Shiny trinkets are shiny.
- +/** + * @file mod/parse_url.php + * + * @todo https://developers.google.com/+/plugins/snippet/ + * + * @verbatim + * + * + * + * + * + *Shiny trinkets are shiny.
+ * + * @endverbatim */ if(!function_exists('deletenode')) { @@ -50,8 +54,33 @@ function completeurl($url, $scheme) { return($complete); } +function parseurl_getsiteinfo_cached($url, $no_guessing = false, $do_oembed = true) { + + if ($url == "") + return false; + + $r = q("SELECT * FROM `parsed_url` WHERE `url` = '%s' AND `guessing` = %d AND `oembed` = %d", + dbesc(normalise_link($url)), intval(!$no_guessing), intval($do_oembed)); + + if ($r) + $data = $r[0]["content"]; + + if (!is_null($data)) { + $data = unserialize($data); + return $data; + } + + $data = parseurl_getsiteinfo($url, $no_guessing, $do_oembed); + + q("INSERT INTO `parsed_url` (`url`, `guessing`, `oembed`, `content`, `created`) VALUES ('%s', %d, %d, '%s', '%s')", + dbesc(normalise_link($url)), intval(!$no_guessing), intval($do_oembed), dbesc(serialize($data)), dbesc(datetime_convert())); + + return $data; +} + function parseurl_getsiteinfo($url, $no_guessing = false, $do_oembed = true, $count = 1) { require_once("include/network.php"); + require_once("include/Photo.php"); $a = get_app(); @@ -97,6 +126,14 @@ function parseurl_getsiteinfo($url, $no_guessing = false, $do_oembed = true, $co return($siteinfo); } + // if the file is too large then exit + if ($curl_info["download_content_length"] > 1000000) + return($siteinfo); + + // if it isn't a HTML file then exit + if (($curl_info["content_type"] != "") AND !strstr(strtolower($curl_info["content_type"]),"html")) + return($siteinfo); + if ($do_oembed) { require_once("include/oembed.php"); @@ -104,15 +141,16 @@ function parseurl_getsiteinfo($url, $no_guessing = false, $do_oembed = true, $co if ($oembed_data->type != "error") $siteinfo["type"] = $oembed_data->type; - } - // if the file is too large then exit - if ($curl_info["download_content_length"] > 1000000) - return($siteinfo); - - // if it isn't a HTML file then exit - if (($curl_info["content_type"] != "") AND !strstr(strtolower($curl_info["content_type"]),"html")) - return($siteinfo); + if (($oembed_data->type == "link") AND ($siteinfo["type"] != "photo")) { + 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; + } + } $stamp1 = microtime(true); @@ -256,7 +294,8 @@ function parseurl_getsiteinfo($url, $no_guessing = false, $do_oembed = true, $co if (isset($keywords)) { $siteinfo["keywords"] = array(); foreach ($keywords as $keyword) - $siteinfo["keywords"][] = trim($keyword); + if (!in_array(trim($keyword), $siteinfo["keywords"])) + $siteinfo["keywords"][] = trim($keyword); } //$list = $xpath->query("head/meta[@property]"); @@ -283,15 +322,6 @@ function parseurl_getsiteinfo($url, $no_guessing = false, $do_oembed = true, $co } } - if (isset($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) { $list = $xpath->query("//img[@src]"); foreach ($list as $node) { @@ -301,7 +331,7 @@ function parseurl_getsiteinfo($url, $no_guessing = false, $do_oembed = true, $co $attr[$attribute->name] = $attribute->value; $src = completeurl($attr["src"], $url); - $photodata = @getimagesize($src); + $photodata = get_photo_info($src); if (($photodata) && ($photodata[0] > 150) and ($photodata[1] > 150)) { if ($photodata[0] > 300) { @@ -318,12 +348,12 @@ function parseurl_getsiteinfo($url, $no_guessing = false, $do_oembed = true, $co } } - } else { + } elseif ($siteinfo["image"] != "") { $src = completeurl($siteinfo["image"], $url); unset($siteinfo["image"]); - $photodata = @getimagesize($src); + $photodata = get_photo_info($src); if (($photodata) && ($photodata[0] > 10) and ($photodata[1] > 10)) $siteinfo["images"][] = array("src"=>$src, @@ -407,6 +437,15 @@ function parse_url_content(&$a) { } } + // add url scheme if missing + $arrurl = parse_url($url); + if (!x($arrurl, 'scheme')) { + if (x($arrurl, 'host')) + $url = "http:".$url; + else + $url = "http://".$url; + } + logger('parse_url: ' . $url); if($textmode) @@ -462,10 +501,7 @@ function parse_url_content(&$a) { $sitedata = ""; - if($siteinfo["title"] == "") { - $sitedata .= sprintf($template,$url,$url,'') . $str_tags; - killme(); - } else { + if($siteinfo["title"] != "") { $text = $siteinfo["text"]; $title = $siteinfo["title"]; }