API
3 months ago
Access
3 months ago
Application
3 months ago
Archive
3 months ago
ArchiveProcessor
3 months ago
Archiver
2 years ago
AssetManager
3 months ago
Auth
6 months ago
Category
6 months ago
Changes
3 months ago
CliMulti
1 year ago
Columns
3 months ago
Concurrency
3 months ago
Config
3 months ago
Container
3 months ago
CronArchive
3 months ago
DataAccess
3 months ago
DataFiles
2 years ago
DataTable
3 months ago
Db
3 months ago
DeviceDetector
1 year ago
Email
2 years ago
Exception
4 months ago
Http
4 months ago
Intl
3 months ago
Log
2 years ago
Mail
1 year ago
Measurable
6 months ago
Menu
3 months ago
Metrics
3 months ago
Notification
6 months ago
Period
3 months ago
Plugin
3 months ago
Policy
3 months ago
ProfessionalServices
1 year ago
Report
1 year ago
ReportRenderer
3 months ago
Request
3 months ago
Scheduler
3 months ago
Segment
3 months ago
Session
3 months ago
Settings
3 months ago
Tracker
3 months ago
Translation
3 months ago
Twig
1 year ago
UpdateCheck
3 months ago
Updater
4 months ago
Updates
3 months ago
Validators
1 year ago
View
6 months ago
ViewDataTable
3 months ago
Visualization
1 year ago
Widget
3 months ago
.htaccess
2 years ago
Access.php
3 months ago
Archive.php
3 months ago
ArchiveProcessor.php
4 months ago
AssetManager.php
3 months ago
Auth.php
6 months ago
AuthResult.php
6 months ago
BaseFactory.php
2 years ago
Cache.php
2 years ago
CacheId.php
4 months ago
CliMulti.php
3 months ago
Common.php
3 months ago
Config.php
3 months ago
Console.php
3 months ago
Context.php
2 years ago
Cookie.php
1 year ago
CronArchive.php
3 months ago
DI.php
3 months ago
DataArray.php
5 months ago
DataTable.php
3 months ago
Date.php
3 months ago
Db.php
3 months ago
DbHelper.php
3 months ago
Development.php
1 year ago
ErrorHandler.php
6 months ago
EventDispatcher.php
1 year ago
ExceptionHandler.php
4 months ago
FileIntegrity.php
3 months ago
Filechecks.php
1 year ago
Filesystem.php
3 months ago
FrontController.php
4 months ago
Http.php
4 months ago
IP.php
1 year ago
Log.php
3 months ago
LogDeleter.php
1 year ago
Mail.php
1 year ago
Metrics.php
3 months ago
NoAccessException.php
2 years ago
Nonce.php
6 months ago
Notification.php
6 months ago
NumberFormatter.php
5 months ago
Option.php
5 months ago
Period.php
3 months ago
Piwik.php
3 months ago
Plugin.php
3 months ago
Process.php
1 year ago
Profiler.php
6 months ago
ProxyHeaders.php
4 months ago
ProxyHttp.php
5 months ago
QuickForm2.php
3 months ago
RankingQuery.php
5 months ago
ReportRenderer.php
3 months ago
Request.php
3 months ago
Segment.php
3 months ago
Sequence.php
6 months ago
Session.php
3 months ago
SettingsPiwik.php
3 months ago
SettingsServer.php
1 year ago
Singleton.php
2 years ago
Site.php
4 months ago
SiteContentDetector.php
3 months ago
SupportedBrowser.php
2 years ago
TCPDF.php
1 year ago
Theme.php
1 year ago
Timer.php
2 years ago
Tracker.php
3 months ago
Twig.php
3 months ago
Unzip.php
1 year ago
UpdateCheck.php
3 months ago
Updater.php
3 months ago
UpdaterErrorException.php
2 years ago
Updates.php
3 months ago
Url.php
3 months ago
UrlHelper.php
3 months ago
Version.php
3 months ago
View.php
3 months ago
bootstrap.php
1 year ago
dispatch.php
2 years ago
testMinimumPhpVersion.php
6 months ago
Segment.php
641 lines
| 1 | <?php |
| 2 | |
| 3 | /** |
| 4 | * Matomo - free/libre analytics platform |
| 5 | * |
| 6 | * @link https://matomo.org |
| 7 | * @license https://www.gnu.org/licenses/gpl-3.0.html GPL v3 or later |
| 8 | */ |
| 9 | namespace Piwik; |
| 10 | |
| 11 | use Exception; |
| 12 | use Piwik\API\Request; |
| 13 | use Piwik\ArchiveProcessor\Rules; |
| 14 | use Piwik\Cache as PiwikCache; |
| 15 | use Piwik\Container\StaticContainer; |
| 16 | use Piwik\DataAccess\LogQueryBuilder; |
| 17 | use Piwik\Plugins\SegmentEditor\SegmentEditor; |
| 18 | use Piwik\Segment\SegmentExpression; |
| 19 | use Piwik\Plugins\SegmentEditor\Model as SegmentEditorModel; |
| 20 | use Piwik\Segment\SegmentsList; |
| 21 | /** |
| 22 | * Limits the set of visits Piwik uses when aggregating analytics data. |
| 23 | * |
| 24 | * A segment is a condition used to filter visits. They can, for example, |
| 25 | * select visits that have a specific browser or come from a specific |
| 26 | * country, or both. |
| 27 | * |
| 28 | * Plugins that aggregate data stored in Piwik can support segments by |
