Enable Lexing of begin_label in the in-procedure statements

Ex. BEGIN.. END, LOOP, REPEAT, WHILE statements

Fix #57
Fix phpmyadmin/phpmyadmin#12344
Fix phpmyadmin/phpmyadmin#12228

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

3 files changed, 78 insertions, 1 deletions

diff --git a/src/Context.php b/src/Context.php
index 0bdc696..3685857 100644
--- a/src/Context.php
+++ b/src/Context.php
@@ -30,6 +30,16 @@ abstract class Context
     const KEYWORD_MAX_LENGTH = 30;
 
     /**
+     * The maximum length of a label.
+     *
+     * @see static::$TOKEN_LABEL
+     * Ref: https://dev.mysql.com/doc/refman/5.7/en/statement-labels.html
+     *
+     * @var int
+     */
+    const LABEL_MAX_LENGTH = 16;
+
+    /**
      * The maximum length of an operator.
      *
      * @see static::$TOKEN_OPERATOR
diff --git a/src/Lexer.php b/src/Lexer.php
index 7fb2356..9de2a7a 100644
--- a/src/Lexer.php
+++ b/src/Lexer.php
@@ -76,7 +76,7 @@ class Lexer
 
         'parseDelimiter', 'parseWhitespace', 'parseNumber',
         'parseComment', 'parseOperator', 'parseBool', 'parseString',
-        'parseSymbol', 'parseKeyword', 'parseUnknown'
+        'parseSymbol', 'parseKeyword', 'parseLabel', 'parseUnknown'
     );
 
     /**
@@ -442,6 +442,61 @@ class Lexer
     }
 
     /**
+     * Parses a label.
+     *
+     * @return Token
+     */
+    public function parseLabel()
+    {
+        $token = '';
+
+        /**
+         * Value to be returned.
+         *
+         * @var Token $ret
+         */
+        $ret = null;
+
+        /**
+         * The value of `$this->last` where `$token` ends in `$this->str`.
+         *
+         * @var int $iEnd
+         */
+        $iEnd = $this->last;
+
+        /**
+         * Whether last parsed character is a whitespace.
+         *
+         * @var bool $lastSpace
+         */
+        $lastSpace = false;
+
+        for ($j = 1; $j < Context::LABEL_MAX_LENGTH && $this->last < $this->len; ++$j, ++$this->last) {
+            // Composed keywords shouldn't have more than one whitespace between
+            // keywords.
+            if (Context::isWhitespace($this->str[$this->last])) {
+                if ($lastSpace) {
+                    --$j; // The size of the keyword didn't increase.
+                    continue;
+                } else {
+                    $lastSpace = true;
+                }
+            } elseif ($this->str[$this->last] === ':') {
+                $token .= $this->str[$this->last];
+                $ret = new Token($token, Token::TYPE_LABEL);
+                $iEnd = $this->last;
+                break;
+            } else {
+                $lastSpace = false;
+            }
+            $token .= $this->str[$this->last];
+        }
+
+        $this->last = $iEnd;
+        return $ret;
+    }
+
+    /**
      * Parses an operator.
      *
      * @return Token
diff --git a/src/Token.php b/src/Token.php
index 3181a44..e5001ef 100644
--- a/src/Token.php
+++ b/src/Token.php
@@ -119,6 +119,18 @@ class Token
      */
     const TYPE_DELIMITER                =  9;
 
+    /**
+     * Labels in LOOP statement, ITERATE statement etc.
+     * For example (only for begin label):
+     *  begin_label: BEGIN [statement_list] END [end_label]
+     *  begin_label: LOOP [statement_list] END LOOP [end_label]
+     *  begin_label: REPEAT [statement_list] ... END REPEAT [end_label]
+     *  begin_label: WHILE ... DO [statement_list] END WHILE [end_label]
+     *
+     * @var int
+     */
+    const TYPE_LABEL                =  10;
+
     // Flags that describe the tokens in more detail.
     // All keywords must have flag 1 so `Context::isKeyword` method doesn't
     // require strict comparison.