| <?php |
|
|
| |
| |
| |
| |
| |
| |
|
|
| namespace Piwik\Segment; |
|
|
| use Exception; |
|
|
| class SegmentExpression |
| { |
| public const AND_DELIMITER = ';'; |
| public const OR_DELIMITER = ','; |
|
|
| public const MATCH_EQUAL = '=='; |
| public const MATCH_NOT_EQUAL = '!='; |
| public const MATCH_GREATER_OR_EQUAL = '>='; |
| public const MATCH_LESS_OR_EQUAL = '<='; |
| public const MATCH_GREATER = '>'; |
| public const MATCH_LESS = '<'; |
| public const MATCH_CONTAINS = '=@'; |
| public const MATCH_DOES_NOT_CONTAIN = '!@'; |
| public const MATCH_STARTS_WITH = '=^'; |
| public const MATCH_ENDS_WITH = '=$'; |
|
|
| public const BOOL_OPERATOR_OR = 'OR'; |
| public const BOOL_OPERATOR_AND = 'AND'; |
| public const BOOL_OPERATOR_END = ''; |
|
|
| |
| |
| |
| public const MATCH_IS_NOT_NULL_NOR_EMPTY = '::NOT_NULL'; |
| public const MATCH_IS_NULL_OR_EMPTY = '::NULL'; |
|
|
| |
| public const MATCH_ACTIONS_CONTAINS = 'IN'; |
| public const MATCH_ACTIONS_NOT_CONTAINS = 'NOTIN'; |
|
|
| |
| |
| |
| |
| public const MATCH_IDVISIT_NOT_IN = 'IDVISIT_NOTIN'; |
|
|
| public const INDEX_OPERAND_NAME = 0; |
| public const INDEX_OPERAND_OPERATOR = 1; |
| public const INDEX_OPERAND_VALUE = 2; |
| public const INDEX_OPERAND_JOIN_COLUMN = 3; |
| public const INDEX_OPERAND_SEGMENT_INFO = 4; |
|
|
| public const SQL_WHERE_DO_NOT_MATCH_ANY_ROW = "(1 = 0)"; |
| public const SQL_WHERE_MATCHES_ALL_ROWS = "(1 = 1)"; |
|
|
| protected $string; |
| protected $valuesBind = []; |
| protected $tree = []; |
| protected $parsedSubExpressions = []; |
|
|
| public function __construct($string) |
| { |
| $this->string = $string; |
| $this->tree = $this->parseTree(); |
| } |
|
|
| public function getSegmentDefinition() |
| { |
| return $this->string; |
| } |
|
|
| public function isEmpty() |
| { |
| return count($this->tree) == 0; |
| } |
|
|
|
|
| public function getSubExpressionCount() |
| { |
| $count = 0; |
| foreach ($this->parsedSubExpressions as $orExpressions) { |
| foreach ($orExpressions as $operand) { |
| $isExpressionColumnPresent = !empty($operand[self::INDEX_OPERAND_NAME]); |
| if ($isExpressionColumnPresent) { |
| ++$count; |
| } |
| } |
| } |
| return $count; |
| } |
|
|
| |
| |
| |
| |
| |
| |
| |
| |
| public function parseSubExpressions() |
| { |
| $parsedSubExpressions = array_map(function (array $orExpressions) { |
| return array_map(function (string $operand) { |
| return $this->parseOperand($operand); |
| }, $orExpressions); |
| }, $this->tree); |
| $this->parsedSubExpressions = $parsedSubExpressions; |
|
|
| return $parsedSubExpressions; |
| } |
|
|
| private function parseOperand(string $operand): array |
| { |
| $operand = urldecode($operand); |
|
|
| $pattern = '/^(.+?)(' . self::MATCH_EQUAL . '|' |
| . self::MATCH_NOT_EQUAL . '|' |
| . self::MATCH_GREATER_OR_EQUAL . '|' |
| . self::MATCH_GREATER . '|' |
| . self::MATCH_LESS_OR_EQUAL . '|' |
| . self::MATCH_LESS . '|' |
| . self::MATCH_CONTAINS . '|' |
| . self::MATCH_DOES_NOT_CONTAIN . '|' |
| . preg_quote(self::MATCH_STARTS_WITH) . '|' |
| . preg_quote(self::MATCH_ENDS_WITH) |
| . '){1}(.*)/'; |
| $match = preg_match($pattern, $operand, $matches); |
| if ($match == 0) { |
