* @author Florin Patan * @copyright 2013 Jeremy Dorn * @license http://opensource.org/licenses/MIT * @link http://github.com/jdorn/sql-formatter * @version 1.2.10 */ class SqlFormatter { // Constants for token types const TOKEN_TYPE_WHITESPACE = 0; const TOKEN_TYPE_WORD = 1; const TOKEN_TYPE_QUOTE = 2; const TOKEN_TYPE_BACKTICK_QUOTE = 3; const TOKEN_TYPE_RESERVED = 4; const TOKEN_TYPE_RESERVED_TOPLEVEL = 5; const TOKEN_TYPE_RESERVED_NEWLINE = 6; const TOKEN_TYPE_BOUNDARY = 7; const TOKEN_TYPE_COMMENT = 8; const TOKEN_TYPE_BLOCK_COMMENT = 9; const TOKEN_TYPE_NUMBER = 10; const TOKEN_TYPE_ERROR = 11; // Constants for different components of a token const TOKEN_TYPE = 0; const TOKEN_VALUE = 1; // Reserved words (for syntax highlighting) protected static $reserved = array( 'ACCESSIBLE', 'ACTION', 'AGAINST', 'AGGREGATE', 'ALGORITHM', 'ALL', 'ALTER', 'ANALYSE', 'ANALYZE', 'AS', 'ASC', 'AUTOCOMMIT', 'AUTO_INCREMENT', 'BACKUP', 'BEGIN', 'BETWEEN', 'BINLOG', 'BOTH', 'CASCADE', 'CASE', 'CHANGE', 'CHANGED', 'CHARACTER SET', 'CHARSET', 'CHECK', 'CHECKSUM', 'COLLATE', 'COLLATION', 'COLUMN', 'COLUMNS', 'COMMENT', 'COMMIT', 'COMMITTED', 'COMPRESSED', 'CONCURRENT', 'CONSTRAINT', 'CONTAINS', 'CONVERT', 'COUNT', 'CREATE', 'CROSS', 'CURRENT_TIMESTAMP', 'DATABASE', 'DATABASES', 'DAY', 'DAY_HOUR', 'DAY_MINUTE', 'DAY_SECOND', 'DEFAULT', 'DEFINER', 'DELAYED', 'DELETE', 'DESC', 'DESCRIBE', 'DETERMINISTIC', 'DISTINCT', 'DISTINCTROW', 'DIV', 'DO', 'DUMPFILE', 'DUPLICATE', 'DYNAMIC', 'ELSE', 'ENCLOSED', 'END', 'ENGINE', 'ENGINE_TYPE', 'ENGINES', 'ESCAPE', 'ESCAPED', 'EVENTS', 'EXECUTE', 'EXISTS', 'EXPLAIN', 'EXTENDED', 'FAST', 'FIELDS', 'FILE', 'FIRST', 'FIXED', 'FLUSH', 'FOR', 'FORCE', 'FOREIGN', 'FULL', 'FULLTEXT', 'FUNCTION', 'GLOBAL', 'GRANT', 'GRANTS', 'GROUP_CONCAT', 'HEAP', 'HIGH_PRIORITY', 'HOSTS', 'HOUR', 'HOUR_MINUTE', 'HOUR_SECOND', 'IDENTIFIED', 'IF', 'IFNULL', 'IGNORE', 'IN', 'INDEX', 'INDEXES', 'INFILE', 'INSERT', 'INSERT_ID', 'INSERT_METHOD', 'INTERVAL', 'INTO', 'INVOKER', 'IS', 'ISOLATION', 'KEY', 'KEYS', 'KILL', 'LAST_INSERT_ID', 'LEADING', 'LEVEL', 'LIKE', 'LINEAR', 'LINES', 'LOAD', 'LOCAL', 'LOCK', 'LOCKS', 'LOGS', 'LOW_PRIORITY', 'MARIA', 'MASTER', 'MASTER_CONNECT_RETRY', 'MASTER_HOST', 'MASTER_LOG_FILE', 'MATCH', 'MEDIUM', 'MERGE', 'MINUTE', 'MINUTE_SECOND', 'MIN_ROWS', 'MODE', 'MODIFY', 'MONTH', 'MRG_MYISAM', 'MYISAM', 'NAMES', 'NATURAL', 'NOT', 'NOW()', 'NULL', 'OFFSET', 'ON', 'OPEN', 'OPTIMIZE', 'OPTION', 'OPTIONALLY', 'ON UPDATE', 'ON DELETE', 'OUTFILE', 'PACK_KEYS', 'PAGE', 'PARTIAL', 'PARTITION', 'PARTITIONS', 'PASSWORD', 'PRIMARY', 'PRIVILEGES', 'PROCEDURE', 'PROCESS', 'PROCESSLIST', 'PURGE', 'QUICK', 'RANGE', 'READ', 'READ_ONLY', 'READ_WRITE', 