| 29 | * using this class when generating aggregation SQL queries. |
| 30 | * |
| 31 | * ### Examples |
| 32 | * |
| 33 | * **Basic usage** |
| 34 | * |
| 35 | * $idSites = array(1,2,3); |
| 36 | * $segmentStr = "browserCode==ff;countryCode==CA"; |
| 37 | * $segment = new Segment($segmentStr, $idSites); |
| 38 | * |
| 39 | * $query = $segment->getSelectQuery( |
| 40 | * $select = "table.col1, table2.col2", |
| 41 | * $from = array("table", "table2"), |
| 42 | * $where = "table.col3 = ?", |
| 43 | * $bind = array(5), |
| 44 | * $orderBy = "table.col1 DESC", |
| 45 | * $groupBy = "table2.col2" |
| 46 | * ); |
| 47 | * |
| 48 | * Db::fetchAll($query['sql'], $query['bind']); |
| 49 | * |
| 50 | * **Creating a _null_ segment** |
| 51 | * |
| 52 | * $idSites = array(1,2,3); |
| 53 | * $segment = new Segment('', $idSites); |
| 54 | * // $segment->getSelectQuery will return a query that selects all visits |
| 55 | * |
| 56 | * @api |
| 57 | */ |
| 58 | class Segment |
| 59 | { |
| 60 | /** |
| 61 | * @var SegmentExpression |
| 62 | */ |
| 63 | protected $segmentExpression = null; |
| 64 | /** |
| 65 | * @var string |
| 66 | */ |
| 67 | protected $string = null; |
| 68 | /** |
| 69 | * @var string |
| 70 | */ |
| 71 | protected $originalString = null; |
| 72 | /** |
| 73 | * @var array |
| 74 | */ |
| 75 | protected $idSites = null; |
| 76 | /** |
| 77 | * @var Date |
| 78 | */ |
| 79 | protected $startDate = null; |
| 80 | /** |
| 81 | * @var Date |
| 82 | */ |
| 83 | protected $endDate = null; |
| 84 | /** |
| 85 | * @var LogQueryBuilder |
| 86 | */ |
| 87 | private $segmentQueryBuilder; |
| 88 | /** |
| 89 | * @var bool |
| 90 | */ |
| 91 | private $isSegmentEncoded; |
| 92 | /** |
| 93 | * @var Exception|null |
| 94 | */ |
| 95 | private $missingDatesException = null; |
| 96 | /** |
| 97 | * Truncate the Segments to 8k |
| 98 | */ |
| 99 | public const SEGMENT_TRUNCATE_LIMIT = 8192; |
| 100 | public const CACHE_KEY = 'segmenthashes'; |
| 101 | public const SEGMENT_HAS_BUILT_CACHE_KEY = 'segmenthashbuilt'; |
| 102 | /** |
| 103 | * Constructor. |
| 104 | * |
| 105 | * When using segments that contain a != or !@ condition on a non visit dimension (e.g. action, conversion, ...) it |
| 106 | * is needed to use a subquery to get correct results. To avoid subqueries that fetch too many data it's required to |
| 107 | * set a startDate and/or an endDate in this case. That date will be used to limit the subquery (along with possibly |
| 108 | * given idSites). If no startDate and endDate is given for such a segment it will generate a query that directly |
| 109 | * joins the according tables, but trigger a php warning as results might be incorrect. |
| 110 | * |
| 111 | * @param string $segmentCondition The segment condition, eg, `'browserCode=ff;countryCode=CA'`. |
| 112 | * @param array $idSites The list of sites the segment will be used with. Some segments are |
| 113 | * dependent on the site, such as goal segments. |
| 114 | * @param Date|null $startDate start date used to limit subqueries |
| 115 | * @param Date|null $endDate end date used to limit subqueries |
| 116 | * @throws |
| 117 | */ |
| 118 | public function __construct($segmentCondition, $idSites, ?\Piwik\Date $startDate = null, ?\Piwik\Date $endDate = null) |
| 119 | { |
| 120 | $this->segmentQueryBuilder = StaticContainer::get('Piwik\\DataAccess\\LogQueryBuilder'); |
| 121 | $segmentCondition = trim($segmentCondition ?: ''); |
| 122 | if (!\Piwik\SettingsPiwik::isSegmentationEnabled() && !empty($segmentCondition)) { |
| 123 | throw new Exception("The Super User has disabled the Segmentation feature."); |
| 124 | } |
| 125 | $this->originalString = $segmentCondition; |
| 126 | if ($startDate instanceof \Piwik\Date) { |
| 127 | $this->startDate = $startDate; |
| 128 | } |
| 129 | if ($endDate instanceof \Piwik\Date) { |
| 130 | $this->endDate = $endDate; |
| 131 | } |
| 132 | // The segment expression can be urlencoded. Unfortunately, both the encoded and decoded versions |
| 133 | // can usually be parsed successfully. To pick the right one, we try both and pick the one w/ more |
| 134 | // successfully parsed subexpressions. |
| 135 | $subexpressionsDecoded = 0; |
| 136 | if (urldecode($segmentCondition) !== $segmentCondition) { |
| 137 | try { |
