friendica
/
friendica
mirror of https://github.com/friendica/friendica
6
0
Fork 1
Code Issues Packages Projects Releases 1 Wiki Activity

Fix formatting in Util/ParseUrl 

- Remove commented out code
- Replace double quotes with simple quotes
- Remove parentheses for language construct
- Add spaces around operators
- Rename variables
This commit is contained in:
Hypolite Petovan 2018-09-02 17:35:55 -04:00
parent 69e7c7feca
commit 8bcc3c6860
1 changed files with 132 additions and 133 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

265
src/Util/ParseUrl.php
View File

@ -118,13 +118,13 @@ class ParseUrl
// Check if the URL does contain a scheme // Check if the URL does contain a scheme
$scheme = parse_url($url, PHP_URL_SCHEME); $scheme = parse_url($url, PHP_URL_SCHEME);
if ($scheme == "") { if ($scheme == '') {
$url = "http://".trim($url, "/"); $url = 'http://' . trim($url, '/');
} }
if ($count > 10) { if ($count > 10) {
logger("parseurl_getsiteinfo: Endless loop detected for ".$url, LOGGER_DEBUG); logger('Endless loop detected for ' . $url, LOGGER_DEBUG);
return($siteinfo); return $siteinfo;
} }
$url = trim($url, "'"); $url = trim($url, "'");
@ -132,220 +132,217 @@ class ParseUrl
$url = Network::stripTrackingQueryParams($url); $url = Network::stripTrackingQueryParams($url);
$siteinfo["url"] = $url; $siteinfo['url'] = $url;
$siteinfo["type"] = "link"; $siteinfo['type'] = 'link';
$data = Network::curl($url); $data = Network::curl($url);
if (!$data['success']) { if (!$data['success']) {
return($siteinfo); return $siteinfo;
} }
// If the file is too large then exit // If the file is too large then exit
if ($data["info"]["download_content_length"] > 1000000) { if ($data['info']['download_content_length'] > 1000000) {
return($siteinfo); return $siteinfo;
} }
// If it isn't a HTML file then exit // If it isn't a HTML file then exit
if (($data["info"]["content_type"] != "") && !strstr(strtolower($data["info"]["content_type"]), "html")) { if (($data['info']['content_type'] != '') && !strstr(strtolower($data['info']['content_type']), 'html')) {
return($siteinfo); return $siteinfo;
} }
$header = $data["header"]; $header = $data['header'];
$body = $data["body"]; $body = $data['body'];
if ($do_oembed) { if ($do_oembed) {
$oembed_data = OEmbed::fetchURL($url); $oembed_data = OEmbed::fetchURL($url);
if (!empty($oembed_data->type)) { if (!empty($oembed_data->type)) {
if (!in_array($oembed_data->type, ["error", "rich", ""])) { if (!in_array($oembed_data->type, ['error', 'rich', ''])) {
$siteinfo["type"] = $oembed_data->type; $siteinfo['type'] = $oembed_data->type;
} }
if (($oembed_data->type == "link") && ($siteinfo["type"] != "photo")) { if (($oembed_data->type == 'link') && ($siteinfo['type'] != 'photo')) {
if (isset($oembed_data->title)) { if (isset($oembed_data->title)) {
$siteinfo["title"] = trim($oembed_data->title); $siteinfo['title'] = trim($oembed_data->title);
} }
if (isset($oembed_data->description)) { if (isset($oembed_data->description)) {
$siteinfo["text"] = trim($oembed_data->description); $siteinfo['text'] = trim($oembed_data->description);
} }
if (isset($oembed_data->thumbnail_url)) { if (isset($oembed_data->thumbnail_url)) {
$siteinfo["image"] = $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 = ""; $charset = '';
if (preg_match('/charset=(.*?)['."'".'"\s\n]/', $header, $matches)) { if (preg_match('/charset=(.*?)[\'"\s\n]/', $header, $matches)) {
$charset = trim(trim(trim(array_pop($matches)), ';,')); $charset = trim(trim(trim(array_pop($matches)), ';,'));
} }
if ($charset == "") { if ($charset == '') {
$charset = "utf-8"; $charset = 'utf-8';
} }