'REFERENCES', 'REGEXP', 'RELOAD', 'RENAME', 'REPAIR', 'REPEATABLE', 'REPLACE', 'REPLICATION', 'RESET', 'RESTORE', 'RESTRICT', 'RETURN', 'RETURNS', 'REVOKE', 'RLIKE', 'ROLLBACK', 'ROW', 'ROWS', 'ROW_FORMAT', 'SECOND', 'SECURITY', 'SEPARATOR', 'SERIALIZABLE', 'SESSION', 'SHARE', 'SHOW', 'SHUTDOWN', 'SLAVE', 'SONAME', 'SOUNDS', 'SQL', 'SQL_CACHE', 'SQL_NO_CACHE', 'START', 'STARTING', 'STATUS', 'STOP', 'STORAGE', 'STRAIGHT_JOIN', 'STRING', 'SUPER', 'TABLE', 'TABLES', 'TEMPORARY', 'TERMINATED', 'THEN', 'TO', 'TRAILING', 'TRANSACTIONAL', 'TRUNCATE', 'TYPE', 'TYPES', 'UNCOMMITTED', 'UNIQUE', 'UNLOCK', 'UNSIGNED', 'USAGE', 'USE', 'USING', 'VARIABLES', 'VIEW', 'WHEN', 'WITH', 'WORK', 'WRITE', 'YEAR_MONTH' ); // For SQL formatting // These keywords will all be on their own line protected static $reserved_toplevel = array( 'SELECT', 'FROM', 'WHERE', 'SET', 'ORDER BY', 'GROUP BY', 'LIMIT', 'DROP', 'VALUES', 'UPDATE', 'HAVING', 'ADD', 'AFTER', 'ALTER TABLE', 'DELETE FROM', 'UNION ALL', 'UNION', 'EXCEPT', 'INTERSECT' ); protected static $reserved_newline = array( 'LEFT JOIN', 'RIGHT JOIN', 'OUTER JOIN', 'INNER JOIN', 'JOIN', 'XOR', 'OR', 'AND' ); // Punctuation that can be used as a boundary between other tokens protected static $boundaries = array(',', ';', ')', '(', '.', '=', '<', '>', '+', '-', '*', '/', '!', '^', '%', '|', '&', '#'); // For HTML syntax highlighting // Styles applied to different token types public static $quote_attributes = 'style="color: blue;"'; public static $backtick_quote_attributes = 'style="color: purple;"'; public static $reserved_attributes = 'style="font-weight:bold;"'; public static $boundary_attributes = ''; public static $number_attributes = 'style="color: green;"'; public static $word_attributes = 'style="color: #333;"'; public static $error_attributes = 'style="background-color: red;"'; public static $comment_attributes = 'style="color: #aaa;"'; public static $pre_attributes = 'style="color: black; background-color: white;"'; // Boolean - whether or not the current environment is the CLI // This affects the type of syntax highlighting // If not defined, it will be determined automatically public static $cli; // For CLI syntax highlighting public static $cli_quote = "\x1b[34;1m"; public static $cli_backtick_quote = "\x1b[35;1m"; public static $cli_reserved = "\x1b[37m"; public static $cli_boundary = ""; public static $cli_number = "\x1b[32;1m"; public static $cli_word = ""; public static $cli_error = "\x1b[31;1;7m"; public static $cli_comment = "\x1b[30;1m"; // The tab character to use when formatting SQL public static $tab = ' '; // This flag tells us if queries need to be enclosed in
 tags
    public static $use_pre = true;
    