| throw new Exception('The segment condition \'' . $operand . '\' is not valid.'); |
| } |
|
|
| $leftMember = $matches[1]; |
| $operation = $matches[2]; |
| $valueRightMember = urldecode($matches[3]); |
|
|
| |
| if ($valueRightMember === '') { |
| if ($operation == self::MATCH_NOT_EQUAL) { |
| $operation = self::MATCH_IS_NOT_NULL_NOR_EMPTY; |
| } elseif ($operation == self::MATCH_EQUAL) { |
| $operation = self::MATCH_IS_NULL_OR_EMPTY; |
| } else { |
| throw new Exception('The segment \'' . $operand . '\' has no value specified. You can leave this value empty ' . |
| 'only when you use the operators: ' . self::MATCH_NOT_EQUAL . ' (is not) or ' . self::MATCH_EQUAL . ' (is)'); |
| } |
| } |
|
|
| return [ |
| self::INDEX_OPERAND_NAME => $leftMember, |
| self::INDEX_OPERAND_OPERATOR => $operation, |
| self::INDEX_OPERAND_VALUE => $valueRightMember, |
| ]; |
| } |
|
|
| |
| |
| |
| |
| public function setSubExpressionsAfterCleanup($parsedSubExpressions) |
| { |
| $this->parsedSubExpressions = $parsedSubExpressions; |
| } |
|
|
| |
| |
| |
| public function parseSubExpressionsIntoSqlExpressions(&$availableTables = array()) |
| { |
| $this->valuesBind = []; |
|
|
| $sqlSubExpressions = array_map(function (array $orExpressions) use (&$availableTables) { |
| return array_map(function (array $operandDefinition) use (&$availableTables) { |
| $operand = $this->getSqlMatchFromDefinition($operandDefinition, $availableTables); |
|
|
| if ($operand[self::INDEX_OPERAND_OPERATOR] !== null) { |
| if (is_array($operand[self::INDEX_OPERAND_OPERATOR])) { |
| $this->valuesBind = array_merge($this->valuesBind, $operand[self::INDEX_OPERAND_OPERATOR]); |
| } else { |
| $this->valuesBind[] = $operand[self::INDEX_OPERAND_OPERATOR]; |
| } |
| } |
|
|
| $operand = $operand[self::INDEX_OPERAND_NAME]; |
| return $operand; |
| }, $orExpressions); |
| }, $this->parsedSubExpressions); |
|
|
| $this->tree = $sqlSubExpressions; |
| } |
|
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| protected function getSqlMatchFromDefinition($def, &$availableTables) |
| { |
| $field = $def[self::INDEX_OPERAND_NAME]; |
| $matchType = $def[self::INDEX_OPERAND_OPERATOR]; |
| $value = $def[self::INDEX_OPERAND_VALUE]; |
| $join = $def[self::INDEX_OPERAND_JOIN_COLUMN] ?? null; |
| $segment = $def[self::INDEX_OPERAND_SEGMENT_INFO] ?? null; |
|
|
| |
| |
| $innerSql = null; |
|
|
| |
| $operandWillNotMatchAnyRow = empty($field) && is_null($value); |
| if ($operandWillNotMatchAnyRow) { |
| if ($matchType == self::MATCH_EQUAL) { |
| |
| |
| $sqlExpression = self::SQL_WHERE_DO_NOT_MATCH_ANY_ROW; |
| } elseif ($matchType == self::MATCH_NOT_EQUAL) { |
| |
| |
| $sqlExpression = self::SQL_WHERE_MATCHES_ALL_ROWS; |
| } elseif ( |
| $matchType == self::MATCH_CONTAINS |
| || $matchType == self::MATCH_DOES_NOT_CONTAIN |
| || $matchType == self::MATCH_STARTS_WITH |
| || $matchType == self::MATCH_ENDS_WITH |
| ) { |
| |
| |
| |
| $sqlExpression = self::SQL_WHERE_DO_NOT_MATCH_ANY_ROW; |
| } else { |
| |
| throw new Exception("Unexpected match type $matchType for your segment. " . |
| "Please report this issue to the Matomo team with the segment you are using."); |
| } |
|
|
