Merge pull request #550 from leepeuker/fix-imdb-rating-scrapper

Update imdb rating scrapping with latest html changes
leepeuker · Dec 3, 2023 · ddeb6a8 · ddeb6a8
2 parents 35de7b9 + a55c5d8
commit ddeb6a8
Show file tree

Hide file tree

Showing 2 changed files with 30 additions and 30 deletions.
diff --git a/src/Api/Imdb/ImdbWebScrapper.php b/src/Api/Imdb/ImdbWebScrapper.php
@@ -70,7 +70,7 @@ private function extractProductionStatus(string $imdbRatingPage) : ?string
 
     private function extractRatingAverage(string $imdbRatingPage, string $imdbId) : ?float
     {
-        preg_match('/iZlgcd">(\d([.,])\d)/', $imdbRatingPage, $averageRatingMatches);
+        preg_match('/cMEQkK">(\d([.,])\d)/', $imdbRatingPage, $averageRatingMatches);
         if (empty($averageRatingMatches[1]) === true) {
             $this->logger->warning('IMDb: Could not extract rating average.', ['url' => $this->urlGenerator->buildMovieUrl($imdbId)]);
 
@@ -83,32 +83,32 @@ private function extractRatingAverage(string $imdbRatingPage, string $imdbId) :
     private function extractRatingVoteCount(string $imdbRatingPage, string $imdbId) : ?int
     {
         // Handle numbers without suffix
-        preg_match('/bjjENQ">([0-9]+)</', $imdbRatingPage, $voteCountMatches);
+        preg_match('/gPVQxL">([0-9]+)</', $imdbRatingPage, $voteCountMatches);
         if (empty($voteCountMatches[1]) === false) {
             return (int)$voteCountMatches[1];
         }
-        preg_match('/bjjENQ">([0-9]{1,3}([.,]?[0-9]{3})+)/', $imdbRatingPage, $voteCountMatches);
+        preg_match('/gPVQxL">([0-9]{1,3}([.,]?[0-9]{3})+)/', $imdbRatingPage, $voteCountMatches);
         if (empty($voteCountMatches[1]) === false) {
             return (int)str_replace([',', '.'], '', $voteCountMatches[1]);
         }
 
         // Handle numbers with K suffix
-        preg_match('/bjjENQ">([0-9]+)K</', $imdbRatingPage, $voteCountMatches);
+        preg_match('/gPVQxL">([0-9]+)K</', $imdbRatingPage, $voteCountMatches);
         if (empty($voteCountMatches[1]) === false) {
             return (int)$voteCountMatches[1] * 1000;
         }
-        preg_match('/bjjENQ">([0-9]{1,3}[.,][0-9]{1,3})K</', $imdbRatingPage, $voteCountMatches);
+        preg_match('/gPVQxL">([0-9]{1,3}[.,][0-9]{1,3})K</', $imdbRatingPage, $voteCountMatches);
         if (empty($voteCountMatches[1]) === false) {
             return (int)((float)$voteCountMatches[1] * 1000);
         }
 
         // Handle simple numbers with M suffix
-        preg_match('/bjjENQ">([0-9]+)M</', $imdbRatingPage, $voteCountMatches);
+        preg_match('/gPVQxL">([0-9]+)M</', $imdbRatingPage, $voteCountMatches);
         if (empty($voteCountMatches[1]) === false) {
             return (int)$voteCountMatches[1] * 1000000;
         }
         // Handle simple numbers with K suffix
-        preg_match('/bjjENQ">([0-9]{1,3}[.,][0-9]{1,3})M</', $imdbRatingPage, $voteCountMatches);
+        preg_match('/gPVQxL">([0-9]{1,3}[.,][0-9]{1,3})M</', $imdbRatingPage, $voteCountMatches);
         if (empty($voteCountMatches[1]) === false) {
             return (int)((float)$voteCountMatches[1] * 1000000);
         }

diff --git a/tests/unit/Api/Imdb/ImdbWebScrapperTest.php b/tests/unit/Api/Imdb/ImdbWebScrapperTest.php
@@ -27,63 +27,63 @@ public function provideFindRatingData() : array
     {
         return [
             [
-                'iZlgcd">7.9</span>
-                bjjENQ">229.240</div>',
+                'cMEQkK">7.9</span>
+                gPVQxL">229.240</div>',
                 ImdbRating::create(7.9, 229240)
             ],
             'returns no rating if current production status is found' => [
                 'hjAonB">Post-production
-                iZlgcd">7.9</span>
-                bjjENQ">229.240</div>',
+                cMEQkK">7.9</span>
+                gPVQxL">229.240</div>',
                 null,
             ],
             [
-                'iZlgcd">7,9</span>
-                bjjENQ">229,240</div>',
+                'cMEQkK">7,9</span>
+                gPVQxL">229,240</div>',
                 ImdbRating::create(7.9, 229240)
             ],
             [
-                'iZlgcd">7,9</span>
-                bjjENQ">229240</div>',
+                'cMEQkK">7,9</span>
+                gPVQxL">229240</div>',
                 ImdbRating::create(7.9, 229240)
             ],
             [
-                'iZlgcd">7,9</span>
-                bjjENQ">1.229,240</div>',
+                'cMEQkK">7,9</span>
+                gPVQxL">1.229,240</div>',
                 ImdbRating::create(7.9, 1229240)
             ],
             [
-                'iZlgcd">7,9</span>
-                bjjENQ">40</div>',
+                'cMEQkK">7,9</span>
+                gPVQxL">40</div>',
                 ImdbRating::create(7.9, 40)
             ],
             [
-                'iZlgcd">7,9</span>
-                bjjENQ">40K</div>',
+                'cMEQkK">7,9</span>
+                gPVQxL">40K</div>',
                 ImdbRating::create(7.9, 40000)
             ],
             [
-                'iZlgcd">7,9</span>
-                bjjENQ">4.1K</div>',
+                'cMEQkK">7,9</span>
+                gPVQxL">4.1K</div>',
                 ImdbRating::create(7.9, 4100)
             ],
             [
-                'iZlgcd">7,9</span>
-                bjjENQ">14.12K</div>',
+                'cMEQkK">7,9</span>
+                gPVQxL">14.12K</div>',
                 ImdbRating::create(7.9, 14120)
             ],
             [
-                'iZlgcd">7,9</span>
-                bjjENQ">10M</div>',
+                'cMEQkK">7,9</span>
+                gPVQxL">10M</div>',
                 ImdbRating::create(7.9, 10000000)
             ],
             [
-                'iZlgcd">7,9</span>
-                bjjENQ">10.1M</div>',
+                'cMEQkK">7,9</span>
+                gPVQxL">10.1M</div>',
                 ImdbRating::create(7.9, 10100000)
             ],
             [
-                'iZlgcd">7,9</span>',
+                'cMEQkK">7,9</span>',
                 null
             ],
             [