Parse LOAD statement properly

Fix #131

Might help to fix phpmyadmin/phpmyadmin#12345

Signed-off-by: Deven Bansod <devenbansod.bits@gmail.com>

8 files changed, 420 insertions, 11 deletions

diff --git a/src/Contexts/ContextMySql50000.php b/src/Contexts/ContextMySql50000.php
index 2ed3035..f949e97 100644
--- a/src/Contexts/ContextMySql50000.php
+++ b/src/Contexts/ContextMySql50000.php
@@ -139,8 +139,8 @@ class ContextMySql50000 extends Context
 
         'GROUP BY' => 7, 'NOT NULL' => 7, 'ORDER BY' => 7, 'SET NULL' => 7,
         'AND CHAIN' => 7, 'FULL JOIN' => 7, 'IF EXISTS' => 7, 'LEFT JOIN' => 7,
-        'LESS THAN' => 7, 'NO ACTION' => 7, 'ON DELETE' => 7, 'ON UPDATE' => 7,
-        'UNION ALL' => 7,
+        'LESS THAN' => 7, 'LOAD DATA' => 7, 'NO ACTION' => 7, 'ON DELETE' => 7,
+        'ON UPDATE' => 7, 'UNION ALL' => 7,
         'CROSS JOIN' => 7, 'ESCAPED BY' => 7, 'FOR UPDATE' => 7, 'INNER JOIN' => 7,
         'LINEAR KEY' => 7, 'NO RELEASE' => 7, 'OR REPLACE' => 7, 'RIGHT JOIN' => 7,
         'ENCLOSED BY' => 7, 'LINEAR HASH' => 7, 'STARTING BY' => 7,
diff --git a/src/Contexts/ContextMySql50100.php b/src/Contexts/ContextMySql50100.php
index e6d9ed3..eaa4872 100644
--- a/src/Contexts/ContextMySql50100.php
+++ b/src/Contexts/ContextMySql50100.php
@@ -153,8 +153,8 @@ class ContextMySql50100 extends Context
 
         'GROUP BY' => 7, 'NOT NULL' => 7, 'ORDER BY' => 7, 'SET NULL' => 7,
         'AND CHAIN' => 7, 'FULL JOIN' => 7, 'IF EXISTS' => 7, 'LEFT JOIN' => 7,
-        'LESS THAN' => 7, 'NO ACTION' => 7, 'ON DELETE' => 7, 'ON UPDATE' => 7,
-        'UNION ALL' => 7,
+        'LESS THAN' => 7, 'LOAD DATA' => 7, 'NO ACTION' => 7, 'ON DELETE' => 7,
+        'ON UPDATE' => 7, 'UNION ALL' => 7,
         'CROSS JOIN' => 7, 'ESCAPED BY' => 7, 'FOR UPDATE' => 7, 'INNER JOIN' => 7,
         'LINEAR KEY' => 7, 'NO RELEASE' => 7, 'OR REPLACE' => 7, 'RIGHT JOIN' => 7,
         'ENCLOSED BY' => 7, 'LINEAR HASH' => 7, 'STARTING BY' => 7,
diff --git a/src/Contexts/ContextMySql50500.php b/src/Contexts/ContextMySql50500.php
index 370319f..663736a 100644
--- a/src/Contexts/ContextMySql50500.php
+++ b/src/Contexts/ContextMySql50500.php
@@ -157,8 +157,8 @@ class ContextMySql50500 extends Context
 
         'GROUP BY' => 7, 'NOT NULL' => 7, 'ORDER BY' => 7, 'SET NULL' => 7,
         'AND CHAIN' => 7, 'FULL JOIN' => 7, 'IF EXISTS' => 7, 'LEFT JOIN' => 7,
-        'LESS THAN' => 7, 'NO ACTION' => 7, 'ON DELETE' => 7, 'ON UPDATE' => 7,
-        'UNION ALL' => 7,
+        'LESS THAN' => 7, 'LOAD DATA' => 7, 'NO ACTION' => 7, 'ON DELETE' => 7,
+        'ON UPDATE' => 7, 'UNION ALL' => 7,
         'CROSS JOIN' => 7, 'ESCAPED BY' => 7, 'FOR UPDATE' => 7, 'INNER JOIN' => 7,
         'LINEAR KEY' => 7, 'NO RELEASE' => 7, 'OR REPLACE' => 7, 'RIGHT JOIN' => 7,
         'ENCLOSED BY' => 7, 'LINEAR HASH' => 7, 'STARTING BY' => 7,
diff --git a/src/Contexts/ContextMySql50600.php b/src/Contexts/ContextMySql50600.php
index 3994a75..8b57a06 100644
--- a/src/Contexts/ContextMySql50600.php
+++ b/src/Contexts/ContextMySql50600.php
@@ -163,8 +163,8 @@ class ContextMySql50600 extends Context
 
