X-Git-Url: https://git.mxchange.org/?a=blobdiff_plain;ds=inline;f=mod%2Fparse_url.php;h=9adee8f65cd8d7d9c261f6d16522ea2ed0859cd9;hb=15ede093cf6dc61b306d43c7fce99c1336623c56;hp=27dac4d5d318565483a392f3a0a7170e2dae4868;hpb=cb05801a9031254f5882038de07a3ee4d5f89dd0;p=friendica.git diff --git a/mod/parse_url.php b/mod/parse_url.php index 27dac4d5d3..9adee8f65c 100644 --- a/mod/parse_url.php +++ b/mod/parse_url.php @@ -1,7 +1,230 @@ + + + +
+Shiny trinkets are shiny.
+ +*/ + +if(!function_exists('deletenode')) { + function deletenode(&$doc, $node) + { + $xpath = new DomXPath($doc); + $list = $xpath->query("//".$node); + foreach ($list as $child) + $child->parentNode->removeChild($child); + } +} + +function completeurl($url, $scheme) { + $urlarr = parse_url($url); + + if (isset($urlarr["scheme"])) + return($url); + + $schemearr = parse_url($scheme); + + $complete = $schemearr["scheme"]."://".$schemearr["host"]; + + if ($schemearr["port"] != "") + $complete .= ":".$schemearr["port"]; + + if(strpos($urlarr['path'],'/') !== 0) + $complete .= '/'; + + $complete .= $urlarr["path"]; + + if ($urlarr["query"] != "") + $complete .= "?".$urlarr["query"]; + + if ($urlarr["fragment"] != "") + $complete .= "#".$urlarr["fragment"]; + + return($complete); +} + +function parseurl_getsiteinfo($url) { + $siteinfo = array(); + + $ch = curl_init(); + curl_setopt($ch, CURLOPT_URL, $url); + curl_setopt($ch, CURLOPT_HEADER, 1); + curl_setopt($ch, CURLOPT_NOBODY, 0); + curl_setopt($ch, CURLOPT_TIMEOUT, 3); + curl_setopt($ch, CURLOPT_RETURNTRANSFER, true); + curl_setopt($ch,CURLOPT_USERAGENT,'Opera/9.64(Windows NT 5.1; U; de) Presto/2.1.1'); + + $header = curl_exec($ch); + curl_close($ch); + + // Fetch the first mentioned charset. Can be in body or header + if (preg_match('/charset=(.*?)['."'".'"\s\n]/', $header, $matches)) + $charset = trim(array_pop($matches)); + else + $charset = "utf-8"; + + $pos = strpos($header, "\r\n\r\n"); + + if ($pos) + $body = trim(substr($header, $pos)); + else + $body = $header; + + $body = mb_convert_encoding($body, "UTF-8", $charset); + $body = mb_convert_encoding($body, 'HTML-ENTITIES', "UTF-8"); + + $doc = new DOMDocument(); + @$doc->loadHTML($body); + + deletenode($doc, 'style'); + deletenode($doc, 'script'); + deletenode($doc, 'option'); + deletenode($doc, 'h1'); + deletenode($doc, 'h2'); + deletenode($doc, 'h3'); + deletenode($doc, 'h4'); + deletenode($doc, 'h5'); + deletenode($doc, 'h6'); + deletenode($doc, 'ol'); + deletenode($doc, 'ul'); + + $xpath = new DomXPath($doc); + + //$list = $xpath->query("head/title"); + $list = $xpath->query("//title"); + foreach ($list as $node) + $siteinfo["title"] = html_entity_decode($node->nodeValue, ENT_QUOTES, "UTF-8"); + + //$list = $xpath->query("head/meta[@name]"); + $list = $xpath->query("//meta[@name]"); + foreach ($list as $node) { + $attr = array(); + if ($node->attributes->length) + 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"] = $attr["content"]; + break; + } + } + + //$list = $xpath->query("head/meta[@property]"); + $list = $xpath->query("//meta[@property]"); + foreach ($list as $node) { + $attr = array(); + if ($node->attributes->length) + foreach ($node->attributes as $attribute) + $attr[$attribute->name] = $attribute->value; + + $attr["content"] = html_entity_decode($attr["content"], ENT_QUOTES, "UTF-8"); + + 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"] = $attr["content"]; + break; + } + } + + if ($siteinfo["image"] == "") { + $list = $xpath->query("//img[@src]"); + foreach ($list as $node) { + $attr = array(); + if ($node->attributes->length) + foreach ($node->attributes as $attribute) + $attr[$attribute->name] = $attribute->value; + + $src = completeurl($attr["src"], $url); + $photodata = @getimagesize($src); + + if (($photodata) && ($photodata[0] > 150) and ($photodata[1] > 150)) { + if ($photodata[0] > 300) { + $photodata[1] = round($photodata[1] * (300 / $photodata[0])); + $photodata[0] = 300; + } + if ($photodata[1] > 300) { + $photodata[0] = round($photodata[0] * (300 / $photodata[1])); + $photodata[1] = 300; + } + $siteinfo["images"][] = array("src"=>$src, + "width"=>$photodata[0], + "height"=>$photodata[1]); + } + + } + } else { + $src = completeurl($siteinfo["image"], $url); + + unset($siteinfo["image"]); + + $photodata = @getimagesize($src); + + if (($photodata) && ($photodata[0] > 10) and ($photodata[1] > 10)) + $siteinfo["images"][] = array("src"=>$src, + "width"=>$photodata[0], + "height"=>$photodata[1]); + } + + if ($siteinfo["text"] == "") { + $text = ""; + + $list = $xpath->query("//div[@class='article']"); + foreach ($list as $node) + if (strlen($node->nodeValue) > 40) + $text .= " ".trim($node->nodeValue); + + if ($text == "") { + $list = $xpath->query("//div[@class='content']"); + foreach ($list as $node) + if (strlen($node->nodeValue) > 40) + $text .= " ".trim($node->nodeValue); + } + + // If none text was found then take the paragraph content + if ($text == "") { + $list = $xpath->query("//p"); + foreach ($list as $node) + if (strlen($node->nodeValue) > 40) + $text .= " ".trim($node->nodeValue); + } + + if ($text != "") { + $text = trim(str_replace(array("\n", "\r"), array(" ", " "), $text)); + + while (strpos($text, " ")) + $text = trim(str_replace(" ", " ", $text)); + + $siteinfo["text"] = html_entity_decode(substr($text,0,350), ENT_QUOTES, "UTF-8").'...'; + } + } -require_once('library/HTML5/Parser.php'); -require_once('library/HTMLPurifier.auto.php'); + return($siteinfo); +} function arr_add_hashes(&$item,$k) { $item = '#' . $item; @@ -16,8 +239,8 @@ function parse_url_content(&$a) { if(local_user() && intval(get_pconfig(local_user(),'system','plaintext'))) $textmode = true; - if($textmode) - $br = (($textmode) ? "\n" : '' . $text . '
' . trim($text) . '