Using an external (Python...) library to fetch FB posts

Obligatory "Fuck you, Zuckerberg"

2 files changed, 32 insertions, 71 deletions

diff --git a/bin/get-fb-content.py b/bin/get-fb-content.py
new file mode 100644
index 0000000..c7238e0
--- /dev/null
+++ b/bin/get-fb-content.py
@@ -0,0 +1,19 @@
+import json
+import logging
+import sys
+from os import path
+
+from facebook_scraper import get_posts
+
+
+BASEDIR = path.dirname(__file__)
+
+posts = []
+for post in get_posts(sys.argv[1], cookies=path.join(BASEDIR, '../config/facebook.com_cookies.txt'), pages=3):
+    posts.append({
+        'id': post['post_id'],
+        'time': str(post['time']),
+        'texts': [t.strip() for t in post['text'].split('\n') if t]
+    })
+
+print(json.dumps(posts))
diff --git a/providers/Facebook.php b/providers/Facebook.php
index d5233c3..7757dff 100644
--- a/providers/Facebook.php
+++ b/providers/Facebook.php
@@ -5,91 +5,25 @@ namespace Providers;
 require_once('HtmlFeed.php');
 require_once('Item.php');
 
-class Facebook extends \Providers\HtmlFeed {
+class Facebook extends \Providers\Provider {
 
     protected $_cacheTimeout = '300 years';
 
     public function __construct($feed, $options) {
         parent::__construct($feed, $options);
-        if (isset($this->_options['dump'])) {
-            $this->_options['force'] = TRUE;
-        }
         if (isset($this->_options['force'])) {
             $this->_cacheTimeout = '1 second';
         }
     }
 
-    protected function __getUserAgent() {
-        return 'Mozilla/5.0 (Windows NT 6.3; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/90.0.4430.51 Safari/537.36';
-    }
-
     protected function _getCachePath() {
         return '../cache/facebook.%s';
     }
 
-    protected function _getFeedUrl($feed) {
-        return sprintf('https://m.facebook.com/%s/posts', $feed);
-    }
-
-    private function _extractTimestamp($block) {
-        $origString = (string)$block->find('abbr')->eq(0)->text();
-        $string = str_replace("\xc2\xa0", ' ', $origString);
-        $string = str_replace(' godzinie ', ' ', $string);
-        $string = str_replace(' o ', ', ', $string);
-        $string = str_replace('Wczoraj', 'Yesterday', $string);
-        $string = str_replace('Dzisiaj', '', $string);
-        $string = str_replace('godz.', 'hours ago', $string);
-        $string = str_replace('min', 'minutes ago', $string);
-        $string = str_replace('Przed chwilą', 'now', $string);
-        $string = str_replace(['stycznia', 'lutego', 'marca', 'kwietnia', 'maja', 'czerwca', 'lipca', 'sierpnia', 'września', 'października', 'listopada', 'grudnia'], ['January', 'February', 'March', 'April', 'May', 'June', 'July', 'August', 'September', 'October', 'November', 'December'], $string);
-        $time = strtotime($string);
-        if (!$time) {
-            var_dump(bin2hex($origString));
-            throw new \Exception('Cannot parse date string: ' . $origString);
-        }
-        return $string;
-    }
-
-    protected function _parseFeedContent($tree) {
-        $items = [];
-        if (isset($this->_options['dump'])) {
-            print($tree->html());
-        }
-        foreach ($tree->find('#timelineBody div[data-ft]') as $header) {
-            $data = json_decode($header->attr()['data-ft'], TRUE);
-            if (isset($this->_options['dump'])) {
-                var_dump($data);
-            }
-            if (isset($data['mf_story_key'])) {
-                $key = $data['mf_story_key'];
-                $texts = [];
-                foreach ($header->find('p, h3') as $paragraph) {
-                    $text = $paragraph->text();
-                    if ($text != 'Więcej') {
-                        $texts[] = $text;
-                    }
-                }
-                if (isset($this->_options['dump'])) {
-                    print_r($data);
-                    print($key);
-                    print(PHP_EOL);
-                    print_r($texts);
-                    print(PHP_EOL);
-                }
-                if (count($texts)) {
-                    $items[$key] = [
-                        'id' => $key,
-                        'time' => $this->_extractTimestamp($header),
-                        'content' => $header->html(),
-                        'texts' => $texts
-                    ];
-                }
-            }
-        }
-        if (isset($this->_options['dump'])) {
-            die();
-        }
-        return array_values($items);
+    protected function _fetchItems() {
+        $jsonContent = [];
+        exec('python3 ' . implode(DIRECTORY_SEPARATOR, [dirname(__FILE__), '..', 'bin', 'get-fb-content.py']) . ' ' . escapeshellarg($this->_feed), $jsonContent);
+        return json_decode(implode(PHP_EOL, $jsonContent), TRUE);
     }
 
     protected function _mapItems($content) {
@@ -110,6 +44,14 @@ class Facebook extends \Providers\HtmlFeed {
         );
     }
 
+    protected function _sortContent($content) {
+        return $content;
+    }
+
+    protected function _spamFilter($items) {
+        return $items;
+    }
+
     public function title() {
         return sprintf("%s's Facebook page posts", $this->_feed);
     }