         'GROUP BY' => 7, 'NOT NULL' => 7, 'ORDER BY' => 7, 'SET NULL' => 7,
         'AND CHAIN' => 7, 'FULL JOIN' => 7, 'IF EXISTS' => 7, 'LEFT JOIN' => 7,
-        'LESS THAN' => 7, 'NO ACTION' => 7, 'ON DELETE' => 7, 'ON UPDATE' => 7,
-        'UNION ALL' => 7,
+        'LESS THAN' => 7, 'LOAD DATA' => 7, 'NO ACTION' => 7, 'ON DELETE' => 7,
+        'ON UPDATE' => 7, 'UNION ALL' => 7,
         'CROSS JOIN' => 7, 'ESCAPED BY' => 7, 'FOR UPDATE' => 7, 'INNER JOIN' => 7,
         'LINEAR KEY' => 7, 'NO RELEASE' => 7, 'OR REPLACE' => 7, 'RIGHT JOIN' => 7,
         'ENCLOSED BY' => 7, 'LINEAR HASH' => 7, 'STARTING BY' => 7,
diff --git a/src/Contexts/ContextMySql50700.php b/src/Contexts/ContextMySql50700.php
index fbc276d..3d3c7c8 100644
--- a/src/Contexts/ContextMySql50700.php
+++ b/src/Contexts/ContextMySql50700.php
@@ -169,8 +169,8 @@ class ContextMySql50700 extends Context
 
