Merge pull request #326 from ArthurHoaro/bug-url

Fixes #325 - Shaarli does not recognize saved links
This commit is contained in:
VirtualTam 2015-08-31 20:31:41 +02:00
commit 6211c498f6
3 changed files with 74 additions and 36 deletions

16
application/Url.php Normal file → Executable file
View File

@ -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'];
}
}

View File

@ -1500,51 +1500,57 @@ 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).
$description = (empty($_GET['description']) ? '' : $_GET['description']); // Get description if it was provided in URL (by the bookmarklet). [Bronco added that]
$tags = (empty($_GET['tags']) ? '' : $_GET['tags'] ); // Get tags if it was provided in URL
$private = (!empty($_GET['private']) && $_GET['private'] === "1" ? 1 : 0); // Get private if it was provided in URL
if (($url!='') && parse_url($url,PHP_URL_SCHEME)=='') $url = 'http://'.$url;
// 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')
{
// Get title if it was provided in URL (by the bookmarklet).
$title = (empty($_GET['title']) ? '' : $_GET['title'] );
// Get description if it was provided in URL (by the bookmarklet). [Bronco added that]
$description = (empty($_GET['description']) ? '' : $_GET['description']);
$tags = (empty($_GET['tags']) ? '' : $_GET['tags'] );
$private = (!empty($_GET['private']) && $_GET['private'] === "1" ? 1 : 0);
// If this is an HTTP(S) link, we try go get the page to extract the title (otherwise we will to straight to the edit form.)
if (empty($title) && strpos($url->getScheme(), 'http') !== false) {
list($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)
{
// Look for charset in html header.
preg_match('#<meta .*charset=.*>#Usi', $data, $meta);
if (strpos($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 { $html_charset = 'utf-8'; }
// If found, extract encoding.
if (!empty($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;
}
}
// 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=='') // In case of empty URL, this is just a text (with a link that points to itself)
{
$url='?'.smallHash($linkdate);
$title='Note: ';
if ($url == '') {
$url = '?' . smallHash($linkdate);
$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;

16
tests/UrlTest.php Normal file → Executable file
View File

@ -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());
}
}