if (($charset != "") && (strtoupper($charset) != "UTF-8")) { if (($charset != '') && (strtoupper($charset) != 'UTF-8')) {
logger("parseurl_getsiteinfo: detected charset ".$charset, LOGGER_DEBUG); logger('detected charset ' . $charset, LOGGER_DEBUG);
//$body = mb_convert_encoding($body, "UTF-8", $charset); $body = iconv($charset, 'UTF-8//TRANSLIT', $body);
$body = iconv($charset, "UTF-8//TRANSLIT", $body);
} }
$body = mb_convert_encoding($body, 'HTML-ENTITIES', "UTF-8"); $body = mb_convert_encoding($body, 'HTML-ENTITIES', 'UTF-8');
$doc = new DOMDocument(); $doc = new DOMDocument();
@$doc->loadHTML($body); @$doc->loadHTML($body);
XML::deleteNode($doc, "style"); XML::deleteNode($doc, 'style');
XML::deleteNode($doc, "script"); XML::deleteNode($doc, 'script');
XML::deleteNode($doc, "option"); XML::deleteNode($doc, 'option');
XML::deleteNode($doc, "h1"); XML::deleteNode($doc, 'h1');
XML::deleteNode($doc, "h2"); XML::deleteNode($doc, 'h2');
XML::deleteNode($doc, "h3"); XML::deleteNode($doc, 'h3');
XML::deleteNode($doc, "h4"); XML::deleteNode($doc, 'h4');
XML::deleteNode($doc, "h5"); XML::deleteNode($doc, 'h5');
XML::deleteNode($doc, "h6"); XML::deleteNode($doc, 'h6');
XML::deleteNode($doc, "ol"); XML::deleteNode($doc, 'ol');
XML::deleteNode($doc, "ul"); XML::deleteNode($doc, 'ul');
$xpath = new DOMXPath($doc); $xpath = new DOMXPath($doc);
$list = $xpath->query("//meta[@content]"); $list = $xpath->query('//meta[@content]');
foreach ($list as $node) { foreach ($list as $node) {
$attr = []; $meta_tag = [];
if ($node->attributes->length) { if ($node->attributes->length) {
foreach ($node->attributes as $attribute) { foreach ($node->attributes as $attribute) {
$attr[$attribute->name] = $attribute->value; $meta_tag[$attribute->name] = $attribute->value;
} }
} }
if (@$attr["http-equiv"] == "refresh") { if (@$meta_tag['http-equiv'] == 'refresh') {
$path = $attr["content"]; $path = $meta_tag['content'];
$pathinfo = explode(";", $path); $pathinfo = explode(';', $path);
$content = ""; $content = '';
foreach ($pathinfo as $value) { foreach ($pathinfo as $value) {
if (substr(strtolower($value), 0, 4) == "url=") { if (substr(strtolower($value), 0, 4) == 'url=') {
$content = substr($value, 4); $content = substr($value, 4);
} }
} }
if ($content != "") { if ($content != '') {
$siteinfo = self::getSiteinfo($content, $no_guessing, $do_oembed, ++$count); $siteinfo = self::getSiteinfo($content, $no_guessing, $do_oembed, ++$count);
return($siteinfo); return $siteinfo;
} }
} }
} }
$list = $xpath->query("//title"); $list = $xpath->query('//title');
if ($list->length > 0) { if ($list->length > 0) {
$siteinfo["title"] = trim($list->item(0)->nodeValue); $siteinfo['title'] = trim($list->item(0)->nodeValue);
} }
//$list = $xpath->query("head/meta[@name]"); $list = $xpath->query('//meta[@name]');
$list = $xpath->query("//meta[@name]");
foreach ($list as $node) { foreach ($list as $node) {
$attr = []; $meta_tag = [];
if ($node->attributes->length) { if ($node->attributes->length) {
foreach ($node->attributes as $attribute) { foreach ($node->attributes as $attribute) {
$attr[$attribute->name] = $attribute->value; $meta_tag[$attribute->name] = $attribute->value;
} }
} }
if (!empty($attr["content"])) { if (!empty($meta_tag['content'])) {
$attr["content"] = trim(html_entity_decode($attr["content"], ENT_QUOTES, "UTF-8")); $meta_tag['content'] = trim(html_entity_decode($meta_tag['content'], ENT_QUOTES, 'UTF-8'));
switch (strtolower($attr["name"])) { switch (strtolower($meta_tag['name'])) {
case "fulltitle": case 'fulltitle':
$siteinfo["title"] = trim($attr["content"]); $siteinfo['title'] = trim($meta_tag['content']);
break; break;