         'GROUP BY' => 7, 'NOT NULL' => 7, 'ORDER BY' => 7, 'SET NULL' => 7,
         'AND CHAIN' => 7, 'FULL JOIN' => 7, 'IF EXISTS' => 7, 'LEFT JOIN' => 7,
-        'LESS THAN' => 7, 'NO ACTION' => 7, 'ON DELETE' => 7, 'ON UPDATE' => 7,
-        'UNION ALL' => 7,
+        'LESS THAN' => 7, 'LOAD DATA' => 7, 'NO ACTION' => 7, 'ON DELETE' => 7,
+        'ON UPDATE' => 7, 'UNION ALL' => 7,
         'CROSS JOIN' => 7, 'ESCAPED BY' => 7, 'FOR UPDATE' => 7, 'INNER JOIN' => 7,
         'LINEAR KEY' => 7, 'NO RELEASE' => 7, 'OR REPLACE' => 7, 'RIGHT JOIN' => 7,
         'ENCLOSED BY' => 7, 'LINEAR HASH' => 7, 'STARTING BY' => 7,
diff --git a/src/Parser.php b/src/Parser.php
index a69cc46..1aaa565 100644
--- a/src/Parser.php
+++ b/src/Parser.php
@@ -69,7 +69,7 @@ class Parser extends Core
         'DO' => '',
         'HANDLER' => '',
         'INSERT' => 'PhpMyAdmin\\SqlParser\\Statements\\InsertStatement',
-        'LOAD' => '',
+        'LOAD DATA' => 'PhpMyAdmin\\SqlParser\\Statements\\LoadStatement',
         'REPLACE' => 'PhpMyAdmin\\SqlParser\\Statements\\ReplaceStatement',
         'SELECT' => 'PhpMyAdmin\\SqlParser\\Statements\\SelectStatement',
         'UPDATE' => 'PhpMyAdmin\\SqlParser\\Statements\\UpdateStatement',
diff --git a/src/Statements/LoadStatement.php b/src/Statements/LoadStatement.php
new file mode 100644
index 0000000..858885e
--- /dev/null
+++ b/src/Statements/LoadStatement.php
@@ -0,0 +1,404 @@
+<?php
+
+/**
+ * `LOAD` statement.
+ */
+
+namespace PhpMyAdmin\SqlParser\Statements;
+
+use PhpMyAdmin\SqlParser\Components\ArrayObj;
+use PhpMyAdmin\SqlParser\Components\Condition;
+use PhpMyAdmin\SqlParser\Components\Expression;
+use PhpMyAdmin\SqlParser\Components\ExpressionArray;
+use PhpMyAdmin\SqlParser\Components\FunctionCall;
+use PhpMyAdmin\SqlParser\Components\IntoKeyword;
+use PhpMyAdmin\SqlParser\Components\JoinKeyword;
+use PhpMyAdmin\SqlParser\Components\Limit;
+use PhpMyAdmin\SqlParser\Components\OrderKeyword;
+use PhpMyAdmin\SqlParser\Components\OptionsArray;
+use PhpMyAdmin\SqlParser\Components\SetOperation;
+use PhpMyAdmin\SqlParser\Parser;
+use PhpMyAdmin\SqlParser\Statement;
+use PhpMyAdmin\SqlParser\Token;
+use PhpMyAdmin\SqlParser\TokensList;
+
+/**
+ * `LOAD` statement.
+ *
+ * LOAD DATA [LOW_PRIORITY | CONCURRENT] [LOCAL] INFILE 'file_name'
+ *   [REPLACE | IGNORE]
+ *   INTO TABLE tbl_name
+ *   [PARTITION (partition_name,...)]
+ *   [CHARACTER SET charset_name]
+ *   [{FIELDS | COLUMNS}
+ *       [TERMINATED BY 'string']
+ *       [[OPTIONALLY] ENCLOSED BY 'char']
+ *       [ESCAPED BY 'char']
+ *   ]
+ *   [LINES
+ *       [STARTING BY 'string']
+ *       [TERMINATED BY 'string']
+ *  ]
+ *   [IGNORE number {LINES | ROWS}]
+ *   [(col_name_or_user_var,...)]
+ *   [SET col_name = expr,...]
+ *
+ *
+ * @category   Statements
+ *
+ * @license    https://www.gnu.org/licenses/gpl-2.0.txt GPL-2.0+
+ */
+class LoadStatement extends Statement
+{
+    /**
+     * Options for `LOAD` statements and their slot ID.
+     *
+     * @var array
+     */
+    public static $OPTIONS = array(
+        'LOW_PRIORITY' => 1,
+        'CONCURRENT' => 1,
+        'LOCAL' => 2
+    );
+
+    /**
+     * FIELDS/COLUMNS Options for `LOAD DATA...INFILE` statements.
+     *
+     * @var array
+     */
+    public static $FIELDS_OPTIONS = array(
+        'TERMINATED BY' => array(1, 'expr'),
+        'OPTIONALLY' => 2,
+        'ENCLOSED BY' => array(3, 'expr'),
+        'ESCAPED BY' => array(4, 'expr'),
+    );
+
+    /**
+     * LINES Options for `LOAD DATA...INFILE` statements.
+     *
+     * @var array
+     */
+    public static $LINES_OPTIONS = array(
+        'STARTING BY' => array(1, 'expr'),
+        'TERMINATED BY' => array(2, 'expr'),
+    );
+
+    /**
+     * File name being used to load data
+     *
+     * @var Expression
+     */
+    public $file_name;
+
+    /**
+     * Table used as destination for this statement.
+     *
+     * @var Expression
+     */
+    public $table;
+
+    /**
+     * Partitions used as source for this statement.
+     *
+     * @var ArrayObj
+     */
+    public $partition;
+
+    /**
+     * Character set used in this statement.
+     *
+     * @var Expression
+     */
+    public $charset_name;
+
+    /**
+     * Options for FIELDS/COLUMNS keyword.
+     *
+     * @var OptionsArray
+     *
+     * @see static::$FIELDS_OPTIONS
+     */
+    public $fields_options;
+
+    /**
+     * Whether to use `FIELDS` or `COLUMNS` while building.
+     *
+     * @var bool
+     */
+    public $fields_keyword;
+
+    /**
+     * Options for OPTIONS keyword.
+     *
+     * @var OptionsArray
+     *
+     * @see static::$LINES_OPTIONS
+     */
+    public $lines_options;
+
+    /**
+     * Column names or user variables
