Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
7.58% |
15 / 198 |
|
7.14% |
1 / 14 |
CRAP | |
0.00% |
0 / 1 |
| GoogleAnalyticsPageViewService | |
7.58% |
15 / 198 |
|
7.14% |
1 / 14 |
2713.91 | |
0.00% |
0 / 1 |
| __construct | |
100.00% |
12 / 12 |
|
100.00% |
1 / 1 |
2 | |||
| setLogger | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| supports | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
6 | |||
| getPageData | |
0.00% |
0 / 77 |
|
0.00% |
0 / 1 |
462 | |||
| getSiteData | |
0.00% |
0 / 35 |
|
0.00% |
0 / 1 |
72 | |||
| getTopPages | |
0.00% |
0 / 37 |
|
0.00% |
0 / 1 |
56 | |||
| extractExpressionsFromRequests | |
0.00% |
0 / 7 |
|
0.00% |
0 / 1 |
30 | |||
| getGAName | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
| getCacheExpiry | |
0.00% |
0 / 2 |
|
0.00% |
0 / 1 |
2 | |||
| verifyApiOptions | |
75.00% |
3 / 4 |
|
0.00% |
0 / 1 |
3.14 | |||
| getEmptyDateRange | |
0.00% |
0 / 7 |
|
0.00% |
0 / 1 |
12 | |||
| getStartEnd | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
2 | |||
| pageTitleForMW | |
0.00% |
0 / 3 |
|
0.00% |
0 / 1 |
2 | |||
| createDimensions | |
0.00% |
0 / 6 |
|
0.00% |
0 / 1 |
6 | |||
| 1 | <?php |
| 2 | |
| 3 | namespace MediaWiki\Extension\PageViewInfoGA; |
| 4 | |
| 5 | use Google\Client; |
| 6 | use Google\Service\AnalyticsReporting; |
| 7 | use Google\Service\AnalyticsReporting\DateRange; |
| 8 | use Google\Service\AnalyticsReporting\Dimension; |
| 9 | use Google\Service\AnalyticsReporting\DimensionFilter; |
| 10 | use Google\Service\AnalyticsReporting\DimensionFilterClause; |
| 11 | use Google\Service\AnalyticsReporting\GetReportsRequest; |
| 12 | use Google\Service\AnalyticsReporting\Metric; |
| 13 | use Google\Service\AnalyticsReporting\OrderBy; |
| 14 | use Google\Service\AnalyticsReporting\ReportRequest; |
| 15 | use Google\Service\AnalyticsReporting\ReportRow; |
| 16 | use InvalidArgumentException; |
| 17 | use MediaWiki\Extension\PageViewInfo\PageViewService; |
| 18 | use MediaWiki\Status\Status; |
| 19 | use Psr\Log\LoggerAwareInterface; |
| 20 | use Psr\Log\LoggerInterface; |
| 21 | use Psr\Log\NullLogger; |
| 22 | use RuntimeException; |
| 23 | use StatusValue; |
| 24 | |
| 25 | /** |
| 26 | * PageViewService implementation for wikis using the Google Analytics |
| 27 | * @see https://developers.google.com/analytics |
| 28 | */ |
| 29 | class GoogleAnalyticsPageViewService implements PageViewService, LoggerAwareInterface { |
| 30 | /** @var LoggerInterface */ |
| 31 | protected $logger; |
| 32 | |
| 33 | /** @var AnalyticsReporting */ |
| 34 | protected $analytics; |
| 35 | |
| 36 | /** @var string Profile(View) ID of the Google Analytics View. */ |
| 37 | protected $profileId; |
| 38 | |
| 39 | /** @var array */ |
| 40 | protected $customMap; |
| 41 | |
| 42 | /** @var bool */ |
| 43 | protected $readCustomDimensions; |
| 44 | |
| 45 | /** @var int UNIX timestamp of 0:00 of the last day with complete data */ |
| 46 | protected $lastCompleteDay; |
| 47 | |
