3 https://developers.google.com/+/plugins/snippet/
5 <meta itemprop="name" content="Toller Titel">
6 <meta itemprop="description" content="Eine tolle Beschreibung">
7 <meta itemprop="image" content="http://maple.libertreeproject.org/images/tree-icon.png">
9 <body itemscope itemtype="http://schema.org/Product">
10 <h1 itemprop="name">Shiny Trinket</h1>
11 <img itemprop="image" src="{image-url}" />
12 <p itemprop="description">Shiny trinkets are shiny.</p>
16 if(!function_exists('deletenode')) {
17 function deletenode(&$doc, $node)
19 $xpath = new DomXPath($doc);
20 $list = $xpath->query("//".$node);
21 foreach ($list as $child)
22 $child->parentNode->removeChild($child);
26 function completeurl($url, $scheme) {
27 $urlarr = parse_url($url);
29 if (isset($urlarr["scheme"]))
32 $schemearr = parse_url($scheme);
34 $complete = $schemearr["scheme"]."://".$schemearr["host"];
36 if (@$schemearr["port"] != "")
37 $complete .= ":".$schemearr["port"];
39 if(strpos($urlarr['path'],'/') !== 0)
42 $complete .= $urlarr["path"];
44 if (@$urlarr["query"] != "")
45 $complete .= "?".$urlarr["query"];
47 if (@$urlarr["fragment"] != "")
48 $complete .= "#".$urlarr["fragment"];
53 function parseurl_getsiteinfo($url, $no_guessing = false, $do_oembed = true, $count = 1) {
58 logger("parseurl_getsiteinfo: Endless loop detected for ".$url, LOGGER_DEBUG);
62 $url = trim($url, "'");
63 $url = trim($url, '"');
64 $siteinfo["url"] = $url;
65 $siteinfo["type"] = "link";
68 curl_setopt($ch, CURLOPT_URL, $url);
69 curl_setopt($ch, CURLOPT_HEADER, 1);
70 curl_setopt($ch, CURLOPT_NOBODY, 0);
71 curl_setopt($ch, CURLOPT_TIMEOUT, 3);
72 curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
73 //curl_setopt($ch, CURLOPT_FOLLOWLOCATION, 1);
74 //curl_setopt($ch,CURLOPT_USERAGENT,' Mozilla/5.0 (Macintosh; Intel Mac OS X 10.8; rv:24.0) Gecko/20100101 Firefox/24.0');
75 curl_setopt($ch,CURLOPT_USERAGENT, "Mozilla/5.0 (compatible; Friendica)");
77 $header = curl_exec($ch);
78 $curl_info = @curl_getinfo($ch);
79 $http_code = $curl_info['http_code'];
82 if ((($curl_info['http_code'] == "301") OR ($curl_info['http_code'] == "302") OR ($curl_info['http_code'] == "303") OR ($curl_info['http_code'] == "307"))
83 AND (($curl_info['redirect_url'] != "") OR ($curl_info['location'] != ""))) {
84 if ($curl_info['redirect_url'] != "")
85 $siteinfo = parseurl_getsiteinfo($curl_info['redirect_url'], $no_guessing, $do_oembed, ++$count);
87 $siteinfo = parseurl_getsiteinfo($curl_info['location'], $no_guessing, $do_oembed, ++$count);
92 require_once("include/oembed.php");
94 $oembed_data = oembed_fetch_url($url);
96 if ($oembed_data->type != "error")
97 $siteinfo["type"] = $oembed_data->type;
100 // Fetch the first mentioned charset. Can be in body or header
102 if (preg_match('/charset=(.*?)['."'".'"\s\n]/', $header, $matches))
103 $charset = trim(array_pop($matches));
108 $pos = strpos($header, "\r\n\r\n");
111 $body = trim(substr($header, $pos));
115 $body = mb_convert_encoding($body, "UTF-8", $charset);
116 $body = mb_convert_encoding($body, 'HTML-ENTITIES', "UTF-8");
118 $doc = new DOMDocument();
119 @$doc->loadHTML($body);
121 deletenode($doc, 'style');
122 deletenode($doc, 'script');
123 deletenode($doc, 'option');
124 deletenode($doc, 'h1');
125 deletenode($doc, 'h2');
126 deletenode($doc, 'h3');
127 deletenode($doc, 'h4');
128 deletenode($doc, 'h5');
129 deletenode($doc, 'h6');
130 deletenode($doc, 'ol');
131 deletenode($doc, 'ul');
133 $xpath = new DomXPath($doc);
135 $list = $xpath->query("//meta[@content]");
136 foreach ($list as $node) {
138 if ($node->attributes->length)
139 foreach ($node->attributes as $attribute)
140 $attr[$attribute->name] = $attribute->value;