    // This flag tells us if SqlFormatted has been initialized
    protected static $init;
   
    // Regular expressions for tokenizing
    protected static $regex_boundaries;
    protected static $regex_reserved;
    protected static $regex_reserved_newline;
    protected static $regex_reserved_toplevel;

    // Cache variables
    // Only tokens shorter than this size will be cached.  Somewhere between 10 and 20 seems to work well for most cases.
    public static $max_cachekey_size = 15;
    protected static $token_cache = array();
    protected static $cache_hits = 0;
    protected static $cache_misses = 0;
   
    /**
     * Get stats about the token cache
     * @return Array An array containing the keys 'hits', 'misses', 'entries', and 'size' in bytes
     */
    public static function getCacheStats() {
        return array(
            'hits'=>self::$cache_hits,
            'misses'=>self::$cache_misses,
            'entries'=>count(self::$token_cache),
            'size'=>strlen(serialize(self::$token_cache))
        );
    }
   
    /**
     * Stuff that only needs to be done once.  Builds regular expressions and sorts the reserved words.
     */
    protected static function init() {
        if(self::$init) return;
       
        // Sort reserved word list from longest word to shortest
        usort(self::$reserved, array('SqlFormatter', 'sortLength'));

        // Set up regular expressions
        self::$regex_boundaries = '('.implode('|',array_map(array('SqlFormatter', 'quote_regex'),self::$boundaries)).')';
        self::$regex_reserved = '('.implode('|',array_map(array('SqlFormatter', 'quote_regex'),self::$reserved)).')';
        self::$regex_reserved_toplevel = str_replace(' ','\\s+','('.implode('|',array_map(array('SqlFormatter', 'quote_regex'),self::$reserved_toplevel)).')');
        self::$regex_reserved_newline = str_replace(' ','\\s+','('.implode('|',array_map(array('SqlFormatter', 'quote_regex'),self::$reserved_newline)).')');

        self::$init = true;
    }
   
    /**
     * Return the next token and token type in a SQL string.
     * Quoted strings, comments, reserved words, whitespace, and punctuation are all their own tokens.
     *
     * @param String $string The SQL string
     * @param array $previous The result of the previous getNextToken() call
     *
     * @return Array An associative array containing the type and value of the token.
     */
    protected static function getNextToken($string, $previous = null)
    {
        // Whitespace
        if (preg_match('/^\s+/',$string,$matches)) {
            return array(
                self::TOKEN_VALUE => $matches[0],
                self::TOKEN_TYPE=>self::TOKEN_TYPE_WHITESPACE
            );
        }
       
        // Comment
        if ($string[0] === '#' || (isset($string[1])&&($string[0]==='-'&&$string[1]==='-') || ($string[0]==='/'&&$string[1]==='*'))) {
            // Comment until end of line
            if ($string[0] === '-' || $string[0] === '#') {
                $last = strpos($string, "\n");
                $type = self::TOKEN_TYPE_COMMENT;
            } else { // Comment until closing comment tag
                $last = strpos($string, "*/", 2) + 2;
                $type = self::TOKEN_TYPE_BLOCK_COMMENT;
            }

            if ($last === false) {
                $last = strlen($string);
            }

            return array(
                self::TOKEN_VALUE => substr($string, 0, $last),
                self::TOKEN_TYPE  => $type
            );
        }

        // Quoted String
        if($string[0]==='"' || $string[0]==='\'' || $string[0]==='`') {
            $return = array(
                self::TOKEN_TYPE => ($string[0]==='`'? self::TOKEN_TYPE_BACKTICK_QUOTE : self::TOKEN_TYPE_QUOTE),
                self::TOKEN_VALUE => $string
            );
            
            // This checks for the following patterns:
            // 1. backtick quoted string using `` to escape
            // 2. double quoted string using "" or \" to escape
            // 3. single quoted string using '' or \' to escape
            if( preg_match('/^(((`[^`]*($|`))+)|(("[^"\\\\]*(?:\\\\.[^"\\\\]*)*("|$))+)|((\'[^\'\\\\]*(?:\\\\.[^\'\\\\]*)*(\'|$))+))/s', $string, $matches)) {
                $return[self::TOKEN_VALUE] = $matches[1];
            }
            
            return $return;
        }
       
        // Number
        if(preg_match('/^([0-9]+(\.[0-9]+)?)($|\s|"\'`|'.self::$regex_boundaries.')/',$string,$matches)) {
            return array(
                self::TOKEN_VALUE => $matches[1],
                self::TOKEN_TYPE=>self::TOKEN_TYPE_NUMBER
            );
        }

        // Boundary Character (punctuation and symbols)
        if(preg_match('/^('.self::$regex_boundaries.')/',$string,$matches)) {           
            return array(
                self::TOKEN_VALUE => $matches[1],
                self::TOKEN_TYPE  => self::TOKEN_TYPE_BOUNDARY
            );
        }