| return array($sqlExpression, $value = null); |
| } |
|
|
| $alsoMatchNULLValues = false; |
| switch ($matchType) { |
| case self::MATCH_EQUAL: |
| $sqlMatch = '%s ='; |
| break; |
| case self::MATCH_NOT_EQUAL: |
| $sqlMatch = '%s <>'; |
| $alsoMatchNULLValues = true; |
| break; |
| case self::MATCH_GREATER: |
| $sqlMatch = '%s >'; |
| break; |
| case self::MATCH_LESS: |
| $sqlMatch = '%s <'; |
| break; |
| case self::MATCH_GREATER_OR_EQUAL: |
| $sqlMatch = '%s >='; |
| break; |
| case self::MATCH_LESS_OR_EQUAL: |
| $sqlMatch = '%s <='; |
| break; |
| case self::MATCH_CONTAINS: |
| $sqlMatch = '%s LIKE'; |
| $value = '%' . $this->escapeLikeString($value) . '%'; |
| break; |
| case self::MATCH_DOES_NOT_CONTAIN: |
| $sqlMatch = '%s NOT LIKE'; |
| $value = '%' . $this->escapeLikeString($value) . '%'; |
| $alsoMatchNULLValues = true; |
| break; |
| case self::MATCH_STARTS_WITH: |
| $sqlMatch = '%s LIKE'; |
| $value = $this->escapeLikeString($value) . '%'; |
| break; |
| case self::MATCH_ENDS_WITH: |
| $sqlMatch = '%s LIKE'; |
| $value = '%' . $this->escapeLikeString($value); |
| break; |
|
|
| case self::MATCH_IS_NOT_NULL_NOR_EMPTY: |
| $sqlMatch = '%s IS NOT NULL AND %s <> \'\' AND %s <> \'0\''; |
| $value = null; |
| break; |
|
|
| case self::MATCH_IS_NULL_OR_EMPTY: |
| $sqlMatch = '%s IS NULL OR %s = \'\' OR %s = \'0\''; |
| $value = null; |
| break; |
|
|
| case self::MATCH_ACTIONS_CONTAINS: |
| |
| |
| |
| |
| $innerSql = $value['SQL']; |
| $sqlMatch = '%s IN (%innerSql%)'; |
| $value = $value['bind']; |
| break; |
| case self::MATCH_ACTIONS_NOT_CONTAINS: |
| |
| |
| |
| |
| $innerSql = $value['sql']; |
| $sqlMatch = '%s NOT IN (%innerSql%)'; |
| $value = $value['bind']; |
| break; |
| default: |
| throw new Exception("Filter contains the match type '" . $matchType . "' which is not supported"); |
| break; |
| } |
|
|
| |
| $alsoMatchNULLValues = $alsoMatchNULLValues && !empty($value); |
|
|
| if ( |
| $matchType === self::MATCH_ACTIONS_CONTAINS || $matchType === self::MATCH_ACTIONS_NOT_CONTAINS |
| || is_null($value) |
| ) { |
| $sqlExpression = "( $sqlMatch )"; |
| } else { |
| if ($alsoMatchNULLValues) { |
| $sqlExpression = "( $field IS NULL OR $sqlMatch ? )"; |
| } else { |
| $sqlExpression = "$sqlMatch ?"; |
| } |
| } |
|
|
| if (!empty($join['field'])) { |
| $sqlExpression = str_replace('%s', $join['field'], $sqlExpression); |
| } else { |
| $sqlExpression = str_replace('%s', $field, $sqlExpression); |
| } |
|
|
| if (!empty($join['discriminator'])) { |
| $sqlExpression = '(' . $sqlExpression . ' AND ' . $join['discriminator'] . ')'; |
| } |
|
|
| $columns = self::parseColumnsFromSqlExpr($sqlExpression); |
| foreach ($columns as $column) { |
| $this->checkFieldIsAvailable($column, $availableTables, $join); |
| } |
|
|
| if (!empty($join['field'])) { |
| $this->checkFieldIsAvailable($join['field'], $availableTables, $join); |
| } |
|
|
| if (!empty($join['joinOn'])) { |
| $joinOnColumns = self::parseColumnsFromSqlExpr($join['joinOn']); |
| foreach ($joinOnColumns as $column) { |
| $this->checkFieldIsAvailable($column, $availableTables, $join); |
| } |
| } |
|
|
| if ($innerSql) { |
| $sqlExpression = str_replace('%innerSql%', $innerSql, $sqlExpression); |
