* Motor Sport Magazine feed provider

1 files changed, 116 insertions, 0 deletions

diff --git a/providers/Motorsport.php b/providers/Motorsport.php
new file mode 100644
index 0000000..f055ad6
--- /dev/null
+++ b/providers/Motorsport.php
@@ -0,0 +1,116 @@
+<?php
+
+namespace Providers;
+
+require_once('Twitter.php');
+require_once('Item.php');
+require_once('../lib/querypath/src/qp.php');
+
+class Motorsport extends \Providers\Twitter {
+
+    protected $_cacheTimeout = '15 minutes';
+
+    public function __construct($feed, $options=[]) {
+        parent::__construct($feed, $options);
+    }
+
+    protected function _getCachePath() {
+        return '../cache/motorsport.%s';
+    }
+
+    protected function _fetchItems() {
+        $content = parent::_fetchItems();
+        $links = [];
+        foreach ($content as $item) {
+            foreach ($item->entities->urls as $url) {
+                $links[$url->expanded_url] = $item;
+            }
+        }
+        $links = array_filter(
+            $links,
+            function($link) {
+                $linkParts = parse_url($link);
+                return $linkParts['host'] != 'twitter.com';
+            },
+            ARRAY_FILTER_USE_KEY
+        );
+        $content = [];
+        foreach ($links as $link => $i) {
+            $item = new \stdClass();
+            $item->title = $i->text;
+            $item->time = $i->created_at;
+            $item->link = $link;
+            if (isset($i->user)) {
+                $item->author = $i->user->screen_name;
+            }
+            $item->content = $this->_getLinkMetadata($this->_getLinkContent($link));
+            $content[] = $item;
+        }
+        return $content;
+    }
+
+    protected function _getLinkContent($link) {
+        $cacheHash = md5($link);
+        $cacheFile = sprintf($this->_getCachePath() . '.%s', $this->_feed, $cacheHash);
+        if (file_exists($cacheFile)) {
+            return file_get_contents($cacheFile);
+        } else {
+            $content = file_get_contents($link);
+            file_put_contents($cacheFile, $content);
+            return $content;
+        }
+    }
+
+    protected function _getLinkMetadata($content) {
+        $tree = htmlqp($content);
+        $meta = $tree->find('meta');
+        $metadata = [];
+        foreach ($meta as $tag) {
+            $attributes = $tag->attr();
+            if (isset($attributes['property'])) {
+                $metadata[$attributes['property']] = $attributes['content'];
+            }
+        }
+        if (isset($metadata['og:url'])) {
+            $originalUrl = parse_url($metadata['og:url']);
+            if (substr($originalUrl['host'], -22) != 'motorsportmagazine.com') {
+                unset($metadata['og:url']);
+            }
+        }
+        return $metadata;
+    }
+
+    protected function _spamFilter($items) {
+        return array_filter(
+            $items,
+            function ($item) {
+                return isset($item->content['article:published_time']);
+            }
+        );
+    }
+
+    protected function _mapItems($content) {
+        $items = [];
+        foreach ($content as $i) {
+            $url = isset($i->content['og:url']) ? $i->content['og:url'] : $i->link;
+            $item = new Item();
+            $item->ID = md5($url);
+            $item->Title = isset($i->content['article:published_time']) ? $i->content['og:title'] : $i->title;
+            $item->Link = $url;
+            $item->Text = isset($i->content['article:published_time']) ? $i->content['og:description'] : $i->content['og:title'];
+            $item->Time = isset($i->content['article:published_time']) ? $i->content['article:published_time'] : $i->time;
+            if (isset($i->author)) {
+                $item->Author = $i->author;
+            }
+            $items[] = $item;
+        }
+        return $items;
+    }
+
+    protected function _sortContent($content) {
+      return $content;
+    }
+
+}
+
+?>