X-Git-Url: https://git.mxchange.org/?a=blobdiff_plain;f=mod%2Fparse_url.php;h=5048f4fa53903ac6ea90d74eb7d827ff6928cf54;hb=d4b2d3bf8a64e1fef8dbd88dbd625acbdef4a0df;hp=cb7268c1df83acd0e14bb9db4e0b1617346e2bc4;hpb=bae844e36767836ebf9dbfed10759dfd1a2fbbd6;p=friendica.git
diff --git a/mod/parse_url.php b/mod/parse_url.php
index cb7268c1df..5048f4fa53 100644
--- a/mod/parse_url.php
+++ b/mod/parse_url.php
@@ -50,8 +50,20 @@ function completeurl($url, $scheme) {
return($complete);
}
-function parseurl_getsiteinfo($url, $no_guessing = false) {
+function parseurl_getsiteinfo($url, $no_guessing = false, $do_oembed = true, $count = 1) {
+
$siteinfo = array();
+
+ if ($count > 10) {
+ logger("parseurl_getsiteinfo: Endless loop detected for ".$url, LOGGER_DEBUG);
+ return($siteinfo);
+ }
+
+ $url = trim($url, "'");
+ $url = trim($url, '"');
+ $siteinfo["url"] = $url;
+ $siteinfo["type"] = "link";
+
$ch = curl_init();
curl_setopt($ch, CURLOPT_URL, $url);
curl_setopt($ch, CURLOPT_HEADER, 1);
@@ -59,39 +71,37 @@ function parseurl_getsiteinfo($url, $no_guessing = false) {
curl_setopt($ch, CURLOPT_TIMEOUT, 3);
curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
//curl_setopt($ch, CURLOPT_FOLLOWLOCATION, 1);
- curl_setopt($ch,CURLOPT_USERAGENT,' Mozilla/5.0 (Macintosh; Intel Mac OS X 10.8; rv:24.0) Gecko/20100101 Firefox/24.0');
+ curl_setopt($ch,CURLOPT_USERAGENT, "Mozilla/5.0 (compatible; ".FRIENDICA_PLATFORM." ".FRIENDICA_VERSION."-".DB_UPDATE_VERSION.")");
$header = curl_exec($ch);
$curl_info = @curl_getinfo($ch);
$http_code = $curl_info['http_code'];
curl_close($ch);
- if ((($curl_info['http_code'] == "301") OR ($curl_info['http_code'] == "302"))
+ if ((($curl_info['http_code'] == "301") OR ($curl_info['http_code'] == "302") OR ($curl_info['http_code'] == "303") OR ($curl_info['http_code'] == "307"))
AND (($curl_info['redirect_url'] != "") OR ($curl_info['location'] != ""))) {
if ($curl_info['redirect_url'] != "")
- $siteinfo = parseurl_getsiteinfo($curl_info['redirect_url']);
+ $siteinfo = parseurl_getsiteinfo($curl_info['redirect_url'], $no_guessing, $do_oembed, ++$count);
else
- $siteinfo = parseurl_getsiteinfo($curl_info['location']);
+ $siteinfo = parseurl_getsiteinfo($curl_info['location'], $no_guessing, $do_oembed, ++$count);
return($siteinfo);
}
- require_once("include/oembed.php");
+ if ($do_oembed) {
+ require_once("include/oembed.php");
- $oembed_data = oembed_fetch_url($url);
+ $oembed_data = oembed_fetch_url($url);
- if ($oembed_data->type == "link") {
- if (isset($oembed_data->title))
- $siteinfo["title"] = $oembed_data->title;
- if (isset($oembed_data->description))
- $siteinfo["text"] = $oembed_data->description;
- if (isset($oembed_data->thumbnail_url))
- $siteinfo["image"] = $oembed_data->thumbnail_url;
+ if ($oembed_data->type != "error")
+ $siteinfo["type"] = $oembed_data->type;
}
// Fetch the first mentioned charset. Can be in body or header
+ $charset = "";
if (preg_match('/charset=(.*?)['."'".'"\s\n]/', $header, $matches))
- $charset = trim(array_pop($matches));
- else
+ $charset = trim(trim(trim(array_pop($matches)), ';,'));
+
+ if ($charset == "")
$charset = "utf-8";
$pos = strpos($header, "\r\n\r\n");
@@ -101,7 +111,12 @@ function parseurl_getsiteinfo($url, $no_guessing = false) {
else
$body = $header;
- $body = mb_convert_encoding($body, "UTF-8", $charset);
+ if (($charset != '') AND (strtoupper($charset) != "UTF-8")) {
+ logger("parseurl_getsiteinfo: detected charset ".$charset, LOGGER_DEBUG);
+ //$body = mb_convert_encoding($body, "UTF-8", $charset);
+ $body = iconv($charset, "UTF-8//TRANSLIT", $body);
+ }
+
$body = mb_convert_encoding($body, 'HTML-ENTITIES', "UTF-8");
$doc = new DOMDocument();
@@ -137,7 +152,7 @@ function parseurl_getsiteinfo($url, $no_guessing = false) {
$content = substr($value, 4);
}
if ($content != "") {
- $siteinfo = parseurl_getsiteinfo($content);
+ $siteinfo = parseurl_getsiteinfo($content, $no_guessing, $do_oembed, ++$count);
return($siteinfo);
}
}
@@ -156,22 +171,38 @@ function parseurl_getsiteinfo($url, $no_guessing = false) {
foreach ($node->attributes as $attribute)
$attr[$attribute->name] = $attribute->value;
- $attr["content"] = html_entity_decode($attr["content"], ENT_QUOTES, "UTF-8");
-
- switch (strtolower($attr["name"])) {
- case "fulltitle":