| 48 | /** @var array Cache for getEmptyDateRange() */ |
| 49 | protected $range; |
| 50 | |
| 51 | /** Google Analytics API restricts number of requests up to 5. */ |
| 52 | public const MAX_REQUEST = 5; |
| 53 | |
| 54 | /** |
| 55 | * @param array $options Associative array. |
| 56 | */ |
| 57 | public function __construct( array $options ) { |
| 58 | $this->verifyApiOptions( $options ); |
| 59 | |
| 60 | // Skip the current day for which only partial information is available |
| 61 | $this->lastCompleteDay = strtotime( '0:0 1 day ago' ); |
| 62 | |
| 63 | $this->logger = new NullLogger(); |
| 64 | |
| 65 | $client = new Client(); |
| 66 | $client->setApplicationName( 'PageViewInfo' ); |
| 67 | if ( $options['credentialsFile'] ) { |
| 68 | $client->setAuthConfig( $options['credentialsFile'] ); |
| 69 | } |
| 70 | |
| 71 | $client->addScope( AnalyticsReporting::ANALYTICS_READONLY ); |
| 72 | $this->analytics = new AnalyticsReporting( $client ); |
| 73 | |
| 74 | $this->profileId = $options['profileId'] ?? false; |
| 75 | $this->customMap = $options['customMap'] ?? false; |
| 76 | $this->readCustomDimensions = $options['readCustomDimensions'] ?? false; |
| 77 | } |
| 78 | |
| 79 | /** |
| 80 | * @inheritDoc |
| 81 | */ |
| 82 | public function setLogger( LoggerInterface $logger ): void { |
| 83 | $this->logger = $logger; |
| 84 | } |
| 85 | |
| 86 | /** |
| 87 | * @inheritDoc |
| 88 | */ |
| 89 | public function supports( $metric, $scope ) { |
| 90 | return in_array( $metric, [ self::METRIC_VIEW, self::METRIC_UNIQUE ] ) && |
| 91 | in_array( $scope, [ self::SCOPE_ARTICLE, self::SCOPE_TOP, self::SCOPE_SITE ] ); |
| 92 | } |
| 93 | |
| 94 | /** |
| 95 | * @inheritDoc |
| 96 | */ |
| 97 | public function getPageData( array $titles, $days, $metric = self::METRIC_VIEW ) { |
| 98 | if ( !$titles ) { |
| 99 | return StatusValue::newGood( [] ); |
| 100 | } |
| 101 | if ( $days <= 0 ) { |
| 102 | throw new InvalidArgumentException( 'Invalid days: ' . $days ); |
| 103 | } |
| 104 | |
| 105 | $readCustomDimensions = $this->readCustomDimensions; |
| 106 | $result = []; |
| 107 | $requests = []; |
| 108 | foreach ( $titles as $title ) { |
| 109 | $result[$title->getPrefixedDBkey()] = $this->getEmptyDateRange( $days ); |
| 110 | |
| 111 | // Create DateRange |
| 112 | $dateRange = new DateRange(); |
| 113 | $dateRange->setStartDate( $days . 'daysAgo' ); |
| 114 | $dateRange->setEndDate( "1daysAgo" ); |
| 115 | |
| 116 | // Create Metrics |
| 117 | $gaMetric = new Metric(); |
| 118 | if ( $metric === self::METRIC_VIEW ) { |
| 119 | $gaMetric->setExpression( 'ga:pageviews' ); |
| 120 | } elseif ( $metric === self::METRIC_UNIQUE ) { |
| 121 | $gaMetric->setExpression( 'ga:uniquePageviews' ); |
| 122 | } else { |
| 123 | throw new InvalidArgumentException( 'Invalid metric: ' . $metric ); |
| 124 | } |
| 125 | |
| 126 | // Create DimensionFilter |
| 127 | $dimensionFilter = new DimensionFilter(); |
| 128 | if ( $readCustomDimensions ) { |
| 129 | // Use custom dimensions instead of ga:pageTitle |