case "description": case 'description':
$siteinfo["text"] = trim($attr["content"]); $siteinfo['text'] = trim($meta_tag['content']);
break; break;
case "thumbnail": case 'thumbnail':
$siteinfo["image"] = $attr["content"]; $siteinfo['image'] = $meta_tag['content'];
break; break;
case "twitter:image": case 'twitter:image':
$siteinfo["image"] = $attr["content"]; $siteinfo['image'] = $meta_tag['content'];
break; break;
case "twitter:image:src": case 'twitter:image:src':
$siteinfo["image"] = $attr["content"]; $siteinfo['image'] = $meta_tag['content'];
break; break;
case "twitter:card": case 'twitter:card':
if (($siteinfo["type"] == "") || ($attr["content"] == "photo")) { if (($siteinfo['type'] == '') || ($meta_tag['content'] == 'photo')) {
$siteinfo["type"] = $attr["content"]; $siteinfo['type'] = $meta_tag['content'];
} }
break; break;
case "twitter:description": case 'twitter:description':
$siteinfo["text"] = trim($attr["content"]); $siteinfo['text'] = trim($meta_tag['content']);
break; break;
case "twitter:title": case 'twitter:title':
$siteinfo["title"] = trim($attr["content"]); $siteinfo['title'] = trim($meta_tag['content']);
break; break;
case "dc.title": case 'dc.title':
$siteinfo["title"] = trim($attr["content"]); $siteinfo['title'] = trim($meta_tag['content']);
break; break;
case "dc.description": case 'dc.description':
$siteinfo["text"] = trim($attr["content"]); $siteinfo['text'] = trim($meta_tag['content']);
break; break;
case "keywords": case 'keywords':
$keywords = explode(",", $attr["content"]); $keywords = explode(',', $meta_tag['content']);
break; break;
case "news_keywords": case 'news_keywords':
$keywords = explode(",", $attr["content"]); $keywords = explode(',', $meta_tag['content']);
break; break;
} }
} }
if ($siteinfo["type"] == "summary") { if ($siteinfo['type'] == 'summary') {
$siteinfo["type"] = "link"; $siteinfo['type'] = 'link';
} }
} }
if (isset($keywords)) { if (isset($keywords)) {
$siteinfo["keywords"] = []; $siteinfo['keywords'] = [];
foreach ($keywords as $keyword) { foreach ($keywords as $keyword) {
if (!in_array(trim($keyword), $siteinfo["keywords"])) { if (!in_array(trim($keyword), $siteinfo['keywords'])) {
$siteinfo["keywords"][] = trim($keyword); $siteinfo['keywords'][] = trim($keyword);
} }
} }
} }
//$list = $xpath->query("head/meta[@property]"); $list = $xpath->query('//meta[@property]');
$list = $xpath->query("//meta[@property]");
foreach ($list as $node) { foreach ($list as $node) {
$attr = []; $meta_tag = [];
if ($node->attributes->length) { if ($node->attributes->length) {
foreach ($node->attributes as $attribute) { foreach ($node->attributes as $attribute) {
$attr[$attribute->name] = $attribute->value; $meta_tag[$attribute->name] = $attribute->value;
} }
} }
if (!empty($attr["content"])) { if (!empty($meta_tag['content'])) {
$attr["content"] = trim(html_entity_decode($attr["content"], ENT_QUOTES, "UTF-8")); $meta_tag['content'] = trim(html_entity_decode($meta_tag['content'], ENT_QUOTES, 'UTF-8'));
switch (strtolower($attr["property"])) { switch (strtolower($meta_tag['property'])) {
case "og:image": case 'og:image':
$siteinfo["image"] = $attr["content"]; $siteinfo['image'] = $meta_tag['content'];
break; break;
case "og:title": case 'og:title':
$siteinfo["title"] = trim($attr["content"]); $siteinfo['title'] = trim($meta_tag['content']);
break; break;
case "og:description": case 'og:description':
$siteinfo["text"] = trim($attr["content"]); $siteinfo['text'] = trim($meta_tag['content']);
break; break;
} }
} }
} }
if ((@$siteinfo["image"] == "") && !$no_guessing) { if ((@$siteinfo['image'] == '') && !$no_guessing) {
$list = $xpath->query("//img[@src]"); $list = $xpath->query('//img[@src]');
