From 39607b20e2973fb271544b69b3594d5a56b547c8 Mon Sep 17 00:00:00 2001
From: Hypolite Petovan <hypolite@mrpetovan.com>
Date: Sun, 18 Dec 2022 23:24:59 -0500
Subject: [PATCH] Convert custom profile field URL values to rel="me" links

---
 src/Content/Text/HTML.php      | 37 +++++++++------------------
 src/Module/Profile/Profile.php | 25 +++++++++++++++---
 src/Util/Strings.php           | 46 +++++++++++++++++++++++-----------
 3 files changed, 66 insertions(+), 42 deletions(-)

diff --git a/src/Content/Text/HTML.php b/src/Content/Text/HTML.php
index 2e55c71d2..fa5a0a590 100644
--- a/src/Content/Text/HTML.php
+++ b/src/Content/Text/HTML.php
@@ -736,22 +736,22 @@ class HTML
 			'[youtube]$2[/youtube]',
 			$s
 		);
-	
+
 		$s = preg_replace(
 			'#<iframe[^>](.*?)https?://www.youtube.com/embed/([A-Za-z0-9\-_=]+)(.*?)</iframe>#ism',
 			'[youtube]$2[/youtube]',
 			$s
 		);
-	
+
 		$s = preg_replace(
 			'#<iframe[^>](.*?)https?://player.vimeo.com/video/([0-9]+)(.*?)</iframe>#ism',
 			'[vimeo]$2[/vimeo]',
 			$s
 		);
-	
+
 		return $s;
 	}
-	
+
 	/**
 	 * transform link href and img src from relative to absolute
 	 *
@@ -764,30 +764,30 @@ class HTML
 		if (empty($base)) {
 			return $text;
 		}
-	
+
 		$base = rtrim($base, '/');
-	
+
 		$base2 = $base . "/";
-	
+
 		// Replace links
 		$pattern = "/<a([^>]*) href=\"(?!http|https|\/)([^\"]*)\"/";
 		$replace = "<a\${1} href=\"" . $base2 . "\${2}\"";
 		$text = preg_replace($pattern, $replace, $text);
-	
+
 		$pattern = "/<a([^>]*) href=\"(?!http|https)([^\"]*)\"/";
 		$replace = "<a\${1} href=\"" . $base . "\${2}\"";
 		$text = preg_replace($pattern, $replace, $text);
-	
+
 		// Replace images
 		$pattern = "/<img([^>]*) src=\"(?!http|https|\/)([^\"]*)\"/";
 		$replace = "<img\${1} src=\"" . $base2 . "\${2}\"";
 		$text = preg_replace($pattern, $replace, $text);
-	
+
 		$pattern = "/<img([^>]*) src=\"(?!http|https)([^\"]*)\"/";
 		$replace = "<img\${1} src=\"" . $base . "\${2}\"";
 		$text = preg_replace($pattern, $replace, $text);
-	
-	
+
+
 		// Done
 		return $text;
 	}
@@ -907,19 +907,6 @@ class HTML
 		return Renderer::replaceMacros(Renderer::getMarkupTemplate('searchbox.tpl'), $values);
 	}
 
-	/**
-	 * Replace naked text hyperlink with HTML formatted hyperlink
-	 *
-	 * @param string $s
-	 * @return string
-	 */
-	public static function toLink(string $s): string
-	{
-		$s = preg_replace("/(https?\:\/\/[a-zA-Z0-9\:\/\-\?\&\;\.\=\_\~\#\'\%\$\!\+]*)/", ' <a href="$1" target="_blank" rel="noopener noreferrer">$1</a>', $s);
-		$s = preg_replace("/\<(.*?)(src|href)=(.*?)\&amp\;(.*?)\>/ism", '<$1$2=$3&$4>', $s);
-		return $s;
-	}
-
 	/**
 	 * Given a HTML text and a set of filtering reasons, adds a content hiding header with the provided reasons
 	 *
diff --git a/src/Module/Profile/Profile.php b/src/Module/Profile/Profile.php
index bed5ffce7..864b8a76e 100644
--- a/src/Module/Profile/Profile.php
+++ b/src/Module/Profile/Profile.php
@@ -26,7 +26,6 @@ use Friendica\Content\Feature;
 use Friendica\Content\ForumManager;
 use Friendica\Content\Nav;
 use Friendica\Content\Text\BBCode;
-use Friendica\Content\Text\HTML;
 use Friendica\Core\Config\Capability\IManageConfigValues;
 use Friendica\Core\Hook;
 use Friendica\Core\L10n;
@@ -48,6 +47,7 @@ use Friendica\Profile\ProfileField\Repository\ProfileField;
 use Friendica\Protocol\ActivityPub;
 use Friendica\Util\DateTimeFormat;
 use Friendica\Util\Profiler;
+use Friendica\Util\Strings;
 use Friendica\Util\Temporal;
 use Psr\Log\LoggerInterface;
 