| 130 | $dimensionFilter->setDimensionName( $this->getGAName( 'mw:page_title' ) ); |
| 131 | $dimensionFilter->setOperator( 'EXACT' ); |
| 132 | $dimensionFilter->setExpressions( [ $title->getPrefixedDBkey() ] ); |
| 133 | } else { |
| 134 | // Use regular expression to filter the title. |
| 135 | // This is not the ideal approach and maybe fails for some titles. |
| 136 | $dimensionFilter->setDimensionName( 'ga:pageTitle' ); |
| 137 | $dimensionFilter->setOperator( 'REGEXP' ); |
| 138 | $dimensionFilter->setExpressions( [ |
| 139 | '^' . str_replace( '_', ' ', $title->getPrefixedDBkey() ) . ' - [^-]+$' ] ); |
| 140 | } |
| 141 | // Create DimensionFilterClause |
| 142 | $dimensionFilterClause = new DimensionFilterClause(); |
| 143 | $dimensionFilterClause->setFilters( [ $dimensionFilter ] ); |
| 144 | |
| 145 | // Create ReportRequest |
| 146 | $request = new ReportRequest(); |
| 147 | $request->setViewId( $this->profileId ); |
| 148 | $request->setDateRanges( [ $dateRange ] ); |
| 149 | $request->setMetrics( [ $gaMetric ] ); |
| 150 | $request->setDimensions( $this->createDimensions( [ |
| 151 | 'ga:date', |
| 152 | $readCustomDimensions ? $this->getGAName( 'mw:page_title' ) : 'ga:pageTitle', |
| 153 | ] ) ); |
| 154 | $request->setDimensionFilterClauses( [ $dimensionFilterClause ] ); |
| 155 | |
| 156 | $requests[] = $request; |
| 157 | } |
| 158 | |
| 159 | $status = StatusValue::newGood(); |
| 160 | for ( $i = 0; $i < count( $requests ); $i += self::MAX_REQUEST ) { |
| 161 | $reqs = array_slice( $requests, $i, self::MAX_REQUEST ); |
| 162 | $body = new GetReportsRequest(); |
| 163 | $body->setReportRequests( $reqs ); |
| 164 | |
| 165 | $reports = []; |
| 166 | try { |
| 167 | $reports = $this->analytics->reports->batchGet( $body )->getReports(); |
| 168 | } catch ( \Google\Service\Exception $e ) { |
| 169 | foreach ( self::extractExpressionsFromRequests( $reqs ) as $exp ) { |
| 170 | if ( !$readCustomDimensions ) { |
| 171 | // $exp is a regular expression for title, strip. |
| 172 | preg_match( '/\^(.+) - \[\^-\]\+\$/', $exp, $matches ); |
| 173 | if ( !$matches ) { |
| 174 | continue; |
| 175 | } |
| 176 | $exp = $matches[1]; |
| 177 | } |
| 178 | $status->success[$exp] = false; |
| 179 | } |
| 180 | $status->error( 'pvi-invalidresponse' ); |
| 181 | } |
| 182 | |
| 183 | foreach ( $reports as $rep ) { |
| 184 | $rows = $rep->getData()->getRows(); |
| 185 | if ( !$rows || !is_array( $rows ) ) { |
| 186 | continue; |
| 187 | } |
| 188 | foreach ( $rows as $row ) { |
| 189 | if ( !( $row instanceof ReportRow ) ) { |
| 190 | continue; |
| 191 | } |
| 192 | $ts = $row->getDimensions()[0]; |
| 193 | $day = substr( $ts, 0, 4 ) . '-' . substr( $ts, 4, 2 ) . '-' . substr( $ts, 6, 2 ); |
| 194 | $count = (int)$row->getMetrics()[0]->getValues()[0]; |
| 195 | $title = $row->getDimensions()[1]; |
| 196 | if ( !$readCustomDimensions ) { |
| 197 | $title = $this->pageTitleForMW( $title ); |
| 198 | } |
| 199 | $result[$title][$day] = $count; |
