X-Git-Url: https://git.mxchange.org/?a=blobdiff_plain;f=mod%2Fparse_url.php;h=a0608f636340cfe8e88b41f5c1bae123204974f6;hb=84189ac775b3e1bc36d9e9d8945624c1bca7fb55;hp=4d894969aadbde21c152ac87895da09aba6915d9;hpb=02a1fc9cd08fba2168895d1892a91d8143323848;p=friendica.git diff --git a/mod/parse_url.php b/mod/parse_url.php index 4d894969aa..a0608f6363 100644 --- a/mod/parse_url.php +++ b/mod/parse_url.php @@ -1,4 +1,18 @@ + + + + +

Shiny Trinket

Shiny trinkets are shiny.

+ +*/ + if(!function_exists('deletenode')) { function deletenode(&$doc, $node) { @@ -19,38 +33,67 @@ function completeurl($url, $scheme) { $complete = $schemearr["scheme"]."://".$schemearr["host"]; - if ($schemearr["port"] != "") + if (@$schemearr["port"] != "") $complete .= ":".$schemearr["port"]; + if(strpos($urlarr['path'],'/') !== 0) + $complete .= '/'; + $complete .= $urlarr["path"]; - if ($urlarr["query"] != "") + if (@$urlarr["query"] != "") $complete .= "?".$urlarr["query"]; - if ($urlarr["fragment"] != "") + if (@$urlarr["fragment"] != "") $complete .= "#".$urlarr["fragment"]; return($complete); } -function parseurl_getsiteinfo($url) { +function parseurl_getsiteinfo($url, $no_guessing = false) { $siteinfo = array(); + $url = trim($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); 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'); + //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'); $header = curl_exec($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") 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']); + else + $siteinfo = parseurl_getsiteinfo($curl_info['location']); + return($siteinfo); + } + + require_once("include/oembed.php"); + + $oembed_data = oembed_fetch_url($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 + + if ($charset == "") $charset = "utf-8"; $pos = strpos($header, "\r\n\r\n"); @@ -80,6 +123,28 @@ function parseurl_getsiteinfo($url) { $xpath = new DomXPath($doc); + $list = $xpath->query("//meta[@content]"); + foreach ($list as $node) { + $attr = array(); + if ($node->attributes->length) + foreach ($node->attributes as $attribute) + $attr[$attribute->name] = $attribute->value; + + if (@$attr["http-equiv"] == 'refresh') { + $path = $attr["content"]; + $pathinfo = explode(";", $path); + $content = ""; + foreach ($pathinfo AS $value) { + if (substr(strtolower($value), 0, 4) == "url=") + $content = substr($value, 4); + } + if ($content != "") { + $siteinfo = parseurl_getsiteinfo($content); + return($siteinfo); + } + } + } + //$list = $xpath->query("head/title"); $list = $xpath->query("//title"); foreach ($list as $node) @@ -93,22 +158,38 @@ function parseurl_getsiteinfo($url) { 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"] == "") + $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 = $xpath->query("head/meta[@property]"); @@ -119,39 +200,49 @@ function parseurl_getsiteinfo($url) { 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"] = $attr["content"]; - break; - } + if ($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"] == "") { - $list = $xpath->query("//img[@src]"); - foreach ($list as $node) { - $attr = array(); - if ($node->attributes->length) - foreach ($node->attributes as $attribute) - $attr[$attribute->name] = $attribute->value; + if ((@$siteinfo["image"] == "") AND !$no_guessing) { + $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); + $photodata = @getimagesize($src); - if (($photodata[0] > 150) and ($photodata[1] > 150)) { + if (($photodata) && ($photodata[0] > 150) and ($photodata[1] > 150)) { if ($photodata[0] > 300) { - $photodata[1] = $photodata[1] * (300 / $photodata[0]); + $photodata[1] = round($photodata[1] * (300 / $photodata[0])); $photodata[0] = 300; } if ($photodata[1] > 300) { - $photodata[0] = $photodata[0] * (300 / $photodata[1]); + $photodata[0] = round($photodata[0] * (300 / $photodata[1])); $photodata[1] = 300; } $siteinfo["images"][] = array("src"=>$src, @@ -159,18 +250,21 @@ function parseurl_getsiteinfo($url) { "height"=>$photodata[1]); } - } - } else { + } + } else { $src = completeurl($siteinfo["image"], $url); - $photodata = getimagesize($src); - if (($photodata[0] > 10) and ($photodata[1] > 10)) + 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 ($siteinfo["text"] == "") { + if ((@$siteinfo["text"] == "") AND (@$siteinfo["title"] != "") AND !$no_guessing) { $text = ""; $list = $xpath->query("//div[@class='article']"); @@ -199,10 +293,12 @@ function parseurl_getsiteinfo($url) { 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($siteinfo); } @@ -216,7 +312,8 @@ function parse_url_content(&$a) { $str_tags = ''; $textmode = false; - if(local_user() && intval(get_pconfig(local_user(),'system','plaintext'))) + + if(local_user() && (! feature_enabled(local_user(),'richtext'))) $textmode = true; //if($textmode) @@ -244,9 +341,9 @@ 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' => $url, 'text' => ''); @@ -260,13 +357,15 @@ function parse_url_content(&$a) { if($url && $title && $text) { - if($textmode) - $text = $br . '[quote]' . trim($text) . '[/quote]' . $br; - else - $text = '