        // A reserved word cannot be preceded by a '.'
        // this makes it so in "mytable.from", "from" is not considered a reserved word
        if (!$previous || !isset($previous[self::TOKEN_VALUE]) || $previous[self::TOKEN_VALUE] !== '.') {
            $upper = strtoupper($string);
            // Top Level Reserved Word
            if(preg_match('/^('.self::$regex_reserved_toplevel.')($|\s|'.self::$regex_boundaries.')/', $upper,$matches)) {
                return array(
                    self::TOKEN_TYPE=>self::TOKEN_TYPE_RESERVED_TOPLEVEL,
                    self::TOKEN_VALUE=>substr($string,0,strlen($matches[1]))
                );
            }
            // Newline Reserved Word
            if(preg_match('/^('.self::$regex_reserved_newline.')($|\s|'.self::$regex_boundaries.')/', $upper,$matches)) {
                return array(
                    self::TOKEN_TYPE=>self::TOKEN_TYPE_RESERVED_NEWLINE,
                    self::TOKEN_VALUE=>substr($string,0,strlen($matches[1]))
                );
            }
            // Other Reserved Word
            if(preg_match('/^('.self::$regex_reserved.')($|\s|'.self::$regex_boundaries.')/', $upper,$matches)) {
                return array(
                    self::TOKEN_TYPE=>self::TOKEN_TYPE_RESERVED,
                    self::TOKEN_VALUE=>substr($string,0,strlen($matches[1]))
                );
            }
        }

        // Non reserved word
        preg_match('/^(.*?)($|\s|["\'`]|'.self::$regex_boundaries.')/',$string,$matches);

        return array(
            self::TOKEN_VALUE => $matches[1],
            self::TOKEN_TYPE  => self::TOKEN_TYPE_WORD
        );
    }

    /**
     * Takes a SQL string and breaks it into tokens.
     * Each token is an associative array with type and value.
     *
     * @param String $string The SQL string
     *
     * @return Array An array of tokens.
     */
    protected static function tokenize($string)
    {
        self::init();
       
        $tokens = array();

        // Used for debugging if there is an error while tokenizing the string
        $original_length = strlen($string);

        // Used to make sure the string keeps shrinking on each iteration
        $old_string_len = strlen($string) + 1;

        $token = null;
       
        $current_length = strlen($string);

        // Keep processing the string until it is empty
        while ($current_length) {
            // If the string stopped shrinking, there was a problem
            if ($old_string_len <= $current_length) {
                $tokens[] = array(
                    self::TOKEN_VALUE=>$string,
                    self::TOKEN_TYPE=>self::TOKEN_TYPE_ERROR
                );
                
                return $tokens;
            }
            $old_string_len =  $current_length;

            // Determine if we can use caching
            if($current_length >= self::$max_cachekey_size) {
                $cacheKey = substr($string,0,self::$max_cachekey_size);
            }
            else {
                $cacheKey = false;
            }

            // See if the token is already cached
            if($cacheKey && isset(self::$token_cache[$cacheKey])) {
                // Retrieve from cache
                $token = self::$token_cache[$cacheKey];
                $token_length = strlen($token[self::TOKEN_VALUE]);
                self::$cache_hits++;
            }
            else {
                // Get the next token and the token type
                $token = self::getNextToken($string, $token);               
                $token_length = strlen($token[self::TOKEN_VALUE]);
                self::$cache_misses++;
               
                // If the token is shorter than the max length, store it in cache
                if($cacheKey && $token_length < self::$max_cachekey_size) {
                    self::$token_cache[$cacheKey] = $token;
                }
            }
           
            $tokens[] = $token;

            // Advance the string
            $string = substr($string, $token_length);
           
            $current_length -= $token_length;
        }

        return $tokens;
    }
   
    /**
     * Format the whitespace in a SQL string to make it easier to read.
     *
     * @param String  $string    The SQL string
     * @param boolean $highlight If true, syntax highlighting will also be performed
     *
     * @return String The SQL string with HTML styles and formatting wrapped in a 
 tag
     */
    public static function format($string, $highlight=true) {
        // This variable will be populated with formatted html
        $return = '';

        // Use an actual tab while formatting and then switch out with self::$tab at the end
        $tab = "\t";

        $indent_level = 0;
        $newline = false;
        $inline_parentheses = false;
        $increase_special_indent = false;
        $increase_block_indent = false;
        $indent_types = array();
        $added_newline = false;
        $inline_count = 0;
        $inline_indented = false;
        
