protected function lookForBaseExpression($sql, &$charPos, &$parsed, $key, &$backtracking)
{
if (!is_numeric($key)) {
if ($key === 'UNION' || $key === 'UNION ALL' || $key === 'expr_type' && isset($this->flippedBacktrackingTypes[$parsed]) || $key === 'select-option' && $parsed !== false || $key === 'alias' && $parsed !== false) {
// we hold the current position and come back after the next base_expr
// we do this, because the next base_expr contains the complete expression/subquery/record
// and we have to look into it too
$backtracking[] = $charPos;
} elseif (($key === 'ref_clause' || $key === 'columns') && $parsed !== false) {
// we hold the current position and come back after n base_expr(s)
// there is an array of sub-elements before (!) the base_expr clause of the current element
// so we go through the sub-elements and must come at the end
$backtracking[] = $charPos;
for ($i = 1; $i < count($parsed); $i++) {
$backtracking[] = false;
// backtracking only after n base_expr!
}
} elseif ($key === 'sub_tree' && $parsed !== false || $key === 'options' && $parsed !== false) {
// we prevent wrong backtracking on subtrees (too much array_pop())
// there is an array of sub-elements after(!) the base_expr clause of the current element
// so we go through the sub-elements and must not come back at the end
for ($i = 1; $i < count($parsed); $i++) {
$backtracking[] = false;
}
} elseif ($key === 'TABLE' || $key === 'create-def' && $parsed !== false) {
// do nothing
} else {
// move the current pos after the keyword
// SELECT, WHERE, INSERT etc.
if (PHPSQLParserConstants::getInstance()->isReserved($key)) {
$charPos = stripos($sql, $key, $charPos);
$charPos += strlen($key);
}
}
}
if (!is_array($parsed)) {
return;
}
foreach ($parsed as $key => $value) {
if ($key === 'base_expr') {
//$this->printPos("0", $sql, $charPos, $key, $value, $backtracking);
$subject = substr($sql, $charPos);
$pos = $this->findPositionWithinString($subject, $value, isset($parsed['expr_type']) ? $parsed['expr_type'] : 'alias');
if ($pos === false) {
throw new UnableToCalculatePositionException($value, $subject);
}
$parsed['position'] = $charPos + $pos;
$charPos += $pos + strlen($value);
//$this->printPos("1", $sql, $charPos, $key, $value, $backtracking);
$oldPos = array_pop($backtracking);
if (isset($oldPos) && $oldPos !== false) {
$charPos = $oldPos;
}
//$this->printPos("2", $sql, $charPos, $key, $value, $backtracking);
} else {
$this->lookForBaseExpression($sql, $charPos, $parsed[$key], $key, $backtracking);
}
}
}
public function process($tokens)
{
$resultList = array();
$category = "";
$prev = "";
foreach ($tokens as $k => $token) {
$upper = strtoupper(trim($token));
if ($this->isWhitespaceToken($token)) {
continue;
}
switch ($upper) {
case 'FROM':
$resultList[] = array('expr_type' => ExpressionType::RESERVED, 'base_expr' => trim($token));
if ($prev === 'INDEX' || $prev === 'COLUMNS') {
continue;
}
$category = $upper;
break;
case 'CREATE':
case 'DATABASE':
case 'FUNCTION':
case 'PROCEDURE':
case 'ENGINE':
case 'TABLE':
case 'FOR':
case 'LIKE':
case 'INDEX':
case 'COLUMNS':
case 'PLUGIN':
case 'PRIVILEGES':
case 'PROCESSLIST':
case 'LOGS':
case 'STATUS':
case 'GLOBAL':
case 'SESSION':
case 'FULL':
case 'GRANTS':
case 'INNODB':
case 'STORAGE':
case 'ENGINES':
case 'OPEN':
case 'BDB':
case 'TRIGGERS':
case 'VARIABLES':
case 'DATABASES':
case 'ERRORS':
case 'TABLES':
case 'WARNINGS':
case 'CHARACTER':
case 'SET':
case 'COLLATION':
$resultList[] = array('expr_type' => ExpressionType::RESERVED, 'base_expr' => trim($token));
$category = $upper;
break;
default:
switch ($prev) {
case 'LIKE':
$resultList[] = array('expr_type' => ExpressionType::CONSTANT, 'base_expr' => $token);
break;
case 'LIMIT':
$limit = array_pop($resultList);
$limit['sub_tree'] = $this->limitProcessor->process(array_slice($tokens, $k));
$resultList[] = $limit;
break;
case 'FROM':
case 'DATABASE':
$resultList[] = array('expr_type' => ExpressionType::DATABASE, 'name' => $token, 'no_quotes' => $this->revokeQuotation($token), 'base_expr' => $token);
break;
case 'FOR':
$resultList[] = array('expr_type' => ExpressionType::USER, 'name' => $token, 'no_quotes' => $this->revokeQuotation($token), 'base_expr' => $token);
break;
case 'INDEX':
case 'COLUMNS':
case 'TABLE':
$resultList[] = array('expr_type' => ExpressionType::TABLE, 'table' => $token, 'no_quotes' => $this->revokeQuotation($token), 'base_expr' => $token);
$category = "TABLENAME";
break;
case 'FUNCTION':
if (PHPSQLParserConstants::getInstance()->isAggregateFunction($upper)) {
$expr_type = ExpressionType::AGGREGATE_FUNCTION;
} else {
$expr_type = ExpressionType::SIMPLE_FUNCTION;
}
$resultList[] = array('expr_type' => $expr_type, 'name' => $token, 'no_quotes' => $this->revokeQuotation($token), 'base_expr' => $token);
break;
case 'PROCEDURE':
$resultList[] = array('expr_type' => ExpressionType::PROCEDURE, 'name' => $token, 'no_quotes' => $this->revokeQuotation($token), 'base_expr' => $token);
break;
case 'ENGINE':
$resultList[] = array('expr_type' => ExpressionType::ENGINE, 'name' => $token, 'no_quotes' => $this->revokeQuotation($token), 'base_expr' => $token);
break;
default:
// ignore
break;
}
break;
}
$prev = $category;
}
return $resultList;
}
public function process($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()) {
$processor = new DefaultProcessor();
$curr->setSubTree($processor->process($this->removeParenthesisFromStart($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($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 = $this->process($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() || $prev->isCustomFunction()) {
// if we have a colref followed by a parenthesis pair,
// it isn't a colref, it is a user-function
// TODO: this should be a method, because we need the same code
// below for unspecified tokens (expressions).
