database error.
@@ -902,12 +902,9 @@ protected function pluginDisable($pks)
 	protected function translateState($item, $category = null)
 	{
 		// If category is present, factor in its states as well
-		if ($category !== null)
+		if ($category !== null && $category == 0)
 		{
-			if ($category == 0)
-			{
-				$item = 0;
-			}
+			$item = 0;
 		}
 
 		// Translate the state

diff --git a/administrator/components/com_finder/helpers/indexer/driver/mysql.php b/administrator/components/com_finder/helpers/indexer/driver/mysql.php
@@ -216,8 +216,7 @@ public function index($item, $format = 'html')
 						if ($group === static::PATH_CONTEXT)
 						{
 							$ip = JFile::stripExt($ip);
-							$ip = str_replace('/', ' ', $ip);
-							$ip = str_replace('-', ' ', $ip);
+							$ip = str_replace(array('/', '-'), ' ', $ip);
 						}
 
 						// Tokenize a string of content and add it to the database.

diff --git a/administrator/components/com_finder/helpers/indexer/driver/postgresql.php b/administrator/components/com_finder/helpers/indexer/driver/postgresql.php
@@ -58,7 +58,7 @@ public function index($item, $format = 'html')
 		$isNew = empty($link->link_id) ? true : false;
 
 		// Check the signatures. If they match, the item is up to date.
-		if (!$isNew && $curSig == $oldSig)
+		if (!$isNew && $curSig === $oldSig)
 		{
 			return $linkId;
 		}
@@ -208,8 +208,7 @@ public function index($item, $format = 'html')
 						if ($group === static::PATH_CONTEXT)
 						{
 							$ip = JFile::stripExt($ip);
-							$ip = str_replace('/', ' ', $ip);
-							$ip = str_replace('-', ' ', $ip);
+							$ip = str_replace(array('/', '-'), ' ', $ip);
 						}
 
 						// Tokenize a string of content and add it to the database.
@@ -328,7 +327,7 @@ public function index($item, $format = 'html')
 			' WHERE ta.term_id = 0'
 		);
 
-		if ($db->loadRow() == null)
+		if ($db->loadRow() === null)
 		{
 			$db->setQuery(
 				'INSERT INTO ' . $db->quoteName('#__finder_terms') .

diff --git a/administrator/components/com_finder/helpers/indexer/driver/sqlsrv.php b/administrator/components/com_finder/helpers/indexer/driver/sqlsrv.php
@@ -66,7 +66,7 @@ public function index($item, $format = 'html')
 		$isNew = empty($link->link_id) ? true : false;
 
 		// Check the signatures. If they match, the item is up to date.
-		if (!$isNew && $curSig == $oldSig)
+		if (!$isNew && $curSig === $oldSig)
 		{
 			return $linkId;
 		}
@@ -216,8 +216,7 @@ public function index($item, $format = 'html')
 						if ($group === static::PATH_CONTEXT)
 						{
 							$ip = JFile::stripExt($ip);
-							$ip = str_replace('/', ' ', $ip);
-							$ip = str_replace('-', ' ', $ip);
+							$ip = str_replace(array('/', '-'), ' ', $ip);
 						}
 
 						// Tokenize a string of content and add it to the database.

diff --git a/administrator/components/com_finder/helpers/indexer/helper.php b/administrator/components/com_finder/helpers/indexer/helper.php
@@ -64,7 +64,7 @@ public static function parse($input, $format = 'html')
 	 * @param   string   $lang    The language of the input.
 	 * @param   boolean  $phrase  Flag to indicate whether input could be a phrase. [optional]
 	 *
-	 * @return  array  An array of FinderIndexerToken objects.
+	 * @return  array|FinderIndexerToken  An array of FinderIndexerToken objects or a single FinderIndexerToken object.
 	 *
 	 * @since   2.5
 	 */
@@ -124,14 +124,14 @@ public static function tokenize($input, $lang, $phrase = false)
 			for ($i = 0, $n = count($terms); $i < $n; $i++)
 			{
 				$charMatches = array();
-				$charCount = preg_match_all('#[\p{Han}]#mui', $terms[$i], $charMatches);
+				$charCount   = preg_match_all('#[\p{Han}]#mui', $terms[$i], $charMatches);
 
 				// Split apart any groups of Chinese characters.
 				for ($j = 0; $j < $charCount; $j++)
 				{
 					$tSplit = StringHelper::str_ireplace($charMatches[0][$j], '', $terms[$i], false);
 
-					if (!empty($tSplit))
+					if ((bool) $tSplit)
 					{
 						$terms[$i] = $tSplit;
 					}
@@ -177,7 +177,12 @@ public static function tokenize($input, $lang, $phrase = false)
 				if ($i2 < $n && isset($tokens[$i2]))
 				{
 					// Tokenize the two word phrase.
-					$token = new FinderIndexerToken(array($tokens[$i]->term, $tokens[$i2]->term), $lang, $lang === 'zh' ? '' : ' ');
+					$token          = new FinderIndexerToken(
+						array(
+							$tokens[$i]->term,
+							$tokens[$i2]->term
+						), $lang, $lang === 'zh' ? '' : ' '
+					);
 					$token->derived = true;
 
 					// Add the token to the stack.
@@ -188,7 +193,13 @@ public static function tokenize($input, $lang, $phrase = false)
 				if ($i3 < $n && isset($tokens[$i3]))
 				{
 					// Tokenize the three word phrase.
-					$token = new FinderIndexerToken(array($tokens[$i]->term, $tokens[$i2]->term, $tokens[$i3]->term), $lang, $lang === 'zh' ? '' : ' ');
+					$token          = new FinderIndexerToken(
+						array(
+							$tokens[$i]->term,
+							$tokens[$i2]->term,
+							$tokens[$i3]->term
+						), $lang, $lang === 'zh' ? '' : ' '
+					);
 					$token->derived = true;
 
