From: ArthurHoaro <arthur@hoa.ro>
Date: Sat, 26 Sep 2020 11:28:38 +0000 (+0200)
Subject: Fix warning if the encoding retrieved from external headers is invalid
X-Git-Tag: v0.12.0-beta-1~3^2
X-Git-Url: https://git.immae.eu/?a=commitdiff_plain;h=refs%2Fpull%2F1569%2Fhead;p=github%2Fshaarli%2FShaarli.git

Fix warning if the encoding retrieved from external headers is invalid

Also fixed the regex to support this failing header: charset="utf-8"\r\n"
---

diff --git a/application/bookmark/LinkUtils.php b/application/bookmark/LinkUtils.php
index 68914fca..e7af4d55 100644
--- a/application/bookmark/LinkUtils.php
+++ b/application/bookmark/LinkUtils.php
@@ -26,7 +26,7 @@ function html_extract_title($html)
  */
 function header_extract_charset($header)
 {
-    preg_match('/charset="?([^; ]+)/i', $header, $match);
+    preg_match('/charset=["\']?([^; "\']+)/i', $header, $match);
     if (! empty($match[1])) {
         return strtolower(trim($match[1]));
     }
diff --git a/application/front/controller/admin/ManageShaareController.php b/application/front/controller/admin/ManageShaareController.php
index ca2da9b5..ffb0dae4 100644
--- a/application/front/controller/admin/ManageShaareController.php
+++ b/application/front/controller/admin/ManageShaareController.php
@@ -69,7 +69,7 @@ class ManageShaareController extends ShaarliAdminController
                         $retrieveDescription
                     )
                 );
-                if (! empty($title) && strtolower($charset) !== 'utf-8') {
+                if (! empty($title) && strtolower($charset) !== 'utf-8' && mb_check_encoding($charset)) {
                     $title = mb_convert_encoding($title, 'utf-8', $charset);
                 }
             }
diff --git a/tests/bookmark/LinkUtilsTest.php b/tests/bookmark/LinkUtilsTest.php
index 7d4a7b89..0d07897b 100644
--- a/tests/bookmark/LinkUtilsTest.php
+++ b/tests/bookmark/LinkUtilsTest.php
@@ -42,6 +42,19 @@ class LinkUtilsTest extends TestCase
         $this->assertEquals(strtolower($charset), header_extract_charset($headers));
     }
 
+    /**
+     * Test headers_extract_charset() when the charset is found with odd quotes.
+     */
+    public function testHeadersExtractExistentCharsetWithQuotes()
+    {
+        $charset = 'x-MacCroatian';
+        $headers = 'text/html; charset="' . $charset . '"otherstuff="test"';
+        $this->assertEquals(strtolower($charset), header_extract_charset($headers));
+
+        $headers = 'text/html; charset=\'' . $charset . '\'otherstuff="test"';
+        $this->assertEquals(strtolower($charset), header_extract_charset($headers));
+    }
+
     /**
      * Test headers_extract_charset() when the charset is not found.
      */