diff options
Diffstat (limited to 'Zotlabs/Module/Linkinfo.php')
| -rw-r--r-- | Zotlabs/Module/Linkinfo.php | 44 |
1 files changed, 36 insertions, 8 deletions
diff --git a/Zotlabs/Module/Linkinfo.php b/Zotlabs/Module/Linkinfo.php index a0ad17e68..7c7dc0e88 100644 --- a/Zotlabs/Module/Linkinfo.php +++ b/Zotlabs/Module/Linkinfo.php @@ -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(); @@ -265,20 +270,43 @@ class Linkinfo extends \Zotlabs\Web\Controller { $attr["content"] = html_entity_decode($attr["content"], ENT_QUOTES, "UTF-8"); switch (strtolower($attr["name"])) { - case 'generator': - $siteinfo['generator'] = $attr['content']; - break; case "fulltitle": - $siteinfo["title"] = $attr["content"]; + $siteinfo["title"] = trim($attr["content"]); break; case "description": - $siteinfo["text"] = $attr["content"]; + $siteinfo["text"] = trim($attr["content"]); + break; + case "thumbnail": + $siteinfo["image"] = $attr["content"]; + break; + case "twitter:image": + $siteinfo["image"] = $attr["content"]; + break; + case "twitter:image:src": + $siteinfo["image"] = $attr["content"]; + break; + case "twitter:card": + if (($siteinfo["type"] == "") || ($attr["content"] == "photo")) { + $siteinfo["type"] = $attr["content"]; + } + break; + case "twitter:description": + $siteinfo["text"] = trim($attr["content"]); + break; + case "twitter:title": + $siteinfo["title"] = trim($attr["content"]); break; case "dc.title": - $siteinfo["title"] = $attr["content"]; + $siteinfo["title"] = trim($attr["content"]); break; case "dc.description": - $siteinfo["text"] = $attr["content"]; + $siteinfo["text"] = trim($attr["content"]); + break; + case "keywords": + $keywords = explode(",", $attr["content"]); + break; + case "news_keywords": + $keywords = explode(",", $attr["content"]); break; } } |