        // Tokenize String
        $original_tokens = self::tokenize($string);
        
        // Remove existing whitespace
        $tokens = array();
        foreach ($original_tokens as $i=>$token) {
            if ($token[self::TOKEN_TYPE] !== self::TOKEN_TYPE_WHITESPACE) {
                $token['i'] = $i;
                $tokens[] = $token;
            }
        }

        // Format token by token
        foreach ($tokens as $i=>$token) {            
            // Get highlighted token if doing syntax highlighting
            if ($highlight) {
                $highlighted = self::highlightToken($token);
            } else { // If returning raw text
                $highlighted = $token[self::TOKEN_VALUE];
            }

            // If we are increasing the special indent level now
            if($increase_special_indent) {
                $indent_level++;
                $increase_special_indent = false;
                array_unshift($indent_types,'special');
            }
            // If we are increasing the block indent level now
            if($increase_block_indent) {
                $indent_level++;
                $increase_block_indent = false;
                array_unshift($indent_types,'block');
            }
           
            // If we need a new line before the token
            if ($newline) {
                $return .= "\n" . str_repeat($tab, $indent_level);
                $newline = false;
                $added_newline = true;
            }
            else {
                $added_newline = false;
            }
            
            // Display comments directly where they appear in the source
            if ($token[self::TOKEN_TYPE] === self::TOKEN_TYPE_COMMENT || $token[self::TOKEN_TYPE] === self::TOKEN_TYPE_BLOCK_COMMENT) {
                if ($token[self::TOKEN_TYPE] === self::TOKEN_TYPE_BLOCK_COMMENT) {
                    $indent = str_repeat($tab,$indent_level);
                    $return .= "\n" . $indent;
                    $highlighted = str_replace("\n","\n".$indent,$highlighted);
                }

                $return .= $highlighted;
                $newline = true;
                continue;
            }
            
            if($inline_parentheses) {
                // End of inline parentheses
                if($token[self::TOKEN_VALUE] === ')') {
                    $return = rtrim($return,' ');
                    
                    if($inline_indented) {
                        array_shift($indent_types);
                        $indent_level --;
                        $return .= "\n" . str_repeat($tab, $indent_level);
                    }
                    
                    $inline_parentheses = false;
                    
                    $return .= $highlighted . ' ';
                    continue;
                }
                
                if($token[self::TOKEN_VALUE] === ',') {
                    if($inline_count >= 30) {
                        $inline_count = 0;
                        $newline = true;
                    }
                }
                
                $inline_count += strlen($token[self::TOKEN_VALUE]);
            }

            // Opening parentheses increase the block indent level and start a new line
            if ($token[self::TOKEN_VALUE] === '(') {
                // First check if this should be an inline parentheses block
                // Examples are "NOW()", "COUNT(*)", "int(10)", key(`somecolumn`), DECIMAL(7,2)
                // Allow up to 3 non-whitespace tokens inside inline parentheses
                $length = 0;
                for($j=1;$j<=250;$j++) {
                    // Reached end of string
                    if(!isset($tokens[$i+$j])) break;
                    
                    $next = $tokens[$i+$j];
                   
                    // Reached closing parentheses, able to inline it
                    if($next[self::TOKEN_VALUE] === ')') {
                        $inline_parentheses = true;
                        $inline_count = 0;
                        $inline_indented = false;
                        break;
                    }
                    
                    // Reached an invalid token for inline parentheses
                    if ($next[self::TOKEN_VALUE]===';' || $next[self::TOKEN_VALUE]==='(') {
                        break;
                    }
                    
                    // Reached an invalid token type for inline parentheses
                    if ($next[self::TOKEN_TYPE]===self::TOKEN_TYPE_RESERVED_TOPLEVEL || $next[self::TOKEN_TYPE]===self::TOKEN_TYPE_RESERVED_NEWLINE || $next[self::TOKEN_TYPE]===self::TOKEN_TYPE_COMMENT || $next[self::TOKEN_TYPE]===self::TOKEN_TYPE_BLOCK_COMMENT) {
                        break;
                    }
                    
                    $length += strlen($next[self::TOKEN_VALUE]);
                }
                
                if($inline_parentheses && $length > 30) {
                    $increase_block_indent = true;
                    $inline_indented = true;
                    $newline = true;
                }
               
