parse url: Emtpy values in meta headers are now ignored. Oembed data has priority. Videos are embedded with some further information.

This commit is contained in:
Michael Vogel 2014-04-04 10:58:31 +02:00
parent 96a41a16fd
commit 12866209e8

View file

@ -53,6 +53,9 @@ function completeurl($url, $scheme) {
function parseurl_getsiteinfo($url, $no_guessing = false) { function parseurl_getsiteinfo($url, $no_guessing = false) {
$siteinfo = array(); $siteinfo = array();
$url = trim($url, "'");
$url = trim($url, '"');
$siteinfo["url"] = $url;
$siteinfo["type"] = "link"; $siteinfo["type"] = "link";
$ch = curl_init(); $ch = curl_init();
@ -69,7 +72,7 @@ function parseurl_getsiteinfo($url, $no_guessing = false) {
$http_code = $curl_info['http_code']; $http_code = $curl_info['http_code'];
curl_close($ch); 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 ($curl_info['location'] != ""))) { AND (($curl_info['redirect_url'] != "") OR ($curl_info['location'] != ""))) {
if ($curl_info['redirect_url'] != "") if ($curl_info['redirect_url'] != "")
$siteinfo = parseurl_getsiteinfo($curl_info['redirect_url']); $siteinfo = parseurl_getsiteinfo($curl_info['redirect_url']);
@ -85,19 +88,12 @@ function parseurl_getsiteinfo($url, $no_guessing = false) {
if ($oembed_data->type != "error") if ($oembed_data->type != "error")
$siteinfo["type"] = $oembed_data->type; $siteinfo["type"] = $oembed_data->type;
if ($oembed_data->type == "link") {
if (isset($oembed_data->title))
$siteinfo["title"] = $oembed_data->title;
if (isset($oembed_data->description))
$siteinfo["text"] = trim($oembed_data->description);
if (isset($oembed_data->thumbnail_url))
$siteinfo["image"] = $oembed_data->thumbnail_url;
}
// Fetch the first mentioned charset. Can be in body or header // Fetch the first mentioned charset. Can be in body or header
$charset = "";
if (preg_match('/charset=(.*?)['."'".'"\s\n]/', $header, $matches)) if (preg_match('/charset=(.*?)['."'".'"\s\n]/', $header, $matches))
$charset = trim(array_pop($matches)); $charset = trim(array_pop($matches));
else
if ($charset == "")
$charset = "utf-8"; $charset = "utf-8";
$pos = strpos($header, "\r\n\r\n"); $pos = strpos($header, "\r\n\r\n");
@ -164,6 +160,7 @@ function parseurl_getsiteinfo($url, $no_guessing = false) {
$attr["content"] = trim(html_entity_decode($attr["content"], ENT_QUOTES, "UTF-8")); $attr["content"] = trim(html_entity_decode($attr["content"], ENT_QUOTES, "UTF-8"));
if ($attr["content"] != "")
switch (strtolower($attr["name"])) { switch (strtolower($attr["name"])) {
case "fulltitle": case "fulltitle":
$siteinfo["title"] = $attr["content"]; $siteinfo["title"] = $attr["content"];
@ -205,6 +202,7 @@ function parseurl_getsiteinfo($url, $no_guessing = false) {
$attr["content"] = trim(html_entity_decode($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"])) { switch (strtolower($attr["property"])) {
case "og:image": case "og:image":
$siteinfo["image"] = $attr["content"]; $siteinfo["image"] = $attr["content"];
@ -218,6 +216,15 @@ function parseurl_getsiteinfo($url, $no_guessing = false) {
} }
} }
if ($oembed_data->type == "link") {
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 !$no_guessing) { if ((@$siteinfo["image"] == "") AND !$no_guessing) {
$list = $xpath->query("//img[@src]"); $list = $xpath->query("//img[@src]");
foreach ($list as $node) { foreach ($list as $node) {
@ -290,6 +297,8 @@ function parseurl_getsiteinfo($url, $no_guessing = false) {
} }
} }
logger("parseurl_getsiteinfo: Siteinfo for ".$url." ".print_r($siteinfo, true), LOGGER_DEBUG);
return($siteinfo); return($siteinfo);
} }
@ -367,6 +376,8 @@ function parse_url_content(&$a) {
$siteinfo = parseurl_getsiteinfo($url); $siteinfo = parseurl_getsiteinfo($url);
$url= $siteinfo["url"];
$sitedata = ""; $sitedata = "";
if($siteinfo["title"] == "") { if($siteinfo["title"] == "") {
@ -379,7 +390,7 @@ function parse_url_content(&$a) {
$image = ""; $image = "";
if(sizeof($siteinfo["images"]) > 0){ if (($siteinfo["type"] != "video") AND (sizeof($siteinfo["images"]) > 0)){
/* Execute below code only if image is present in siteinfo */ /* Execute below code only if image is present in siteinfo */
$total_images = 0; $total_images = 0;
@ -421,7 +432,7 @@ function parse_url_content(&$a) {
$sitedata .= trim($result); $sitedata .= trim($result);
if (($siteinfo["type"] == "video") AND ($url != "")) if (($siteinfo["type"] == "video") AND ($url != ""))
echo "[video]".$url."[/video]"; echo "[class=type-video]".$sitedata."[/class]";
elseif (($siteinfo["type"] != "photo")) elseif (($siteinfo["type"] != "photo"))
echo "[class=type-link]".$sitedata."[/class]"; echo "[class=type-link]".$sitedata."[/class]";
else else