- $siteinfo["title"] = $attr["content"];
- break;
- case "description":
- $siteinfo["text"] = $attr["content"];
- break;
- case "dc.title":
- $siteinfo["title"] = $attr["content"];
- break;
- case "dc.description":
- $siteinfo["text"] = $attr["content"];
- break;
- }
+ $attr["content"] = trim(html_entity_decode($attr["content"], ENT_QUOTES, "UTF-8"));
+
+ if ($attr["content"] != "")
+ switch (strtolower($attr["name"])) {
+ case "fulltitle":
+ $siteinfo["title"] = $attr["content"];
+ break;
+ case "description":
+ $siteinfo["text"] = $attr["content"];
+ break;
+ case "twitter:image":
+ $siteinfo["image"] = $attr["content"];
+ break;
+ case "twitter:card":
+ if (($siteinfo["type"] == "") OR ($attr["content"] == "photo"))
+ $siteinfo["type"] = $attr["content"];
+ break;
+ case "twitter:description":
+ $siteinfo["text"] = $attr["content"];
+ break;
+ case "twitter:title":
+ $siteinfo["title"] = $attr["content"];
+ break;
+ case "dc.title":
+ $siteinfo["title"] = $attr["content"];
+ break;
+ case "dc.description":
+ $siteinfo["text"] = $attr["content"];
+ break;
+ }
+ if ($siteinfo["type"] == "summary")
+ $siteinfo["type"] = "link";
}
//$list = $xpath->query("head/meta[@property]");
@@ -182,19 +213,29 @@ function parseurl_getsiteinfo($url, $no_guessing = false) {
foreach ($node->attributes as $attribute)
$attr[$attribute->name] = $attribute->value;
- $attr["content"] = html_entity_decode($attr["content"], ENT_QUOTES, "UTF-8");
+ $attr["content"] = trim(html_entity_decode($attr["content"], ENT_QUOTES, "UTF-8"));
+
+ if ($attr["content"] != "")
+ switch (strtolower($attr["property"])) {
+ case "og:image":
+ $siteinfo["image"] = $attr["content"];
+ break;
+ case "og:title":
+ $siteinfo["title"] = $attr["content"];
+ break;
+ case "og:description":
+ $siteinfo["text"] = $attr["content"];
+ break;
+ }
+ }
- switch (strtolower($attr["property"])) {
- case "og:image":
- $siteinfo["image"] = $attr["content"];
- break;
- case "og:title":
- $siteinfo["title"] = $attr["content"];
- break;
- case "og:description":
- $siteinfo["text"] = $attr["content"];
- break;
- }
+ if (isset($oembed_data) AND ($oembed_data->type == "link") AND ($siteinfo["type"] != "photo")) {
+ if (isset($oembed_data->title) AND (trim($oembed_data->title) != ""))
+ $siteinfo["title"] = $oembed_data->title;
+ if (isset($oembed_data->description) AND (trim($oembed_data->description) != ""))
+ $siteinfo["text"] = trim($oembed_data->description);
+ if (isset($oembed_data->thumbnail_url) AND (trim($oembed_data->thumbnail_url) != ""))
+ $siteinfo["image"] = $oembed_data->thumbnail_url;
}
if ((@$siteinfo["image"] == "") AND !$no_guessing) {
@@ -265,10 +306,12 @@ function parseurl_getsiteinfo($url, $no_guessing = false) {
while (strpos($text, " "))
$text = trim(str_replace(" ", " ", $text));
- $siteinfo["text"] = html_entity_decode(substr($text,0,350), ENT_QUOTES, "UTF-8").'...';
+ $siteinfo["text"] = trim(html_entity_decode(substr($text,0,350), ENT_QUOTES, "UTF-8").'...');
}
}
+ logger("parseurl_getsiteinfo: Siteinfo for ".$url." ".print_r($siteinfo, true), LOGGER_DEBUG);
+
return($siteinfo);
}
@@ -311,9 +354,9 @@ function parse_url_content(&$a) {
logger('parse_url: ' . $url);
if($textmode)
- $template = $br . '[bookmark=%s]%s[/bookmark]%s' . $br;
+ $template = '[bookmark=%s]%s[/bookmark]%s';
else
- $template = "
%s%s
";
+ $template = "%s%s";
$arr = array('url' => $url, 'text' => '');
@@ -327,13 +370,15 @@ function parse_url_content(&$a) {
if($url && $title && $text) {
- if($textmode)
- $text = $br . '[quote]' . trim($text) . '[/quote]' . $br;
- else
- $text = '
' . trim($text) . '
' . htmlspecialchars(trim($text)) . '
'.trim($text).'
'.htmlspecialchars(trim($text)).''; } - if($image) { + if($image) $text = $br.$br.$image.$text; - } + else + $text = $br.$text; + $title = str_replace(array("\r","\n"),array('',''),$title); $result = sprintf($template,$url,($title) ? $title : $url,$text) . $str_tags; logger('parse_url: returns: ' . $result); - echo trim($result); + $sitedata .= trim($result); + + if (($siteinfo["type"] == "video") AND ($url != "")) + echo "[class=type-video]".$sitedata."[/class]"; + elseif (($siteinfo["type"] != "photo")) + echo "[class=type-link]".$sitedata."[/class]"; + else + echo "[class=type-photo]".$title.$br.$image."[/class]"; + killme(); } +?>