| 200 | $status->success[$title] = true; |
| 201 | } |
| 202 | } |
| 203 | } |
| 204 | |
| 205 | // Fills success even if the title is not included in responses. |
| 206 | // https://github.com/femiwiki/PageViewInfoGA/issues/46 |
| 207 | foreach ( $titles as $title ) { |
| 208 | if ( !in_array( $title, $status->success ) ) { |
| 209 | $status->success[$title->getPrefixedDBkey()] = false; |
| 210 | } |
| 211 | } |
| 212 | $status->successCount = count( array_filter( $status->success ) ); |
| 213 | $status->failCount = count( $status->success ) - $status->successCount; |
| 214 | $status->setResult( (bool)$status->successCount, $result ); |
| 215 | return $status; |
| 216 | } |
| 217 | |
| 218 | /** |
| 219 | * @inheritDoc |
| 220 | */ |
| 221 | public function getSiteData( $days, $metric = self::METRIC_VIEW ) { |
| 222 | if ( $metric !== self::METRIC_VIEW && $metric !== self::METRIC_UNIQUE ) { |
| 223 | throw new InvalidArgumentException( 'Invalid metric: ' . $metric ); |
| 224 | } |
| 225 | if ( $days <= 0 ) { |
| 226 | throw new InvalidArgumentException( 'Invalid days: ' . $days ); |
| 227 | } |
| 228 | $result = $this->getEmptyDateRange( $days ); |
| 229 | |
| 230 | // Create the DateRange object. |
| 231 | $dateRange = new DateRange(); |
| 232 | $dateRange->setStartDate( $days . 'daysAgo' ); |
| 233 | $dateRange->setEndDate( '1daysAgo' ); |
| 234 | |
| 235 | // Create the Metrics object. |
| 236 | $gaMetric = new Metric(); |
| 237 | if ( $metric === self::METRIC_VIEW ) { |
| 238 | $gaMetric->setExpression( 'ga:pageviews' ); |
| 239 | } elseif ( $metric === self::METRIC_UNIQUE ) { |
| 240 | $gaMetric->setExpression( 'ga:uniquePageviews' ); |
| 241 | } else { |
| 242 | throw new InvalidArgumentException( 'Invalid metric: ' . $metric ); |
| 243 | } |
| 244 | |
| 245 | // Create the Dimension object. |
| 246 | $dimension = new Dimension(); |
| 247 | $dimension->setName( 'ga:date' ); |
| 248 | |
| 249 | // Create the ReportRequest object. |
| 250 | $request = new ReportRequest(); |
| 251 | $request->setViewId( $this->profileId ); |
| 252 | $request->setDateRanges( [ $dateRange ] ); |
| 253 | $request->setMetrics( [ $gaMetric ] ); |
| 254 | $request->setDimensions( [ $dimension ] ); |
| 255 | |
| 256 | $body = new GetReportsRequest(); |
| 257 | $body->setReportRequests( [ $request ] ); |
| 258 | |
| 259 | $status = Status::newGood(); |
| 260 | try { |
| 261 | $data = $this->analytics->reports->batchGet( $body ); |
| 262 | $rows = $data->getReports()[0]->getData()->getRows(); |
| 263 | |
| 264 | foreach ( $rows as $row ) { |
| 265 | $ts = $row->dimensions[0]; |
| 266 | $day = substr( $ts, 0, 4 ) . '-' . substr( $ts, 4, 2 ) . '-' . substr( $ts, 6, 2 ); |
| 267 | $count = (int)$row->metrics[0]->values[0]; |
| 268 | $result[$day] = $count; |
| 269 | } |
| 270 | $status->setResult( $status->isOK(), $result ); |
| 271 | } catch ( RuntimeException $e ) { |
| 272 | $status->fatal( 'pvi-invalidresponse' ); |
| 273 | } |
| 274 | return $status; |
