3 files changed, 75 insertions, 37 deletions
diff --git a/application/Url.php b/application/Url.php
index 23356f39..02a4395d 100644..100755
--- a/application/Url.php
+++ b/application/Url.php
@@ -81,6 +81,10 @@ class Url
    public function __construct($url)
    {
        $this->parts = parse_url($url);
+        if (!empty($url) && empty($this->parts['scheme'])) {
+            $this->parts['scheme'] = 'http';
+        }
    }
    /**
@@ -147,4 +151,16 @@ class Url
        $this->cleanupFragment();
        return $this->__toString();
    }
+    /**
+     * Get URL scheme.
+     *
+     * @return string the URL scheme or false if none is provided.
+     */
+    public function getScheme() {
+        if (!isset($this->parts['scheme'])) {
+            return false;
+        }
+        return $this->parts['scheme'];
+    }
 }
diff --git a/index.php b/index.php
index 8e04fa3e..ba70b29c 100755
--- a/index.php
+++ b/index.php
@@ -1485,51 +1485,57 @@ function renderPage()
        $url->cleanup();
        $link_is_new = false;
-        $link = $LINKSDB->getLinkFromUrl($url); // Check if URL is not already in database (in this case, we will edit the existing link)
+        // Check if URL is not already in database (in this case, we will edit the existing link)
+        $link = $LINKSDB->getLinkFromUrl((string)$url);
        if (!$link)
        {
-            $link_is_new = true;  // This is a new link
+            $link_is_new = true;
            $linkdate = strval(date('Ymd_His'));
-            $title = (empty($_GET['title']) ? '' : $_GET['title'] ); // Get title if it was provided in URL (by the bookmarklet).
+            // Get title if it was provided in URL (by the bookmarklet).
-            $description = (empty($_GET['description']) ? '' : $_GET['description']); // Get description if it was provided in URL (by the bookmarklet). [Bronco added that]
+            $title = (empty($_GET['title']) ? '' : $_GET['title'] );
-            $tags = (empty($_GET['tags']) ? '' : $_GET['tags'] ); // Get tags if it was provided in URL
+            // Get description if it was provided in URL (by the bookmarklet). [Bronco added that]
-            $private = (!empty($_GET['private']) && $_GET['private'] === "1" ? 1 : 0); // Get private if it was provided in URL
+            $description = (empty($_GET['description']) ? '' : $_GET['description']);
-            if (($url!='') && parse_url($url,PHP_URL_SCHEME)=='') $url = 'http://'.$url;
+            $tags = (empty($_GET['tags']) ? '' : $_GET['tags'] );
+            $private = (!empty($_GET['private']) && $_GET['private'] === "1" ? 1 : 0);
            // If this is an HTTP link, we try go get the page to extract the title (otherwise we will to straight to the edit form.)
-            if (empty($title) && parse_url($url,PHP_URL_SCHEME)=='http')
+            if (empty($title) && $url->getScheme() == 'http') {
-            {
                list($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)
+                if (strpos($status,'200 OK')!==false) {
-                                         {
+                    // Look for charset in html header.
-                        // Look for charset in html header.
+                    preg_match('#<meta .*charset=.*>#Usi', $data, $meta);
-                                                preg_match('#<meta .*charset=.*>#Usi', $data, $meta);
+                    // If found, extract encoding.
-                                                // If found, extract encoding.
+                    if (!empty($meta[0])) {
-                                                if (!empty($meta[0]))
+                        // Get encoding specified in header.
-                                                {
+                        preg_match('#charset="?(.*)"#si', $meta[0], $enc);
-                                                        // Get encoding specified in header.
+                        // If charset not found, use utf-8.
-                                                        preg_match('#charset="?(.*)"#si', $meta[0], $enc);
+                        $html_charset = (!empty($enc[1])) ? strtolower($enc[1]) : 'utf-8';
-                                                        // If charset not found, use utf-8.
+                    }
-                                                        $html_charset = (!empty($enc[1])) ? strtolower($enc[1]) : 'utf-8';
+                    else {
-                                                }
+                        $html_charset = 'utf-8';
-                                                else { $html_charset = 'utf-8'; }
+                    }
-                                                // Extract title
+                    // Extract title
-                                                $title = html_extract_title($data);
+                    $title = html_extract_title($data);
-                                                if (!empty($title))
+                    if (!empty($title)) {
-                                                {
+                        // Re-encode title in utf-8 if necessary.
-                                                        // Re-encode title in utf-8 if necessary.
