Fix #22 : Validate order of clauses in the parsing of statements

src/Parser.php

@@ -531,6 +531,9 @@ public function parse()
                 $lastStatement->last = $statement->last;
 
                 $unionType = false;
+
+                // Validate clause order
+                $statement->validateClauseOrder($this, $list);
                 continue;
             }
 
@@ -556,9 +559,15 @@ public function parse()
                     }
                     $lastTransaction = null;
                 }
+
+                // Validate clause order
+                $statement->validateClauseOrder($this, $list);
                 continue;
             }
 
+            // Validate clause order
+            $statement->validateClauseOrder($this, $list);
+
             // Finally, storing the statement.
             if ($lastTransaction !== null) {
                 $lastTransaction->statements[] = $statement;

src/Statement.php

@@ -422,4 +422,47 @@ public function __toString()
     {
         return $this->build();
     }
+
+    /**
+     * Validates the order of the clauses in parsed statement
+     * Ideally this should be called after successfully
+     * completing the parsing of each statement
+     *
+     * @param Parser     $parser The instance that requests parsing.
+     * @param TokensList $list   The list of tokens to be parsed.
+     *
+     * @return boolean
+     */
+    public function validateClauseOrder($parser, $list)
+    {
+        $clauses = array_flip(array_keys($this->getClauses()));
+
+        if (empty($clauses)
+            || count($clauses) == 0
+        ) {
+            return true;
+        }
+
+        $minIdx = -1;
+        foreach ($clauses as $clauseType => $index) {
+            $clauseStartIdx = Utils\Query::getClauseStartOffset(
+                $this,
+                $list,
+                $clauseType
+            );
+
+            if ($clauseStartIdx != -1 && $clauseStartIdx < $minIdx) {
+                $token = $list->tokens[$clauseStartIdx];
+                $parser->error(
+                    __('Unexpected ordering of clauses.'),
+                    $token
+                );
+                return false;
+            } elseif ($clauseStartIdx != -1) {
+                $minIdx = $clauseStartIdx;
+            }
+        }
+
+        return true;
+    }
 }

src/Utils/Query.php

@@ -783,4 +783,66 @@ public static function getFirstStatement($query, $delimiter = null)
 
         return array(trim($statement), $query, $delimiter);
     }
+
+    /**
+     * Gets a starting offset of a specific clause.
+     *
+     * @param Statement  $statement The parsed query that has to be modified.
+     * @param TokensList $list      The list of tokens.
+     * @param string     $clause    The clause to be returned.
+     *
+     * @return int
+     */
+    public static function getClauseStartOffset($statement, $list, $clause)
+    {
+
+        /**
+         * The index of the current clause.
+         *
+         * @var int $currIdx
+         */
+        $currIdx = 0;
+
+        /**
+         * The count of brackets.
+         * We keep track of them so we won't insert the clause in a subquery.
+         *
+         * @var int $brackets
+         */
+        $brackets = 0;
+
+        /**
+         * The clauses of this type of statement and their index.
+         *
+         * @var array $clauses
+         */
+        $clauses = array_flip(array_keys($statement->getClauses()));
+
+        for ($i = $statement->first; $i <= $statement->last; ++$i) {
+            $token = $list->tokens[$i];
+
+            if ($token->type === Token::TYPE_COMMENT) {
+                continue;
+            }
+
+            if ($token->type === Token::TYPE_OPERATOR) {
+                if ($token->value === '(') {
+                    ++$brackets;
+                } elseif ($token->value === ')') {
+                    --$brackets;
+                }
+            }
+
+            if ($brackets == 0) {
+                if (($token->type === Token::TYPE_KEYWORD)
+                    && (isset($clauses[$token->value]))
+                    && ($clause === $token->value)
+                ) {
+                    return $i;
+                }
+            }
+        }
+
+        return -1;
+    }
 }

tests/Parser/SelectStatementTest.php

@@ -54,6 +54,7 @@ public function testSelectProvider()
             array('parser/parseSelectJoinNaturalLeftOuter'),
             array('parser/parseSelectJoinNaturalRightOuter'),
             array('parser/parseSelectJoinMultiple'),
+            array('parser/parseSelectWrongOrder'),
         );
     }
 }

