diff options
Diffstat (limited to 'core/Segment.php')
| -rw-r--r-- | core/Segment.php | 124 |
1 files changed, 116 insertions, 8 deletions
diff --git a/core/Segment.php b/core/Segment.php index 8b46e4a811..8911d06247 100644 --- a/core/Segment.php +++ b/core/Segment.php @@ -1,6 +1,6 @@ <?php /** - * Piwik - free/libre analytics platform + * Matomo - free/libre analytics platform * * @link https://matomo.org * @license http://www.gnu.org/licenses/gpl-3.0.html GPL v3 or later @@ -76,6 +76,16 @@ class Segment protected $idSites = null; /** + * @var Date + */ + protected $startDate = null; + + /** + * @var Date + */ + protected $endDate = null; + + /** * @var LogQueryBuilder */ private $segmentQueryBuilder; @@ -93,12 +103,20 @@ class Segment /** * Constructor. * + * When using segments that contain a != or !@ condition on a non visit dimension (e.g. action, conversion, ...) it + * is needed to use a subquery to get correct results. To avoid subqueries that fetch too many data it's required to + * set a startDate and/or an endDate in this case. That date will be used to limit the subquery (along with possibly + * given idSites). If no startDate and endDate is given for such a segment it will generate a query that directly + * joins the according tables, but trigger a php warning as results might be incorrect. + * * @param string $segmentCondition The segment condition, eg, `'browserCode=ff;countryCode=CA'`. * @param array $idSites The list of sites the segment will be used with. Some segments are * dependent on the site, such as goal segments. + * @param Date|null $startDate start date used to limit subqueries + * @param Date|null $endDate end date used to limit subqueries * @throws */ - public function __construct($segmentCondition, $idSites) + public function __construct($segmentCondition, $idSites, Date $startDate = null, Date $endDate = null) { $this->segmentQueryBuilder = StaticContainer::get('Piwik\DataAccess\LogQueryBuilder'); @@ -111,6 +129,14 @@ class Segment $this->originalString = $segmentCondition; + if ($startDate instanceof Date) { + $this->startDate = $startDate; + } + + if ($endDate instanceof Date) { + $this->endDate = $endDate; + } + // The segment expression can be urlencoded. Unfortunately, both the encoded and decoded versions // can usually be parsed successfully. To pick the right one, we try both and pick the one w/ more // successfully parsed subexpressions. @@ -195,6 +221,12 @@ class Segment $string = substr($string, 0, self::SEGMENT_TRUNCATE_LIMIT); $this->string = $string; + + if (empty($idSites)) { + $idSites = []; + } else if (!is_array($idSites)) { + $idSites = [$idSites]; + } $this->idSites = $idSites; $segment = new SegmentExpression($string); $this->segmentExpression = $segment; @@ -209,8 +241,7 @@ class Segment $cleanedExpressions = array(); foreach ($expressions as $expression) { $operand = $expression[SegmentExpression::INDEX_OPERAND]; - $cleanedExpression = $this->getCleanedExpression($operand); - $expression[SegmentExpression::INDEX_OPERAND] = $cleanedExpression; + $expression[SegmentExpression::INDEX_OPERAND] = $this->getCleanedExpression($operand); $cleanedExpressions[] = $expression; } @@ -226,7 +257,8 @@ class Segment $availableSegment = $this->getSegmentByName($name); - if (!empty($availableSegment['unionOfSegments'])) { + // We leave segments using !@ and != operands untouched for segments not on log_visit table as they will be build using a subquery + if (!