+                        $title = ($html_charset == 'iso-8859-1') ? utf8_encode($title) : $title;
-                                                        $title = ($html_charset == 'iso-8859-1') ? utf8_encode($title) : $title;
+                    }
-                                                }
+                }
-                                        }
            }
-            if ($url=='') // In case of empty URL, this is just a text (with a link that points to itself)
+            if ($url == '') {
-            {
+                $url = '?' . smallHash($linkdate);
-                $url='?'.smallHash($linkdate);
+                $title = 'Note: ';
-                $title='Note: ';
            }
-            $link = array('linkdate'=>$linkdate,'title'=>$title,'url'=>$url,'description'=>$description,'tags'=>$tags,'private'=>$private);
+            $link = array(
+                'linkdate' => $linkdate,
+                'title' => $title,
+                'url' => (string)$url,
+                'description' => $description,
+                'tags' => $tags,
+                'private' => $private
+            );
        }
        $PAGE = new pageBuilder;
diff --git a/tests/UrlTest.php b/tests/UrlTest.php
index a39630f1..c848e88e 100644..100755
--- a/tests/UrlTest.php
+++ b/tests/UrlTest.php
@@ -151,4 +151,20 @@ class UrlTest extends PHPUnit_Framework_TestCase
            $url->cleanup()
        );
    }
+    /**
+     * Test default http scheme.
+     */
+    public function testDefaultScheme() {
+        $url = new Url(self::$baseUrl);
+        $this->assertEquals('http', $url->getScheme());
+        $url = new Url('domain.tld');
+        $this->assertEquals('http', $url->getScheme());
+        $url = new Url('ssh://domain.tld');
+        $this->assertEquals('ssh', $url->getScheme());
+        $url = new Url('ftp://domain.tld');
+        $this->assertEquals('ftp', $url->getScheme());
+        $url = new Url('git://domain.tld/push?pull=clone#checkout');
+        $this->assertEquals('git', $url->getScheme());
+    }
 }

diff --git a/application/Url.php b/application/Url.php index 23356f39..02a4395d 100644..100755 --- a/application/Url.php +++ b/application/Url.php
@@ -81,6 +81,10 @@ class Url
81	public function __construct($url)	81	public function __construct($url)
82	{	82	{
83	$this->parts = parse_url($url);	83	$this->parts = parse_url($url);
		84
		85	if (!empty($url) && empty($this->parts['scheme'])) {
		86	$this->parts['scheme'] = 'http';
		87	}
84	}	88	}
85		89
86	/**	90	/**
@@ -147,4 +151,16 @@ class Url
147	$this->cleanupFragment();	151	$this->cleanupFragment();
148	return $this->__toString();	152	return $this->__toString();
149	}	153	}
		154
		155	/**
		156	* Get URL scheme.
		157	*
		158	* @return string the URL scheme or false if none is provided.
		159	*/
		160	public function getScheme() {
		161	if (!isset($this->parts['scheme'])) {
		162	return false;
		163	}
		164	return $this->parts['scheme'];
		165	}
150	}	166	}


diff --git a/index.php b/index.php index 8e04fa3e..ba70b29c 100755 --- a/index.php +++ b/index.php
@@ -1485,51 +1485,57 @@ function renderPage()
1485	$url->cleanup();	1485	$url->cleanup();
1486		1486
1487	$link_is_new = false;	1487	$link_is_new = false;
1488	$link = $LINKSDB->getLinkFromUrl($url); // Check if URL is not already in database (in this case, we will edit the existing link)	1488	// Check if URL is not already in database (in this case, we will edit the existing link)
		1489	$link = $LINKSDB->getLinkFromUrl((string)$url);
1489	if (!$link)	1490	if (!$link)
1490	{	1491	{
1491	$link_is_new = true; // This is a new link	1492	$link_is_new = true;
1492	$linkdate = strval(date('Ymd_His'));	1493	$linkdate = strval(date('Ymd_His'));
1493	$title = (empty($_GET['title']) ? '' : $_GET['title'] ); // Get title if it was provided in URL (by the bookmarklet).	1494	// Get title if it was provided in URL (by the bookmarklet).
1494	$description = (empty($_GET['description']) ? '' : $_GET['description']); // Get description if it was provided in URL (by the bookmarklet). [Bronco added that]	1495	$title = (empty($_GET['title']) ? '' : $_GET['title'] );
1495	$tags = (empty($_GET['tags']) ? '' : $_GET['tags'] ); // Get tags if it was provided in URL	1496	// Get description if it was provided in URL (by the bookmarklet). [Bronco added that]
1496	$private = (!empty($_GET['private']) && $_GET['private'] === "1" ? 1 : 0); // Get private if it was provided in URL	1497	$description = (empty($_GET['description']) ? '' : $_GET['description']);
1497	if (($url!='') && parse_url($url,PHP_URL_SCHEME)=='') $url = 'http://'.$url;	1498	$tags = (empty($_GET['tags']) ? '' : $_GET['tags'] );
		1499	$private = (!empty($_GET['private']) && $_GET['private'] === "1" ? 1 : 0);
1498	// If this is an HTTP link, we try go get the page to extract the title (otherwise we will to straight to the edit form.)	1500	// If this is an HTTP link, we try go get the page to extract the title (otherwise we will to straight to the edit form.)
