We only need to fetch the content once

[friendica.git] / src / Util / ParseUrl.php

diff --git a/src/Util/ParseUrl.php b/src/Util/ParseUrl.php
index f2a5eccd395247956c9e39a4e22c037d9c0567fa..70c5279cc84a42ce9314a202375cb2f1382c6fd5 100644 (file)
--- a/src/Util/ParseUrl.php
+++ b/src/Util/ParseUrl.php
@@ -9,10 +9,10 @@ use DOMDocument;
 use DOMXPath;
 use Friendica\Content\OEmbed;
 use Friendica\Core\Addon;
+use Friendica\Core\Logger;
 use Friendica\Database\DBA;
 use Friendica\Object\Image;
-
-require_once 'include/dba.php';
+use Friendica\Util\Strings;
 
 /**
  * @brief Class with methods for extracting certain content from an url
@@ -48,7 +48,7 @@ class ParseUrl
                }
 
                $parsed_url = DBA::selectFirst('parsed_url', ['content'],
-                       ['url' => normalise_link($url), 'guessing' => !$no_guessing, 'oembed' => $do_oembed]
+                       ['url' => Strings::normaliseLink($url), 'guessing' => !$no_guessing, 'oembed' => $do_oembed]
                );
                if (!empty($parsed_url['content'])) {
                        $data = unserialize($parsed_url['content']);
@@ -60,7 +60,7 @@ class ParseUrl
                DBA::insert(
                        'parsed_url',
                        [
-                               'url' => normalise_link($url), 'guessing' => !$no_guessing,
+                               'url' => Strings::normaliseLink($url), 'guessing' => !$no_guessing,
                                'oembed' => $do_oembed, 'content' => serialize($data),
                                'created' => DateTimeFormat::utcNow()
                        ],
@@ -111,7 +111,7 @@ class ParseUrl
         */
        public static function getSiteinfo($url, $no_guessing = false, $do_oembed = true, $count = 1)
        {
-               $a = get_app();
+               $a = \get_app();
 
                $siteinfo = [];
 
@@ -123,7 +123,7 @@ class ParseUrl
                }
 
                if ($count > 10) {
-                       logger('Endless loop detected for ' . $url, LOGGER_DEBUG);
+                       Logger::log('Endless loop detected for ' . $url, Logger::DEBUG);
                        return $siteinfo;
                }
 
@@ -141,7 +141,7 @@ class ParseUrl
                }
 
                // If the file is too large then exit
-               if ($curlResult->getInfo()['download_content_length'] > 1000000) {
+               if (defaults($curlResult->getInfo(), 'download_content_length', 0) > 1000000) {
                        return $siteinfo;
                }
 
@@ -182,12 +182,11 @@ class ParseUrl
                        $charset = trim(trim(trim(array_pop($matches)), ';,'));
                }
 
-               if ($charset == '') {
-                       $charset = 'utf-8';
-               }
+               if ($charset && strtoupper($charset) != 'UTF-8') {
+                       // See https://github.com/friendica/friendica/issues/5470#issuecomment-418351211
+                       $charset = str_ireplace('latin-1', 'latin1', $charset);
 
-               if (($charset != '') && (strtoupper($charset) != 'UTF-8')) {
-                       logger('detected charset ' . $charset, LOGGER_DEBUG);
+                       Logger::log('detected charset ' . $charset, Logger::DEBUG);
                        $body = iconv($charset, 'UTF-8//TRANSLIT', $body);
                }
 
@@ -421,7 +420,7 @@ class ParseUrl
                        }
                }
 
-               logger('Siteinfo for ' . $url . ' ' . print_r($siteinfo, true), LOGGER_DEBUG);
+               Logger::log('Siteinfo for ' . $url . ' ' . print_r($siteinfo, true), Logger::DEBUG);
 
                Addon::callHooks('getsiteinfo', $siteinfo);