| 138 | $this->initializeSegment(urldecode($segmentCondition), $idSites); |
| 139 | $subexpressionsDecoded = $this->segmentExpression->getSubExpressionCount(); |
| 140 | } catch (Exception $e) { |
| 141 | // ignore |
| 142 | } |
| 143 | } |
| 144 | $subexpressionsRaw = 0; |
| 145 | try { |
| 146 | $this->initializeSegment($segmentCondition, $idSites); |
| 147 | $subexpressionsRaw = $this->segmentExpression->getSubExpressionCount(); |
| 148 | } catch (Exception $e) { |
| 149 | // ignore |
| 150 | } |
| 151 | if ($subexpressionsRaw > $subexpressionsDecoded) { |
| 152 | // segment initialized above |
| 153 | $this->isSegmentEncoded = \false; |
| 154 | } else { |
| 155 | $this->initializeSegment(urldecode($segmentCondition), $idSites); |
| 156 | $this->isSegmentEncoded = \true; |
| 157 | } |
| 158 | } |
| 159 | /** |
| 160 | * Checks if the provided segmentCondition is valid and available for the given idSites |
| 161 | * |
| 162 | * @params array $idSites |
| 163 | * @api since Matomo 5.3.0 |
| 164 | */ |
| 165 | public static function isAvailable(string $segmentCondition, array $idSites) : bool |
| 166 | { |
| 167 | try { |
| 168 | new self($segmentCondition, $idSites); |
| 169 | } catch (Exception $e) { |
| 170 | return \false; |
| 171 | } |
| 172 | return \true; |
| 173 | } |
| 174 | /** |
| 175 | * Returns the segment expression. |
| 176 | * @return SegmentExpression |
| 177 | * @api since Piwik 3.2.0 |
| 178 | */ |
| 179 | public function getSegmentExpression() |
| 180 | { |
| 181 | return $this->segmentExpression; |
| 182 | } |
| 183 | /** |
| 184 | * @throws Exception |
| 185 | */ |
| 186 | private function getAvailableSegments() |
| 187 | { |
| 188 | // start cache |
| 189 | $cache = PiwikCache::getTransientCache(); |
| 190 | //covert cache id |
| 191 | $cacheId = 'API.getSegmentsMetadata.' . \Piwik\SettingsPiwik::getPiwikInstanceId() . '.' . implode(",", $this->idSites); |
| 192 | //fetch cache lockId |
| 193 | $availableSegments = $cache->fetch($cacheId); |
| 194 | // segment metadata |
| 195 | if (empty($availableSegments)) { |
| 196 | $availableSegments = Request::processRequest('API.getSegmentsMetadata', array('idSites' => $this->idSites, '_hideImplementationData' => 0, 'filter_limit' => -1, 'filter_offset' => 0, '_showAllSegments' => 1), []); |
| 197 | // index by segment name |
| 198 | $availableSegments = array_column($availableSegments, null, 'segment'); |
| 199 | // remove segments we don't have permission to use |
| 200 | foreach ($availableSegments as $segment => $segmentInfo) { |
| 201 | if (isset($segmentInfo['permission']) && $segmentInfo['permission'] != 1) { |
| 202 | $availableSegments[$segment] = null; |
| 203 | } |
| 204 | } |
| 205 | $cache->save($cacheId, $availableSegments); |
| 206 | } |
| 207 | return $availableSegments; |
| 208 | } |
| 209 | private function getSegmentByName($name) |
| 210 | { |
| 211 | $segments = $this->getAvailableSegments(); |
| 212 | if (array_key_exists($name, $segments)) { |
| 213 | if ($segments[$name] === null) { |
| 214 | throw new \Piwik\NoAccessException("You do not have enough permission to access the segment " . $name); |
| 215 | } |
| 216 | return $segments[$name]; |
| 217 | } |
| 218 | throw new Exception("Segment '{$name}' is not a supported segment."); |
| 219 | } |
| 220 | /** |
| 221 | * @param $string |
| 222 | * @param $idSites |
| 223 | * @throws Exception |
| 224 | */ |
| 225 | protected function initializeSegment($string, $idSites) |
| 226 | { |
| 227 | // As a preventive measure, we restrict the filter size to a safe limit |
| 228 | $string = substr($string, 0, self::SEGMENT_TRUNCATE_LIMIT); |
| 229 | $this->string = $string; |
| 230 | if (empty($idSites)) { |
| 231 | $idSites = []; |
| 232 | } elseif (!is_array($idSites)) { |
| 233 | $idSites = [$idSites]; |
| 234 | } |
| 235 | $this->idSites = $idSites; |
| 236 | $segment = new SegmentExpression($string); |
| 237 | $this->segmentExpression = $segment; |
| 238 | // parse segments |
| 239 | $expressions = $segment->parseSubExpressions(); |
| 240 | $expressions = $this->getExpressionsWithUnionsResolved($expressions); |
| 241 | $expressions = $this->mergeSubqueryExpressionsInTree($expressions); |
| 242 | // convert segments name to sql segment |
| 243 | // check that user is allowed to view this segment |
| 244 | // and apply a filter to the value to match if necessary (to map DB fields format) |
