Db
2 years ago
Handler
2 years ago
Visit
2 years ago
Action.php
2 years ago
ActionPageview.php
2 years ago
Cache.php
2 years ago
Db.php
2 years ago
Failures.php
2 years ago
FingerprintSalt.php
2 years ago
GoalManager.php
2 years ago
Handler.php
2 years ago
IgnoreCookie.php
2 years ago
LogTable.php
2 years ago
Model.php
2 years ago
PageUrl.php
2 years ago
Request.php
2 years ago
RequestProcessor.php
2 years ago
RequestSet.php
2 years ago
Response.php
2 years ago
ScheduledTasksRunner.php
2 years ago
Settings.php
2 years ago
TableLogAction.php
2 years ago
TrackerCodeGenerator.php
2 years ago
TrackerConfig.php
2 years ago
Visit.php
2 years ago
VisitExcluded.php
2 years ago
VisitInterface.php
2 years ago
Visitor.php
2 years ago
VisitorNotFoundInDb.php
2 years ago
VisitorRecognizer.php
2 years ago
TableLogAction.php
266 lines
| 1 | <?php |
| 2 | |
| 3 | /** |
| 4 | * Matomo - free/libre analytics platform |
| 5 | * |
| 6 | * @link https://matomo.org |
| 7 | * @license http://www.gnu.org/licenses/gpl-3.0.html GPL v3 or later |
| 8 | * |
| 9 | */ |
| 10 | namespace Piwik\Tracker; |
| 11 | |
| 12 | use Piwik\Common; |
| 13 | use Piwik\Segment\SegmentExpression; |
| 14 | /** |
| 15 | * This class is used to query Action IDs from the log_action table. |
| 16 | * |
| 17 | * A pageview, outlink, download or site search are made of several "Action IDs" |
| 18 | * For example pageview is idaction_url and idaction_name. |
| 19 | * |
| 20 | */ |
| 21 | class TableLogAction |
| 22 | { |
| 23 | /** |
| 24 | * This function will find the idaction from the lookup table log_action, |
| 25 | * given an Action name, type, and an optional URL Prefix. |
| 26 | * |
| 27 | * This is used to record Page URLs, Page Titles, Ecommerce items SKUs, item names, item categories |
| 28 | * |
| 29 | * If the action name does not exist in the lookup table, it will INSERT it |
| 30 | * @param array $actionsNameAndType Array of one or many (name,type) |
| 31 | * @return array Returns the an array (Field name => idaction) |
| 32 | */ |
| 33 | public static function loadIdsAction($actionsNameAndType) |
| 34 | { |
| 35 | // Add url prefix if not set |
| 36 | foreach ($actionsNameAndType as &$action) { |
| 37 | if (2 == count($action)) { |
| 38 | $action[] = null; |
| 39 | } |
| 40 | } |
| 41 | $actionIds = self::queryIdsAction($actionsNameAndType); |
| 42 | [$queriedIds, $fieldNamesToInsert] = self::processIdsToInsert($actionsNameAndType, $actionIds); |
| 43 | $insertedIds = self::insertNewIdsAction($actionsNameAndType, $fieldNamesToInsert); |
| 44 | $queriedIds = $queriedIds + $insertedIds; |
| 45 | return $queriedIds; |
| 46 | } |
| 47 | /** |
| 48 | * @param $matchType |
| 49 | * @param $actionType |
| 50 | * @return string |
| 51 | * @throws \Exception |
| 52 | */ |
| 53 | private static function getSelectQueryWhereNameContains($matchType, $actionType) |
| 54 | { |
| 55 | // now, we handle the cases =@ (contains) and !@ (does not contain) |
| 56 | // build the expression based on the match type |
| 57 | $sql = 'SELECT idaction FROM ' . Common::prefixTable('log_action') . ' WHERE %s AND type = ' . $actionType . ' )'; |
| 58 | switch ($matchType) { |
| 59 | case SegmentExpression::MATCH_CONTAINS: |
| 60 | // use concat to make sure, no %s occurs because some plugins use %s in their sql |
| 61 | $where = '( name LIKE CONCAT(\'%\', ?, \'%\') '; |
| 62 | break; |
| 63 | case SegmentExpression::MATCH_DOES_NOT_CONTAIN: |
| 64 | $where = '( name NOT LIKE CONCAT(\'%\', ?, \'%\') '; |
