summaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
authorJeremy Dorn <jeremy@jeremydorn.com>2013-07-10 23:30:32 -0700
committerJeremy Dorn <jeremy@jeremydorn.com>2013-07-10 23:30:32 -0700
commit7dc6794a53571d5ccbe10716d20e2e3a7101c00c (patch)
treece446f32d83fdbd0d21ed9b3b6c890ca02d91d04
parenta8546539505e6dc388303ee89dcde7c84c338dd4 (diff)
parentf0bbbe193836e760c1baa581578abe19e48f2d54 (diff)
downloadsql-formatter-7dc6794a53571d5ccbe10716d20e2e3a7101c00c.zip
sql-formatter-7dc6794a53571d5ccbe10716d20e2e3a7101c00c.tar.gz
sql-formatter-7dc6794a53571d5ccbe10716d20e2e3a7101c00c.tar.bz2
Merge branch 'master' of https://github.com/adityagoyal/sql-formatter into adityagoyal-master
-rw-r--r--lib/SqlFormatter.php85
-rw-r--r--tests/SqlFormatterTest.php2
2 files changed, 73 insertions, 14 deletions
diff --git a/lib/SqlFormatter.php b/lib/SqlFormatter.php
index 0e5e781..5908003 100644
--- a/lib/SqlFormatter.php
+++ b/lib/SqlFormatter.php
@@ -26,6 +26,7 @@ class SqlFormatter
const TOKEN_TYPE_BLOCK_COMMENT = 9;
const TOKEN_TYPE_NUMBER = 10;
const TOKEN_TYPE_ERROR = 11;
+ const TOKEN_TYPE_FUNCTION = 12;
// Constants for different components of a token
const TOKEN_TYPE = 0;
@@ -36,7 +37,7 @@ class SqlFormatter
'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',
+ 'CONSTRAINT', 'CONTAINS', 'CONVERT', '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',
@@ -46,7 +47,7 @@ class SqlFormatter
'LINES', 'LOAD', 'LOCAL', 'LOCK', 'LOCKS', 'LOGS', 'LOW_PRIORITY', 'MARIA', 'MASTER', 'MASTER_CONNECT_RETRY', 'MASTER_HOST', 'MASTER_LOG_FILE',
'MATCH','MAX_CONNECTIONS_PER_HOUR', 'MAX_QUERIES_PER_HOUR', 'MAX_ROWS', 'MAX_UPDATES_PER_HOUR', 'MAX_USER_CONNECTIONS',
'MEDIUM', 'MERGE', 'MINUTE', 'MINUTE_SECOND', 'MIN_ROWS', 'MODE', 'MODIFY',
- 'MONTH', 'MRG_MYISAM', 'MYISAM', 'NAMES', 'NATURAL', 'NOT', 'NOW()', 'NULL', 'OFFSET', 'ON', 'OPEN', 'OPTIMIZE', 'OPTION', 'OPTIONALLY',
+ 'MONTH', 'MRG_MYISAM', 'MYISAM', 'NAMES', 'NATURAL', 'NOT', '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', 'RAID0', 'RAID_CHUNKS', 'RAID_CHUNKSIZE','RAID_TYPE', 'READ', 'READ_ONLY',
'READ_WRITE', 'REFERENCES', 'REGEXP', 'RELOAD', 'RENAME', 'REPAIR', 'REPEATABLE', 'REPLACE', 'REPLICATION', 'RESET', 'RESTORE', 'RESTRICT',
@@ -71,6 +72,35 @@ class SqlFormatter
'LEFT OUTER JOIN', 'RIGHT OUTER JOIN', 'LEFT JOIN', 'RIGHT JOIN', 'OUTER JOIN', 'INNER JOIN', 'JOIN', 'XOR', 'OR', 'AND'
);
+ protected static $mysql_function = array (