| 245 | $cleanedExpressions = array_map(function (array $orExpressions) { |
| 246 | return array_map(function (array $operand) { |
| 247 | return $this->getCleanedExpression($operand); |
| 248 | }, $orExpressions); |
| 249 | }, $expressions); |
| 250 | $segment->setSubExpressionsAfterCleanup($cleanedExpressions); |
| 251 | } |
| 252 | private function getExpressionsWithUnionsResolved(array $expressions) : array |
| 253 | { |
| 254 | $expressionsWithUnions = array_map(function ($orExpressions) { |
| 255 | $mappedOrExpressions = []; |
| 256 | foreach ($orExpressions as $operand) { |
| 257 | $name = $operand[SegmentExpression::INDEX_OPERAND_NAME]; |
| 258 | $availableSegment = $this->getSegmentByName($name); |
| 259 | // We leave segments using !@ and != operands untouched for segments not on log_visit table as they will be build using a subquery |
| 260 | if (!$this->doesSegmentNeedSubquery($operand[SegmentExpression::INDEX_OPERAND_OPERATOR], $name) && !empty($availableSegment['unionOfSegments'])) { |
| 261 | foreach ($availableSegment['unionOfSegments'] as $segmentNameOfUnion) { |
| 262 | $operand[SegmentExpression::INDEX_OPERAND_NAME] = $segmentNameOfUnion; |
| 263 | $mappedOrExpressions[] = $operand; |
| 264 | } |
| 265 | } else { |
| 266 | $mappedOrExpressions[] = $operand; |
| 267 | } |
| 268 | } |
| 269 | return $mappedOrExpressions; |
| 270 | }, $expressions); |
| 271 | return $expressionsWithUnions; |
| 272 | } |
| 273 | private function isVisitSegment($name) |
| 274 | { |
| 275 | $availableSegment = $this->getSegmentByName($name); |
| 276 | if (!empty($availableSegment['unionOfSegments'])) { |
| 277 | foreach ($availableSegment['unionOfSegments'] as $segmentNameOfUnion) { |
| 278 | $unionSegment = $this->getSegmentByName($segmentNameOfUnion); |
| 279 | if (strpos($unionSegment['sqlSegment'], 'log_visit.') === 0) { |
| 280 | return \true; |
| 281 | } |
| 282 | } |
| 283 | } elseif (strpos($availableSegment['sqlSegment'], 'log_visit.') === 0) { |
| 284 | return \true; |
| 285 | } |
| 286 | return \false; |
| 287 | } |
| 288 | private function doesSegmentNeedSubquery($operator, $segmentName) |
| 289 | { |
| 290 | $requiresSubQuery = in_array($operator, [SegmentExpression::MATCH_DOES_NOT_CONTAIN, SegmentExpression::MATCH_NOT_EQUAL]) && !$this->isVisitSegment($segmentName); |
| 291 | if ($requiresSubQuery && empty($this->startDate) && empty($this->endDate)) { |
| 292 | if (\Piwik\Development::isEnabled()) { |
| 293 | $this->missingDatesException = new Exception(); |
| 294 | } |
| 295 | return \false; |
| 296 | } |
| 297 | return $requiresSubQuery; |
| 298 | } |
| 299 | private function getInvertedOperatorForSubQuery($operator) |
| 300 | { |
| 301 | if ($operator === SegmentExpression::MATCH_DOES_NOT_CONTAIN) { |
| 302 | return SegmentExpression::MATCH_CONTAINS; |
| 303 | } elseif ($operator === SegmentExpression::MATCH_NOT_EQUAL) { |
| 304 | return SegmentExpression::MATCH_EQUAL; |
| 305 | } |
| 306 | throw new Exception("Operator not support for subqueries"); |
| 307 | } |
| 308 | /** |
| 309 | * Returns `true` if the segment is empty, `false` if otherwise. |
| 310 | */ |
| 311 | public function isEmpty() |
| 312 | { |
| 313 | return $this->segmentExpression->isEmpty(); |
| 314 | } |
| 315 | /** |
| 316 | * Detects whether the Piwik instance is configured to be able to archive this segment. It checks whether the segment |
| 317 | * will be either archived via browser or cli archiving. It does not check if the segment has been archived. If you |
| 318 | * want to know whether the segment has been archived, the actual report data needs to be requested. |
| 319 | * |
| 320 | * This method does not take any date/period into consideration. Meaning a Piwik instance might be able to archive |
| 321 | * this segment in general, but not for a certain period if eg the archiving of range dates is disabled. |
| 322 | * |
| 323 | * @return bool |
| 324 | */ |
| 325 | public function willBeArchived() |
| 326 | { |
| 327 | if ($this->isEmpty()) { |
| 328 | return \true; |
| 329 | } |
| 330 | $idSites = $this->idSites; |
| 331 | return Rules::isRequestAuthorizedToArchive() || Rules::isBrowserArchivingAvailableForSegments() || Rules::isSegmentPreProcessed($idSites, $this); |
| 332 | } |