142 if (@$attr["http-equiv"] == 'refresh') {
143 $path = $attr["content"];
144 $pathinfo = explode(";", $path);
146 foreach ($pathinfo AS $value) {
147 if (substr(strtolower($value), 0, 4) == "url=")
148 $content = substr($value, 4);
150 if ($content != "") {
151 $siteinfo = parseurl_getsiteinfo($content, $no_guessing, $do_oembed, ++$count);
157 //$list = $xpath->query("head/title");
158 $list = $xpath->query("//title");
159 foreach ($list as $node)
160 $siteinfo["title"] = html_entity_decode($node->nodeValue, ENT_QUOTES, "UTF-8");
162 //$list = $xpath->query("head/meta[@name]");
163 $list = $xpath->query("//meta[@name]");
164 foreach ($list as $node) {
166 if ($node->attributes->length)
167 foreach ($node->attributes as $attribute)
168 $attr[$attribute->name] = $attribute->value;
170 $attr["content"] = trim(html_entity_decode($attr["content"], ENT_QUOTES, "UTF-8"));
172 if ($attr["content"] != "")
173 switch (strtolower($attr["name"])) {
175 $siteinfo["title"] = $attr["content"];
178 $siteinfo["text"] = $attr["content"];
180 case "twitter:image":
181 $siteinfo["image"] = $attr["content"];
184 if (($siteinfo["type"] == "") OR ($attr["content"] == "photo"))
185 $siteinfo["type"] = $attr["content"];
187 case "twitter:description":
188 $siteinfo["text"] = $attr["content"];
190 case "twitter:title":
191 $siteinfo["title"] = $attr["content"];
194 $siteinfo["title"] = $attr["content"];
196 case "dc.description":
197 $siteinfo["text"] = $attr["content"];
200 if ($siteinfo["type"] == "summary")
201 $siteinfo["type"] = "link";
204 //$list = $xpath->query("head/meta[@property]");
205 $list = $xpath->query("//meta[@property]");
206 foreach ($list as $node) {
208 if ($node->attributes->length)
209 foreach ($node->attributes as $attribute)
210 $attr[$attribute->name] = $attribute->value;
212 $attr["content"] = trim(html_entity_decode($attr["content"], ENT_QUOTES, "UTF-8"));
214 if ($attr["content"] != "")
215 switch (strtolower($attr["property"])) {
217 $siteinfo["image"] = $attr["content"];
220 $siteinfo["title"] = $attr["content"];
222 case "og:description":
223 $siteinfo["text"] = $attr["content"];
228 if (isset($oembed_data) AND ($oembed_data->type == "link") AND ($siteinfo["type"] != "photo")) {
229 if (isset($oembed_data->title) AND (trim($oembed_data->title) != ""))
230 $siteinfo["title"] = $oembed_data->title;
231 if (isset($oembed_data->description) AND (trim($oembed_data->description) != ""))
232 $siteinfo["text"] = trim($oembed_data->description);
233 if (isset($oembed_data->thumbnail_url) AND (trim($oembed_data->thumbnail_url) != ""))
234 $siteinfo["image"] = $oembed_data->thumbnail_url;
237 if ((@$siteinfo["image"] == "") AND !$no_guessing) {
238 $list = $xpath->query("//img[@src]");
239 foreach ($list as $node) {
241 if ($node->attributes->length)
242 foreach ($node->attributes as $attribute)
243 $attr[$attribute->name] = $attribute->value;
245 $src = completeurl($attr["src"], $url);
246 $photodata = @getimagesize($src);
248 if (($photodata) && ($photodata[0] > 150) and ($photodata[1] > 150)) {
249 if ($photodata[0] > 300) {
250 $photodata[1] = round($photodata[1] * (300 / $photodata[0]));
253 if ($photodata[1] > 300) {
254 $photodata[0] = round($photodata[0] * (300 / $photodata[1]));
257 $siteinfo["images"][] = array("src"=>$src,
258 "width"=>$photodata[0],
259 "height"=>$photodata[1]);
264 $src = completeurl($siteinfo["image"], $url);
266 unset($siteinfo["image"]);
268 $photodata = @getimagesize($src);
270 if (($photodata) && ($photodata[0] > 10) and ($photodata[1] > 10))
271 $siteinfo["images"][] = array("src"=>$src,
272 "width"=>$photodata[0],
273 "height"=>$photodata[1]);
