Get titles, names and companies from imdb search dom

Author: hmerritt

src/HtmlPieces.php

@@ -19,7 +19,7 @@ class HtmlPieces
      * @param string $element
      * @return string
      */
-    public function get(object $page, string $element, array $options = [])
+    public function get(object $page, string $element)
     {
         //  Initiate dom object
         //  -> handles page scraping
@@ -124,6 +124,48 @@ public function get(object $page, string $element, array $options = [])
                 return $technical_specs;
                 break;
 
+            case "titles":
+            case "names":
+            case "companies":
+                $response = [];
+                $sections = $page->find(".findSection");
+                if (count($sections) > 0)
+                {
+                    foreach ($sections as $section)
+                    {
+                        $sectionName = @strtolower($section->find(".findSectionHeader")->text);
+                        if ($sectionName === $element) {
+                            $sectionRows = $section->find(".findList tr");
+                            if (count($sectionRows) > 0)
+                            {
+                                foreach ($sectionRows as $sectionRow)
+                                {
+                                    $row = [];
+
+                                    $link = $dom->find($sectionRow, 'td.result_text a');
+                                    $row["title"] = $link->text;
+                                    if ($row["title"] == "") {
+                                        continue;
+                                    }
+
+                                    $row["image"] = $dom->find($sectionRow, 'td.primary_photo img')->src;
+                                    if (preg_match('/@/', $row["image"]))
+                                    {
+                                        $row["image"] = preg_split('~@(?=[^@]*$)~', $row["image"])[0] . "@.jpg";
+                                    }
+                                    $row["image"] = empty($row["image"]) ? "" : $row["image"];
+
+                                    $row["id"] = $this->extractImdbId($link->href);
+
+                                    array_push($response, $row);
+                                }
+                            }
+                        }
+                    }
+                }
+                return $response;
+                break;
+
             default:
                 return "";
         }

src/imdb.php

@@ -11,13 +11,6 @@
 class Imdb
 {
 
-    /**
-     * IMDB base url
-     *
-     * @var string
-     */
-    protected $baseUrl = 'https://www.imdb.com/';
-
     /**
      * Returns default options combined with any user options
      *
@@ -66,8 +59,8 @@ public function film(string $filmId, array $options = []): array
         //  -> handles finding specific content from the dom
         $htmlPieces = new HtmlPieces;
 
-        //  Load imdb page and parse the dom
-        $page = $dom->fetch($this->baseUrl."title/".$filmId, $options);
+        //  Load imdb film page and parse the dom
+        $page = $dom->fetch("https://www.imdb.com/title/".$filmId, $options);
 
         //  Add all film data to response $store
         $response->add("id", $filmId);
@@ -81,9 +74,10 @@ public function film(string $filmId, array $options = []): array
         $response->add("trailer", $htmlPieces->get($page, "trailer"));
         $response->add("cast", $htmlPieces->get($page, "cast"));
 
-        //  Technical specs
+        //  If techSpecs is enabled in user $options
+        //  -> Make a second request to load the full techSpecs page
         if ($options["techSpecs"]) {
-                $page_techSpecs = $dom->fetch($this->baseUrl."title/".$filmId.'/technical', $options);
+                $page_techSpecs = $dom->fetch("https://www.imdb.com/title/$filmId/technical", $options);
                 $response->add("technical_specs", $htmlPieces->get($page_techSpecs, "technical_specs"));
         }
         else {
@@ -109,9 +103,21 @@ public function search(string $search, array $options = []): array
         // -> handles what the api returns
         $response = new Response;
 
-        $response->add("titles", []);
-        $response->add("names", []);
-        $response->add("companies", []);
+        //  Initiate dom object
+        //  -> handles page scraping
+        $dom = new Dom;
+
+        //  Initiate html-pieces object
+        //  -> handles finding specific content from the dom
+        $htmlPieces = new HtmlPieces;
+
+        //  Load imdb search page and parse the dom
+        $page = $dom->fetch("https://www.imdb.com/find?q=$search&s=all", $options);
+
+        //  Add all search data to response $store
+        $response->add("titles", $htmlPieces->get($page, "titles"));
+        $response->add("names", $htmlPieces->get($page, "names"));
+        $response->add("companies", $htmlPieces->get($page, "companies"));
 
         return $response->return();
     }

tests/index.php

@@ -14,7 +14,7 @@
 // Initialise Imdb
 // Load film data
 $imdb = new Imdb();
-$film = $imdb->film($q);  // tt0816692  tt8633464
+$film = $imdb->search($q);  // tt0816692  tt8633464
 
 // Return loaded film data
 echo json_encode($film, JSON_PRETTY_PRINT);