From e84a7e8c2a9a26040d77f5e33382cfa470cbaf0a Mon Sep 17 00:00:00 2001
From: Michael <heluecht@pirati.ca>
Date: Sat, 30 Sep 2023 15:51:53 +0000
Subject: [PATCH] Additional languages / new hook for additional language
 detection

---
 doc/Addons.md      |  8 ++++++++
 src/Core/L10n.php  | 30 ++++++++++++++++++++++++------
 src/Model/Item.php | 14 ++++++++++++--
 3 files changed, 44 insertions(+), 8 deletions(-)

diff --git a/doc/Addons.md b/doc/Addons.md
index 567e5b275..497717e6f 100644
--- a/doc/Addons.md
+++ b/doc/Addons.md
@@ -221,6 +221,13 @@ Please note: body contents are bbcode - not HTML
 Called when receiving a post from another source. This may also be used to post local activity or system generated messages.
 `$b` is the item array of information to be stored in the database and the item body is bbcode.
 
+### get_language
+Called after the language detection. This can be used for alternative language detection methods.
+`$data` is an array:
+
+- **text**: The text that is analyzed.
+- **detected**: The array with the original language detection. Can be overwritten via an addon.
+
 ### addon_settings
 Called when generating the HTML for the addon settings page.
 `$data` is an array containing:
@@ -800,6 +807,7 @@ Here is a complete list of all hook callbacks with file locations (as of 24-Sep-
 
 ### src/Model/Item.php
 
+    Hook::callAll('get_language', $item);
     Hook::callAll('post_local', $item);
     Hook::callAll('post_remote', $item);
     Hook::callAll('post_local_end', $posted_item);
diff --git a/src/Core/L10n.php b/src/Core/L10n.php
index 636cb1190..7fd7fc4e8 100644
--- a/src/Core/L10n.php
+++ b/src/Core/L10n.php
@@ -397,13 +397,31 @@ class L10n
 				// See https://github.com/friendica/friendica/issues/10511
 				// Persian is manually added to language detection until a persian translation is provided for the interface, at
 				// which point it will be automatically available through `getAvailableLanguages()` and this should be removed.
-				// Additionally Portuguese, Ukrainian, traditional Chinese and Welsh are added to that list.
+				// Additionally some more languages are added to that list that are used in the Fediverse.
 				$additional_langs = [
-					'cy'      => 'Cymraeg',
-					'uk'      => 'Українська',
-					'pt-PT'   => 'Português',
-					'zh-hant' => '繁體',
-					'fa'      => 'فارسی'
+					'af'         => 'Afrikaans',
+					'cy'         => 'Cymraeg',
+					'el-monoton' => 'Ελληνικά',
+					'eu'         => 'euskara',
+					'fa'         => 'فارسی',
+					'gl'         => 'Galego',
+					'hi'         => 'हिन्दी',
+					'hr'         => 'Hrvatski',
+					'id'         => 'bahasa Indonesia',
+					'ko'         => '한국인',
+					'lt'         => 'lietuvių',
+					'lv'         => 'latviešu',
+					'sk'         => 'slovenský',
+					'sl'         => 'Slovenščina',
+					'sw'         => 'Kiswahili',
+					'th'         => 'แบบไทย',
+					'tl'         => 'Wikang Tagalog',
+					'tr'         => 'Türkçe',
+					'pt-PT'      => 'Português',
+					'uk'         => 'Українська',
+					'uz'         => 'Ўзбек',
+					'vi'         => 'Tiếng Việt',
+					'zh-hant'    => '繁體',
 				];
 				$langs = array_merge($additional_langs, $langs);
 				ksort($langs);
diff --git a/src/Model/Item.php b/src/Model/Item.php
index 41aae0413..c529c2c16 100644
--- a/src/Model/Item.php
+++ b/src/Model/Item.php
@@ -2017,7 +2017,7 @@ class Item
 		$naked_body = BBCode::toPlaintext($naked_body);
 
 		// Remove possibly remaining links
-		$naked_body = preg_replace(Strings::autoLinkRegEx(), '', $naked_body);
+		$naked_body = trim(preg_replace(Strings::autoLinkRegEx(), '', $naked_body));
 
 		if (empty($naked_body)) {
 			return [];
@@ -2029,7 +2029,17 @@ class Item
 		$availableLanguages = DI::l10n()->convertForLanguageDetection($availableLanguages);
 
 		$ld = new Language(array_keys($availableLanguages));
-		return $ld->detect($naked_body)->limit(0, $count)->close() ?: [];
+		$languages = $ld->detect($naked_body)->limit(0, $count)->close() ?: [];
+
+		$data = [
+			'text'     => $naked_body,
+			'detected' => $languages,
+		];
+
+		Hook::callAll('get_language', $data);
+		$languages = $data['detected'];
+
+		return $languages;
 	}
 
 	/**