@@ -204,7 +204,11 @@ class Profile extends BaseProfile
 		}
 
 		if ($profile['homepage']) {
-			$basic_fields += self::buildField('homepage', $this->t('Homepage:'), HTML::toLink($profile['homepage']));
+			$basic_fields += self::buildField(
+				'homepage',
+				$this->t('Homepage:'),
+				$this->tryRelMe($profile['homepage']) ?: $profile['homepage']
+			);
 		}
 
 		if (
@@ -245,7 +249,7 @@ class Profile extends BaseProfile
 			$custom_fields += self::buildField(
 				'custom_' . $profile_field->order,
 				$profile_field->label,
-				BBCode::convertForUriId($profile['uri-id'], $profile_field->value),
+				$this->tryRelMe($profile_field->value) ?: BBCode::convertForUriId($profile['uri-id'], $profile_field->value),
 				'aprofile custom'
 			);
 		}
@@ -359,4 +363,19 @@ class Profile extends BaseProfile
 
 		return $htmlhead;
 	}
+
+	/**
+	 * Check if the input is an HTTP(S) link and returns a rel="me" link if yes, empty string if not
+	 *
+	 * @param string $input
+	 * @return string
+	 */
+	private function tryRelMe(string $input): string
+	{
+		if (preg_match(Strings::onlyLinkRegEx(), trim($input))) {
+			return '<a href="' . trim($input) . '" target="_blank" rel="noopener noreferrer me">' . trim($input) . '</a>';
+		}
+
+		return '';
+	}
 }
diff --git a/src/Util/Strings.php b/src/Util/Strings.php
index 379f2a252..c5d5c760a 100644
--- a/src/Util/Strings.php
+++ b/src/Util/Strings.php
@@ -380,29 +380,47 @@ class Strings
 	 * Returns the regular expression string to match URLs in a given text
 	 *
 	 * @return string
-	 * @see https://daringfireball.net/2010/07/improved_regex_for_matching_urls
 	 */
 	public static function autoLinkRegEx(): string
 	{
 		return '@
-(?<![=\'\]"/])			# Not preceded by [, =, \', ], ", /
+(?<![=\'\]"/]) # Not preceded by [, =, \', ], ", /
 \b
-(							   # Capture 1: entire matched URL
-  https?://							   # http or https protocol
+(              # Capture 1: entire matched URL
+  ' . self::linkRegEx() . '
+)@xiu';
+	}
+
+	/**
+	 * Returns the regular expression string to match only an HTTP URL
+	 *
+	 * @return string
+	 */
+	public static function onlyLinkRegEx(): string
+	{
+		return '@^' . self::linkRegEx() . '$@xiu';
+	}
+
+	/**
+	 * @return string
+	 * @see https://daringfireball.net/2010/07/improved_regex_for_matching_urls
+	 */
+	private static function linkRegEx(): string
+	{
+		return 'https?://                   # http or https protocol
   (?:
-	[^/\s\xA0`!()\[\]{};:\'",<>?«»“”‘’.]	  # Domain can\'t start with a .
-	[^/\s\xA0`!()\[\]{};:\'",<>?«»“”‘’]+	  # Domain can\'t end with a .
+	[^/\s\xA0`!()\[\]{};:\'",<>?«»“”‘’.]    # Domain can\'t start with a .
+	[^/\s\xA0`!()\[\]{};:\'",<>?«»“”‘’]+    # Domain can\'t end with a .
 	\.
 	[^/\s\xA0`!()\[\]{};:\'".,<>?«»“”‘’]+/? # Followed by a slash
   )
-  (?:								   # One or more:
-	[^\s\xA0()<>]+						   # Run of non-space, non-()<>
-	|								   #   or
-	\(([^\s\xA0()<>]+|(\([^\s()<>]+\)))*\) # balanced parens, up to 2 levels
-	|								   #   or
-	[^\s\xA0`!()\[\]{};:\'".,<>?«»“”‘’]	 # not a space or one of these punct chars
-  )*
-)@xiu';
+  (?:                                       # One or more:
+	[^\s\xA0()<>]+                            # Run of non-space, non-()<>
+	|                                         #   or
+	\(([^\s\xA0()<>]+|(\([^\s()<>]+\)))*\)    # balanced parens, up to 2 levels
+	|								          #   or
+	[^\s\xA0`!()\[\]{};:\'".,<>?«»“”‘’]	      # not a space or one of these punct chars
+  )*';
 	}
 
 	/**