From 1d02214e1253b86622262988ad7ecce1f5647006 Mon Sep 17 00:00:00 2001
From: Dag <me@dvikan.no>
Date: Sat, 4 Jan 2025 19:43:48 +0100
Subject: [PATCH] feat: extract simple_html_dom max_file_size to config (#4395)

---
 bridges/YoutubeBridge.php             | 7 -------
 config.default.ini.php                | 9 +++++++++
 lib/bootstrap.php                     | 4 ----
 lib/simplehtmldom/simple_html_dom.php | 5 +++--
 4 files changed, 12 insertions(+), 13 deletions(-)

diff --git a/bridges/YoutubeBridge.php b/bridges/YoutubeBridge.php
index 647b1c42..12cdaec4 100644
--- a/bridges/YoutubeBridge.php
+++ b/bridges/YoutubeBridge.php
@@ -1,12 +1,5 @@
 <?php
 
-/**
-* RssBridgeYoutube
-* Returns the newest videos
-* WARNING: to parse big playlists (over ~90 videos), you need to edit simple_html_dom.php:
-* change: define('MAX_FILE_SIZE', 600000);
-* into:   define('MAX_FILE_SIZE', 900000);  (or more)
-*/
 class YoutubeBridge extends BridgeAbstract
 {
     const NAME = 'YouTube Bridge';
diff --git a/config.default.ini.php b/config.default.ini.php
index 1045d6c3..25f35508 100644
--- a/config.default.ini.php
+++ b/config.default.ini.php
@@ -46,7 +46,11 @@ enable_debug_mode = false
 ; Whether to enable maintenance mode. If enabled, feed requests receive 503 Service Unavailable
 enable_maintenance_mode = false
 
+; Max file size for simple_html_dom in bytes (10000000 => 10 MB)
+max_file_size = 10000000
+
 [http]
+
 ; Operation timeout in seconds
 timeout = 15
 
@@ -70,6 +74,7 @@ type = "file"
 custom_timeout = false
 
 [admin]
+
 ; Advertise an email address where people can reach the administrator.
 ; This address is displayed on the main page, visible to everyone!
 ; ""    = Disabled (default)
@@ -86,6 +91,7 @@ telegram = ""
 donations = true
 
 [proxy]
+
 ; The HTTP proxy to tunnel requests through
 ; https://curl.se/libcurl/c/CURLOPT_PROXY.html
 ; ""    = Proxy disabled (default)
@@ -135,6 +141,7 @@ report_limit = 1
 ; --- Cache specific configuration ---------------------------------------------
 
 [FileCache]
+
 ; The root folder to store files in.
 ; "" = Use the cache folder in the repository (default)
 path = ""
@@ -142,6 +149,7 @@ path = ""
 enable_purge = true
 
 [SQLiteCache]
+
 ; Filepath of the sqlite db file
 file = "cache.sqlite"
 ; Whether to actually delete data when purging
@@ -150,6 +158,7 @@ enable_purge = true
 timeout = 5000
 
 [MemcachedCache]
+
 host = "localhost"
 port = 11211
 
diff --git a/lib/bootstrap.php b/lib/bootstrap.php
index 36b13e19..8a7c62a1 100644
--- a/lib/bootstrap.php
+++ b/lib/bootstrap.php
@@ -7,10 +7,6 @@ if (is_file(__DIR__ . '/../vendor/autoload.php')) {
 const PATH_LIB_CACHES = __DIR__ . '/../caches/';
 const PATH_CACHE = __DIR__ . '/../cache/';
 
-// Allow larger files for simple_html_dom
-// todo: extract to config (if possible)
-const MAX_FILE_SIZE = 10000000;
-
 // Files
 $files = [
     __DIR__ . '/../lib/html.php',
diff --git a/lib/simplehtmldom/simple_html_dom.php b/lib/simplehtmldom/simple_html_dom.php
index 170f6fb0..e8b3a727 100644
--- a/lib/simplehtmldom/simple_html_dom.php
+++ b/lib/simplehtmldom/simple_html_dom.php
@@ -114,8 +114,9 @@ function str_get_html(
     if (empty($str)) {
         throw new \Exception('Refusing to parse empty string input');
     }
-    if (strlen($str) > MAX_FILE_SIZE) {
-        throw new \Exception('Refusing to parse too big input');
+
+    if (strlen($str) > Configuration::getConfig('system', 'max_file_size')) {
+        throw new \Exception('simple_html_dom: Refusing to parse too big input: ' . strlen($str));
     }
 
 	return $dom->load($str, $lowercase, $stripRN);