1 files changed, 23 insertions, 2 deletions
diff --git a/index.php b/index.php
index 8b5c912a..d88f471c 100644
--- a/index.php
+++ b/index.php
@@ -1545,8 +1545,29 @@ function renderPage()
            {
                list($status,$headers,$data) = getHTTP($url,4); // Short timeout to keep the application responsive.
                // FIXME: Decode charset according to specified in either 1) HTTP response headers or 2) <head> in html
-                if (strpos($status,'200 OK')!==false) $title=html_entity_decode(html_extract_title($data),ENT_QUOTES,'UTF-8');
+                if (strpos($status,'200 OK')!==false)
+                                         {
+                        // Look for charset in html header.
+                                                preg_match('#<meta .*charset=.*>#Usi', $data, $meta);
+ 
+                                                // If found, extract encoding.
+                                                if (!empty($meta[0]))
+                                                {
+                                                        // Get encoding specified in header.
+                                                        preg_match('#charset="?(.*)"#si', $meta[0], $enc);
+                                                        // If charset not found, use utf-8.
+                                                        $html_charset = (!empty($enc[1])) ? strtolower($enc[1]) : 'utf-8';
+                                                }
+                                                else { $html_charset = 'utf-8'; }
+ 
+                                                // Extract title
+                                                $title = html_extract_title($data);
+                                                if (!empty($title))
+                                                {
+                                                        // Re-encode title in utf-8 if necessary.
+                                                        $title = ($html_charset == 'iso-8859-1') ? utf8_encode($title) : $title;
+                                                }
+                                        }
            }
            if ($url=='') $url='?'.smallHash($linkdate); // In case of empty URL, this is just a text (with a link that point to itself)
            $link = array('linkdate'=>$linkdate,'title'=>$title,'url'=>$url,'description'=>$description,'tags'=>$tags,'private'=>0);

diff --git a/index.php b/index.php index 8b5c912a..d88f471c 100644 --- a/index.php +++ b/index.php
@@ -1545,8 +1545,29 @@ function renderPage()
1545	{	1545	{
1546	list($status,$headers,$data) = getHTTP($url,4); // Short timeout to keep the application responsive.	1546	list($status,$headers,$data) = getHTTP($url,4); // Short timeout to keep the application responsive.
1547	// FIXME: Decode charset according to specified in either 1) HTTP response headers or 2) <head> in html	1547	// FIXME: Decode charset according to specified in either 1) HTTP response headers or 2) <head> in html
1548	if (strpos($status,'200 OK')!==false) $title=html_entity_decode(html_extract_title($data),ENT_QUOTES,'UTF-8');	1548	if (strpos($status,'200 OK')!==false)
1549		1549	{
		1550	// Look for charset in html header.
		1551	preg_match('#<meta .charset=.>#Usi', $data, $meta);
		1552
		1553	// If found, extract encoding.
		1554	if (!empty($meta[0]))
		1555	{
		1556	// Get encoding specified in header.
		1557	preg_match('#charset="?(.*)"#si', $meta[0], $enc);
		1558	// If charset not found, use utf-8.
		1559	$html_charset = (!empty($enc[1])) ? strtolower($enc[1]) : 'utf-8';
		1560	}
		1561	else { $html_charset = 'utf-8'; }
		1562
		1563	// Extract title
		1564	$title = html_extract_title($data);
		1565	if (!empty($title))
		1566	{
		1567	// Re-encode title in utf-8 if necessary.
		1568	$title = ($html_charset == 'iso-8859-1') ? utf8_encode($title) : $title;
		1569	}
		1570	}
1550	}	1571	}
1551	if ($url=='') $url='?'.smallHash($linkdate); // In case of empty URL, this is just a text (with a link that point to itself)	1572	if ($url=='') $url='?'.smallHash($linkdate); // In case of empty URL, this is just a text (with a link that point to itself)
1552	$link = array('linkdate'=>$linkdate,'title'=>$title,'url'=>$url,'description'=>$description,'tags'=>$tags,'private'=>0);	1573	$link = array('linkdate'=>$linkdate,'title'=>$title,'url'=>$url,'description'=>$description,'tags'=>$tags,'private'=>0);