+     *
+     * @var ExpressionArray
+     */
+    public $col_name_or_user_var;
+
+    /**
+     * SET clause's updated values(optional)
+     *
+     * @var SetOperation[]
+     */
+    public $set;
+
+    /**
+     * Ignore 'number' LINES/ROWS
+     *
+     * @var Expression
+     */
+    public $ignore_number;
+
+    /**
+     * REPLACE/IGNORE Keyword
+     *
+     * @var string
+     */
+    public $replace_ignore;
+
+    /**
+     * LINES/ROWS Keyword
+     *
+     * @var string
+     */
+    public $lines_rows;
+
+    /**
+     * @return string
+     */
+    public function build()
+    {
+        $ret = 'LOAD DATA ' . $this->options
+            . ' INFILE ' . $this->file_name;
+
+        if ($this->replace_ignore !== null) {
+            $ret .= ' ' . trim($this->replace_ignore);
+        }
+
+        $ret .= ' INTO TABLE ' . $this->table;
+
+        if ($this->partition !== null && count($this->partition) > 0) {
+            $ret .= ' PARTITION ' . ArrayObj::build($this->partition);
+        }
+
+        if ($this->charset_name !== null) {
+            $ret .= ' CHARACTER SET ' . $this->charset_name;
+        }
+
+        if ($this->fields_keyword !== null) {
+            $ret .= ' ' . $this->fields_keyword . ' ' . $this->fields_options;
+        }
+
+        if ($this->lines_options !== null && count($this->lines_options) > 0) {
+            $ret .= ' LINES ' . $this->lines_options;
+        }
+
+        if ($this->ignore_number !== null) {
+            $ret .= ' IGNORE ' . $this->ignore_number . ' ' . $this->lines_rows;
+        }
+
+        if ($this->col_name_or_user_var !== null && count($this->col_name_or_user_var) > 0) {
+            $ret .= ' ' . ExpressionArray::build($this->col_name_or_user_var);
+        }
+
+        if ($this->set !== null && count($this->set) > 0) {
+            $ret .= ' SET ' . SetOperation::build($this->set);
+        }
+
+        return $ret;
+    }
+
+    /**
+     * @param Parser     $parser the instance that requests parsing
+     * @param TokensList $list   the list of tokens to be parsed
+     */
+    public function parse(Parser $parser, TokensList $list)
+    {
+        ++$list->idx; // Skipping `LOAD DATA`.
+
+        // parse any options if provided
+        $this->options = OptionsArray::parse(
+            $parser,
+            $list,
+            static::$OPTIONS
+        );
+        ++$list->idx;
+
+                /**
+         * The state of the parser.
+         *
+         * @var int
+         */
+        $state = 0;
+
+        for (; $list->idx < $list->count; ++$list->idx) {
+            /**
+             * Token parsed at this moment.
+             *
+             * @var Token
+             */
+            $token = $list->tokens[$list->idx];
+
+            // End of statement.
+            if ($token->type === Token::TYPE_DELIMITER) {
+                break;
+            }
+
+            // Skipping whitespaces and comments.
+            if (($token->type === Token::TYPE_WHITESPACE) || ($token->type === Token::TYPE_COMMENT)) {
+                continue;
+            }
+
+            if ($state === 0) {
+                if ($token->type === Token::TYPE_KEYWORD
+                    && $token->keyword !== 'INFILE'
+                ) {
+                    $parser->error('Unexpected keyword.', $token);
+                    break;
+                } elseif ($token->type !== Token::TYPE_KEYWORD) {
+                    $parser->error('Unexpected token1.', $token);
+                    break;
+                }
+
+                ++$list->idx;
+                $this->file_name = Expression::parse(
+                    $parser,
+                    $list,
+                    array('parseField' => 'file')
+                );
+                $state = 1;
+            } elseif ($state === 1) {
+                if (($token->type === Token::TYPE_KEYWORD)
+                    && ($token->keyword === 'REPLACE'
+                    || $token->keyword === 'IGNORE')
+                ) {
+                    $this->replace_ignore = trim($token->keyword);
+                } elseif ($token->type === Token::TYPE_KEYWORD
+                    && $token->keyword === 'INTO'
+                ) {
+                    $state = 2;
+                }
+            } elseif ($state === 2) {
+                if ($token->type === Token::TYPE_KEYWORD
+                    && $token->keyword === 'TABLE'
+                ) {
+                    $list->idx++;
+                    $this->table = Expression::parse($parser, $list, array('parseField' => 'table'));
+                    $state = 3;
+                } else {
+                    $parser->error('Unexpected token2.', $token);
+                    break;