+ 'ABS', 'ACOS', 'ADDDATE', 'ADDTIME', 'AES_DECRYPT', 'AES_ENCRYPT', 'AREA', 'ASBINARY', 'ASCII', 'ASIN', 'ASTEXT', 'ATAN', 'ATAN2',
+ 'AVG', 'BDMPOLYFROMTEXT', 'BDMPOLYFROMWKB', 'BDPOLYFROMTEXT', 'BDPOLYFROMWKB', 'BENCHMARK', 'BIN', 'BIT_AND', 'BIT_COUNT', 'BIT_LENGTH',
+ 'BIT_OR', 'BIT_XOR', 'BOUNDARY', 'BUFFER', 'CAST', 'CEIL', 'CEILING', 'CENTROID', 'CHAR', 'CHARACTER_LENGTH', 'CHARSET', 'CHAR_LENGTH',
+ 'COALESCE', 'COERCIBILITY', 'COLLATION', 'COMPRESS', 'CONCAT', 'CONCAT_WS', 'CONNECTION_ID', 'CONTAINS', 'CONV', 'CONVERT', 'CONVERT_TZ',
+ 'CONVEXHULL', 'COS', 'COT', 'COUNT', 'CRC32', 'CROSSES', 'CURDATE', 'CURRENT_DATE', 'CURRENT_TIME', 'CURRENT_TIMESTAMP', 'CURRENT_USER',
+ 'CURTIME', 'DATABASE', 'DATE', 'DATEDIFF', 'DATE_ADD', 'DATE_DIFF', 'DATE_FORMAT', 'DATE_SUB', 'DAY', 'DAYNAME', 'DAYOFMONTH', 'DAYOFWEEK',
+ 'DAYOFYEAR', 'DECODE', 'DEFAULT', 'DEGREES', 'DES_DECRYPT', 'DES_ENCRYPT', 'DIFFERENCE', 'DIMENSION', 'DISJOINT', 'DISTANCE', 'ELT', 'ENCODE',
+ 'ENCRYPT', 'ENDPOINT', 'ENVELOPE', 'EQUALS', 'EXP', 'EXPORT_SET', 'EXTERIORRING', 'EXTRACT', 'EXTRACTVALUE', 'FIELD', 'FIND_IN_SET', 'FLOOR',
+ 'FORMAT', 'FOUND_ROWS', 'FROM_DAYS', 'FROM_UNIXTIME', 'GEOMCOLLFROMTEXT', 'GEOMCOLLFROMWKB', 'GEOMETRYCOLLECTION', 'GEOMETRYCOLLECTIONFROMTEXT',
+ 'GEOMETRYCOLLECTIONFROMWKB', 'GEOMETRYFROMTEXT', 'GEOMETRYFROMWKB', 'GEOMETRYN', 'GEOMETRYTYPE', 'GEOMFROMTEXT', 'GEOMFROMWKB', 'GET_FORMAT',
+ 'GET_LOCK', 'GLENGTH', 'GREATEST', 'GROUP_CONCAT', 'GROUP_UNIQUE_USERS', 'HEX', 'HOUR', 'IF', 'IFNULL', 'INET_ATON', 'INET_NTOA', 'INSERT', 'INSTR',
+ 'INTERIORRINGN', 'INTERSECTION', 'INTERSECTS', 'INTERVAL', 'ISCLOSED', 'ISEMPTY', 'ISNULL', 'ISRING', 'ISSIMPLE', 'IS_FREE_LOCK', 'IS_USED_LOCK',
+ 'LAST_DAY', 'LAST_INSERT_ID', 'LCASE', 'LEAST', 'LEFT', 'LENGTH', 'LINEFROMTEXT', 'LINEFROMWKB', 'LINESTRING', 'LINESTRINGFROMTEXT', 'LINESTRINGFROMWKB',
+ 'LN', 'LOAD_FILE', 'LOCALTIME', 'LOCALTIMESTAMP', 'LOCATE', 'LOG', 'LOG10', 'LOG2', 'LOWER', 'LPAD', 'LTRIM', 'MAKEDATE', 'MAKETIME', 'MAKE_SET',
+ 'MASTER_POS_WAIT', 'MAX', 'MBRCONTAINS', 'MBRDISJOINT', 'MBREQUAL', 'MBRINTERSECTS', 'MBROVERLAPS', 'MBRTOUCHES', 'MBRWITHIN', 'MD5', 'MICROSECOND',
+ 'MID', 'MIN', 'MINUTE', 'MLINEFROMTEXT', 'MLINEFROMWKB', 'MOD', 'MONTH', 'MONTHNAME', 'MPOINTFROMTEXT', 'MPOINTFROMWKB', 'MPOLYFROMTEXT', 'MPOLYFROMWKB',
+ 'MULTILINESTRING', 'MULTILINESTRINGFROMTEXT', 'MULTILINESTRINGFROMWKB', 'MULTIPOINT', 'MULTIPOINTFROMTEXT', 'MULTIPOINTFROMWKB', 'MULTIPOLYGON',