| 333 | protected function getCleanedExpression(array $expression) : array |
| 334 | { |
| 335 | $name = $expression[SegmentExpression::INDEX_OPERAND_NAME]; |
| 336 | $matchType = $expression[SegmentExpression::INDEX_OPERAND_OPERATOR]; |
| 337 | $value = $expression[SegmentExpression::INDEX_OPERAND_VALUE]; |
| 338 | if (empty($this->idSites)) { |
| 339 | $segmentsList = SegmentsList::get(); |
| 340 | } else { |
| 341 | $segmentsList = \Piwik\Context::changeIdSite(implode(',', $this->idSites), function () { |
| 342 | return SegmentsList::get(); |
| 343 | }); |
| 344 | } |
| 345 | $segmentObject = $segmentsList->getSegment($name); |
| 346 | $sqlName = $segmentObject ? $segmentObject->getSqlSegment() : null; |
| 347 | $joinTable = null; |
| 348 | if ($segmentObject && $segmentObject->dimension && $segmentObject->dimension->getDbColumnJoin()) { |
| 349 | $join = $segmentObject->dimension->getDbColumnJoin(); |
| 350 | $dbDiscriminator = $segmentObject->dimension->getDbDiscriminator(); |
| 351 | // we append alias since an archive query may add the table with a different join. we could eg add $table_$segmentName but |
| 352 | // then we would join an extra table per segment when we ideally want to join each table only once. However, we still need |
| 353 | // to see which table/column it joins to join it accurately each table extra if the same table is joined with different columns; |
| 354 | $tableAlias = $join->getTable() . '_segment_' . str_replace('.', '', $sqlName ?: ''); |
| 355 | $joinTable = ['table' => $join->getTable(), 'tableAlias' => $tableAlias, 'field' => $tableAlias . '.' . $join->getTargetColumn(), 'joinOn' => $sqlName . ' = ' . $tableAlias . '.' . $join->getColumn()]; |
| 356 | if ($dbDiscriminator) { |
| 357 | $joinTable['discriminator'] = $tableAlias . '.' . $dbDiscriminator->getColumn() . ' = \'' . $dbDiscriminator->getValue() . '\''; |
| 358 | } |
| 359 | } |
| 360 | if ($matchType == SegmentExpression::MATCH_IDVISIT_NOT_IN) { |
| 361 | $segmentObj = new \Piwik\Segment($value, $this->idSites, $this->startDate, $this->endDate); |
| 362 | $select = 'log_visit.idvisit'; |
| 363 | $from = 'log_visit'; |
| 364 | $datetimeField = 'visit_last_action_time'; |
| 365 | $where = []; |
| 366 | $bind = []; |
| 367 | if (!empty($this->idSites)) { |
| 368 | $where[] = "{$from}.idsite IN (" . \Piwik\Common::getSqlStringFieldsArray($this->idSites) . ")"; |
| 369 | $bind = $this->idSites; |
| 370 | } |
| 371 | if ($this->startDate instanceof \Piwik\Date) { |
| 372 | $where[] = "{$from}.{$datetimeField} >= ?"; |
| 373 | $bind[] = $this->startDate->toString(\Piwik\Date::DATE_TIME_FORMAT); |
| 374 | } |
| 375 | if ($this->endDate instanceof \Piwik\Date) { |
| 376 | $where[] = "{$from}.{$datetimeField} <= ?"; |
| 377 | $bind[] = $this->endDate->toString(\Piwik\Date::DATE_TIME_FORMAT); |
| 378 | } |
| 379 | $logQueryBuilder = StaticContainer::get('Piwik\\DataAccess\\LogQueryBuilder'); |
| 380 | $forceGroupByBackup = $logQueryBuilder->getForcedInnerGroupBySubselect(); |
| 381 | $logQueryBuilder->forceInnerGroupBySubselect(LogQueryBuilder::FORCE_INNER_GROUP_BY_NO_SUBSELECT); |
| 382 | $query = $segmentObj->getSelectQuery($select, $from, implode(' AND ', $where), $bind); |
| 383 | $logQueryBuilder->forceInnerGroupBySubselect($forceGroupByBackup); |
| 384 | return ['log_visit.idvisit', SegmentExpression::MATCH_ACTIONS_NOT_CONTAINS, $query, null, null]; |
| 385 | } |
| 386 | if (empty($segmentObject)) { |
| 387 | throw new Exception("Segment '{$name}' is not a supported segment."); |
| 388 | } |
| 389 | $segment = $this->getSegmentByName($name); |
| 390 | if ($matchType != SegmentExpression::MATCH_IS_NOT_NULL_NOR_EMPTY && $matchType != SegmentExpression::MATCH_IS_NULL_OR_EMPTY) { |
| 391 | if (isset($segment['sqlFilterValue'])) { |
| 392 | $value = call_user_func($segment['sqlFilterValue'], $value, $segment['sqlSegment']); |
| 393 | } |
| 394 | // apply presentation filter |
| 395 | if (isset($segment['sqlFilter'])) { |
| 396 | $value = call_user_func($segment['sqlFilter'], $value, $segment['sqlSegment'], $matchType, $name); |
| 397 | if (is_null($value)) { |
| 398 | // null is returned in TableLogAction::getIdActionFromSegment() |
| 399 | return array(null, $matchType, null, null, $segment); |
| 400 | } |
| 401 | // sqlFilter-callbacks might return arrays for more complex cases |