+                }
+            } elseif ($state >= 3 && $state <= 7) {
+                if ($token->type === Token::TYPE_KEYWORD) {
+                    $newState = $this->parseKeywordsAccordingToState(
+                        $parser, $list, $state
+                    );
+                    if ($newState === $state) {
+                        // Avoid infinite loop
+                        break;
+                    }
+                } elseif ($token->type === Token::TYPE_OPERATOR
+                    && $token->token === '('
+                ) {
+                    $this->col_name_or_user_var
+                        = ExpressionArray::parse($parser, $list);
+                    $state = 7;
+                } else {
+                    $parser->error('Unexpected token3.', $token);
+                    break;
+                }
+            }
+        }
+
+        --$list->idx;
+    }
+
+    public function parseFileOptions(Parser $parser, TokensList $list, $keyword = 'FIELDS')
+    {
+        ++$list->idx;
+
+        if ($keyword === 'FIELDS' || $keyword === 'COLUMNS') {
+            // parse field options
+            $this->fields_options = OptionsArray::parse(
+                $parser,
+                $list,
+                static::$FIELDS_OPTIONS
+            );
+
+            $this->fields_keyword = $keyword;
+        } else {
+            // parse line options
+            $this->lines_options = OptionsArray::parse(
+                $parser,
+                $list,
+                static::$LINES_OPTIONS
+            );
+        }
+    }
+
+
+    public function parseKeywordsAccordingToState($parser, $list, $state) {
+        $token = $list->tokens[$list->idx];
+
+        switch ($state) {
+            case 3:
+                if ($token->keyword === 'PARTITION') {
+                    $list->idx++;
+                    $this->partition = ArrayObj::parse($parser, $list);
+                    $state = 4;
+                    return $state;
+                }
+            case 4:
+                if ($token->keyword === 'CHARACTER SET') {
+                    $list->idx++;
+                    $this->charset_name = Expression::parse($parser, $list);
+                    $state = 5;
+                    return $state;
+                }
+            case 5:
+                if ($token->keyword === 'FIELDS'
+                    || $token->keyword === 'COLUMNS'
+                    || $token->keyword === 'LINES'
+                ) {
+                    $this->parseFileOptions($parser, $list, $token->value);
+                    $state = 6;
+                    return $state;
+                }
+            case 6:
+                if ($token->keyword === 'IGNORE') {
+                    $list->idx++;
+
+                    $this->ignore_number = Expression::parse($parser, $list);
+                    $nextToken = $list->getNextOfType(Token::TYPE_KEYWORD);
+
+                    if ($nextToken->type === Token::TYPE_KEYWORD
+                        && (($nextToken->keyword === 'LINES')
+                        || ($nextToken->keyword === 'ROWS'))
+                    ) {
+                        $this->lines_rows = $nextToken->token;
+                    }
+                    $state = 7;
+                    return $state;
+                }
+            case 7:
+                if ($token->keyword === 'SET') {
+                    $list->idx++;
+                    $this->set = SetOperation::parse($parser, $list);
+                    $state = 8;
+                    return $state;
+                }
+            default:
+        }
+        return $state;
+    }
+
+}
diff --git a/src/Utils/Query.php b/src/Utils/Query.php
index 5ad701e..a62537e 100644
--- a/src/Utils/Query.php
+++ b/src/Utils/Query.php
@@ -20,6 +20,7 @@ use PhpMyAdmin\SqlParser\Statements\DeleteStatement;
 use PhpMyAdmin\SqlParser\Statements\DropStatement;
 use PhpMyAdmin\SqlParser\Statements\ExplainStatement;
 use PhpMyAdmin\SqlParser\Statements\InsertStatement;
+use PhpMyAdmin\SqlParser\Statements\LoadStatement;
 use PhpMyAdmin\SqlParser\Statements\OptimizeStatement;
 use PhpMyAdmin\SqlParser\Statements\RenameStatement;
 use PhpMyAdmin\SqlParser\Statements\RepairStatement;
@@ -340,6 +341,10 @@ class Query
             $flags['querytype'] = 'INSERT';
             $flags['is_affected'] = true;
             $flags['is_insert'] = true;
+        } elseif ($statement instanceof LoadStatement) {
+            $flags['querytype'] = 'LOAD';
+            $flags['is_affected'] = true;
+            $flags['is_insert'] = true;
         } elseif ($statement instanceof ReplaceStatement) {
             $flags['querytype'] = 'REPLACE';
             $flags['is_affected'] = true;