+ 'MULTIPOLYGONFROMTEXT', 'MULTIPOLYGONFROMWKB', 'NAME_CONST', 'NOW', 'NULLIF', 'NUMGEOMETRIES', 'NUMINTERIORRINGS', 'NUMPOINTS', 'OCT', 'OCTET_LENGTH',
+ 'OLD_PASSWORD', 'ORD', 'OVERLAPS', 'PASSWORD', 'PERIOD_ADD', 'PERIOD_DIFF', 'PI', 'POINT', 'POINTFROMTEXT', 'POINTFROMWKB', 'POINTN', 'POINTONSURFACE',
+ 'POLYFROMTEXT', 'POLYFROMWKB', 'POLYGON', 'POLYGONFROMTEXT', 'POLYGONFROMWKB', 'POSITION', 'POW', 'POWER', 'QUARTER', 'QUOTE', 'RADIANS', 'RAND',
+ 'RELATED', 'RELEASE_LOCK', 'REPEAT', 'REPLACE', 'REVERSE', 'RIGHT', 'ROUND', 'ROW_COUNT', 'RPAD', 'RTRIM', 'SCHEMA', 'SECOND', 'SEC_TO_TIME',
+ 'SESSION_USER', 'SHA', 'SHA1', 'SIGN', 'SIN', 'SLEEP', 'SOUNDEX', 'SPACE', 'SQRT', 'SRID', 'STARTPOINT', 'STD', 'STDDEV', 'STDDEV_POP', 'STDDEV_SAMP',
+ 'STRCMP', 'STR_TO_DATE', 'SUBDATE', 'SUBSTR', 'SUBSTRING', 'SUBSTRING_INDEX', 'SUBTIME', 'SUM', 'SYMDIFFERENCE', 'SYSDATE', 'SYSTEM_USER', 'TAN',
+ 'TIME', 'TIMEDIFF', 'TIMESTAMP', 'TIMESTAMPADD', 'TIMESTAMPDIFF', 'TIME_FORMAT', 'TIME_TO_SEC', 'TOUCHES', 'TO_DAYS', 'TRIM', 'TRUNCATE', 'UCASE',
+ 'UNCOMPRESS', 'UNCOMPRESSED_LENGTH', 'UNHEX', 'UNIQUE_USERS', 'UNIX_TIMESTAMP', 'UPDATEXML', 'UPPER', 'USER', 'UTC_DATE', 'UTC_TIME', 'UTC_TIMESTAMP',
+ 'UUID', 'VARIANCE', 'VAR_POP', 'VAR_SAMP', 'VERSION', 'WEEK', 'WEEKDAY', 'WEEKOFYEAR', 'WITHIN', 'X', 'Y', 'YEAR', 'YEARWEEK'
+);
+
// Punctuation that can be used as a boundary between other tokens
protected static $boundaries = array(',', ';', ')', '(', '.', '=', '<', '>', '+', '-', '*', '/', '!', '^', '%', '|', '&', '#');
@@ -85,6 +115,7 @@ class SqlFormatter
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;"';
+ public static $mysql_functions = 'style="color: red;"';
// Boolean - whether or not the current environment is the CLI
// This affects the type of syntax highlighting
@@ -100,6 +131,7 @@ class SqlFormatter
public static $cli_word = "";
public static $cli_error = "\x1b[31;1;7m";
public static $cli_comment = "\x1b[30;1m";
+ public static $cli_functions = "\x1b[37m";
// The tab character to use when formatting SQL
public static $tab = ' ';
@@ -115,6 +147,7 @@ class SqlFormatter
protected static $regex_reserved;
protected static $regex_reserved_newline;
protected static $regex_reserved_toplevel;
+ protected static $regex_mysql_function;
// Cache variables
// Only tokens shorter than this size will be cached. Somewhere between 10 and 20 seems to work well for most cases.