| 402 | // e.g. see TableLogAction::getIdActionFromSegment() |
| 403 | if (is_array($value) && isset($value['SQL'])) { |
| 404 | // Special case: returned value is a sub sql expression! |
| 405 | $matchType = SegmentExpression::MATCH_ACTIONS_CONTAINS; |
| 406 | $joinTable = null; |
| 407 | } |
| 408 | if (is_array($value) && isset($value['value'])) { |
| 409 | $value = $value['value']; |
| 410 | $joinTable = !empty($value['joinTable']); |
| 411 | } |
| 412 | } |
| 413 | } |
| 414 | return array($sqlName, $matchType, $value, $joinTable, $segment); |
| 415 | } |
| 416 | /** |
| 417 | * Returns the segment condition. |
| 418 | * |
| 419 | * @return string |
| 420 | */ |
| 421 | public function getString() |
| 422 | { |
| 423 | return $this->string; |
| 424 | } |
| 425 | /** |
| 426 | * Returns a hash of the segment condition, or the empty string if the segment |
| 427 | * condition is empty. |
| 428 | * |
| 429 | * @return string |
| 430 | */ |
| 431 | public function getHash() |
| 432 | { |
| 433 | if (empty($this->string)) { |
| 434 | return ''; |
| 435 | } |
| 436 | return self::getSegmentHash($this->string); |
| 437 | } |
| 438 | public static function getSegmentHash($definition) |
| 439 | { |
| 440 | $cache = \Piwik\Cache::getEagerCache(); |
| 441 | $cacheKey = self::CACHE_KEY . md5($definition); |
| 442 | if ($cache->contains($cacheKey)) { |
| 443 | return $cache->fetch($cacheKey); |
| 444 | } |
| 445 | $defaultHash = md5(urldecode($definition)); |
| 446 | // if the cache for segments already built, but this segment was not found, |
| 447 | // we return the default segment, this can be a segment from url or |
| 448 | // something like "visitorType==new" |
| 449 | if ($cache->contains(self::SEGMENT_HAS_BUILT_CACHE_KEY)) { |
| 450 | return $defaultHash; |
| 451 | } |
| 452 | // the segment hash is not built yet, let's do it |
| 453 | $model = new SegmentEditorModel(); |
| 454 | $segments = $model->getAllSegmentsAndIgnoreVisibility(); |
| 455 | foreach ($segments as $segment) { |
| 456 | $cacheKeyTemp = self::CACHE_KEY . md5($segment['definition']); |
| 457 | $cache->save($cacheKeyTemp, $segment['hash']); |
| 458 | $cacheKeyTemp = self::CACHE_KEY . md5(urldecode($segment['definition'])); |
| 459 | $cache->save($cacheKeyTemp, $segment['hash']); |
| 460 | $cacheKeyTemp = self::CACHE_KEY . md5(urlencode($segment['definition'])); |
| 461 | $cache->save($cacheKeyTemp, $segment['hash']); |
| 462 | } |
| 463 | $cache->save(self::SEGMENT_HAS_BUILT_CACHE_KEY, \true); |
| 464 | // if we found the segment, return it's hash, but maybe this |
| 465 | // segment is not stored in the db, return the default |
| 466 | if ($cache->contains($cacheKey)) { |
| 467 | return $cache->fetch($cacheKey); |
| 468 | } |
| 469 | return $defaultHash; |
| 470 | } |
| 471 | /** |
| 472 | * Extend an SQL query that aggregates data over one of the 'log_' tables with segment expressions. |
| 473 | * |
| 474 | * @param string $select The select clause. Should NOT include the **SELECT** just the columns, eg, |
| 475 | * `'t1.col1 as col1, t2.col2 as col2'`. |
| 476 | * @param array|string $from Array of table names (without prefix), eg, `array('log_visit', 'log_conversion')`. |
| 477 | * @param false|string $where (optional) Where clause, eg, `'t1.col1 = ? AND t2.col2 = ?'`. |
| 478 | * @param array|string $bind (optional) Bind parameters, eg, `array($col1Value, $col2Value)`. |
| 479 | * @param false|string $orderBy (optional) Order by clause, eg, `"t1.col1 ASC"`. |
| 480 | * @param false|string $groupBy (optional) Group by clause, eg, `"t2.col2"`. |
| 481 | * @param int $limit Limit number of result to $limit |
| 482 | * @param int $offset Specified the offset of the first row to return |
| 483 | * @param bool $forceGroupBy Force the group by and not using a subquery. Note: This may make the query slower see https://github.com/matomo-org/matomo/issues/9200#issuecomment-183641293 |
| 484 | * A $groupBy value needs to be set for this to work. |
| 485 | * @param int If set to value >= 1 then the Select query (and All inner queries) will be LIMIT'ed by this value. |
| 486 | * Use only when you're not aggregating or it will sample the data. |
| 487 | * @return array{sql: string, bind: array<scalar>} The entire select query. |
| 488 | */ |