| 275 | } |
| 276 | |
| 277 | /** |
| 278 | * @inheritDoc |
| 279 | */ |
| 280 | public function getTopPages( $metric = self::METRIC_VIEW ) { |
| 281 | $result = []; |
| 282 | if ( !in_array( $metric, [ self::METRIC_VIEW, self::METRIC_UNIQUE ] ) ) { |
| 283 | throw new InvalidArgumentException( 'Invalid metric: ' . $metric ); |
| 284 | } |
| 285 | |
| 286 | // Create the DateRange object. |
| 287 | $dateRange = new DateRange(); |
| 288 | $dateRange->setStartDate( '2daysAgo' ); |
| 289 | $dateRange->setEndDate( '1daysAgo' ); |
| 290 | |
| 291 | // Create the Metrics object and OrderBy object. |
| 292 | $gaMetric = new Metric(); |
| 293 | $orderBy = new OrderBy(); |
| 294 | $orderBy->setSortOrder( 'DESCENDING' ); |
| 295 | if ( $metric === self::METRIC_VIEW ) { |
| 296 | $gaMetric->setExpression( 'ga:pageviews' ); |
| 297 | $orderBy->setFieldName( 'ga:pageviews' ); |
| 298 | } elseif ( $metric === self::METRIC_UNIQUE ) { |
| 299 | $gaMetric->setExpression( 'ga:uniquePageviews' ); |
| 300 | $orderBy->setFieldName( 'ga:uniquePageviews' ); |
| 301 | } |
| 302 | |
| 303 | // Create the Dimension object. |
| 304 | $dimension = new Dimension(); |
| 305 | $dimension->setName( $this->readCustomDimensions ? $this->getGAName( 'mw:page_title' ) : 'ga:pageTitle' ); |
| 306 | |
| 307 | // Create the ReportRequest object. |
| 308 | $request = new ReportRequest(); |
| 309 | $request->setViewId( $this->profileId ); |
| 310 | $request->setDateRanges( [ $dateRange ] ); |
| 311 | $request->setMetrics( [ $gaMetric ] ); |
| 312 | $request->setDimensions( [ $dimension ] ); |
| 313 | $request->setOrderBys( [ $orderBy ] ); |
| 314 | |
| 315 | $body = new GetReportsRequest(); |
| 316 | $body->setReportRequests( [ $request ] ); |
| 317 | |
| 318 | $status = Status::newGood(); |
| 319 | try { |
| 320 | $data = $this->analytics->reports->batchGet( $body ); |
| 321 | $rows = $data->getReports()[0]->getData()->getRows(); |
| 322 | |
| 323 | foreach ( $rows as $row ) { |
| 324 | $title = $row->dimensions[0]; |
| 325 | $title = $this->pageTitleForMW( $title ); |
| 326 | $count = (int)$row->metrics[0]->values[0]; |
| 327 | $result[$title] = $count; |
| 328 | } |
| 329 | $status->setResult( $status->isOK(), $result ); |
| 330 | } catch ( RuntimeException $e ) { |
| 331 | $status->fatal( 'pvi-invalidresponse' ); |
| 332 | } |
| 333 | return $status; |
| 334 | } |
| 335 | |
| 336 | /** |
| 337 | * @param ReportRequest[] $requests |
| 338 | * @return string[] |
| 339 | */ |
| 340 | protected static function extractExpressionsFromRequests( $requests ) { |
| 341 | $exps = []; |
| 342 | foreach ( $requests as $req ) { |
| 343 | foreach ( $req->getDimensionFilterClauses() as $clause ) { |
| 344 | foreach ( $clause->getFilters() as $filter ) { |
| 345 | foreach ( $filter->getExpressions() as $exp ) { |
| 346 | $exps[] = $exp; |
| 347 | } |
| 348 | } |
| 349 | } |
| 350 | } |
| 351 | return $exps; |
| 352 | } |
| 353 | |
| 354 | /** |
| 355 | * @param string $mwName |
| 356 | * @return string |