| 65 | break; |
| 66 | case SegmentExpression::MATCH_STARTS_WITH: |
| 67 | // use concat to make sure, no %s occurs because some plugins use %s in their sql |
| 68 | $where = '( name LIKE CONCAT(?, \'%\') '; |
| 69 | break; |
| 70 | case SegmentExpression::MATCH_ENDS_WITH: |
| 71 | // use concat to make sure, no %s occurs because some plugins use %s in their sql |
| 72 | $where = '( name LIKE CONCAT(\'%\', ?) '; |
| 73 | break; |
| 74 | default: |
| 75 | throw new \Exception("This match type {$matchType} is not available for action-segments."); |
| 76 | break; |
| 77 | } |
| 78 | $sql = sprintf($sql, $where); |
| 79 | return $sql; |
| 80 | } |
| 81 | private static function insertNewIdsAction($actionsNameAndType, $fieldNamesToInsert) |
| 82 | { |
| 83 | // Then, we insert all new actions in the lookup table |
| 84 | $inserted = array(); |
| 85 | foreach ($fieldNamesToInsert as $fieldName) { |
| 86 | [$name, $type, $urlPrefix] = $actionsNameAndType[$fieldName]; |
| 87 | $actionId = self::getModel()->createNewIdAction($name, $type, $urlPrefix); |
| 88 | Common::printDebug("Recorded a new action (" . \Piwik\Tracker\Action::getTypeAsString($type) . ") in the lookup table: " . $name . " (idaction = " . $actionId . ")"); |
| 89 | $inserted[$fieldName] = $actionId; |
| 90 | } |
| 91 | return $inserted; |
| 92 | } |
| 93 | private static function getModel() |
| 94 | { |
| 95 | return new \Piwik\Tracker\Model(); |
| 96 | } |
| 97 | private static function queryIdsAction($actionsNameAndType) |
| 98 | { |
| 99 | $toQuery = array(); |
| 100 | foreach ($actionsNameAndType as &$actionNameType) { |
| 101 | [$name, $type, $urlPrefix] = $actionNameType; |
| 102 | $toQuery[] = array('name' => $name, 'type' => $type); |
| 103 | } |
| 104 | $actionIds = self::getModel()->getIdsAction($toQuery); |
| 105 | return $actionIds; |
| 106 | } |
| 107 | private static function processIdsToInsert($actionsNameAndType, $actionIds) |
| 108 | { |
| 109 | // For the Actions found in the lookup table, add the idaction in the array, |
| 110 | // If not found in lookup table, queue for INSERT |
| 111 | $fieldNamesToInsert = $fieldNameToActionId = array(); |
| 112 | foreach ($actionsNameAndType as $fieldName => &$actionNameType) { |
| 113 | @(list($name, $type, $urlPrefix) = $actionNameType); |
| 114 | if (empty($name)) { |
| 115 | $fieldNameToActionId[$fieldName] = false; |
| 116 | continue; |
| 117 | } |
| 118 | $found = false; |
| 119 | foreach ($actionIds as $row) { |
| 120 | if ($name == $row['name'] && $type == $row['type']) { |
| 121 | $found = true; |
| 122 | $fieldNameToActionId[$fieldName] = $row['idaction']; |
| 123 | continue; |
| 124 | } |
| 125 | } |
| 126 | if (!$found) { |
| 127 | $fieldNamesToInsert[] = $fieldName; |
| 128 | } |
| 129 | } |
| 130 | return array($fieldNameToActionId, $fieldNamesToInsert); |
| 131 | } |
| 132 | /** |
| 133 | * Convert segment expression to an action ID or an SQL expression. |
| 134 | * |
| 135 | * This method is used as a sqlFilter-callback for the segments of this plugin. |
| 136 | * Usually, these callbacks only return a value that should be compared to the |
| 137 | * column in the database. In this case, that doesn't work since multiple IDs |
| 138 | * can match an expression (e.g. "pageUrl=@foo"). |
| 139 | * @param string $valueToMatch |
| 140 | * @param string $sqlField |
| 141 | * @param string $matchType |
| 142 | * @param string $segmentName |
| 143 | * @throws \Exception |
| 144 | * @return array|int|string |