| 489 | public function getSelectQuery($select, $from, $where = \false, $bind = array(), $orderBy = \false, $groupBy = \false, $limit = 0, $offset = 0, $forceGroupBy = \false, bool $withRollup = \false) |
| 490 | { |
| 491 | if (\Piwik\Development::isEnabled() && !empty($this->missingDatesException)) { |
| 492 | $e = new Exception(); |
| 493 | \Piwik\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 segment: ' . "\n\nCreation stacktrace:\n" . $this->missingDatesException->getTraceAsString() . "\n\nUsage stacktrace:\n" . $e->getTraceAsString()); |
| 494 | } |
| 495 | $segmentExpression = $this->segmentExpression; |
| 496 | $limitAndOffset = null; |
| 497 | if ($limit > 0) { |
| 498 | $limitAndOffset = (int) $offset . ', ' . (int) $limit; |
| 499 | } |
| 500 | try { |
| 501 | if ($forceGroupBy && $groupBy) { |
| 502 | $this->segmentQueryBuilder->forceInnerGroupBySubselect(LogQueryBuilder::FORCE_INNER_GROUP_BY_NO_SUBSELECT); |
| 503 | } |
| 504 | $result = $this->segmentQueryBuilder->getSelectQueryString($segmentExpression, $select, $from, $where, $bind, $groupBy, $orderBy, $limitAndOffset, $withRollup); |
| 505 | } catch (Exception $e) { |
| 506 | if ($forceGroupBy && $groupBy) { |
| 507 | $this->segmentQueryBuilder->forceInnerGroupBySubselect(''); |
| 508 | } |
| 509 | throw $e; |
| 510 | } |
| 511 | if ($forceGroupBy && $groupBy) { |
| 512 | $this->segmentQueryBuilder->forceInnerGroupBySubselect(''); |
| 513 | } |
| 514 | return $result; |
| 515 | } |
| 516 | /** |
| 517 | * Returns the segment string. |
| 518 | * |
| 519 | * @return string |
| 520 | */ |
| 521 | public function __toString() |
| 522 | { |
| 523 | return (string) $this->getString(); |
| 524 | } |
| 525 | /** |
| 526 | * Combines this segment with another segment condition, if the segment condition is not already |
| 527 | * in the segment. |
| 528 | * |
| 529 | * The combination is naive in that it does not take order of operations into account. |
| 530 | * |
| 531 | * @param string $segment |
| 532 | * @param string $operator The operator to use. Should be either SegmentExpression::AND_DELIMITER |
| 533 | * or SegmentExpression::OR_DELIMITER. |
| 534 | * @param string $segmentCondition The segment condition to add. |
| 535 | * @return string |
| 536 | * @throws Exception |
| 537 | */ |
| 538 | public static function combine($segment, $operator, $segmentCondition) |
| 539 | { |
| 540 | if (empty($segment)) { |
| 541 | return $segmentCondition; |
| 542 | } |
| 543 | if (empty($segmentCondition) || self::containsCondition($segment, $operator, $segmentCondition)) { |
| 544 | return $segment; |
| 545 | } |
| 546 | return $segment . $operator . $segmentCondition; |
| 547 | } |
| 548 | private static function containsCondition($segment, $operator, $segmentCondition) |
| 549 | { |
| 550 | // check when segment/condition are of same encoding |
| 551 | return strpos($segment, $operator . $segmentCondition) !== \false || strpos($segment, $segmentCondition . $operator) !== \false || strpos($segment, urlencode($operator . $segmentCondition)) !== \false || strpos($segment, urlencode($segmentCondition . $operator)) !== \false || strpos($segment, $operator . urlencode($segmentCondition)) !== \false || strpos($segment, urlencode($segmentCondition) . $operator) !== \false || strpos($segment, $operator . urldecode($segmentCondition)) !== \false || strpos($segment, urldecode($segmentCondition) . $operator) !== \false || $segment === $segmentCondition || $segment === urlencode($segmentCondition) || $segment === urldecode($segmentCondition); |
| 552 | } |
| 553 | public function getStoredSegmentName($idSite) |
| 554 | { |
| 555 | $segment = $this->getString(); |
| 556 | if (empty($segment)) { |
| 557 | return \Piwik\Piwik::translate('SegmentEditor_DefaultAllVisits'); |
| 558 | } |
| 559 | $availableSegments = SegmentEditor::getAllSegmentsForSite($idSite); |
| 560 | $foundStoredSegment = null; |
| 561 | foreach ($availableSegments as $storedSegment) { |
| 562 | if ($storedSegment['definition'] == $segment || $storedSegment['definition'] == urldecode($segment) || $storedSegment['definition'] == urlencode($segment) || $storedSegment['definition'] == $this->originalString || $storedSegment['definition'] == urldecode($this->originalString) || $storedSegment['definition'] == urlencode($this->originalString)) { |
