From 9601f3be1ee5be711671a922c3282339aeb1ddf4 Mon Sep 17 00:00:00 2001
From: emkael <emkael@tlen.pl>
Date: Mon, 31 Jan 2022 21:39:23 +0100
Subject: Another round of FB crap

---
 _cron/fb-cache-files            |   3 ++-
 bin/fb-scrape/.gitignore        |   2 ++
 bin/fb-scrape/get-fb-content.py |  42 ++++++++++++++++++++++++++++++++++++++++
 bin/get-fb-content.py           |  20 -------------------
 bin/refresh-fb-cache.php        |   2 +-
 config/facebook.com_cookies.txt | Bin 1154 -> 875 bytes
 providers/Facebook.php          |  19 +++++++++++-------
 7 files changed, 59 insertions(+), 29 deletions(-)
 create mode 100644 bin/fb-scrape/.gitignore
 create mode 100644 bin/fb-scrape/get-fb-content.py
 delete mode 100644 bin/get-fb-content.py

diff --git a/_cron/fb-cache-files b/_cron/fb-cache-files
index c6b2cb1..4e116d3 100644
--- a/_cron/fb-cache-files
+++ b/_cron/fb-cache-files
@@ -1,2 +1,3 @@
-*/10 * * * * $SITEPATH/bin/refresh-fb-cache.php
+DIRENV_LOG_FORMAT=""
+14 * * * * $SITEPATH/bin/refresh-fb-cache.php
 10 * * * * find $SITEPATH/cache -size 6c
diff --git a/bin/fb-scrape/.gitignore b/bin/fb-scrape/.gitignore
new file mode 100644
index 0000000..772944e
--- /dev/null
+++ b/bin/fb-scrape/.gitignore
@@ -0,0 +1,2 @@
+.envrc
+.direnv
diff --git a/bin/fb-scrape/get-fb-content.py b/bin/fb-scrape/get-fb-content.py
new file mode 100644
index 0000000..5b22f70
--- /dev/null
+++ b/bin/fb-scrape/get-fb-content.py
@@ -0,0 +1,42 @@
+import json
+import logging
+import sys
+import time
+import warnings
+from os import path
+from pytz_deprecation_shim import PytzUsageWarning
+from random import randint
+
+from facebook_scraper import get_posts, enable_logging
+from requests.exceptions import RequestException
+
+debug = len(sys.argv) > 2 and sys.argv[2] == 'debug'
+
+if debug:
+    enable_logging(logging.DEBUG)
+
+warnings.filterwarnings(
+    action='ignore',
+    message=r'A low page limit'
+)
+warnings.filterwarnings(
+    action='ignore',
+    category=PytzUsageWarning
+)
+
+BASEDIR = path.dirname(__file__)
+
+posts = []
+try:
+    for post in get_posts(sys.argv[1], pages=2, cookies=path.join(BASEDIR, '../../config/facebook.com_cookies.txt')):
+        posts.append({
+            'id': post['post_id'],
+            'time': str(post['time']),
+            'texts': [t.strip() for t in post['text'].split('\n') if t] if post['text'] else [],
+            'images': post['images']
+        })
+        time.sleep(randint(10, 15))
+except RequestException:
+    pass
+
+print(json.dumps(posts))
diff --git a/bin/get-fb-content.py b/bin/get-fb-content.py
deleted file mode 100644
index 180c7b4..0000000
--- a/bin/get-fb-content.py
+++ /dev/null
@@ -1,20 +0,0 @@
-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],
-        'images': post['images']
-    })
-
-print(json.dumps(posts))
diff --git a/bin/refresh-fb-cache.php b/bin/refresh-fb-cache.php
index 4b702ef..ca50146 100755
--- a/bin/refresh-fb-cache.php
+++ b/bin/refresh-fb-cache.php
@@ -25,7 +25,7 @@ foreach ($cacheFiles as $file) {
     }
 }
 
-$filesToFetch = ['Piwoteka', 'fermentlodz', '103731068463865', $fileToFetch];
+$filesToFetch = ['Piwoteka', 'fermentlodz', $fileToFetch];
 
 foreach ($filesToFetch as $feed) {
     sleep(rand(60, 90));
diff --git a/config/facebook.com_cookies.txt b/config/facebook.com_cookies.txt
index 26d3c0e..42160a8 100644
Binary files a/config/facebook.com_cookies.txt and b/config/facebook.com_cookies.txt differ
diff --git a/providers/Facebook.php b/providers/Facebook.php
index 5eb588e..20f5028 100644
--- a/providers/Facebook.php
+++ b/providers/Facebook.php
@@ -22,7 +22,10 @@ class Facebook extends \Providers\Provider {
 
     protected function _fetchItems() {
         $jsonContent = [];
-        exec('python3 ' . implode(DIRECTORY_SEPARATOR, [dirname(__FILE__), '..', 'bin', 'get-fb-content.py']) . ' ' . escapeshellarg($this->_feed), $jsonContent);
+        exec('direnv exec ' .
+             implode(DIRECTORY_SEPARATOR, [dirname(__FILE__), '..', 'bin', 'fb-scrape']) . ' ' .
+             'python ' . implode(DIRECTORY_SEPARATOR, [dirname(__FILE__), '..', 'bin', 'fb-scrape', 'get-fb-content.py']) . ' ' .
+             escapeshellarg($this->_feed), $jsonContent);
         return json_decode(implode(PHP_EOL, $jsonContent), TRUE);
     }
 
@@ -33,12 +36,14 @@ class Facebook extends \Providers\Provider {
                 if (!count($texts)) {
                     $texts[] = '';
                 }
-                $texts = array_merge(
-                    $texts,
-                    array_map(function($i) {
-                        return sprintf('<img src="%s" />', $i);
-                    }, $obj['images'])
-                );
+                if ($obj['images']) {
+                    $texts = array_merge(
+                        $texts,
+                        array_map(function($i) {
+                            return sprintf('<img src="%s" />', $i);
+                        }, $obj['images'])
+                    );
+                }
                 $item = new Item();
                 $item->ID = $obj['id'];
                 $item->Link = sprintf(
-- 
cgit v1.2.3