Fixed #73 - Can't Poch url with special caracter

This commit is contained in:
nicosomb 2013-04-23 15:25:12 +02:00
parent 3e7188185d
commit 2987031bc3
2 changed files with 19 additions and 18 deletions

View file

@ -16,6 +16,7 @@ if (!is_dir('db/')) {
define ('ABS_PATH', 'assets/');
define ('CONVERT_LINKS_FOOTNOTES', TRUE);
define ('REVERT_FORCED_PARAGRAPH_ELEMENTS',FALSE);
define ('DOWNLOAD_PICTURES', TRUE);
$storage_type = 'sqlite'; # sqlite or file

View file

@ -39,9 +39,9 @@ function get_external_file($url)
curl_setopt($curl, CURLOPT_RETURNTRANSFER, true);
curl_setopt($curl, CURLOPT_HEADER, false);
// FOR SSL do not verified certificate
// FOR SSL do not verified certificate
curl_setopt($curl, CURLOPT_SSL_VERIFYPEER, FALSE);
curl_setopt($curl, CURLOPT_AUTOREFERER, TRUE );
curl_setopt($curl, CURLOPT_AUTOREFERER, TRUE );
// FeedBurner requires a proper USER-AGENT...
curl_setopt($curl, CURL_HTTP_VERSION_1_1, true);
@ -59,14 +59,14 @@ function get_external_file($url)
// create http context and add timeout and user-agent
$context = stream_context_create(array(
'http'=>array('timeout' => $timeout,
'header'=> "User-Agent: ".$useragent, /*spoot Mozilla Firefox*/
'follow_location' => true),
// FOR SSL do not verified certificate
'ssl' => array('verify_peer' => false,
'allow_self_signed' => true)
)
);
'http'=>array('timeout' => $timeout,
'header'=> "User-Agent: ".$useragent, /*spoot Mozilla Firefox*/
'follow_location' => true),
// FOR SSL do not verified certificate
'ssl' => array('verify_peer' => false,
'allow_self_signed' => true)
)
);
// only download page lesser than 4MB
$data = @file_get_contents($url, false, $context, -1, 4000000); // We download at most 4 MB from source.
@ -122,19 +122,19 @@ function prepare_url($url)
$i=strpos($url,'#xtor=RSS-'); if ($i!==false) $url=substr($url,0,$i);
$title = $url;
$html = Encoding::toUTF8(get_external_file($url,15));
// If get_external_file if not able to retrieve HTTPS content try the same URL with HTTP protocol
if (!preg_match('!^https?://!i', $url) && (!isset($html) || strlen($html) <= 0)) {
$url = 'http://' . $url;
$html = Encoding::toUTF8(get_external_file($url,15));
}
$html = Encoding::toUTF8(get_external_file($url,15));
// If get_external_file if not able to retrieve HTTPS content try the same URL with HTTP protocol
if (!preg_match('!^https?://!i', $url) && (!isset($html) || strlen($html) <= 0)) {
$url = 'http://' . $url;
$html = Encoding::toUTF8(get_external_file($url,15));
}
if (isset($html) and strlen($html) > 0)
{
$r = new Readability($html, $url);
$r->convertLinksToFootnotes = CONVERT_LINKS_FOOTNOTES;
$r->revertForcedParagraphElements = REVERT_FORCED_PARAGRAPH_ELEMENTS;
$r->revertForcedParagraphElements = REVERT_FORCED_PARAGRAPH_ELEMENTS;
if($r->init())
{