tests/data/parser/parseSelectWrongOrder.in

@@ -0,0 +1 @@
+SELECT pid, name2 FROM tablename LIMIT 10 WHERE pid = 20

tests/data/parser/parseSelectWrongOrder.out

@@ -0,0 +1 @@
+a:4:{s:5:"query";s:56:"SELECT pid, name2 FROM tablename LIMIT 10 WHERE pid = 20";s:5:"lexer";O:15:"SqlParser\Lexer":8:{s:6:"strict";b:0;s:3:"str";s:56:"SELECT pid, name2 FROM tablename LIMIT 10 WHERE pid = 20";s:3:"len";i:56;s:4:"last";i:56;s:4:"list";O:20:"SqlParser\TokensList":3:{s:6:"tokens";a:23:{i:0;O:15:"SqlParser\Token":5:{s:5:"token";s:6:"SELECT";s:5:"value";s:6:"SELECT";s:4:"type";i:1;s:5:"flags";i:3;s:8:"position";i:0;}i:1;O:15:"SqlParser\Token":5:{s:5:"token";s:1:" ";s:5:"value";s:1:" ";s:4:"type";i:3;s:5:"flags";i:0;s:8:"position";i:6;}i:2;O:15:"SqlParser\Token":5:{s:5:"token";s:3:"pid";s:5:"value";s:3:"pid";s:4:"type";i:0;s:5:"flags";i:0;s:8:"position";i:7;}i:3;O:15:"SqlParser\Token":5:{s:5:"token";s:1:",";s:5:"value";s:1:",";s:4:"type";i:2;s:5:"flags";i:16;s:8:"position";i:10;}i:4;O:15:"SqlParser\Token":5:{s:5:"token";s:1:" ";s:5:"value";s:1:" ";s:4:"type";i:3;s:5:"flags";i:0;s:8:"position";i:11;}i:5;O:15:"SqlParser\Token":5:{s:5:"token";s:5:"name2";s:5:"value";s:5:"name2";s:4:"type";i:0;s:5:"flags";i:0;s:8:"position";i:12;}i:6;O:15:"SqlParser\Token":5:{s:5:"token";s:1:" ";s:5:"value";s:1:" ";s:4:"type";i:3;s:5:"flags";i:0;s:8:"position";i:17;}i:7;O:15:"SqlParser\Token":5:{s:5:"token";s:4:"FROM";s:5:"value";s:4:"FROM";s:4:"type";i:1;s:5:"flags";i:3;s:8:"position";i:18;}i:8;O:15:"SqlParser\Token":5:{s:5:"token";s:1:" ";s:5:"value";s:1:" ";s:4:"type";i:3;s:5:"flags";i:0;s:8:"position";i:22;}i:9;O:15:"SqlParser\Token":5:{s:5:"token";s:9:"tablename";s:5:"value";s:9:"tablename";s:4:"type";i:0;s:5:"flags";i:0;s:8:"position";i:23;}i:10;O:15:"SqlParser\Token":5:{s:5:"token";s:1:" ";s:5:"value";s:1:" ";s:4:"type";i:3;s:5:"flags";i:0;s:8:"position";i:32;}i:11;O:15:"SqlParser\Token":5:{s:5:"token";s:5:"LIMIT";s:5:"value";s:5:"LIMIT";s:4:"type";i:1;s:5:"flags";i:3;s:8:"position";i:33;}i:12;O:15:"SqlParser\Token":5:{s:5:"token";s:1:" ";s:5:"value";s:1:" ";s:4:"type";i:3;s:5:"flags";i:0;s:8:"position";i:38;}i:13;O:15:"SqlParser\Token":5:{s:5:"token";s:2:"10";s:5:"value";i:10;s:4:"type";i:6;s:5:"flags";i:0;s:8:"position";i:39;}i:14;O:15:"SqlParser\Token":5:{s:5:"token";s:1:" ";s:5:"value";s:1:" ";s:4:"type";i:3;s:5:"flags";i:0;s:8:"position";i:41;}i:15;O:15:"SqlParser\Token":5:{s:5:"token";s:5:"WHERE";s:5:"value";s:5:"WHERE";s:4:"type";i:1;s:5:"flags";i:3;s:8:"position";i:42;}i:16;O:15:"SqlParser\Token":5:{s:5:"token";s:1:" ";s:5:"value";s:1:" ";s:4:"type";i:3;s:5:"flags";i:0;s:8:"position";i:47;}i:17;O:15:"SqlParser\Token":5:{s:5:"token";s:3:"pid";s:5:"value";s:3:"pid";s:4:"type";i:0;s:5:"flags";i:0;s:8:"position";i:48;}i:18;O:15:"SqlParser\Token":5:{s:5:"token";s:1:" ";s:5:"value";s:1:" ";s:4:"type";i:3;s:5:"flags";i:0;s:8:"position";i:51;}i:19;O:15:"SqlParser\Token":5:{s:5:"token";s:1:"=";s:5:"value";s:1:"=";s:4:"type";i:2;s:5:"flags";i:2;s:8:"position";i:52;}i:20;O:15:"SqlParser\Token":5:{s:5:"token";s:1:" ";s:5:"value";s:1:" ";s:4:"type";i:3;s:5:"flags";i:0;s:8:"position";i:53;}i:21;O:15:"SqlParser\Token":5:{s:5:"token";s:2:"20";s:5:"value";i:20;s:4:"type";i:6;s:5:"flags";i:0;s:8:"position";i:54;}i:22;O:15:"SqlParser\Token":5:{s:5:"token";N;s:5:"value";N;s:4:"type";i:9;s:5:"flags";i:0;s:8:"position";N;}}s:5:"count";i:23;s:3:"idx";i:23;}s:9:"delimiter";s:1:";";s:12:"delimiterLen";i:1;s:6:"errors";a:0:{}}s:6:"parser";O:16:"SqlParser\Parser":5:{s:4:"list";r:8;s:6:"strict";b:0;s:6:"errors";a:0:{}s:10:"statements";a:1:{i:0;O:36:"SqlParser\Statements\SelectStatement":15:{s:4:"expr";a:2:{i:0;O:31:"SqlParser\Components\Expression":7:{s:8:"database";N;s:5:"table";N;s:6:"column";s:3:"pid";s:4:"expr";s:3:"pid";s:5:"alias";N;s:8:"function";N;s:8:"subquery";N;}i:1;O:31:"SqlParser\Components\Expression":7:{s:8:"database";N;s:5:"table";N;s:6:"column";s:5:"name2";s:4:"expr";s:5:"name2";s:5:"alias";N;s:8:"function";N;s:8:"subquery";N;}}s:4:"from";a:1:{i:0;O:31:"SqlParser\Components\Expression":7:{s:8:"database";N;s:5:"table";s:9:"tablename";s:6:"column";N;s:4:"expr";s:9:"tablename";s:5:"alias";N;s:8:"function";N;s:8:"subquery";N;}}s:9:"partition";N;s:5:"where";a:1:{i:0;O:30:"SqlParser\Components\Condition":3:{s:11:"identifiers";a:1:{i:0;s:3:"pid";}s:10:"isOperator";b:0;s:4:"expr";s:8:"pid = 20";}}s:5:"group";N;s:6:"having";N;s:5:"order";N;s:5:"limit";O:26:"SqlParser\Components\Limit":2:{s:6:"offset";i:0;s:8:"rowCount";i:10;}s:9:"procedure";N;s:4:"into";N;s:4:"join";N;s:5:"union";a:0:{}s:7:"options";O:33:"SqlParser\Components\OptionsArray":1:{s:7:"options";a:0:{}}s:5:"first";i:0;s:4:"last";i:21;}}s:8:"brackets";i:0;}s:6:"errors";a:2:{s:5:"lexer";a:0:{}s:6:"parser";a:1:{i:0;a:3:{i:0;s:31:"Unexpected ordering of clauses.";i:1;r:76;i:2;i:0;}}}}