$this->doesSegmentNeedSubquery($operand[SegmentExpression::INDEX_OPERAND_OPERATOR], $name) && !empty($availableSegment['unionOfSegments'])) { $count = 0; foreach ($availableSegment['unionOfSegments'] as $segmentNameOfUnion) { $count++; @@ -252,6 +284,51 @@ class Segment return $expressionsWithUnions; } + private function isVisitSegment($name) + { + $availableSegment = $this->getSegmentByName($name); + + if (!empty($availableSegment['unionOfSegments'])) { + foreach ($availableSegment['unionOfSegments'] as $segmentNameOfUnion) { + $unionSegment = $this->getSegmentByName($segmentNameOfUnion); + if (strpos($unionSegment['sqlSegment'], 'log_visit.') === 0) { + return true; + } + } + } else if (strpos($availableSegment['sqlSegment'], 'log_visit.') === 0) { + return true; + } + + return false; + } + + private function doesSegmentNeedSubquery($operator, $segmentName) + { + $requiresSubQuery = in_array($operator, [ + SegmentExpression::MATCH_DOES_NOT_CONTAIN, + SegmentExpression::MATCH_NOT_EQUAL + ]) && !$this->isVisitSegment($segmentName); + + if ($requiresSubQuery && empty($this->startDate) && empty($this->endDate)) { + $e = new Exception(); + Log::warning("Avoiding segment subquery due to missing start date and/or an end date. Please ensure a start date and/or end date is set when initializing a segment if it's used to build a query. Stacktrace:\n" . $e->getTraceAsString()); + return false; + } + + return $requiresSubQuery; + } + + private function getInvertedOperatorForSubQuery($operator) + { + if ($operator === SegmentExpression::MATCH_DOES_NOT_CONTAIN) { + return SegmentExpression::MATCH_CONTAINS; + } else if ($operator === SegmentExpression::MATCH_NOT_EQUAL) { + return SegmentExpression::MATCH_EQUAL; + } + + throw new Exception("Operator not support for subqueries"); + } + /** * Returns `true` if the segment is empty, `false` if otherwise. */ @@ -277,9 +354,6 @@ class Segment } $idSites = $this->idSites; - if (!is_array($idSites)) { - $idSites = array($this->idSites); - } return Rules::isRequestAuthorizedToArchive() || Rules::isBrowserArchivingAvailableForSegments() @@ -297,6 +371,40 @@ class Segment $segment = $this->getSegmentByName($name); $sqlName = $segment['sqlSegment']; + // Build subqueries for segments that are not on log_visit table but use !@ or != as operator + // This is required to ensure segments like actionUrl!@value really do not include any visit having an action containing `value` + if ($this->doesSegmentNeedSubquery($matchType, $name)) { + $operator = $this->getInvertedOperatorForSubQuery($matchType); + $stringSegment = $name . $operator . $value; + $segmentObj = new Segment($stringSegment, $this->idSites, $this->startDate, $this->endDate); + + $select = 'log_visit.idvisit'; + $from = 'log_visit'; + $datetimeField = 'visit_last_action_time'; + $where = []; + $bind = []; + if (!empty($this->idSites)) { + $where[] = "$from.idsite IN (" . Common::getSqlStringFieldsArray($this->idSites) . ")"; + $bind = $this->idSites; + } + if ($this->startDate instanceof Date) { + $where[] = "$from.$datetimeField >= ?"; + $bind[] = $this->startDate->toString(Date::DATE_TIME_FORMAT); + } + if ($this->endDate instanceof Date) { + $where[] = "$from.$datetimeField <= ?"; + $bind[] = $this->endDate->toString(Date::DATE_TIME_FORMAT); + } + + $logQueryBuilder = StaticContainer::get('Piwik\DataAccess\LogQueryBuilder'); + $forceGroupByBackup = $logQueryBuilder->getForcedInnerGroupBySubselect(); + $logQueryBuilder->forceInnerGroupBySubselect(LogQueryBuilder::FORCE_INNER_GROUP_BY_NO_SUBSELECT); + $query = $segmentObj->getSelectQuery($select, $from, implode(' AND ', $where), $bind); + $logQueryBuilder->forceInnerGroupBySubselect($forceGroupByBackup); + + return ['log_visit.idvisit', SegmentExpression::MATCH_ACTIONS_NOT_CONTAINS, $query]; + } + if ($matchType != SegmentExpression::MATCH_IS_NOT_NULL_NOR_EMPTY && $matchType != SegmentExpression::MATCH_IS_NULL_OR_EMPTY) { |