Merge #287 - Fix #285 - Fixing conflict in lexer when '*/*' was in a query.

Pull-request: #287
Fixes: #285

Signed-off-by: William Desportes <[email protected]>

Author: williamdes

src/Context.php

@@ -336,18 +336,28 @@ public static function isComment($str, $end = false)
         if ($len === 0) {
             return null;
         }
+
+        // If comment is Bash style (#):
         if ($str[0] === '#') {
             return Token::FLAG_COMMENT_BASH;
-        } elseif (($len > 1) && ($str[0] === '/') && ($str[1] === '*')) {
-            return (($len > 2) && ($str[2] === '!')) ?
+        }
+        // If comment is opening C style (/*), warning, it could be a MySQL command (/*!)
+        if (($len > 1) && ($str[0] === '/') && ($str[1] === '*')) {
+            return ($len > 2) && ($str[2] === '!') ?
                 Token::FLAG_COMMENT_MYSQL_CMD : Token::FLAG_COMMENT_C;
-        } elseif (($len > 1) && ($str[0] === '*') && ($str[1] === '/')) {
+        }
+        // If comment is closing C style (*/), warning, it could conflicts with wildcard and a real opening C style.
+        // It would looks like the following valid SQL statement: "SELECT */* comment */ FROM...".
+        if (($len > 1) && ($str[0] === '*') && ($str[1] === '/')) {
             return Token::FLAG_COMMENT_C;
-        } elseif (($len > 2) && ($str[0] === '-')
+        }
+        // If comment is SQL style (--\s?):
+        if (($len > 2) && ($str[0] === '-')
             && ($str[1] === '-') && static::isWhitespace($str[2])
         ) {
             return Token::FLAG_COMMENT_SQL;
-        } elseif (($len === 2) && $end && ($str[0] === '-') && ($str[1] === '-')) {
+        }
+        if (($len === 2) && $end && ($str[0] === '-') && ($str[1] === '-')) {
             return Token::FLAG_COMMENT_SQL;
         }
 

src/Lexer.php

@@ -559,6 +559,19 @@ public function parseComment()
         if (++$this->last < $this->len) {
             $token .= $this->str[$this->last];
             if (Context::isComment($token)) {
+                // There might be a conflict with "*" operator here, when string is "*/*".
+                // This can occurs in the following statements:
+                // - "SELECT */* comment */ FROM ..."
+                // - "SELECT 2*/* comment */3 AS `six`;"
+                $next = $this->last+1;
+                if (($next < $this->len) && $this->str[$next] === '*') {
+                    // Conflict in "*/*": first "*" was not for ending a comment.
+                    // Stop here and let other parsing method define the true behavior of that first star.
+                    $this->last = $iBak;
+
+                    return null;
+                }
+
                 $flags = Token::FLAG_COMMENT_C;
 
                 // This comment already ended. It may be a part of a

tests/Lexer/LexerTest.php

@@ -84,7 +84,8 @@ public function lexProvider()
             array('lexer/lexWhitespace'),
             array('lexer/lexLabel1'),
             array('lexer/lexLabel2'),
-            array('lexer/lexNoLabel')
+            array('lexer/lexNoLabel'),
+            array('lexer/lexWildcardThenComment')
         );
     }
 }

tests/data/lexer/lexWildcardThenComment.in

@@ -0,0 +1,33 @@
+SELECT */* comment */
+
+SELECT /* comment */*
+
+SELECT 2*/* comment */3
+
+SELECT 2/* comment */*3
+
+SELECT */*
+comment
+on
+multiple
+lines
+*/FROM
+
+DELETE foo.*/* foo */ USING
+
+DELETE foo.*/* foo */,bar.*/*bar*/ USING
+
+SELECT `*`/*with comment*/ AS star_field
+
+SELECT `*`,*/*with comment*/
+
+DELETE a.*/*multi
+line /* with C open tag
+comment inside */ USING
+
+SELECT 2*/* operator */3 + 3/* operator */*2,/* start wildcard */*/* end wildcard */
+
+SELECT `*`/*a*/*/*b*/`*`
+
+-- invalid queries
+/* SELECT */*