harukin/core
Archived
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Merge branch 'patch-22' into 'dev'

Browse Source 
More precise content encoding detection on conversion

See merge request hubzilla/core!1363
This commit is contained in:
Mario 2018-11-03 13:21:22 +01:00
commit 34e3f3907a
1 changed files with 6 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
Zotlabs/Module/Linkinfo.php
View File

@ -228,8 +228,13 @@ class Linkinfo extends \Zotlabs\Web\Controller {
$header = $result['header'];
$body = $result['body'];
// Check codepage in HTTP headers or HTML if not exist
$cp = (preg_match('/Content-Type: text\/html; charset=(.+)\r\n/i', $header, $o) ? $o[1] : '');
if(empty($cp))
$cp = (preg_match('/meta.+content=["|\']text\/html; charset=([^"|\']+)/i', $body, $o) ? $o[1] : 'AUTO');
$body = mb_convert_encoding($body, 'UTF-8', (preg_match('/meta.+content=["|\']text\/html;\s+charset=([^"|\']+)/i', $body, $o) ? $o[1] : 'UTF-8'));
$body = mb_convert_encoding($body, 'UTF-8', $cp);
$body = mb_convert_encoding($body, 'HTML-ENTITIES', "UTF-8");
$doc = new \DOMDocument();