foreach ($list as $node) { foreach ($list as $node) {
$attr = []; $img_tag = [];
if ($node->attributes->length) { if ($node->attributes->length) {
foreach ($node->attributes as $attribute) { foreach ($node->attributes as $attribute) {
$attr[$attribute->name] = $attribute->value; $img_tag[$attribute->name] = $attribute->value;
} }
} }
$src = self::completeUrl($attr["src"], $url); $src = self::completeUrl($img_tag['src'], $url);
$photodata = Image::getInfoFromURL($src); $photodata = Image::getInfoFromURL($src);
if (($photodata) && ($photodata[0] > 150) && ($photodata[1] > 150)) { if (($photodata) && ($photodata[0] > 150) && ($photodata[1] > 150)) {
@ -357,70 +354,72 @@ class ParseUrl
$photodata[0] = round($photodata[0] * (300 / $photodata[1])); $photodata[0] = round($photodata[0] * (300 / $photodata[1]));
$photodata[1] = 300; $photodata[1] = 300;
} }
$siteinfo["images"][] = ["src" => $src, $siteinfo['images'][] = [
"width" => $photodata[0], 'src' => $src,
"height" => $photodata[1]]; 'width' => $photodata[0],
'height' => $photodata[1]
];
} }
} }
} elseif (!empty($siteinfo["image"])) { } elseif (!empty($siteinfo['image'])) {
$src = self::completeUrl($siteinfo["image"], $url); $src = self::completeUrl($siteinfo['image'], $url);
unset($siteinfo["image"]); unset($siteinfo['image']);
$photodata = Image::getInfoFromURL($src); $photodata = Image::getInfoFromURL($src);
if (($photodata) && ($photodata[0] > 10) && ($photodata[1] > 10)) { if (($photodata) && ($photodata[0] > 10) && ($photodata[1] > 10)) {
$siteinfo["images"][] = ["src" => $src, $siteinfo['images'][] = ['src' => $src,
"width" => $photodata[0], 'width' => $photodata[0],
"height" => $photodata[1]]; 'height' => $photodata[1]];
} }
} }
if ((@$siteinfo["text"] == "") && (@$siteinfo["title"] != "") && !$no_guessing) { if ((@$siteinfo['text'] == '') && (@$siteinfo['title'] != '') && !$no_guessing) {
$text = ""; $text = '';
$list = $xpath->query("//div[@class='article']"); $list = $xpath->query('//div[@class="article"]');
foreach ($list as $node) { foreach ($list as $node) {
if (strlen($node->nodeValue) > 40) { if (strlen($node->nodeValue) > 40) {
$text .= " ".trim($node->nodeValue); $text .= ' ' . trim($node->nodeValue);
} }
} }
if ($text == "") { if ($text == '') {
$list = $xpath->query("//div[@class='content']"); $list = $xpath->query('//div[@class="content"]');
foreach ($list as $node) { foreach ($list as $node) {
if (strlen($node->nodeValue) > 40) { if (strlen($node->nodeValue) > 40) {
$text .= " ".trim($node->nodeValue); $text .= ' ' . trim($node->nodeValue);
} }
} }
} }
// If none text was found then take the paragraph content // If none text was found then take the paragraph content
if ($text == "") { if ($text == '') {
$list = $xpath->query("//p"); $list = $xpath->query('//p');
foreach ($list as $node) { foreach ($list as $node) {
if (strlen($node->nodeValue) > 40) { if (strlen($node->nodeValue) > 40) {
$text .= " ".trim($node->nodeValue); $text .= ' ' . trim($node->nodeValue);
} }
} }
} }
if ($text != "") { if ($text != '') {
$text = trim(str_replace(["\n", "\r"], [" ", " "], $text)); $text = trim(str_replace(["\n", "\r"], [' ', ' '], $text));
while (strpos($text, " ")) { while (strpos($text, ' ')) {
$text = trim(str_replace(" ", " ", $text)); $text = trim(str_replace(' ', ' ', $text));
} }
$siteinfo["text"] = trim(html_entity_decode(substr($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); logger('Siteinfo for ' . $url . ' ' . print_r($siteinfo, true), LOGGER_DEBUG);
Addon::callHooks("getsiteinfo", $siteinfo); Addon::callHooks('getsiteinfo', $siteinfo);
return($siteinfo); return $siteinfo;
} }
/** /**