/**
* Some sections are just lists of expressions, like the WHERE and HAVING clauses.
* This function processes these sections. Recursive.
*/
private function process_expr_list($tokens)
{
$resultList = array();
$skip_next = false;
$prev = new ExpressionToken();
foreach ($tokens as $k => $v) {
$curr = new ExpressionToken($k, $v);
if ($curr->isWhitespaceToken()) {
continue;
}
if ($skip_next) {
# skip the next non-whitespace token
$skip_next = false;
continue;
}
/* is it a subquery?*/
if ($curr->isSubQueryToken()) {
$curr->setSubTree($this->parse($this->removeParenthesisFromStart($curr->getTrim())));
$curr->setTokenType(ExpressionType::SUBQUERY);
} elseif ($curr->isEnclosedWithinParenthesis()) {
/* is it an in-list? */
$localTokenList = $this->splitSQLIntoTokens($this->removeParenthesisFromStart($curr->getTrim()));
if ($prev->getUpper() === 'IN') {
foreach ($localTokenList as $k => $v) {
$tmpToken = new ExpressionToken($k, $v);
if ($tmpToken->isCommaToken()) {
unset($localTokenList[$k]);
}
}
$localTokenList = array_values($localTokenList);
$curr->setSubTree($this->process_expr_list($localTokenList));
$curr->setTokenType(ExpressionType::IN_LIST);
} elseif ($prev->getUpper() === 'AGAINST') {
$match_mode = false;
foreach ($localTokenList as $k => $v) {
$tmpToken = new ExpressionToken($k, $v);
switch ($tmpToken->getUpper()) {
case 'WITH':
$match_mode = 'WITH QUERY EXPANSION';
break;
case 'IN':
$match_mode = 'IN BOOLEAN MODE';
break;
default:
}
if ($match_mode !== false) {
unset($localTokenList[$k]);
}
}
$tmpToken = $this->process_expr_list($localTokenList);
if ($match_mode !== false) {
$match_mode = new ExpressionToken(0, $match_mode);
$match_mode->setTokenType(ExpressionType::MATCH_MODE);
$tmpToken[] = $match_mode->toArray();
}
$curr->setSubTree($tmpToken);
$curr->setTokenType(ExpressionType::MATCH_ARGUMENTS);
$prev->setTokenType(ExpressionType::SIMPLE_FUNCTION);
} elseif ($prev->isColumnReference() || $prev->isFunction() || $prev->isAggregateFunction()) {
# if we have a colref followed by a parenthesis pair,
# it isn't a colref, it is a user-function
$localExpr = new ExpressionToken();
$tmpExprList = array();
foreach ($localTokenList as $k => $v) {
$tmpToken = new ExpressionToken($k, $v);
if (!$tmpToken->isCommaToken()) {
$localExpr->addToken($v);
$tmpExprList[] = $v;
} else {
# an expression could have multiple parts split by operands
# if we have a comma, it is a split-point for expressions
$tmpExprList = array_values($tmpExprList);
$localExprList = $this->process_expr_list($tmpExprList);
if (count($localExprList) > 1) {
$localExpr->setSubTree($localExprList);
$localExpr->setTokenType(ExpressionType::EXPRESSION);
$localExprList = $localExpr->toArray();
$localExprList['alias'] = false;
$localExprList = array($localExprList);
}
if (!$curr->getSubTree()) {
$curr->setSubTree($localExprList);
} else {
$tmpExprList = $curr->getSubTree();
$curr->setSubTree(array_merge($tmpExprList, $localExprList));
}
$tmpExprList = array();
$localExpr = new ExpressionToken();
}
}
$tmpExprList = array_values($tmpExprList);
$localExprList = $this->process_expr_list($tmpExprList);
if (count($localExprList) > 1) {
$localExpr->setSubTree($localExprList);
$localExpr->setTokenType(ExpressionType::EXPRESSION);
$localExprList = $localExpr->toArray();
$localExprList['alias'] = false;
$localExprList = array($localExprList);
}
if (!$curr->getSubTree()) {
$curr->setSubTree($localExprList);
} else {
$tmpExprList = $curr->getSubTree();
$curr->setSubTree(array_merge($tmpExprList, $localExprList));
}
$prev->setSubTree($curr->getSubTree());
if ($prev->isColumnReference()) {
$prev->setTokenType(ExpressionType::SIMPLE_FUNCTION);
}
array_pop($resultList);
$curr = $prev;
}
# we have parenthesis, but it seems to be an expression
if ($curr->isUnspecified()) {