| } |
|
|
| return array($sqlExpression, $value); |
| } |
|
|
| |
| |
| |
| |
| public static function parseColumnsFromSqlExpr($field) |
| { |
| preg_match_all('/[^@a-zA-Z0-9_]?`?([@a-zA-Z_][@a-zA-Z0-9_]*`?\.`?[a-zA-Z0-9_`]+)`?\b/', $field, $matches); |
| $result = isset($matches[1]) ? $matches[1] : []; |
| |
| $result = array_filter($result, function ($value) { |
| return strpos($value, '@') === false; |
| }); |
| $result = array_map(function ($item) { |
| return str_replace('`', '', $item); |
| }, $result); |
| $result = array_unique($result); |
| $result = array_values($result); |
| return $result; |
| } |
|
|
| |
| |
| |
| |
| |
| |
| |
| private function checkFieldIsAvailable($field, &$availableTables, $join) |
| { |
| $fieldParts = explode('.', $field); |
|
|
| $table = count($fieldParts) == 2 ? $fieldParts[0] : false; |
|
|
| |
| |
| $table = preg_replace('/^[A-Z_]+\(/', '', $table); |
|
|
| $tableExists = !$table || in_array($table, $availableTables); |
|
|
| if ($tableExists) { |
| return; |
| } |
|
|
| if (is_array($availableTables)) { |
| foreach ($availableTables as $availableTable) { |
| if (is_array($availableTable)) { |
| if (!isset($availableTable['tableAlias']) && $availableTable['table'] === $table) { |
| return; |
| } elseif (isset($availableTable['tableAlias']) && $availableTable['tableAlias'] === $table) { |
| return; |
| } |
| } |
| } |
| } |
|
|
| if ( |
| $join |
| && ((empty($join['tableAlias']) && $table == $join['table']) |
| || $table == $join['tableAlias'] |
| ) |
| ) { |
| $availableTables[] = $join; |
| } else { |
| $availableTables[] = $table; |
| } |
| } |
|
|
| |
| |
| |
| |
| |
| private function escapeLikeString($str) |
| { |
| if (false !== strpos($str, '%')) { |
| $str = str_replace("%", "\%", $str); |
| } |
|
|
| if (false !== strpos($str, '_')) { |
| $str = str_replace("_", "\_", $str); |
| } |
|
|
| return $str; |
| } |
|
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| protected function parseTree() |
| { |
| $segmentStr = trim($this->string); |
| if (empty($segmentStr)) { |
| return []; |
| } |
|
|
| |
| $ands = array_filter(preg_split('/' . self::AND_DELIMITER . '(?!$)(?<!\\\\' . self::AND_DELIMITER . ')/', $segmentStr)); |
| return array_map(function ($and) { |
| |
| $ors = preg_split('/' . self::OR_DELIMITER . '(?!$)(?<!\\\\' . self::OR_DELIMITER . ')/', $and); |
|
|
| |
| return array_map(function ($or) { |
| return str_replace( |
| ['\\' . self::AND_DELIMITER, '\\' . self::OR_DELIMITER], |
| [self::AND_DELIMITER, self::OR_DELIMITER], |
| $or |
| ); |
| }, $ors); |
| }, $ands); |
| } |
|
|
| |
| |
| |
| |
| |
| |
| |
| public function getSql() |
| { |
| if ($this->isEmpty()) { |
| throw new Exception("Invalid segment, please specify a valid segment."); |
| } |
|
|
| $andExpressions = $this->tree; |
| $andExpressions = array_map(function ($orExpressions) { |
| if (count($orExpressions) == 1) { |
| return $orExpressions[0]; |
| } |
|
|
| return '( ' . implode(' OR ', $orExpressions) . ')'; |
| }, $andExpressions); |
|
|
| $sql = implode(' AND ', $andExpressions); |
|
|
| return array( |
| 'where' => $sql, |
| 'bind' => $this->valuesBind, |
| ); |
| } |
| } |
|
|