@@ -151,6 +184,8 @@ class SqlFormatter
self::$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::$reserved_newline)).')');
+ self::$regex_mysql_function = '('.implode('|',array_map(array('SqlFormatter', 'quote_regex'),self::$mysql_function)).')';
+
self::$init = true;
}
@@ -255,6 +290,20 @@ class SqlFormatter
}
}
+ // A function must be suceeded by '('
+ // this makes it so in "count(", which considers it as a mysql function.
+
+ $upper = strtoupper($string);
+ // MySQL function
+ if(preg_match('/^('.self::$regex_mysql_function.'[(]|\s|[)])/', $upper,$matches)) {
+
+ return array(
+ self::TOKEN_TYPE=>self::TOKEN_TYPE_FUNCTION,
+ self::TOKEN_VALUE=>substr($string,0,strlen($matches[1])-1)
+ );
+ }
+
+
// Non reserved word
preg_match('/^(.*?)($|\s|["\'`]|'.self::$regex_boundaries.')/',$string,$matches);
@@ -699,7 +748,7 @@ class SqlFormatter
$result .= $token[self::TOKEN_VALUE];
}
- $result = self::remove_blank_lines($result);
+ $result = self::format( $result,false);
return $result;
}
@@ -790,6 +839,9 @@ class SqlFormatter
elseif($type===self::TOKEN_TYPE_COMMENT || $type===self::TOKEN_TYPE_BLOCK_COMMENT) {
return self::highlightComment($token);
}
+ elseif($type===self::TOKEN_TYPE_FUNCTION) {
+ return self::highlightMySQLFunction($token);
+ }
return $token;
}
@@ -843,6 +895,23 @@ class SqlFormatter
return '<span ' . self::$reserved_attributes . '>' . $value . '</span>';
}
}
+
+ /**
+ * Highlights mysql functions
+ *
+ * @param String $value The token's value
+ *
+ * @return String HTML code of the highlighted token.
+ */
+ protected static function highlightMySQLFunction($value)
+ {
+ if(self::is_cli()) {
+ return self::$cli_reserved . $value . "\x1b[0m";//not sure as doesn't have this environment for now.Need to check
+ }
+ else {
+ return '<span ' . self::$mysql_functions . '>' . $value . '</span>';
+ }
+ }
/**
* Highlights a boundary token
@@ -982,14 +1051,4 @@ class SqlFormatter
else return php_sapi_name() === 'cli';
}
- /**
- * Removes blank line between and before sql statements.
- *@param String $sql from which blank lines need to be removed
- *
- * @return String sql without blank lines
- */
- private static function remove_blank_lines($sql)
- {
- return preg_replace('/^\n+|^[\t\s]*\n+/m','',$sql);
- }
}
diff --git a/tests/SqlFormatterTest.php b/tests/SqlFormatterTest.php
index 2c587e4..333483c 100644
--- a/tests/SqlFormatterTest.php
+++ b/tests/SqlFormatterTest.php
@@ -72,7 +72,7 @@ class SqlFormatterTest extends PHPUnit_Framework_TestCase {
}
function testRemoveComments() {
- $expected = "SELECT\n * FROM\n MyTable";
+ $expected = SqlFormatter::format("SELECT\n * FROM\n MyTable",false);
$sql = "/* this is a comment */SELECT#This is another comment\n * FROM-- One final comment\n MyTable";
$actual = SqlFormatter::removeComments($sql);