$curr->setSubTree($this->process_expr_list($localTokenList));
$curr->setTokenType(ExpressionType::BRACKET_EXPRESSION);
}
} elseif ($curr->isVariableToken()) {
// a variable
$curr->setTokenType($this->getVariableType($curr->getUpper()));
$curr->setSubTree(false);
} else {
/* it is either an operator, a colref or a constant */
switch ($curr->getUpper()) {
case '*':
$curr->setSubTree(false);
#no subtree
# single or first element of expression list -> all-column-alias
if (empty($resultList)) {
$curr->setTokenType(ExpressionType::COLREF);
break;
}
# if the last token is colref, const or expression
# then * is an operator
# but if the previous colref ends with a dot, the * is the all-columns-alias
if (!$prev->isColumnReference() && !$prev->isConstant() && !$prev->isExpression() && !$prev->isBracketExpression() && !$prev->isAggregateFunction()) {
$curr->setTokenType(ExpressionType::COLREF);
break;
}
if ($prev->isColumnReference() && $prev->endsWith(".")) {
$prev->addToken('*');
# tablealias dot *
continue 2;
# skip the current token
}
$curr->setTokenType(ExpressionType::OPERATOR);
break;
case 'AND':
case '&&':
case 'BETWEEN':
case 'AND':
case 'BINARY':
case '&':
case '~':
case '|':
case '^':
case 'DIV':
case '/':
case '<=>':
case '=':
case '>=':
case '>':
case 'IS':
case 'NOT':
case '<<':
case '<=':
case '<':
case 'LIKE':
case '%':
case '!=':
case '<>':
case 'REGEXP':
case '!':
case '||':
case 'OR':
case '>>':
case 'RLIKE':
case 'SOUNDS':
case 'XOR':
case 'IN':
$curr->setSubTree(false);
$curr->setTokenType(ExpressionType::OPERATOR);
break;
case 'NULL':
$curr->setSubTree(false);
$curr->setTokenType(ExpressionType::CONSTANT);
break;
case '-':
case '+':
// differ between preceding sign and operator
$curr->setSubTree(false);
if ($prev->isColumnReference() || $prev->isFunction() || $prev->isAggregateFunction() || $prev->isConstant() || $prev->isSubQuery() || $prev->isExpression() || $prev->isBracketExpression()) {
$curr->setTokenType(ExpressionType::OPERATOR);
} else {
$curr->setTokenType(ExpressionType::SIGN);
}
break;
default:
$curr->setSubTree(false);
switch ($curr->getToken(0)) {
case "'":
case '"':
# it is a string literal
$curr->setTokenType(ExpressionType::CONSTANT);
break;
case '`':
# it is an escaped colum name
$curr->setTokenType(ExpressionType::COLREF);
break;
default:
if (is_numeric($curr->getToken())) {
if ($prev->isSign()) {
$prev->addToken($curr->getToken());
# it is a negative numeric constant
$prev->setTokenType(ExpressionType::CONSTANT);
continue 3;
# skip current token
} else {
$curr->setTokenType(ExpressionType::CONSTANT);
}
} else {
$curr->setTokenType(ExpressionType::COLREF);
}
break;
}
}
}
/* is a reserved word? */
if (!$curr->isOperator() && !$curr->isInList() && !$curr->isFunction() && !$curr->isAggregateFunction() && in_array($curr->getUpper(), parent::$reserved)) {
if (in_array($curr->getUpper(), parent::$aggregateFunctions)) {
$curr->setTokenType(ExpressionType::AGGREGATE_FUNCTION);
} elseif ($curr->getUpper() === 'NULL') {
// it is a reserved word, but we would like to set it as constant
$curr->setTokenType(ExpressionType::CONSTANT);
} else {
if (in_array($curr->getUpper(), parent::$parameterizedFunctions)) {
// issue 60: check functions with parameters
// -> colref (we check parameters later)
// -> if there is no parameter, we leave the colref
$curr->setTokenType(ExpressionType::COLREF);
} elseif (in_array($curr->getUpper(), parent::$functions)) {
$curr->setTokenType(ExpressionType::SIMPLE_FUNCTION);
} else {
$curr->setTokenType(ExpressionType::RESERVED);
}
}
}
if ($curr->isUnspecified()) {
$curr->setTokenType(ExpressionType::EXPRESSION);
$curr->setSubTree($this->process_expr_list($this->splitSQLIntoTokens($curr->getTrim())));
}
$resultList[] = $curr;
$prev = $curr;
}
// end of for-loop
return $this->toArray($resultList);
}