                // Take out the preceding space unless there was whitespace there in the original query
                if (isset($original_tokens[$token['i']-1]) && $original_tokens[$token['i']-1][self::TOKEN_TYPE] !== self::TOKEN_TYPE_WHITESPACE) {
                    $return = rtrim($return,' ');
                }
               
                if(!$inline_parentheses) {
                    $increase_block_indent = true;
                    // Add a newline after the parentheses
                    $newline = true;
                }
                
            }
           
            // Closing parentheses decrease the block indent level
            elseif ($token[self::TOKEN_VALUE] === ')') {
                // Remove whitespace before the closing parentheses
                $return = rtrim($return,' ');
               
                $indent_level--;
       
                // Reset indent level
                while($j=array_shift($indent_types)) {
                    if($j==='special') {
                        $indent_level--;
                    }
                    else {
                        break;
                    }
                }
               
                if($indent_level < 0) {
                    // This is an error
                    $indent_level = 0;
                   
                    if ($highlight) {
                        $return .= "\n".self::highlightError($token[self::TOKEN_VALUE]);
                        continue;
                    }
                }
               
                // Add a newline before the closing parentheses (if not already added)
                if(!$added_newline) {
                    $return .= "\n" . str_repeat($tab, $indent_level);
                }               
            }
           
            // Commas start a new line (unless within inline parentheses)
            elseif ($token[self::TOKEN_VALUE] === ',' && !$inline_parentheses) {
                $newline = true;
            }
           
            // Top level reserved words start a new line and increase the special indent level
            elseif ($token[self::TOKEN_TYPE] === self::TOKEN_TYPE_RESERVED_TOPLEVEL) {
                $increase_special_indent = true;
               
                // If the last indent type was 'special', decrease the special indent for this round
                reset($indent_types);
                if(current($indent_types)==='special') {
                    $indent_level--;
                    array_shift($indent_types);
                }
               
                // Add a newline after the top level reserved word
                $newline = true;
                // Add a newline before the top level reserved word (if not already added)
                if(!$added_newline) {
                    $return .= "\n" . str_repeat($tab, $indent_level);
                }
                // If we already added a newline, redo the indentation since it may be different now
                else {
                    $return = rtrim($return,$tab).str_repeat($tab, $indent_level);
                }

                // If the token may have extra whitespace
                if (strpos($token[self::TOKEN_VALUE],' ')!==false || strpos($token[self::TOKEN_VALUE],"\n")!==false || strpos($token[self::TOKEN_VALUE],"\t")!==false) {
                    $highlighted = preg_replace('/\s+/',' ',$highlighted);
                }
            }

            // Newline reserved words start a new line
            elseif ($token[self::TOKEN_TYPE] === self::TOKEN_TYPE_RESERVED_NEWLINE) {
                // Add a newline before the reserved word (if not already added)
                if(!$added_newline) {
                    $return .= "\n" . str_repeat($tab, $indent_level);
                }

                // If the token may have extra whitespace
                if (strpos($token[self::TOKEN_VALUE],' ')!==false || strpos($token[self::TOKEN_VALUE],"\n")!==false || strpos($token[self::TOKEN_VALUE],"\t")!==false) {
                    $highlighted = preg_replace('/\s+/',' ',$highlighted);
                }
            }
            
            // Multiple boundary characters in a row should not have spaces between them (not including parentheses)
            elseif($token[self::TOKEN_TYPE] === self::TOKEN_TYPE_BOUNDARY) {
                if($tokens[$i-1][self::TOKEN_TYPE] === self::TOKEN_TYPE_BOUNDARY) {
                    if (isset($original_tokens[$token['i']-1]) && $original_tokens[$token['i']-1][self::TOKEN_TYPE] !== self::TOKEN_TYPE_WHITESPACE) {
                        $return = rtrim($return,' ');
                    }
                }
            }

            // If the token shouldn't have a space before it
            if ($token[self::TOKEN_VALUE] === '.' || $token[self::TOKEN_VALUE] === ',' || $token[self::TOKEN_VALUE] === ';') {
                $return = rtrim($return, ' ');
            }
           
            $return .= $highlighted.' ';

            // If the token shouldn't have a space after it
            if ($token[self::TOKEN_VALUE] === '(' || $token[self::TOKEN_VALUE] === '.') {
                $return = rtrim($return,' ');
            }
        }

        // If there are unmatched parentheses
        if ($highlight && array_search('block',$indent_types) !== false) {
            $return .= "\n".self::highlightError("WARNING: unclosed parentheses or section");
        }
       