 					// Add the token to the stack.
@@ -265,7 +276,7 @@ public static function addContentType($title, $mime = null)
 	{
 		static $types;
 
-		$db = JFactory::getDbo();
+		$db    = JFactory::getDbo();
 		$query = $db->getQuery(true);
 
 		// Check if the types are loaded.
@@ -432,7 +443,7 @@ public static function getContentPath($url)
 		}
 
 		// Build the relative route.
-		$uri = $router->build($url);
+		$uri   = $router->build($url);
 		$route = $uri->toString(array('path', 'query', 'fragment'));
 		$route = str_replace(JUri::base(true) . '/', '', $route);
 
@@ -450,7 +461,7 @@ public static function getContentPath($url)
 	 * @since   2.5
 	 * @throws  Exception on database error.
 	 */
-	public static function getContentExtras(FinderIndexerResult &$item)
+	public static function getContentExtras(FinderIndexerResult $item)
 	{
 		// Get the event dispatcher.
 		$dispatcher = JEventDispatcher::getInstance();
@@ -504,7 +515,7 @@ public static function prepareContent($text, $params = null, FinderIndexerResult
 		}
 
 		// Create a mock content object.
-		$content = JTable::getInstance('Content');
+		$content       = JTable::getInstance('Content');
 		$content->text = $text;
 
 		if ($item)

diff --git a/administrator/components/com_finder/helpers/indexer/indexer.php b/administrator/components/com_finder/helpers/indexer/indexer.php
@@ -177,7 +177,7 @@ public static function getInstance()
 	public static function getState()
 	{
 		// First, try to load from the internal state.
-		if (!empty(static::$state))
+		if ((bool) static::$state)
 		{
 			return static::$state;
 		}
@@ -443,8 +443,7 @@ protected function tokenizeToDb($input, $context, $lang, $format)
 				// Parse, tokenise and add tokens to the database.
 				$count = $this->tokenizeToDbShort($string, $context, $lang, $format, $count);
 
-				unset($string);
-				unset($tokens);
+				unset($string, $tokens);
 			}
 
 			return $count;
@@ -513,36 +512,40 @@ protected function addTokensToDb($tokens, $context = '')
 
 		$query = clone $this->addTokensToDbQueryTemplate;
 
-		// Check if a single FinderIndexerToken object was given and make it to be an array of FinderIndexerToken objects
-		$tokens = is_array($tokens) ? $tokens : array($tokens);
-
 		// Count the number of token values.
 		$values = 0;
 
-		// Break into chunks of no more than 1000 items
-		$chunks = array_chunk($tokens, 1000);
-
-		foreach ($chunks as $tokens)
+		// Iterate through the tokens to create SQL value sets.
+		if (!is_a($tokens, 'FinderIndexerToken'))
 		{
-			$query->clear('values');
-
-			// Iterate through the tokens to create SQL value sets.
 			foreach ($tokens as $token)
 			{
 				$query->values(
 					$db->quote($token->term) . ', '
 					. $db->quote($token->stem) . ', '
 					. (int) $token->common . ', '
 					. (int) $token->phrase . ', '
-					. (float) $token->weight . ', '
+					. $db->escape((float) $token->weight) . ', '
 					. (int) $context . ', '
 					. $db->quote($token->language)
 				);
 				++$values;
 			}
-
-			$db->setQuery($query)->execute();
 		}
+		else
+		{
+			$query->values(
+				$db->quote($tokens->term) . ', '
+				. $db->quote($tokens->stem) . ', '
+				. (int) $tokens->common . ', '
+				. (int) $tokens->phrase . ', '
+				. $db->escape((float) $tokens->weight) . ', '
+				. (int) $context . ', '
+				. $db->quote($tokens->language)
+			);
+			++$values;
+		}
+		$db->setQuery($query)->execute();
 
 		return $values;
 	}

diff --git a/administrator/components/com_finder/helpers/indexer/parser/html.php b/administrator/components/com_finder/helpers/indexer/parser/html.php
@@ -98,6 +98,8 @@ protected function process($input)
 	 * @param   string  $endTag    String representing the end tag.
 	 *
 	 * @return  string with blocks removed.
+	 *
+	 * @since   3.4
 	 */
 	private function removeBlocks($input, $startTag, $endTag)
 	{

diff --git a/administrator/components/com_finder/helpers/indexer/parser/rtf.php b/administrator/components/com_finder/helpers/indexer/parser/rtf.php
@@ -30,12 +30,12 @@ class FinderIndexerParserRtf extends FinderIndexerParser
 	protected function process($input)
 	{
 		// Remove embedded pictures.
-		$input = preg_replace('#{\\\pict[^}]*}#mis', '', $input);
+		$input = preg_replace('#{\\\pict[^}]*}#mi', '', $input);
 
 		// Remove control characters.
 		$input = str_replace(array('{', '}', "\\\n"), array(' ', ' ', "\n"), $input);
-		$input = preg_replace('#\\\([^;]+?);#mis', ' ', $input);
-		$input = preg_replace('#\\\[\'a-zA-Z0-9]+#mis', ' ', $input);
+		$input = preg_replace('#\\\([^;]+?);#m', ' ', $input);
+		$input = preg_replace('#\\\[\'a-zA-Z0-9]+#mi', ' ', $input);
 
 		return $input;
 	}