$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 = $this->process($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 = $this->process($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 ($prev->isColumnReference()) {
if (PHPSQLParserConstants::isCustomFunction($prev->getUpper())) {
$prev->setTokenType(ExpressionType::CUSTOM_FUNCTION);
} else {
$prev->setTokenType(ExpressionType::SIMPLE_FUNCTION);
}
$prev->setNoQuotes(null);
}
array_pop($resultList);
$curr = $prev;
}
// we have parenthesis, but it seems to be an expression
if ($curr->isUnspecified()) {
$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 = $this->process($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 = $this->process($tmpExprList);
$curr->setTokenType(ExpressionType::BRACKET_EXPRESSION);
if (!$curr->getSubTree()) {
$curr->setSubTree($localExprList);
} else {
$tmpExprList = $curr->getSubTree();
$curr->setSubTree(array_merge($tmpExprList, $localExprList));
}
}
} elseif ($curr->isVariableToken()) {
# a variable
# it can be quoted
$curr->setTokenType($this->getVariableType($curr->getUpper()));
$curr->setSubTree(false);
$curr->setNoQuotes(trim(trim($curr->getToken()), '@'), "`'\"");
} else {
/* it is either an operator, a colref or a constant */
switch ($curr->getUpper()) {
case '*':
$curr->setSubTree(false);
// o subtree
// single or first element of expression list -> all-column-alias
if (empty($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() && !$prev->isVariable()) {
$curr->setTokenType(ExpressionType::COLREF);
break;
}
if ($prev->isColumnReference() && $prev->endsWith(".")) {
$prev->addToken('*');
// tablealias dot *
continue 2;
// skip the current token
}
$curr->setTokenType(ExpressionType::OPERATOR);
break;
case ':=':
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() || $prev->isVariable() || $prev->isCustomFunction()) {
$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);
$curr->setNoQuotes($curr->getToken());
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);
$curr->setNoQuotes($curr->getToken());
}
break;
}
}
}
/* is a reserved word? */
if (!$curr->isOperator() && !$curr->isInList() && !$curr->isFunction() && !$curr->isAggregateFunction() && !$curr->isCustomFunction() && PHPSQLParserConstants::getInstance()->isReserved($curr->getUpper())) {
if (PHPSQLParserConstants::getInstance()->isCustomFunction($curr->getUpper())) {
$curr->setTokenType(ExpressionType::CUSTOM_FUNCTION);
$curr->setNoQuotes(null);
} elseif (PHPSQLParserConstants::getInstance()->isAggregateFunction($curr->getUpper())) {
$curr->setTokenType(ExpressionType::AGGREGATE_FUNCTION);
$curr->setNoQuotes(null);
} elseif ($curr->getUpper() === 'NULL') {
// it is a reserved word, but we would like to set it as constant
$curr->setTokenType(ExpressionType::CONSTANT);
} else {
if (PHPSQLParserConstants::getInstance()->isParameterizedFunction($curr->getUpper())) {
// issue 60: check functions with parameters
// -> colref (we check parameters later)
// -> if there is no parameter, we leave the colref
$curr->setTokenType(ExpressionType::COLREF);
} elseif (PHPSQLParserConstants::getInstance()->isFunction($curr->getUpper())) {
$curr->setTokenType(ExpressionType::SIMPLE_FUNCTION);
$curr->setNoQuotes(null);
} else {
$curr->setTokenType(ExpressionType::RESERVED);
$curr->setNoQuotes(null);
}
}
}
// issue 94, INTERVAL 1 MONTH
if ($curr->isConstant() && PHPSQLParserConstants::getInstance()->isParameterizedFunction($prev->getUpper())) {
$prev->setTokenType(ExpressionType::RESERVED);
$prev->setNoQuotes(null);
}
if ($prev->isConstant() && PHPSQLParserConstants::getInstance()->isParameterizedFunction($curr->getUpper())) {
$curr->setTokenType(ExpressionType::RESERVED);
$curr->setNoQuotes(null);
}
if ($curr->isUnspecified()) {
$curr->setTokenType(ExpressionType::EXPRESSION);
$curr->setNoQuotes(null);
$curr->setSubTree($this->process($this->splitSQLIntoTokens($curr->getTrim())));
}
$resultList[] = $curr;
$prev = $curr;
}
// end of for-loop
return $this->toArray($resultList);
}