Options for YT feeds: minimum video duration and extracting links from descriptionHEADmaster

1 files changed, 95 insertions, 0 deletions

diff --git a/providers/Youtube.php b/providers/Youtube.php
index bd38a52..03d0e27 100644
--- a/providers/Youtube.php
+++ b/providers/Youtube.php
@@ -3,9 +3,20 @@
 namespace Providers;
 
 require_once('Rss.php');
+require_once('../lib/php-youtube-api/src/autoload.php');
 
 class Youtube extends Rss {
 
+    const DEFAULT_MIN_DURATION = 60;
+
+    public function __construct($feed, $options=[]) {
+        parent::__construct($feed, $options);
+        $config = json_decode(file_get_contents('../config/youtube.json'), TRUE);
+        $this->_youtube = new \Madcoda\Youtube\Youtube([
+            'key' => $config['key']
+        ]);
+    }
+
     protected function _getFeedUrl($feed) {
         return 'https://www.youtube.com/feeds/videos.xml?channel_id=' . $feed;
     }
@@ -14,6 +25,90 @@ class Youtube extends Rss {
         return '../cache/youtube.%s';
     }
 
+    protected function _mapRssItem($item) {
+        $itemObj = parent::_mapRssItem($item);
+        if (array_key_exists('descriptionlink', $this->_options)) {
+            $linkConfig = explode(':', $this->_options['descriptionlink']);
+            $linkDomain = NULL;
+            if (!is_numeric($linkConfig[0])) {
+                $linkDomain = array_shift($linkConfig);
+                if (!$linkConfig) {
+                    $linkConfig = ['1'];
+                }
+            }
+            $linkIndex = intval($linkConfig[0]) - 1;
+            $textLinks = [];
+            if (preg_match_all('#(?:(?:https?|ftp):\/\/)?[\w/\-?=%.]+\.[\w/\-&?=%.]+#', $itemObj->Text, $textLinks)) {
+                $textLinks = $textLinks[0];
+                if ($linkDomain) {
+                    $textLinks = array_values(array_filter($textLinks, function($link) use($linkDomain) {
+                        if($linkParts = parse_url($link)) {
+                            return $linkParts['host'] == $linkDomain;
+                        }
+                    }));
+                }
+                if (count($textLinks) > $linkIndex) {
+                    $itemObj->Link = $textLinks[$linkIndex];
+                }
+            }
+        }
+        return $itemObj;
+    }
+
+    private function _getVideoMetadata($ids) {
+        $cacheFile = sprintf($this->_getCachePath() . '.metadata', $this->_feed);
+        $cacheData = [];
+        if (file_exists($cacheFile)) {
+            $cacheData = unserialize(file_get_contents($cacheFile));
+        }
+        $oldIDs = array_keys($cacheData);
+        $newIDs = array_diff($ids, $oldIDs);
+        if ($newIDs) {
+            $metadata = $this->_youtube->getVideosInfo($newIDs);
+            foreach ($metadata as $video) {
+                $cacheData[$video->id] = $video;
+            }
+        }
+        file_put_contents($cacheFile, serialize($cacheData));
+        return array_filter($cacheData, function($item) use($ids) {
+            return in_array($item->id, $ids);
+        });
+    }
+
+    private function _getVideoDurations($ids) {
+        $metadata = $this->_getVideoMetadata($ids);
+        $return = [];
+        foreach ($ids as $id) {
+            $duration = new \DateInterval($metadata[$id]->contentDetails->duration);
+            $return[$id] = $duration->h * 3600 + $duration->i * 60 + $duration->s;
+        }
+        return $return;
+    }
+
+    private function _getVideoID($item) {
+        return preg_replace('/^yt:video:/', '', $item->ID);
+    }
+
+    protected function _filterItemContent($items) {
+        $items = parent::_filterItemContent($items);
+
+        if (array_key_exists('duration', $this->_options)) {
+            $duration = intval($this->_options['duration']);
+            $videoIDs = array_map(array($this, '_getVideoID'), $items);
+            $durations = $this->_getVideoDurations($videoIDs);
+            $minDuration = self::DEFAULT_MIN_DURATION;
+            if (is_numeric($this->_options['duration'])) {
+                $minDuration = intval($this->_options['duration']);
+            }
+            $items = array_filter($items, function($item) use($minDuration, $durations) {
+                $videoID = $this->_getVideoID($item);
+                return $durations[$videoID] >= $minDuration;
+            });
+        }
+
+        return $items;
+    }
+
 }
 
 ?>