| 145 | */ |
| 146 | public static function getIdActionFromSegment($valueToMatch, $sqlField, $matchType, $segmentName) |
| 147 | { |
| 148 | if ($segmentName === 'actionType') { |
| 149 | $actionType = (int) $valueToMatch; |
| 150 | $valueToMatch = array(); |
| 151 | $sql = 'SELECT idaction FROM ' . Common::prefixTable('log_action') . ' WHERE type = ' . $actionType . ' )'; |
| 152 | } else { |
| 153 | $actionType = self::guessActionTypeFromSegment($segmentName); |
| 154 | $valueToMatch = self::removeProtocolIfSegmentStoredWithoutIt($valueToMatch, $actionType, $segmentName); |
| 155 | $unsanitizedValue = $valueToMatch; |
| 156 | $valueToMatch = self::normaliseActionString($actionType, $valueToMatch); |
| 157 | if ($matchType == SegmentExpression::MATCH_EQUAL || $matchType == SegmentExpression::MATCH_NOT_EQUAL) { |
| 158 | $idAction = self::getModel()->getIdActionMatchingNameAndType($valueToMatch, $actionType); |
| 159 | // If action can't be found normalized try search for it with original value |
| 160 | // This can eg happen for outlinks that contain a & see https://github.com/matomo-org/matomo/issues/11806 |
| 161 | if (empty($idAction)) { |
| 162 | $idAction = self::getModel()->getIdActionMatchingNameAndType($unsanitizedValue, $actionType); |
| 163 | // Action is not found (eg. &segment=pageTitle==Větrnásssssss) |
| 164 | if (empty($idAction)) { |
| 165 | $idAction = null; |
| 166 | } |
| 167 | } |
| 168 | return $idAction; |
| 169 | } |
| 170 | // "name contains $string" match can match several idaction so we cannot return yet an idaction |
| 171 | // special case |
| 172 | $sql = self::getSelectQueryWhereNameContains($matchType, $actionType); |
| 173 | } |
| 174 | return array( |
| 175 | // mark that the returned value is an sql-expression instead of a literal value |
| 176 | 'SQL' => $sql, |
| 177 | 'bind' => $valueToMatch, |
| 178 | ); |
| 179 | } |
| 180 | /** |
| 181 | * @param $segmentName |
| 182 | * @return int |
| 183 | * @throws \Exception |
| 184 | */ |
| 185 | private static function guessActionTypeFromSegment($segmentName) |
| 186 | { |
| 187 | $exactMatch = array('outlinkUrl' => \Piwik\Tracker\Action::TYPE_OUTLINK, 'downloadUrl' => \Piwik\Tracker\Action::TYPE_DOWNLOAD, 'eventUrl' => \Piwik\Tracker\Action::TYPE_EVENT, 'eventAction' => \Piwik\Tracker\Action::TYPE_EVENT_ACTION, 'eventCategory' => \Piwik\Tracker\Action::TYPE_EVENT_CATEGORY, 'eventName' => \Piwik\Tracker\Action::TYPE_EVENT_NAME, 'contentPiece' => \Piwik\Tracker\Action::TYPE_CONTENT_PIECE, 'contentTarget' => \Piwik\Tracker\Action::TYPE_CONTENT_TARGET, 'contentName' => \Piwik\Tracker\Action::TYPE_CONTENT_NAME, 'contentInteraction' => \Piwik\Tracker\Action::TYPE_CONTENT_INTERACTION, 'productName' => \Piwik\Tracker\Action::TYPE_ECOMMERCE_ITEM_NAME, 'productSku' => \Piwik\Tracker\Action::TYPE_ECOMMERCE_ITEM_SKU, 'productViewName' => \Piwik\Tracker\Action::TYPE_ECOMMERCE_ITEM_NAME, 'productViewSku' => \Piwik\Tracker\Action::TYPE_ECOMMERCE_ITEM_SKU); |
| 188 | if (!empty($exactMatch[$segmentName])) { |
| 189 | return $exactMatch[$segmentName]; |
| 190 | } |
| 191 | if (stripos($segmentName, 'pageurl') !== false) { |
| 192 | return \Piwik\Tracker\Action::TYPE_PAGE_URL; |
| 193 | } elseif (stripos($segmentName, 'pagetitle') !== false) { |
| 194 | return \Piwik\Tracker\Action::TYPE_PAGE_TITLE; |
| 195 | } elseif (stripos($segmentName, 'sitesearch') !== false) { |
| 196 | return \Piwik\Tracker\Action::TYPE_SITE_SEARCH; |