        // Replace tab characters with the configuration tab character
        $return = trim(str_replace("\t",self::$tab,$return));

        if ($highlight) {
            $return = self::output($return);
        }

        return $return;       
    }

    /**
     * Add syntax highlighting to a SQL string
     *
     * @param String $string The SQL string
     *
     * @return String The SQL string with HTML styles applied
     */
    public static function highlight($string)
    {
        $tokens = self::tokenize($string);

        $return = '';

        foreach ($tokens as $token) {
            $return .= self::highlightToken($token);
        }

        return self::output($return);
    }

    /**
     * Split a SQL string into multiple queries.
     * Uses ";" as a query delimiter.
     *
     * @param String $string The SQL string
     *
     * @return Array An array of individual query strings without trailing semicolons
     */
    public static function splitQuery($string)
    {
        $queries = array();
        $current_query = '';
        $empty = true;

        $tokens = self::tokenize($string);
        
        foreach ($tokens as $token) {            
            // If this is a query separator
            if ($token[self::TOKEN_VALUE] === ';') {
                if (!$empty) {
                    $queries[] = $current_query.';';
                }
                $current_query = '';
                $empty = true;
                continue;
            }
            
            // If this is a non-empty character
            if($token[self::TOKEN_TYPE] !== self::TOKEN_TYPE_WHITESPACE && $token[self::TOKEN_TYPE] !== self::TOKEN_TYPE_COMMENT && $token[self::TOKEN_TYPE] !== self::TOKEN_TYPE_BLOCK_COMMENT) {
                $empty = false;
            }

            $current_query .= $token[self::TOKEN_VALUE];
        }

        if (!$empty) {
            $queries[] = trim($current_query);
        }

        return $queries;
    }

    /**
     * Remove all comments from a SQL string
     *
     * @param String $string The SQL string
     *
     * @return String The SQL string without comments
     */
    public static function removeComments($string)
    {
        $result = '';

        $tokens = self::tokenize($string);

        foreach ($tokens as $token) {
            // Skip comment tokens
            if ($token[self::TOKEN_TYPE] === self::TOKEN_TYPE_COMMENT || $token[self::TOKEN_TYPE] === self::TOKEN_TYPE_BLOCK_COMMENT) {
                continue;
            }

            $result .= $token[self::TOKEN_VALUE];
        }

        return $result;
    }

    /**
     * Compress a query by collapsing white space and removing comments
     *
     * @param String $string The SQL string
     *
     * @return String The SQL string without comments
     */
    public static function compress($string)
    {
        $result = '';

        $tokens = self::tokenize($string);


        $whitespace = true;
        foreach ($tokens as $token) {
            // Skip comment tokens
            if ($token[self::TOKEN_TYPE] === self::TOKEN_TYPE_COMMENT || $token[self::TOKEN_TYPE] === self::TOKEN_TYPE_BLOCK_COMMENT) {
                continue;
            }
            
            if($token[self::TOKEN_TYPE] === self::TOKEN_TYPE_WHITESPACE) {
                // If the last token was whitespace, don't add another one
                if($whitespace) {
                    continue;
                }
                else {
                    $whitespace = true;
                    // Convert all whitespace to a single space
                    $token[self::TOKEN_VALUE] = ' ';
                }
            }
            else {
                $whitespace = false;
            }

            $result .= $token[self::TOKEN_VALUE];
        }

        return rtrim($result);
    }

    /**
     * Highlights a token depending on its type.
     *
     * @param Array $token An associative array containing type and value.
     *
     * @return String HTML code of the highlighted token.
     */
    protected static function highlightToken($token)
    {
        $type = $token[self::TOKEN_TYPE];
        
        if(self::is_cli()) {
            $token = $token[self::TOKEN_VALUE];
        }
        else {
            $token = htmlentities($token[self::TOKEN_VALUE],ENT_COMPAT,'UTF-8');
        }