1499	if (empty($title) && parse_url($url,PHP_URL_SCHEME)=='http')	1501	if (empty($title) && $url->getScheme() == 'http') {
1500	{
1501	list($status,$headers,$data) = getHTTP($url,4); // Short timeout to keep the application responsive.	1502	list($status,$headers,$data) = getHTTP($url,4); // Short timeout to keep the application responsive.
1502	// FIXME: Decode charset according to specified in either 1) HTTP response headers or 2) <head> in html	1503	// FIXME: Decode charset according to specified in either 1) HTTP response headers or 2) <head> in html
1503	if (strpos($status,'200 OK')!==false)	1504	if (strpos($status,'200 OK')!==false) {
1504	{	1505	// Look for charset in html header.
1505	// Look for charset in html header.	1506	preg_match('#<meta .charset=.>#Usi', $data, $meta);
1506	preg_match('#<meta .charset=.>#Usi', $data, $meta);	1507
1507		1508	// If found, extract encoding.
1508	// If found, extract encoding.	1509	if (!empty($meta[0])) {
1509	if (!empty($meta[0]))	1510	// Get encoding specified in header.
1510	{	1511	preg_match('#charset="?(.*)"#si', $meta[0], $enc);
1511	// Get encoding specified in header.	1512	// If charset not found, use utf-8.
1512	preg_match('#charset="?(.*)"#si', $meta[0], $enc);	1513	$html_charset = (!empty($enc[1])) ? strtolower($enc[1]) : 'utf-8';
1513	// If charset not found, use utf-8.	1514	}
1514	$html_charset = (!empty($enc[1])) ? strtolower($enc[1]) : 'utf-8';	1515	else {
1515	}	1516	$html_charset = 'utf-8';
1516	else { $html_charset = 'utf-8'; }	1517	}
1517		1518
1518	// Extract title	1519	// Extract title
1519	$title = html_extract_title($data);	1520	$title = html_extract_title($data);
1520	if (!empty($title))	1521	if (!empty($title)) {
1521	{	1522	// Re-encode title in utf-8 if necessary.
1522	// Re-encode title in utf-8 if necessary.	1523	$title = ($html_charset == 'iso-8859-1') ? utf8_encode($title) : $title;
1523	$title = ($html_charset == 'iso-8859-1') ? utf8_encode($title) : $title;	1524	}
1524	}	1525	}
1525	}
1526	}	1526	}
1527	if ($url=='') // In case of empty URL, this is just a text (with a link that points to itself)	1527	if ($url == '') {
1528	{	1528	$url = '?' . smallHash($linkdate);
1529	$url='?'.smallHash($linkdate);	1529	$title = 'Note: ';
1530	$title='Note: ';
1531	}	1530	}
1532	$link = array('linkdate'=>$linkdate,'title'=>$title,'url'=>$url,'description'=>$description,'tags'=>$tags,'private'=>$private);	1531	$link = array(
		1532	'linkdate' => $linkdate,
		1533	'title' => $title,
		1534	'url' => (string)$url,
		1535	'description' => $description,
		1536	'tags' => $tags,
		1537	'private' => $private
		1538	);
1533	}	1539	}
1534		1540
1535	$PAGE = new pageBuilder;	1541	$PAGE = new pageBuilder;


diff --git a/tests/UrlTest.php b/tests/UrlTest.php index a39630f1..c848e88e 100644..100755 --- a/tests/UrlTest.php +++ b/tests/UrlTest.php
@@ -151,4 +151,20 @@ class UrlTest extends PHPUnit_Framework_TestCase
151	$url->cleanup()	151	$url->cleanup()
152	);	152	);
153	}	153	}
		154
		155	/**
		156	* Test default http scheme.
		157	*/
		158	public function testDefaultScheme() {
		159	$url = new Url(self::$baseUrl);
		160	$this->assertEquals('http', $url->getScheme());
		161	$url = new Url('domain.tld');
		162	$this->assertEquals('http', $url->getScheme());
		163	$url = new Url('ssh://domain.tld');
		164	$this->assertEquals('ssh', $url->getScheme());
		165	$url = new Url('ftp://domain.tld');
		166	$this->assertEquals('ftp', $url->getScheme());
		167	$url = new Url('git://domain.tld/push?pull=clone#checkout');
		168	$this->assertEquals('git', $url->getScheme());
		169	}
154	}	170	}