| 197 | } elseif (stripos($segmentName, 'productcategory') !== false || stripos($segmentName, 'productviewcategory') !== false) { |
| 198 | return \Piwik\Tracker\Action::TYPE_ECOMMERCE_ITEM_CATEGORY; |
| 199 | } else { |
| 200 | throw new \Exception("We cannot guess the action type from the segment {$segmentName}."); |
| 201 | } |
| 202 | } |
| 203 | /** |
| 204 | * This function will sanitize or not if it's needed for the specified action type |
| 205 | * |
| 206 | * URLs (Download URL, Outlink URL) are stored raw (unsanitized) |
| 207 | * while other action types are stored Sanitized |
| 208 | * |
| 209 | * @param $actionType |
| 210 | * @param $actionString |
| 211 | * @return string |
| 212 | */ |
| 213 | private static function normaliseActionString($actionType, $actionString) |
| 214 | { |
| 215 | $actionString = Common::unsanitizeInputValue($actionString); |
| 216 | if (self::isActionTypeStoredUnsanitized($actionType)) { |
| 217 | return $actionString; |
| 218 | } |
| 219 | return Common::sanitizeInputValue($actionString); |
| 220 | } |
| 221 | /** |
| 222 | * @param $actionType |
| 223 | * @return bool |
| 224 | */ |
| 225 | private static function isActionTypeStoredUnsanitized($actionType) |
| 226 | { |
| 227 | $actionsTypesStoredUnsanitized = array(\Piwik\Tracker\Action::TYPE_DOWNLOAD, \Piwik\Tracker\Action::TYPE_OUTLINK, \Piwik\Tracker\Action::TYPE_PAGE_URL, \Piwik\Tracker\Action::TYPE_CONTENT); |
| 228 | return in_array($actionType, $actionsTypesStoredUnsanitized); |
| 229 | } |
| 230 | public static function removeProtocolIfSegmentStoredWithoutIt($url, $actionType, $segmentName) |
| 231 | { |
| 232 | if ($actionType == \Piwik\Tracker\Action::TYPE_PAGE_URL || $segmentName == 'eventUrl') { |
| 233 | // for urls trim protocol and www because it is not recorded in the db |
| 234 | $url = preg_replace('@^http[s]?://(www\\.)?@i', '', $url); |
| 235 | } |
| 236 | return $url; |
| 237 | } |
| 238 | /** |
| 239 | * Returns an idaction value to match an idaction column by searching log_action, if $matchType is |
| 240 | * SegmentExpression::MATCH_EQUAL or SegmentExpression::MATCH_NOT_EQUAL. This method is used |
| 241 | * to optimize segment conditions involving idaction queries, avoiding a join by querying the log_action |
| 242 | * table beforehand. |
| 243 | * |
| 244 | * Should be used as the $sqlFilter property for idaction dimensions that use `ActionNameJoin`. |
| 245 | * |
| 246 | * @param string $value the value in the segment condition |
| 247 | * @param string $sqlField the table column of the segment condition |
| 248 | * @param string $matchType the SegmentExpression match type, eg, `SegmentExpression::MATCH_NOT_EQUAL` |
| 249 | * @param string $segmentName the name of the segment, ie, `pageUrl` |
| 250 | * @return array|null|string |
| 251 | */ |
| 252 | public static function getOptimizedIdActionSqlMatch($value, $sqlField, $matchType, $segmentName) |
| 253 | { |
| 254 | if ($matchType == SegmentExpression::MATCH_EQUAL || $matchType == SegmentExpression::MATCH_NOT_EQUAL) { |
| 255 | $result = self::getIdActionFromSegment($value, $sqlField, $matchType, $segmentName); |
| 256 | if (is_numeric($result)) { |
| 257 | return ['value' => $result, 'joinTable' => false]; |
| 258 | } |
| 259 | return $result; |
| 260 | } |
| 261 | $actionType = self::guessActionTypeFromSegment($segmentName); |
| 262 | $value = self::removeProtocolIfSegmentStoredWithoutIt($value, $actionType, $segmentName); |
| 263 | return $value; |
| 264 | } |
| 265 | } |
| 266 |