        if($type===self::TOKEN_TYPE_BOUNDARY) {
            return self::highlightBoundary($token);
        }
        elseif($type===self::TOKEN_TYPE_WORD) {
            return self::highlightWord($token);
        }
        elseif($type===self::TOKEN_TYPE_BACKTICK_QUOTE) {
            return self::highlightBacktickQuote($token);
        }
        elseif($type===self::TOKEN_TYPE_QUOTE) {
            return self::highlightQuote($token);
        }
        elseif($type===self::TOKEN_TYPE_RESERVED) {
            return self::highlightReservedWord($token);
        }
        elseif($type===self::TOKEN_TYPE_RESERVED_TOPLEVEL) {
            return self::highlightReservedWord($token);
        }
        elseif($type===self::TOKEN_TYPE_RESERVED_NEWLINE) {
            return self::highlightReservedWord($token);
        }
        elseif($type===self::TOKEN_TYPE_NUMBER) {
            return self::highlightNumber($token);
        }
        elseif($type===self::TOKEN_TYPE_COMMENT || $type===self::TOKEN_TYPE_BLOCK_COMMENT) {
            return self::highlightComment($token);
        }
        
        return $token;
    }

    /**
     * Highlights a quoted string
     *
     * @param String $value The token's value
     *
     * @return String HTML code of the highlighted token.
     */
    protected static function highlightQuote($value)
    {
        if(self::is_cli()) {
            return self::$cli_quote . $value . "\x1b[0m";
        }
        else {
            return '' . $value . '';
        }
    }

    /**
     * Highlights a backtick quoted string
     *
     * @param String $value The token's value
     *
     * @return String HTML code of the highlighted token.
     */
    protected static function highlightBacktickQuote($value) {
        if(self::is_cli()) {
            return self::$cli_backtick_quote . $value . "\x1b[0m";
        }
        else {
            return '' . $value . '';
        }
    }

    /**
     * Highlights a reserved word
     *
     * @param String $value The token's value
     *
     * @return String HTML code of the highlighted token.
     */
    protected static function highlightReservedWord($value)
    {
        if(self::is_cli()) {
            return self::$cli_reserved . $value . "\x1b[0m";
        }
        else {
            return '' . $value . '';
        }
    }

    /**
     * Highlights a boundary token
     *
     * @param String $value The token's value
     *
     * @return String HTML code of the highlighted token.
     */
    protected static function highlightBoundary($value)
    {
        if($value==='(' || $value===')') return $value;
        
        if(self::is_cli()) {
            return self::$cli_boundary . $value . "\x1b[0m";
        }
        else {
            return '' . $value . '';
        }
    }

    /**
     * Highlights a number
     *
     * @param String $value The token's value
     *
     * @return String HTML code of the highlighted token.
     */
    protected static function highlightNumber($value)
    {
        if(self::is_cli()) {
            return self::$cli_number . $value . "\x1b[0m";
        }
        else {
            return '' . $value . '';
        }
    }

    /**
     * Highlights an error
     *
     * @param String $value The token's value
     *
     * @return String HTML code of the highlighted token.
     */
    protected static function highlightError($value)
    {
        if(self::is_cli()) {
            return self::$cli_error . $value . "\x1b[0m";
        }
        else {
            return '' . $value . '';
        }
    }

    /**
     * Highlights a comment
     *
     * @param String $value The token's value
     *
     * @return String HTML code of the highlighted token.
     */
    protected static function highlightComment($value)
    {
        if(self::is_cli()) {
            return self::$cli_comment . $value . "\x1b[0m";
        }
        else {
            return '' . $value . '';
        }
    }

    /**
     * Highlights a word token
     *
     * @param String $value The token's value
     *
     * @return String HTML code of the highlighted token.
     */
    protected static function highlightWord($value)
    {
        if(self::is_cli()) {
            return self::$cli_word . $value . "\x1b[0m";
        }
        else {
            return '' . $value . '';
        }
    }

    /**
     * Helper function for sorting the list of reserved words by length
     *
     * @param String $a The first string
     * @param String $b The second string
     *
     * @return int The comparison of the string lengths
     */
    private static function sortLength($a, $b)
    {
        return strlen($b) - strlen($a);
    }
   
    /**
     * Helper function for building regular expressions for reserved words and boundary characters
     *
     * @param String $a The string to be quoted
     *
     * @return String The quoted string
     */
    private static function quote_regex($a)
    {
        return preg_quote($a,'/');
    }
    
    /**
     * Helper function for building string output
     *
     * @param String $string The string to be quoted
     *
     * @return String The quoted string
     */
    private static function output($string)
    {
        if(self::is_cli()) {
            return $string."\n";
        }
        else {
            $string=trim($string);
            if(!self::$use_pre) {
                return $string;
            }
            return '
' . $string . '
'; } } private static function is_cli() { if(isset(self::$cli)) return self::$cli; else return php_sapi_name() === 'cli'; } }