| 563 | $foundStoredSegment = $storedSegment; |
| 564 | } |
| 565 | } |
| 566 | if (isset($foundStoredSegment)) { |
| 567 | // segment name is stored sanitized |
| 568 | return \Piwik\Common::unsanitizeInputValues($foundStoredSegment['name']); |
| 569 | } |
| 570 | return $this->isSegmentEncoded ? urldecode($segment) : $segment; |
| 571 | } |
| 572 | public function getOriginalString() |
| 573 | { |
| 574 | return $this->originalString; |
| 575 | } |
| 576 | /** |
| 577 | * Build subqueries for segments that are not on log_visit table but use !@ or != as operator |
| 578 | * This is required to ensure segments like actionUrl!@value really do not include any visit having an action containing `value` |
| 579 | * |
| 580 | * Adjacent segment conditions that both require subqueries are merged here into single NOT IN sql subqueries, |
| 581 | * which improves performance. |
| 582 | * |
| 583 | * Subquery segment conditions that are next to each other in a chain of OR's are merged together and |
| 584 | * subquery segment conditions that are next to each other in a chain of AND's, but are also alone and not |
| 585 | * a part of an OR expression, are merged. |
| 586 | * |
| 587 | * The operands for the merged conditions in the parsed intermediate structure use the special MATCH_IDVISIT_NOT_IN |
| 588 | * operator. |
| 589 | */ |
| 590 | private function mergeSubqueryExpressionsInTree(array $tree) : array |
| 591 | { |
| 592 | $andExpressions = array_map(function ($orExpressions) { |
| 593 | return $this->mergeSubqueryExpressionsInExpr($orExpressions, \false); |
| 594 | }, $tree); |
| 595 | $mappedAndExpressions = $this->mergeSubqueryExpressionsInExpr($andExpressions, \true); |
| 596 | return $mappedAndExpressions; |
| 597 | } |
| 598 | private function mergeSubqueryExpressionsInExpr(array $expressions, bool $isAndChain) : array |
| 599 | { |
| 600 | // nothing to merge if there's only one expression |
| 601 | if (!$isAndChain && count($expressions) <= 1) { |
| 602 | return $expressions; |
| 603 | } |
| 604 | $mappedExpressions = []; |
| 605 | $idvisitNotInExpressions = []; |
| 606 | foreach ($expressions as $childExpressionsOrOperand) { |
| 607 | // if this is an AND chain w/ more than one sub-expression being OR-ed together, we can't do anything about the NOT IN subqueries there |
| 608 | if ($isAndChain && count($childExpressionsOrOperand) > 1) { |
| 609 | $mappedExpressions[] = $childExpressionsOrOperand; |
| 610 | continue; |
| 611 | } |
| 612 | $operand = $isAndChain ? $childExpressionsOrOperand[0] : $childExpressionsOrOperand; |
| 613 | $name = $operand[SegmentExpression::INDEX_OPERAND_NAME]; |
| 614 | $matchType = $operand[SegmentExpression::INDEX_OPERAND_OPERATOR]; |
| 615 | $value = $operand[SegmentExpression::INDEX_OPERAND_VALUE]; |
| 616 | if (!$this->doesSegmentNeedSubquery($matchType, $name)) { |
| 617 | $mappedExpressions[] = $childExpressionsOrOperand; |
| 618 | continue; |
| 619 | } |
| 620 | // if the segment is pageTitle!=def, then NOT IN sql will have to be idvisit NOT IN (... WHERE pageTitle == def), |
| 621 | // so we must invert the operator before we create a MATCH_IDVISIT_NOT_IN operand below |
| 622 | $operator = $this->getInvertedOperatorForSubQuery($matchType); |
| 623 | $idvisitNotInExpressions[] = $name . $operator . $this->escapeSegmentValue($value); |
| 624 | } |
| 625 | if (!empty($idvisitNotInExpressions)) { |
| 626 | $newOperand = [SegmentExpression::INDEX_OPERAND_NAME => null, SegmentExpression::INDEX_OPERAND_OPERATOR => SegmentExpression::MATCH_IDVISIT_NOT_IN, SegmentExpression::INDEX_OPERAND_VALUE => implode($isAndChain ? SegmentExpression::OR_DELIMITER : SegmentExpression::AND_DELIMITER, $idvisitNotInExpressions)]; |
| 627 | $mappedExpressions[] = $isAndChain ? [$newOperand] : $newOperand; |
| 628 | } |
| 629 | return $mappedExpressions; |
| 630 | } |
| 631 | /** |
| 632 | * Escapes segment expression delimiters in a segment value with a backslash if not already done. |
| 633 | */ |
| 634 | private function escapeSegmentValue(string $value) : string |
| 635 | { |
| 636 | $delimiterPattern = SegmentExpression::AND_DELIMITER . SegmentExpression::OR_DELIMITER; |
| 637 | $pattern = '/((?<!\\\\)[' . preg_quote($delimiterPattern) . '])/'; |
| 638 | return preg_replace($pattern, '\\\\$1', $value); |
| 639 | } |
| 640 | } |
| 641 |