276 if ((@$siteinfo["text"] == "") AND (@$siteinfo["title"] != "") AND !$no_guessing) {
279 $list = $xpath->query("//div[@class='article']");
280 foreach ($list as $node)
281 if (strlen($node->nodeValue) > 40)
282 $text .= " ".trim($node->nodeValue);
285 $list = $xpath->query("//div[@class='content']");
286 foreach ($list as $node)
287 if (strlen($node->nodeValue) > 40)
288 $text .= " ".trim($node->nodeValue);
291 // If none text was found then take the paragraph content
293 $list = $xpath->query("//p");
294 foreach ($list as $node)
295 if (strlen($node->nodeValue) > 40)
296 $text .= " ".trim($node->nodeValue);
300 $text = trim(str_replace(array("\n", "\r"), array(" ", " "), $text));
302 while (strpos($text, " "))
303 $text = trim(str_replace(" ", " ", $text));
305 $siteinfo["text"] = trim(html_entity_decode(substr($text,0,350), ENT_QUOTES, "UTF-8").'...');
309 logger("parseurl_getsiteinfo: Siteinfo for ".$url." ".print_r($siteinfo, true), LOGGER_DEBUG);
314 function arr_add_hashes(&$item,$k) {
318 function parse_url_content(&$a) {
325 if(local_user() && (! feature_enabled(local_user(),'richtext')))
329 $br = (($textmode) ? "\n" : '<br />');
331 if(x($_GET,'binurl'))
332 $url = trim(hex2bin($_GET['binurl']));
334 $url = trim($_GET['url']);
337 $title = strip_tags(trim($_GET['title']));
339 if($_GET['description'])
340 $text = strip_tags(trim($_GET['description']));
343 $arr_tags = str_getcsv($_GET['tags']);
344 if(count($arr_tags)) {
345 array_walk($arr_tags,'arr_add_hashes');
346 $str_tags = $br . implode(' ',$arr_tags) . $br;
350 logger('parse_url: ' . $url);
353 $template = '[bookmark=%s]%s[/bookmark]%s';
355 $template = "<a class=\"bookmark\" href=\"%s\" >%s</a>%s";
357 $arr = array('url' => $url, 'text' => '');
359 call_hooks('parse_link', $arr);
361 if(strlen($arr['text'])) {
367 if($url && $title && $text) {
369 $title = str_replace(array("\r","\n"),array('',''),$title);
372 $text = '[quote]' . trim($text) . '[/quote]' . $br;
374 $text = '<blockquote>' . htmlspecialchars(trim($text)) . '</blockquote><br />';
375 $title = htmlspecialchars($title);
378 $result = sprintf($template,$url,($title) ? $title : $url,$text) . $str_tags;
380 logger('parse_url (unparsed): returns: ' . $result);
386 $siteinfo = parseurl_getsiteinfo($url);
388 $url= $siteinfo["url"];
392 if($siteinfo["title"] == "") {
393 $sitedata .= sprintf($template,$url,$url,'') . $str_tags;
396 $text = $siteinfo["text"];
397 $title = $siteinfo["title"];
402 if (($siteinfo["type"] != "video") AND (sizeof($siteinfo["images"]) > 0)){
403 /* Execute below code only if image is present in siteinfo */
406 $max_images = get_config('system','max_bookmark_images');
407 if($max_images === false)
410 $max_images = intval($max_images);
412 foreach ($siteinfo["images"] as $imagedata) {
414 $image .= '[img='.$imagedata["width"].'x'.$imagedata["height"].']'.$imagedata["src"].'[/img]' . "\n";
416 $image .= '<img height="'.$imagedata["height"].'" width="'.$imagedata["width"].'" src="'.$imagedata["src"].'" alt="photo" /><br />';
418 if($max_images && $max_images >= $total_images)
425 $text = '[quote]'.trim($text).'[/quote]';
427 $text = '<blockquote>'.htmlspecialchars(trim($text)).'</blockquote>';
431 $text = $br.$br.$image.$text;
435 $title = str_replace(array("\r","\n"),array('',''),$title);
437 $result = sprintf($template,$url,($title) ? $title : $url,$text) . $str_tags;
439 logger('parse_url: returns: ' . $result);
441 $sitedata .= trim($result);
443 if (($siteinfo["type"] == "video") AND ($url != ""))
444 echo "[class=type-video]".$sitedata."[/class]";
445 elseif (($siteinfo["type"] != "photo"))
446 echo "[class=type-link]".$sitedata."[/class]";
448 echo "[class=type-photo]".$title.$br.$image."[/class]";