PageDiff provider

1 files changed, 110 insertions, 0 deletions

diff --git a/providers/Pagediff.php b/providers/Pagediff.php
new file mode 100644
index 0000000..8679532
--- /dev/null
+++ b/providers/Pagediff.php
@@ -0,0 +1,110 @@
+<?php
+
+namespace Providers;
+
+require_once('HtmlFeed.php');
+require_once('Item.php');
+
+class Pagediff extends \Providers\HtmlFeed {
+
+    protected $_cacheTimeout = '1 hour';
+
+    public function __construct($feed, $options=[]) {
+        $config = json_decode(file_get_contents('../config/pagediff.json'), TRUE);
+        if (!isset($config[$feed])) {
+            throw new \Exception(sprintf('Feed %s not configured', $feed));
+        }
+        $this->_config = $config[$feed];
+        parent::__construct($feed, $options);
+    }
+
+    protected function _getCachePath() {
+        return '../cache/pagediff.%s';
+    }
+
+    protected function _getFeedUrl($feed) {
+        return $this->_config['url'];
+    }
+
+    private function _getItemCachePath() {
+        return sprintf('../cache/pagediff.items.%s', $this->_feed);
+    }
+
+    private function _getCachedContent() {
+        if (!file_exists($this->_getItemCachePath())) {
+            return [];
+        }
+        return unserialize(
+            file_get_contents(
+                $this->_getItemCachePath()
+            )
+        );
+    }
+
+    private function _saveCachedContent($content) {
+        return file_put_contents(
+            $this->_getItemCachePath(),
+            serialize($content)
+        );
+    }
+
+    private function _getContentFromSelector($tree, $selector) {
+        $node = $tree->find($selector['node']);
+        if ($node->count() == 0) {
+            return NULL;
+        }
+        if ($node->count() != 1) {
+            if (isset($selector['index'])) {
+                $node = $node->eq($selector['index']);
+            } else {
+                $node = $node->first();
+            }
+        }
+        if (isset($selector['html'])) {
+            return $node->innerHTML();
+        }
+        if (isset($selector['attr'])) {
+            $text = $node->attr()[$selector['attr']];
+        } else {
+            $text = $node->text();
+        }
+        if (isset($selector['transform'])) {
+            $text = sprintf($selector['transform'], $text);
+        }
+        return $text;
+    }
+
+    protected function _parseFeedContent($tree) {
+        $selectors = $this->_config['selectors'];
+        $items = $this->_getCachedContent();
+        $currentItem = [];
+        foreach (['id', 'link', 'name', 'text'] as $type) {
+            $currentItem[$type] = $this->_getContentFromSelector($tree, $selectors[$type]);
+        }
+        $currentItem['time'] = date('Y-m-d H:i:s');
+        if (!count($items) || $currentItem['id'] != $items[0]['id']) {
+            $items = array_merge([$currentItem], $items);
+            $this->_saveCachedContent($items);
+        }
+        return $items;
+    }
+
+    protected function _mapItems($items) {
+        return array_map(function($item) {
+            $i = new Item();
+            $i->ID = $item['id'];
+            $i->Title = $item['name'];
+            $i->Time = $item['time'];
+            $i->Text = $item['text'];
+            $i->Link = $item['link'];
+            return $i;
+        }, $items);
+    }
+
+    public function title() {
+        return $this->_config['title'];
+    }
+
+}
+
+?>