From d498d1520027f005a278f19daad6c0486f74abe4 Mon Sep 17 00:00:00 2001
From: Michael <heluecht@pirati.ca>
Date: Tue, 16 Mar 2021 07:04:16 +0000
Subject: [PATCH] Avoid double fetches

---
 src/Content/OEmbed.php | 72 ++++++++++++++++++++++--------------------
 src/Util/ParseUrl.php  | 20 +++++++++---
 2 files changed, 54 insertions(+), 38 deletions(-)

diff --git a/src/Content/OEmbed.php b/src/Content/OEmbed.php
index 4e26e45f9..bf3c113b7 100644
--- a/src/Content/OEmbed.php
+++ b/src/Content/OEmbed.php
@@ -62,11 +62,12 @@ class OEmbed
 	 *
 	 * @param string $embedurl     The URL from which the data should be fetched.
 	 * @param bool   $no_rich_type If set to true rich type content won't be fetched.
+	 * @param bool   $use_parseurl Use the "ParseUrl" functionality to add additional data
 	 *
 	 * @return \Friendica\Object\OEmbed
 	 * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 	 */
-	public static function fetchURL($embedurl, $no_rich_type = false)
+	public static function fetchURL($embedurl, bool $no_rich_type = false, bool $use_parseurl = true)
 	{
 		$embedurl = trim($embedurl, '\'"');
 
@@ -149,48 +150,51 @@ class OEmbed
 		}
 
 		// Improve the OEmbed data with data from OpenGraph, Twitter cards and other sources
-		$data = ParseUrl::getSiteinfoCached($embedurl, true, false);
-		if (($oembed->type == 'error') && empty($data['title']) && empty($data['text'])) {
-			return $oembed;
-		}
+		if ($use_parseurl) {
+			$data = ParseUrl::getSiteinfoCached($embedurl, true, false);
 
-		if ($no_rich_type || ($oembed->type == 'error')) {
-			$oembed->html = '';
-			$oembed->type = $data['type'];
-
-			if ($oembed->type == 'photo') {
-				$oembed->url = $data['url'];
+			if (($oembed->type == 'error') && empty($data['title']) && empty($data['text'])) {
+				return $oembed;
 			}
-		}
 
-		if (!empty($data['title'])) {
-			$oembed->title = $data['title'];
-		}
+			if ($no_rich_type || ($oembed->type == 'error')) {
+				$oembed->html = '';
+				$oembed->type = $data['type'];
 
-		if (!empty($data['text'])) {
-			$oembed->description = $data['text'];
-		}
+				if ($oembed->type == 'photo') {
+					$oembed->url = $data['url'];
+				}
+			}
 
-		if (!empty($data['publisher_name'])) {
-			$oembed->provider_name = $data['publisher_name'];
-		}
+			if (!empty($data['title'])) {
+				$oembed->title = $data['title'];
+			}
 
-		if (!empty($data['publisher_url'])) {
-			$oembed->provider_url = $data['publisher_url'];
-		}
+			if (!empty($data['text'])) {
+				$oembed->description = $data['text'];
+			}
 
-		if (!empty($data['author_name'])) {
-			$oembed->author_name = $data['author_name'];
-		}
+			if (!empty($data['publisher_name'])) {
+				$oembed->provider_name = $data['publisher_name'];
+			}
 
-		if (!empty($data['author_url'])) {
-			$oembed->author_url = $data['author_url'];
-		}
+			if (!empty($data['publisher_url'])) {
+				$oembed->provider_url = $data['publisher_url'];
+			}
 
-		if (!empty($data['images'])) {
-			$oembed->thumbnail_url = $data['images'][0]['src'];
-			$oembed->thumbnail_width = $data['images'][0]['width'];
-			$oembed->thumbnail_height = $data['images'][0]['height'];
+			if (!empty($data['author_name'])) {
+				$oembed->author_name = $data['author_name'];
+			}
+
+			if (!empty($data['author_url'])) {
+				$oembed->author_url = $data['author_url'];
+			}
+
+			if (!empty($data['images'])) {
+				$oembed->thumbnail_url = $data['images'][0]['src'];
+				$oembed->thumbnail_width = $data['images'][0]['width'];
+				$oembed->thumbnail_height = $data['images'][0]['height'];
+			}
 		}
 
 		Hook::callAll('oembed_fetch_url', $embedurl, $oembed);
diff --git a/src/Util/ParseUrl.php b/src/Util/ParseUrl.php
index 8bde05f23..cca158c26 100644
--- a/src/Util/ParseUrl.php
+++ b/src/Util/ParseUrl.php
@@ -241,7 +241,7 @@ class ParseUrl
 		$body = $curlResult->getBody();
 
 		if ($do_oembed) {
-			$oembed_data = OEmbed::fetchURL($url);
+			$oembed_data = OEmbed::fetchURL($url, false, false);
 
 			if (!empty($oembed_data->type)) {
 				if (!in_array($oembed_data->type, ['error', 'rich', 'image', 'video', 'audio', ''])) {
@@ -250,13 +250,25 @@ class ParseUrl
 
 				// See https://github.com/friendica/friendica/pull/5763#discussion_r217913178
 				if ($siteinfo['type'] != 'photo') {
-					if (isset($oembed_data->title)) {
+					if (!empty($oembed_data->title)) {
 						$siteinfo['title'] = trim($oembed_data->title);
 					}
-					if (isset($oembed_data->description)) {
+					if (!empty($oembed_data->description)) {
 						$siteinfo['text'] = trim($oembed_data->description);
 					}
-					if (isset($oembed_data->thumbnail_url)) {
+					if (!empty($oembed_data->author_name)) {
+						$siteinfo['author_name'] = trim($oembed_data->author_name);
+					}
+					if (!empty($oembed_data->author_url)) {
+						$siteinfo['author_url'] = trim($oembed_data->author_url);
+					}
+					if (!empty($oembed_data->provider_name)) {
+						$siteinfo['publisher_name'] = trim($oembed_data->provider_name);
+					}
+					if (!empty($oembed_data->provider_url)) {
+						$siteinfo['publisher_url'] = trim($oembed_data->provider_url);
+					}
+					if (!empty($oembed_data->thumbnail_url)) {
 						$siteinfo['image'] = $oembed_data->thumbnail_url;
 					}
 				}