' . trim($text) . '

'; - $title = str_replace(array("\r","\n"),array('',''),$title); + if($textmode) + $text = '[quote]' . trim($text) . '[/quote]' . $br; + else { + $text = '

' . htmlspecialchars(trim($text)) . '

'; + $title = htmlspecialchars($title); + } + $result = sprintf($template,$url,($title) ? $title : $url,$text) . $str_tags; logger('parse_url (unparsed): returns: ' . $result); @@ -277,9 +376,12 @@ function parse_url_content(&$a) { $siteinfo = parseurl_getsiteinfo($url); + $url= $siteinfo["url"]; + + $sitedata = ""; + if($siteinfo["title"] == "") { - echo print_r($siteinfo, true); - //echo sprintf($template,$url,$url,'') . $str_tags; + $sitedata .= sprintf($template,$url,$url,'') . $str_tags; killme(); } else { $text = $siteinfo["text"]; @@ -288,55 +390,53 @@ function parse_url_content(&$a) { $image = ""; - foreach ($siteinfo["images"] as $imagedata) - if($textmode) - $image .= '[img='.$imagedata["width"].'x'.$imagedata["height"].']'.$imagedata["src"].'[/img]'; + if (($siteinfo["type"] != "video") AND (sizeof($siteinfo["images"]) > 0)){ + /* Execute below code only if image is present in siteinfo */ + + $total_images = 0; + $max_images = get_config('system','max_bookmark_images'); + if($max_images === false) + $max_images = 2; else - $image .= '

'; - -/* if ($image != "") { - $i = fetch_url($image); - if($i) { - require_once('include/Photo.php'); - // guess mimetype from headers or filename - $type = guess_image_type($image,true); - - $ph = new Photo($i, $type); - if($ph->is_valid()) { - if($ph->getWidth() > 300 || $ph->getHeight() > 300) { - $ph->scaleImage(300); - $new_width = $ph->getWidth(); - $new_height = $ph->getHeight(); - if($textmode) - $image = $br . $br . '[img=' . $new_width . 'x' . $new_height . ']' . $image . '[/img]'; - else - $image = '

'; - } else { - if($textmode) - $image = $br.$br.'[img]'.$image.'[/img]'; - else - $image = '

'; - } - } + $max_images = intval($max_images); + + foreach ($siteinfo["images"] as $imagedata) { + if($textmode) + $image .= '[img='.$imagedata["width"].'x'.$imagedata["height"].']'.$imagedata["src"].'[/img]' . "\n"; + else + $image .= '

'; + $total_images ++; + if($max_images && $max_images >= $total_images) + break; } - }*/ + } if(strlen($text)) { if($textmode) - $text = $br.'[quote]'.trim($text).'[/quote]'.$br ; + $text = '[quote]'.trim($text).'[/quote]'; else - $text = '

'.trim($text).'

'; + $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(); }