| 357 | */ |
| 358 | protected function getGAName( $mwName ) { |
| 359 | $flipped = array_flip( $this->customMap ); |
| 360 | return 'ga:' . $flipped[$mwName]; |
| 361 | } |
| 362 | |
| 363 | /** |
| 364 | * @inheritDoc |
| 365 | */ |
| 366 | public function getCacheExpiry( $metric, $scope ) { |
| 367 | // data is valid until the end of the day |
| 368 | $endOfDay = strtotime( '0:0 next day' ); |
| 369 | return $endOfDay - time(); |
| 370 | } |
| 371 | |
| 372 | /** |
| 373 | * @param array $apiOptions |
| 374 | * @throws InvalidArgumentException |
| 375 | */ |
| 376 | protected function verifyApiOptions( array $apiOptions ) { |
| 377 | if ( !isset( $apiOptions['credentialsFile'] ) ) { |
| 378 | throw new InvalidArgumentException( "'credentialsFile' is required" ); |
| 379 | } elseif ( !isset( $apiOptions['profileId'] ) ) { |
| 380 | throw new InvalidArgumentException( "'profileId' is required" ); |
| 381 | } |
| 382 | } |
| 383 | |
| 384 | /** |
| 385 | * The API omits dates if there is no data. Fill it with nulls to make client-side |
| 386 | * processing easier. |
| 387 | * @param int $days |
| 388 | * @return array YYYY-MM-DD => null |
| 389 | */ |
| 390 | protected function getEmptyDateRange( $days ) { |
| 391 | if ( !$this->range ) { |
| 392 | $this->range = []; |
| 393 | // we only care about the date part, so add some hours to avoid errors when there is a |
| 394 | // leap second or some other weirdness |
| 395 | $end = $this->lastCompleteDay + 12 * 3600; |
| 396 | $start = $end - ( $days - 1 ) * 24 * 3600; |
| 397 | for ( $ts = $start; $ts <= $end; $ts += 24 * 3600 ) { |
| 398 | $this->range[gmdate( 'Y-m-d', $ts )] = null; |
| 399 | } |
| 400 | } |
| 401 | return $this->range; |
| 402 | } |
| 403 | |
| 404 | /** |
| 405 | * Get start and end timestamp in YYYYMMDDHH format |
| 406 | * @param int $days |
| 407 | * @return string[] |
| 408 | */ |
| 409 | protected function getStartEnd( $days ) { |
| 410 | $end = $this->lastCompleteDay + 12 * 3600; |
| 411 | $start = $end - ( $days - 1 ) * 24 * 3600; |
| 412 | return [ gmdate( 'Ymd', $start ) . '00', gmdate( 'Ymd', $end ) . '00' ]; |
| 413 | } |
| 414 | |
| 415 | /** |
| 416 | * @param string $gaTitle |
| 417 | * @return string title text converted MediaWiki-friendly |
| 418 | */ |
| 419 | protected static function pageTitleForMW( $gaTitle ) { |
| 420 | // TODO: Use "pagetitle" and "pagetitle-view-mainpage" messages |
| 421 | $title = preg_replace( '/ - [^-]+$/', '', $gaTitle ); |
| 422 | $title = preg_replace( '/ /', '_', $title ); |
| 423 | |
| 424 | return $title; |
| 425 | } |
| 426 | |
| 427 | /** |
| 428 | * @param string[] $names |
| 429 | * @return Dimension[] |
| 430 | */ |
| 431 | protected function createDimensions( $names ) { |
| 432 | $dimensions = []; |
| 433 | foreach ( $names as $name ) { |
| 434 | $dimension = new Dimension(); |
| 435 | $dimension->setName( $name ); |
| 436 | $dimensions[] = $dimension; |
| 437 | } |
| 438 | return $dimensions; |
| 439 | } |
| 440 | } |