imp: Allow to get the correct duration of Youtube videos

flusio · May 29, 2024 · e132baa · e132baa
1 parent bf0fdb6
commit e132baa
Show file tree

Hide file tree

Showing 3 changed files with 104 additions and 4 deletions.
diff --git a/lib/SpiderBits/src/DomExtractor.php b/lib/SpiderBits/src/DomExtractor.php
@@ -124,6 +124,46 @@ public static function content(Dom $dom): string
         return $main_node->text();
     }
 
+    /**
+     * Return the duration in minutes of the element in the DOM.
+     *
+     * If the DOM declares an element with itemprop="duration", return the
+     * corresponding duration. Otherwise, estimate the duration from the
+     * content itself by counting the number of words in the content and divide
+     * by an average reading speed (i.e. 200 words per minute).
+     */
+    public static function duration(Dom $dom): int
+    {
+        // Search for a node having the attribute itemprop="duration
+        // @see https://schema.org/docs/gs.html
+        $duration_node = $dom->select('//*[@itemprop = "duration"]/attribute::content');
+
+        if ($duration_node) {
+            try {
+                $interval = new \DateInterval($duration_node->text());
+                // Convert the interval to minutes
+                $duration = $interval->y * 12 * 30 * 24 * 60;
+                $duration += $interval->m * 30 * 24 * 60;
+                $duration += $interval->d * 24 * 60;
+                $duration += $interval->h * 60;
+                $duration += $interval->i;
+                if ($interval->s >= 30) {
+                    $duration += 1;
+                }
+                return $duration;
+            } catch (\Exception $e) {
+                // Do nothing and fallback to the other mode
+            }
+        }
+
+        // If there is no duration node (or if its content can't be parsed),
+        // roughly estimate the duration from the DOM content.
+        $content = self::content($dom);
+        $words = array_filter(explode(' ', $content));
+        $average_reading_speed = 200;
+        return intval(count($words) / $average_reading_speed);
+    }
+
     /**
      * Return the autodiscovered feeds URLs (RSS and Atom).
      *

diff --git a/src/services/LinkFetcher.php b/src/services/LinkFetcher.php
@@ -269,10 +269,7 @@ public function fetchUrl(string $url): array
             }
         }
 
-        // And roughly estimate the reading time
-        $content = \SpiderBits\DomExtractor::content($dom);
-        $words = array_filter(explode(' ', $content));
-        $info['reading_time'] = intval(count($words) / 200);
+        $info['reading_time'] = \SpiderBits\DomExtractor::duration($dom);
 
         // Get the illustration URL if any
         $url_illustration = \SpiderBits\DomExtractor::illustration($dom);

diff --git a/tests/lib/SpiderBits/DomExtractorTest.php b/tests/lib/SpiderBits/DomExtractorTest.php
@@ -4,6 +4,8 @@
 
 class DomExtractorTest extends \PHPUnit\Framework\TestCase
 {
+    use \tests\FakerHelper;
+
     public function testTitle(): void
     {
         $dom = Dom::fromText(<<<HTML
@@ -280,6 +282,67 @@ public function testContentStripsScripts(): void
         $this->assertSame('This is main', $content);
     }
 
+    public function testDuration(): void
+    {
+        /** @var string */
+        $content = $this->fake('words', 400, true);
+        $dom = Dom::fromText(<<<HTML
+            <html>
+                <body>
+                    <main>
+                        {$content}
+                    </main>
+                </body>
+            </html>
+        HTML);
+
+        $duration = DomExtractor::duration($dom);
+
+        $this->assertSame(2, $duration);
+    }
+
+    public function testDurationWithItempropAttribute(): void
+    {
+        /** @var string */
+        $content = $this->fake('words', 400, true);
+        $dom = Dom::fromText(<<<HTML
+            <html>
+                <body>
+                    <meta itemprop="duration" content="PT41M35S" />
+
+                    <main>
+                        {$content}
+                    </main>
+                </body>
+            </html>
+        HTML);
+
+        $duration = DomExtractor::duration($dom);
+
+        $this->assertSame(42, $duration);
+    }
+
+    public function testDurationWithInvalidItempropAttribute(): void
+    {
+        /** @var string */
+        $content = $this->fake('words', 400, true);
+        $dom = Dom::fromText(<<<HTML
+            <html>
+                <body>
+                    <meta itemprop="duration" content="NotIso8601" />
+
+                    <main>
+                        {$content}
+                    </main>
+                </body>
+            </html>
+        HTML);
+
+        $duration = DomExtractor::duration($dom);
+
+        $this->assertSame(2, $duration);
+    }
+
     public function testFeeds(): void
     {
         $dom = Dom::fromText(<<<HTML